This commit is contained in:
Miku AuahDark
2021-11-13 14:36:39 +08:00
parent c802045cb2
commit a3ea2d5ca0
207 changed files with 32782 additions and 1544 deletions
@@ -0,0 +1,190 @@
----------------------------------------------------------------------------
-- LuaJIT bytecode listing module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module lists the bytecode of a Lua function. If it's loaded by -jbc
-- it hooks into the parser and lists all functions of a chunk as they
-- are parsed.
--
-- Example usage:
--
-- luajit -jbc -e 'local x=0; for i=1,1e6 do x=x+i end; print(x)'
-- luajit -jbc=- foo.lua
-- luajit -jbc=foo.list foo.lua
--
-- Default output is to stderr. To redirect the output to a file, pass a
-- filename as an argument (use '-' for stdout) or set the environment
-- variable LUAJIT_LISTFILE. The file is overwritten every time the module
-- is started.
--
-- This module can also be used programmatically:
--
-- local bc = require("jit.bc")
--
-- local function foo() print("hello") end
--
-- bc.dump(foo) --> -- BYTECODE -- [...]
-- print(bc.line(foo, 2)) --> 0002 KSTR 1 1 ; "hello"
--
-- local out = {
-- -- Do something with each line:
-- write = function(t, ...) io.write(...) end,
-- close = function(t) end,
-- flush = function(t) end,
-- }
-- bc.dump(foo, out)
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local bit = require("bit")
local sub, gsub, format = string.sub, string.gsub, string.format
local byte, band, shr = string.byte, bit.band, bit.rshift
local funcinfo, funcbc, funck = jutil.funcinfo, jutil.funcbc, jutil.funck
local funcuvname = jutil.funcuvname
local bcnames = vmdef.bcnames
local stdout, stderr = io.stdout, io.stderr
------------------------------------------------------------------------------
local function ctlsub(c)
if c == "\n" then return "\\n"
elseif c == "\r" then return "\\r"
elseif c == "\t" then return "\\t"
else return format("\\%03d", byte(c))
end
end
-- Return one bytecode line.
local function bcline(func, pc, prefix)
local ins, m = funcbc(func, pc)
if not ins then return end
local ma, mb, mc = band(m, 7), band(m, 15*8), band(m, 15*128)
local a = band(shr(ins, 8), 0xff)
local oidx = 6*band(ins, 0xff)
local op = sub(bcnames, oidx+1, oidx+6)
local s = format("%04d %s %-6s %3s ",
pc, prefix or " ", op, ma == 0 and "" or a)
local d = shr(ins, 16)
if mc == 13*128 then -- BCMjump
return format("%s=> %04d\n", s, pc+d-0x7fff)
end
if mb ~= 0 then
d = band(d, 0xff)
elseif mc == 0 then
return s.."\n"
end
local kc
if mc == 10*128 then -- BCMstr
kc = funck(func, -d-1)
kc = format(#kc > 40 and '"%.40s"~' or '"%s"', gsub(kc, "%c", ctlsub))
elseif mc == 9*128 then -- BCMnum
kc = funck(func, d)
if op == "TSETM " then kc = kc - 2^52 end
elseif mc == 12*128 then -- BCMfunc
local fi = funcinfo(funck(func, -d-1))
if fi.ffid then
kc = vmdef.ffnames[fi.ffid]
else
kc = fi.loc
end
elseif mc == 5*128 then -- BCMuv
kc = funcuvname(func, d)
end
if ma == 5 then -- BCMuv
local ka = funcuvname(func, a)
if kc then kc = ka.." ; "..kc else kc = ka end
end
if mb ~= 0 then
local b = shr(ins, 24)
if kc then return format("%s%3d %3d ; %s\n", s, b, d, kc) end
return format("%s%3d %3d\n", s, b, d)
end
if kc then return format("%s%3d ; %s\n", s, d, kc) end
if mc == 7*128 and d > 32767 then d = d - 65536 end -- BCMlits
return format("%s%3d\n", s, d)
end
-- Collect branch targets of a function.
local function bctargets(func)
local target = {}
for pc=1,1000000000 do
local ins, m = funcbc(func, pc)
if not ins then break end
if band(m, 15*128) == 13*128 then target[pc+shr(ins, 16)-0x7fff] = true end
end
return target
end
-- Dump bytecode instructions of a function.
local function bcdump(func, out, all)
if not out then out = stdout end
local fi = funcinfo(func)
if all and fi.children then
for n=-1,-1000000000,-1 do
local k = funck(func, n)
if not k then break end
if type(k) == "proto" then bcdump(k, out, true) end
end
end
out:write(format("-- BYTECODE -- %s-%d\n", fi.loc, fi.lastlinedefined))
local target = bctargets(func)
for pc=1,1000000000 do
local s = bcline(func, pc, target[pc] and "=>")
if not s then break end
out:write(s)
end
out:write("\n")
out:flush()
end
------------------------------------------------------------------------------
-- Active flag and output file handle.
local active, out
-- List handler.
local function h_list(func)
return bcdump(func, out)
end
-- Detach list handler.
local function bclistoff()
if active then
active = false
jit.attach(h_list)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach list handler.
local function bcliston(outfile)
if active then bclistoff() end
if not outfile then outfile = os.getenv("LUAJIT_LISTFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stderr
end
jit.attach(h_list, "bc")
active = true
end
-- Public module functions.
return {
line = bcline,
dump = bcdump,
targets = bctargets,
on = bcliston,
off = bclistoff,
start = bcliston -- For -j command line option.
}
@@ -0,0 +1,665 @@
----------------------------------------------------------------------------
-- LuaJIT module to save/list bytecode.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module saves or lists the bytecode for an input file.
-- It's run by the -b command line option.
--
------------------------------------------------------------------------------
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local bit = require("bit")
-- Symbol name prefix for LuaJIT bytecode.
local LJBC_PREFIX = "luaJIT_BC_"
local type, assert = type, assert
local format = string.format
local tremove, tconcat = table.remove, table.concat
------------------------------------------------------------------------------
local function usage()
io.stderr:write[[
Save LuaJIT bytecode: luajit -b[options] input output
-l Only list bytecode.
-s Strip debug info (default).
-g Keep debug info.
-n name Set module name (default: auto-detect from input name).
-t type Set output file type (default: auto-detect from output name).
-a arch Override architecture for object files (default: native).
-o os Override OS for object files (default: native).
-e chunk Use chunk string as input.
-- Stop handling options.
- Use stdin as input and/or stdout as output.
File types: c h obj o raw (default)
]]
os.exit(1)
end
local function check(ok, ...)
if ok then return ok, ... end
io.stderr:write("luajit: ", ...)
io.stderr:write("\n")
os.exit(1)
end
local function readfile(input)
if type(input) == "function" then return input end
if input == "-" then input = nil end
return check(loadfile(input))
end
local function savefile(name, mode)
if name == "-" then return io.stdout end
return check(io.open(name, mode))
end
------------------------------------------------------------------------------
local map_type = {
raw = "raw", c = "c", h = "h", o = "obj", obj = "obj",
}
local map_arch = {
x86 = { e = "le", b = 32, m = 3, p = 0x14c, },
x64 = { e = "le", b = 64, m = 62, p = 0x8664, },
arm = { e = "le", b = 32, m = 40, p = 0x1c0, },
arm64 = { e = "le", b = 64, m = 183, p = 0xaa64, },
arm64be = { e = "be", b = 64, m = 183, },
ppc = { e = "be", b = 32, m = 20, },
mips = { e = "be", b = 32, m = 8, f = 0x50001006, },
mipsel = { e = "le", b = 32, m = 8, f = 0x50001006, },
mips64 = { e = "be", b = 64, m = 8, f = 0x80000007, },
mips64el = { e = "le", b = 64, m = 8, f = 0x80000007, },
mips64r6 = { e = "be", b = 64, m = 8, f = 0xa0000407, },
mips64r6el = { e = "le", b = 64, m = 8, f = 0xa0000407, },
}
local map_os = {
linux = true, windows = true, osx = true, freebsd = true, netbsd = true,
openbsd = true, dragonfly = true, solaris = true,
}
local function checkarg(str, map, err)
str = str:lower()
local s = check(map[str], "unknown ", err)
return type(s) == "string" and s or str
end
local function detecttype(str)
local ext = str:lower():match("%.(%a+)$")
return map_type[ext] or "raw"
end
local function checkmodname(str)
check(str:match("^[%w_.%-]+$"), "bad module name")
return str:gsub("[%.%-]", "_")
end
local function detectmodname(str)
if type(str) == "string" then
local tail = str:match("[^/\\]+$")
if tail then str = tail end
local head = str:match("^(.*)%.[^.]*$")
if head then str = head end
str = str:match("^[%w_.%-]+")
else
str = nil
end
check(str, "cannot derive module name, use -n name")
return str:gsub("[%.%-]", "_")
end
------------------------------------------------------------------------------
local function bcsave_tail(fp, output, s)
local ok, err = fp:write(s)
if ok and output ~= "-" then ok, err = fp:close() end
check(ok, "cannot write ", output, ": ", err)
end
local function bcsave_raw(output, s)
local fp = savefile(output, "wb")
bcsave_tail(fp, output, s)
end
local function bcsave_c(ctx, output, s)
local fp = savefile(output, "w")
if ctx.type == "c" then
fp:write(format([[
#ifdef __cplusplus
extern "C"
#endif
#ifdef _WIN32
__declspec(dllexport)
#endif
const unsigned char %s%s[] = {
]], LJBC_PREFIX, ctx.modname))
else
fp:write(format([[
#define %s%s_SIZE %d
static const unsigned char %s%s[] = {
]], LJBC_PREFIX, ctx.modname, #s, LJBC_PREFIX, ctx.modname))
end
local t, n, m = {}, 0, 0
for i=1,#s do
local b = tostring(string.byte(s, i))
m = m + #b + 1
if m > 78 then
fp:write(tconcat(t, ",", 1, n), ",\n")
n, m = 0, #b + 1
end
n = n + 1
t[n] = b
end
bcsave_tail(fp, output, tconcat(t, ",", 1, n).."\n};\n")
end
local function bcsave_elfobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct {
uint8_t emagic[4], eclass, eendian, eversion, eosabi, eabiversion, epad[7];
uint16_t type, machine;
uint32_t version;
uint32_t entry, phofs, shofs;
uint32_t flags;
uint16_t ehsize, phentsize, phnum, shentsize, shnum, shstridx;
} ELF32header;
typedef struct {
uint8_t emagic[4], eclass, eendian, eversion, eosabi, eabiversion, epad[7];
uint16_t type, machine;
uint32_t version;
uint64_t entry, phofs, shofs;
uint32_t flags;
uint16_t ehsize, phentsize, phnum, shentsize, shnum, shstridx;
} ELF64header;
typedef struct {
uint32_t name, type, flags, addr, ofs, size, link, info, align, entsize;
} ELF32sectheader;
typedef struct {
uint32_t name, type;
uint64_t flags, addr, ofs, size;
uint32_t link, info;
uint64_t align, entsize;
} ELF64sectheader;
typedef struct {
uint32_t name, value, size;
uint8_t info, other;
uint16_t sectidx;
} ELF32symbol;
typedef struct {
uint32_t name;
uint8_t info, other;
uint16_t sectidx;
uint64_t value, size;
} ELF64symbol;
typedef struct {
ELF32header hdr;
ELF32sectheader sect[6];
ELF32symbol sym[2];
uint8_t space[4096];
} ELF32obj;
typedef struct {
ELF64header hdr;
ELF64sectheader sect[6];
ELF64symbol sym[2];
uint8_t space[4096];
} ELF64obj;
]]
local symname = LJBC_PREFIX..ctx.modname
local ai = assert(map_arch[ctx.arch])
local is64, isbe = ai.b == 64, ai.e == "be"
-- Handle different host/target endianess.
local function f32(x) return x end
local f16, fofs = f32, f32
if ffi.abi("be") ~= isbe then
f32 = bit.bswap
function f16(x) return bit.rshift(bit.bswap(x), 16) end
if is64 then
local two32 = ffi.cast("int64_t", 2^32)
function fofs(x) return bit.bswap(x)*two32 end
else
fofs = f32
end
end
-- Create ELF object and fill in header.
local o = ffi.new(is64 and "ELF64obj" or "ELF32obj")
local hdr = o.hdr
if ctx.os == "bsd" or ctx.os == "other" then -- Determine native hdr.eosabi.
local bf = assert(io.open("/bin/ls", "rb"))
local bs = bf:read(9)
bf:close()
ffi.copy(o, bs, 9)
check(hdr.emagic[0] == 127, "no support for writing native object files")
else
hdr.emagic = "\127ELF"
hdr.eosabi = ({ freebsd=9, netbsd=2, openbsd=12, solaris=6 })[ctx.os] or 0
end
hdr.eclass = is64 and 2 or 1
hdr.eendian = isbe and 2 or 1
hdr.eversion = 1
hdr.type = f16(1)
hdr.machine = f16(ai.m)
hdr.flags = f32(ai.f or 0)
hdr.version = f32(1)
hdr.shofs = fofs(ffi.offsetof(o, "sect"))
hdr.ehsize = f16(ffi.sizeof(hdr))
hdr.shentsize = f16(ffi.sizeof(o.sect[0]))
hdr.shnum = f16(6)
hdr.shstridx = f16(2)
-- Fill in sections and symbols.
local sofs, ofs = ffi.offsetof(o, "space"), 1
for i,name in ipairs{
".symtab", ".shstrtab", ".strtab", ".rodata", ".note.GNU-stack",
} do
local sect = o.sect[i]
sect.align = fofs(1)
sect.name = f32(ofs)
ffi.copy(o.space+ofs, name)
ofs = ofs + #name+1
end
o.sect[1].type = f32(2) -- .symtab
o.sect[1].link = f32(3)
o.sect[1].info = f32(1)
o.sect[1].align = fofs(8)
o.sect[1].ofs = fofs(ffi.offsetof(o, "sym"))
o.sect[1].entsize = fofs(ffi.sizeof(o.sym[0]))
o.sect[1].size = fofs(ffi.sizeof(o.sym))
o.sym[1].name = f32(1)
o.sym[1].sectidx = f16(4)
o.sym[1].size = fofs(#s)
o.sym[1].info = 17
o.sect[2].type = f32(3) -- .shstrtab
o.sect[2].ofs = fofs(sofs)
o.sect[2].size = fofs(ofs)
o.sect[3].type = f32(3) -- .strtab
o.sect[3].ofs = fofs(sofs + ofs)
o.sect[3].size = fofs(#symname+2)
ffi.copy(o.space+ofs+1, symname)
ofs = ofs + #symname + 2
o.sect[4].type = f32(1) -- .rodata
o.sect[4].flags = fofs(2)
o.sect[4].ofs = fofs(sofs + ofs)
o.sect[4].size = fofs(#s)
o.sect[5].type = f32(1) -- .note.GNU-stack
o.sect[5].ofs = fofs(sofs + ofs + #s)
-- Write ELF object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, ffi.sizeof(o)-4096+ofs))
bcsave_tail(fp, output, s)
end
local function bcsave_peobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct {
uint16_t arch, nsects;
uint32_t time, symtabofs, nsyms;
uint16_t opthdrsz, flags;
} PEheader;
typedef struct {
char name[8];
uint32_t vsize, vaddr, size, ofs, relocofs, lineofs;
uint16_t nreloc, nline;
uint32_t flags;
} PEsection;
typedef struct __attribute((packed)) {
union {
char name[8];
uint32_t nameref[2];
};
uint32_t value;
int16_t sect;
uint16_t type;
uint8_t scl, naux;
} PEsym;
typedef struct __attribute((packed)) {
uint32_t size;
uint16_t nreloc, nline;
uint32_t cksum;
uint16_t assoc;
uint8_t comdatsel, unused[3];
} PEsymaux;
typedef struct {
PEheader hdr;
PEsection sect[2];
// Must be an even number of symbol structs.
PEsym sym0;
PEsymaux sym0aux;
PEsym sym1;
PEsymaux sym1aux;
PEsym sym2;
PEsym sym3;
uint32_t strtabsize;
uint8_t space[4096];
} PEobj;
]]
local symname = LJBC_PREFIX..ctx.modname
local ai = assert(map_arch[ctx.arch])
local is64 = ai.b == 64
local symexport = " /EXPORT:"..symname..",DATA "
-- The file format is always little-endian. Swap if the host is big-endian.
local function f32(x) return x end
local f16 = f32
if ffi.abi("be") then
f32 = bit.bswap
function f16(x) return bit.rshift(bit.bswap(x), 16) end
end
-- Create PE object and fill in header.
local o = ffi.new("PEobj")
local hdr = o.hdr
hdr.arch = f16(assert(ai.p))
hdr.nsects = f16(2)
hdr.symtabofs = f32(ffi.offsetof(o, "sym0"))
hdr.nsyms = f32(6)
-- Fill in sections and symbols.
o.sect[0].name = ".drectve"
o.sect[0].size = f32(#symexport)
o.sect[0].flags = f32(0x00100a00)
o.sym0.sect = f16(1)
o.sym0.scl = 3
o.sym0.name = ".drectve"
o.sym0.naux = 1
o.sym0aux.size = f32(#symexport)
o.sect[1].name = ".rdata"
o.sect[1].size = f32(#s)
o.sect[1].flags = f32(0x40300040)
o.sym1.sect = f16(2)
o.sym1.scl = 3
o.sym1.name = ".rdata"
o.sym1.naux = 1
o.sym1aux.size = f32(#s)
o.sym2.sect = f16(2)
o.sym2.scl = 2
o.sym2.nameref[1] = f32(4)
o.sym3.sect = f16(-1)
o.sym3.scl = 2
o.sym3.value = f32(1)
o.sym3.name = "@feat.00" -- Mark as SafeSEH compliant.
ffi.copy(o.space, symname)
local ofs = #symname + 1
o.strtabsize = f32(ofs + 4)
o.sect[0].ofs = f32(ffi.offsetof(o, "space") + ofs)
ffi.copy(o.space + ofs, symexport)
ofs = ofs + #symexport
o.sect[1].ofs = f32(ffi.offsetof(o, "space") + ofs)
-- Write PE object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, ffi.sizeof(o)-4096+ofs))
bcsave_tail(fp, output, s)
end
local function bcsave_machobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct
{
uint32_t magic, cputype, cpusubtype, filetype, ncmds, sizeofcmds, flags;
} mach_header;
typedef struct
{
mach_header; uint32_t reserved;
} mach_header_64;
typedef struct {
uint32_t cmd, cmdsize;
char segname[16];
uint32_t vmaddr, vmsize, fileoff, filesize;
uint32_t maxprot, initprot, nsects, flags;
} mach_segment_command;
typedef struct {
uint32_t cmd, cmdsize;
char segname[16];
uint64_t vmaddr, vmsize, fileoff, filesize;
uint32_t maxprot, initprot, nsects, flags;
} mach_segment_command_64;
typedef struct {
char sectname[16], segname[16];
uint32_t addr, size;
uint32_t offset, align, reloff, nreloc, flags;
uint32_t reserved1, reserved2;
} mach_section;
typedef struct {
char sectname[16], segname[16];
uint64_t addr, size;
uint32_t offset, align, reloff, nreloc, flags;
uint32_t reserved1, reserved2, reserved3;
} mach_section_64;
typedef struct {
uint32_t cmd, cmdsize, symoff, nsyms, stroff, strsize;
} mach_symtab_command;
typedef struct {
int32_t strx;
uint8_t type, sect;
int16_t desc;
uint32_t value;
} mach_nlist;
typedef struct {
uint32_t strx;
uint8_t type, sect;
uint16_t desc;
uint64_t value;
} mach_nlist_64;
typedef struct
{
uint32_t magic, nfat_arch;
} mach_fat_header;
typedef struct
{
uint32_t cputype, cpusubtype, offset, size, align;
} mach_fat_arch;
typedef struct {
struct {
mach_header hdr;
mach_segment_command seg;
mach_section sec;
mach_symtab_command sym;
} arch[1];
mach_nlist sym_entry;
uint8_t space[4096];
} mach_obj;
typedef struct {
struct {
mach_header_64 hdr;
mach_segment_command_64 seg;
mach_section_64 sec;
mach_symtab_command sym;
} arch[1];
mach_nlist_64 sym_entry;
uint8_t space[4096];
} mach_obj_64;
typedef struct {
mach_fat_header fat;
mach_fat_arch fat_arch[2];
struct {
mach_header hdr;
mach_segment_command seg;
mach_section sec;
mach_symtab_command sym;
} arch[2];
mach_nlist sym_entry;
uint8_t space[4096];
} mach_fat_obj;
]]
local symname = '_'..LJBC_PREFIX..ctx.modname
local isfat, is64, align, mobj = false, false, 4, "mach_obj"
if ctx.arch == "x64" then
is64, align, mobj = true, 8, "mach_obj_64"
elseif ctx.arch == "arm" then
isfat, mobj = true, "mach_fat_obj"
elseif ctx.arch == "arm64" then
is64, align, isfat, mobj = true, 8, true, "mach_fat_obj"
else
check(ctx.arch == "x86", "unsupported architecture for OSX")
end
local function aligned(v, a) return bit.band(v+a-1, -a) end
local be32 = bit.bswap -- Mach-O FAT is BE, supported archs are LE.
-- Create Mach-O object and fill in header.
local o = ffi.new(mobj)
local mach_size = aligned(ffi.offsetof(o, "space")+#symname+2, align)
local cputype = ({ x86={7}, x64={0x01000007}, arm={7,12}, arm64={0x01000007,0x0100000c} })[ctx.arch]
local cpusubtype = ({ x86={3}, x64={3}, arm={3,9}, arm64={3,0} })[ctx.arch]
if isfat then
o.fat.magic = be32(0xcafebabe)
o.fat.nfat_arch = be32(#cpusubtype)
end
-- Fill in sections and symbols.
for i=0,#cpusubtype-1 do
local ofs = 0
if isfat then
local a = o.fat_arch[i]
a.cputype = be32(cputype[i+1])
a.cpusubtype = be32(cpusubtype[i+1])
-- Subsequent slices overlap each other to share data.
ofs = ffi.offsetof(o, "arch") + i*ffi.sizeof(o.arch[0])
a.offset = be32(ofs)
a.size = be32(mach_size-ofs+#s)
end
local a = o.arch[i]
a.hdr.magic = is64 and 0xfeedfacf or 0xfeedface
a.hdr.cputype = cputype[i+1]
a.hdr.cpusubtype = cpusubtype[i+1]
a.hdr.filetype = 1
a.hdr.ncmds = 2
a.hdr.sizeofcmds = ffi.sizeof(a.seg)+ffi.sizeof(a.sec)+ffi.sizeof(a.sym)
a.seg.cmd = is64 and 0x19 or 0x1
a.seg.cmdsize = ffi.sizeof(a.seg)+ffi.sizeof(a.sec)
a.seg.vmsize = #s
a.seg.fileoff = mach_size-ofs
a.seg.filesize = #s
a.seg.maxprot = 1
a.seg.initprot = 1
a.seg.nsects = 1
ffi.copy(a.sec.sectname, "__data")
ffi.copy(a.sec.segname, "__DATA")
a.sec.size = #s
a.sec.offset = mach_size-ofs
a.sym.cmd = 2
a.sym.cmdsize = ffi.sizeof(a.sym)
a.sym.symoff = ffi.offsetof(o, "sym_entry")-ofs
a.sym.nsyms = 1
a.sym.stroff = ffi.offsetof(o, "sym_entry")+ffi.sizeof(o.sym_entry)-ofs
a.sym.strsize = aligned(#symname+2, align)
end
o.sym_entry.type = 0xf
o.sym_entry.sect = 1
o.sym_entry.strx = 1
ffi.copy(o.space+1, symname)
-- Write Macho-O object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, mach_size))
bcsave_tail(fp, output, s)
end
local function bcsave_obj(ctx, output, s)
local ok, ffi = pcall(require, "ffi")
check(ok, "FFI library required to write this file type")
if ctx.os == "windows" then
return bcsave_peobj(ctx, output, s, ffi)
elseif ctx.os == "osx" then
return bcsave_machobj(ctx, output, s, ffi)
else
return bcsave_elfobj(ctx, output, s, ffi)
end
end
------------------------------------------------------------------------------
local function bclist(input, output)
local f = readfile(input)
require("jit.bc").dump(f, savefile(output, "w"), true)
end
local function bcsave(ctx, input, output)
local f = readfile(input)
local s = string.dump(f, ctx.strip)
local t = ctx.type
if not t then
t = detecttype(output)
ctx.type = t
end
if t == "raw" then
bcsave_raw(output, s)
else
if not ctx.modname then ctx.modname = detectmodname(input) end
if t == "obj" then
bcsave_obj(ctx, output, s)
else
bcsave_c(ctx, output, s)
end
end
end
local function docmd(...)
local arg = {...}
local n = 1
local list = false
local ctx = {
strip = true, arch = jit.arch, os = jit.os:lower(),
type = false, modname = false,
}
while n <= #arg do
local a = arg[n]
if type(a) == "string" and a:sub(1, 1) == "-" and a ~= "-" then
tremove(arg, n)
if a == "--" then break end
for m=2,#a do
local opt = a:sub(m, m)
if opt == "l" then
list = true
elseif opt == "s" then
ctx.strip = true
elseif opt == "g" then
ctx.strip = false
else
if arg[n] == nil or m ~= #a then usage() end
if opt == "e" then
if n ~= 1 then usage() end
arg[1] = check(loadstring(arg[1]))
elseif opt == "n" then
ctx.modname = checkmodname(tremove(arg, n))
elseif opt == "t" then
ctx.type = checkarg(tremove(arg, n), map_type, "file type")
elseif opt == "a" then
ctx.arch = checkarg(tremove(arg, n), map_arch, "architecture")
elseif opt == "o" then
ctx.os = checkarg(tremove(arg, n), map_os, "OS name")
else
usage()
end
end
end
else
n = n + 1
end
end
if list then
if #arg == 0 or #arg > 2 then usage() end
bclist(arg[1], arg[2] or "-")
else
if #arg ~= 2 then usage() end
bcsave(ctx, arg[1], arg[2])
end
end
------------------------------------------------------------------------------
-- Public module functions.
return {
start = docmd -- Process -b command line option.
}
@@ -0,0 +1,689 @@
----------------------------------------------------------------------------
-- LuaJIT ARM disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles most user-mode ARMv7 instructions
-- NYI: Advanced SIMD and VFP instructions.
------------------------------------------------------------------------------
local type = type
local sub, byte, format = string.sub, string.byte, string.format
local match, gmatch = string.match, string.gmatch
local concat = table.concat
local bit = require("bit")
local band, bor, ror, tohex = bit.band, bit.bor, bit.ror, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Opcode maps
------------------------------------------------------------------------------
local map_loadc = {
shift = 8, mask = 15,
[10] = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "vmovFmDN", "vstmFNdr",
_ = {
shift = 21, mask = 1,
[0] = "vstrFdl",
{ shift = 16, mask = 15, [13] = "vpushFdr", _ = "vstmdbFNdr", }
},
},
{
shift = 23, mask = 3,
[0] = "vmovFDNm",
{ shift = 16, mask = 15, [13] = "vpopFdr", _ = "vldmFNdr", },
_ = {
shift = 21, mask = 1,
[0] = "vldrFdl", "vldmdbFNdr",
},
},
},
[11] = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "vmovGmDN", "vstmGNdr",
_ = {
shift = 21, mask = 1,
[0] = "vstrGdl",
{ shift = 16, mask = 15, [13] = "vpushGdr", _ = "vstmdbGNdr", }
},
},
{
shift = 23, mask = 3,
[0] = "vmovGDNm",
{ shift = 16, mask = 15, [13] = "vpopGdr", _ = "vldmGNdr", },
_ = {
shift = 21, mask = 1,
[0] = "vldrGdl", "vldmdbGNdr",
},
},
},
_ = {
shift = 0, mask = 0 -- NYI ldc, mcrr, mrrc.
},
}
local map_vfps = {
shift = 6, mask = 0x2c001,
[0] = "vmlaF.dnm", "vmlsF.dnm",
[0x04000] = "vnmlsF.dnm", [0x04001] = "vnmlaF.dnm",
[0x08000] = "vmulF.dnm", [0x08001] = "vnmulF.dnm",
[0x0c000] = "vaddF.dnm", [0x0c001] = "vsubF.dnm",
[0x20000] = "vdivF.dnm",
[0x24000] = "vfnmsF.dnm", [0x24001] = "vfnmaF.dnm",
[0x28000] = "vfmaF.dnm", [0x28001] = "vfmsF.dnm",
[0x2c000] = "vmovF.dY",
[0x2c001] = {
shift = 7, mask = 0x1e01,
[0] = "vmovF.dm", "vabsF.dm",
[0x0200] = "vnegF.dm", [0x0201] = "vsqrtF.dm",
[0x0800] = "vcmpF.dm", [0x0801] = "vcmpeF.dm",
[0x0a00] = "vcmpzF.d", [0x0a01] = "vcmpzeF.d",
[0x0e01] = "vcvtG.dF.m",
[0x1000] = "vcvt.f32.u32Fdm", [0x1001] = "vcvt.f32.s32Fdm",
[0x1800] = "vcvtr.u32F.dm", [0x1801] = "vcvt.u32F.dm",
[0x1a00] = "vcvtr.s32F.dm", [0x1a01] = "vcvt.s32F.dm",
},
}
local map_vfpd = {
shift = 6, mask = 0x2c001,
[0] = "vmlaG.dnm", "vmlsG.dnm",
[0x04000] = "vnmlsG.dnm", [0x04001] = "vnmlaG.dnm",
[0x08000] = "vmulG.dnm", [0x08001] = "vnmulG.dnm",
[0x0c000] = "vaddG.dnm", [0x0c001] = "vsubG.dnm",
[0x20000] = "vdivG.dnm",
[0x24000] = "vfnmsG.dnm", [0x24001] = "vfnmaG.dnm",
[0x28000] = "vfmaG.dnm", [0x28001] = "vfmsG.dnm",
[0x2c000] = "vmovG.dY",
[0x2c001] = {
shift = 7, mask = 0x1e01,
[0] = "vmovG.dm", "vabsG.dm",
[0x0200] = "vnegG.dm", [0x0201] = "vsqrtG.dm",
[0x0800] = "vcmpG.dm", [0x0801] = "vcmpeG.dm",
[0x0a00] = "vcmpzG.d", [0x0a01] = "vcmpzeG.d",
[0x0e01] = "vcvtF.dG.m",
[0x1000] = "vcvt.f64.u32GdFm", [0x1001] = "vcvt.f64.s32GdFm",
[0x1800] = "vcvtr.u32FdG.m", [0x1801] = "vcvt.u32FdG.m",
[0x1a00] = "vcvtr.s32FdG.m", [0x1a01] = "vcvt.s32FdG.m",
},
}
local map_datac = {
shift = 24, mask = 1,
[0] = {
shift = 4, mask = 1,
[0] = {
shift = 8, mask = 15,
[10] = map_vfps,
[11] = map_vfpd,
-- NYI cdp, mcr, mrc.
},
{
shift = 8, mask = 15,
[10] = {
shift = 20, mask = 15,
[0] = "vmovFnD", "vmovFDn",
[14] = "vmsrD",
[15] = { shift = 12, mask = 15, [15] = "vmrs", _ = "vmrsD", },
},
},
},
"svcT",
}
local map_loadcu = {
shift = 0, mask = 0, -- NYI unconditional CP load/store.
}
local map_datacu = {
shift = 0, mask = 0, -- NYI unconditional CP data.
}
local map_simddata = {
shift = 0, mask = 0, -- NYI SIMD data.
}
local map_simdload = {
shift = 0, mask = 0, -- NYI SIMD load/store, preload.
}
local map_preload = {
shift = 0, mask = 0, -- NYI preload.
}
local map_media = {
shift = 20, mask = 31,
[0] = false,
{ --01
shift = 5, mask = 7,
[0] = "sadd16DNM", "sasxDNM", "ssaxDNM", "ssub16DNM",
"sadd8DNM", false, false, "ssub8DNM",
},
{ --02
shift = 5, mask = 7,
[0] = "qadd16DNM", "qasxDNM", "qsaxDNM", "qsub16DNM",
"qadd8DNM", false, false, "qsub8DNM",
},
{ --03
shift = 5, mask = 7,
[0] = "shadd16DNM", "shasxDNM", "shsaxDNM", "shsub16DNM",
"shadd8DNM", false, false, "shsub8DNM",
},
false,
{ --05
shift = 5, mask = 7,
[0] = "uadd16DNM", "uasxDNM", "usaxDNM", "usub16DNM",
"uadd8DNM", false, false, "usub8DNM",
},
{ --06
shift = 5, mask = 7,
[0] = "uqadd16DNM", "uqasxDNM", "uqsaxDNM", "uqsub16DNM",
"uqadd8DNM", false, false, "uqsub8DNM",
},
{ --07
shift = 5, mask = 7,
[0] = "uhadd16DNM", "uhasxDNM", "uhsaxDNM", "uhsub16DNM",
"uhadd8DNM", false, false, "uhsub8DNM",
},
{ --08
shift = 5, mask = 7,
[0] = "pkhbtDNMU", false, "pkhtbDNMU",
{ shift = 16, mask = 15, [15] = "sxtb16DMU", _ = "sxtab16DNMU", },
"pkhbtDNMU", "selDNM", "pkhtbDNMU",
},
false,
{ --0a
shift = 5, mask = 7,
[0] = "ssatDxMu", "ssat16DxM", "ssatDxMu",
{ shift = 16, mask = 15, [15] = "sxtbDMU", _ = "sxtabDNMU", },
"ssatDxMu", false, "ssatDxMu",
},
{ --0b
shift = 5, mask = 7,
[0] = "ssatDxMu", "revDM", "ssatDxMu",
{ shift = 16, mask = 15, [15] = "sxthDMU", _ = "sxtahDNMU", },
"ssatDxMu", "rev16DM", "ssatDxMu",
},
{ --0c
shift = 5, mask = 7,
[3] = { shift = 16, mask = 15, [15] = "uxtb16DMU", _ = "uxtab16DNMU", },
},
false,
{ --0e
shift = 5, mask = 7,
[0] = "usatDwMu", "usat16DwM", "usatDwMu",
{ shift = 16, mask = 15, [15] = "uxtbDMU", _ = "uxtabDNMU", },
"usatDwMu", false, "usatDwMu",
},
{ --0f
shift = 5, mask = 7,
[0] = "usatDwMu", "rbitDM", "usatDwMu",
{ shift = 16, mask = 15, [15] = "uxthDMU", _ = "uxtahDNMU", },
"usatDwMu", "revshDM", "usatDwMu",
},
{ --10
shift = 12, mask = 15,
[15] = {
shift = 5, mask = 7,
"smuadNMS", "smuadxNMS", "smusdNMS", "smusdxNMS",
},
_ = {
shift = 5, mask = 7,
[0] = "smladNMSD", "smladxNMSD", "smlsdNMSD", "smlsdxNMSD",
},
},
false, false, false,
{ --14
shift = 5, mask = 7,
[0] = "smlaldDNMS", "smlaldxDNMS", "smlsldDNMS", "smlsldxDNMS",
},
{ --15
shift = 5, mask = 7,
[0] = { shift = 12, mask = 15, [15] = "smmulNMS", _ = "smmlaNMSD", },
{ shift = 12, mask = 15, [15] = "smmulrNMS", _ = "smmlarNMSD", },
false, false, false, false,
"smmlsNMSD", "smmlsrNMSD",
},
false, false,
{ --18
shift = 5, mask = 7,
[0] = { shift = 12, mask = 15, [15] = "usad8NMS", _ = "usada8NMSD", },
},
false,
{ --1a
shift = 5, mask = 3, [2] = "sbfxDMvw",
},
{ --1b
shift = 5, mask = 3, [2] = "sbfxDMvw",
},
{ --1c
shift = 5, mask = 3,
[0] = { shift = 0, mask = 15, [15] = "bfcDvX", _ = "bfiDMvX", },
},
{ --1d
shift = 5, mask = 3,
[0] = { shift = 0, mask = 15, [15] = "bfcDvX", _ = "bfiDMvX", },
},
{ --1e
shift = 5, mask = 3, [2] = "ubfxDMvw",
},
{ --1f
shift = 5, mask = 3, [2] = "ubfxDMvw",
},
}
local map_load = {
shift = 21, mask = 9,
{
shift = 20, mask = 5,
[0] = "strtDL", "ldrtDL", [4] = "strbtDL", [5] = "ldrbtDL",
},
_ = {
shift = 20, mask = 5,
[0] = "strDL", "ldrDL", [4] = "strbDL", [5] = "ldrbDL",
}
}
local map_load1 = {
shift = 4, mask = 1,
[0] = map_load, map_media,
}
local map_loadm = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "stmdaNR", "stmNR",
{ shift = 16, mask = 63, [45] = "pushR", _ = "stmdbNR", }, "stmibNR",
},
{
shift = 23, mask = 3,
[0] = "ldmdaNR", { shift = 16, mask = 63, [61] = "popR", _ = "ldmNR", },
"ldmdbNR", "ldmibNR",
},
}
local map_data = {
shift = 21, mask = 15,
[0] = "andDNPs", "eorDNPs", "subDNPs", "rsbDNPs",
"addDNPs", "adcDNPs", "sbcDNPs", "rscDNPs",
"tstNP", "teqNP", "cmpNP", "cmnNP",
"orrDNPs", "movDPs", "bicDNPs", "mvnDPs",
}
local map_mul = {
shift = 21, mask = 7,
[0] = "mulNMSs", "mlaNMSDs", "umaalDNMS", "mlsDNMS",
"umullDNMSs", "umlalDNMSs", "smullDNMSs", "smlalDNMSs",
}
local map_sync = {
shift = 20, mask = 15, -- NYI: brackets around N. R(D+1) for ldrexd/strexd.
[0] = "swpDMN", false, false, false,
"swpbDMN", false, false, false,
"strexDMN", "ldrexDN", "strexdDN", "ldrexdDN",
"strexbDMN", "ldrexbDN", "strexhDN", "ldrexhDN",
}
local map_mulh = {
shift = 21, mask = 3,
[0] = { shift = 5, mask = 3,
[0] = "smlabbNMSD", "smlatbNMSD", "smlabtNMSD", "smlattNMSD", },
{ shift = 5, mask = 3,
[0] = "smlawbNMSD", "smulwbNMS", "smlawtNMSD", "smulwtNMS", },
{ shift = 5, mask = 3,
[0] = "smlalbbDNMS", "smlaltbDNMS", "smlalbtDNMS", "smlalttDNMS", },
{ shift = 5, mask = 3,
[0] = "smulbbNMS", "smultbNMS", "smulbtNMS", "smulttNMS", },
}
local map_misc = {
shift = 4, mask = 7,
-- NYI: decode PSR bits of msr.
[0] = { shift = 21, mask = 1, [0] = "mrsD", "msrM", },
{ shift = 21, mask = 3, "bxM", false, "clzDM", },
{ shift = 21, mask = 3, "bxjM", },
{ shift = 21, mask = 3, "blxM", },
false,
{ shift = 21, mask = 3, [0] = "qaddDMN", "qsubDMN", "qdaddDMN", "qdsubDMN", },
false,
{ shift = 21, mask = 3, "bkptK", },
}
local map_datar = {
shift = 4, mask = 9,
[9] = {
shift = 5, mask = 3,
[0] = { shift = 24, mask = 1, [0] = map_mul, map_sync, },
{ shift = 20, mask = 1, [0] = "strhDL", "ldrhDL", },
{ shift = 20, mask = 1, [0] = "ldrdDL", "ldrsbDL", },
{ shift = 20, mask = 1, [0] = "strdDL", "ldrshDL", },
},
_ = {
shift = 20, mask = 25,
[16] = { shift = 7, mask = 1, [0] = map_misc, map_mulh, },
_ = {
shift = 0, mask = 0xffffffff,
[bor(0xe1a00000)] = "nop",
_ = map_data,
}
},
}
local map_datai = {
shift = 20, mask = 31, -- NYI: decode PSR bits of msr. Decode imm12.
[16] = "movwDW", [20] = "movtDW",
[18] = { shift = 0, mask = 0xf00ff, [0] = "nopv6", _ = "msrNW", },
[22] = "msrNW",
_ = map_data,
}
local map_branch = {
shift = 24, mask = 1,
[0] = "bB", "blB"
}
local map_condins = {
[0] = map_datar, map_datai, map_load, map_load1,
map_loadm, map_branch, map_loadc, map_datac
}
-- NYI: setend.
local map_uncondins = {
[0] = false, map_simddata, map_simdload, map_preload,
false, "blxB", map_loadcu, map_datacu,
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "sp", "lr", "pc",
}
local map_cond = {
[0] = "eq", "ne", "hs", "lo", "mi", "pl", "vs", "vc",
"hi", "ls", "ge", "lt", "gt", "le", "al",
}
local map_shift = { [0] = "lsl", "lsr", "asr", "ror", }
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then
extra = "\t->"..sym
elseif band(ctx.op, 0x0e000000) ~= 0x0a000000 then
extra = "\t; 0x"..tohex(ctx.rel)
end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-5s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-5s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
-- Format operand 2 of load/store opcodes.
local function fmtload(ctx, op, pos)
local base = map_gpr[band(rshift(op, 16), 15)]
local x, ofs
local ext = (band(op, 0x04000000) == 0)
if not ext and band(op, 0x02000000) == 0 then
ofs = band(op, 4095)
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
ofs = "#"..ofs
elseif ext and band(op, 0x00400000) ~= 0 then
ofs = band(op, 15) + band(rshift(op, 4), 0xf0)
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
ofs = "#"..ofs
else
ofs = map_gpr[band(op, 15)]
if ext or band(op, 0xfe0) == 0 then
elseif band(op, 0xfe0) == 0x60 then
ofs = format("%s, rrx", ofs)
else
local sh = band(rshift(op, 7), 31)
if sh == 0 then sh = 32 end
ofs = format("%s, %s #%d", ofs, map_shift[band(rshift(op, 5), 3)], sh)
end
if band(op, 0x00800000) == 0 then ofs = "-"..ofs end
end
if ofs == "#0" then
x = format("[%s]", base)
elseif band(op, 0x01000000) == 0 then
x = format("[%s], %s", base, ofs)
else
x = format("[%s, %s]", base, ofs)
end
if band(op, 0x01200000) == 0x01200000 then x = x.."!" end
return x
end
-- Format operand 2 of vector load/store opcodes.
local function fmtvload(ctx, op, pos)
local base = map_gpr[band(rshift(op, 16), 15)]
local ofs = band(op, 255)*4
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
if ofs == 0 then
return format("[%s]", base)
else
return format("[%s, #%d]", base, ofs)
end
end
local function fmtvr(op, vr, sh0, sh1)
if vr == "s" then
return format("s%d", 2*band(rshift(op, sh0), 15)+band(rshift(op, sh1), 1))
else
return format("d%d", band(rshift(op, sh0), 15)+band(rshift(op, sh1-4), 16))
end
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
local op = bor(lshift(b3, 24), lshift(b2, 16), lshift(b1, 8), b0)
local operands = {}
local suffix = ""
local last, name, pat
local vr
ctx.op = op
ctx.rel = nil
local cond = rshift(op, 28)
local opat
if cond == 15 then
opat = map_uncondins[band(rshift(op, 25), 7)]
else
if cond ~= 14 then suffix = map_cond[cond] end
opat = map_condins[band(rshift(op, 25), 7)]
end
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
opat = opat[band(rshift(op, opat.shift), opat.mask)] or opat._
end
name, pat = match(opat, "^([a-z0-9]*)(.*)")
if sub(pat, 1, 1) == "." then
local s2, p2 = match(pat, "^([a-z0-9.]*)(.*)")
suffix = suffix..s2
pat = p2
end
for p in gmatch(pat, ".") do
local x = nil
if p == "D" then
x = map_gpr[band(rshift(op, 12), 15)]
elseif p == "N" then
x = map_gpr[band(rshift(op, 16), 15)]
elseif p == "S" then
x = map_gpr[band(rshift(op, 8), 15)]
elseif p == "M" then
x = map_gpr[band(op, 15)]
elseif p == "d" then
x = fmtvr(op, vr, 12, 22)
elseif p == "n" then
x = fmtvr(op, vr, 16, 7)
elseif p == "m" then
x = fmtvr(op, vr, 0, 5)
elseif p == "P" then
if band(op, 0x02000000) ~= 0 then
x = ror(band(op, 255), 2*band(rshift(op, 8), 15))
else
x = map_gpr[band(op, 15)]
if band(op, 0xff0) ~= 0 then
operands[#operands+1] = x
local s = map_shift[band(rshift(op, 5), 3)]
local r = nil
if band(op, 0xf90) == 0 then
if s == "ror" then s = "rrx" else r = "#32" end
elseif band(op, 0x10) == 0 then
r = "#"..band(rshift(op, 7), 31)
else
r = map_gpr[band(rshift(op, 8), 15)]
end
if name == "mov" then name = s; x = r
elseif r then x = format("%s %s", s, r)
else x = s end
end
end
elseif p == "L" then
x = fmtload(ctx, op, pos)
elseif p == "l" then
x = fmtvload(ctx, op, pos)
elseif p == "B" then
local addr = ctx.addr + pos + 8 + arshift(lshift(op, 8), 6)
if cond == 15 then addr = addr + band(rshift(op, 23), 2) end
ctx.rel = addr
x = "0x"..tohex(addr)
elseif p == "F" then
vr = "s"
elseif p == "G" then
vr = "d"
elseif p == "." then
suffix = suffix..(vr == "s" and ".f32" or ".f64")
elseif p == "R" then
if band(op, 0x00200000) ~= 0 and #operands == 1 then
operands[1] = operands[1].."!"
end
local t = {}
for i=0,15 do
if band(rshift(op, i), 1) == 1 then t[#t+1] = map_gpr[i] end
end
x = "{"..concat(t, ", ").."}"
elseif p == "r" then
if band(op, 0x00200000) ~= 0 and #operands == 2 then
operands[1] = operands[1].."!"
end
local s = tonumber(sub(last, 2))
local n = band(op, 255)
if vr == "d" then n = rshift(n, 1) end
operands[#operands] = format("{%s-%s%d}", last, vr, s+n-1)
elseif p == "W" then
x = band(op, 0x0fff) + band(rshift(op, 4), 0xf000)
elseif p == "T" then
x = "#0x"..tohex(band(op, 0x00ffffff), 6)
elseif p == "U" then
x = band(rshift(op, 7), 31)
if x == 0 then x = nil end
elseif p == "u" then
x = band(rshift(op, 7), 31)
if band(op, 0x40) == 0 then
if x == 0 then x = nil else x = "lsl #"..x end
else
if x == 0 then x = "asr #32" else x = "asr #"..x end
end
elseif p == "v" then
x = band(rshift(op, 7), 31)
elseif p == "w" then
x = band(rshift(op, 16), 31)
elseif p == "x" then
x = band(rshift(op, 16), 31) + 1
elseif p == "X" then
x = band(rshift(op, 16), 31) - last + 1
elseif p == "Y" then
x = band(rshift(op, 12), 0xf0) + band(op, 0x0f)
elseif p == "K" then
x = "#0x"..tohex(band(rshift(op, 4), 0x0000fff0) + band(op, 15), 4)
elseif p == "s" then
if band(op, 0x00100000) ~= 0 then suffix = "s"..suffix end
else
assert(false)
end
if x then
last = x
if type(x) == "number" then x = "#"..x end
operands[#operands+1] = x
end
end
return putop(ctx, name..suffix, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
ctx.pos = ofs
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 16 then return map_gpr[r] end
return "d"..(r-16)
end
-- Public module functions.
return {
create = create,
disass = disass,
regname = regname
}
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,12 @@
----------------------------------------------------------------------------
-- LuaJIT ARM64BE disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- ARM64 instructions are always little-endian. So just forward to the
-- common ARM64 disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
return require((string.match(..., ".*%.") or "").."dis_arm64")
@@ -0,0 +1,694 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT/X license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles all standard MIPS32R1/R2 instructions.
-- Default mode is big-endian, but see: dis_mipsel.lua
------------------------------------------------------------------------------
local type = type
local byte, format = string.byte, string.format
local match, gmatch = string.match, string.gmatch
local concat = table.concat
local bit = require("bit")
local band, bor, tohex = bit.band, bit.bor, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Extended opcode maps common to all MIPS releases
------------------------------------------------------------------------------
local map_srl = { shift = 21, mask = 1, [0] = "srlDTA", "rotrDTA", }
local map_srlv = { shift = 6, mask = 1, [0] = "srlvDTS", "rotrvDTS", }
local map_cop0 = {
shift = 25, mask = 1,
[0] = {
shift = 21, mask = 15,
[0] = "mfc0TDW", [4] = "mtc0TDW",
[10] = "rdpgprDT",
[11] = { shift = 5, mask = 1, [0] = "diT0", "eiT0", },
[14] = "wrpgprDT",
}, {
shift = 0, mask = 63,
[1] = "tlbr", [2] = "tlbwi", [6] = "tlbwr", [8] = "tlbp",
[24] = "eret", [31] = "deret",
[32] = "wait",
},
}
------------------------------------------------------------------------------
-- Primary and extended opcode maps for MIPS R1-R5
------------------------------------------------------------------------------
local map_movci = { shift = 16, mask = 1, [0] = "movfDSC", "movtDSC", }
local map_special = {
shift = 0, mask = 63,
[0] = { shift = 0, mask = -1, [0] = "nop", _ = "sllDTA" },
map_movci, map_srl, "sraDTA",
"sllvDTS", false, map_srlv, "sravDTS",
"jrS", "jalrD1S", "movzDST", "movnDST",
"syscallY", "breakY", false, "sync",
"mfhiD", "mthiS", "mfloD", "mtloS",
"dsllvDST", false, "dsrlvDST", "dsravDST",
"multST", "multuST", "divST", "divuST",
"dmultST", "dmultuST", "ddivST", "ddivuST",
"addDST", "addu|moveDST0", "subDST", "subu|neguDS0T",
"andDST", "or|moveDST0", "xorDST", "nor|notDST0",
false, false, "sltDST", "sltuDST",
"daddDST", "dadduDST", "dsubDST", "dsubuDST",
"tgeSTZ", "tgeuSTZ", "tltSTZ", "tltuSTZ",
"teqSTZ", false, "tneSTZ", false,
"dsllDTA", false, "dsrlDTA", "dsraDTA",
"dsll32DTA", false, "dsrl32DTA", "dsra32DTA",
}
local map_special2 = {
shift = 0, mask = 63,
[0] = "maddST", "madduST", "mulDST", false,
"msubST", "msubuST",
[32] = "clzDS", [33] = "cloDS",
[63] = "sdbbpY",
}
local map_bshfl = {
shift = 6, mask = 31,
[2] = "wsbhDT",
[16] = "sebDT",
[24] = "sehDT",
}
local map_dbshfl = {
shift = 6, mask = 31,
[2] = "dsbhDT",
[5] = "dshdDT",
}
local map_special3 = {
shift = 0, mask = 63,
[0] = "extTSAK", [1] = "dextmTSAP", [3] = "dextTSAK",
[4] = "insTSAL", [6] = "dinsuTSEQ", [7] = "dinsTSAL",
[32] = map_bshfl, [36] = map_dbshfl, [59] = "rdhwrTD",
}
local map_regimm = {
shift = 16, mask = 31,
[0] = "bltzSB", "bgezSB", "bltzlSB", "bgezlSB",
false, false, false, false,
"tgeiSI", "tgeiuSI", "tltiSI", "tltiuSI",
"teqiSI", false, "tneiSI", false,
"bltzalSB", "bgezalSB", "bltzallSB", "bgezallSB",
false, false, false, false,
false, false, false, false,
false, false, false, "synciSO",
}
local map_cop1s = {
shift = 0, mask = 63,
[0] = "add.sFGH", "sub.sFGH", "mul.sFGH", "div.sFGH",
"sqrt.sFG", "abs.sFG", "mov.sFG", "neg.sFG",
"round.l.sFG", "trunc.l.sFG", "ceil.l.sFG", "floor.l.sFG",
"round.w.sFG", "trunc.w.sFG", "ceil.w.sFG", "floor.w.sFG",
false,
{ shift = 16, mask = 1, [0] = "movf.sFGC", "movt.sFGC" },
"movz.sFGT", "movn.sFGT",
false, "recip.sFG", "rsqrt.sFG", false,
false, false, false, false,
false, false, false, false,
false, "cvt.d.sFG", false, false,
"cvt.w.sFG", "cvt.l.sFG", "cvt.ps.sFGH", false,
false, false, false, false,
false, false, false, false,
"c.f.sVGH", "c.un.sVGH", "c.eq.sVGH", "c.ueq.sVGH",
"c.olt.sVGH", "c.ult.sVGH", "c.ole.sVGH", "c.ule.sVGH",
"c.sf.sVGH", "c.ngle.sVGH", "c.seq.sVGH", "c.ngl.sVGH",
"c.lt.sVGH", "c.nge.sVGH", "c.le.sVGH", "c.ngt.sVGH",
}
local map_cop1d = {
shift = 0, mask = 63,
[0] = "add.dFGH", "sub.dFGH", "mul.dFGH", "div.dFGH",
"sqrt.dFG", "abs.dFG", "mov.dFG", "neg.dFG",
"round.l.dFG", "trunc.l.dFG", "ceil.l.dFG", "floor.l.dFG",
"round.w.dFG", "trunc.w.dFG", "ceil.w.dFG", "floor.w.dFG",
false,
{ shift = 16, mask = 1, [0] = "movf.dFGC", "movt.dFGC" },
"movz.dFGT", "movn.dFGT",
false, "recip.dFG", "rsqrt.dFG", false,
false, false, false, false,
false, false, false, false,
"cvt.s.dFG", false, false, false,
"cvt.w.dFG", "cvt.l.dFG", false, false,
false, false, false, false,
false, false, false, false,
"c.f.dVGH", "c.un.dVGH", "c.eq.dVGH", "c.ueq.dVGH",
"c.olt.dVGH", "c.ult.dVGH", "c.ole.dVGH", "c.ule.dVGH",
"c.df.dVGH", "c.ngle.dVGH", "c.deq.dVGH", "c.ngl.dVGH",
"c.lt.dVGH", "c.nge.dVGH", "c.le.dVGH", "c.ngt.dVGH",
}
local map_cop1ps = {
shift = 0, mask = 63,
[0] = "add.psFGH", "sub.psFGH", "mul.psFGH", false,
false, "abs.psFG", "mov.psFG", "neg.psFG",
false, false, false, false,
false, false, false, false,
false,
{ shift = 16, mask = 1, [0] = "movf.psFGC", "movt.psFGC" },
"movz.psFGT", "movn.psFGT",
false, false, false, false,
false, false, false, false,
false, false, false, false,
"cvt.s.puFG", false, false, false,
false, false, false, false,
"cvt.s.plFG", false, false, false,
"pll.psFGH", "plu.psFGH", "pul.psFGH", "puu.psFGH",
"c.f.psVGH", "c.un.psVGH", "c.eq.psVGH", "c.ueq.psVGH",
"c.olt.psVGH", "c.ult.psVGH", "c.ole.psVGH", "c.ule.psVGH",
"c.psf.psVGH", "c.ngle.psVGH", "c.pseq.psVGH", "c.ngl.psVGH",
"c.lt.psVGH", "c.nge.psVGH", "c.le.psVGH", "c.ngt.psVGH",
}
local map_cop1w = {
shift = 0, mask = 63,
[32] = "cvt.s.wFG", [33] = "cvt.d.wFG",
}
local map_cop1l = {
shift = 0, mask = 63,
[32] = "cvt.s.lFG", [33] = "cvt.d.lFG",
}
local map_cop1bc = {
shift = 16, mask = 3,
[0] = "bc1fCB", "bc1tCB", "bc1flCB", "bc1tlCB",
}
local map_cop1 = {
shift = 21, mask = 31,
[0] = "mfc1TG", "dmfc1TG", "cfc1TG", "mfhc1TG",
"mtc1TG", "dmtc1TG", "ctc1TG", "mthc1TG",
map_cop1bc, false, false, false,
false, false, false, false,
map_cop1s, map_cop1d, false, false,
map_cop1w, map_cop1l, map_cop1ps,
}
local map_cop1x = {
shift = 0, mask = 63,
[0] = "lwxc1FSX", "ldxc1FSX", false, false,
false, "luxc1FSX", false, false,
"swxc1FSX", "sdxc1FSX", false, false,
false, "suxc1FSX", false, "prefxMSX",
false, false, false, false,
false, false, false, false,
false, false, false, false,
false, false, "alnv.psFGHS", false,
"madd.sFRGH", "madd.dFRGH", false, false,
false, false, "madd.psFRGH", false,
"msub.sFRGH", "msub.dFRGH", false, false,
false, false, "msub.psFRGH", false,
"nmadd.sFRGH", "nmadd.dFRGH", false, false,
false, false, "nmadd.psFRGH", false,
"nmsub.sFRGH", "nmsub.dFRGH", false, false,
false, false, "nmsub.psFRGH", false,
}
local map_pri = {
[0] = map_special, map_regimm, "jJ", "jalJ",
"beq|beqz|bST00B", "bne|bnezST0B", "blezSB", "bgtzSB",
"addiTSI", "addiu|liTS0I", "sltiTSI", "sltiuTSI",
"andiTSU", "ori|liTS0U", "xoriTSU", "luiTU",
map_cop0, map_cop1, false, map_cop1x,
"beql|beqzlST0B", "bnel|bnezlST0B", "blezlSB", "bgtzlSB",
"daddiTSI", "daddiuTSI", false, false,
map_special2, "jalxJ", false, map_special3,
"lbTSO", "lhTSO", "lwlTSO", "lwTSO",
"lbuTSO", "lhuTSO", "lwrTSO", false,
"sbTSO", "shTSO", "swlTSO", "swTSO",
false, false, "swrTSO", "cacheNSO",
"llTSO", "lwc1HSO", "lwc2TSO", "prefNSO",
false, "ldc1HSO", "ldc2TSO", "ldTSO",
"scTSO", "swc1HSO", "swc2TSO", false,
false, "sdc1HSO", "sdc2TSO", "sdTSO",
}
------------------------------------------------------------------------------
-- Primary and extended opcode maps for MIPS R6
------------------------------------------------------------------------------
local map_mul_r6 = { shift = 6, mask = 3, [2] = "mulDST", [3] = "muhDST" }
local map_mulu_r6 = { shift = 6, mask = 3, [2] = "muluDST", [3] = "muhuDST" }
local map_div_r6 = { shift = 6, mask = 3, [2] = "divDST", [3] = "modDST" }
local map_divu_r6 = { shift = 6, mask = 3, [2] = "divuDST", [3] = "moduDST" }
local map_dmul_r6 = { shift = 6, mask = 3, [2] = "dmulDST", [3] = "dmuhDST" }
local map_dmulu_r6 = { shift = 6, mask = 3, [2] = "dmuluDST", [3] = "dmuhuDST" }
local map_ddiv_r6 = { shift = 6, mask = 3, [2] = "ddivDST", [3] = "dmodDST" }
local map_ddivu_r6 = { shift = 6, mask = 3, [2] = "ddivuDST", [3] = "dmoduDST" }
local map_special_r6 = {
shift = 0, mask = 63,
[0] = { shift = 0, mask = -1, [0] = "nop", _ = "sllDTA" },
false, map_srl, "sraDTA",
"sllvDTS", false, map_srlv, "sravDTS",
"jrS", "jalrD1S", false, false,
"syscallY", "breakY", false, "sync",
"clzDS", "cloDS", "dclzDS", "dcloDS",
"dsllvDST", "dlsaDSTA", "dsrlvDST", "dsravDST",
map_mul_r6, map_mulu_r6, map_div_r6, map_divu_r6,
map_dmul_r6, map_dmulu_r6, map_ddiv_r6, map_ddivu_r6,
"addDST", "addu|moveDST0", "subDST", "subu|neguDS0T",
"andDST", "or|moveDST0", "xorDST", "nor|notDST0",
false, false, "sltDST", "sltuDST",
"daddDST", "dadduDST", "dsubDST", "dsubuDST",
"tgeSTZ", "tgeuSTZ", "tltSTZ", "tltuSTZ",
"teqSTZ", "seleqzDST", "tneSTZ", "selnezDST",
"dsllDTA", false, "dsrlDTA", "dsraDTA",
"dsll32DTA", false, "dsrl32DTA", "dsra32DTA",
}
local map_bshfl_r6 = {
shift = 9, mask = 3,
[1] = "alignDSTa",
_ = {
shift = 6, mask = 31,
[0] = "bitswapDT",
[2] = "wsbhDT",
[16] = "sebDT",
[24] = "sehDT",
}
}
local map_dbshfl_r6 = {
shift = 9, mask = 3,
[1] = "dalignDSTa",
_ = {
shift = 6, mask = 31,
[0] = "dbitswapDT",
[2] = "dsbhDT",
[5] = "dshdDT",
}
}
local map_special3_r6 = {
shift = 0, mask = 63,
[0] = "extTSAK", [1] = "dextmTSAP", [3] = "dextTSAK",
[4] = "insTSAL", [6] = "dinsuTSEQ", [7] = "dinsTSAL",
[32] = map_bshfl_r6, [36] = map_dbshfl_r6, [59] = "rdhwrTD",
}
local map_regimm_r6 = {
shift = 16, mask = 31,
[0] = "bltzSB", [1] = "bgezSB",
[6] = "dahiSI", [30] = "datiSI",
[23] = "sigrieI", [31] = "synciSO",
}
local map_pcrel_r6 = {
shift = 19, mask = 3,
[0] = "addiupcS2", "lwpcS2", "lwupcS2", {
shift = 18, mask = 1,
[0] = "ldpcS3", { shift = 16, mask = 3, [2] = "auipcSI", [3] = "aluipcSI" }
}
}
local map_cop1s_r6 = {
shift = 0, mask = 63,
[0] = "add.sFGH", "sub.sFGH", "mul.sFGH", "div.sFGH",
"sqrt.sFG", "abs.sFG", "mov.sFG", "neg.sFG",
"round.l.sFG", "trunc.l.sFG", "ceil.l.sFG", "floor.l.sFG",
"round.w.sFG", "trunc.w.sFG", "ceil.w.sFG", "floor.w.sFG",
"sel.sFGH", false, false, false,
"seleqz.sFGH", "recip.sFG", "rsqrt.sFG", "selnez.sFGH",
"maddf.sFGH", "msubf.sFGH", "rint.sFG", "class.sFG",
"min.sFGH", "mina.sFGH", "max.sFGH", "maxa.sFGH",
false, "cvt.d.sFG", false, false,
"cvt.w.sFG", "cvt.l.sFG",
}
local map_cop1d_r6 = {
shift = 0, mask = 63,
[0] = "add.dFGH", "sub.dFGH", "mul.dFGH", "div.dFGH",
"sqrt.dFG", "abs.dFG", "mov.dFG", "neg.dFG",
"round.l.dFG", "trunc.l.dFG", "ceil.l.dFG", "floor.l.dFG",
"round.w.dFG", "trunc.w.dFG", "ceil.w.dFG", "floor.w.dFG",
"sel.dFGH", false, false, false,
"seleqz.dFGH", "recip.dFG", "rsqrt.dFG", "selnez.dFGH",
"maddf.dFGH", "msubf.dFGH", "rint.dFG", "class.dFG",
"min.dFGH", "mina.dFGH", "max.dFGH", "maxa.dFGH",
"cvt.s.dFG", false, false, false,
"cvt.w.dFG", "cvt.l.dFG",
}
local map_cop1w_r6 = {
shift = 0, mask = 63,
[0] = "cmp.af.sFGH", "cmp.un.sFGH", "cmp.eq.sFGH", "cmp.ueq.sFGH",
"cmp.lt.sFGH", "cmp.ult.sFGH", "cmp.le.sFGH", "cmp.ule.sFGH",
"cmp.saf.sFGH", "cmp.sun.sFGH", "cmp.seq.sFGH", "cmp.sueq.sFGH",
"cmp.slt.sFGH", "cmp.sult.sFGH", "cmp.sle.sFGH", "cmp.sule.sFGH",
false, "cmp.or.sFGH", "cmp.une.sFGH", "cmp.ne.sFGH",
false, false, false, false,
false, "cmp.sor.sFGH", "cmp.sune.sFGH", "cmp.sne.sFGH",
false, false, false, false,
"cvt.s.wFG", "cvt.d.wFG",
}
local map_cop1l_r6 = {
shift = 0, mask = 63,
[0] = "cmp.af.dFGH", "cmp.un.dFGH", "cmp.eq.dFGH", "cmp.ueq.dFGH",
"cmp.lt.dFGH", "cmp.ult.dFGH", "cmp.le.dFGH", "cmp.ule.dFGH",
"cmp.saf.dFGH", "cmp.sun.dFGH", "cmp.seq.dFGH", "cmp.sueq.dFGH",
"cmp.slt.dFGH", "cmp.sult.dFGH", "cmp.sle.dFGH", "cmp.sule.dFGH",
false, "cmp.or.dFGH", "cmp.une.dFGH", "cmp.ne.dFGH",
false, false, false, false,
false, "cmp.sor.dFGH", "cmp.sune.dFGH", "cmp.sne.dFGH",
false, false, false, false,
"cvt.s.lFG", "cvt.d.lFG",
}
local map_cop1_r6 = {
shift = 21, mask = 31,
[0] = "mfc1TG", "dmfc1TG", "cfc1TG", "mfhc1TG",
"mtc1TG", "dmtc1TG", "ctc1TG", "mthc1TG",
false, "bc1eqzHB", false, false,
false, "bc1nezHB", false, false,
map_cop1s_r6, map_cop1d_r6, false, false,
map_cop1w_r6, map_cop1l_r6,
}
local function maprs_popTS(rs, rt)
if rt == 0 then return 0 elseif rs == 0 then return 1
elseif rs == rt then return 2 else return 3 end
end
local map_pop06_r6 = {
maprs = maprs_popTS, [0] = "blezSB", "blezalcTB", "bgezalcTB", "bgeucSTB"
}
local map_pop07_r6 = {
maprs = maprs_popTS, [0] = "bgtzSB", "bgtzalcTB", "bltzalcTB", "bltucSTB"
}
local map_pop26_r6 = {
maprs = maprs_popTS, "blezcTB", "bgezcTB", "bgecSTB"
}
local map_pop27_r6 = {
maprs = maprs_popTS, "bgtzcTB", "bltzcTB", "bltcSTB"
}
local function maprs_popS(rs, rt)
if rs == 0 then return 0 else return 1 end
end
local map_pop66_r6 = {
maprs = maprs_popS, [0] = "jicTI", "beqzcSb"
}
local map_pop76_r6 = {
maprs = maprs_popS, [0] = "jialcTI", "bnezcSb"
}
local function maprs_popST(rs, rt)
if rs >= rt then return 0 elseif rs == 0 then return 1 else return 2 end
end
local map_pop10_r6 = {
maprs = maprs_popST, [0] = "bovcSTB", "beqzalcTB", "beqcSTB"
}
local map_pop30_r6 = {
maprs = maprs_popST, [0] = "bnvcSTB", "bnezalcTB", "bnecSTB"
}
local map_pri_r6 = {
[0] = map_special_r6, map_regimm_r6, "jJ", "jalJ",
"beq|beqz|bST00B", "bne|bnezST0B", map_pop06_r6, map_pop07_r6,
map_pop10_r6, "addiu|liTS0I", "sltiTSI", "sltiuTSI",
"andiTSU", "ori|liTS0U", "xoriTSU", "aui|luiTS0U",
map_cop0, map_cop1_r6, false, false,
false, false, map_pop26_r6, map_pop27_r6,
map_pop30_r6, "daddiuTSI", false, false,
false, "dauiTSI", false, map_special3_r6,
"lbTSO", "lhTSO", false, "lwTSO",
"lbuTSO", "lhuTSO", false, false,
"sbTSO", "shTSO", false, "swTSO",
false, false, false, false,
false, "lwc1HSO", "bc#", false,
false, "ldc1HSO", map_pop66_r6, "ldTSO",
false, "swc1HSO", "balc#", map_pcrel_r6,
false, "sdc1HSO", map_pop76_r6, "sdTSO",
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
"r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
"r24", "r25", "r26", "r27", "r28", "sp", "r30", "ra",
}
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then extra = "\t->"..sym end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-7s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-7s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
local function get_be(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
return bor(lshift(b0, 24), lshift(b1, 16), lshift(b2, 8), b3)
end
local function get_le(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
return bor(lshift(b3, 24), lshift(b2, 16), lshift(b1, 8), b0)
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local op = ctx:get()
local operands = {}
local last = nil
ctx.op = op
ctx.rel = nil
local opat = ctx.map_pri[rshift(op, 26)]
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
if opat.maprs then
opat = opat[opat.maprs(band(rshift(op,21),31), band(rshift(op,16),31))]
else
opat = opat[band(rshift(op, opat.shift), opat.mask)] or opat._
end
end
local name, pat = match(opat, "^([a-z0-9_.]*)(.*)")
local altname, pat2 = match(pat, "|([a-z0-9_.|]*)(.*)")
if altname then pat = pat2 end
for p in gmatch(pat, ".") do
local x = nil
if p == "S" then
x = map_gpr[band(rshift(op, 21), 31)]
elseif p == "T" then
x = map_gpr[band(rshift(op, 16), 31)]
elseif p == "D" then
x = map_gpr[band(rshift(op, 11), 31)]
elseif p == "F" then
x = "f"..band(rshift(op, 6), 31)
elseif p == "G" then
x = "f"..band(rshift(op, 11), 31)
elseif p == "H" then
x = "f"..band(rshift(op, 16), 31)
elseif p == "R" then
x = "f"..band(rshift(op, 21), 31)
elseif p == "A" then
x = band(rshift(op, 6), 31)
elseif p == "a" then
x = band(rshift(op, 6), 7)
elseif p == "E" then
x = band(rshift(op, 6), 31) + 32
elseif p == "M" then
x = band(rshift(op, 11), 31)
elseif p == "N" then
x = band(rshift(op, 16), 31)
elseif p == "C" then
x = band(rshift(op, 18), 7)
if x == 0 then x = nil end
elseif p == "K" then
x = band(rshift(op, 11), 31) + 1
elseif p == "P" then
x = band(rshift(op, 11), 31) + 33
elseif p == "L" then
x = band(rshift(op, 11), 31) - last + 1
elseif p == "Q" then
x = band(rshift(op, 11), 31) - last + 33
elseif p == "I" then
x = arshift(lshift(op, 16), 16)
elseif p == "2" then
x = arshift(lshift(op, 13), 11)
elseif p == "3" then
x = arshift(lshift(op, 14), 11)
elseif p == "U" then
x = band(op, 0xffff)
elseif p == "O" then
local disp = arshift(lshift(op, 16), 16)
operands[#operands] = format("%d(%s)", disp, last)
elseif p == "X" then
local index = map_gpr[band(rshift(op, 16), 31)]
operands[#operands] = format("%s(%s)", index, last)
elseif p == "B" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 16), 14) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "b" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 11), 9) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "#" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 6), 4) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "J" then
local a = ctx.addr + ctx.pos
x = a - band(a, 0x0fffffff) + band(op, 0x03ffffff)*4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "V" then
x = band(rshift(op, 8), 7)
if x == 0 then x = nil end
elseif p == "W" then
x = band(op, 7)
if x == 0 then x = nil end
elseif p == "Y" then
x = band(rshift(op, 6), 0x000fffff)
if x == 0 then x = nil end
elseif p == "Z" then
x = band(rshift(op, 6), 1023)
if x == 0 then x = nil end
elseif p == "0" then
if last == "r0" or last == 0 then
local n = #operands
operands[n] = nil
last = operands[n-1]
if altname then
local a1, a2 = match(altname, "([^|]*)|(.*)")
if a1 then name, altname = a1, a2
else name = altname end
end
end
elseif p == "1" then
if last == "ra" then
operands[#operands] = nil
end
else
assert(false)
end
if x then operands[#operands+1] = x; last = x end
end
return putop(ctx, name, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
stop = stop - stop % 4
ctx.pos = ofs - ofs % 4
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
ctx.get = get_be
ctx.map_pri = map_pri
return ctx
end
local function create_el(code, addr, out)
local ctx = create(code, addr, out)
ctx.get = get_le
return ctx
end
local function create_r6(code, addr, out)
local ctx = create(code, addr, out)
ctx.map_pri = map_pri_r6
return ctx
end
local function create_r6_el(code, addr, out)
local ctx = create(code, addr, out)
ctx.get = get_le
ctx.map_pri = map_pri_r6
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
local function disass_el(code, addr, out)
create_el(code, addr, out):disass()
end
local function disass_r6(code, addr, out)
create_r6(code, addr, out):disass()
end
local function disass_r6_el(code, addr, out)
create_r6_el(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 32 then return map_gpr[r] end
return "f"..(r-32)
end
-- Public module functions.
return {
create = create,
create_el = create_el,
create_r6 = create_r6,
create_r6_el = create_r6_el,
disass = disass,
disass_el = disass_el,
disass_r6 = disass_r6,
disass_r6_el = disass_r6_el,
regname = regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the big-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create,
disass = dis_mips.disass,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64EL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_el,
disass = dis_mips.disass_el,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64R6 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the r6 big-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_r6,
disass = dis_mips.disass_r6,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64R6EL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the r6 little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_r6_el,
disass = dis_mips.disass_r6_el,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPSEL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_el,
disass = dis_mips.disass_el,
regname = dis_mips.regname
}
@@ -0,0 +1,591 @@
----------------------------------------------------------------------------
-- LuaJIT PPC disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT/X license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles all common, non-privileged 32/64 bit PowerPC instructions
-- plus the e500 SPE instructions and some Cell/Xenon extensions.
--
-- NYI: VMX, VMX128
------------------------------------------------------------------------------
local type = type
local byte, format = string.byte, string.format
local match, gmatch, gsub = string.match, string.gmatch, string.gsub
local concat = table.concat
local bit = require("bit")
local band, bor, tohex = bit.band, bit.bor, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Primary and extended opcode maps
------------------------------------------------------------------------------
local map_crops = {
shift = 1, mask = 1023,
[0] = "mcrfXX",
[33] = "crnor|crnotCCC=", [129] = "crandcCCC",
[193] = "crxor|crclrCCC%", [225] = "crnandCCC",
[257] = "crandCCC", [289] = "creqv|crsetCCC%",
[417] = "crorcCCC", [449] = "cror|crmoveCCC=",
[16] = "b_lrKB", [528] = "b_ctrKB",
[150] = "isync",
}
local map_rlwinm = setmetatable({
shift = 0, mask = -1,
},
{ __index = function(t, x)
local rot = band(rshift(x, 11), 31)
local mb = band(rshift(x, 6), 31)
local me = band(rshift(x, 1), 31)
if mb == 0 and me == 31-rot then
return "slwiRR~A."
elseif me == 31 and mb == 32-rot then
return "srwiRR~-A."
else
return "rlwinmRR~AAA."
end
end
})
local map_rld = {
shift = 2, mask = 7,
[0] = "rldiclRR~HM.", "rldicrRR~HM.", "rldicRR~HM.", "rldimiRR~HM.",
{
shift = 1, mask = 1,
[0] = "rldclRR~RM.", "rldcrRR~RM.",
},
}
local map_ext = setmetatable({
shift = 1, mask = 1023,
[0] = "cmp_YLRR", [32] = "cmpl_YLRR",
[4] = "twARR", [68] = "tdARR",
[8] = "subfcRRR.", [40] = "subfRRR.",
[104] = "negRR.", [136] = "subfeRRR.",
[200] = "subfzeRR.", [232] = "subfmeRR.",
[520] = "subfcoRRR.", [552] = "subfoRRR.",
[616] = "negoRR.", [648] = "subfeoRRR.",
[712] = "subfzeoRR.", [744] = "subfmeoRR.",
[9] = "mulhduRRR.", [73] = "mulhdRRR.", [233] = "mulldRRR.",
[457] = "divduRRR.", [489] = "divdRRR.",
[745] = "mulldoRRR.",
[969] = "divduoRRR.", [1001] = "divdoRRR.",
[10] = "addcRRR.", [138] = "addeRRR.",
[202] = "addzeRR.", [234] = "addmeRR.", [266] = "addRRR.",
[522] = "addcoRRR.", [650] = "addeoRRR.",
[714] = "addzeoRR.", [746] = "addmeoRR.", [778] = "addoRRR.",
[11] = "mulhwuRRR.", [75] = "mulhwRRR.", [235] = "mullwRRR.",
[459] = "divwuRRR.", [491] = "divwRRR.",
[747] = "mullwoRRR.",
[971] = "divwouRRR.", [1003] = "divwoRRR.",
[15] = "iselltRRR", [47] = "iselgtRRR", [79] = "iseleqRRR",
[144] = { shift = 20, mask = 1, [0] = "mtcrfRZ~", "mtocrfRZ~", },
[19] = { shift = 20, mask = 1, [0] = "mfcrR", "mfocrfRZ", },
[371] = { shift = 11, mask = 1023, [392] = "mftbR", [424] = "mftbuR", },
[339] = {
shift = 11, mask = 1023,
[32] = "mferR", [256] = "mflrR", [288] = "mfctrR", [16] = "mfspefscrR",
},
[467] = {
shift = 11, mask = 1023,
[32] = "mtxerR", [256] = "mtlrR", [288] = "mtctrR", [16] = "mtspefscrR",
},
[20] = "lwarxRR0R", [84] = "ldarxRR0R",
[21] = "ldxRR0R", [53] = "lduxRRR",
[149] = "stdxRR0R", [181] = "stduxRRR",
[341] = "lwaxRR0R", [373] = "lwauxRRR",
[23] = "lwzxRR0R", [55] = "lwzuxRRR",
[87] = "lbzxRR0R", [119] = "lbzuxRRR",
[151] = "stwxRR0R", [183] = "stwuxRRR",
[215] = "stbxRR0R", [247] = "stbuxRRR",
[279] = "lhzxRR0R", [311] = "lhzuxRRR",
[343] = "lhaxRR0R", [375] = "lhauxRRR",
[407] = "sthxRR0R", [439] = "sthuxRRR",
[54] = "dcbst-R0R", [86] = "dcbf-R0R",
[150] = "stwcxRR0R.", [214] = "stdcxRR0R.",
[246] = "dcbtst-R0R", [278] = "dcbt-R0R",
[310] = "eciwxRR0R", [438] = "ecowxRR0R",
[470] = "dcbi-RR",
[598] = {
shift = 21, mask = 3,
[0] = "sync", "lwsync", "ptesync",
},
[758] = "dcba-RR",
[854] = "eieio", [982] = "icbi-R0R", [1014] = "dcbz-R0R",
[26] = "cntlzwRR~", [58] = "cntlzdRR~",
[122] = "popcntbRR~",
[154] = "prtywRR~", [186] = "prtydRR~",
[28] = "andRR~R.", [60] = "andcRR~R.", [124] = "nor|notRR~R=.",
[284] = "eqvRR~R.", [316] = "xorRR~R.",
[412] = "orcRR~R.", [444] = "or|mrRR~R=.", [476] = "nandRR~R.",
[508] = "cmpbRR~R",
[512] = "mcrxrX",
[532] = "ldbrxRR0R", [660] = "stdbrxRR0R",
[533] = "lswxRR0R", [597] = "lswiRR0A",
[661] = "stswxRR0R", [725] = "stswiRR0A",
[534] = "lwbrxRR0R", [662] = "stwbrxRR0R",
[790] = "lhbrxRR0R", [918] = "sthbrxRR0R",
[535] = "lfsxFR0R", [567] = "lfsuxFRR",
[599] = "lfdxFR0R", [631] = "lfduxFRR",
[663] = "stfsxFR0R", [695] = "stfsuxFRR",
[727] = "stfdxFR0R", [759] = "stfduxFR0R",
[855] = "lfiwaxFR0R",
[983] = "stfiwxFR0R",
[24] = "slwRR~R.",
[27] = "sldRR~R.", [536] = "srwRR~R.",
[792] = "srawRR~R.", [824] = "srawiRR~A.",
[794] = "sradRR~R.", [826] = "sradiRR~H.", [827] = "sradiRR~H.",
[922] = "extshRR~.", [954] = "extsbRR~.", [986] = "extswRR~.",
[539] = "srdRR~R.",
},
{ __index = function(t, x)
if band(x, 31) == 15 then return "iselRRRC" end
end
})
local map_ld = {
shift = 0, mask = 3,
[0] = "ldRRE", "lduRRE", "lwaRRE",
}
local map_std = {
shift = 0, mask = 3,
[0] = "stdRRE", "stduRRE",
}
local map_fps = {
shift = 5, mask = 1,
{
shift = 1, mask = 15,
[0] = false, false, "fdivsFFF.", false,
"fsubsFFF.", "faddsFFF.", "fsqrtsF-F.", false,
"fresF-F.", "fmulsFF-F.", "frsqrtesF-F.", false,
"fmsubsFFFF~.", "fmaddsFFFF~.", "fnmsubsFFFF~.", "fnmaddsFFFF~.",
}
}
local map_fpd = {
shift = 5, mask = 1,
[0] = {
shift = 1, mask = 1023,
[0] = "fcmpuXFF", [32] = "fcmpoXFF", [64] = "mcrfsXX",
[38] = "mtfsb1A.", [70] = "mtfsb0A.", [134] = "mtfsfiA>>-A>",
[8] = "fcpsgnFFF.", [40] = "fnegF-F.", [72] = "fmrF-F.",
[136] = "fnabsF-F.", [264] = "fabsF-F.",
[12] = "frspF-F.",
[14] = "fctiwF-F.", [15] = "fctiwzF-F.",
[583] = "mffsF.", [711] = "mtfsfZF.",
[392] = "frinF-F.", [424] = "frizF-F.",
[456] = "fripF-F.", [488] = "frimF-F.",
[814] = "fctidF-F.", [815] = "fctidzF-F.", [846] = "fcfidF-F.",
},
{
shift = 1, mask = 15,
[0] = false, false, "fdivFFF.", false,
"fsubFFF.", "faddFFF.", "fsqrtF-F.", "fselFFFF~.",
"freF-F.", "fmulFF-F.", "frsqrteF-F.", false,
"fmsubFFFF~.", "fmaddFFFF~.", "fnmsubFFFF~.", "fnmaddFFFF~.",
}
}
local map_spe = {
shift = 0, mask = 2047,
[512] = "evaddwRRR", [514] = "evaddiwRAR~",
[516] = "evsubwRRR~", [518] = "evsubiwRAR~",
[520] = "evabsRR", [521] = "evnegRR",
[522] = "evextsbRR", [523] = "evextshRR", [524] = "evrndwRR",
[525] = "evcntlzwRR", [526] = "evcntlswRR",
[527] = "brincRRR",
[529] = "evandRRR", [530] = "evandcRRR", [534] = "evxorRRR",
[535] = "evor|evmrRRR=", [536] = "evnor|evnotRRR=",
[537] = "eveqvRRR", [539] = "evorcRRR", [542] = "evnandRRR",
[544] = "evsrwuRRR", [545] = "evsrwsRRR",
[546] = "evsrwiuRRA", [547] = "evsrwisRRA",
[548] = "evslwRRR", [550] = "evslwiRRA",
[552] = "evrlwRRR", [553] = "evsplatiRS",
[554] = "evrlwiRRA", [555] = "evsplatfiRS",
[556] = "evmergehiRRR", [557] = "evmergeloRRR",
[558] = "evmergehiloRRR", [559] = "evmergelohiRRR",
[560] = "evcmpgtuYRR", [561] = "evcmpgtsYRR",
[562] = "evcmpltuYRR", [563] = "evcmpltsYRR",
[564] = "evcmpeqYRR",
[632] = "evselRRR", [633] = "evselRRRW",
[634] = "evselRRRW", [635] = "evselRRRW",
[636] = "evselRRRW", [637] = "evselRRRW",
[638] = "evselRRRW", [639] = "evselRRRW",
[640] = "evfsaddRRR", [641] = "evfssubRRR",
[644] = "evfsabsRR", [645] = "evfsnabsRR", [646] = "evfsnegRR",
[648] = "evfsmulRRR", [649] = "evfsdivRRR",
[652] = "evfscmpgtYRR", [653] = "evfscmpltYRR", [654] = "evfscmpeqYRR",
[656] = "evfscfuiR-R", [657] = "evfscfsiR-R",
[658] = "evfscfufR-R", [659] = "evfscfsfR-R",
[660] = "evfsctuiR-R", [661] = "evfsctsiR-R",
[662] = "evfsctufR-R", [663] = "evfsctsfR-R",
[664] = "evfsctuizR-R", [666] = "evfsctsizR-R",
[668] = "evfststgtYRR", [669] = "evfststltYRR", [670] = "evfststeqYRR",
[704] = "efsaddRRR", [705] = "efssubRRR",
[708] = "efsabsRR", [709] = "efsnabsRR", [710] = "efsnegRR",
[712] = "efsmulRRR", [713] = "efsdivRRR",
[716] = "efscmpgtYRR", [717] = "efscmpltYRR", [718] = "efscmpeqYRR",
[719] = "efscfdR-R",
[720] = "efscfuiR-R", [721] = "efscfsiR-R",
[722] = "efscfufR-R", [723] = "efscfsfR-R",
[724] = "efsctuiR-R", [725] = "efsctsiR-R",
[726] = "efsctufR-R", [727] = "efsctsfR-R",
[728] = "efsctuizR-R", [730] = "efsctsizR-R",
[732] = "efststgtYRR", [733] = "efststltYRR", [734] = "efststeqYRR",
[736] = "efdaddRRR", [737] = "efdsubRRR",
[738] = "efdcfuidR-R", [739] = "efdcfsidR-R",
[740] = "efdabsRR", [741] = "efdnabsRR", [742] = "efdnegRR",
[744] = "efdmulRRR", [745] = "efddivRRR",
[746] = "efdctuidzR-R", [747] = "efdctsidzR-R",
[748] = "efdcmpgtYRR", [749] = "efdcmpltYRR", [750] = "efdcmpeqYRR",
[751] = "efdcfsR-R",
[752] = "efdcfuiR-R", [753] = "efdcfsiR-R",
[754] = "efdcfufR-R", [755] = "efdcfsfR-R",
[756] = "efdctuiR-R", [757] = "efdctsiR-R",
[758] = "efdctufR-R", [759] = "efdctsfR-R",
[760] = "efdctuizR-R", [762] = "efdctsizR-R",
[764] = "efdtstgtYRR", [765] = "efdtstltYRR", [766] = "efdtsteqYRR",
[768] = "evlddxRR0R", [769] = "evlddRR8",
[770] = "evldwxRR0R", [771] = "evldwRR8",
[772] = "evldhxRR0R", [773] = "evldhRR8",
[776] = "evlhhesplatxRR0R", [777] = "evlhhesplatRR2",
[780] = "evlhhousplatxRR0R", [781] = "evlhhousplatRR2",
[782] = "evlhhossplatxRR0R", [783] = "evlhhossplatRR2",
[784] = "evlwhexRR0R", [785] = "evlwheRR4",
[788] = "evlwhouxRR0R", [789] = "evlwhouRR4",
[790] = "evlwhosxRR0R", [791] = "evlwhosRR4",
[792] = "evlwwsplatxRR0R", [793] = "evlwwsplatRR4",
[796] = "evlwhsplatxRR0R", [797] = "evlwhsplatRR4",
[800] = "evstddxRR0R", [801] = "evstddRR8",
[802] = "evstdwxRR0R", [803] = "evstdwRR8",
[804] = "evstdhxRR0R", [805] = "evstdhRR8",
[816] = "evstwhexRR0R", [817] = "evstwheRR4",
[820] = "evstwhoxRR0R", [821] = "evstwhoRR4",
[824] = "evstwwexRR0R", [825] = "evstwweRR4",
[828] = "evstwwoxRR0R", [829] = "evstwwoRR4",
[1027] = "evmhessfRRR", [1031] = "evmhossfRRR", [1032] = "evmheumiRRR",
[1033] = "evmhesmiRRR", [1035] = "evmhesmfRRR", [1036] = "evmhoumiRRR",
[1037] = "evmhosmiRRR", [1039] = "evmhosmfRRR", [1059] = "evmhessfaRRR",
[1063] = "evmhossfaRRR", [1064] = "evmheumiaRRR", [1065] = "evmhesmiaRRR",
[1067] = "evmhesmfaRRR", [1068] = "evmhoumiaRRR", [1069] = "evmhosmiaRRR",
[1071] = "evmhosmfaRRR", [1095] = "evmwhssfRRR", [1096] = "evmwlumiRRR",
[1100] = "evmwhumiRRR", [1101] = "evmwhsmiRRR", [1103] = "evmwhsmfRRR",
[1107] = "evmwssfRRR", [1112] = "evmwumiRRR", [1113] = "evmwsmiRRR",
[1115] = "evmwsmfRRR", [1127] = "evmwhssfaRRR", [1128] = "evmwlumiaRRR",
[1132] = "evmwhumiaRRR", [1133] = "evmwhsmiaRRR", [1135] = "evmwhsmfaRRR",
[1139] = "evmwssfaRRR", [1144] = "evmwumiaRRR", [1145] = "evmwsmiaRRR",
[1147] = "evmwsmfaRRR",
[1216] = "evaddusiaawRR", [1217] = "evaddssiaawRR",
[1218] = "evsubfusiaawRR", [1219] = "evsubfssiaawRR",
[1220] = "evmraRR",
[1222] = "evdivwsRRR", [1223] = "evdivwuRRR",
[1224] = "evaddumiaawRR", [1225] = "evaddsmiaawRR",
[1226] = "evsubfumiaawRR", [1227] = "evsubfsmiaawRR",
[1280] = "evmheusiaawRRR", [1281] = "evmhessiaawRRR",
[1283] = "evmhessfaawRRR", [1284] = "evmhousiaawRRR",
[1285] = "evmhossiaawRRR", [1287] = "evmhossfaawRRR",
[1288] = "evmheumiaawRRR", [1289] = "evmhesmiaawRRR",
[1291] = "evmhesmfaawRRR", [1292] = "evmhoumiaawRRR",
[1293] = "evmhosmiaawRRR", [1295] = "evmhosmfaawRRR",
[1320] = "evmhegumiaaRRR", [1321] = "evmhegsmiaaRRR",
[1323] = "evmhegsmfaaRRR", [1324] = "evmhogumiaaRRR",
[1325] = "evmhogsmiaaRRR", [1327] = "evmhogsmfaaRRR",
[1344] = "evmwlusiaawRRR", [1345] = "evmwlssiaawRRR",
[1352] = "evmwlumiaawRRR", [1353] = "evmwlsmiaawRRR",
[1363] = "evmwssfaaRRR", [1368] = "evmwumiaaRRR",
[1369] = "evmwsmiaaRRR", [1371] = "evmwsmfaaRRR",
[1408] = "evmheusianwRRR", [1409] = "evmhessianwRRR",
[1411] = "evmhessfanwRRR", [1412] = "evmhousianwRRR",
[1413] = "evmhossianwRRR", [1415] = "evmhossfanwRRR",
[1416] = "evmheumianwRRR", [1417] = "evmhesmianwRRR",
[1419] = "evmhesmfanwRRR", [1420] = "evmhoumianwRRR",
[1421] = "evmhosmianwRRR", [1423] = "evmhosmfanwRRR",
[1448] = "evmhegumianRRR", [1449] = "evmhegsmianRRR",
[1451] = "evmhegsmfanRRR", [1452] = "evmhogumianRRR",
[1453] = "evmhogsmianRRR", [1455] = "evmhogsmfanRRR",
[1472] = "evmwlusianwRRR", [1473] = "evmwlssianwRRR",
[1480] = "evmwlumianwRRR", [1481] = "evmwlsmianwRRR",
[1491] = "evmwssfanRRR", [1496] = "evmwumianRRR",
[1497] = "evmwsmianRRR", [1499] = "evmwsmfanRRR",
}
local map_pri = {
[0] = false, false, "tdiARI", "twiARI",
map_spe, false, false, "mulliRRI",
"subficRRI", false, "cmpl_iYLRU", "cmp_iYLRI",
"addicRRI", "addic.RRI", "addi|liRR0I", "addis|lisRR0I",
"b_KBJ", "sc", "bKJ", map_crops,
"rlwimiRR~AAA.", map_rlwinm, false, "rlwnmRR~RAA.",
"oriNRR~U", "orisRR~U", "xoriRR~U", "xorisRR~U",
"andi.RR~U", "andis.RR~U", map_rld, map_ext,
"lwzRRD", "lwzuRRD", "lbzRRD", "lbzuRRD",
"stwRRD", "stwuRRD", "stbRRD", "stbuRRD",
"lhzRRD", "lhzuRRD", "lhaRRD", "lhauRRD",
"sthRRD", "sthuRRD", "lmwRRD", "stmwRRD",
"lfsFRD", "lfsuFRD", "lfdFRD", "lfduFRD",
"stfsFRD", "stfsuFRD", "stfdFRD", "stfduFRD",
false, false, map_ld, map_fps,
false, false, map_std, map_fpd,
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "sp", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
"r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
"r24", "r25", "r26", "r27", "r28", "r29", "r30", "r31",
}
local map_cond = { [0] = "lt", "gt", "eq", "so", "ge", "le", "ne", "ns", }
-- Format a condition bit.
local function condfmt(cond)
if cond <= 3 then
return map_cond[band(cond, 3)]
else
return format("4*cr%d+%s", rshift(cond, 2), map_cond[band(cond, 3)])
end
end
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then extra = "\t->"..sym end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-7s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-7s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
local op = bor(lshift(b0, 24), lshift(b1, 16), lshift(b2, 8), b3)
local operands = {}
local last = nil
local rs = 21
ctx.op = op
ctx.rel = nil
local opat = map_pri[rshift(b0, 2)]
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
opat = opat[band(rshift(op, opat.shift), opat.mask)]
end
local name, pat = match(opat, "^([a-z0-9_.]*)(.*)")
local altname, pat2 = match(pat, "|([a-z0-9_.]*)(.*)")
if altname then pat = pat2 end
for p in gmatch(pat, ".") do
local x = nil
if p == "R" then
x = map_gpr[band(rshift(op, rs), 31)]
rs = rs - 5
elseif p == "F" then
x = "f"..band(rshift(op, rs), 31)
rs = rs - 5
elseif p == "A" then
x = band(rshift(op, rs), 31)
rs = rs - 5
elseif p == "S" then
x = arshift(lshift(op, 27-rs), 27)
rs = rs - 5
elseif p == "I" then
x = arshift(lshift(op, 16), 16)
elseif p == "U" then
x = band(op, 0xffff)
elseif p == "D" or p == "E" then
local disp = arshift(lshift(op, 16), 16)
if p == "E" then disp = band(disp, -4) end
if last == "r0" then last = "0" end
operands[#operands] = format("%d(%s)", disp, last)
elseif p >= "2" and p <= "8" then
local disp = band(rshift(op, rs), 31) * p
if last == "r0" then last = "0" end
operands[#operands] = format("%d(%s)", disp, last)
elseif p == "H" then
x = band(rshift(op, rs), 31) + lshift(band(op, 2), 4)
rs = rs - 5
elseif p == "M" then
x = band(rshift(op, rs), 31) + band(op, 0x20)
elseif p == "C" then
x = condfmt(band(rshift(op, rs), 31))
rs = rs - 5
elseif p == "B" then
local bo = rshift(op, 21)
local cond = band(rshift(op, 16), 31)
local cn = ""
rs = rs - 10
if band(bo, 4) == 0 then
cn = band(bo, 2) == 0 and "dnz" or "dz"
if band(bo, 0x10) == 0 then
cn = cn..(band(bo, 8) == 0 and "f" or "t")
end
if band(bo, 0x10) == 0 then x = condfmt(cond) end
name = name..(band(bo, 1) == band(rshift(op, 15), 1) and "-" or "+")
elseif band(bo, 0x10) == 0 then
cn = map_cond[band(cond, 3) + (band(bo, 8) == 0 and 4 or 0)]
if cond > 3 then x = "cr"..rshift(cond, 2) end
name = name..(band(bo, 1) == band(rshift(op, 15), 1) and "-" or "+")
end
name = gsub(name, "_", cn)
elseif p == "J" then
x = arshift(lshift(op, 27-rs), 29-rs)*4
if band(op, 2) == 0 then x = ctx.addr + pos + x end
ctx.rel = x
x = "0x"..tohex(x)
elseif p == "K" then
if band(op, 1) ~= 0 then name = name.."l" end
if band(op, 2) ~= 0 then name = name.."a" end
elseif p == "X" or p == "Y" then
x = band(rshift(op, rs+2), 7)
if x == 0 and p == "Y" then x = nil else x = "cr"..x end
rs = rs - 5
elseif p == "W" then
x = "cr"..band(op, 7)
elseif p == "Z" then
x = band(rshift(op, rs-4), 255)
rs = rs - 10
elseif p == ">" then
operands[#operands] = rshift(operands[#operands], 1)
elseif p == "0" then
if last == "r0" then
operands[#operands] = nil
if altname then name = altname end
end
elseif p == "L" then
name = gsub(name, "_", band(op, 0x00200000) ~= 0 and "d" or "w")
elseif p == "." then
if band(op, 1) == 1 then name = name.."." end
elseif p == "N" then
if op == 0x60000000 then name = "nop"; break end
elseif p == "~" then
local n = #operands
operands[n-1], operands[n] = operands[n], operands[n-1]
elseif p == "=" then
local n = #operands
if last == operands[n-1] then
operands[n] = nil
name = altname
end
elseif p == "%" then
local n = #operands
if last == operands[n-1] and last == operands[n-2] then
operands[n] = nil
operands[n-1] = nil
name = altname
end
elseif p == "-" then
rs = rs - 5
else
assert(false)
end
if x then operands[#operands+1] = x; last = x end
end
return putop(ctx, name, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
stop = stop - stop % 4
ctx.pos = ofs - ofs % 4
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 32 then return map_gpr[r] end
return "f"..(r-32)
end
-- Public module functions.
return {
create = create,
disass = disass,
regname = regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT x64 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the 64 bit functions from the combined
-- x86/x64 disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_x86 = require((string.match(..., ".*%.") or "").."dis_x86")
return {
create = dis_x86.create64,
disass = dis_x86.disass64,
regname = dis_x86.regname64
}
@@ -0,0 +1,953 @@
----------------------------------------------------------------------------
-- LuaJIT x86/x64 disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- Sending small code snippets to an external disassembler and mixing the
-- output with our own stuff was too fragile. So I had to bite the bullet
-- and write yet another x86 disassembler. Oh well ...
--
-- The output format is very similar to what ndisasm generates. But it has
-- been developed independently by looking at the opcode tables from the
-- Intel and AMD manuals. The supported instruction set is quite extensive
-- and reflects what a current generation Intel or AMD CPU implements in
-- 32 bit and 64 bit mode. Yes, this includes MMX, SSE, SSE2, SSE3, SSSE3,
-- SSE4.1, SSE4.2, SSE4a, AVX, AVX2 and even privileged and hypervisor
-- (VMX/SVM) instructions.
--
-- Notes:
-- * The (useless) a16 prefix, 3DNow and pre-586 opcodes are unsupported.
-- * No attempt at optimization has been made -- it's fast enough for my needs.
------------------------------------------------------------------------------
local type = type
local sub, byte, format = string.sub, string.byte, string.format
local match, gmatch, gsub = string.match, string.gmatch, string.gsub
local lower, rep = string.lower, string.rep
local bit = require("bit")
local tohex = bit.tohex
-- Map for 1st opcode byte in 32 bit mode. Ugly? Well ... read on.
local map_opc1_32 = {
--0x
[0]="addBmr","addVmr","addBrm","addVrm","addBai","addVai","push es","pop es",
"orBmr","orVmr","orBrm","orVrm","orBai","orVai","push cs","opc2*",
--1x
"adcBmr","adcVmr","adcBrm","adcVrm","adcBai","adcVai","push ss","pop ss",
"sbbBmr","sbbVmr","sbbBrm","sbbVrm","sbbBai","sbbVai","push ds","pop ds",
--2x
"andBmr","andVmr","andBrm","andVrm","andBai","andVai","es:seg","daa",
"subBmr","subVmr","subBrm","subVrm","subBai","subVai","cs:seg","das",
--3x
"xorBmr","xorVmr","xorBrm","xorVrm","xorBai","xorVai","ss:seg","aaa",
"cmpBmr","cmpVmr","cmpBrm","cmpVrm","cmpBai","cmpVai","ds:seg","aas",
--4x
"incVR","incVR","incVR","incVR","incVR","incVR","incVR","incVR",
"decVR","decVR","decVR","decVR","decVR","decVR","decVR","decVR",
--5x
"pushUR","pushUR","pushUR","pushUR","pushUR","pushUR","pushUR","pushUR",
"popUR","popUR","popUR","popUR","popUR","popUR","popUR","popUR",
--6x
"sz*pushaw,pusha","sz*popaw,popa","boundVrm","arplWmr",
"fs:seg","gs:seg","o16:","a16",
"pushUi","imulVrmi","pushBs","imulVrms",
"insb","insVS","outsb","outsVS",
--7x
"joBj","jnoBj","jbBj","jnbBj","jzBj","jnzBj","jbeBj","jaBj",
"jsBj","jnsBj","jpeBj","jpoBj","jlBj","jgeBj","jleBj","jgBj",
--8x
"arith!Bmi","arith!Vmi","arith!Bmi","arith!Vms",
"testBmr","testVmr","xchgBrm","xchgVrm",
"movBmr","movVmr","movBrm","movVrm",
"movVmg","leaVrm","movWgm","popUm",
--9x
"nop*xchgVaR|pause|xchgWaR|repne nop","xchgVaR","xchgVaR","xchgVaR",
"xchgVaR","xchgVaR","xchgVaR","xchgVaR",
"sz*cbw,cwde,cdqe","sz*cwd,cdq,cqo","call farViw","wait",
"sz*pushfw,pushf","sz*popfw,popf","sahf","lahf",
--Ax
"movBao","movVao","movBoa","movVoa",
"movsb","movsVS","cmpsb","cmpsVS",
"testBai","testVai","stosb","stosVS",
"lodsb","lodsVS","scasb","scasVS",
--Bx
"movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi",
"movVRI","movVRI","movVRI","movVRI","movVRI","movVRI","movVRI","movVRI",
--Cx
"shift!Bmu","shift!Vmu","retBw","ret","vex*3$lesVrm","vex*2$ldsVrm","movBmi","movVmi",
"enterBwu","leave","retfBw","retf","int3","intBu","into","iretVS",
--Dx
"shift!Bm1","shift!Vm1","shift!Bmc","shift!Vmc","aamBu","aadBu","salc","xlatb",
"fp*0","fp*1","fp*2","fp*3","fp*4","fp*5","fp*6","fp*7",
--Ex
"loopneBj","loopeBj","loopBj","sz*jcxzBj,jecxzBj,jrcxzBj",
"inBau","inVau","outBua","outVua",
"callVj","jmpVj","jmp farViw","jmpBj","inBad","inVad","outBda","outVda",
--Fx
"lock:","int1","repne:rep","rep:","hlt","cmc","testb!Bm","testv!Vm",
"clc","stc","cli","sti","cld","std","incb!Bm","incd!Vm",
}
assert(#map_opc1_32 == 255)
-- Map for 1st opcode byte in 64 bit mode (overrides only).
local map_opc1_64 = setmetatable({
[0x06]=false, [0x07]=false, [0x0e]=false,
[0x16]=false, [0x17]=false, [0x1e]=false, [0x1f]=false,
[0x27]=false, [0x2f]=false, [0x37]=false, [0x3f]=false,
[0x60]=false, [0x61]=false, [0x62]=false, [0x63]="movsxdVrDmt", [0x67]="a32:",
[0x40]="rex*", [0x41]="rex*b", [0x42]="rex*x", [0x43]="rex*xb",
[0x44]="rex*r", [0x45]="rex*rb", [0x46]="rex*rx", [0x47]="rex*rxb",
[0x48]="rex*w", [0x49]="rex*wb", [0x4a]="rex*wx", [0x4b]="rex*wxb",
[0x4c]="rex*wr", [0x4d]="rex*wrb", [0x4e]="rex*wrx", [0x4f]="rex*wrxb",
[0x82]=false, [0x9a]=false, [0xc4]="vex*3", [0xc5]="vex*2", [0xce]=false,
[0xd4]=false, [0xd5]=false, [0xd6]=false, [0xea]=false,
}, { __index = map_opc1_32 })
-- Map for 2nd opcode byte (0F xx). True CISC hell. Hey, I told you.
-- Prefix dependent MMX/SSE opcodes: (none)|rep|o16|repne, -|F3|66|F2
local map_opc2 = {
--0x
[0]="sldt!Dmp","sgdt!Ump","larVrm","lslVrm",nil,"syscall","clts","sysret",
"invd","wbinvd",nil,"ud1",nil,"$prefetch!Bm","femms","3dnowMrmu",
--1x
"movupsXrm|movssXrvm|movupdXrm|movsdXrvm",
"movupsXmr|movssXmvr|movupdXmr|movsdXmvr",
"movhlpsXrm$movlpsXrm|movsldupXrm|movlpdXrm|movddupXrm",
"movlpsXmr||movlpdXmr",
"unpcklpsXrvm||unpcklpdXrvm",
"unpckhpsXrvm||unpckhpdXrvm",
"movlhpsXrm$movhpsXrm|movshdupXrm|movhpdXrm",
"movhpsXmr||movhpdXmr",
"$prefetcht!Bm","hintnopVm","hintnopVm","hintnopVm",
"hintnopVm","hintnopVm","hintnopVm","hintnopVm",
--2x
"movUmx$","movUmy$","movUxm$","movUym$","movUmz$",nil,"movUzm$",nil,
"movapsXrm||movapdXrm",
"movapsXmr||movapdXmr",
"cvtpi2psXrMm|cvtsi2ssXrvVmt|cvtpi2pdXrMm|cvtsi2sdXrvVmt",
"movntpsXmr|movntssXmr|movntpdXmr|movntsdXmr",
"cvttps2piMrXm|cvttss2siVrXm|cvttpd2piMrXm|cvttsd2siVrXm",
"cvtps2piMrXm|cvtss2siVrXm|cvtpd2piMrXm|cvtsd2siVrXm",
"ucomissXrm||ucomisdXrm",
"comissXrm||comisdXrm",
--3x
"wrmsr","rdtsc","rdmsr","rdpmc","sysenter","sysexit",nil,"getsec",
"opc3*38",nil,"opc3*3a",nil,nil,nil,nil,nil,
--4x
"cmovoVrm","cmovnoVrm","cmovbVrm","cmovnbVrm",
"cmovzVrm","cmovnzVrm","cmovbeVrm","cmovaVrm",
"cmovsVrm","cmovnsVrm","cmovpeVrm","cmovpoVrm",
"cmovlVrm","cmovgeVrm","cmovleVrm","cmovgVrm",
--5x
"movmskpsVrXm$||movmskpdVrXm$","sqrtpsXrm|sqrtssXrm|sqrtpdXrm|sqrtsdXrm",
"rsqrtpsXrm|rsqrtssXrvm","rcppsXrm|rcpssXrvm",
"andpsXrvm||andpdXrvm","andnpsXrvm||andnpdXrvm",
"orpsXrvm||orpdXrvm","xorpsXrvm||xorpdXrvm",
"addpsXrvm|addssXrvm|addpdXrvm|addsdXrvm","mulpsXrvm|mulssXrvm|mulpdXrvm|mulsdXrvm",
"cvtps2pdXrm|cvtss2sdXrvm|cvtpd2psXrm|cvtsd2ssXrvm",
"cvtdq2psXrm|cvttps2dqXrm|cvtps2dqXrm",
"subpsXrvm|subssXrvm|subpdXrvm|subsdXrvm","minpsXrvm|minssXrvm|minpdXrvm|minsdXrvm",
"divpsXrvm|divssXrvm|divpdXrvm|divsdXrvm","maxpsXrvm|maxssXrvm|maxpdXrvm|maxsdXrvm",
--6x
"punpcklbwPrvm","punpcklwdPrvm","punpckldqPrvm","packsswbPrvm",
"pcmpgtbPrvm","pcmpgtwPrvm","pcmpgtdPrvm","packuswbPrvm",
"punpckhbwPrvm","punpckhwdPrvm","punpckhdqPrvm","packssdwPrvm",
"||punpcklqdqXrvm","||punpckhqdqXrvm",
"movPrVSm","movqMrm|movdquXrm|movdqaXrm",
--7x
"pshufwMrmu|pshufhwXrmu|pshufdXrmu|pshuflwXrmu","pshiftw!Pvmu",
"pshiftd!Pvmu","pshiftq!Mvmu||pshiftdq!Xvmu",
"pcmpeqbPrvm","pcmpeqwPrvm","pcmpeqdPrvm","emms*|",
"vmreadUmr||extrqXmuu$|insertqXrmuu$","vmwriteUrm||extrqXrm$|insertqXrm$",
nil,nil,
"||haddpdXrvm|haddpsXrvm","||hsubpdXrvm|hsubpsXrvm",
"movVSmMr|movqXrm|movVSmXr","movqMmr|movdquXmr|movdqaXmr",
--8x
"joVj","jnoVj","jbVj","jnbVj","jzVj","jnzVj","jbeVj","jaVj",
"jsVj","jnsVj","jpeVj","jpoVj","jlVj","jgeVj","jleVj","jgVj",
--9x
"setoBm","setnoBm","setbBm","setnbBm","setzBm","setnzBm","setbeBm","setaBm",
"setsBm","setnsBm","setpeBm","setpoBm","setlBm","setgeBm","setleBm","setgBm",
--Ax
"push fs","pop fs","cpuid","btVmr","shldVmru","shldVmrc",nil,nil,
"push gs","pop gs","rsm","btsVmr","shrdVmru","shrdVmrc","fxsave!Dmp","imulVrm",
--Bx
"cmpxchgBmr","cmpxchgVmr","$lssVrm","btrVmr",
"$lfsVrm","$lgsVrm","movzxVrBmt","movzxVrWmt",
"|popcntVrm","ud2Dp","bt!Vmu","btcVmr",
"bsfVrm","bsrVrm|lzcntVrm|bsrWrm","movsxVrBmt","movsxVrWmt",
--Cx
"xaddBmr","xaddVmr",
"cmppsXrvmu|cmpssXrvmu|cmppdXrvmu|cmpsdXrvmu","$movntiVmr|",
"pinsrwPrvWmu","pextrwDrPmu",
"shufpsXrvmu||shufpdXrvmu","$cmpxchg!Qmp",
"bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR",
--Dx
"||addsubpdXrvm|addsubpsXrvm","psrlwPrvm","psrldPrvm","psrlqPrvm",
"paddqPrvm","pmullwPrvm",
"|movq2dqXrMm|movqXmr|movdq2qMrXm$","pmovmskbVrMm||pmovmskbVrXm",
"psubusbPrvm","psubuswPrvm","pminubPrvm","pandPrvm",
"paddusbPrvm","padduswPrvm","pmaxubPrvm","pandnPrvm",
--Ex
"pavgbPrvm","psrawPrvm","psradPrvm","pavgwPrvm",
"pmulhuwPrvm","pmulhwPrvm",
"|cvtdq2pdXrm|cvttpd2dqXrm|cvtpd2dqXrm","$movntqMmr||$movntdqXmr",
"psubsbPrvm","psubswPrvm","pminswPrvm","porPrvm",
"paddsbPrvm","paddswPrvm","pmaxswPrvm","pxorPrvm",
--Fx
"|||lddquXrm","psllwPrvm","pslldPrvm","psllqPrvm",
"pmuludqPrvm","pmaddwdPrvm","psadbwPrvm","maskmovqMrm||maskmovdquXrm$",
"psubbPrvm","psubwPrvm","psubdPrvm","psubqPrvm",
"paddbPrvm","paddwPrvm","padddPrvm","ud",
}
assert(map_opc2[255] == "ud")
-- Map for three-byte opcodes. Can't wait for their next invention.
local map_opc3 = {
["38"] = { -- [66] 0f 38 xx
--0x
[0]="pshufbPrvm","phaddwPrvm","phadddPrvm","phaddswPrvm",
"pmaddubswPrvm","phsubwPrvm","phsubdPrvm","phsubswPrvm",
"psignbPrvm","psignwPrvm","psigndPrvm","pmulhrswPrvm",
"||permilpsXrvm","||permilpdXrvm",nil,nil,
--1x
"||pblendvbXrma",nil,nil,nil,
"||blendvpsXrma","||blendvpdXrma","||permpsXrvm","||ptestXrm",
"||broadcastssXrm","||broadcastsdXrm","||broadcastf128XrlXm",nil,
"pabsbPrm","pabswPrm","pabsdPrm",nil,
--2x
"||pmovsxbwXrm","||pmovsxbdXrm","||pmovsxbqXrm","||pmovsxwdXrm",
"||pmovsxwqXrm","||pmovsxdqXrm",nil,nil,
"||pmuldqXrvm","||pcmpeqqXrvm","||$movntdqaXrm","||packusdwXrvm",
"||maskmovpsXrvm","||maskmovpdXrvm","||maskmovpsXmvr","||maskmovpdXmvr",
--3x
"||pmovzxbwXrm","||pmovzxbdXrm","||pmovzxbqXrm","||pmovzxwdXrm",
"||pmovzxwqXrm","||pmovzxdqXrm","||permdXrvm","||pcmpgtqXrvm",
"||pminsbXrvm","||pminsdXrvm","||pminuwXrvm","||pminudXrvm",
"||pmaxsbXrvm","||pmaxsdXrvm","||pmaxuwXrvm","||pmaxudXrvm",
--4x
"||pmulddXrvm","||phminposuwXrm",nil,nil,
nil,"||psrlvVSXrvm","||psravdXrvm","||psllvVSXrvm",
--5x
[0x58] = "||pbroadcastdXrlXm",[0x59] = "||pbroadcastqXrlXm",
[0x5a] = "||broadcasti128XrlXm",
--7x
[0x78] = "||pbroadcastbXrlXm",[0x79] = "||pbroadcastwXrlXm",
--8x
[0x8c] = "||pmaskmovXrvVSm",
[0x8e] = "||pmaskmovVSmXvr",
--9x
[0x96] = "||fmaddsub132pHXrvm",[0x97] = "||fmsubadd132pHXrvm",
[0x98] = "||fmadd132pHXrvm",[0x99] = "||fmadd132sHXrvm",
[0x9a] = "||fmsub132pHXrvm",[0x9b] = "||fmsub132sHXrvm",
[0x9c] = "||fnmadd132pHXrvm",[0x9d] = "||fnmadd132sHXrvm",
[0x9e] = "||fnmsub132pHXrvm",[0x9f] = "||fnmsub132sHXrvm",
--Ax
[0xa6] = "||fmaddsub213pHXrvm",[0xa7] = "||fmsubadd213pHXrvm",
[0xa8] = "||fmadd213pHXrvm",[0xa9] = "||fmadd213sHXrvm",
[0xaa] = "||fmsub213pHXrvm",[0xab] = "||fmsub213sHXrvm",
[0xac] = "||fnmadd213pHXrvm",[0xad] = "||fnmadd213sHXrvm",
[0xae] = "||fnmsub213pHXrvm",[0xaf] = "||fnmsub213sHXrvm",
--Bx
[0xb6] = "||fmaddsub231pHXrvm",[0xb7] = "||fmsubadd231pHXrvm",
[0xb8] = "||fmadd231pHXrvm",[0xb9] = "||fmadd231sHXrvm",
[0xba] = "||fmsub231pHXrvm",[0xbb] = "||fmsub231sHXrvm",
[0xbc] = "||fnmadd231pHXrvm",[0xbd] = "||fnmadd231sHXrvm",
[0xbe] = "||fnmsub231pHXrvm",[0xbf] = "||fnmsub231sHXrvm",
--Dx
[0xdc] = "||aesencXrvm", [0xdd] = "||aesenclastXrvm",
[0xde] = "||aesdecXrvm", [0xdf] = "||aesdeclastXrvm",
--Fx
[0xf0] = "|||crc32TrBmt",[0xf1] = "|||crc32TrVmt",
[0xf7] = "| sarxVrmv| shlxVrmv| shrxVrmv",
},
["3a"] = { -- [66] 0f 3a xx
--0x
[0x00]="||permqXrmu","||permpdXrmu","||pblenddXrvmu",nil,
"||permilpsXrmu","||permilpdXrmu","||perm2f128Xrvmu",nil,
"||roundpsXrmu","||roundpdXrmu","||roundssXrvmu","||roundsdXrvmu",
"||blendpsXrvmu","||blendpdXrvmu","||pblendwXrvmu","palignrPrvmu",
--1x
nil,nil,nil,nil,
"||pextrbVmXru","||pextrwVmXru","||pextrVmSXru","||extractpsVmXru",
"||insertf128XrvlXmu","||extractf128XlXmYru",nil,nil,
nil,nil,nil,nil,
--2x
"||pinsrbXrvVmu","||insertpsXrvmu","||pinsrXrvVmuS",nil,
--3x
[0x38] = "||inserti128Xrvmu",[0x39] = "||extracti128XlXmYru",
--4x
[0x40] = "||dppsXrvmu",
[0x41] = "||dppdXrvmu",
[0x42] = "||mpsadbwXrvmu",
[0x44] = "||pclmulqdqXrvmu",
[0x46] = "||perm2i128Xrvmu",
[0x4a] = "||blendvpsXrvmb",[0x4b] = "||blendvpdXrvmb",
[0x4c] = "||pblendvbXrvmb",
--6x
[0x60] = "||pcmpestrmXrmu",[0x61] = "||pcmpestriXrmu",
[0x62] = "||pcmpistrmXrmu",[0x63] = "||pcmpistriXrmu",
[0xdf] = "||aeskeygenassistXrmu",
--Fx
[0xf0] = "||| rorxVrmu",
},
}
-- Map for VMX/SVM opcodes 0F 01 C0-FF (sgdt group with register operands).
local map_opcvm = {
[0xc1]="vmcall",[0xc2]="vmlaunch",[0xc3]="vmresume",[0xc4]="vmxoff",
[0xc8]="monitor",[0xc9]="mwait",
[0xd8]="vmrun",[0xd9]="vmmcall",[0xda]="vmload",[0xdb]="vmsave",
[0xdc]="stgi",[0xdd]="clgi",[0xde]="skinit",[0xdf]="invlpga",
[0xf8]="swapgs",[0xf9]="rdtscp",
}
-- Map for FP opcodes. And you thought stack machines are simple?
local map_opcfp = {
-- D8-DF 00-BF: opcodes with a memory operand.
-- D8
[0]="faddFm","fmulFm","fcomFm","fcompFm","fsubFm","fsubrFm","fdivFm","fdivrFm",
"fldFm",nil,"fstFm","fstpFm","fldenvVm","fldcwWm","fnstenvVm","fnstcwWm",
-- DA
"fiaddDm","fimulDm","ficomDm","ficompDm",
"fisubDm","fisubrDm","fidivDm","fidivrDm",
-- DB
"fildDm","fisttpDm","fistDm","fistpDm",nil,"fld twordFmp",nil,"fstp twordFmp",
-- DC
"faddGm","fmulGm","fcomGm","fcompGm","fsubGm","fsubrGm","fdivGm","fdivrGm",
-- DD
"fldGm","fisttpQm","fstGm","fstpGm","frstorDmp",nil,"fnsaveDmp","fnstswWm",
-- DE
"fiaddWm","fimulWm","ficomWm","ficompWm",
"fisubWm","fisubrWm","fidivWm","fidivrWm",
-- DF
"fildWm","fisttpWm","fistWm","fistpWm",
"fbld twordFmp","fildQm","fbstp twordFmp","fistpQm",
-- xx C0-FF: opcodes with a pseudo-register operand.
-- D8
"faddFf","fmulFf","fcomFf","fcompFf","fsubFf","fsubrFf","fdivFf","fdivrFf",
-- D9
"fldFf","fxchFf",{"fnop"},nil,
{"fchs","fabs",nil,nil,"ftst","fxam"},
{"fld1","fldl2t","fldl2e","fldpi","fldlg2","fldln2","fldz"},
{"f2xm1","fyl2x","fptan","fpatan","fxtract","fprem1","fdecstp","fincstp"},
{"fprem","fyl2xp1","fsqrt","fsincos","frndint","fscale","fsin","fcos"},
-- DA
"fcmovbFf","fcmoveFf","fcmovbeFf","fcmovuFf",nil,{nil,"fucompp"},nil,nil,
-- DB
"fcmovnbFf","fcmovneFf","fcmovnbeFf","fcmovnuFf",
{nil,nil,"fnclex","fninit"},"fucomiFf","fcomiFf",nil,
-- DC
"fadd toFf","fmul toFf",nil,nil,
"fsub toFf","fsubr toFf","fdivr toFf","fdiv toFf",
-- DD
"ffreeFf",nil,"fstFf","fstpFf","fucomFf","fucompFf",nil,nil,
-- DE
"faddpFf","fmulpFf",nil,{nil,"fcompp"},
"fsubrpFf","fsubpFf","fdivrpFf","fdivpFf",
-- DF
nil,nil,nil,nil,{"fnstsw ax"},"fucomipFf","fcomipFf",nil,
}
assert(map_opcfp[126] == "fcomipFf")
-- Map for opcode groups. The subkey is sp from the ModRM byte.
local map_opcgroup = {
arith = { "add", "or", "adc", "sbb", "and", "sub", "xor", "cmp" },
shift = { "rol", "ror", "rcl", "rcr", "shl", "shr", "sal", "sar" },
testb = { "testBmi", "testBmi", "not", "neg", "mul", "imul", "div", "idiv" },
testv = { "testVmi", "testVmi", "not", "neg", "mul", "imul", "div", "idiv" },
incb = { "inc", "dec" },
incd = { "inc", "dec", "callUmp", "$call farDmp",
"jmpUmp", "$jmp farDmp", "pushUm" },
sldt = { "sldt", "str", "lldt", "ltr", "verr", "verw" },
sgdt = { "vm*$sgdt", "vm*$sidt", "$lgdt", "vm*$lidt",
"smsw", nil, "lmsw", "vm*$invlpg" },
bt = { nil, nil, nil, nil, "bt", "bts", "btr", "btc" },
cmpxchg = { nil, "sz*,cmpxchg8bQmp,cmpxchg16bXmp", nil, nil,
nil, nil, "vmptrld|vmxon|vmclear", "vmptrst" },
pshiftw = { nil, nil, "psrlw", nil, "psraw", nil, "psllw" },
pshiftd = { nil, nil, "psrld", nil, "psrad", nil, "pslld" },
pshiftq = { nil, nil, "psrlq", nil, nil, nil, "psllq" },
pshiftdq = { nil, nil, "psrlq", "psrldq", nil, nil, "psllq", "pslldq" },
fxsave = { "$fxsave", "$fxrstor", "$ldmxcsr", "$stmxcsr",
nil, "lfenceDp$", "mfenceDp$", "sfenceDp$clflush" },
prefetch = { "prefetch", "prefetchw" },
prefetcht = { "prefetchnta", "prefetcht0", "prefetcht1", "prefetcht2" },
}
------------------------------------------------------------------------------
-- Maps for register names.
local map_regs = {
B = { "al", "cl", "dl", "bl", "ah", "ch", "dh", "bh",
"r8b", "r9b", "r10b", "r11b", "r12b", "r13b", "r14b", "r15b" },
B64 = { "al", "cl", "dl", "bl", "spl", "bpl", "sil", "dil",
"r8b", "r9b", "r10b", "r11b", "r12b", "r13b", "r14b", "r15b" },
W = { "ax", "cx", "dx", "bx", "sp", "bp", "si", "di",
"r8w", "r9w", "r10w", "r11w", "r12w", "r13w", "r14w", "r15w" },
D = { "eax", "ecx", "edx", "ebx", "esp", "ebp", "esi", "edi",
"r8d", "r9d", "r10d", "r11d", "r12d", "r13d", "r14d", "r15d" },
Q = { "rax", "rcx", "rdx", "rbx", "rsp", "rbp", "rsi", "rdi",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15" },
M = { "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
"mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7" }, -- No x64 ext!
X = { "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7",
"xmm8", "xmm9", "xmm10", "xmm11", "xmm12", "xmm13", "xmm14", "xmm15" },
Y = { "ymm0", "ymm1", "ymm2", "ymm3", "ymm4", "ymm5", "ymm6", "ymm7",
"ymm8", "ymm9", "ymm10", "ymm11", "ymm12", "ymm13", "ymm14", "ymm15" },
}
local map_segregs = { "es", "cs", "ss", "ds", "fs", "gs", "segr6", "segr7" }
-- Maps for size names.
local map_sz2n = {
B = 1, W = 2, D = 4, Q = 8, M = 8, X = 16, Y = 32,
}
local map_sz2prefix = {
B = "byte", W = "word", D = "dword",
Q = "qword",
M = "qword", X = "xword", Y = "yword",
F = "dword", G = "qword", -- No need for sizes/register names for these two.
}
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local code, pos, hex = ctx.code, ctx.pos, ""
local hmax = ctx.hexdump
if hmax > 0 then
for i=ctx.start,pos-1 do
hex = hex..format("%02X", byte(code, i, i))
end
if #hex > hmax then hex = sub(hex, 1, hmax)..". "
else hex = hex..rep(" ", hmax-#hex+2) end
end
if operands then text = text.." "..operands end
if ctx.o16 then text = "o16 "..text; ctx.o16 = false end
if ctx.a32 then text = "a32 "..text; ctx.a32 = false end
if ctx.rep then text = ctx.rep.." "..text; ctx.rep = false end
if ctx.rex then
local t = (ctx.rexw and "w" or "")..(ctx.rexr and "r" or "")..
(ctx.rexx and "x" or "")..(ctx.rexb and "b" or "")..
(ctx.vexl and "l" or "")
if ctx.vexv and ctx.vexv ~= 0 then t = t.."v"..ctx.vexv end
if t ~= "" then text = ctx.rex.."."..t.." "..gsub(text, "^ ", "")
elseif ctx.rex == "vex" then text = gsub("v"..text, "^v ", "") end
ctx.rexw = false; ctx.rexr = false; ctx.rexx = false; ctx.rexb = false
ctx.rex = false; ctx.vexl = false; ctx.vexv = false
end
if ctx.seg then
local text2, n = gsub(text, "%[", "["..ctx.seg..":")
if n == 0 then text = ctx.seg.." "..text else text = text2 end
ctx.seg = false
end
if ctx.lock then text = "lock "..text; ctx.lock = false end
local imm = ctx.imm
if imm then
local sym = ctx.symtab[imm]
if sym then text = text.."\t->"..sym end
end
ctx.out(format("%08x %s%s\n", ctx.addr+ctx.start, hex, text))
ctx.mrm = false
ctx.vexv = false
ctx.start = pos
ctx.imm = nil
end
-- Clear all prefix flags.
local function clearprefixes(ctx)
ctx.o16 = false; ctx.seg = false; ctx.lock = false; ctx.rep = false
ctx.rexw = false; ctx.rexr = false; ctx.rexx = false; ctx.rexb = false
ctx.rex = false; ctx.a32 = false; ctx.vexl = false
end
-- Fallback for incomplete opcodes at the end.
local function incomplete(ctx)
ctx.pos = ctx.stop+1
clearprefixes(ctx)
return putop(ctx, "(incomplete)")
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
clearprefixes(ctx)
return putop(ctx, "(unknown)")
end
-- Return an immediate of the specified size.
local function getimm(ctx, pos, n)
if pos+n-1 > ctx.stop then return incomplete(ctx) end
local code = ctx.code
if n == 1 then
local b1 = byte(code, pos, pos)
return b1
elseif n == 2 then
local b1, b2 = byte(code, pos, pos+1)
return b1+b2*256
else
local b1, b2, b3, b4 = byte(code, pos, pos+3)
local imm = b1+b2*256+b3*65536+b4*16777216
ctx.imm = imm
return imm
end
end
-- Process pattern string and generate the operands.
local function putpat(ctx, name, pat)
local operands, regs, sz, mode, sp, rm, sc, rx, sdisp
local code, pos, stop, vexl = ctx.code, ctx.pos, ctx.stop, ctx.vexl
-- Chars used: 1DFGHIMPQRSTUVWXYabcdfgijlmoprstuvwxyz
for p in gmatch(pat, ".") do
local x = nil
if p == "V" or p == "U" then
if ctx.rexw then sz = "Q"; ctx.rexw = false
elseif ctx.o16 then sz = "W"; ctx.o16 = false
elseif p == "U" and ctx.x64 then sz = "Q"
else sz = "D" end
regs = map_regs[sz]
elseif p == "T" then
if ctx.rexw then sz = "Q"; ctx.rexw = false else sz = "D" end
regs = map_regs[sz]
elseif p == "B" then
sz = "B"
regs = ctx.rex and map_regs.B64 or map_regs.B
elseif match(p, "[WDQMXYFG]") then
sz = p
if sz == "X" and vexl then sz = "Y"; ctx.vexl = false end
regs = map_regs[sz]
elseif p == "P" then
sz = ctx.o16 and "X" or "M"; ctx.o16 = false
if sz == "X" and vexl then sz = "Y"; ctx.vexl = false end
regs = map_regs[sz]
elseif p == "H" then
name = name..(ctx.rexw and "d" or "s")
ctx.rexw = false
elseif p == "S" then
name = name..lower(sz)
elseif p == "s" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = imm <= 127 and format("+0x%02x", imm)
or format("-0x%02x", 256-imm)
pos = pos+1
elseif p == "u" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = format("0x%02x", imm)
pos = pos+1
elseif p == "b" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = regs[imm/16+1]
pos = pos+1
elseif p == "w" then
local imm = getimm(ctx, pos, 2); if not imm then return end
x = format("0x%x", imm)
pos = pos+2
elseif p == "o" then -- [offset]
if ctx.x64 then
local imm1 = getimm(ctx, pos, 4); if not imm1 then return end
local imm2 = getimm(ctx, pos+4, 4); if not imm2 then return end
x = format("[0x%08x%08x]", imm2, imm1)
pos = pos+8
else
local imm = getimm(ctx, pos, 4); if not imm then return end
x = format("[0x%08x]", imm)
pos = pos+4
end
elseif p == "i" or p == "I" then
local n = map_sz2n[sz]
if n == 8 and ctx.x64 and p == "I" then
local imm1 = getimm(ctx, pos, 4); if not imm1 then return end
local imm2 = getimm(ctx, pos+4, 4); if not imm2 then return end
x = format("0x%08x%08x", imm2, imm1)
else
if n == 8 then n = 4 end
local imm = getimm(ctx, pos, n); if not imm then return end
if sz == "Q" and (imm < 0 or imm > 0x7fffffff) then
imm = (0xffffffff+1)-imm
x = format(imm > 65535 and "-0x%08x" or "-0x%x", imm)
else
x = format(imm > 65535 and "0x%08x" or "0x%x", imm)
end
end
pos = pos+n
elseif p == "j" then
local n = map_sz2n[sz]
if n == 8 then n = 4 end
local imm = getimm(ctx, pos, n); if not imm then return end
if sz == "B" and imm > 127 then imm = imm-256
elseif imm > 2147483647 then imm = imm-4294967296 end
pos = pos+n
imm = imm + pos + ctx.addr
if imm > 4294967295 and not ctx.x64 then imm = imm-4294967296 end
ctx.imm = imm
if sz == "W" then
x = format("word 0x%04x", imm%65536)
elseif ctx.x64 then
local lo = imm % 0x1000000
x = format("0x%02x%06x", (imm-lo) / 0x1000000, lo)
else
x = "0x"..tohex(imm)
end
elseif p == "R" then
local r = byte(code, pos-1, pos-1)%8
if ctx.rexb then r = r + 8; ctx.rexb = false end
x = regs[r+1]
elseif p == "a" then x = regs[1]
elseif p == "c" then x = "cl"
elseif p == "d" then x = "dx"
elseif p == "1" then x = "1"
else
if not mode then
mode = ctx.mrm
if not mode then
if pos > stop then return incomplete(ctx) end
mode = byte(code, pos, pos)
pos = pos+1
end
rm = mode%8; mode = (mode-rm)/8
sp = mode%8; mode = (mode-sp)/8
sdisp = ""
if mode < 3 then
if rm == 4 then
if pos > stop then return incomplete(ctx) end
sc = byte(code, pos, pos)
pos = pos+1
rm = sc%8; sc = (sc-rm)/8
rx = sc%8; sc = (sc-rx)/8
if ctx.rexx then rx = rx + 8; ctx.rexx = false end
if rx == 4 then rx = nil end
end
if mode > 0 or rm == 5 then
local dsz = mode
if dsz ~= 1 then dsz = 4 end
local disp = getimm(ctx, pos, dsz); if not disp then return end
if mode == 0 then rm = nil end
if rm or rx or (not sc and ctx.x64 and not ctx.a32) then
if dsz == 1 and disp > 127 then
sdisp = format("-0x%x", 256-disp)
elseif disp >= 0 and disp <= 0x7fffffff then
sdisp = format("+0x%x", disp)
else
sdisp = format("-0x%x", (0xffffffff+1)-disp)
end
else
sdisp = format(ctx.x64 and not ctx.a32 and
not (disp >= 0 and disp <= 0x7fffffff)
and "0xffffffff%08x" or "0x%08x", disp)
end
pos = pos+dsz
end
end
if rm and ctx.rexb then rm = rm + 8; ctx.rexb = false end
if ctx.rexr then sp = sp + 8; ctx.rexr = false end
end
if p == "m" then
if mode == 3 then x = regs[rm+1]
else
local aregs = ctx.a32 and map_regs.D or ctx.aregs
local srm, srx = "", ""
if rm then srm = aregs[rm+1]
elseif not sc and ctx.x64 and not ctx.a32 then srm = "rip" end
ctx.a32 = false
if rx then
if rm then srm = srm.."+" end
srx = aregs[rx+1]
if sc > 0 then srx = srx.."*"..(2^sc) end
end
x = format("[%s%s%s]", srm, srx, sdisp)
end
if mode < 3 and
(not match(pat, "[aRrgp]") or match(pat, "t")) then -- Yuck.
x = map_sz2prefix[sz].." "..x
end
elseif p == "r" then x = regs[sp+1]
elseif p == "g" then x = map_segregs[sp+1]
elseif p == "p" then -- Suppress prefix.
elseif p == "f" then x = "st"..rm
elseif p == "x" then
if sp == 0 and ctx.lock and not ctx.x64 then
x = "CR8"; ctx.lock = false
else
x = "CR"..sp
end
elseif p == "v" then
if ctx.vexv then
x = regs[ctx.vexv+1]; ctx.vexv = false
end
elseif p == "y" then x = "DR"..sp
elseif p == "z" then x = "TR"..sp
elseif p == "l" then vexl = false
elseif p == "t" then
else
error("bad pattern `"..pat.."'")
end
end
if x then operands = operands and operands..", "..x or x end
end
ctx.pos = pos
return putop(ctx, name, operands)
end
-- Forward declaration.
local map_act
-- Fetch and cache MRM byte.
local function getmrm(ctx)
local mrm = ctx.mrm
if not mrm then
local pos = ctx.pos
if pos > ctx.stop then return nil end
mrm = byte(ctx.code, pos, pos)
ctx.pos = pos+1
ctx.mrm = mrm
end
return mrm
end
-- Dispatch to handler depending on pattern.
local function dispatch(ctx, opat, patgrp)
if not opat then return unknown(ctx) end
if match(opat, "%|") then -- MMX/SSE variants depending on prefix.
local p
if ctx.rep then
p = ctx.rep=="rep" and "%|([^%|]*)" or "%|[^%|]*%|[^%|]*%|([^%|]*)"
ctx.rep = false
elseif ctx.o16 then p = "%|[^%|]*%|([^%|]*)"; ctx.o16 = false
else p = "^[^%|]*" end
opat = match(opat, p)
if not opat then return unknown(ctx) end
-- ctx.rep = false; ctx.o16 = false
--XXX fails for 66 f2 0f 38 f1 06 crc32 eax,WORD PTR [esi]
--XXX remove in branches?
end
if match(opat, "%$") then -- reg$mem variants.
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
opat = match(opat, mrm >= 192 and "^[^%$]*" or "%$(.*)")
if opat == "" then return unknown(ctx) end
end
if opat == "" then return unknown(ctx) end
local name, pat = match(opat, "^([a-z0-9 ]*)(.*)")
if pat == "" and patgrp then pat = patgrp end
return map_act[sub(pat, 1, 1)](ctx, name, pat)
end
-- Get a pattern from an opcode map and dispatch to handler.
local function dispatchmap(ctx, opcmap)
local pos = ctx.pos
local opat = opcmap[byte(ctx.code, pos, pos)]
pos = pos + 1
ctx.pos = pos
return dispatch(ctx, opat)
end
-- Map for action codes. The key is the first char after the name.
map_act = {
-- Simple opcodes without operands.
[""] = function(ctx, name, pat)
return putop(ctx, name)
end,
-- Operand size chars fall right through.
B = putpat, W = putpat, D = putpat, Q = putpat,
V = putpat, U = putpat, T = putpat,
M = putpat, X = putpat, P = putpat,
F = putpat, G = putpat, Y = putpat,
H = putpat,
-- Collect prefixes.
[":"] = function(ctx, name, pat)
ctx[pat == ":" and name or sub(pat, 2)] = name
if ctx.pos - ctx.start > 5 then return unknown(ctx) end -- Limit #prefixes.
end,
-- Chain to special handler specified by name.
["*"] = function(ctx, name, pat)
return map_act[name](ctx, name, sub(pat, 2))
end,
-- Use named subtable for opcode group.
["!"] = function(ctx, name, pat)
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
return dispatch(ctx, map_opcgroup[name][((mrm-(mrm%8))/8)%8+1], sub(pat, 2))
end,
-- o16,o32[,o64] variants.
sz = function(ctx, name, pat)
if ctx.o16 then ctx.o16 = false
else
pat = match(pat, ",(.*)")
if ctx.rexw then
local p = match(pat, ",(.*)")
if p then pat = p; ctx.rexw = false end
end
end
pat = match(pat, "^[^,]*")
return dispatch(ctx, pat)
end,
-- Two-byte opcode dispatch.
opc2 = function(ctx, name, pat)
return dispatchmap(ctx, map_opc2)
end,
-- Three-byte opcode dispatch.
opc3 = function(ctx, name, pat)
return dispatchmap(ctx, map_opc3[pat])
end,
-- VMX/SVM dispatch.
vm = function(ctx, name, pat)
return dispatch(ctx, map_opcvm[ctx.mrm])
end,
-- Floating point opcode dispatch.
fp = function(ctx, name, pat)
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
local rm = mrm%8
local idx = pat*8 + ((mrm-rm)/8)%8
if mrm >= 192 then idx = idx + 64 end
local opat = map_opcfp[idx]
if type(opat) == "table" then opat = opat[rm+1] end
return dispatch(ctx, opat)
end,
-- REX prefix.
rex = function(ctx, name, pat)
if ctx.rex then return unknown(ctx) end -- Only 1 REX or VEX prefix allowed.
for p in gmatch(pat, ".") do ctx["rex"..p] = true end
ctx.rex = "rex"
end,
-- VEX prefix.
vex = function(ctx, name, pat)
if ctx.rex then return unknown(ctx) end -- Only 1 REX or VEX prefix allowed.
ctx.rex = "vex"
local pos = ctx.pos
if ctx.mrm then
ctx.mrm = nil
pos = pos-1
end
local b = byte(ctx.code, pos, pos)
if not b then return incomplete(ctx) end
pos = pos+1
if b < 128 then ctx.rexr = true end
local m = 1
if pat == "3" then
m = b%32; b = (b-m)/32
local nb = b%2; b = (b-nb)/2
if nb == 0 then ctx.rexb = true end
local nx = b%2
if nx == 0 then ctx.rexx = true end
b = byte(ctx.code, pos, pos)
if not b then return incomplete(ctx) end
pos = pos+1
if b >= 128 then ctx.rexw = true end
end
ctx.pos = pos
local map
if m == 1 then map = map_opc2
elseif m == 2 then map = map_opc3["38"]
elseif m == 3 then map = map_opc3["3a"]
else return unknown(ctx) end
local p = b%4; b = (b-p)/4
if p == 1 then ctx.o16 = "o16"
elseif p == 2 then ctx.rep = "rep"
elseif p == 3 then ctx.rep = "repne" end
local l = b%2; b = (b-l)/2
if l ~= 0 then ctx.vexl = true end
ctx.vexv = (-1-b)%16
return dispatchmap(ctx, map)
end,
-- Special case for nop with REX prefix.
nop = function(ctx, name, pat)
return dispatch(ctx, ctx.rex and pat or "nop")
end,
-- Special case for 0F 77.
emms = function(ctx, name, pat)
if ctx.rex ~= "vex" then
return putop(ctx, "emms")
elseif ctx.vexl then
ctx.vexl = false
return putop(ctx, "zeroall")
else
return putop(ctx, "zeroupper")
end
end,
}
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
ofs = ofs + 1
ctx.start = ofs
ctx.pos = ofs
ctx.stop = stop
ctx.imm = nil
ctx.mrm = false
clearprefixes(ctx)
while ctx.pos <= stop do dispatchmap(ctx, ctx.map1) end
if ctx.pos ~= ctx.start then incomplete(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = (addr or 0) - 1
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 16
ctx.x64 = false
ctx.map1 = map_opc1_32
ctx.aregs = map_regs.D
return ctx
end
local function create64(code, addr, out)
local ctx = create(code, addr, out)
ctx.x64 = true
ctx.map1 = map_opc1_64
ctx.aregs = map_regs.Q
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
local function disass64(code, addr, out)
create64(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 8 then return map_regs.D[r+1] end
return map_regs.X[r-7]
end
local function regname64(r)
if r < 16 then return map_regs.Q[r+1] end
return map_regs.X[r-15]
end
-- Public module functions.
return {
create = create,
create64 = create64,
disass = disass,
disass64 = disass64,
regname = regname,
regname64 = regname64
}
@@ -0,0 +1,726 @@
----------------------------------------------------------------------------
-- LuaJIT compiler dump module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module can be used to debug the JIT compiler itself. It dumps the
-- code representations and structures used in various compiler stages.
--
-- Example usage:
--
-- luajit -jdump -e "local x=0; for i=1,1e6 do x=x+i end; print(x)"
-- luajit -jdump=im -e "for i=1,1000 do for j=1,1000 do end end" | less -R
-- luajit -jdump=is myapp.lua | less -R
-- luajit -jdump=-b myapp.lua
-- luajit -jdump=+aH,myapp.html myapp.lua
-- luajit -jdump=ixT,myapp.dump myapp.lua
--
-- The first argument specifies the dump mode. The second argument gives
-- the output file name. Default output is to stdout, unless the environment
-- variable LUAJIT_DUMPFILE is set. The file is overwritten every time the
-- module is started.
--
-- Different features can be turned on or off with the dump mode. If the
-- mode starts with a '+', the following features are added to the default
-- set of features; a '-' removes them. Otherwise the features are replaced.
--
-- The following dump features are available (* marks the default):
--
-- * t Print a line for each started, ended or aborted trace (see also -jv).
-- * b Dump the traced bytecode.
-- * i Dump the IR (intermediate representation).
-- r Augment the IR with register/stack slots.
-- s Dump the snapshot map.
-- * m Dump the generated machine code.
-- x Print each taken trace exit.
-- X Print each taken trace exit and the contents of all registers.
-- a Print the IR of aborted traces, too.
--
-- The output format can be set with the following characters:
--
-- T Plain text output.
-- A ANSI-colored text output
-- H Colorized HTML + CSS output.
--
-- The default output format is plain text. It's set to ANSI-colored text
-- if the COLORTERM variable is set. Note: this is independent of any output
-- redirection, which is actually considered a feature.
--
-- You probably want to use less -R to enjoy viewing ANSI-colored text from
-- a pipe or a file. Add this to your ~/.bashrc: export LESS="-R"
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local funcinfo, funcbc = jutil.funcinfo, jutil.funcbc
local traceinfo, traceir, tracek = jutil.traceinfo, jutil.traceir, jutil.tracek
local tracemc, tracesnap = jutil.tracemc, jutil.tracesnap
local traceexitstub, ircalladdr = jutil.traceexitstub, jutil.ircalladdr
local bit = require("bit")
local band, shr, tohex = bit.band, bit.rshift, bit.tohex
local sub, gsub, format = string.sub, string.gsub, string.format
local byte, rep = string.byte, string.rep
local type, tostring = type, tostring
local stdout, stderr = io.stdout, io.stderr
-- Load other modules on-demand.
local bcline, disass
-- Active flag, output file handle and dump mode.
local active, out, dumpmode
------------------------------------------------------------------------------
local symtabmt = { __index = false }
local symtab = {}
local nexitsym = 0
-- Fill nested symbol table with per-trace exit stub addresses.
local function fillsymtab_tr(tr, nexit)
local t = {}
symtabmt.__index = t
if jit.arch:sub(1, 4) == "mips" then
t[traceexitstub(tr, 0)] = "exit"
return
end
for i=0,nexit-1 do
local addr = traceexitstub(tr, i)
if addr < 0 then addr = addr + 2^32 end
t[addr] = tostring(i)
end
local addr = traceexitstub(tr, nexit)
if addr then t[addr] = "stack_check" end
end
-- Fill symbol table with trace exit stub addresses.
local function fillsymtab(tr, nexit)
local t = symtab
if nexitsym == 0 then
local maskaddr = jit.arch == "arm" and -2
local ircall = vmdef.ircall
for i=0,#ircall do
local addr = ircalladdr(i)
if addr ~= 0 then
if maskaddr then addr = band(addr, maskaddr) end
if addr < 0 then addr = addr + 2^32 end
t[addr] = ircall[i]
end
end
end
if nexitsym == 1000000 then -- Per-trace exit stubs.
fillsymtab_tr(tr, nexit)
elseif nexit > nexitsym then -- Shared exit stubs.
for i=nexitsym,nexit-1 do
local addr = traceexitstub(i)
if addr == nil then -- Fall back to per-trace exit stubs.
fillsymtab_tr(tr, nexit)
setmetatable(symtab, symtabmt)
nexit = 1000000
break
end
if addr < 0 then addr = addr + 2^32 end
t[addr] = tostring(i)
end
nexitsym = nexit
end
return t
end
local function dumpwrite(s)
out:write(s)
end
-- Disassemble machine code.
local function dump_mcode(tr)
local info = traceinfo(tr)
if not info then return end
local mcode, addr, loop = tracemc(tr)
if not mcode then return end
if not disass then disass = require("jit.dis_"..jit.arch) end
if addr < 0 then addr = addr + 2^32 end
out:write("---- TRACE ", tr, " mcode ", #mcode, "\n")
local ctx = disass.create(mcode, addr, dumpwrite)
ctx.hexdump = 0
ctx.symtab = fillsymtab(tr, info.nexit)
if loop ~= 0 then
symtab[addr+loop] = "LOOP"
ctx:disass(0, loop)
out:write("->LOOP:\n")
ctx:disass(loop, #mcode-loop)
symtab[addr+loop] = nil
else
ctx:disass(0, #mcode)
end
end
------------------------------------------------------------------------------
local irtype_text = {
[0] = "nil",
"fal",
"tru",
"lud",
"str",
"p32",
"thr",
"pro",
"fun",
"p64",
"cdt",
"tab",
"udt",
"flt",
"num",
"i8 ",
"u8 ",
"i16",
"u16",
"int",
"u32",
"i64",
"u64",
"sfp",
}
local colortype_ansi = {
[0] = "%s",
"%s",
"%s",
"\027[36m%s\027[m",
"\027[32m%s\027[m",
"%s",
"\027[1m%s\027[m",
"%s",
"\027[1m%s\027[m",
"%s",
"\027[33m%s\027[m",
"\027[31m%s\027[m",
"\027[36m%s\027[m",
"\027[34m%s\027[m",
"\027[34m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
}
local function colorize_text(s)
return s
end
local function colorize_ansi(s, t, extra)
local out = format(colortype_ansi[t], s)
if extra then out = "\027[3m"..out end
return out
end
local irtype_ansi = setmetatable({},
{ __index = function(tab, t)
local s = colorize_ansi(irtype_text[t], t); tab[t] = s; return s; end })
local html_escape = { ["<"] = "&lt;", [">"] = "&gt;", ["&"] = "&amp;", }
local function colorize_html(s, t, extra)
s = gsub(s, "[<>&]", html_escape)
return format('<span class="irt_%s%s">%s</span>',
irtype_text[t], extra and " irt_extra" or "", s)
end
local irtype_html = setmetatable({},
{ __index = function(tab, t)
local s = colorize_html(irtype_text[t], t); tab[t] = s; return s; end })
local header_html = [[
<style type="text/css">
background { background: #ffffff; color: #000000; }
pre.ljdump {
font-size: 10pt;
background: #f0f4ff;
color: #000000;
border: 1px solid #bfcfff;
padding: 0.5em;
margin-left: 2em;
margin-right: 2em;
}
span.irt_str { color: #00a000; }
span.irt_thr, span.irt_fun { color: #404040; font-weight: bold; }
span.irt_tab { color: #c00000; }
span.irt_udt, span.irt_lud { color: #00c0c0; }
span.irt_num { color: #4040c0; }
span.irt_int, span.irt_i8, span.irt_u8, span.irt_i16, span.irt_u16 { color: #b040b0; }
span.irt_extra { font-style: italic; }
</style>
]]
local colorize, irtype
-- Lookup tables to convert some literals into names.
local litname = {
["SLOAD "] = setmetatable({}, { __index = function(t, mode)
local s = ""
if band(mode, 1) ~= 0 then s = s.."P" end
if band(mode, 2) ~= 0 then s = s.."F" end
if band(mode, 4) ~= 0 then s = s.."T" end
if band(mode, 8) ~= 0 then s = s.."C" end
if band(mode, 16) ~= 0 then s = s.."R" end
if band(mode, 32) ~= 0 then s = s.."I" end
if band(mode, 64) ~= 0 then s = s.."K" end
t[mode] = s
return s
end}),
["XLOAD "] = { [0] = "", "R", "V", "RV", "U", "RU", "VU", "RVU", },
["CONV "] = setmetatable({}, { __index = function(t, mode)
local s = irtype[band(mode, 31)]
s = irtype[band(shr(mode, 5), 31)].."."..s
if band(mode, 0x800) ~= 0 then s = s.." sext" end
local c = shr(mode, 12)
if c == 1 then s = s.." none"
elseif c == 2 then s = s.." index"
elseif c == 3 then s = s.." check" end
t[mode] = s
return s
end}),
["FLOAD "] = vmdef.irfield,
["FREF "] = vmdef.irfield,
["FPMATH"] = vmdef.irfpm,
["TMPREF"] = { [0] = "", "IN", "OUT", "INOUT", "", "", "OUT2", "INOUT2" },
["BUFHDR"] = { [0] = "RESET", "APPEND", "WRITE" },
["TOSTR "] = { [0] = "INT", "NUM", "CHAR" },
}
local function ctlsub(c)
if c == "\n" then return "\\n"
elseif c == "\r" then return "\\r"
elseif c == "\t" then return "\\t"
else return format("\\%03d", byte(c))
end
end
local function fmtfunc(func, pc)
local fi = funcinfo(func, pc)
if fi.loc then
return fi.loc
elseif fi.ffid then
return vmdef.ffnames[fi.ffid]
elseif fi.addr then
return format("C:%x", fi.addr)
else
return "(?)"
end
end
local function formatk(tr, idx, sn)
local k, t, slot = tracek(tr, idx)
local tn = type(k)
local s
if tn == "number" then
if t < 12 then
s = k == 0 and "NULL" or format("[0x%08x]", k)
elseif band(sn or 0, 0x30000) ~= 0 then
s = band(sn, 0x20000) ~= 0 and "contpc" or "ftsz"
elseif k == 2^52+2^51 then
s = "bias"
else
s = format(0 < k and k < 0x1p-1026 and "%+a" or "%+.14g", k)
end
elseif tn == "string" then
s = format(#k > 20 and '"%.20s"~' or '"%s"', gsub(k, "%c", ctlsub))
elseif tn == "function" then
s = fmtfunc(k)
elseif tn == "table" then
s = format("{%p}", k)
elseif tn == "userdata" then
if t == 12 then
s = format("userdata:%p", k)
else
s = format("[%p]", k)
if s == "[NULL]" then s = "NULL" end
end
elseif t == 21 then -- int64_t
s = sub(tostring(k), 1, -3)
if sub(s, 1, 1) ~= "-" then s = "+"..s end
elseif sn == 0x1057fff then -- SNAP(1, SNAP_FRAME | SNAP_NORESTORE, REF_NIL)
return "----" -- Special case for LJ_FR2 slot 1.
else
s = tostring(k) -- For primitives.
end
s = colorize(format("%-4s", s), t, band(sn or 0, 0x100000) ~= 0)
if slot then
s = format("%s @%d", s, slot)
end
return s
end
local function printsnap(tr, snap)
local n = 2
for s=0,snap[1]-1 do
local sn = snap[n]
if shr(sn, 24) == s then
n = n + 1
local ref = band(sn, 0xffff) - 0x8000 -- REF_BIAS
if ref < 0 then
out:write(formatk(tr, ref, sn))
elseif band(sn, 0x80000) ~= 0 then -- SNAP_SOFTFPNUM
out:write(colorize(format("%04d/%04d", ref, ref+1), 14))
else
local m, ot, op1, op2 = traceir(tr, ref)
out:write(colorize(format("%04d", ref), band(ot, 31), band(sn, 0x100000) ~= 0))
end
out:write(band(sn, 0x10000) == 0 and " " or "|") -- SNAP_FRAME
else
out:write("---- ")
end
end
out:write("]\n")
end
-- Dump snapshots (not interleaved with IR).
local function dump_snap(tr)
out:write("---- TRACE ", tr, " snapshots\n")
for i=0,1000000000 do
local snap = tracesnap(tr, i)
if not snap then break end
out:write(format("#%-3d %04d [ ", i, snap[0]))
printsnap(tr, snap)
end
end
-- Return a register name or stack slot for a rid/sp location.
local function ridsp_name(ridsp, ins)
if not disass then disass = require("jit.dis_"..jit.arch) end
local rid, slot = band(ridsp, 0xff), shr(ridsp, 8)
if rid == 253 or rid == 254 then
return (slot == 0 or slot == 255) and " {sink" or format(" {%04d", ins-slot)
end
if ridsp > 255 then return format("[%x]", slot*4) end
if rid < 128 then return disass.regname(rid) end
return ""
end
-- Dump CALL* function ref and return optional ctype.
local function dumpcallfunc(tr, ins)
local ctype
if ins > 0 then
local m, ot, op1, op2 = traceir(tr, ins)
if band(ot, 31) == 0 then -- nil type means CARG(func, ctype).
ins = op1
ctype = formatk(tr, op2)
end
end
if ins < 0 then
out:write(format("[0x%x](", tonumber((tracek(tr, ins)))))
else
out:write(format("%04d (", ins))
end
return ctype
end
-- Recursively gather CALL* args and dump them.
local function dumpcallargs(tr, ins)
if ins < 0 then
out:write(formatk(tr, ins))
else
local m, ot, op1, op2 = traceir(tr, ins)
local oidx = 6*shr(ot, 8)
local op = sub(vmdef.irnames, oidx+1, oidx+6)
if op == "CARG " then
dumpcallargs(tr, op1)
if op2 < 0 then
out:write(" ", formatk(tr, op2))
else
out:write(" ", format("%04d", op2))
end
else
out:write(format("%04d", ins))
end
end
end
-- Dump IR and interleaved snapshots.
local function dump_ir(tr, dumpsnap, dumpreg)
local info = traceinfo(tr)
if not info then return end
local nins = info.nins
out:write("---- TRACE ", tr, " IR\n")
local irnames = vmdef.irnames
local snapref = 65536
local snap, snapno
if dumpsnap then
snap = tracesnap(tr, 0)
snapref = snap[0]
snapno = 0
end
for ins=1,nins do
if ins >= snapref then
if dumpreg then
out:write(format(".... SNAP #%-3d [ ", snapno))
else
out:write(format(".... SNAP #%-3d [ ", snapno))
end
printsnap(tr, snap)
snapno = snapno + 1
snap = tracesnap(tr, snapno)
snapref = snap and snap[0] or 65536
end
local m, ot, op1, op2, ridsp = traceir(tr, ins)
local oidx, t = 6*shr(ot, 8), band(ot, 31)
local op = sub(irnames, oidx+1, oidx+6)
if op == "LOOP " then
if dumpreg then
out:write(format("%04d ------------ LOOP ------------\n", ins))
else
out:write(format("%04d ------ LOOP ------------\n", ins))
end
elseif op ~= "NOP " and op ~= "CARG " and
(dumpreg or op ~= "RENAME") then
local rid = band(ridsp, 255)
if dumpreg then
out:write(format("%04d %-6s", ins, ridsp_name(ridsp, ins)))
else
out:write(format("%04d ", ins))
end
out:write(format("%s%s %s %s ",
(rid == 254 or rid == 253) and "}" or
(band(ot, 128) == 0 and " " or ">"),
band(ot, 64) == 0 and " " or "+",
irtype[t], op))
local m1, m2 = band(m, 3), band(m, 3*4)
if sub(op, 1, 4) == "CALL" then
local ctype
if m2 == 1*4 then -- op2 == IRMlit
out:write(format("%-10s (", vmdef.ircall[op2]))
else
ctype = dumpcallfunc(tr, op2)
end
if op1 ~= -1 then dumpcallargs(tr, op1) end
out:write(")")
if ctype then out:write(" ctype ", ctype) end
elseif op == "CNEW " and op2 == -1 then
out:write(formatk(tr, op1))
elseif m1 ~= 3 then -- op1 != IRMnone
if op1 < 0 then
out:write(formatk(tr, op1))
else
out:write(format(m1 == 0 and "%04d" or "#%-3d", op1))
end
if m2 ~= 3*4 then -- op2 != IRMnone
if m2 == 1*4 then -- op2 == IRMlit
local litn = litname[op]
if litn and litn[op2] then
out:write(" ", litn[op2])
elseif op == "UREFO " or op == "UREFC " then
out:write(format(" #%-3d", shr(op2, 8)))
else
out:write(format(" #%-3d", op2))
end
elseif op2 < 0 then
out:write(" ", formatk(tr, op2))
else
out:write(format(" %04d", op2))
end
end
end
out:write("\n")
end
end
if snap then
if dumpreg then
out:write(format(".... SNAP #%-3d [ ", snapno))
else
out:write(format(".... SNAP #%-3d [ ", snapno))
end
printsnap(tr, snap)
end
end
------------------------------------------------------------------------------
local recprefix = ""
local recdepth = 0
-- Format trace error message.
local function fmterr(err, info)
if type(err) == "number" then
if type(info) == "function" then info = fmtfunc(info) end
err = format(vmdef.traceerr[err], info)
end
return err
end
-- Dump trace states.
local function dump_trace(what, tr, func, pc, otr, oex)
if what == "stop" or (what == "abort" and dumpmode.a) then
if dumpmode.i then dump_ir(tr, dumpmode.s, dumpmode.r and what == "stop")
elseif dumpmode.s then dump_snap(tr) end
if dumpmode.m then dump_mcode(tr) end
end
if what == "start" then
if dumpmode.H then out:write('<pre class="ljdump">\n') end
out:write("---- TRACE ", tr, " ", what)
if otr then out:write(" ", otr, "/", oex == -1 and "stitch" or oex) end
out:write(" ", fmtfunc(func, pc), "\n")
elseif what == "stop" or what == "abort" then
out:write("---- TRACE ", tr, " ", what)
if what == "abort" then
out:write(" ", fmtfunc(func, pc), " -- ", fmterr(otr, oex), "\n")
else
local info = traceinfo(tr)
local link, ltype = info.link, info.linktype
if link == tr or link == 0 then
out:write(" -> ", ltype, "\n")
elseif ltype == "root" then
out:write(" -> ", link, "\n")
else
out:write(" -> ", link, " ", ltype, "\n")
end
end
if dumpmode.H then out:write("</pre>\n\n") else out:write("\n") end
else
if what == "flush" then symtab, nexitsym = {}, 0 end
out:write("---- TRACE ", what, "\n\n")
end
out:flush()
end
-- Dump recorded bytecode.
local function dump_record(tr, func, pc, depth)
if depth ~= recdepth then
recdepth = depth
recprefix = rep(" .", depth)
end
local line
if pc >= 0 then
line = bcline(func, pc, recprefix)
if dumpmode.H then line = gsub(line, "[<>&]", html_escape) end
else
line = "0000 "..recprefix.." FUNCC \n"
end
if pc <= 0 then
out:write(sub(line, 1, -2), " ; ", fmtfunc(func), "\n")
else
out:write(line)
end
if pc >= 0 and band(funcbc(func, pc), 0xff) < 16 then -- ORDER BC
out:write(bcline(func, pc+1, recprefix)) -- Write JMP for cond.
end
end
------------------------------------------------------------------------------
local gpr64 = jit.arch:match("64")
local fprmips32 = jit.arch == "mips" or jit.arch == "mipsel"
-- Dump taken trace exits.
local function dump_texit(tr, ex, ngpr, nfpr, ...)
out:write("---- TRACE ", tr, " exit ", ex, "\n")
if dumpmode.X then
local regs = {...}
if gpr64 then
for i=1,ngpr do
out:write(format(" %016x", regs[i]))
if i % 4 == 0 then out:write("\n") end
end
else
for i=1,ngpr do
out:write(" ", tohex(regs[i]))
if i % 8 == 0 then out:write("\n") end
end
end
if fprmips32 then
for i=1,nfpr,2 do
out:write(format(" %+17.14g", regs[ngpr+i]))
if i % 8 == 7 then out:write("\n") end
end
else
for i=1,nfpr do
out:write(format(" %+17.14g", regs[ngpr+i]))
if i % 4 == 0 then out:write("\n") end
end
end
end
end
------------------------------------------------------------------------------
-- Detach dump handlers.
local function dumpoff()
if active then
active = false
jit.attach(dump_texit)
jit.attach(dump_record)
jit.attach(dump_trace)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach dump handlers.
local function dumpon(opt, outfile)
if active then dumpoff() end
local term = os.getenv("TERM")
local colormode = (term and term:match("color") or os.getenv("COLORTERM")) and "A" or "T"
if opt then
opt = gsub(opt, "[TAH]", function(mode) colormode = mode; return ""; end)
end
local m = { t=true, b=true, i=true, m=true, }
if opt and opt ~= "" then
local o = sub(opt, 1, 1)
if o ~= "+" and o ~= "-" then m = {} end
for i=1,#opt do m[sub(opt, i, i)] = (o ~= "-") end
end
dumpmode = m
if m.t or m.b or m.i or m.s or m.m then
jit.attach(dump_trace, "trace")
end
if m.b then
jit.attach(dump_record, "record")
if not bcline then bcline = require("jit.bc").line end
end
if m.x or m.X then
jit.attach(dump_texit, "texit")
end
if not outfile then outfile = os.getenv("LUAJIT_DUMPFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stdout
end
m[colormode] = true
if colormode == "A" then
colorize = colorize_ansi
irtype = irtype_ansi
elseif colormode == "H" then
colorize = colorize_html
irtype = irtype_html
out:write(header_html)
else
colorize = colorize_text
irtype = irtype_text
end
active = true
end
-- Public module functions.
return {
on = dumpon,
off = dumpoff,
start = dumpon -- For -j command line option.
}
@@ -0,0 +1,312 @@
----------------------------------------------------------------------------
-- LuaJIT profiler.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module is a simple command line interface to the built-in
-- low-overhead profiler of LuaJIT.
--
-- The lower-level API of the profiler is accessible via the "jit.profile"
-- module or the luaJIT_profile_* C API.
--
-- Example usage:
--
-- luajit -jp myapp.lua
-- luajit -jp=s myapp.lua
-- luajit -jp=-s myapp.lua
-- luajit -jp=vl myapp.lua
-- luajit -jp=G,profile.txt myapp.lua
--
-- The following dump features are available:
--
-- f Stack dump: function name, otherwise module:line. Default mode.
-- F Stack dump: ditto, but always prepend module.
-- l Stack dump: module:line.
-- <number> stack dump depth (callee < caller). Default: 1.
-- -<number> Inverse stack dump depth (caller > callee).
-- s Split stack dump after first stack level. Implies abs(depth) >= 2.
-- p Show full path for module names.
-- v Show VM states. Can be combined with stack dumps, e.g. vf or fv.
-- z Show zones. Can be combined with stack dumps, e.g. zf or fz.
-- r Show raw sample counts. Default: show percentages.
-- a Annotate excerpts from source code files.
-- A Annotate complete source code files.
-- G Produce raw output suitable for graphical tools (e.g. flame graphs).
-- m<number> Minimum sample percentage to be shown. Default: 3.
-- i<number> Sampling interval in milliseconds. Default: 10.
--
----------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local profile = require("jit.profile")
local vmdef = require("jit.vmdef")
local math = math
local pairs, ipairs, tonumber, floor = pairs, ipairs, tonumber, math.floor
local sort, format = table.sort, string.format
local stdout = io.stdout
local zone -- Load jit.zone module on demand.
-- Output file handle.
local out
------------------------------------------------------------------------------
local prof_ud
local prof_states, prof_split, prof_min, prof_raw, prof_fmt, prof_depth
local prof_ann, prof_count1, prof_count2, prof_samples
local map_vmmode = {
N = "Compiled",
I = "Interpreted",
C = "C code",
G = "Garbage Collector",
J = "JIT Compiler",
}
-- Profiler callback.
local function prof_cb(th, samples, vmmode)
prof_samples = prof_samples + samples
local key_stack, key_stack2, key_state
-- Collect keys for sample.
if prof_states then
if prof_states == "v" then
key_state = map_vmmode[vmmode] or vmmode
else
key_state = zone:get() or "(none)"
end
end
if prof_fmt then
key_stack = profile.dumpstack(th, prof_fmt, prof_depth)
key_stack = key_stack:gsub("%[builtin#(%d+)%]", function(x)
return vmdef.ffnames[tonumber(x)]
end)
if prof_split == 2 then
local k1, k2 = key_stack:match("(.-) [<>] (.*)")
if k2 then key_stack, key_stack2 = k1, k2 end
elseif prof_split == 3 then
key_stack2 = profile.dumpstack(th, "l", 1)
end
end
-- Order keys.
local k1, k2
if prof_split == 1 then
if key_state then
k1 = key_state
if key_stack then k2 = key_stack end
end
elseif key_stack then
k1 = key_stack
if key_stack2 then k2 = key_stack2 elseif key_state then k2 = key_state end
end
-- Coalesce samples in one or two levels.
if k1 then
local t1 = prof_count1
t1[k1] = (t1[k1] or 0) + samples
if k2 then
local t2 = prof_count2
local t3 = t2[k1]
if not t3 then t3 = {}; t2[k1] = t3 end
t3[k2] = (t3[k2] or 0) + samples
end
end
end
------------------------------------------------------------------------------
-- Show top N list.
local function prof_top(count1, count2, samples, indent)
local t, n = {}, 0
for k in pairs(count1) do
n = n + 1
t[n] = k
end
sort(t, function(a, b) return count1[a] > count1[b] end)
for i=1,n do
local k = t[i]
local v = count1[k]
local pct = floor(v*100/samples + 0.5)
if pct < prof_min then break end
if not prof_raw then
out:write(format("%s%2d%% %s\n", indent, pct, k))
elseif prof_raw == "r" then
out:write(format("%s%5d %s\n", indent, v, k))
else
out:write(format("%s %d\n", k, v))
end
if count2 then
local r = count2[k]
if r then
prof_top(r, nil, v, (prof_split == 3 or prof_split == 1) and " -- " or
(prof_depth < 0 and " -> " or " <- "))
end
end
end
end
-- Annotate source code
local function prof_annotate(count1, samples)
local files = {}
local ms = 0
for k, v in pairs(count1) do
local pct = floor(v*100/samples + 0.5)
ms = math.max(ms, v)
if pct >= prof_min then
local file, line = k:match("^(.*):(%d+)$")
if not file then file = k; line = 0 end
local fl = files[file]
if not fl then fl = {}; files[file] = fl; files[#files+1] = file end
line = tonumber(line)
fl[line] = prof_raw and v or pct
end
end
sort(files)
local fmtv, fmtn = " %3d%% | %s\n", " | %s\n"
if prof_raw then
local n = math.max(5, math.ceil(math.log10(ms)))
fmtv = "%"..n.."d | %s\n"
fmtn = (" "):rep(n).." | %s\n"
end
local ann = prof_ann
for _, file in ipairs(files) do
local f0 = file:byte()
if f0 == 40 or f0 == 91 then
out:write(format("\n====== %s ======\n[Cannot annotate non-file]\n", file))
break
end
local fp, err = io.open(file)
if not fp then
out:write(format("====== ERROR: %s: %s\n", file, err))
break
end
out:write(format("\n====== %s ======\n", file))
local fl = files[file]
local n, show = 1, false
if ann ~= 0 then
for i=1,ann do
if fl[i] then show = true; out:write("@@ 1 @@\n"); break end
end
end
for line in fp:lines() do
if line:byte() == 27 then
out:write("[Cannot annotate bytecode file]\n")
break
end
local v = fl[n]
if ann ~= 0 then
local v2 = fl[n+ann]
if show then
if v2 then show = n+ann elseif v then show = n
elseif show+ann < n then show = false end
elseif v2 then
show = n+ann
out:write(format("@@ %d @@\n", n))
end
if not show then goto next end
end
if v then
out:write(format(fmtv, v, line))
else
out:write(format(fmtn, line))
end
::next::
n = n + 1
end
fp:close()
end
end
------------------------------------------------------------------------------
-- Finish profiling and dump result.
local function prof_finish()
if prof_ud then
profile.stop()
local samples = prof_samples
if samples == 0 then
if prof_raw ~= true then out:write("[No samples collected]\n") end
return
end
if prof_ann then
prof_annotate(prof_count1, samples)
else
prof_top(prof_count1, prof_count2, samples, "")
end
prof_count1 = nil
prof_count2 = nil
prof_ud = nil
if out ~= stdout then out:close() end
end
end
-- Start profiling.
local function prof_start(mode)
local interval = ""
mode = mode:gsub("i%d*", function(s) interval = s; return "" end)
prof_min = 3
mode = mode:gsub("m(%d+)", function(s) prof_min = tonumber(s); return "" end)
prof_depth = 1
mode = mode:gsub("%-?%d+", function(s) prof_depth = tonumber(s); return "" end)
local m = {}
for c in mode:gmatch(".") do m[c] = c end
prof_states = m.z or m.v
if prof_states == "z" then zone = require("jit.zone") end
local scope = m.l or m.f or m.F or (prof_states and "" or "f")
local flags = (m.p or "")
prof_raw = m.r
if m.s then
prof_split = 2
if prof_depth == -1 or m["-"] then prof_depth = -2
elseif prof_depth == 1 then prof_depth = 2 end
elseif mode:find("[fF].*l") then
scope = "l"
prof_split = 3
else
prof_split = (scope == "" or mode:find("[zv].*[lfF]")) and 1 or 0
end
prof_ann = m.A and 0 or (m.a and 3)
if prof_ann then
scope = "l"
prof_fmt = "pl"
prof_split = 0
prof_depth = 1
elseif m.G and scope ~= "" then
prof_fmt = flags..scope.."Z;"
prof_depth = -100
prof_raw = true
prof_min = 0
elseif scope == "" then
prof_fmt = false
else
local sc = prof_split == 3 and m.f or m.F or scope
prof_fmt = flags..sc..(prof_depth >= 0 and "Z < " or "Z > ")
end
prof_count1 = {}
prof_count2 = {}
prof_samples = 0
profile.start(scope:lower()..interval, prof_cb)
prof_ud = newproxy(true)
getmetatable(prof_ud).__gc = prof_finish
end
------------------------------------------------------------------------------
local function start(mode, outfile)
if not outfile then outfile = os.getenv("LUAJIT_PROFILEFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stdout
end
prof_start(mode or "f")
end
-- Public module functions.
return {
start = start, -- For -j command line option.
stop = prof_finish
}
@@ -0,0 +1,170 @@
----------------------------------------------------------------------------
-- Verbose mode of the LuaJIT compiler.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module shows verbose information about the progress of the
-- JIT compiler. It prints one line for each generated trace. This module
-- is useful to see which code has been compiled or where the compiler
-- punts and falls back to the interpreter.
--
-- Example usage:
--
-- luajit -jv -e "for i=1,1000 do for j=1,1000 do end end"
-- luajit -jv=myapp.out myapp.lua
--
-- Default output is to stderr. To redirect the output to a file, pass a
-- filename as an argument (use '-' for stdout) or set the environment
-- variable LUAJIT_VERBOSEFILE. The file is overwritten every time the
-- module is started.
--
-- The output from the first example should look like this:
--
-- [TRACE 1 (command line):1 loop]
-- [TRACE 2 (1/3) (command line):1 -> 1]
--
-- The first number in each line is the internal trace number. Next are
-- the file name ('(command line)') and the line number (':1') where the
-- trace has started. Side traces also show the parent trace number and
-- the exit number where they are attached to in parentheses ('(1/3)').
-- An arrow at the end shows where the trace links to ('-> 1'), unless
-- it loops to itself.
--
-- In this case the inner loop gets hot and is traced first, generating
-- a root trace. Then the last exit from the 1st trace gets hot, too,
-- and triggers generation of the 2nd trace. The side trace follows the
-- path along the outer loop and *around* the inner loop, back to its
-- start, and then links to the 1st trace. Yes, this may seem unusual,
-- if you know how traditional compilers work. Trace compilers are full
-- of surprises like this -- have fun! :-)
--
-- Aborted traces are shown like this:
--
-- [TRACE --- foo.lua:44 -- leaving loop in root trace at foo:lua:50]
--
-- Don't worry -- trace aborts are quite common, even in programs which
-- can be fully compiled. The compiler may retry several times until it
-- finds a suitable trace.
--
-- Of course this doesn't work with features that are not-yet-implemented
-- (NYI error messages). The VM simply falls back to the interpreter. This
-- may not matter at all if the particular trace is not very high up in
-- the CPU usage profile. Oh, and the interpreter is quite fast, too.
--
-- Also check out the -jdump module, which prints all the gory details.
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local funcinfo, traceinfo = jutil.funcinfo, jutil.traceinfo
local type, format = type, string.format
local stdout, stderr = io.stdout, io.stderr
-- Active flag and output file handle.
local active, out
------------------------------------------------------------------------------
local startloc, startex
local function fmtfunc(func, pc)
local fi = funcinfo(func, pc)
if fi.loc then
return fi.loc
elseif fi.ffid then
return vmdef.ffnames[fi.ffid]
elseif fi.addr then
return format("C:%x", fi.addr)
else
return "(?)"
end
end
-- Format trace error message.
local function fmterr(err, info)
if type(err) == "number" then
if type(info) == "function" then info = fmtfunc(info) end
err = format(vmdef.traceerr[err], info)
end
return err
end
-- Dump trace states.
local function dump_trace(what, tr, func, pc, otr, oex)
if what == "start" then
startloc = fmtfunc(func, pc)
startex = otr and "("..otr.."/"..(oex == -1 and "stitch" or oex)..") " or ""
else
if what == "abort" then
local loc = fmtfunc(func, pc)
if loc ~= startloc then
out:write(format("[TRACE --- %s%s -- %s at %s]\n",
startex, startloc, fmterr(otr, oex), loc))
else
out:write(format("[TRACE --- %s%s -- %s]\n",
startex, startloc, fmterr(otr, oex)))
end
elseif what == "stop" then
local info = traceinfo(tr)
local link, ltype = info.link, info.linktype
if ltype == "interpreter" then
out:write(format("[TRACE %3s %s%s -- fallback to interpreter]\n",
tr, startex, startloc))
elseif ltype == "stitch" then
out:write(format("[TRACE %3s %s%s %s %s]\n",
tr, startex, startloc, ltype, fmtfunc(func, pc)))
elseif link == tr or link == 0 then
out:write(format("[TRACE %3s %s%s %s]\n",
tr, startex, startloc, ltype))
elseif ltype == "root" then
out:write(format("[TRACE %3s %s%s -> %d]\n",
tr, startex, startloc, link))
else
out:write(format("[TRACE %3s %s%s -> %d %s]\n",
tr, startex, startloc, link, ltype))
end
else
out:write(format("[TRACE %s]\n", what))
end
out:flush()
end
end
------------------------------------------------------------------------------
-- Detach dump handlers.
local function dumpoff()
if active then
active = false
jit.attach(dump_trace)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach dump handlers.
local function dumpon(outfile)
if active then dumpoff() end
if not outfile then outfile = os.getenv("LUAJIT_VERBOSEFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stderr
end
jit.attach(dump_trace, "trace")
active = true
end
-- Public module functions.
return {
on = dumpon,
off = dumpoff,
start = dumpon -- For -j command line option.
}
@@ -0,0 +1,395 @@
-- This is a generated file. DO NOT EDIT!
return {
bcnames = "ISLT ISGE ISLE ISGT ISEQV ISNEV ISEQS ISNES ISEQN ISNEN ISEQP ISNEP ISTC ISFC IST ISF ISTYPEISNUM MOV NOT UNM LEN ADDVN SUBVN MULVN DIVVN MODVN ADDNV SUBNV MULNV DIVNV MODNV ADDVV SUBVV MULVV DIVVV MODVV POW CAT KSTR KCDATAKSHORTKNUM KPRI KNIL UGET USETV USETS USETN USETP UCLO FNEW TNEW TDUP GGET GSET TGETV TGETS TGETB TGETR TSETV TSETS TSETB TSETM TSETR CALLM CALL CALLMTCALLT ITERC ITERN VARG ISNEXTRETM RET RET0 RET1 FORI JFORI FORL IFORL JFORL ITERL IITERLJITERLLOOP ILOOP JLOOP JMP FUNCF IFUNCFJFUNCFFUNCV IFUNCVJFUNCVFUNCC FUNCCW",
irnames = "LT GE LE GT ULT UGE ULE UGT EQ NE ABC RETF NOP BASE PVAL GCSTEPHIOP LOOP USE PHI RENAMEPROF KPRI KINT KGC KPTR KKPTR KNULL KNUM KINT64KSLOT BNOT BSWAP BAND BOR BXOR BSHL BSHR BSAR BROL BROR ADD SUB MUL DIV MOD POW NEG ABS LDEXP MIN MAX FPMATHADDOV SUBOV MULOV AREF HREFK HREF NEWREFUREFO UREFC FREF TMPREFSTRREFLREF ALOAD HLOAD ULOAD FLOAD XLOAD SLOAD VLOAD ALEN ASTOREHSTOREUSTOREFSTOREXSTORESNEW XSNEW TNEW TDUP CNEW CNEWI BUFHDRBUFPUTBUFSTRTBAR OBAR XBAR CONV TOBIT TOSTR STRTO CALLN CALLA CALLL CALLS CALLXSCARG ",
irfpm = { [0]="floor", "ceil", "trunc", "sqrt", "log", "log2", "other", },
irfield = { [0]="str.len", "func.env", "func.pc", "func.ffid", "thread.env", "tab.meta", "tab.array", "tab.node", "tab.asize", "tab.hmask", "tab.nomm", "udata.meta", "udata.udtype", "udata.file", "sbuf.w", "sbuf.e", "sbuf.b", "sbuf.l", "sbuf.ref", "sbuf.r", "cdata.ctypeid", "cdata.ptr", "cdata.int", "cdata.int64", "cdata.int64_4", },
ircall = {
[0]="lj_str_cmp",
"lj_str_find",
"lj_str_new",
"lj_strscan_num",
"lj_strfmt_int",
"lj_strfmt_num",
"lj_strfmt_char",
"lj_strfmt_putint",
"lj_strfmt_putnum",
"lj_strfmt_putquoted",
"lj_strfmt_putfxint",
"lj_strfmt_putfnum_int",
"lj_strfmt_putfnum_uint",
"lj_strfmt_putfnum",
"lj_strfmt_putfstr",
"lj_strfmt_putfchar",
"lj_buf_putmem",
"lj_buf_putstr",
"lj_buf_putchar",
"lj_buf_putstr_reverse",
"lj_buf_putstr_lower",
"lj_buf_putstr_upper",
"lj_buf_putstr_rep",
"lj_buf_puttab",
"lj_bufx_set",
"lj_bufx_more",
"lj_serialize_put",
"lj_serialize_get",
"lj_serialize_encode",
"lj_serialize_decode",
"lj_buf_tostr",
"lj_tab_new_ah",
"lj_tab_new1",
"lj_tab_dup",
"lj_tab_clear",
"lj_tab_newkey",
"lj_tab_keyindex",
"lj_vm_next",
"lj_tab_len",
"lj_tab_len_hint",
"lj_gc_step_jit",
"lj_gc_barrieruv",
"lj_mem_newgco",
"lj_prng_u64d",
"lj_vm_modi",
"log10",
"exp",
"sin",
"cos",
"tan",
"asin",
"acos",
"atan",
"sinh",
"cosh",
"tanh",
"fputc",
"fwrite",
"fflush",
"lj_vm_floor",
"lj_vm_ceil",
"lj_vm_trunc",
"sqrt",
"log",
"lj_vm_log2",
"lj_vm_powi",
"pow",
"atan2",
"ldexp",
"lj_vm_tobit",
"softfp_add",
"softfp_sub",
"softfp_mul",
"softfp_div",
"softfp_cmp",
"softfp_i2d",
"softfp_d2i",
"lj_vm_sfmin",
"lj_vm_sfmax",
"lj_vm_tointg",
"softfp_ui2d",
"softfp_f2d",
"softfp_d2ui",
"softfp_d2f",
"softfp_i2f",
"softfp_ui2f",
"softfp_f2i",
"softfp_f2ui",
"fp64_l2d",
"fp64_ul2d",
"fp64_l2f",
"fp64_ul2f",
"fp64_d2l",
"fp64_d2ul",
"fp64_f2l",
"fp64_f2ul",
"lj_carith_divi64",
"lj_carith_divu64",
"lj_carith_modi64",
"lj_carith_modu64",
"lj_carith_powi64",
"lj_carith_powu64",
"lj_cdata_newv",
"lj_cdata_setfin",
"strlen",
"memcpy",
"memset",
"lj_vm_errno",
"lj_carith_mul64",
"lj_carith_shl64",
"lj_carith_shr64",
"lj_carith_sar64",
"lj_carith_rol64",
"lj_carith_ror64",
},
traceerr = {
[0]="error thrown or hook called during recording",
"trace too short",
"trace too long",
"trace too deep",
"too many snapshots",
"blacklisted",
"retry recording",
"NYI: bytecode %d",
"leaving loop in root trace",
"inner loop in root trace",
"loop unroll limit reached",
"bad argument type",
"JIT compilation disabled for function",
"call unroll limit reached",
"down-recursion, restarting",
"NYI: unsupported variant of FastFunc %s",
"NYI: return to lower frame",
"store with nil or NaN key",
"missing metamethod",
"looping index lookup",
"NYI: mixed sparse/dense table",
"symbol not in cache",
"NYI: unsupported C type conversion",
"NYI: unsupported C function type",
"guard would always fail",
"too many PHIs",
"persistent type instability",
"failed to allocate mcode memory",
"machine code too long",
"hit mcode limit (retrying)",
"too many spill slots",
"inconsistent register allocation",
"NYI: cannot assemble IR instruction %d",
"NYI: PHI shuffling too complex",
"NYI: register coalescing too complex",
},
ffnames = {
[0]="Lua",
"C",
"assert",
"type",
"next",
"pairs",
"ipairs_aux",
"ipairs",
"getmetatable",
"setmetatable",
"getfenv",
"setfenv",
"rawget",
"rawset",
"rawequal",
"unpack",
"select",
"tonumber",
"tostring",
"error",
"pcall",
"xpcall",
"loadfile",
"load",
"loadstring",
"dofile",
"gcinfo",
"collectgarbage",
"newproxy",
"print",
"coroutine.status",
"coroutine.running",
"coroutine.isyieldable",
"coroutine.create",
"coroutine.yield",
"coroutine.resume",
"coroutine.wrap_aux",
"coroutine.wrap",
"math.abs",
"math.floor",
"math.ceil",
"math.sqrt",
"math.log10",
"math.exp",
"math.sin",
"math.cos",
"math.tan",
"math.asin",
"math.acos",
"math.atan",
"math.sinh",
"math.cosh",
"math.tanh",
"math.frexp",
"math.modf",
"math.log",
"math.atan2",
"math.pow",
"math.fmod",
"math.ldexp",
"math.min",
"math.max",
"math.random",
"math.randomseed",
"bit.tobit",
"bit.bnot",
"bit.bswap",
"bit.lshift",
"bit.rshift",
"bit.arshift",
"bit.rol",
"bit.ror",
"bit.band",
"bit.bor",
"bit.bxor",
"bit.tohex",
"string.byte",
"string.char",
"string.sub",
"string.rep",
"string.reverse",
"string.lower",
"string.upper",
"string.dump",
"string.find",
"string.match",
"string.gmatch_aux",
"string.gmatch",
"string.gsub",
"string.format",
"table.maxn",
"table.insert",
"table.concat",
"table.sort",
"table.new",
"table.clear",
"io.method.close",
"io.method.read",
"io.method.write",
"io.method.flush",
"io.method.seek",
"io.method.setvbuf",
"io.method.lines",
"io.method.__gc",
"io.method.__tostring",
"io.open",
"io.popen",
"io.tmpfile",
"io.close",
"io.read",
"io.write",
"io.flush",
"io.input",
"io.output",
"io.lines",
"io.type",
"os.execute",
"os.remove",
"os.rename",
"os.tmpname",
"os.getenv",
"os.exit",
"os.clock",
"os.date",
"os.time",
"os.difftime",
"os.setlocale",
"debug.getregistry",
"debug.getmetatable",
"debug.setmetatable",
"debug.getfenv",
"debug.setfenv",
"debug.getinfo",
"debug.getlocal",
"debug.setlocal",
"debug.getupvalue",
"debug.setupvalue",
"debug.upvalueid",
"debug.upvaluejoin",
"debug.sethook",
"debug.gethook",
"debug.debug",
"debug.traceback",
"jit.on",
"jit.off",
"jit.flush",
"jit.status",
"jit.security",
"jit.attach",
"jit.util.funcinfo",
"jit.util.funcbc",
"jit.util.funck",
"jit.util.funcuvname",
"jit.util.traceinfo",
"jit.util.traceir",
"jit.util.tracek",
"jit.util.tracesnap",
"jit.util.tracemc",
"jit.util.traceexitstub",
"jit.util.ircalladdr",
"jit.opt.start",
"jit.profile.start",
"jit.profile.stop",
"jit.profile.dumpstack",
"ffi.meta.__index",
"ffi.meta.__newindex",
"ffi.meta.__eq",
"ffi.meta.__len",
"ffi.meta.__lt",
"ffi.meta.__le",
"ffi.meta.__concat",
"ffi.meta.__call",
"ffi.meta.__add",
"ffi.meta.__sub",
"ffi.meta.__mul",
"ffi.meta.__div",
"ffi.meta.__mod",
"ffi.meta.__pow",
"ffi.meta.__unm",
"ffi.meta.__tostring",
"ffi.meta.__pairs",
"ffi.meta.__ipairs",
"ffi.clib.__index",
"ffi.clib.__newindex",
"ffi.clib.__gc",
"ffi.callback.free",
"ffi.callback.set",
"ffi.cdef",
"ffi.new",
"ffi.cast",
"ffi.typeof",
"ffi.typeinfo",
"ffi.istype",
"ffi.sizeof",
"ffi.alignof",
"ffi.offsetof",
"ffi.errno",
"ffi.string",
"ffi.copy",
"ffi.fill",
"ffi.abi",
"ffi.metatype",
"ffi.gc",
"ffi.load",
"buffer.method.free",
"buffer.method.reset",
"buffer.method.skip",
"buffer.method.set",
"buffer.method.put",
"buffer.method.putf",
"buffer.method.get",
"buffer.method.putcdata",
"buffer.method.reserve",
"buffer.method.commit",
"buffer.method.ref",
"buffer.method.encode",
"buffer.method.decode",
"buffer.method.__gc",
"buffer.method.__tostring",
"buffer.method.__len",
"buffer.new",
"buffer.encode",
"buffer.decode",
},
}
@@ -0,0 +1,45 @@
----------------------------------------------------------------------------
-- LuaJIT profiler zones.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module implements a simple hierarchical zone model.
--
-- Example usage:
--
-- local zone = require("jit.zone")
-- zone("AI")
-- ...
-- zone("A*")
-- ...
-- print(zone:get()) --> "A*"
-- ...
-- zone()
-- ...
-- print(zone:get()) --> "AI"
-- ...
-- zone()
--
----------------------------------------------------------------------------
local remove = table.remove
return setmetatable({
flush = function(t)
for i=#t,1,-1 do t[i] = nil end
end,
get = function(t)
return t[#t]
end
}, {
__call = function(t, zone)
if zone then
t[#t+1] = zone
else
return (assert(remove(t), "empty zone stack"))
end
end
})
@@ -0,0 +1,190 @@
----------------------------------------------------------------------------
-- LuaJIT bytecode listing module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module lists the bytecode of a Lua function. If it's loaded by -jbc
-- it hooks into the parser and lists all functions of a chunk as they
-- are parsed.
--
-- Example usage:
--
-- luajit -jbc -e 'local x=0; for i=1,1e6 do x=x+i end; print(x)'
-- luajit -jbc=- foo.lua
-- luajit -jbc=foo.list foo.lua
--
-- Default output is to stderr. To redirect the output to a file, pass a
-- filename as an argument (use '-' for stdout) or set the environment
-- variable LUAJIT_LISTFILE. The file is overwritten every time the module
-- is started.
--
-- This module can also be used programmatically:
--
-- local bc = require("jit.bc")
--
-- local function foo() print("hello") end
--
-- bc.dump(foo) --> -- BYTECODE -- [...]
-- print(bc.line(foo, 2)) --> 0002 KSTR 1 1 ; "hello"
--
-- local out = {
-- -- Do something with each line:
-- write = function(t, ...) io.write(...) end,
-- close = function(t) end,
-- flush = function(t) end,
-- }
-- bc.dump(foo, out)
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local bit = require("bit")
local sub, gsub, format = string.sub, string.gsub, string.format
local byte, band, shr = string.byte, bit.band, bit.rshift
local funcinfo, funcbc, funck = jutil.funcinfo, jutil.funcbc, jutil.funck
local funcuvname = jutil.funcuvname
local bcnames = vmdef.bcnames
local stdout, stderr = io.stdout, io.stderr
------------------------------------------------------------------------------
local function ctlsub(c)
if c == "\n" then return "\\n"
elseif c == "\r" then return "\\r"
elseif c == "\t" then return "\\t"
else return format("\\%03d", byte(c))
end
end
-- Return one bytecode line.
local function bcline(func, pc, prefix)
local ins, m = funcbc(func, pc)
if not ins then return end
local ma, mb, mc = band(m, 7), band(m, 15*8), band(m, 15*128)
local a = band(shr(ins, 8), 0xff)
local oidx = 6*band(ins, 0xff)
local op = sub(bcnames, oidx+1, oidx+6)
local s = format("%04d %s %-6s %3s ",
pc, prefix or " ", op, ma == 0 and "" or a)
local d = shr(ins, 16)
if mc == 13*128 then -- BCMjump
return format("%s=> %04d\n", s, pc+d-0x7fff)
end
if mb ~= 0 then
d = band(d, 0xff)
elseif mc == 0 then
return s.."\n"
end
local kc
if mc == 10*128 then -- BCMstr
kc = funck(func, -d-1)
kc = format(#kc > 40 and '"%.40s"~' or '"%s"', gsub(kc, "%c", ctlsub))
elseif mc == 9*128 then -- BCMnum
kc = funck(func, d)
if op == "TSETM " then kc = kc - 2^52 end
elseif mc == 12*128 then -- BCMfunc
local fi = funcinfo(funck(func, -d-1))
if fi.ffid then
kc = vmdef.ffnames[fi.ffid]
else
kc = fi.loc
end
elseif mc == 5*128 then -- BCMuv
kc = funcuvname(func, d)
end
if ma == 5 then -- BCMuv
local ka = funcuvname(func, a)
if kc then kc = ka.." ; "..kc else kc = ka end
end
if mb ~= 0 then
local b = shr(ins, 24)
if kc then return format("%s%3d %3d ; %s\n", s, b, d, kc) end
return format("%s%3d %3d\n", s, b, d)
end
if kc then return format("%s%3d ; %s\n", s, d, kc) end
if mc == 7*128 and d > 32767 then d = d - 65536 end -- BCMlits
return format("%s%3d\n", s, d)
end
-- Collect branch targets of a function.
local function bctargets(func)
local target = {}
for pc=1,1000000000 do
local ins, m = funcbc(func, pc)
if not ins then break end
if band(m, 15*128) == 13*128 then target[pc+shr(ins, 16)-0x7fff] = true end
end
return target
end
-- Dump bytecode instructions of a function.
local function bcdump(func, out, all)
if not out then out = stdout end
local fi = funcinfo(func)
if all and fi.children then
for n=-1,-1000000000,-1 do
local k = funck(func, n)
if not k then break end
if type(k) == "proto" then bcdump(k, out, true) end
end
end
out:write(format("-- BYTECODE -- %s-%d\n", fi.loc, fi.lastlinedefined))
local target = bctargets(func)
for pc=1,1000000000 do
local s = bcline(func, pc, target[pc] and "=>")
if not s then break end
out:write(s)
end
out:write("\n")
out:flush()
end
------------------------------------------------------------------------------
-- Active flag and output file handle.
local active, out
-- List handler.
local function h_list(func)
return bcdump(func, out)
end
-- Detach list handler.
local function bclistoff()
if active then
active = false
jit.attach(h_list)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach list handler.
local function bcliston(outfile)
if active then bclistoff() end
if not outfile then outfile = os.getenv("LUAJIT_LISTFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stderr
end
jit.attach(h_list, "bc")
active = true
end
-- Public module functions.
return {
line = bcline,
dump = bcdump,
targets = bctargets,
on = bcliston,
off = bclistoff,
start = bcliston -- For -j command line option.
}
@@ -0,0 +1,665 @@
----------------------------------------------------------------------------
-- LuaJIT module to save/list bytecode.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module saves or lists the bytecode for an input file.
-- It's run by the -b command line option.
--
------------------------------------------------------------------------------
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local bit = require("bit")
-- Symbol name prefix for LuaJIT bytecode.
local LJBC_PREFIX = "luaJIT_BC_"
local type, assert = type, assert
local format = string.format
local tremove, tconcat = table.remove, table.concat
------------------------------------------------------------------------------
local function usage()
io.stderr:write[[
Save LuaJIT bytecode: luajit -b[options] input output
-l Only list bytecode.
-s Strip debug info (default).
-g Keep debug info.
-n name Set module name (default: auto-detect from input name).
-t type Set output file type (default: auto-detect from output name).
-a arch Override architecture for object files (default: native).
-o os Override OS for object files (default: native).
-e chunk Use chunk string as input.
-- Stop handling options.
- Use stdin as input and/or stdout as output.
File types: c h obj o raw (default)
]]
os.exit(1)
end
local function check(ok, ...)
if ok then return ok, ... end
io.stderr:write("luajit: ", ...)
io.stderr:write("\n")
os.exit(1)
end
local function readfile(input)
if type(input) == "function" then return input end
if input == "-" then input = nil end
return check(loadfile(input))
end
local function savefile(name, mode)
if name == "-" then return io.stdout end
return check(io.open(name, mode))
end
------------------------------------------------------------------------------
local map_type = {
raw = "raw", c = "c", h = "h", o = "obj", obj = "obj",
}
local map_arch = {
x86 = { e = "le", b = 32, m = 3, p = 0x14c, },
x64 = { e = "le", b = 64, m = 62, p = 0x8664, },
arm = { e = "le", b = 32, m = 40, p = 0x1c0, },
arm64 = { e = "le", b = 64, m = 183, p = 0xaa64, },
arm64be = { e = "be", b = 64, m = 183, },
ppc = { e = "be", b = 32, m = 20, },
mips = { e = "be", b = 32, m = 8, f = 0x50001006, },
mipsel = { e = "le", b = 32, m = 8, f = 0x50001006, },
mips64 = { e = "be", b = 64, m = 8, f = 0x80000007, },
mips64el = { e = "le", b = 64, m = 8, f = 0x80000007, },
mips64r6 = { e = "be", b = 64, m = 8, f = 0xa0000407, },
mips64r6el = { e = "le", b = 64, m = 8, f = 0xa0000407, },
}
local map_os = {
linux = true, windows = true, osx = true, freebsd = true, netbsd = true,
openbsd = true, dragonfly = true, solaris = true,
}
local function checkarg(str, map, err)
str = str:lower()
local s = check(map[str], "unknown ", err)
return type(s) == "string" and s or str
end
local function detecttype(str)
local ext = str:lower():match("%.(%a+)$")
return map_type[ext] or "raw"
end
local function checkmodname(str)
check(str:match("^[%w_.%-]+$"), "bad module name")
return str:gsub("[%.%-]", "_")
end
local function detectmodname(str)
if type(str) == "string" then
local tail = str:match("[^/\\]+$")
if tail then str = tail end
local head = str:match("^(.*)%.[^.]*$")
if head then str = head end
str = str:match("^[%w_.%-]+")
else
str = nil
end
check(str, "cannot derive module name, use -n name")
return str:gsub("[%.%-]", "_")
end
------------------------------------------------------------------------------
local function bcsave_tail(fp, output, s)
local ok, err = fp:write(s)
if ok and output ~= "-" then ok, err = fp:close() end
check(ok, "cannot write ", output, ": ", err)
end
local function bcsave_raw(output, s)
local fp = savefile(output, "wb")
bcsave_tail(fp, output, s)
end
local function bcsave_c(ctx, output, s)
local fp = savefile(output, "w")
if ctx.type == "c" then
fp:write(format([[
#ifdef __cplusplus
extern "C"
#endif
#ifdef _WIN32
__declspec(dllexport)
#endif
const unsigned char %s%s[] = {
]], LJBC_PREFIX, ctx.modname))
else
fp:write(format([[
#define %s%s_SIZE %d
static const unsigned char %s%s[] = {
]], LJBC_PREFIX, ctx.modname, #s, LJBC_PREFIX, ctx.modname))
end
local t, n, m = {}, 0, 0
for i=1,#s do
local b = tostring(string.byte(s, i))
m = m + #b + 1
if m > 78 then
fp:write(tconcat(t, ",", 1, n), ",\n")
n, m = 0, #b + 1
end
n = n + 1
t[n] = b
end
bcsave_tail(fp, output, tconcat(t, ",", 1, n).."\n};\n")
end
local function bcsave_elfobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct {
uint8_t emagic[4], eclass, eendian, eversion, eosabi, eabiversion, epad[7];
uint16_t type, machine;
uint32_t version;
uint32_t entry, phofs, shofs;
uint32_t flags;
uint16_t ehsize, phentsize, phnum, shentsize, shnum, shstridx;
} ELF32header;
typedef struct {
uint8_t emagic[4], eclass, eendian, eversion, eosabi, eabiversion, epad[7];
uint16_t type, machine;
uint32_t version;
uint64_t entry, phofs, shofs;
uint32_t flags;
uint16_t ehsize, phentsize, phnum, shentsize, shnum, shstridx;
} ELF64header;
typedef struct {
uint32_t name, type, flags, addr, ofs, size, link, info, align, entsize;
} ELF32sectheader;
typedef struct {
uint32_t name, type;
uint64_t flags, addr, ofs, size;
uint32_t link, info;
uint64_t align, entsize;
} ELF64sectheader;
typedef struct {
uint32_t name, value, size;
uint8_t info, other;
uint16_t sectidx;
} ELF32symbol;
typedef struct {
uint32_t name;
uint8_t info, other;
uint16_t sectidx;
uint64_t value, size;
} ELF64symbol;
typedef struct {
ELF32header hdr;
ELF32sectheader sect[6];
ELF32symbol sym[2];
uint8_t space[4096];
} ELF32obj;
typedef struct {
ELF64header hdr;
ELF64sectheader sect[6];
ELF64symbol sym[2];
uint8_t space[4096];
} ELF64obj;
]]
local symname = LJBC_PREFIX..ctx.modname
local ai = assert(map_arch[ctx.arch])
local is64, isbe = ai.b == 64, ai.e == "be"
-- Handle different host/target endianess.
local function f32(x) return x end
local f16, fofs = f32, f32
if ffi.abi("be") ~= isbe then
f32 = bit.bswap
function f16(x) return bit.rshift(bit.bswap(x), 16) end
if is64 then
local two32 = ffi.cast("int64_t", 2^32)
function fofs(x) return bit.bswap(x)*two32 end
else
fofs = f32
end
end
-- Create ELF object and fill in header.
local o = ffi.new(is64 and "ELF64obj" or "ELF32obj")
local hdr = o.hdr
if ctx.os == "bsd" or ctx.os == "other" then -- Determine native hdr.eosabi.
local bf = assert(io.open("/bin/ls", "rb"))
local bs = bf:read(9)
bf:close()
ffi.copy(o, bs, 9)
check(hdr.emagic[0] == 127, "no support for writing native object files")
else
hdr.emagic = "\127ELF"
hdr.eosabi = ({ freebsd=9, netbsd=2, openbsd=12, solaris=6 })[ctx.os] or 0
end
hdr.eclass = is64 and 2 or 1
hdr.eendian = isbe and 2 or 1
hdr.eversion = 1
hdr.type = f16(1)
hdr.machine = f16(ai.m)
hdr.flags = f32(ai.f or 0)
hdr.version = f32(1)
hdr.shofs = fofs(ffi.offsetof(o, "sect"))
hdr.ehsize = f16(ffi.sizeof(hdr))
hdr.shentsize = f16(ffi.sizeof(o.sect[0]))
hdr.shnum = f16(6)
hdr.shstridx = f16(2)
-- Fill in sections and symbols.
local sofs, ofs = ffi.offsetof(o, "space"), 1
for i,name in ipairs{
".symtab", ".shstrtab", ".strtab", ".rodata", ".note.GNU-stack",
} do
local sect = o.sect[i]
sect.align = fofs(1)
sect.name = f32(ofs)
ffi.copy(o.space+ofs, name)
ofs = ofs + #name+1
end
o.sect[1].type = f32(2) -- .symtab
o.sect[1].link = f32(3)
o.sect[1].info = f32(1)
o.sect[1].align = fofs(8)
o.sect[1].ofs = fofs(ffi.offsetof(o, "sym"))
o.sect[1].entsize = fofs(ffi.sizeof(o.sym[0]))
o.sect[1].size = fofs(ffi.sizeof(o.sym))
o.sym[1].name = f32(1)
o.sym[1].sectidx = f16(4)
o.sym[1].size = fofs(#s)
o.sym[1].info = 17
o.sect[2].type = f32(3) -- .shstrtab
o.sect[2].ofs = fofs(sofs)
o.sect[2].size = fofs(ofs)
o.sect[3].type = f32(3) -- .strtab
o.sect[3].ofs = fofs(sofs + ofs)
o.sect[3].size = fofs(#symname+2)
ffi.copy(o.space+ofs+1, symname)
ofs = ofs + #symname + 2
o.sect[4].type = f32(1) -- .rodata
o.sect[4].flags = fofs(2)
o.sect[4].ofs = fofs(sofs + ofs)
o.sect[4].size = fofs(#s)
o.sect[5].type = f32(1) -- .note.GNU-stack
o.sect[5].ofs = fofs(sofs + ofs + #s)
-- Write ELF object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, ffi.sizeof(o)-4096+ofs))
bcsave_tail(fp, output, s)
end
local function bcsave_peobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct {
uint16_t arch, nsects;
uint32_t time, symtabofs, nsyms;
uint16_t opthdrsz, flags;
} PEheader;
typedef struct {
char name[8];
uint32_t vsize, vaddr, size, ofs, relocofs, lineofs;
uint16_t nreloc, nline;
uint32_t flags;
} PEsection;
typedef struct __attribute((packed)) {
union {
char name[8];
uint32_t nameref[2];
};
uint32_t value;
int16_t sect;
uint16_t type;
uint8_t scl, naux;
} PEsym;
typedef struct __attribute((packed)) {
uint32_t size;
uint16_t nreloc, nline;
uint32_t cksum;
uint16_t assoc;
uint8_t comdatsel, unused[3];
} PEsymaux;
typedef struct {
PEheader hdr;
PEsection sect[2];
// Must be an even number of symbol structs.
PEsym sym0;
PEsymaux sym0aux;
PEsym sym1;
PEsymaux sym1aux;
PEsym sym2;
PEsym sym3;
uint32_t strtabsize;
uint8_t space[4096];
} PEobj;
]]
local symname = LJBC_PREFIX..ctx.modname
local ai = assert(map_arch[ctx.arch])
local is64 = ai.b == 64
local symexport = " /EXPORT:"..symname..",DATA "
-- The file format is always little-endian. Swap if the host is big-endian.
local function f32(x) return x end
local f16 = f32
if ffi.abi("be") then
f32 = bit.bswap
function f16(x) return bit.rshift(bit.bswap(x), 16) end
end
-- Create PE object and fill in header.
local o = ffi.new("PEobj")
local hdr = o.hdr
hdr.arch = f16(assert(ai.p))
hdr.nsects = f16(2)
hdr.symtabofs = f32(ffi.offsetof(o, "sym0"))
hdr.nsyms = f32(6)
-- Fill in sections and symbols.
o.sect[0].name = ".drectve"
o.sect[0].size = f32(#symexport)
o.sect[0].flags = f32(0x00100a00)
o.sym0.sect = f16(1)
o.sym0.scl = 3
o.sym0.name = ".drectve"
o.sym0.naux = 1
o.sym0aux.size = f32(#symexport)
o.sect[1].name = ".rdata"
o.sect[1].size = f32(#s)
o.sect[1].flags = f32(0x40300040)
o.sym1.sect = f16(2)
o.sym1.scl = 3
o.sym1.name = ".rdata"
o.sym1.naux = 1
o.sym1aux.size = f32(#s)
o.sym2.sect = f16(2)
o.sym2.scl = 2
o.sym2.nameref[1] = f32(4)
o.sym3.sect = f16(-1)
o.sym3.scl = 2
o.sym3.value = f32(1)
o.sym3.name = "@feat.00" -- Mark as SafeSEH compliant.
ffi.copy(o.space, symname)
local ofs = #symname + 1
o.strtabsize = f32(ofs + 4)
o.sect[0].ofs = f32(ffi.offsetof(o, "space") + ofs)
ffi.copy(o.space + ofs, symexport)
ofs = ofs + #symexport
o.sect[1].ofs = f32(ffi.offsetof(o, "space") + ofs)
-- Write PE object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, ffi.sizeof(o)-4096+ofs))
bcsave_tail(fp, output, s)
end
local function bcsave_machobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct
{
uint32_t magic, cputype, cpusubtype, filetype, ncmds, sizeofcmds, flags;
} mach_header;
typedef struct
{
mach_header; uint32_t reserved;
} mach_header_64;
typedef struct {
uint32_t cmd, cmdsize;
char segname[16];
uint32_t vmaddr, vmsize, fileoff, filesize;
uint32_t maxprot, initprot, nsects, flags;
} mach_segment_command;
typedef struct {
uint32_t cmd, cmdsize;
char segname[16];
uint64_t vmaddr, vmsize, fileoff, filesize;
uint32_t maxprot, initprot, nsects, flags;
} mach_segment_command_64;
typedef struct {
char sectname[16], segname[16];
uint32_t addr, size;
uint32_t offset, align, reloff, nreloc, flags;
uint32_t reserved1, reserved2;
} mach_section;
typedef struct {
char sectname[16], segname[16];
uint64_t addr, size;
uint32_t offset, align, reloff, nreloc, flags;
uint32_t reserved1, reserved2, reserved3;
} mach_section_64;
typedef struct {
uint32_t cmd, cmdsize, symoff, nsyms, stroff, strsize;
} mach_symtab_command;
typedef struct {
int32_t strx;
uint8_t type, sect;
int16_t desc;
uint32_t value;
} mach_nlist;
typedef struct {
uint32_t strx;
uint8_t type, sect;
uint16_t desc;
uint64_t value;
} mach_nlist_64;
typedef struct
{
uint32_t magic, nfat_arch;
} mach_fat_header;
typedef struct
{
uint32_t cputype, cpusubtype, offset, size, align;
} mach_fat_arch;
typedef struct {
struct {
mach_header hdr;
mach_segment_command seg;
mach_section sec;
mach_symtab_command sym;
} arch[1];
mach_nlist sym_entry;
uint8_t space[4096];
} mach_obj;
typedef struct {
struct {
mach_header_64 hdr;
mach_segment_command_64 seg;
mach_section_64 sec;
mach_symtab_command sym;
} arch[1];
mach_nlist_64 sym_entry;
uint8_t space[4096];
} mach_obj_64;
typedef struct {
mach_fat_header fat;
mach_fat_arch fat_arch[2];
struct {
mach_header hdr;
mach_segment_command seg;
mach_section sec;
mach_symtab_command sym;
} arch[2];
mach_nlist sym_entry;
uint8_t space[4096];
} mach_fat_obj;
]]
local symname = '_'..LJBC_PREFIX..ctx.modname
local isfat, is64, align, mobj = false, false, 4, "mach_obj"
if ctx.arch == "x64" then
is64, align, mobj = true, 8, "mach_obj_64"
elseif ctx.arch == "arm" then
isfat, mobj = true, "mach_fat_obj"
elseif ctx.arch == "arm64" then
is64, align, isfat, mobj = true, 8, true, "mach_fat_obj"
else
check(ctx.arch == "x86", "unsupported architecture for OSX")
end
local function aligned(v, a) return bit.band(v+a-1, -a) end
local be32 = bit.bswap -- Mach-O FAT is BE, supported archs are LE.
-- Create Mach-O object and fill in header.
local o = ffi.new(mobj)
local mach_size = aligned(ffi.offsetof(o, "space")+#symname+2, align)
local cputype = ({ x86={7}, x64={0x01000007}, arm={7,12}, arm64={0x01000007,0x0100000c} })[ctx.arch]
local cpusubtype = ({ x86={3}, x64={3}, arm={3,9}, arm64={3,0} })[ctx.arch]
if isfat then
o.fat.magic = be32(0xcafebabe)
o.fat.nfat_arch = be32(#cpusubtype)
end
-- Fill in sections and symbols.
for i=0,#cpusubtype-1 do
local ofs = 0
if isfat then
local a = o.fat_arch[i]
a.cputype = be32(cputype[i+1])
a.cpusubtype = be32(cpusubtype[i+1])
-- Subsequent slices overlap each other to share data.
ofs = ffi.offsetof(o, "arch") + i*ffi.sizeof(o.arch[0])
a.offset = be32(ofs)
a.size = be32(mach_size-ofs+#s)
end
local a = o.arch[i]
a.hdr.magic = is64 and 0xfeedfacf or 0xfeedface
a.hdr.cputype = cputype[i+1]
a.hdr.cpusubtype = cpusubtype[i+1]
a.hdr.filetype = 1
a.hdr.ncmds = 2
a.hdr.sizeofcmds = ffi.sizeof(a.seg)+ffi.sizeof(a.sec)+ffi.sizeof(a.sym)
a.seg.cmd = is64 and 0x19 or 0x1
a.seg.cmdsize = ffi.sizeof(a.seg)+ffi.sizeof(a.sec)
a.seg.vmsize = #s
a.seg.fileoff = mach_size-ofs
a.seg.filesize = #s
a.seg.maxprot = 1
a.seg.initprot = 1
a.seg.nsects = 1
ffi.copy(a.sec.sectname, "__data")
ffi.copy(a.sec.segname, "__DATA")
a.sec.size = #s
a.sec.offset = mach_size-ofs
a.sym.cmd = 2
a.sym.cmdsize = ffi.sizeof(a.sym)
a.sym.symoff = ffi.offsetof(o, "sym_entry")-ofs
a.sym.nsyms = 1
a.sym.stroff = ffi.offsetof(o, "sym_entry")+ffi.sizeof(o.sym_entry)-ofs
a.sym.strsize = aligned(#symname+2, align)
end
o.sym_entry.type = 0xf
o.sym_entry.sect = 1
o.sym_entry.strx = 1
ffi.copy(o.space+1, symname)
-- Write Macho-O object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, mach_size))
bcsave_tail(fp, output, s)
end
local function bcsave_obj(ctx, output, s)
local ok, ffi = pcall(require, "ffi")
check(ok, "FFI library required to write this file type")
if ctx.os == "windows" then
return bcsave_peobj(ctx, output, s, ffi)
elseif ctx.os == "osx" then
return bcsave_machobj(ctx, output, s, ffi)
else
return bcsave_elfobj(ctx, output, s, ffi)
end
end
------------------------------------------------------------------------------
local function bclist(input, output)
local f = readfile(input)
require("jit.bc").dump(f, savefile(output, "w"), true)
end
local function bcsave(ctx, input, output)
local f = readfile(input)
local s = string.dump(f, ctx.strip)
local t = ctx.type
if not t then
t = detecttype(output)
ctx.type = t
end
if t == "raw" then
bcsave_raw(output, s)
else
if not ctx.modname then ctx.modname = detectmodname(input) end
if t == "obj" then
bcsave_obj(ctx, output, s)
else
bcsave_c(ctx, output, s)
end
end
end
local function docmd(...)
local arg = {...}
local n = 1
local list = false
local ctx = {
strip = true, arch = jit.arch, os = jit.os:lower(),
type = false, modname = false,
}
while n <= #arg do
local a = arg[n]
if type(a) == "string" and a:sub(1, 1) == "-" and a ~= "-" then
tremove(arg, n)
if a == "--" then break end
for m=2,#a do
local opt = a:sub(m, m)
if opt == "l" then
list = true
elseif opt == "s" then
ctx.strip = true
elseif opt == "g" then
ctx.strip = false
else
if arg[n] == nil or m ~= #a then usage() end
if opt == "e" then
if n ~= 1 then usage() end
arg[1] = check(loadstring(arg[1]))
elseif opt == "n" then
ctx.modname = checkmodname(tremove(arg, n))
elseif opt == "t" then
ctx.type = checkarg(tremove(arg, n), map_type, "file type")
elseif opt == "a" then
ctx.arch = checkarg(tremove(arg, n), map_arch, "architecture")
elseif opt == "o" then
ctx.os = checkarg(tremove(arg, n), map_os, "OS name")
else
usage()
end
end
end
else
n = n + 1
end
end
if list then
if #arg == 0 or #arg > 2 then usage() end
bclist(arg[1], arg[2] or "-")
else
if #arg ~= 2 then usage() end
bcsave(ctx, arg[1], arg[2])
end
end
------------------------------------------------------------------------------
-- Public module functions.
return {
start = docmd -- Process -b command line option.
}
@@ -0,0 +1,689 @@
----------------------------------------------------------------------------
-- LuaJIT ARM disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles most user-mode ARMv7 instructions
-- NYI: Advanced SIMD and VFP instructions.
------------------------------------------------------------------------------
local type = type
local sub, byte, format = string.sub, string.byte, string.format
local match, gmatch = string.match, string.gmatch
local concat = table.concat
local bit = require("bit")
local band, bor, ror, tohex = bit.band, bit.bor, bit.ror, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Opcode maps
------------------------------------------------------------------------------
local map_loadc = {
shift = 8, mask = 15,
[10] = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "vmovFmDN", "vstmFNdr",
_ = {
shift = 21, mask = 1,
[0] = "vstrFdl",
{ shift = 16, mask = 15, [13] = "vpushFdr", _ = "vstmdbFNdr", }
},
},
{
shift = 23, mask = 3,
[0] = "vmovFDNm",
{ shift = 16, mask = 15, [13] = "vpopFdr", _ = "vldmFNdr", },
_ = {
shift = 21, mask = 1,
[0] = "vldrFdl", "vldmdbFNdr",
},
},
},
[11] = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "vmovGmDN", "vstmGNdr",
_ = {
shift = 21, mask = 1,
[0] = "vstrGdl",
{ shift = 16, mask = 15, [13] = "vpushGdr", _ = "vstmdbGNdr", }
},
},
{
shift = 23, mask = 3,
[0] = "vmovGDNm",
{ shift = 16, mask = 15, [13] = "vpopGdr", _ = "vldmGNdr", },
_ = {
shift = 21, mask = 1,
[0] = "vldrGdl", "vldmdbGNdr",
},
},
},
_ = {
shift = 0, mask = 0 -- NYI ldc, mcrr, mrrc.
},
}
local map_vfps = {
shift = 6, mask = 0x2c001,
[0] = "vmlaF.dnm", "vmlsF.dnm",
[0x04000] = "vnmlsF.dnm", [0x04001] = "vnmlaF.dnm",
[0x08000] = "vmulF.dnm", [0x08001] = "vnmulF.dnm",
[0x0c000] = "vaddF.dnm", [0x0c001] = "vsubF.dnm",
[0x20000] = "vdivF.dnm",
[0x24000] = "vfnmsF.dnm", [0x24001] = "vfnmaF.dnm",
[0x28000] = "vfmaF.dnm", [0x28001] = "vfmsF.dnm",
[0x2c000] = "vmovF.dY",
[0x2c001] = {
shift = 7, mask = 0x1e01,
[0] = "vmovF.dm", "vabsF.dm",
[0x0200] = "vnegF.dm", [0x0201] = "vsqrtF.dm",
[0x0800] = "vcmpF.dm", [0x0801] = "vcmpeF.dm",
[0x0a00] = "vcmpzF.d", [0x0a01] = "vcmpzeF.d",
[0x0e01] = "vcvtG.dF.m",
[0x1000] = "vcvt.f32.u32Fdm", [0x1001] = "vcvt.f32.s32Fdm",
[0x1800] = "vcvtr.u32F.dm", [0x1801] = "vcvt.u32F.dm",
[0x1a00] = "vcvtr.s32F.dm", [0x1a01] = "vcvt.s32F.dm",
},
}
local map_vfpd = {
shift = 6, mask = 0x2c001,
[0] = "vmlaG.dnm", "vmlsG.dnm",
[0x04000] = "vnmlsG.dnm", [0x04001] = "vnmlaG.dnm",
[0x08000] = "vmulG.dnm", [0x08001] = "vnmulG.dnm",
[0x0c000] = "vaddG.dnm", [0x0c001] = "vsubG.dnm",
[0x20000] = "vdivG.dnm",
[0x24000] = "vfnmsG.dnm", [0x24001] = "vfnmaG.dnm",
[0x28000] = "vfmaG.dnm", [0x28001] = "vfmsG.dnm",
[0x2c000] = "vmovG.dY",
[0x2c001] = {
shift = 7, mask = 0x1e01,
[0] = "vmovG.dm", "vabsG.dm",
[0x0200] = "vnegG.dm", [0x0201] = "vsqrtG.dm",
[0x0800] = "vcmpG.dm", [0x0801] = "vcmpeG.dm",
[0x0a00] = "vcmpzG.d", [0x0a01] = "vcmpzeG.d",
[0x0e01] = "vcvtF.dG.m",
[0x1000] = "vcvt.f64.u32GdFm", [0x1001] = "vcvt.f64.s32GdFm",
[0x1800] = "vcvtr.u32FdG.m", [0x1801] = "vcvt.u32FdG.m",
[0x1a00] = "vcvtr.s32FdG.m", [0x1a01] = "vcvt.s32FdG.m",
},
}
local map_datac = {
shift = 24, mask = 1,
[0] = {
shift = 4, mask = 1,
[0] = {
shift = 8, mask = 15,
[10] = map_vfps,
[11] = map_vfpd,
-- NYI cdp, mcr, mrc.
},
{
shift = 8, mask = 15,
[10] = {
shift = 20, mask = 15,
[0] = "vmovFnD", "vmovFDn",
[14] = "vmsrD",
[15] = { shift = 12, mask = 15, [15] = "vmrs", _ = "vmrsD", },
},
},
},
"svcT",
}
local map_loadcu = {
shift = 0, mask = 0, -- NYI unconditional CP load/store.
}
local map_datacu = {
shift = 0, mask = 0, -- NYI unconditional CP data.
}
local map_simddata = {
shift = 0, mask = 0, -- NYI SIMD data.
}
local map_simdload = {
shift = 0, mask = 0, -- NYI SIMD load/store, preload.
}
local map_preload = {
shift = 0, mask = 0, -- NYI preload.
}
local map_media = {
shift = 20, mask = 31,
[0] = false,
{ --01
shift = 5, mask = 7,
[0] = "sadd16DNM", "sasxDNM", "ssaxDNM", "ssub16DNM",
"sadd8DNM", false, false, "ssub8DNM",
},
{ --02
shift = 5, mask = 7,
[0] = "qadd16DNM", "qasxDNM", "qsaxDNM", "qsub16DNM",
"qadd8DNM", false, false, "qsub8DNM",
},
{ --03
shift = 5, mask = 7,
[0] = "shadd16DNM", "shasxDNM", "shsaxDNM", "shsub16DNM",
"shadd8DNM", false, false, "shsub8DNM",
},
false,
{ --05
shift = 5, mask = 7,
[0] = "uadd16DNM", "uasxDNM", "usaxDNM", "usub16DNM",
"uadd8DNM", false, false, "usub8DNM",
},
{ --06
shift = 5, mask = 7,
[0] = "uqadd16DNM", "uqasxDNM", "uqsaxDNM", "uqsub16DNM",
"uqadd8DNM", false, false, "uqsub8DNM",
},
{ --07
shift = 5, mask = 7,
[0] = "uhadd16DNM", "uhasxDNM", "uhsaxDNM", "uhsub16DNM",
"uhadd8DNM", false, false, "uhsub8DNM",
},
{ --08
shift = 5, mask = 7,
[0] = "pkhbtDNMU", false, "pkhtbDNMU",
{ shift = 16, mask = 15, [15] = "sxtb16DMU", _ = "sxtab16DNMU", },
"pkhbtDNMU", "selDNM", "pkhtbDNMU",
},
false,
{ --0a
shift = 5, mask = 7,
[0] = "ssatDxMu", "ssat16DxM", "ssatDxMu",
{ shift = 16, mask = 15, [15] = "sxtbDMU", _ = "sxtabDNMU", },
"ssatDxMu", false, "ssatDxMu",
},
{ --0b
shift = 5, mask = 7,
[0] = "ssatDxMu", "revDM", "ssatDxMu",
{ shift = 16, mask = 15, [15] = "sxthDMU", _ = "sxtahDNMU", },
"ssatDxMu", "rev16DM", "ssatDxMu",
},
{ --0c
shift = 5, mask = 7,
[3] = { shift = 16, mask = 15, [15] = "uxtb16DMU", _ = "uxtab16DNMU", },
},
false,
{ --0e
shift = 5, mask = 7,
[0] = "usatDwMu", "usat16DwM", "usatDwMu",
{ shift = 16, mask = 15, [15] = "uxtbDMU", _ = "uxtabDNMU", },
"usatDwMu", false, "usatDwMu",
},
{ --0f
shift = 5, mask = 7,
[0] = "usatDwMu", "rbitDM", "usatDwMu",
{ shift = 16, mask = 15, [15] = "uxthDMU", _ = "uxtahDNMU", },
"usatDwMu", "revshDM", "usatDwMu",
},
{ --10
shift = 12, mask = 15,
[15] = {
shift = 5, mask = 7,
"smuadNMS", "smuadxNMS", "smusdNMS", "smusdxNMS",
},
_ = {
shift = 5, mask = 7,
[0] = "smladNMSD", "smladxNMSD", "smlsdNMSD", "smlsdxNMSD",
},
},
false, false, false,
{ --14
shift = 5, mask = 7,
[0] = "smlaldDNMS", "smlaldxDNMS", "smlsldDNMS", "smlsldxDNMS",
},
{ --15
shift = 5, mask = 7,
[0] = { shift = 12, mask = 15, [15] = "smmulNMS", _ = "smmlaNMSD", },
{ shift = 12, mask = 15, [15] = "smmulrNMS", _ = "smmlarNMSD", },
false, false, false, false,
"smmlsNMSD", "smmlsrNMSD",
},
false, false,
{ --18
shift = 5, mask = 7,
[0] = { shift = 12, mask = 15, [15] = "usad8NMS", _ = "usada8NMSD", },
},
false,
{ --1a
shift = 5, mask = 3, [2] = "sbfxDMvw",
},
{ --1b
shift = 5, mask = 3, [2] = "sbfxDMvw",
},
{ --1c
shift = 5, mask = 3,
[0] = { shift = 0, mask = 15, [15] = "bfcDvX", _ = "bfiDMvX", },
},
{ --1d
shift = 5, mask = 3,
[0] = { shift = 0, mask = 15, [15] = "bfcDvX", _ = "bfiDMvX", },
},
{ --1e
shift = 5, mask = 3, [2] = "ubfxDMvw",
},
{ --1f
shift = 5, mask = 3, [2] = "ubfxDMvw",
},
}
local map_load = {
shift = 21, mask = 9,
{
shift = 20, mask = 5,
[0] = "strtDL", "ldrtDL", [4] = "strbtDL", [5] = "ldrbtDL",
},
_ = {
shift = 20, mask = 5,
[0] = "strDL", "ldrDL", [4] = "strbDL", [5] = "ldrbDL",
}
}
local map_load1 = {
shift = 4, mask = 1,
[0] = map_load, map_media,
}
local map_loadm = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "stmdaNR", "stmNR",
{ shift = 16, mask = 63, [45] = "pushR", _ = "stmdbNR", }, "stmibNR",
},
{
shift = 23, mask = 3,
[0] = "ldmdaNR", { shift = 16, mask = 63, [61] = "popR", _ = "ldmNR", },
"ldmdbNR", "ldmibNR",
},
}
local map_data = {
shift = 21, mask = 15,
[0] = "andDNPs", "eorDNPs", "subDNPs", "rsbDNPs",
"addDNPs", "adcDNPs", "sbcDNPs", "rscDNPs",
"tstNP", "teqNP", "cmpNP", "cmnNP",
"orrDNPs", "movDPs", "bicDNPs", "mvnDPs",
}
local map_mul = {
shift = 21, mask = 7,
[0] = "mulNMSs", "mlaNMSDs", "umaalDNMS", "mlsDNMS",
"umullDNMSs", "umlalDNMSs", "smullDNMSs", "smlalDNMSs",
}
local map_sync = {
shift = 20, mask = 15, -- NYI: brackets around N. R(D+1) for ldrexd/strexd.
[0] = "swpDMN", false, false, false,
"swpbDMN", false, false, false,
"strexDMN", "ldrexDN", "strexdDN", "ldrexdDN",
"strexbDMN", "ldrexbDN", "strexhDN", "ldrexhDN",
}
local map_mulh = {
shift = 21, mask = 3,
[0] = { shift = 5, mask = 3,
[0] = "smlabbNMSD", "smlatbNMSD", "smlabtNMSD", "smlattNMSD", },
{ shift = 5, mask = 3,
[0] = "smlawbNMSD", "smulwbNMS", "smlawtNMSD", "smulwtNMS", },
{ shift = 5, mask = 3,
[0] = "smlalbbDNMS", "smlaltbDNMS", "smlalbtDNMS", "smlalttDNMS", },
{ shift = 5, mask = 3,
[0] = "smulbbNMS", "smultbNMS", "smulbtNMS", "smulttNMS", },
}
local map_misc = {
shift = 4, mask = 7,
-- NYI: decode PSR bits of msr.
[0] = { shift = 21, mask = 1, [0] = "mrsD", "msrM", },
{ shift = 21, mask = 3, "bxM", false, "clzDM", },
{ shift = 21, mask = 3, "bxjM", },
{ shift = 21, mask = 3, "blxM", },
false,
{ shift = 21, mask = 3, [0] = "qaddDMN", "qsubDMN", "qdaddDMN", "qdsubDMN", },
false,
{ shift = 21, mask = 3, "bkptK", },
}
local map_datar = {
shift = 4, mask = 9,
[9] = {
shift = 5, mask = 3,
[0] = { shift = 24, mask = 1, [0] = map_mul, map_sync, },
{ shift = 20, mask = 1, [0] = "strhDL", "ldrhDL", },
{ shift = 20, mask = 1, [0] = "ldrdDL", "ldrsbDL", },
{ shift = 20, mask = 1, [0] = "strdDL", "ldrshDL", },
},
_ = {
shift = 20, mask = 25,
[16] = { shift = 7, mask = 1, [0] = map_misc, map_mulh, },
_ = {
shift = 0, mask = 0xffffffff,
[bor(0xe1a00000)] = "nop",
_ = map_data,
}
},
}
local map_datai = {
shift = 20, mask = 31, -- NYI: decode PSR bits of msr. Decode imm12.
[16] = "movwDW", [20] = "movtDW",
[18] = { shift = 0, mask = 0xf00ff, [0] = "nopv6", _ = "msrNW", },
[22] = "msrNW",
_ = map_data,
}
local map_branch = {
shift = 24, mask = 1,
[0] = "bB", "blB"
}
local map_condins = {
[0] = map_datar, map_datai, map_load, map_load1,
map_loadm, map_branch, map_loadc, map_datac
}
-- NYI: setend.
local map_uncondins = {
[0] = false, map_simddata, map_simdload, map_preload,
false, "blxB", map_loadcu, map_datacu,
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "sp", "lr", "pc",
}
local map_cond = {
[0] = "eq", "ne", "hs", "lo", "mi", "pl", "vs", "vc",
"hi", "ls", "ge", "lt", "gt", "le", "al",
}
local map_shift = { [0] = "lsl", "lsr", "asr", "ror", }
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then
extra = "\t->"..sym
elseif band(ctx.op, 0x0e000000) ~= 0x0a000000 then
extra = "\t; 0x"..tohex(ctx.rel)
end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-5s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-5s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
-- Format operand 2 of load/store opcodes.
local function fmtload(ctx, op, pos)
local base = map_gpr[band(rshift(op, 16), 15)]
local x, ofs
local ext = (band(op, 0x04000000) == 0)
if not ext and band(op, 0x02000000) == 0 then
ofs = band(op, 4095)
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
ofs = "#"..ofs
elseif ext and band(op, 0x00400000) ~= 0 then
ofs = band(op, 15) + band(rshift(op, 4), 0xf0)
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
ofs = "#"..ofs
else
ofs = map_gpr[band(op, 15)]
if ext or band(op, 0xfe0) == 0 then
elseif band(op, 0xfe0) == 0x60 then
ofs = format("%s, rrx", ofs)
else
local sh = band(rshift(op, 7), 31)
if sh == 0 then sh = 32 end
ofs = format("%s, %s #%d", ofs, map_shift[band(rshift(op, 5), 3)], sh)
end
if band(op, 0x00800000) == 0 then ofs = "-"..ofs end
end
if ofs == "#0" then
x = format("[%s]", base)
elseif band(op, 0x01000000) == 0 then
x = format("[%s], %s", base, ofs)
else
x = format("[%s, %s]", base, ofs)
end
if band(op, 0x01200000) == 0x01200000 then x = x.."!" end
return x
end
-- Format operand 2 of vector load/store opcodes.
local function fmtvload(ctx, op, pos)
local base = map_gpr[band(rshift(op, 16), 15)]
local ofs = band(op, 255)*4
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
if ofs == 0 then
return format("[%s]", base)
else
return format("[%s, #%d]", base, ofs)
end
end
local function fmtvr(op, vr, sh0, sh1)
if vr == "s" then
return format("s%d", 2*band(rshift(op, sh0), 15)+band(rshift(op, sh1), 1))
else
return format("d%d", band(rshift(op, sh0), 15)+band(rshift(op, sh1-4), 16))
end
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
local op = bor(lshift(b3, 24), lshift(b2, 16), lshift(b1, 8), b0)
local operands = {}
local suffix = ""
local last, name, pat
local vr
ctx.op = op
ctx.rel = nil
local cond = rshift(op, 28)
local opat
if cond == 15 then
opat = map_uncondins[band(rshift(op, 25), 7)]
else
if cond ~= 14 then suffix = map_cond[cond] end
opat = map_condins[band(rshift(op, 25), 7)]
end
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
opat = opat[band(rshift(op, opat.shift), opat.mask)] or opat._
end
name, pat = match(opat, "^([a-z0-9]*)(.*)")
if sub(pat, 1, 1) == "." then
local s2, p2 = match(pat, "^([a-z0-9.]*)(.*)")
suffix = suffix..s2
pat = p2
end
for p in gmatch(pat, ".") do
local x = nil
if p == "D" then
x = map_gpr[band(rshift(op, 12), 15)]
elseif p == "N" then
x = map_gpr[band(rshift(op, 16), 15)]
elseif p == "S" then
x = map_gpr[band(rshift(op, 8), 15)]
elseif p == "M" then
x = map_gpr[band(op, 15)]
elseif p == "d" then
x = fmtvr(op, vr, 12, 22)
elseif p == "n" then
x = fmtvr(op, vr, 16, 7)
elseif p == "m" then
x = fmtvr(op, vr, 0, 5)
elseif p == "P" then
if band(op, 0x02000000) ~= 0 then
x = ror(band(op, 255), 2*band(rshift(op, 8), 15))
else
x = map_gpr[band(op, 15)]
if band(op, 0xff0) ~= 0 then
operands[#operands+1] = x
local s = map_shift[band(rshift(op, 5), 3)]
local r = nil
if band(op, 0xf90) == 0 then
if s == "ror" then s = "rrx" else r = "#32" end
elseif band(op, 0x10) == 0 then
r = "#"..band(rshift(op, 7), 31)
else
r = map_gpr[band(rshift(op, 8), 15)]
end
if name == "mov" then name = s; x = r
elseif r then x = format("%s %s", s, r)
else x = s end
end
end
elseif p == "L" then
x = fmtload(ctx, op, pos)
elseif p == "l" then
x = fmtvload(ctx, op, pos)
elseif p == "B" then
local addr = ctx.addr + pos + 8 + arshift(lshift(op, 8), 6)
if cond == 15 then addr = addr + band(rshift(op, 23), 2) end
ctx.rel = addr
x = "0x"..tohex(addr)
elseif p == "F" then
vr = "s"
elseif p == "G" then
vr = "d"
elseif p == "." then
suffix = suffix..(vr == "s" and ".f32" or ".f64")
elseif p == "R" then
if band(op, 0x00200000) ~= 0 and #operands == 1 then
operands[1] = operands[1].."!"
end
local t = {}
for i=0,15 do
if band(rshift(op, i), 1) == 1 then t[#t+1] = map_gpr[i] end
end
x = "{"..concat(t, ", ").."}"
elseif p == "r" then
if band(op, 0x00200000) ~= 0 and #operands == 2 then
operands[1] = operands[1].."!"
end
local s = tonumber(sub(last, 2))
local n = band(op, 255)
if vr == "d" then n = rshift(n, 1) end
operands[#operands] = format("{%s-%s%d}", last, vr, s+n-1)
elseif p == "W" then
x = band(op, 0x0fff) + band(rshift(op, 4), 0xf000)
elseif p == "T" then
x = "#0x"..tohex(band(op, 0x00ffffff), 6)
elseif p == "U" then
x = band(rshift(op, 7), 31)
if x == 0 then x = nil end
elseif p == "u" then
x = band(rshift(op, 7), 31)
if band(op, 0x40) == 0 then
if x == 0 then x = nil else x = "lsl #"..x end
else
if x == 0 then x = "asr #32" else x = "asr #"..x end
end
elseif p == "v" then
x = band(rshift(op, 7), 31)
elseif p == "w" then
x = band(rshift(op, 16), 31)
elseif p == "x" then
x = band(rshift(op, 16), 31) + 1
elseif p == "X" then
x = band(rshift(op, 16), 31) - last + 1
elseif p == "Y" then
x = band(rshift(op, 12), 0xf0) + band(op, 0x0f)
elseif p == "K" then
x = "#0x"..tohex(band(rshift(op, 4), 0x0000fff0) + band(op, 15), 4)
elseif p == "s" then
if band(op, 0x00100000) ~= 0 then suffix = "s"..suffix end
else
assert(false)
end
if x then
last = x
if type(x) == "number" then x = "#"..x end
operands[#operands+1] = x
end
end
return putop(ctx, name..suffix, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
ctx.pos = ofs
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 16 then return map_gpr[r] end
return "d"..(r-16)
end
-- Public module functions.
return {
create = create,
disass = disass,
regname = regname
}
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,12 @@
----------------------------------------------------------------------------
-- LuaJIT ARM64BE disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- ARM64 instructions are always little-endian. So just forward to the
-- common ARM64 disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
return require((string.match(..., ".*%.") or "").."dis_arm64")
@@ -0,0 +1,694 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT/X license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles all standard MIPS32R1/R2 instructions.
-- Default mode is big-endian, but see: dis_mipsel.lua
------------------------------------------------------------------------------
local type = type
local byte, format = string.byte, string.format
local match, gmatch = string.match, string.gmatch
local concat = table.concat
local bit = require("bit")
local band, bor, tohex = bit.band, bit.bor, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Extended opcode maps common to all MIPS releases
------------------------------------------------------------------------------
local map_srl = { shift = 21, mask = 1, [0] = "srlDTA", "rotrDTA", }
local map_srlv = { shift = 6, mask = 1, [0] = "srlvDTS", "rotrvDTS", }
local map_cop0 = {
shift = 25, mask = 1,
[0] = {
shift = 21, mask = 15,
[0] = "mfc0TDW", [4] = "mtc0TDW",
[10] = "rdpgprDT",
[11] = { shift = 5, mask = 1, [0] = "diT0", "eiT0", },
[14] = "wrpgprDT",
}, {
shift = 0, mask = 63,
[1] = "tlbr", [2] = "tlbwi", [6] = "tlbwr", [8] = "tlbp",
[24] = "eret", [31] = "deret",
[32] = "wait",
},
}
------------------------------------------------------------------------------
-- Primary and extended opcode maps for MIPS R1-R5
------------------------------------------------------------------------------
local map_movci = { shift = 16, mask = 1, [0] = "movfDSC", "movtDSC", }
local map_special = {
shift = 0, mask = 63,
[0] = { shift = 0, mask = -1, [0] = "nop", _ = "sllDTA" },
map_movci, map_srl, "sraDTA",
"sllvDTS", false, map_srlv, "sravDTS",
"jrS", "jalrD1S", "movzDST", "movnDST",
"syscallY", "breakY", false, "sync",
"mfhiD", "mthiS", "mfloD", "mtloS",
"dsllvDST", false, "dsrlvDST", "dsravDST",
"multST", "multuST", "divST", "divuST",
"dmultST", "dmultuST", "ddivST", "ddivuST",
"addDST", "addu|moveDST0", "subDST", "subu|neguDS0T",
"andDST", "or|moveDST0", "xorDST", "nor|notDST0",
false, false, "sltDST", "sltuDST",
"daddDST", "dadduDST", "dsubDST", "dsubuDST",
"tgeSTZ", "tgeuSTZ", "tltSTZ", "tltuSTZ",
"teqSTZ", false, "tneSTZ", false,
"dsllDTA", false, "dsrlDTA", "dsraDTA",
"dsll32DTA", false, "dsrl32DTA", "dsra32DTA",
}
local map_special2 = {
shift = 0, mask = 63,
[0] = "maddST", "madduST", "mulDST", false,
"msubST", "msubuST",
[32] = "clzDS", [33] = "cloDS",
[63] = "sdbbpY",
}
local map_bshfl = {
shift = 6, mask = 31,
[2] = "wsbhDT",
[16] = "sebDT",
[24] = "sehDT",
}
local map_dbshfl = {
shift = 6, mask = 31,
[2] = "dsbhDT",
[5] = "dshdDT",
}
local map_special3 = {
shift = 0, mask = 63,
[0] = "extTSAK", [1] = "dextmTSAP", [3] = "dextTSAK",
[4] = "insTSAL", [6] = "dinsuTSEQ", [7] = "dinsTSAL",
[32] = map_bshfl, [36] = map_dbshfl, [59] = "rdhwrTD",
}
local map_regimm = {
shift = 16, mask = 31,
[0] = "bltzSB", "bgezSB", "bltzlSB", "bgezlSB",
false, false, false, false,
"tgeiSI", "tgeiuSI", "tltiSI", "tltiuSI",
"teqiSI", false, "tneiSI", false,
"bltzalSB", "bgezalSB", "bltzallSB", "bgezallSB",
false, false, false, false,
false, false, false, false,
false, false, false, "synciSO",
}
local map_cop1s = {
shift = 0, mask = 63,
[0] = "add.sFGH", "sub.sFGH", "mul.sFGH", "div.sFGH",
"sqrt.sFG", "abs.sFG", "mov.sFG", "neg.sFG",
"round.l.sFG", "trunc.l.sFG", "ceil.l.sFG", "floor.l.sFG",
"round.w.sFG", "trunc.w.sFG", "ceil.w.sFG", "floor.w.sFG",
false,
{ shift = 16, mask = 1, [0] = "movf.sFGC", "movt.sFGC" },
"movz.sFGT", "movn.sFGT",
false, "recip.sFG", "rsqrt.sFG", false,
false, false, false, false,
false, false, false, false,
false, "cvt.d.sFG", false, false,
"cvt.w.sFG", "cvt.l.sFG", "cvt.ps.sFGH", false,
false, false, false, false,
false, false, false, false,
"c.f.sVGH", "c.un.sVGH", "c.eq.sVGH", "c.ueq.sVGH",
"c.olt.sVGH", "c.ult.sVGH", "c.ole.sVGH", "c.ule.sVGH",
"c.sf.sVGH", "c.ngle.sVGH", "c.seq.sVGH", "c.ngl.sVGH",
"c.lt.sVGH", "c.nge.sVGH", "c.le.sVGH", "c.ngt.sVGH",
}
local map_cop1d = {
shift = 0, mask = 63,
[0] = "add.dFGH", "sub.dFGH", "mul.dFGH", "div.dFGH",
"sqrt.dFG", "abs.dFG", "mov.dFG", "neg.dFG",
"round.l.dFG", "trunc.l.dFG", "ceil.l.dFG", "floor.l.dFG",
"round.w.dFG", "trunc.w.dFG", "ceil.w.dFG", "floor.w.dFG",
false,
{ shift = 16, mask = 1, [0] = "movf.dFGC", "movt.dFGC" },
"movz.dFGT", "movn.dFGT",
false, "recip.dFG", "rsqrt.dFG", false,
false, false, false, false,
false, false, false, false,
"cvt.s.dFG", false, false, false,
"cvt.w.dFG", "cvt.l.dFG", false, false,
false, false, false, false,
false, false, false, false,
"c.f.dVGH", "c.un.dVGH", "c.eq.dVGH", "c.ueq.dVGH",
"c.olt.dVGH", "c.ult.dVGH", "c.ole.dVGH", "c.ule.dVGH",
"c.df.dVGH", "c.ngle.dVGH", "c.deq.dVGH", "c.ngl.dVGH",
"c.lt.dVGH", "c.nge.dVGH", "c.le.dVGH", "c.ngt.dVGH",
}
local map_cop1ps = {
shift = 0, mask = 63,
[0] = "add.psFGH", "sub.psFGH", "mul.psFGH", false,
false, "abs.psFG", "mov.psFG", "neg.psFG",
false, false, false, false,
false, false, false, false,
false,
{ shift = 16, mask = 1, [0] = "movf.psFGC", "movt.psFGC" },
"movz.psFGT", "movn.psFGT",
false, false, false, false,
false, false, false, false,
false, false, false, false,
"cvt.s.puFG", false, false, false,
false, false, false, false,
"cvt.s.plFG", false, false, false,
"pll.psFGH", "plu.psFGH", "pul.psFGH", "puu.psFGH",
"c.f.psVGH", "c.un.psVGH", "c.eq.psVGH", "c.ueq.psVGH",
"c.olt.psVGH", "c.ult.psVGH", "c.ole.psVGH", "c.ule.psVGH",
"c.psf.psVGH", "c.ngle.psVGH", "c.pseq.psVGH", "c.ngl.psVGH",
"c.lt.psVGH", "c.nge.psVGH", "c.le.psVGH", "c.ngt.psVGH",
}
local map_cop1w = {
shift = 0, mask = 63,
[32] = "cvt.s.wFG", [33] = "cvt.d.wFG",
}
local map_cop1l = {
shift = 0, mask = 63,
[32] = "cvt.s.lFG", [33] = "cvt.d.lFG",
}
local map_cop1bc = {
shift = 16, mask = 3,
[0] = "bc1fCB", "bc1tCB", "bc1flCB", "bc1tlCB",
}
local map_cop1 = {
shift = 21, mask = 31,
[0] = "mfc1TG", "dmfc1TG", "cfc1TG", "mfhc1TG",
"mtc1TG", "dmtc1TG", "ctc1TG", "mthc1TG",
map_cop1bc, false, false, false,
false, false, false, false,
map_cop1s, map_cop1d, false, false,
map_cop1w, map_cop1l, map_cop1ps,
}
local map_cop1x = {
shift = 0, mask = 63,
[0] = "lwxc1FSX", "ldxc1FSX", false, false,
false, "luxc1FSX", false, false,
"swxc1FSX", "sdxc1FSX", false, false,
false, "suxc1FSX", false, "prefxMSX",
false, false, false, false,
false, false, false, false,
false, false, false, false,
false, false, "alnv.psFGHS", false,
"madd.sFRGH", "madd.dFRGH", false, false,
false, false, "madd.psFRGH", false,
"msub.sFRGH", "msub.dFRGH", false, false,
false, false, "msub.psFRGH", false,
"nmadd.sFRGH", "nmadd.dFRGH", false, false,
false, false, "nmadd.psFRGH", false,
"nmsub.sFRGH", "nmsub.dFRGH", false, false,
false, false, "nmsub.psFRGH", false,
}
local map_pri = {
[0] = map_special, map_regimm, "jJ", "jalJ",
"beq|beqz|bST00B", "bne|bnezST0B", "blezSB", "bgtzSB",
"addiTSI", "addiu|liTS0I", "sltiTSI", "sltiuTSI",
"andiTSU", "ori|liTS0U", "xoriTSU", "luiTU",
map_cop0, map_cop1, false, map_cop1x,
"beql|beqzlST0B", "bnel|bnezlST0B", "blezlSB", "bgtzlSB",
"daddiTSI", "daddiuTSI", false, false,
map_special2, "jalxJ", false, map_special3,
"lbTSO", "lhTSO", "lwlTSO", "lwTSO",
"lbuTSO", "lhuTSO", "lwrTSO", false,
"sbTSO", "shTSO", "swlTSO", "swTSO",
false, false, "swrTSO", "cacheNSO",
"llTSO", "lwc1HSO", "lwc2TSO", "prefNSO",
false, "ldc1HSO", "ldc2TSO", "ldTSO",
"scTSO", "swc1HSO", "swc2TSO", false,
false, "sdc1HSO", "sdc2TSO", "sdTSO",
}
------------------------------------------------------------------------------
-- Primary and extended opcode maps for MIPS R6
------------------------------------------------------------------------------
local map_mul_r6 = { shift = 6, mask = 3, [2] = "mulDST", [3] = "muhDST" }
local map_mulu_r6 = { shift = 6, mask = 3, [2] = "muluDST", [3] = "muhuDST" }
local map_div_r6 = { shift = 6, mask = 3, [2] = "divDST", [3] = "modDST" }
local map_divu_r6 = { shift = 6, mask = 3, [2] = "divuDST", [3] = "moduDST" }
local map_dmul_r6 = { shift = 6, mask = 3, [2] = "dmulDST", [3] = "dmuhDST" }
local map_dmulu_r6 = { shift = 6, mask = 3, [2] = "dmuluDST", [3] = "dmuhuDST" }
local map_ddiv_r6 = { shift = 6, mask = 3, [2] = "ddivDST", [3] = "dmodDST" }
local map_ddivu_r6 = { shift = 6, mask = 3, [2] = "ddivuDST", [3] = "dmoduDST" }
local map_special_r6 = {
shift = 0, mask = 63,
[0] = { shift = 0, mask = -1, [0] = "nop", _ = "sllDTA" },
false, map_srl, "sraDTA",
"sllvDTS", false, map_srlv, "sravDTS",
"jrS", "jalrD1S", false, false,
"syscallY", "breakY", false, "sync",
"clzDS", "cloDS", "dclzDS", "dcloDS",
"dsllvDST", "dlsaDSTA", "dsrlvDST", "dsravDST",
map_mul_r6, map_mulu_r6, map_div_r6, map_divu_r6,
map_dmul_r6, map_dmulu_r6, map_ddiv_r6, map_ddivu_r6,
"addDST", "addu|moveDST0", "subDST", "subu|neguDS0T",
"andDST", "or|moveDST0", "xorDST", "nor|notDST0",
false, false, "sltDST", "sltuDST",
"daddDST", "dadduDST", "dsubDST", "dsubuDST",
"tgeSTZ", "tgeuSTZ", "tltSTZ", "tltuSTZ",
"teqSTZ", "seleqzDST", "tneSTZ", "selnezDST",
"dsllDTA", false, "dsrlDTA", "dsraDTA",
"dsll32DTA", false, "dsrl32DTA", "dsra32DTA",
}
local map_bshfl_r6 = {
shift = 9, mask = 3,
[1] = "alignDSTa",
_ = {
shift = 6, mask = 31,
[0] = "bitswapDT",
[2] = "wsbhDT",
[16] = "sebDT",
[24] = "sehDT",
}
}
local map_dbshfl_r6 = {
shift = 9, mask = 3,
[1] = "dalignDSTa",
_ = {
shift = 6, mask = 31,
[0] = "dbitswapDT",
[2] = "dsbhDT",
[5] = "dshdDT",
}
}
local map_special3_r6 = {
shift = 0, mask = 63,
[0] = "extTSAK", [1] = "dextmTSAP", [3] = "dextTSAK",
[4] = "insTSAL", [6] = "dinsuTSEQ", [7] = "dinsTSAL",
[32] = map_bshfl_r6, [36] = map_dbshfl_r6, [59] = "rdhwrTD",
}
local map_regimm_r6 = {
shift = 16, mask = 31,
[0] = "bltzSB", [1] = "bgezSB",
[6] = "dahiSI", [30] = "datiSI",
[23] = "sigrieI", [31] = "synciSO",
}
local map_pcrel_r6 = {
shift = 19, mask = 3,
[0] = "addiupcS2", "lwpcS2", "lwupcS2", {
shift = 18, mask = 1,
[0] = "ldpcS3", { shift = 16, mask = 3, [2] = "auipcSI", [3] = "aluipcSI" }
}
}
local map_cop1s_r6 = {
shift = 0, mask = 63,
[0] = "add.sFGH", "sub.sFGH", "mul.sFGH", "div.sFGH",
"sqrt.sFG", "abs.sFG", "mov.sFG", "neg.sFG",
"round.l.sFG", "trunc.l.sFG", "ceil.l.sFG", "floor.l.sFG",
"round.w.sFG", "trunc.w.sFG", "ceil.w.sFG", "floor.w.sFG",
"sel.sFGH", false, false, false,
"seleqz.sFGH", "recip.sFG", "rsqrt.sFG", "selnez.sFGH",
"maddf.sFGH", "msubf.sFGH", "rint.sFG", "class.sFG",
"min.sFGH", "mina.sFGH", "max.sFGH", "maxa.sFGH",
false, "cvt.d.sFG", false, false,
"cvt.w.sFG", "cvt.l.sFG",
}
local map_cop1d_r6 = {
shift = 0, mask = 63,
[0] = "add.dFGH", "sub.dFGH", "mul.dFGH", "div.dFGH",
"sqrt.dFG", "abs.dFG", "mov.dFG", "neg.dFG",
"round.l.dFG", "trunc.l.dFG", "ceil.l.dFG", "floor.l.dFG",
"round.w.dFG", "trunc.w.dFG", "ceil.w.dFG", "floor.w.dFG",
"sel.dFGH", false, false, false,
"seleqz.dFGH", "recip.dFG", "rsqrt.dFG", "selnez.dFGH",
"maddf.dFGH", "msubf.dFGH", "rint.dFG", "class.dFG",
"min.dFGH", "mina.dFGH", "max.dFGH", "maxa.dFGH",
"cvt.s.dFG", false, false, false,
"cvt.w.dFG", "cvt.l.dFG",
}
local map_cop1w_r6 = {
shift = 0, mask = 63,
[0] = "cmp.af.sFGH", "cmp.un.sFGH", "cmp.eq.sFGH", "cmp.ueq.sFGH",
"cmp.lt.sFGH", "cmp.ult.sFGH", "cmp.le.sFGH", "cmp.ule.sFGH",
"cmp.saf.sFGH", "cmp.sun.sFGH", "cmp.seq.sFGH", "cmp.sueq.sFGH",
"cmp.slt.sFGH", "cmp.sult.sFGH", "cmp.sle.sFGH", "cmp.sule.sFGH",
false, "cmp.or.sFGH", "cmp.une.sFGH", "cmp.ne.sFGH",
false, false, false, false,
false, "cmp.sor.sFGH", "cmp.sune.sFGH", "cmp.sne.sFGH",
false, false, false, false,
"cvt.s.wFG", "cvt.d.wFG",
}
local map_cop1l_r6 = {
shift = 0, mask = 63,
[0] = "cmp.af.dFGH", "cmp.un.dFGH", "cmp.eq.dFGH", "cmp.ueq.dFGH",
"cmp.lt.dFGH", "cmp.ult.dFGH", "cmp.le.dFGH", "cmp.ule.dFGH",
"cmp.saf.dFGH", "cmp.sun.dFGH", "cmp.seq.dFGH", "cmp.sueq.dFGH",
"cmp.slt.dFGH", "cmp.sult.dFGH", "cmp.sle.dFGH", "cmp.sule.dFGH",
false, "cmp.or.dFGH", "cmp.une.dFGH", "cmp.ne.dFGH",
false, false, false, false,
false, "cmp.sor.dFGH", "cmp.sune.dFGH", "cmp.sne.dFGH",
false, false, false, false,
"cvt.s.lFG", "cvt.d.lFG",
}
local map_cop1_r6 = {
shift = 21, mask = 31,
[0] = "mfc1TG", "dmfc1TG", "cfc1TG", "mfhc1TG",
"mtc1TG", "dmtc1TG", "ctc1TG", "mthc1TG",
false, "bc1eqzHB", false, false,
false, "bc1nezHB", false, false,
map_cop1s_r6, map_cop1d_r6, false, false,
map_cop1w_r6, map_cop1l_r6,
}
local function maprs_popTS(rs, rt)
if rt == 0 then return 0 elseif rs == 0 then return 1
elseif rs == rt then return 2 else return 3 end
end
local map_pop06_r6 = {
maprs = maprs_popTS, [0] = "blezSB", "blezalcTB", "bgezalcTB", "bgeucSTB"
}
local map_pop07_r6 = {
maprs = maprs_popTS, [0] = "bgtzSB", "bgtzalcTB", "bltzalcTB", "bltucSTB"
}
local map_pop26_r6 = {
maprs = maprs_popTS, "blezcTB", "bgezcTB", "bgecSTB"
}
local map_pop27_r6 = {
maprs = maprs_popTS, "bgtzcTB", "bltzcTB", "bltcSTB"
}
local function maprs_popS(rs, rt)
if rs == 0 then return 0 else return 1 end
end
local map_pop66_r6 = {
maprs = maprs_popS, [0] = "jicTI", "beqzcSb"
}
local map_pop76_r6 = {
maprs = maprs_popS, [0] = "jialcTI", "bnezcSb"
}
local function maprs_popST(rs, rt)
if rs >= rt then return 0 elseif rs == 0 then return 1 else return 2 end
end
local map_pop10_r6 = {
maprs = maprs_popST, [0] = "bovcSTB", "beqzalcTB", "beqcSTB"
}
local map_pop30_r6 = {
maprs = maprs_popST, [0] = "bnvcSTB", "bnezalcTB", "bnecSTB"
}
local map_pri_r6 = {
[0] = map_special_r6, map_regimm_r6, "jJ", "jalJ",
"beq|beqz|bST00B", "bne|bnezST0B", map_pop06_r6, map_pop07_r6,
map_pop10_r6, "addiu|liTS0I", "sltiTSI", "sltiuTSI",
"andiTSU", "ori|liTS0U", "xoriTSU", "aui|luiTS0U",
map_cop0, map_cop1_r6, false, false,
false, false, map_pop26_r6, map_pop27_r6,
map_pop30_r6, "daddiuTSI", false, false,
false, "dauiTSI", false, map_special3_r6,
"lbTSO", "lhTSO", false, "lwTSO",
"lbuTSO", "lhuTSO", false, false,
"sbTSO", "shTSO", false, "swTSO",
false, false, false, false,
false, "lwc1HSO", "bc#", false,
false, "ldc1HSO", map_pop66_r6, "ldTSO",
false, "swc1HSO", "balc#", map_pcrel_r6,
false, "sdc1HSO", map_pop76_r6, "sdTSO",
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
"r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
"r24", "r25", "r26", "r27", "r28", "sp", "r30", "ra",
}
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then extra = "\t->"..sym end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-7s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-7s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
local function get_be(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
return bor(lshift(b0, 24), lshift(b1, 16), lshift(b2, 8), b3)
end
local function get_le(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
return bor(lshift(b3, 24), lshift(b2, 16), lshift(b1, 8), b0)
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local op = ctx:get()
local operands = {}
local last = nil
ctx.op = op
ctx.rel = nil
local opat = ctx.map_pri[rshift(op, 26)]
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
if opat.maprs then
opat = opat[opat.maprs(band(rshift(op,21),31), band(rshift(op,16),31))]
else
opat = opat[band(rshift(op, opat.shift), opat.mask)] or opat._
end
end
local name, pat = match(opat, "^([a-z0-9_.]*)(.*)")
local altname, pat2 = match(pat, "|([a-z0-9_.|]*)(.*)")
if altname then pat = pat2 end
for p in gmatch(pat, ".") do
local x = nil
if p == "S" then
x = map_gpr[band(rshift(op, 21), 31)]
elseif p == "T" then
x = map_gpr[band(rshift(op, 16), 31)]
elseif p == "D" then
x = map_gpr[band(rshift(op, 11), 31)]
elseif p == "F" then
x = "f"..band(rshift(op, 6), 31)
elseif p == "G" then
x = "f"..band(rshift(op, 11), 31)
elseif p == "H" then
x = "f"..band(rshift(op, 16), 31)
elseif p == "R" then
x = "f"..band(rshift(op, 21), 31)
elseif p == "A" then
x = band(rshift(op, 6), 31)
elseif p == "a" then
x = band(rshift(op, 6), 7)
elseif p == "E" then
x = band(rshift(op, 6), 31) + 32
elseif p == "M" then
x = band(rshift(op, 11), 31)
elseif p == "N" then
x = band(rshift(op, 16), 31)
elseif p == "C" then
x = band(rshift(op, 18), 7)
if x == 0 then x = nil end
elseif p == "K" then
x = band(rshift(op, 11), 31) + 1
elseif p == "P" then
x = band(rshift(op, 11), 31) + 33
elseif p == "L" then
x = band(rshift(op, 11), 31) - last + 1
elseif p == "Q" then
x = band(rshift(op, 11), 31) - last + 33
elseif p == "I" then
x = arshift(lshift(op, 16), 16)
elseif p == "2" then
x = arshift(lshift(op, 13), 11)
elseif p == "3" then
x = arshift(lshift(op, 14), 11)
elseif p == "U" then
x = band(op, 0xffff)
elseif p == "O" then
local disp = arshift(lshift(op, 16), 16)
operands[#operands] = format("%d(%s)", disp, last)
elseif p == "X" then
local index = map_gpr[band(rshift(op, 16), 31)]
operands[#operands] = format("%s(%s)", index, last)
elseif p == "B" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 16), 14) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "b" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 11), 9) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "#" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 6), 4) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "J" then
local a = ctx.addr + ctx.pos
x = a - band(a, 0x0fffffff) + band(op, 0x03ffffff)*4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "V" then
x = band(rshift(op, 8), 7)
if x == 0 then x = nil end
elseif p == "W" then
x = band(op, 7)
if x == 0 then x = nil end
elseif p == "Y" then
x = band(rshift(op, 6), 0x000fffff)
if x == 0 then x = nil end
elseif p == "Z" then
x = band(rshift(op, 6), 1023)
if x == 0 then x = nil end
elseif p == "0" then
if last == "r0" or last == 0 then
local n = #operands
operands[n] = nil
last = operands[n-1]
if altname then
local a1, a2 = match(altname, "([^|]*)|(.*)")
if a1 then name, altname = a1, a2
else name = altname end
end
end
elseif p == "1" then
if last == "ra" then
operands[#operands] = nil
end
else
assert(false)
end
if x then operands[#operands+1] = x; last = x end
end
return putop(ctx, name, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
stop = stop - stop % 4
ctx.pos = ofs - ofs % 4
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
ctx.get = get_be
ctx.map_pri = map_pri
return ctx
end
local function create_el(code, addr, out)
local ctx = create(code, addr, out)
ctx.get = get_le
return ctx
end
local function create_r6(code, addr, out)
local ctx = create(code, addr, out)
ctx.map_pri = map_pri_r6
return ctx
end
local function create_r6_el(code, addr, out)
local ctx = create(code, addr, out)
ctx.get = get_le
ctx.map_pri = map_pri_r6
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
local function disass_el(code, addr, out)
create_el(code, addr, out):disass()
end
local function disass_r6(code, addr, out)
create_r6(code, addr, out):disass()
end
local function disass_r6_el(code, addr, out)
create_r6_el(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 32 then return map_gpr[r] end
return "f"..(r-32)
end
-- Public module functions.
return {
create = create,
create_el = create_el,
create_r6 = create_r6,
create_r6_el = create_r6_el,
disass = disass,
disass_el = disass_el,
disass_r6 = disass_r6,
disass_r6_el = disass_r6_el,
regname = regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the big-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create,
disass = dis_mips.disass,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64EL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_el,
disass = dis_mips.disass_el,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64R6 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the r6 big-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_r6,
disass = dis_mips.disass_r6,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64R6EL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the r6 little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_r6_el,
disass = dis_mips.disass_r6_el,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPSEL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_el,
disass = dis_mips.disass_el,
regname = dis_mips.regname
}
@@ -0,0 +1,591 @@
----------------------------------------------------------------------------
-- LuaJIT PPC disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT/X license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles all common, non-privileged 32/64 bit PowerPC instructions
-- plus the e500 SPE instructions and some Cell/Xenon extensions.
--
-- NYI: VMX, VMX128
------------------------------------------------------------------------------
local type = type
local byte, format = string.byte, string.format
local match, gmatch, gsub = string.match, string.gmatch, string.gsub
local concat = table.concat
local bit = require("bit")
local band, bor, tohex = bit.band, bit.bor, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Primary and extended opcode maps
------------------------------------------------------------------------------
local map_crops = {
shift = 1, mask = 1023,
[0] = "mcrfXX",
[33] = "crnor|crnotCCC=", [129] = "crandcCCC",
[193] = "crxor|crclrCCC%", [225] = "crnandCCC",
[257] = "crandCCC", [289] = "creqv|crsetCCC%",
[417] = "crorcCCC", [449] = "cror|crmoveCCC=",
[16] = "b_lrKB", [528] = "b_ctrKB",
[150] = "isync",
}
local map_rlwinm = setmetatable({
shift = 0, mask = -1,
},
{ __index = function(t, x)
local rot = band(rshift(x, 11), 31)
local mb = band(rshift(x, 6), 31)
local me = band(rshift(x, 1), 31)
if mb == 0 and me == 31-rot then
return "slwiRR~A."
elseif me == 31 and mb == 32-rot then
return "srwiRR~-A."
else
return "rlwinmRR~AAA."
end
end
})
local map_rld = {
shift = 2, mask = 7,
[0] = "rldiclRR~HM.", "rldicrRR~HM.", "rldicRR~HM.", "rldimiRR~HM.",
{
shift = 1, mask = 1,
[0] = "rldclRR~RM.", "rldcrRR~RM.",
},
}
local map_ext = setmetatable({
shift = 1, mask = 1023,
[0] = "cmp_YLRR", [32] = "cmpl_YLRR",
[4] = "twARR", [68] = "tdARR",
[8] = "subfcRRR.", [40] = "subfRRR.",
[104] = "negRR.", [136] = "subfeRRR.",
[200] = "subfzeRR.", [232] = "subfmeRR.",
[520] = "subfcoRRR.", [552] = "subfoRRR.",
[616] = "negoRR.", [648] = "subfeoRRR.",
[712] = "subfzeoRR.", [744] = "subfmeoRR.",
[9] = "mulhduRRR.", [73] = "mulhdRRR.", [233] = "mulldRRR.",
[457] = "divduRRR.", [489] = "divdRRR.",
[745] = "mulldoRRR.",
[969] = "divduoRRR.", [1001] = "divdoRRR.",
[10] = "addcRRR.", [138] = "addeRRR.",
[202] = "addzeRR.", [234] = "addmeRR.", [266] = "addRRR.",
[522] = "addcoRRR.", [650] = "addeoRRR.",
[714] = "addzeoRR.", [746] = "addmeoRR.", [778] = "addoRRR.",
[11] = "mulhwuRRR.", [75] = "mulhwRRR.", [235] = "mullwRRR.",
[459] = "divwuRRR.", [491] = "divwRRR.",
[747] = "mullwoRRR.",
[971] = "divwouRRR.", [1003] = "divwoRRR.",
[15] = "iselltRRR", [47] = "iselgtRRR", [79] = "iseleqRRR",
[144] = { shift = 20, mask = 1, [0] = "mtcrfRZ~", "mtocrfRZ~", },
[19] = { shift = 20, mask = 1, [0] = "mfcrR", "mfocrfRZ", },
[371] = { shift = 11, mask = 1023, [392] = "mftbR", [424] = "mftbuR", },
[339] = {
shift = 11, mask = 1023,
[32] = "mferR", [256] = "mflrR", [288] = "mfctrR", [16] = "mfspefscrR",
},
[467] = {
shift = 11, mask = 1023,
[32] = "mtxerR", [256] = "mtlrR", [288] = "mtctrR", [16] = "mtspefscrR",
},
[20] = "lwarxRR0R", [84] = "ldarxRR0R",
[21] = "ldxRR0R", [53] = "lduxRRR",
[149] = "stdxRR0R", [181] = "stduxRRR",
[341] = "lwaxRR0R", [373] = "lwauxRRR",
[23] = "lwzxRR0R", [55] = "lwzuxRRR",
[87] = "lbzxRR0R", [119] = "lbzuxRRR",
[151] = "stwxRR0R", [183] = "stwuxRRR",
[215] = "stbxRR0R", [247] = "stbuxRRR",
[279] = "lhzxRR0R", [311] = "lhzuxRRR",
[343] = "lhaxRR0R", [375] = "lhauxRRR",
[407] = "sthxRR0R", [439] = "sthuxRRR",
[54] = "dcbst-R0R", [86] = "dcbf-R0R",
[150] = "stwcxRR0R.", [214] = "stdcxRR0R.",
[246] = "dcbtst-R0R", [278] = "dcbt-R0R",
[310] = "eciwxRR0R", [438] = "ecowxRR0R",
[470] = "dcbi-RR",
[598] = {
shift = 21, mask = 3,
[0] = "sync", "lwsync", "ptesync",
},
[758] = "dcba-RR",
[854] = "eieio", [982] = "icbi-R0R", [1014] = "dcbz-R0R",
[26] = "cntlzwRR~", [58] = "cntlzdRR~",
[122] = "popcntbRR~",
[154] = "prtywRR~", [186] = "prtydRR~",
[28] = "andRR~R.", [60] = "andcRR~R.", [124] = "nor|notRR~R=.",
[284] = "eqvRR~R.", [316] = "xorRR~R.",
[412] = "orcRR~R.", [444] = "or|mrRR~R=.", [476] = "nandRR~R.",
[508] = "cmpbRR~R",
[512] = "mcrxrX",
[532] = "ldbrxRR0R", [660] = "stdbrxRR0R",
[533] = "lswxRR0R", [597] = "lswiRR0A",
[661] = "stswxRR0R", [725] = "stswiRR0A",
[534] = "lwbrxRR0R", [662] = "stwbrxRR0R",
[790] = "lhbrxRR0R", [918] = "sthbrxRR0R",
[535] = "lfsxFR0R", [567] = "lfsuxFRR",
[599] = "lfdxFR0R", [631] = "lfduxFRR",
[663] = "stfsxFR0R", [695] = "stfsuxFRR",
[727] = "stfdxFR0R", [759] = "stfduxFR0R",
[855] = "lfiwaxFR0R",
[983] = "stfiwxFR0R",
[24] = "slwRR~R.",
[27] = "sldRR~R.", [536] = "srwRR~R.",
[792] = "srawRR~R.", [824] = "srawiRR~A.",
[794] = "sradRR~R.", [826] = "sradiRR~H.", [827] = "sradiRR~H.",
[922] = "extshRR~.", [954] = "extsbRR~.", [986] = "extswRR~.",
[539] = "srdRR~R.",
},
{ __index = function(t, x)
if band(x, 31) == 15 then return "iselRRRC" end
end
})
local map_ld = {
shift = 0, mask = 3,
[0] = "ldRRE", "lduRRE", "lwaRRE",
}
local map_std = {
shift = 0, mask = 3,
[0] = "stdRRE", "stduRRE",
}
local map_fps = {
shift = 5, mask = 1,
{
shift = 1, mask = 15,
[0] = false, false, "fdivsFFF.", false,
"fsubsFFF.", "faddsFFF.", "fsqrtsF-F.", false,
"fresF-F.", "fmulsFF-F.", "frsqrtesF-F.", false,
"fmsubsFFFF~.", "fmaddsFFFF~.", "fnmsubsFFFF~.", "fnmaddsFFFF~.",
}
}
local map_fpd = {
shift = 5, mask = 1,
[0] = {
shift = 1, mask = 1023,
[0] = "fcmpuXFF", [32] = "fcmpoXFF", [64] = "mcrfsXX",
[38] = "mtfsb1A.", [70] = "mtfsb0A.", [134] = "mtfsfiA>>-A>",
[8] = "fcpsgnFFF.", [40] = "fnegF-F.", [72] = "fmrF-F.",
[136] = "fnabsF-F.", [264] = "fabsF-F.",
[12] = "frspF-F.",
[14] = "fctiwF-F.", [15] = "fctiwzF-F.",
[583] = "mffsF.", [711] = "mtfsfZF.",
[392] = "frinF-F.", [424] = "frizF-F.",
[456] = "fripF-F.", [488] = "frimF-F.",
[814] = "fctidF-F.", [815] = "fctidzF-F.", [846] = "fcfidF-F.",
},
{
shift = 1, mask = 15,
[0] = false, false, "fdivFFF.", false,
"fsubFFF.", "faddFFF.", "fsqrtF-F.", "fselFFFF~.",
"freF-F.", "fmulFF-F.", "frsqrteF-F.", false,
"fmsubFFFF~.", "fmaddFFFF~.", "fnmsubFFFF~.", "fnmaddFFFF~.",
}
}
local map_spe = {
shift = 0, mask = 2047,
[512] = "evaddwRRR", [514] = "evaddiwRAR~",
[516] = "evsubwRRR~", [518] = "evsubiwRAR~",
[520] = "evabsRR", [521] = "evnegRR",
[522] = "evextsbRR", [523] = "evextshRR", [524] = "evrndwRR",
[525] = "evcntlzwRR", [526] = "evcntlswRR",
[527] = "brincRRR",
[529] = "evandRRR", [530] = "evandcRRR", [534] = "evxorRRR",
[535] = "evor|evmrRRR=", [536] = "evnor|evnotRRR=",
[537] = "eveqvRRR", [539] = "evorcRRR", [542] = "evnandRRR",
[544] = "evsrwuRRR", [545] = "evsrwsRRR",
[546] = "evsrwiuRRA", [547] = "evsrwisRRA",
[548] = "evslwRRR", [550] = "evslwiRRA",
[552] = "evrlwRRR", [553] = "evsplatiRS",
[554] = "evrlwiRRA", [555] = "evsplatfiRS",
[556] = "evmergehiRRR", [557] = "evmergeloRRR",
[558] = "evmergehiloRRR", [559] = "evmergelohiRRR",
[560] = "evcmpgtuYRR", [561] = "evcmpgtsYRR",
[562] = "evcmpltuYRR", [563] = "evcmpltsYRR",
[564] = "evcmpeqYRR",
[632] = "evselRRR", [633] = "evselRRRW",
[634] = "evselRRRW", [635] = "evselRRRW",
[636] = "evselRRRW", [637] = "evselRRRW",
[638] = "evselRRRW", [639] = "evselRRRW",
[640] = "evfsaddRRR", [641] = "evfssubRRR",
[644] = "evfsabsRR", [645] = "evfsnabsRR", [646] = "evfsnegRR",
[648] = "evfsmulRRR", [649] = "evfsdivRRR",
[652] = "evfscmpgtYRR", [653] = "evfscmpltYRR", [654] = "evfscmpeqYRR",
[656] = "evfscfuiR-R", [657] = "evfscfsiR-R",
[658] = "evfscfufR-R", [659] = "evfscfsfR-R",
[660] = "evfsctuiR-R", [661] = "evfsctsiR-R",
[662] = "evfsctufR-R", [663] = "evfsctsfR-R",
[664] = "evfsctuizR-R", [666] = "evfsctsizR-R",
[668] = "evfststgtYRR", [669] = "evfststltYRR", [670] = "evfststeqYRR",
[704] = "efsaddRRR", [705] = "efssubRRR",
[708] = "efsabsRR", [709] = "efsnabsRR", [710] = "efsnegRR",
[712] = "efsmulRRR", [713] = "efsdivRRR",
[716] = "efscmpgtYRR", [717] = "efscmpltYRR", [718] = "efscmpeqYRR",
[719] = "efscfdR-R",
[720] = "efscfuiR-R", [721] = "efscfsiR-R",
[722] = "efscfufR-R", [723] = "efscfsfR-R",
[724] = "efsctuiR-R", [725] = "efsctsiR-R",
[726] = "efsctufR-R", [727] = "efsctsfR-R",
[728] = "efsctuizR-R", [730] = "efsctsizR-R",
[732] = "efststgtYRR", [733] = "efststltYRR", [734] = "efststeqYRR",
[736] = "efdaddRRR", [737] = "efdsubRRR",
[738] = "efdcfuidR-R", [739] = "efdcfsidR-R",
[740] = "efdabsRR", [741] = "efdnabsRR", [742] = "efdnegRR",
[744] = "efdmulRRR", [745] = "efddivRRR",
[746] = "efdctuidzR-R", [747] = "efdctsidzR-R",
[748] = "efdcmpgtYRR", [749] = "efdcmpltYRR", [750] = "efdcmpeqYRR",
[751] = "efdcfsR-R",
[752] = "efdcfuiR-R", [753] = "efdcfsiR-R",
[754] = "efdcfufR-R", [755] = "efdcfsfR-R",
[756] = "efdctuiR-R", [757] = "efdctsiR-R",
[758] = "efdctufR-R", [759] = "efdctsfR-R",
[760] = "efdctuizR-R", [762] = "efdctsizR-R",
[764] = "efdtstgtYRR", [765] = "efdtstltYRR", [766] = "efdtsteqYRR",
[768] = "evlddxRR0R", [769] = "evlddRR8",
[770] = "evldwxRR0R", [771] = "evldwRR8",
[772] = "evldhxRR0R", [773] = "evldhRR8",
[776] = "evlhhesplatxRR0R", [777] = "evlhhesplatRR2",
[780] = "evlhhousplatxRR0R", [781] = "evlhhousplatRR2",
[782] = "evlhhossplatxRR0R", [783] = "evlhhossplatRR2",
[784] = "evlwhexRR0R", [785] = "evlwheRR4",
[788] = "evlwhouxRR0R", [789] = "evlwhouRR4",
[790] = "evlwhosxRR0R", [791] = "evlwhosRR4",
[792] = "evlwwsplatxRR0R", [793] = "evlwwsplatRR4",
[796] = "evlwhsplatxRR0R", [797] = "evlwhsplatRR4",
[800] = "evstddxRR0R", [801] = "evstddRR8",
[802] = "evstdwxRR0R", [803] = "evstdwRR8",
[804] = "evstdhxRR0R", [805] = "evstdhRR8",
[816] = "evstwhexRR0R", [817] = "evstwheRR4",
[820] = "evstwhoxRR0R", [821] = "evstwhoRR4",
[824] = "evstwwexRR0R", [825] = "evstwweRR4",
[828] = "evstwwoxRR0R", [829] = "evstwwoRR4",
[1027] = "evmhessfRRR", [1031] = "evmhossfRRR", [1032] = "evmheumiRRR",
[1033] = "evmhesmiRRR", [1035] = "evmhesmfRRR", [1036] = "evmhoumiRRR",
[1037] = "evmhosmiRRR", [1039] = "evmhosmfRRR", [1059] = "evmhessfaRRR",
[1063] = "evmhossfaRRR", [1064] = "evmheumiaRRR", [1065] = "evmhesmiaRRR",
[1067] = "evmhesmfaRRR", [1068] = "evmhoumiaRRR", [1069] = "evmhosmiaRRR",
[1071] = "evmhosmfaRRR", [1095] = "evmwhssfRRR", [1096] = "evmwlumiRRR",
[1100] = "evmwhumiRRR", [1101] = "evmwhsmiRRR", [1103] = "evmwhsmfRRR",
[1107] = "evmwssfRRR", [1112] = "evmwumiRRR", [1113] = "evmwsmiRRR",
[1115] = "evmwsmfRRR", [1127] = "evmwhssfaRRR", [1128] = "evmwlumiaRRR",
[1132] = "evmwhumiaRRR", [1133] = "evmwhsmiaRRR", [1135] = "evmwhsmfaRRR",
[1139] = "evmwssfaRRR", [1144] = "evmwumiaRRR", [1145] = "evmwsmiaRRR",
[1147] = "evmwsmfaRRR",
[1216] = "evaddusiaawRR", [1217] = "evaddssiaawRR",
[1218] = "evsubfusiaawRR", [1219] = "evsubfssiaawRR",
[1220] = "evmraRR",
[1222] = "evdivwsRRR", [1223] = "evdivwuRRR",
[1224] = "evaddumiaawRR", [1225] = "evaddsmiaawRR",
[1226] = "evsubfumiaawRR", [1227] = "evsubfsmiaawRR",
[1280] = "evmheusiaawRRR", [1281] = "evmhessiaawRRR",
[1283] = "evmhessfaawRRR", [1284] = "evmhousiaawRRR",
[1285] = "evmhossiaawRRR", [1287] = "evmhossfaawRRR",
[1288] = "evmheumiaawRRR", [1289] = "evmhesmiaawRRR",
[1291] = "evmhesmfaawRRR", [1292] = "evmhoumiaawRRR",
[1293] = "evmhosmiaawRRR", [1295] = "evmhosmfaawRRR",
[1320] = "evmhegumiaaRRR", [1321] = "evmhegsmiaaRRR",
[1323] = "evmhegsmfaaRRR", [1324] = "evmhogumiaaRRR",
[1325] = "evmhogsmiaaRRR", [1327] = "evmhogsmfaaRRR",
[1344] = "evmwlusiaawRRR", [1345] = "evmwlssiaawRRR",
[1352] = "evmwlumiaawRRR", [1353] = "evmwlsmiaawRRR",
[1363] = "evmwssfaaRRR", [1368] = "evmwumiaaRRR",
[1369] = "evmwsmiaaRRR", [1371] = "evmwsmfaaRRR",
[1408] = "evmheusianwRRR", [1409] = "evmhessianwRRR",
[1411] = "evmhessfanwRRR", [1412] = "evmhousianwRRR",
[1413] = "evmhossianwRRR", [1415] = "evmhossfanwRRR",
[1416] = "evmheumianwRRR", [1417] = "evmhesmianwRRR",
[1419] = "evmhesmfanwRRR", [1420] = "evmhoumianwRRR",
[1421] = "evmhosmianwRRR", [1423] = "evmhosmfanwRRR",
[1448] = "evmhegumianRRR", [1449] = "evmhegsmianRRR",
[1451] = "evmhegsmfanRRR", [1452] = "evmhogumianRRR",
[1453] = "evmhogsmianRRR", [1455] = "evmhogsmfanRRR",
[1472] = "evmwlusianwRRR", [1473] = "evmwlssianwRRR",
[1480] = "evmwlumianwRRR", [1481] = "evmwlsmianwRRR",
[1491] = "evmwssfanRRR", [1496] = "evmwumianRRR",
[1497] = "evmwsmianRRR", [1499] = "evmwsmfanRRR",
}
local map_pri = {
[0] = false, false, "tdiARI", "twiARI",
map_spe, false, false, "mulliRRI",
"subficRRI", false, "cmpl_iYLRU", "cmp_iYLRI",
"addicRRI", "addic.RRI", "addi|liRR0I", "addis|lisRR0I",
"b_KBJ", "sc", "bKJ", map_crops,
"rlwimiRR~AAA.", map_rlwinm, false, "rlwnmRR~RAA.",
"oriNRR~U", "orisRR~U", "xoriRR~U", "xorisRR~U",
"andi.RR~U", "andis.RR~U", map_rld, map_ext,
"lwzRRD", "lwzuRRD", "lbzRRD", "lbzuRRD",
"stwRRD", "stwuRRD", "stbRRD", "stbuRRD",
"lhzRRD", "lhzuRRD", "lhaRRD", "lhauRRD",
"sthRRD", "sthuRRD", "lmwRRD", "stmwRRD",
"lfsFRD", "lfsuFRD", "lfdFRD", "lfduFRD",
"stfsFRD", "stfsuFRD", "stfdFRD", "stfduFRD",
false, false, map_ld, map_fps,
false, false, map_std, map_fpd,
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "sp", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
"r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
"r24", "r25", "r26", "r27", "r28", "r29", "r30", "r31",
}
local map_cond = { [0] = "lt", "gt", "eq", "so", "ge", "le", "ne", "ns", }
-- Format a condition bit.
local function condfmt(cond)
if cond <= 3 then
return map_cond[band(cond, 3)]
else
return format("4*cr%d+%s", rshift(cond, 2), map_cond[band(cond, 3)])
end
end
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then extra = "\t->"..sym end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-7s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-7s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
local op = bor(lshift(b0, 24), lshift(b1, 16), lshift(b2, 8), b3)
local operands = {}
local last = nil
local rs = 21
ctx.op = op
ctx.rel = nil
local opat = map_pri[rshift(b0, 2)]
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
opat = opat[band(rshift(op, opat.shift), opat.mask)]
end
local name, pat = match(opat, "^([a-z0-9_.]*)(.*)")
local altname, pat2 = match(pat, "|([a-z0-9_.]*)(.*)")
if altname then pat = pat2 end
for p in gmatch(pat, ".") do
local x = nil
if p == "R" then
x = map_gpr[band(rshift(op, rs), 31)]
rs = rs - 5
elseif p == "F" then
x = "f"..band(rshift(op, rs), 31)
rs = rs - 5
elseif p == "A" then
x = band(rshift(op, rs), 31)
rs = rs - 5
elseif p == "S" then
x = arshift(lshift(op, 27-rs), 27)
rs = rs - 5
elseif p == "I" then
x = arshift(lshift(op, 16), 16)
elseif p == "U" then
x = band(op, 0xffff)
elseif p == "D" or p == "E" then
local disp = arshift(lshift(op, 16), 16)
if p == "E" then disp = band(disp, -4) end
if last == "r0" then last = "0" end
operands[#operands] = format("%d(%s)", disp, last)
elseif p >= "2" and p <= "8" then
local disp = band(rshift(op, rs), 31) * p
if last == "r0" then last = "0" end
operands[#operands] = format("%d(%s)", disp, last)
elseif p == "H" then
x = band(rshift(op, rs), 31) + lshift(band(op, 2), 4)
rs = rs - 5
elseif p == "M" then
x = band(rshift(op, rs), 31) + band(op, 0x20)
elseif p == "C" then
x = condfmt(band(rshift(op, rs), 31))
rs = rs - 5
elseif p == "B" then
local bo = rshift(op, 21)
local cond = band(rshift(op, 16), 31)
local cn = ""
rs = rs - 10
if band(bo, 4) == 0 then
cn = band(bo, 2) == 0 and "dnz" or "dz"
if band(bo, 0x10) == 0 then
cn = cn..(band(bo, 8) == 0 and "f" or "t")
end
if band(bo, 0x10) == 0 then x = condfmt(cond) end
name = name..(band(bo, 1) == band(rshift(op, 15), 1) and "-" or "+")
elseif band(bo, 0x10) == 0 then
cn = map_cond[band(cond, 3) + (band(bo, 8) == 0 and 4 or 0)]
if cond > 3 then x = "cr"..rshift(cond, 2) end
name = name..(band(bo, 1) == band(rshift(op, 15), 1) and "-" or "+")
end
name = gsub(name, "_", cn)
elseif p == "J" then
x = arshift(lshift(op, 27-rs), 29-rs)*4
if band(op, 2) == 0 then x = ctx.addr + pos + x end
ctx.rel = x
x = "0x"..tohex(x)
elseif p == "K" then
if band(op, 1) ~= 0 then name = name.."l" end
if band(op, 2) ~= 0 then name = name.."a" end
elseif p == "X" or p == "Y" then
x = band(rshift(op, rs+2), 7)
if x == 0 and p == "Y" then x = nil else x = "cr"..x end
rs = rs - 5
elseif p == "W" then
x = "cr"..band(op, 7)
elseif p == "Z" then
x = band(rshift(op, rs-4), 255)
rs = rs - 10
elseif p == ">" then
operands[#operands] = rshift(operands[#operands], 1)
elseif p == "0" then
if last == "r0" then
operands[#operands] = nil
if altname then name = altname end
end
elseif p == "L" then
name = gsub(name, "_", band(op, 0x00200000) ~= 0 and "d" or "w")
elseif p == "." then
if band(op, 1) == 1 then name = name.."." end
elseif p == "N" then
if op == 0x60000000 then name = "nop"; break end
elseif p == "~" then
local n = #operands
operands[n-1], operands[n] = operands[n], operands[n-1]
elseif p == "=" then
local n = #operands
if last == operands[n-1] then
operands[n] = nil
name = altname
end
elseif p == "%" then
local n = #operands
if last == operands[n-1] and last == operands[n-2] then
operands[n] = nil
operands[n-1] = nil
name = altname
end
elseif p == "-" then
rs = rs - 5
else
assert(false)
end
if x then operands[#operands+1] = x; last = x end
end
return putop(ctx, name, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
stop = stop - stop % 4
ctx.pos = ofs - ofs % 4
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 32 then return map_gpr[r] end
return "f"..(r-32)
end
-- Public module functions.
return {
create = create,
disass = disass,
regname = regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT x64 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the 64 bit functions from the combined
-- x86/x64 disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_x86 = require((string.match(..., ".*%.") or "").."dis_x86")
return {
create = dis_x86.create64,
disass = dis_x86.disass64,
regname = dis_x86.regname64
}
@@ -0,0 +1,953 @@
----------------------------------------------------------------------------
-- LuaJIT x86/x64 disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- Sending small code snippets to an external disassembler and mixing the
-- output with our own stuff was too fragile. So I had to bite the bullet
-- and write yet another x86 disassembler. Oh well ...
--
-- The output format is very similar to what ndisasm generates. But it has
-- been developed independently by looking at the opcode tables from the
-- Intel and AMD manuals. The supported instruction set is quite extensive
-- and reflects what a current generation Intel or AMD CPU implements in
-- 32 bit and 64 bit mode. Yes, this includes MMX, SSE, SSE2, SSE3, SSSE3,
-- SSE4.1, SSE4.2, SSE4a, AVX, AVX2 and even privileged and hypervisor
-- (VMX/SVM) instructions.
--
-- Notes:
-- * The (useless) a16 prefix, 3DNow and pre-586 opcodes are unsupported.
-- * No attempt at optimization has been made -- it's fast enough for my needs.
------------------------------------------------------------------------------
local type = type
local sub, byte, format = string.sub, string.byte, string.format
local match, gmatch, gsub = string.match, string.gmatch, string.gsub
local lower, rep = string.lower, string.rep
local bit = require("bit")
local tohex = bit.tohex
-- Map for 1st opcode byte in 32 bit mode. Ugly? Well ... read on.
local map_opc1_32 = {
--0x
[0]="addBmr","addVmr","addBrm","addVrm","addBai","addVai","push es","pop es",
"orBmr","orVmr","orBrm","orVrm","orBai","orVai","push cs","opc2*",
--1x
"adcBmr","adcVmr","adcBrm","adcVrm","adcBai","adcVai","push ss","pop ss",
"sbbBmr","sbbVmr","sbbBrm","sbbVrm","sbbBai","sbbVai","push ds","pop ds",
--2x
"andBmr","andVmr","andBrm","andVrm","andBai","andVai","es:seg","daa",
"subBmr","subVmr","subBrm","subVrm","subBai","subVai","cs:seg","das",
--3x
"xorBmr","xorVmr","xorBrm","xorVrm","xorBai","xorVai","ss:seg","aaa",
"cmpBmr","cmpVmr","cmpBrm","cmpVrm","cmpBai","cmpVai","ds:seg","aas",
--4x
"incVR","incVR","incVR","incVR","incVR","incVR","incVR","incVR",
"decVR","decVR","decVR","decVR","decVR","decVR","decVR","decVR",
--5x
"pushUR","pushUR","pushUR","pushUR","pushUR","pushUR","pushUR","pushUR",
"popUR","popUR","popUR","popUR","popUR","popUR","popUR","popUR",
--6x
"sz*pushaw,pusha","sz*popaw,popa","boundVrm","arplWmr",
"fs:seg","gs:seg","o16:","a16",
"pushUi","imulVrmi","pushBs","imulVrms",
"insb","insVS","outsb","outsVS",
--7x
"joBj","jnoBj","jbBj","jnbBj","jzBj","jnzBj","jbeBj","jaBj",
"jsBj","jnsBj","jpeBj","jpoBj","jlBj","jgeBj","jleBj","jgBj",
--8x
"arith!Bmi","arith!Vmi","arith!Bmi","arith!Vms",
"testBmr","testVmr","xchgBrm","xchgVrm",
"movBmr","movVmr","movBrm","movVrm",
"movVmg","leaVrm","movWgm","popUm",
--9x
"nop*xchgVaR|pause|xchgWaR|repne nop","xchgVaR","xchgVaR","xchgVaR",
"xchgVaR","xchgVaR","xchgVaR","xchgVaR",
"sz*cbw,cwde,cdqe","sz*cwd,cdq,cqo","call farViw","wait",
"sz*pushfw,pushf","sz*popfw,popf","sahf","lahf",
--Ax
"movBao","movVao","movBoa","movVoa",
"movsb","movsVS","cmpsb","cmpsVS",
"testBai","testVai","stosb","stosVS",
"lodsb","lodsVS","scasb","scasVS",
--Bx
"movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi",
"movVRI","movVRI","movVRI","movVRI","movVRI","movVRI","movVRI","movVRI",
--Cx
"shift!Bmu","shift!Vmu","retBw","ret","vex*3$lesVrm","vex*2$ldsVrm","movBmi","movVmi",
"enterBwu","leave","retfBw","retf","int3","intBu","into","iretVS",
--Dx
"shift!Bm1","shift!Vm1","shift!Bmc","shift!Vmc","aamBu","aadBu","salc","xlatb",
"fp*0","fp*1","fp*2","fp*3","fp*4","fp*5","fp*6","fp*7",
--Ex
"loopneBj","loopeBj","loopBj","sz*jcxzBj,jecxzBj,jrcxzBj",
"inBau","inVau","outBua","outVua",
"callVj","jmpVj","jmp farViw","jmpBj","inBad","inVad","outBda","outVda",
--Fx
"lock:","int1","repne:rep","rep:","hlt","cmc","testb!Bm","testv!Vm",
"clc","stc","cli","sti","cld","std","incb!Bm","incd!Vm",
}
assert(#map_opc1_32 == 255)
-- Map for 1st opcode byte in 64 bit mode (overrides only).
local map_opc1_64 = setmetatable({
[0x06]=false, [0x07]=false, [0x0e]=false,
[0x16]=false, [0x17]=false, [0x1e]=false, [0x1f]=false,
[0x27]=false, [0x2f]=false, [0x37]=false, [0x3f]=false,
[0x60]=false, [0x61]=false, [0x62]=false, [0x63]="movsxdVrDmt", [0x67]="a32:",
[0x40]="rex*", [0x41]="rex*b", [0x42]="rex*x", [0x43]="rex*xb",
[0x44]="rex*r", [0x45]="rex*rb", [0x46]="rex*rx", [0x47]="rex*rxb",
[0x48]="rex*w", [0x49]="rex*wb", [0x4a]="rex*wx", [0x4b]="rex*wxb",
[0x4c]="rex*wr", [0x4d]="rex*wrb", [0x4e]="rex*wrx", [0x4f]="rex*wrxb",
[0x82]=false, [0x9a]=false, [0xc4]="vex*3", [0xc5]="vex*2", [0xce]=false,
[0xd4]=false, [0xd5]=false, [0xd6]=false, [0xea]=false,
}, { __index = map_opc1_32 })
-- Map for 2nd opcode byte (0F xx). True CISC hell. Hey, I told you.
-- Prefix dependent MMX/SSE opcodes: (none)|rep|o16|repne, -|F3|66|F2
local map_opc2 = {
--0x
[0]="sldt!Dmp","sgdt!Ump","larVrm","lslVrm",nil,"syscall","clts","sysret",
"invd","wbinvd",nil,"ud1",nil,"$prefetch!Bm","femms","3dnowMrmu",
--1x
"movupsXrm|movssXrvm|movupdXrm|movsdXrvm",
"movupsXmr|movssXmvr|movupdXmr|movsdXmvr",
"movhlpsXrm$movlpsXrm|movsldupXrm|movlpdXrm|movddupXrm",
"movlpsXmr||movlpdXmr",
"unpcklpsXrvm||unpcklpdXrvm",
"unpckhpsXrvm||unpckhpdXrvm",
"movlhpsXrm$movhpsXrm|movshdupXrm|movhpdXrm",
"movhpsXmr||movhpdXmr",
"$prefetcht!Bm","hintnopVm","hintnopVm","hintnopVm",
"hintnopVm","hintnopVm","hintnopVm","hintnopVm",
--2x
"movUmx$","movUmy$","movUxm$","movUym$","movUmz$",nil,"movUzm$",nil,
"movapsXrm||movapdXrm",
"movapsXmr||movapdXmr",
"cvtpi2psXrMm|cvtsi2ssXrvVmt|cvtpi2pdXrMm|cvtsi2sdXrvVmt",
"movntpsXmr|movntssXmr|movntpdXmr|movntsdXmr",
"cvttps2piMrXm|cvttss2siVrXm|cvttpd2piMrXm|cvttsd2siVrXm",
"cvtps2piMrXm|cvtss2siVrXm|cvtpd2piMrXm|cvtsd2siVrXm",
"ucomissXrm||ucomisdXrm",
"comissXrm||comisdXrm",
--3x
"wrmsr","rdtsc","rdmsr","rdpmc","sysenter","sysexit",nil,"getsec",
"opc3*38",nil,"opc3*3a",nil,nil,nil,nil,nil,
--4x
"cmovoVrm","cmovnoVrm","cmovbVrm","cmovnbVrm",
"cmovzVrm","cmovnzVrm","cmovbeVrm","cmovaVrm",
"cmovsVrm","cmovnsVrm","cmovpeVrm","cmovpoVrm",
"cmovlVrm","cmovgeVrm","cmovleVrm","cmovgVrm",
--5x
"movmskpsVrXm$||movmskpdVrXm$","sqrtpsXrm|sqrtssXrm|sqrtpdXrm|sqrtsdXrm",
"rsqrtpsXrm|rsqrtssXrvm","rcppsXrm|rcpssXrvm",
"andpsXrvm||andpdXrvm","andnpsXrvm||andnpdXrvm",
"orpsXrvm||orpdXrvm","xorpsXrvm||xorpdXrvm",
"addpsXrvm|addssXrvm|addpdXrvm|addsdXrvm","mulpsXrvm|mulssXrvm|mulpdXrvm|mulsdXrvm",
"cvtps2pdXrm|cvtss2sdXrvm|cvtpd2psXrm|cvtsd2ssXrvm",
"cvtdq2psXrm|cvttps2dqXrm|cvtps2dqXrm",
"subpsXrvm|subssXrvm|subpdXrvm|subsdXrvm","minpsXrvm|minssXrvm|minpdXrvm|minsdXrvm",
"divpsXrvm|divssXrvm|divpdXrvm|divsdXrvm","maxpsXrvm|maxssXrvm|maxpdXrvm|maxsdXrvm",
--6x
"punpcklbwPrvm","punpcklwdPrvm","punpckldqPrvm","packsswbPrvm",
"pcmpgtbPrvm","pcmpgtwPrvm","pcmpgtdPrvm","packuswbPrvm",
"punpckhbwPrvm","punpckhwdPrvm","punpckhdqPrvm","packssdwPrvm",
"||punpcklqdqXrvm","||punpckhqdqXrvm",
"movPrVSm","movqMrm|movdquXrm|movdqaXrm",
--7x
"pshufwMrmu|pshufhwXrmu|pshufdXrmu|pshuflwXrmu","pshiftw!Pvmu",
"pshiftd!Pvmu","pshiftq!Mvmu||pshiftdq!Xvmu",
"pcmpeqbPrvm","pcmpeqwPrvm","pcmpeqdPrvm","emms*|",
"vmreadUmr||extrqXmuu$|insertqXrmuu$","vmwriteUrm||extrqXrm$|insertqXrm$",
nil,nil,
"||haddpdXrvm|haddpsXrvm","||hsubpdXrvm|hsubpsXrvm",
"movVSmMr|movqXrm|movVSmXr","movqMmr|movdquXmr|movdqaXmr",
--8x
"joVj","jnoVj","jbVj","jnbVj","jzVj","jnzVj","jbeVj","jaVj",
"jsVj","jnsVj","jpeVj","jpoVj","jlVj","jgeVj","jleVj","jgVj",
--9x
"setoBm","setnoBm","setbBm","setnbBm","setzBm","setnzBm","setbeBm","setaBm",
"setsBm","setnsBm","setpeBm","setpoBm","setlBm","setgeBm","setleBm","setgBm",
--Ax
"push fs","pop fs","cpuid","btVmr","shldVmru","shldVmrc",nil,nil,
"push gs","pop gs","rsm","btsVmr","shrdVmru","shrdVmrc","fxsave!Dmp","imulVrm",
--Bx
"cmpxchgBmr","cmpxchgVmr","$lssVrm","btrVmr",
"$lfsVrm","$lgsVrm","movzxVrBmt","movzxVrWmt",
"|popcntVrm","ud2Dp","bt!Vmu","btcVmr",
"bsfVrm","bsrVrm|lzcntVrm|bsrWrm","movsxVrBmt","movsxVrWmt",
--Cx
"xaddBmr","xaddVmr",
"cmppsXrvmu|cmpssXrvmu|cmppdXrvmu|cmpsdXrvmu","$movntiVmr|",
"pinsrwPrvWmu","pextrwDrPmu",
"shufpsXrvmu||shufpdXrvmu","$cmpxchg!Qmp",
"bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR",
--Dx
"||addsubpdXrvm|addsubpsXrvm","psrlwPrvm","psrldPrvm","psrlqPrvm",
"paddqPrvm","pmullwPrvm",
"|movq2dqXrMm|movqXmr|movdq2qMrXm$","pmovmskbVrMm||pmovmskbVrXm",
"psubusbPrvm","psubuswPrvm","pminubPrvm","pandPrvm",
"paddusbPrvm","padduswPrvm","pmaxubPrvm","pandnPrvm",
--Ex
"pavgbPrvm","psrawPrvm","psradPrvm","pavgwPrvm",
"pmulhuwPrvm","pmulhwPrvm",
"|cvtdq2pdXrm|cvttpd2dqXrm|cvtpd2dqXrm","$movntqMmr||$movntdqXmr",
"psubsbPrvm","psubswPrvm","pminswPrvm","porPrvm",
"paddsbPrvm","paddswPrvm","pmaxswPrvm","pxorPrvm",
--Fx
"|||lddquXrm","psllwPrvm","pslldPrvm","psllqPrvm",
"pmuludqPrvm","pmaddwdPrvm","psadbwPrvm","maskmovqMrm||maskmovdquXrm$",
"psubbPrvm","psubwPrvm","psubdPrvm","psubqPrvm",
"paddbPrvm","paddwPrvm","padddPrvm","ud",
}
assert(map_opc2[255] == "ud")
-- Map for three-byte opcodes. Can't wait for their next invention.
local map_opc3 = {
["38"] = { -- [66] 0f 38 xx
--0x
[0]="pshufbPrvm","phaddwPrvm","phadddPrvm","phaddswPrvm",
"pmaddubswPrvm","phsubwPrvm","phsubdPrvm","phsubswPrvm",
"psignbPrvm","psignwPrvm","psigndPrvm","pmulhrswPrvm",
"||permilpsXrvm","||permilpdXrvm",nil,nil,
--1x
"||pblendvbXrma",nil,nil,nil,
"||blendvpsXrma","||blendvpdXrma","||permpsXrvm","||ptestXrm",
"||broadcastssXrm","||broadcastsdXrm","||broadcastf128XrlXm",nil,
"pabsbPrm","pabswPrm","pabsdPrm",nil,
--2x
"||pmovsxbwXrm","||pmovsxbdXrm","||pmovsxbqXrm","||pmovsxwdXrm",
"||pmovsxwqXrm","||pmovsxdqXrm",nil,nil,
"||pmuldqXrvm","||pcmpeqqXrvm","||$movntdqaXrm","||packusdwXrvm",
"||maskmovpsXrvm","||maskmovpdXrvm","||maskmovpsXmvr","||maskmovpdXmvr",
--3x
"||pmovzxbwXrm","||pmovzxbdXrm","||pmovzxbqXrm","||pmovzxwdXrm",
"||pmovzxwqXrm","||pmovzxdqXrm","||permdXrvm","||pcmpgtqXrvm",
"||pminsbXrvm","||pminsdXrvm","||pminuwXrvm","||pminudXrvm",
"||pmaxsbXrvm","||pmaxsdXrvm","||pmaxuwXrvm","||pmaxudXrvm",
--4x
"||pmulddXrvm","||phminposuwXrm",nil,nil,
nil,"||psrlvVSXrvm","||psravdXrvm","||psllvVSXrvm",
--5x
[0x58] = "||pbroadcastdXrlXm",[0x59] = "||pbroadcastqXrlXm",
[0x5a] = "||broadcasti128XrlXm",
--7x
[0x78] = "||pbroadcastbXrlXm",[0x79] = "||pbroadcastwXrlXm",
--8x
[0x8c] = "||pmaskmovXrvVSm",
[0x8e] = "||pmaskmovVSmXvr",
--9x
[0x96] = "||fmaddsub132pHXrvm",[0x97] = "||fmsubadd132pHXrvm",
[0x98] = "||fmadd132pHXrvm",[0x99] = "||fmadd132sHXrvm",
[0x9a] = "||fmsub132pHXrvm",[0x9b] = "||fmsub132sHXrvm",
[0x9c] = "||fnmadd132pHXrvm",[0x9d] = "||fnmadd132sHXrvm",
[0x9e] = "||fnmsub132pHXrvm",[0x9f] = "||fnmsub132sHXrvm",
--Ax
[0xa6] = "||fmaddsub213pHXrvm",[0xa7] = "||fmsubadd213pHXrvm",
[0xa8] = "||fmadd213pHXrvm",[0xa9] = "||fmadd213sHXrvm",
[0xaa] = "||fmsub213pHXrvm",[0xab] = "||fmsub213sHXrvm",
[0xac] = "||fnmadd213pHXrvm",[0xad] = "||fnmadd213sHXrvm",
[0xae] = "||fnmsub213pHXrvm",[0xaf] = "||fnmsub213sHXrvm",
--Bx
[0xb6] = "||fmaddsub231pHXrvm",[0xb7] = "||fmsubadd231pHXrvm",
[0xb8] = "||fmadd231pHXrvm",[0xb9] = "||fmadd231sHXrvm",
[0xba] = "||fmsub231pHXrvm",[0xbb] = "||fmsub231sHXrvm",
[0xbc] = "||fnmadd231pHXrvm",[0xbd] = "||fnmadd231sHXrvm",
[0xbe] = "||fnmsub231pHXrvm",[0xbf] = "||fnmsub231sHXrvm",
--Dx
[0xdc] = "||aesencXrvm", [0xdd] = "||aesenclastXrvm",
[0xde] = "||aesdecXrvm", [0xdf] = "||aesdeclastXrvm",
--Fx
[0xf0] = "|||crc32TrBmt",[0xf1] = "|||crc32TrVmt",
[0xf7] = "| sarxVrmv| shlxVrmv| shrxVrmv",
},
["3a"] = { -- [66] 0f 3a xx
--0x
[0x00]="||permqXrmu","||permpdXrmu","||pblenddXrvmu",nil,
"||permilpsXrmu","||permilpdXrmu","||perm2f128Xrvmu",nil,
"||roundpsXrmu","||roundpdXrmu","||roundssXrvmu","||roundsdXrvmu",
"||blendpsXrvmu","||blendpdXrvmu","||pblendwXrvmu","palignrPrvmu",
--1x
nil,nil,nil,nil,
"||pextrbVmXru","||pextrwVmXru","||pextrVmSXru","||extractpsVmXru",
"||insertf128XrvlXmu","||extractf128XlXmYru",nil,nil,
nil,nil,nil,nil,
--2x
"||pinsrbXrvVmu","||insertpsXrvmu","||pinsrXrvVmuS",nil,
--3x
[0x38] = "||inserti128Xrvmu",[0x39] = "||extracti128XlXmYru",
--4x
[0x40] = "||dppsXrvmu",
[0x41] = "||dppdXrvmu",
[0x42] = "||mpsadbwXrvmu",
[0x44] = "||pclmulqdqXrvmu",
[0x46] = "||perm2i128Xrvmu",
[0x4a] = "||blendvpsXrvmb",[0x4b] = "||blendvpdXrvmb",
[0x4c] = "||pblendvbXrvmb",
--6x
[0x60] = "||pcmpestrmXrmu",[0x61] = "||pcmpestriXrmu",
[0x62] = "||pcmpistrmXrmu",[0x63] = "||pcmpistriXrmu",
[0xdf] = "||aeskeygenassistXrmu",
--Fx
[0xf0] = "||| rorxVrmu",
},
}
-- Map for VMX/SVM opcodes 0F 01 C0-FF (sgdt group with register operands).
local map_opcvm = {
[0xc1]="vmcall",[0xc2]="vmlaunch",[0xc3]="vmresume",[0xc4]="vmxoff",
[0xc8]="monitor",[0xc9]="mwait",
[0xd8]="vmrun",[0xd9]="vmmcall",[0xda]="vmload",[0xdb]="vmsave",
[0xdc]="stgi",[0xdd]="clgi",[0xde]="skinit",[0xdf]="invlpga",
[0xf8]="swapgs",[0xf9]="rdtscp",
}
-- Map for FP opcodes. And you thought stack machines are simple?
local map_opcfp = {
-- D8-DF 00-BF: opcodes with a memory operand.
-- D8
[0]="faddFm","fmulFm","fcomFm","fcompFm","fsubFm","fsubrFm","fdivFm","fdivrFm",
"fldFm",nil,"fstFm","fstpFm","fldenvVm","fldcwWm","fnstenvVm","fnstcwWm",
-- DA
"fiaddDm","fimulDm","ficomDm","ficompDm",
"fisubDm","fisubrDm","fidivDm","fidivrDm",
-- DB
"fildDm","fisttpDm","fistDm","fistpDm",nil,"fld twordFmp",nil,"fstp twordFmp",
-- DC
"faddGm","fmulGm","fcomGm","fcompGm","fsubGm","fsubrGm","fdivGm","fdivrGm",
-- DD
"fldGm","fisttpQm","fstGm","fstpGm","frstorDmp",nil,"fnsaveDmp","fnstswWm",
-- DE
"fiaddWm","fimulWm","ficomWm","ficompWm",
"fisubWm","fisubrWm","fidivWm","fidivrWm",
-- DF
"fildWm","fisttpWm","fistWm","fistpWm",
"fbld twordFmp","fildQm","fbstp twordFmp","fistpQm",
-- xx C0-FF: opcodes with a pseudo-register operand.
-- D8
"faddFf","fmulFf","fcomFf","fcompFf","fsubFf","fsubrFf","fdivFf","fdivrFf",
-- D9
"fldFf","fxchFf",{"fnop"},nil,
{"fchs","fabs",nil,nil,"ftst","fxam"},
{"fld1","fldl2t","fldl2e","fldpi","fldlg2","fldln2","fldz"},
{"f2xm1","fyl2x","fptan","fpatan","fxtract","fprem1","fdecstp","fincstp"},
{"fprem","fyl2xp1","fsqrt","fsincos","frndint","fscale","fsin","fcos"},
-- DA
"fcmovbFf","fcmoveFf","fcmovbeFf","fcmovuFf",nil,{nil,"fucompp"},nil,nil,
-- DB
"fcmovnbFf","fcmovneFf","fcmovnbeFf","fcmovnuFf",
{nil,nil,"fnclex","fninit"},"fucomiFf","fcomiFf",nil,
-- DC
"fadd toFf","fmul toFf",nil,nil,
"fsub toFf","fsubr toFf","fdivr toFf","fdiv toFf",
-- DD
"ffreeFf",nil,"fstFf","fstpFf","fucomFf","fucompFf",nil,nil,
-- DE
"faddpFf","fmulpFf",nil,{nil,"fcompp"},
"fsubrpFf","fsubpFf","fdivrpFf","fdivpFf",
-- DF
nil,nil,nil,nil,{"fnstsw ax"},"fucomipFf","fcomipFf",nil,
}
assert(map_opcfp[126] == "fcomipFf")
-- Map for opcode groups. The subkey is sp from the ModRM byte.
local map_opcgroup = {
arith = { "add", "or", "adc", "sbb", "and", "sub", "xor", "cmp" },
shift = { "rol", "ror", "rcl", "rcr", "shl", "shr", "sal", "sar" },
testb = { "testBmi", "testBmi", "not", "neg", "mul", "imul", "div", "idiv" },
testv = { "testVmi", "testVmi", "not", "neg", "mul", "imul", "div", "idiv" },
incb = { "inc", "dec" },
incd = { "inc", "dec", "callUmp", "$call farDmp",
"jmpUmp", "$jmp farDmp", "pushUm" },
sldt = { "sldt", "str", "lldt", "ltr", "verr", "verw" },
sgdt = { "vm*$sgdt", "vm*$sidt", "$lgdt", "vm*$lidt",
"smsw", nil, "lmsw", "vm*$invlpg" },
bt = { nil, nil, nil, nil, "bt", "bts", "btr", "btc" },
cmpxchg = { nil, "sz*,cmpxchg8bQmp,cmpxchg16bXmp", nil, nil,
nil, nil, "vmptrld|vmxon|vmclear", "vmptrst" },
pshiftw = { nil, nil, "psrlw", nil, "psraw", nil, "psllw" },
pshiftd = { nil, nil, "psrld", nil, "psrad", nil, "pslld" },
pshiftq = { nil, nil, "psrlq", nil, nil, nil, "psllq" },
pshiftdq = { nil, nil, "psrlq", "psrldq", nil, nil, "psllq", "pslldq" },
fxsave = { "$fxsave", "$fxrstor", "$ldmxcsr", "$stmxcsr",
nil, "lfenceDp$", "mfenceDp$", "sfenceDp$clflush" },
prefetch = { "prefetch", "prefetchw" },
prefetcht = { "prefetchnta", "prefetcht0", "prefetcht1", "prefetcht2" },
}
------------------------------------------------------------------------------
-- Maps for register names.
local map_regs = {
B = { "al", "cl", "dl", "bl", "ah", "ch", "dh", "bh",
"r8b", "r9b", "r10b", "r11b", "r12b", "r13b", "r14b", "r15b" },
B64 = { "al", "cl", "dl", "bl", "spl", "bpl", "sil", "dil",
"r8b", "r9b", "r10b", "r11b", "r12b", "r13b", "r14b", "r15b" },
W = { "ax", "cx", "dx", "bx", "sp", "bp", "si", "di",
"r8w", "r9w", "r10w", "r11w", "r12w", "r13w", "r14w", "r15w" },
D = { "eax", "ecx", "edx", "ebx", "esp", "ebp", "esi", "edi",
"r8d", "r9d", "r10d", "r11d", "r12d", "r13d", "r14d", "r15d" },
Q = { "rax", "rcx", "rdx", "rbx", "rsp", "rbp", "rsi", "rdi",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15" },
M = { "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
"mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7" }, -- No x64 ext!
X = { "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7",
"xmm8", "xmm9", "xmm10", "xmm11", "xmm12", "xmm13", "xmm14", "xmm15" },
Y = { "ymm0", "ymm1", "ymm2", "ymm3", "ymm4", "ymm5", "ymm6", "ymm7",
"ymm8", "ymm9", "ymm10", "ymm11", "ymm12", "ymm13", "ymm14", "ymm15" },
}
local map_segregs = { "es", "cs", "ss", "ds", "fs", "gs", "segr6", "segr7" }
-- Maps for size names.
local map_sz2n = {
B = 1, W = 2, D = 4, Q = 8, M = 8, X = 16, Y = 32,
}
local map_sz2prefix = {
B = "byte", W = "word", D = "dword",
Q = "qword",
M = "qword", X = "xword", Y = "yword",
F = "dword", G = "qword", -- No need for sizes/register names for these two.
}
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local code, pos, hex = ctx.code, ctx.pos, ""
local hmax = ctx.hexdump
if hmax > 0 then
for i=ctx.start,pos-1 do
hex = hex..format("%02X", byte(code, i, i))
end
if #hex > hmax then hex = sub(hex, 1, hmax)..". "
else hex = hex..rep(" ", hmax-#hex+2) end
end
if operands then text = text.." "..operands end
if ctx.o16 then text = "o16 "..text; ctx.o16 = false end
if ctx.a32 then text = "a32 "..text; ctx.a32 = false end
if ctx.rep then text = ctx.rep.." "..text; ctx.rep = false end
if ctx.rex then
local t = (ctx.rexw and "w" or "")..(ctx.rexr and "r" or "")..
(ctx.rexx and "x" or "")..(ctx.rexb and "b" or "")..
(ctx.vexl and "l" or "")
if ctx.vexv and ctx.vexv ~= 0 then t = t.."v"..ctx.vexv end
if t ~= "" then text = ctx.rex.."."..t.." "..gsub(text, "^ ", "")
elseif ctx.rex == "vex" then text = gsub("v"..text, "^v ", "") end
ctx.rexw = false; ctx.rexr = false; ctx.rexx = false; ctx.rexb = false
ctx.rex = false; ctx.vexl = false; ctx.vexv = false
end
if ctx.seg then
local text2, n = gsub(text, "%[", "["..ctx.seg..":")
if n == 0 then text = ctx.seg.." "..text else text = text2 end
ctx.seg = false
end
if ctx.lock then text = "lock "..text; ctx.lock = false end
local imm = ctx.imm
if imm then
local sym = ctx.symtab[imm]
if sym then text = text.."\t->"..sym end
end
ctx.out(format("%08x %s%s\n", ctx.addr+ctx.start, hex, text))
ctx.mrm = false
ctx.vexv = false
ctx.start = pos
ctx.imm = nil
end
-- Clear all prefix flags.
local function clearprefixes(ctx)
ctx.o16 = false; ctx.seg = false; ctx.lock = false; ctx.rep = false
ctx.rexw = false; ctx.rexr = false; ctx.rexx = false; ctx.rexb = false
ctx.rex = false; ctx.a32 = false; ctx.vexl = false
end
-- Fallback for incomplete opcodes at the end.
local function incomplete(ctx)
ctx.pos = ctx.stop+1
clearprefixes(ctx)
return putop(ctx, "(incomplete)")
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
clearprefixes(ctx)
return putop(ctx, "(unknown)")
end
-- Return an immediate of the specified size.
local function getimm(ctx, pos, n)
if pos+n-1 > ctx.stop then return incomplete(ctx) end
local code = ctx.code
if n == 1 then
local b1 = byte(code, pos, pos)
return b1
elseif n == 2 then
local b1, b2 = byte(code, pos, pos+1)
return b1+b2*256
else
local b1, b2, b3, b4 = byte(code, pos, pos+3)
local imm = b1+b2*256+b3*65536+b4*16777216
ctx.imm = imm
return imm
end
end
-- Process pattern string and generate the operands.
local function putpat(ctx, name, pat)
local operands, regs, sz, mode, sp, rm, sc, rx, sdisp
local code, pos, stop, vexl = ctx.code, ctx.pos, ctx.stop, ctx.vexl
-- Chars used: 1DFGHIMPQRSTUVWXYabcdfgijlmoprstuvwxyz
for p in gmatch(pat, ".") do
local x = nil
if p == "V" or p == "U" then
if ctx.rexw then sz = "Q"; ctx.rexw = false
elseif ctx.o16 then sz = "W"; ctx.o16 = false
elseif p == "U" and ctx.x64 then sz = "Q"
else sz = "D" end
regs = map_regs[sz]
elseif p == "T" then
if ctx.rexw then sz = "Q"; ctx.rexw = false else sz = "D" end
regs = map_regs[sz]
elseif p == "B" then
sz = "B"
regs = ctx.rex and map_regs.B64 or map_regs.B
elseif match(p, "[WDQMXYFG]") then
sz = p
if sz == "X" and vexl then sz = "Y"; ctx.vexl = false end
regs = map_regs[sz]
elseif p == "P" then
sz = ctx.o16 and "X" or "M"; ctx.o16 = false
if sz == "X" and vexl then sz = "Y"; ctx.vexl = false end
regs = map_regs[sz]
elseif p == "H" then
name = name..(ctx.rexw and "d" or "s")
ctx.rexw = false
elseif p == "S" then
name = name..lower(sz)
elseif p == "s" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = imm <= 127 and format("+0x%02x", imm)
or format("-0x%02x", 256-imm)
pos = pos+1
elseif p == "u" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = format("0x%02x", imm)
pos = pos+1
elseif p == "b" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = regs[imm/16+1]
pos = pos+1
elseif p == "w" then
local imm = getimm(ctx, pos, 2); if not imm then return end
x = format("0x%x", imm)
pos = pos+2
elseif p == "o" then -- [offset]
if ctx.x64 then
local imm1 = getimm(ctx, pos, 4); if not imm1 then return end
local imm2 = getimm(ctx, pos+4, 4); if not imm2 then return end
x = format("[0x%08x%08x]", imm2, imm1)
pos = pos+8
else
local imm = getimm(ctx, pos, 4); if not imm then return end
x = format("[0x%08x]", imm)
pos = pos+4
end
elseif p == "i" or p == "I" then
local n = map_sz2n[sz]
if n == 8 and ctx.x64 and p == "I" then
local imm1 = getimm(ctx, pos, 4); if not imm1 then return end
local imm2 = getimm(ctx, pos+4, 4); if not imm2 then return end
x = format("0x%08x%08x", imm2, imm1)
else
if n == 8 then n = 4 end
local imm = getimm(ctx, pos, n); if not imm then return end
if sz == "Q" and (imm < 0 or imm > 0x7fffffff) then
imm = (0xffffffff+1)-imm
x = format(imm > 65535 and "-0x%08x" or "-0x%x", imm)
else
x = format(imm > 65535 and "0x%08x" or "0x%x", imm)
end
end
pos = pos+n
elseif p == "j" then
local n = map_sz2n[sz]
if n == 8 then n = 4 end
local imm = getimm(ctx, pos, n); if not imm then return end
if sz == "B" and imm > 127 then imm = imm-256
elseif imm > 2147483647 then imm = imm-4294967296 end
pos = pos+n
imm = imm + pos + ctx.addr
if imm > 4294967295 and not ctx.x64 then imm = imm-4294967296 end
ctx.imm = imm
if sz == "W" then
x = format("word 0x%04x", imm%65536)
elseif ctx.x64 then
local lo = imm % 0x1000000
x = format("0x%02x%06x", (imm-lo) / 0x1000000, lo)
else
x = "0x"..tohex(imm)
end
elseif p == "R" then
local r = byte(code, pos-1, pos-1)%8
if ctx.rexb then r = r + 8; ctx.rexb = false end
x = regs[r+1]
elseif p == "a" then x = regs[1]
elseif p == "c" then x = "cl"
elseif p == "d" then x = "dx"
elseif p == "1" then x = "1"
else
if not mode then
mode = ctx.mrm
if not mode then
if pos > stop then return incomplete(ctx) end
mode = byte(code, pos, pos)
pos = pos+1
end
rm = mode%8; mode = (mode-rm)/8
sp = mode%8; mode = (mode-sp)/8
sdisp = ""
if mode < 3 then
if rm == 4 then
if pos > stop then return incomplete(ctx) end
sc = byte(code, pos, pos)
pos = pos+1
rm = sc%8; sc = (sc-rm)/8
rx = sc%8; sc = (sc-rx)/8
if ctx.rexx then rx = rx + 8; ctx.rexx = false end
if rx == 4 then rx = nil end
end
if mode > 0 or rm == 5 then
local dsz = mode
if dsz ~= 1 then dsz = 4 end
local disp = getimm(ctx, pos, dsz); if not disp then return end
if mode == 0 then rm = nil end
if rm or rx or (not sc and ctx.x64 and not ctx.a32) then
if dsz == 1 and disp > 127 then
sdisp = format("-0x%x", 256-disp)
elseif disp >= 0 and disp <= 0x7fffffff then
sdisp = format("+0x%x", disp)
else
sdisp = format("-0x%x", (0xffffffff+1)-disp)
end
else
sdisp = format(ctx.x64 and not ctx.a32 and
not (disp >= 0 and disp <= 0x7fffffff)
and "0xffffffff%08x" or "0x%08x", disp)
end
pos = pos+dsz
end
end
if rm and ctx.rexb then rm = rm + 8; ctx.rexb = false end
if ctx.rexr then sp = sp + 8; ctx.rexr = false end
end
if p == "m" then
if mode == 3 then x = regs[rm+1]
else
local aregs = ctx.a32 and map_regs.D or ctx.aregs
local srm, srx = "", ""
if rm then srm = aregs[rm+1]
elseif not sc and ctx.x64 and not ctx.a32 then srm = "rip" end
ctx.a32 = false
if rx then
if rm then srm = srm.."+" end
srx = aregs[rx+1]
if sc > 0 then srx = srx.."*"..(2^sc) end
end
x = format("[%s%s%s]", srm, srx, sdisp)
end
if mode < 3 and
(not match(pat, "[aRrgp]") or match(pat, "t")) then -- Yuck.
x = map_sz2prefix[sz].." "..x
end
elseif p == "r" then x = regs[sp+1]
elseif p == "g" then x = map_segregs[sp+1]
elseif p == "p" then -- Suppress prefix.
elseif p == "f" then x = "st"..rm
elseif p == "x" then
if sp == 0 and ctx.lock and not ctx.x64 then
x = "CR8"; ctx.lock = false
else
x = "CR"..sp
end
elseif p == "v" then
if ctx.vexv then
x = regs[ctx.vexv+1]; ctx.vexv = false
end
elseif p == "y" then x = "DR"..sp
elseif p == "z" then x = "TR"..sp
elseif p == "l" then vexl = false
elseif p == "t" then
else
error("bad pattern `"..pat.."'")
end
end
if x then operands = operands and operands..", "..x or x end
end
ctx.pos = pos
return putop(ctx, name, operands)
end
-- Forward declaration.
local map_act
-- Fetch and cache MRM byte.
local function getmrm(ctx)
local mrm = ctx.mrm
if not mrm then
local pos = ctx.pos
if pos > ctx.stop then return nil end
mrm = byte(ctx.code, pos, pos)
ctx.pos = pos+1
ctx.mrm = mrm
end
return mrm
end
-- Dispatch to handler depending on pattern.
local function dispatch(ctx, opat, patgrp)
if not opat then return unknown(ctx) end
if match(opat, "%|") then -- MMX/SSE variants depending on prefix.
local p
if ctx.rep then
p = ctx.rep=="rep" and "%|([^%|]*)" or "%|[^%|]*%|[^%|]*%|([^%|]*)"
ctx.rep = false
elseif ctx.o16 then p = "%|[^%|]*%|([^%|]*)"; ctx.o16 = false
else p = "^[^%|]*" end
opat = match(opat, p)
if not opat then return unknown(ctx) end
-- ctx.rep = false; ctx.o16 = false
--XXX fails for 66 f2 0f 38 f1 06 crc32 eax,WORD PTR [esi]
--XXX remove in branches?
end
if match(opat, "%$") then -- reg$mem variants.
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
opat = match(opat, mrm >= 192 and "^[^%$]*" or "%$(.*)")
if opat == "" then return unknown(ctx) end
end
if opat == "" then return unknown(ctx) end
local name, pat = match(opat, "^([a-z0-9 ]*)(.*)")
if pat == "" and patgrp then pat = patgrp end
return map_act[sub(pat, 1, 1)](ctx, name, pat)
end
-- Get a pattern from an opcode map and dispatch to handler.
local function dispatchmap(ctx, opcmap)
local pos = ctx.pos
local opat = opcmap[byte(ctx.code, pos, pos)]
pos = pos + 1
ctx.pos = pos
return dispatch(ctx, opat)
end
-- Map for action codes. The key is the first char after the name.
map_act = {
-- Simple opcodes without operands.
[""] = function(ctx, name, pat)
return putop(ctx, name)
end,
-- Operand size chars fall right through.
B = putpat, W = putpat, D = putpat, Q = putpat,
V = putpat, U = putpat, T = putpat,
M = putpat, X = putpat, P = putpat,
F = putpat, G = putpat, Y = putpat,
H = putpat,
-- Collect prefixes.
[":"] = function(ctx, name, pat)
ctx[pat == ":" and name or sub(pat, 2)] = name
if ctx.pos - ctx.start > 5 then return unknown(ctx) end -- Limit #prefixes.
end,
-- Chain to special handler specified by name.
["*"] = function(ctx, name, pat)
return map_act[name](ctx, name, sub(pat, 2))
end,
-- Use named subtable for opcode group.
["!"] = function(ctx, name, pat)
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
return dispatch(ctx, map_opcgroup[name][((mrm-(mrm%8))/8)%8+1], sub(pat, 2))
end,
-- o16,o32[,o64] variants.
sz = function(ctx, name, pat)
if ctx.o16 then ctx.o16 = false
else
pat = match(pat, ",(.*)")
if ctx.rexw then
local p = match(pat, ",(.*)")
if p then pat = p; ctx.rexw = false end
end
end
pat = match(pat, "^[^,]*")
return dispatch(ctx, pat)
end,
-- Two-byte opcode dispatch.
opc2 = function(ctx, name, pat)
return dispatchmap(ctx, map_opc2)
end,
-- Three-byte opcode dispatch.
opc3 = function(ctx, name, pat)
return dispatchmap(ctx, map_opc3[pat])
end,
-- VMX/SVM dispatch.
vm = function(ctx, name, pat)
return dispatch(ctx, map_opcvm[ctx.mrm])
end,
-- Floating point opcode dispatch.
fp = function(ctx, name, pat)
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
local rm = mrm%8
local idx = pat*8 + ((mrm-rm)/8)%8
if mrm >= 192 then idx = idx + 64 end
local opat = map_opcfp[idx]
if type(opat) == "table" then opat = opat[rm+1] end
return dispatch(ctx, opat)
end,
-- REX prefix.
rex = function(ctx, name, pat)
if ctx.rex then return unknown(ctx) end -- Only 1 REX or VEX prefix allowed.
for p in gmatch(pat, ".") do ctx["rex"..p] = true end
ctx.rex = "rex"
end,
-- VEX prefix.
vex = function(ctx, name, pat)
if ctx.rex then return unknown(ctx) end -- Only 1 REX or VEX prefix allowed.
ctx.rex = "vex"
local pos = ctx.pos
if ctx.mrm then
ctx.mrm = nil
pos = pos-1
end
local b = byte(ctx.code, pos, pos)
if not b then return incomplete(ctx) end
pos = pos+1
if b < 128 then ctx.rexr = true end
local m = 1
if pat == "3" then
m = b%32; b = (b-m)/32
local nb = b%2; b = (b-nb)/2
if nb == 0 then ctx.rexb = true end
local nx = b%2
if nx == 0 then ctx.rexx = true end
b = byte(ctx.code, pos, pos)
if not b then return incomplete(ctx) end
pos = pos+1
if b >= 128 then ctx.rexw = true end
end
ctx.pos = pos
local map
if m == 1 then map = map_opc2
elseif m == 2 then map = map_opc3["38"]
elseif m == 3 then map = map_opc3["3a"]
else return unknown(ctx) end
local p = b%4; b = (b-p)/4
if p == 1 then ctx.o16 = "o16"
elseif p == 2 then ctx.rep = "rep"
elseif p == 3 then ctx.rep = "repne" end
local l = b%2; b = (b-l)/2
if l ~= 0 then ctx.vexl = true end
ctx.vexv = (-1-b)%16
return dispatchmap(ctx, map)
end,
-- Special case for nop with REX prefix.
nop = function(ctx, name, pat)
return dispatch(ctx, ctx.rex and pat or "nop")
end,
-- Special case for 0F 77.
emms = function(ctx, name, pat)
if ctx.rex ~= "vex" then
return putop(ctx, "emms")
elseif ctx.vexl then
ctx.vexl = false
return putop(ctx, "zeroall")
else
return putop(ctx, "zeroupper")
end
end,
}
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
ofs = ofs + 1
ctx.start = ofs
ctx.pos = ofs
ctx.stop = stop
ctx.imm = nil
ctx.mrm = false
clearprefixes(ctx)
while ctx.pos <= stop do dispatchmap(ctx, ctx.map1) end
if ctx.pos ~= ctx.start then incomplete(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = (addr or 0) - 1
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 16
ctx.x64 = false
ctx.map1 = map_opc1_32
ctx.aregs = map_regs.D
return ctx
end
local function create64(code, addr, out)
local ctx = create(code, addr, out)
ctx.x64 = true
ctx.map1 = map_opc1_64
ctx.aregs = map_regs.Q
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
local function disass64(code, addr, out)
create64(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 8 then return map_regs.D[r+1] end
return map_regs.X[r-7]
end
local function regname64(r)
if r < 16 then return map_regs.Q[r+1] end
return map_regs.X[r-15]
end
-- Public module functions.
return {
create = create,
create64 = create64,
disass = disass,
disass64 = disass64,
regname = regname,
regname64 = regname64
}
@@ -0,0 +1,726 @@
----------------------------------------------------------------------------
-- LuaJIT compiler dump module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module can be used to debug the JIT compiler itself. It dumps the
-- code representations and structures used in various compiler stages.
--
-- Example usage:
--
-- luajit -jdump -e "local x=0; for i=1,1e6 do x=x+i end; print(x)"
-- luajit -jdump=im -e "for i=1,1000 do for j=1,1000 do end end" | less -R
-- luajit -jdump=is myapp.lua | less -R
-- luajit -jdump=-b myapp.lua
-- luajit -jdump=+aH,myapp.html myapp.lua
-- luajit -jdump=ixT,myapp.dump myapp.lua
--
-- The first argument specifies the dump mode. The second argument gives
-- the output file name. Default output is to stdout, unless the environment
-- variable LUAJIT_DUMPFILE is set. The file is overwritten every time the
-- module is started.
--
-- Different features can be turned on or off with the dump mode. If the
-- mode starts with a '+', the following features are added to the default
-- set of features; a '-' removes them. Otherwise the features are replaced.
--
-- The following dump features are available (* marks the default):
--
-- * t Print a line for each started, ended or aborted trace (see also -jv).
-- * b Dump the traced bytecode.
-- * i Dump the IR (intermediate representation).
-- r Augment the IR with register/stack slots.
-- s Dump the snapshot map.
-- * m Dump the generated machine code.
-- x Print each taken trace exit.
-- X Print each taken trace exit and the contents of all registers.
-- a Print the IR of aborted traces, too.
--
-- The output format can be set with the following characters:
--
-- T Plain text output.
-- A ANSI-colored text output
-- H Colorized HTML + CSS output.
--
-- The default output format is plain text. It's set to ANSI-colored text
-- if the COLORTERM variable is set. Note: this is independent of any output
-- redirection, which is actually considered a feature.
--
-- You probably want to use less -R to enjoy viewing ANSI-colored text from
-- a pipe or a file. Add this to your ~/.bashrc: export LESS="-R"
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local funcinfo, funcbc = jutil.funcinfo, jutil.funcbc
local traceinfo, traceir, tracek = jutil.traceinfo, jutil.traceir, jutil.tracek
local tracemc, tracesnap = jutil.tracemc, jutil.tracesnap
local traceexitstub, ircalladdr = jutil.traceexitstub, jutil.ircalladdr
local bit = require("bit")
local band, shr, tohex = bit.band, bit.rshift, bit.tohex
local sub, gsub, format = string.sub, string.gsub, string.format
local byte, rep = string.byte, string.rep
local type, tostring = type, tostring
local stdout, stderr = io.stdout, io.stderr
-- Load other modules on-demand.
local bcline, disass
-- Active flag, output file handle and dump mode.
local active, out, dumpmode
------------------------------------------------------------------------------
local symtabmt = { __index = false }
local symtab = {}
local nexitsym = 0
-- Fill nested symbol table with per-trace exit stub addresses.
local function fillsymtab_tr(tr, nexit)
local t = {}
symtabmt.__index = t
if jit.arch:sub(1, 4) == "mips" then
t[traceexitstub(tr, 0)] = "exit"
return
end
for i=0,nexit-1 do
local addr = traceexitstub(tr, i)
if addr < 0 then addr = addr + 2^32 end
t[addr] = tostring(i)
end
local addr = traceexitstub(tr, nexit)
if addr then t[addr] = "stack_check" end
end
-- Fill symbol table with trace exit stub addresses.
local function fillsymtab(tr, nexit)
local t = symtab
if nexitsym == 0 then
local maskaddr = jit.arch == "arm" and -2
local ircall = vmdef.ircall
for i=0,#ircall do
local addr = ircalladdr(i)
if addr ~= 0 then
if maskaddr then addr = band(addr, maskaddr) end
if addr < 0 then addr = addr + 2^32 end
t[addr] = ircall[i]
end
end
end
if nexitsym == 1000000 then -- Per-trace exit stubs.
fillsymtab_tr(tr, nexit)
elseif nexit > nexitsym then -- Shared exit stubs.
for i=nexitsym,nexit-1 do
local addr = traceexitstub(i)
if addr == nil then -- Fall back to per-trace exit stubs.
fillsymtab_tr(tr, nexit)
setmetatable(symtab, symtabmt)
nexit = 1000000
break
end
if addr < 0 then addr = addr + 2^32 end
t[addr] = tostring(i)
end
nexitsym = nexit
end
return t
end
local function dumpwrite(s)
out:write(s)
end
-- Disassemble machine code.
local function dump_mcode(tr)
local info = traceinfo(tr)
if not info then return end
local mcode, addr, loop = tracemc(tr)
if not mcode then return end
if not disass then disass = require("jit.dis_"..jit.arch) end
if addr < 0 then addr = addr + 2^32 end
out:write("---- TRACE ", tr, " mcode ", #mcode, "\n")
local ctx = disass.create(mcode, addr, dumpwrite)
ctx.hexdump = 0
ctx.symtab = fillsymtab(tr, info.nexit)
if loop ~= 0 then
symtab[addr+loop] = "LOOP"
ctx:disass(0, loop)
out:write("->LOOP:\n")
ctx:disass(loop, #mcode-loop)
symtab[addr+loop] = nil
else
ctx:disass(0, #mcode)
end
end
------------------------------------------------------------------------------
local irtype_text = {
[0] = "nil",
"fal",
"tru",
"lud",
"str",
"p32",
"thr",
"pro",
"fun",
"p64",
"cdt",
"tab",
"udt",
"flt",
"num",
"i8 ",
"u8 ",
"i16",
"u16",
"int",
"u32",
"i64",
"u64",
"sfp",
}
local colortype_ansi = {
[0] = "%s",
"%s",
"%s",
"\027[36m%s\027[m",
"\027[32m%s\027[m",
"%s",
"\027[1m%s\027[m",
"%s",
"\027[1m%s\027[m",
"%s",
"\027[33m%s\027[m",
"\027[31m%s\027[m",
"\027[36m%s\027[m",
"\027[34m%s\027[m",
"\027[34m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
}
local function colorize_text(s)
return s
end
local function colorize_ansi(s, t, extra)
local out = format(colortype_ansi[t], s)
if extra then out = "\027[3m"..out end
return out
end
local irtype_ansi = setmetatable({},
{ __index = function(tab, t)
local s = colorize_ansi(irtype_text[t], t); tab[t] = s; return s; end })
local html_escape = { ["<"] = "&lt;", [">"] = "&gt;", ["&"] = "&amp;", }
local function colorize_html(s, t, extra)
s = gsub(s, "[<>&]", html_escape)
return format('<span class="irt_%s%s">%s</span>',
irtype_text[t], extra and " irt_extra" or "", s)
end
local irtype_html = setmetatable({},
{ __index = function(tab, t)
local s = colorize_html(irtype_text[t], t); tab[t] = s; return s; end })
local header_html = [[
<style type="text/css">
background { background: #ffffff; color: #000000; }
pre.ljdump {
font-size: 10pt;
background: #f0f4ff;
color: #000000;
border: 1px solid #bfcfff;
padding: 0.5em;
margin-left: 2em;
margin-right: 2em;
}
span.irt_str { color: #00a000; }
span.irt_thr, span.irt_fun { color: #404040; font-weight: bold; }
span.irt_tab { color: #c00000; }
span.irt_udt, span.irt_lud { color: #00c0c0; }
span.irt_num { color: #4040c0; }
span.irt_int, span.irt_i8, span.irt_u8, span.irt_i16, span.irt_u16 { color: #b040b0; }
span.irt_extra { font-style: italic; }
</style>
]]
local colorize, irtype
-- Lookup tables to convert some literals into names.
local litname = {
["SLOAD "] = setmetatable({}, { __index = function(t, mode)
local s = ""
if band(mode, 1) ~= 0 then s = s.."P" end
if band(mode, 2) ~= 0 then s = s.."F" end
if band(mode, 4) ~= 0 then s = s.."T" end
if band(mode, 8) ~= 0 then s = s.."C" end
if band(mode, 16) ~= 0 then s = s.."R" end
if band(mode, 32) ~= 0 then s = s.."I" end
if band(mode, 64) ~= 0 then s = s.."K" end
t[mode] = s
return s
end}),
["XLOAD "] = { [0] = "", "R", "V", "RV", "U", "RU", "VU", "RVU", },
["CONV "] = setmetatable({}, { __index = function(t, mode)
local s = irtype[band(mode, 31)]
s = irtype[band(shr(mode, 5), 31)].."."..s
if band(mode, 0x800) ~= 0 then s = s.." sext" end
local c = shr(mode, 12)
if c == 1 then s = s.." none"
elseif c == 2 then s = s.." index"
elseif c == 3 then s = s.." check" end
t[mode] = s
return s
end}),
["FLOAD "] = vmdef.irfield,
["FREF "] = vmdef.irfield,
["FPMATH"] = vmdef.irfpm,
["TMPREF"] = { [0] = "", "IN", "OUT", "INOUT", "", "", "OUT2", "INOUT2" },
["BUFHDR"] = { [0] = "RESET", "APPEND", "WRITE" },
["TOSTR "] = { [0] = "INT", "NUM", "CHAR" },
}
local function ctlsub(c)
if c == "\n" then return "\\n"
elseif c == "\r" then return "\\r"
elseif c == "\t" then return "\\t"
else return format("\\%03d", byte(c))
end
end
local function fmtfunc(func, pc)
local fi = funcinfo(func, pc)
if fi.loc then
return fi.loc
elseif fi.ffid then
return vmdef.ffnames[fi.ffid]
elseif fi.addr then
return format("C:%x", fi.addr)
else
return "(?)"
end
end
local function formatk(tr, idx, sn)
local k, t, slot = tracek(tr, idx)
local tn = type(k)
local s
if tn == "number" then
if t < 12 then
s = k == 0 and "NULL" or format("[0x%08x]", k)
elseif band(sn or 0, 0x30000) ~= 0 then
s = band(sn, 0x20000) ~= 0 and "contpc" or "ftsz"
elseif k == 2^52+2^51 then
s = "bias"
else
s = format(0 < k and k < 0x1p-1026 and "%+a" or "%+.14g", k)
end
elseif tn == "string" then
s = format(#k > 20 and '"%.20s"~' or '"%s"', gsub(k, "%c", ctlsub))
elseif tn == "function" then
s = fmtfunc(k)
elseif tn == "table" then
s = format("{%p}", k)
elseif tn == "userdata" then
if t == 12 then
s = format("userdata:%p", k)
else
s = format("[%p]", k)
if s == "[NULL]" then s = "NULL" end
end
elseif t == 21 then -- int64_t
s = sub(tostring(k), 1, -3)
if sub(s, 1, 1) ~= "-" then s = "+"..s end
elseif sn == 0x1057fff then -- SNAP(1, SNAP_FRAME | SNAP_NORESTORE, REF_NIL)
return "----" -- Special case for LJ_FR2 slot 1.
else
s = tostring(k) -- For primitives.
end
s = colorize(format("%-4s", s), t, band(sn or 0, 0x100000) ~= 0)
if slot then
s = format("%s @%d", s, slot)
end
return s
end
local function printsnap(tr, snap)
local n = 2
for s=0,snap[1]-1 do
local sn = snap[n]
if shr(sn, 24) == s then
n = n + 1
local ref = band(sn, 0xffff) - 0x8000 -- REF_BIAS
if ref < 0 then
out:write(formatk(tr, ref, sn))
elseif band(sn, 0x80000) ~= 0 then -- SNAP_SOFTFPNUM
out:write(colorize(format("%04d/%04d", ref, ref+1), 14))
else
local m, ot, op1, op2 = traceir(tr, ref)
out:write(colorize(format("%04d", ref), band(ot, 31), band(sn, 0x100000) ~= 0))
end
out:write(band(sn, 0x10000) == 0 and " " or "|") -- SNAP_FRAME
else
out:write("---- ")
end
end
out:write("]\n")
end
-- Dump snapshots (not interleaved with IR).
local function dump_snap(tr)
out:write("---- TRACE ", tr, " snapshots\n")
for i=0,1000000000 do
local snap = tracesnap(tr, i)
if not snap then break end
out:write(format("#%-3d %04d [ ", i, snap[0]))
printsnap(tr, snap)
end
end
-- Return a register name or stack slot for a rid/sp location.
local function ridsp_name(ridsp, ins)
if not disass then disass = require("jit.dis_"..jit.arch) end
local rid, slot = band(ridsp, 0xff), shr(ridsp, 8)
if rid == 253 or rid == 254 then
return (slot == 0 or slot == 255) and " {sink" or format(" {%04d", ins-slot)
end
if ridsp > 255 then return format("[%x]", slot*4) end
if rid < 128 then return disass.regname(rid) end
return ""
end
-- Dump CALL* function ref and return optional ctype.
local function dumpcallfunc(tr, ins)
local ctype
if ins > 0 then
local m, ot, op1, op2 = traceir(tr, ins)
if band(ot, 31) == 0 then -- nil type means CARG(func, ctype).
ins = op1
ctype = formatk(tr, op2)
end
end
if ins < 0 then
out:write(format("[0x%x](", tonumber((tracek(tr, ins)))))
else
out:write(format("%04d (", ins))
end
return ctype
end
-- Recursively gather CALL* args and dump them.
local function dumpcallargs(tr, ins)
if ins < 0 then
out:write(formatk(tr, ins))
else
local m, ot, op1, op2 = traceir(tr, ins)
local oidx = 6*shr(ot, 8)
local op = sub(vmdef.irnames, oidx+1, oidx+6)
if op == "CARG " then
dumpcallargs(tr, op1)
if op2 < 0 then
out:write(" ", formatk(tr, op2))
else
out:write(" ", format("%04d", op2))
end
else
out:write(format("%04d", ins))
end
end
end
-- Dump IR and interleaved snapshots.
local function dump_ir(tr, dumpsnap, dumpreg)
local info = traceinfo(tr)
if not info then return end
local nins = info.nins
out:write("---- TRACE ", tr, " IR\n")
local irnames = vmdef.irnames
local snapref = 65536
local snap, snapno
if dumpsnap then
snap = tracesnap(tr, 0)
snapref = snap[0]
snapno = 0
end
for ins=1,nins do
if ins >= snapref then
if dumpreg then
out:write(format(".... SNAP #%-3d [ ", snapno))
else
out:write(format(".... SNAP #%-3d [ ", snapno))
end
printsnap(tr, snap)
snapno = snapno + 1
snap = tracesnap(tr, snapno)
snapref = snap and snap[0] or 65536
end
local m, ot, op1, op2, ridsp = traceir(tr, ins)
local oidx, t = 6*shr(ot, 8), band(ot, 31)
local op = sub(irnames, oidx+1, oidx+6)
if op == "LOOP " then
if dumpreg then
out:write(format("%04d ------------ LOOP ------------\n", ins))
else
out:write(format("%04d ------ LOOP ------------\n", ins))
end
elseif op ~= "NOP " and op ~= "CARG " and
(dumpreg or op ~= "RENAME") then
local rid = band(ridsp, 255)
if dumpreg then
out:write(format("%04d %-6s", ins, ridsp_name(ridsp, ins)))
else
out:write(format("%04d ", ins))
end
out:write(format("%s%s %s %s ",
(rid == 254 or rid == 253) and "}" or
(band(ot, 128) == 0 and " " or ">"),
band(ot, 64) == 0 and " " or "+",
irtype[t], op))
local m1, m2 = band(m, 3), band(m, 3*4)
if sub(op, 1, 4) == "CALL" then
local ctype
if m2 == 1*4 then -- op2 == IRMlit
out:write(format("%-10s (", vmdef.ircall[op2]))
else
ctype = dumpcallfunc(tr, op2)
end
if op1 ~= -1 then dumpcallargs(tr, op1) end
out:write(")")
if ctype then out:write(" ctype ", ctype) end
elseif op == "CNEW " and op2 == -1 then
out:write(formatk(tr, op1))
elseif m1 ~= 3 then -- op1 != IRMnone
if op1 < 0 then
out:write(formatk(tr, op1))
else
out:write(format(m1 == 0 and "%04d" or "#%-3d", op1))
end
if m2 ~= 3*4 then -- op2 != IRMnone
if m2 == 1*4 then -- op2 == IRMlit
local litn = litname[op]
if litn and litn[op2] then
out:write(" ", litn[op2])
elseif op == "UREFO " or op == "UREFC " then
out:write(format(" #%-3d", shr(op2, 8)))
else
out:write(format(" #%-3d", op2))
end
elseif op2 < 0 then
out:write(" ", formatk(tr, op2))
else
out:write(format(" %04d", op2))
end
end
end
out:write("\n")
end
end
if snap then
if dumpreg then
out:write(format(".... SNAP #%-3d [ ", snapno))
else
out:write(format(".... SNAP #%-3d [ ", snapno))
end
printsnap(tr, snap)
end
end
------------------------------------------------------------------------------
local recprefix = ""
local recdepth = 0
-- Format trace error message.
local function fmterr(err, info)
if type(err) == "number" then
if type(info) == "function" then info = fmtfunc(info) end
err = format(vmdef.traceerr[err], info)
end
return err
end
-- Dump trace states.
local function dump_trace(what, tr, func, pc, otr, oex)
if what == "stop" or (what == "abort" and dumpmode.a) then
if dumpmode.i then dump_ir(tr, dumpmode.s, dumpmode.r and what == "stop")
elseif dumpmode.s then dump_snap(tr) end
if dumpmode.m then dump_mcode(tr) end
end
if what == "start" then
if dumpmode.H then out:write('<pre class="ljdump">\n') end
out:write("---- TRACE ", tr, " ", what)
if otr then out:write(" ", otr, "/", oex == -1 and "stitch" or oex) end
out:write(" ", fmtfunc(func, pc), "\n")
elseif what == "stop" or what == "abort" then
out:write("---- TRACE ", tr, " ", what)
if what == "abort" then
out:write(" ", fmtfunc(func, pc), " -- ", fmterr(otr, oex), "\n")
else
local info = traceinfo(tr)
local link, ltype = info.link, info.linktype
if link == tr or link == 0 then
out:write(" -> ", ltype, "\n")
elseif ltype == "root" then
out:write(" -> ", link, "\n")
else
out:write(" -> ", link, " ", ltype, "\n")
end
end
if dumpmode.H then out:write("</pre>\n\n") else out:write("\n") end
else
if what == "flush" then symtab, nexitsym = {}, 0 end
out:write("---- TRACE ", what, "\n\n")
end
out:flush()
end
-- Dump recorded bytecode.
local function dump_record(tr, func, pc, depth)
if depth ~= recdepth then
recdepth = depth
recprefix = rep(" .", depth)
end
local line
if pc >= 0 then
line = bcline(func, pc, recprefix)
if dumpmode.H then line = gsub(line, "[<>&]", html_escape) end
else
line = "0000 "..recprefix.." FUNCC \n"
end
if pc <= 0 then
out:write(sub(line, 1, -2), " ; ", fmtfunc(func), "\n")
else
out:write(line)
end
if pc >= 0 and band(funcbc(func, pc), 0xff) < 16 then -- ORDER BC
out:write(bcline(func, pc+1, recprefix)) -- Write JMP for cond.
end
end
------------------------------------------------------------------------------
local gpr64 = jit.arch:match("64")
local fprmips32 = jit.arch == "mips" or jit.arch == "mipsel"
-- Dump taken trace exits.
local function dump_texit(tr, ex, ngpr, nfpr, ...)
out:write("---- TRACE ", tr, " exit ", ex, "\n")
if dumpmode.X then
local regs = {...}
if gpr64 then
for i=1,ngpr do
out:write(format(" %016x", regs[i]))
if i % 4 == 0 then out:write("\n") end
end
else
for i=1,ngpr do
out:write(" ", tohex(regs[i]))
if i % 8 == 0 then out:write("\n") end
end
end
if fprmips32 then
for i=1,nfpr,2 do
out:write(format(" %+17.14g", regs[ngpr+i]))
if i % 8 == 7 then out:write("\n") end
end
else
for i=1,nfpr do
out:write(format(" %+17.14g", regs[ngpr+i]))
if i % 4 == 0 then out:write("\n") end
end
end
end
end
------------------------------------------------------------------------------
-- Detach dump handlers.
local function dumpoff()
if active then
active = false
jit.attach(dump_texit)
jit.attach(dump_record)
jit.attach(dump_trace)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach dump handlers.
local function dumpon(opt, outfile)
if active then dumpoff() end
local term = os.getenv("TERM")
local colormode = (term and term:match("color") or os.getenv("COLORTERM")) and "A" or "T"
if opt then
opt = gsub(opt, "[TAH]", function(mode) colormode = mode; return ""; end)
end
local m = { t=true, b=true, i=true, m=true, }
if opt and opt ~= "" then
local o = sub(opt, 1, 1)
if o ~= "+" and o ~= "-" then m = {} end
for i=1,#opt do m[sub(opt, i, i)] = (o ~= "-") end
end
dumpmode = m
if m.t or m.b or m.i or m.s or m.m then
jit.attach(dump_trace, "trace")
end
if m.b then
jit.attach(dump_record, "record")
if not bcline then bcline = require("jit.bc").line end
end
if m.x or m.X then
jit.attach(dump_texit, "texit")
end
if not outfile then outfile = os.getenv("LUAJIT_DUMPFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stdout
end
m[colormode] = true
if colormode == "A" then
colorize = colorize_ansi
irtype = irtype_ansi
elseif colormode == "H" then
colorize = colorize_html
irtype = irtype_html
out:write(header_html)
else
colorize = colorize_text
irtype = irtype_text
end
active = true
end
-- Public module functions.
return {
on = dumpon,
off = dumpoff,
start = dumpon -- For -j command line option.
}
@@ -0,0 +1,312 @@
----------------------------------------------------------------------------
-- LuaJIT profiler.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module is a simple command line interface to the built-in
-- low-overhead profiler of LuaJIT.
--
-- The lower-level API of the profiler is accessible via the "jit.profile"
-- module or the luaJIT_profile_* C API.
--
-- Example usage:
--
-- luajit -jp myapp.lua
-- luajit -jp=s myapp.lua
-- luajit -jp=-s myapp.lua
-- luajit -jp=vl myapp.lua
-- luajit -jp=G,profile.txt myapp.lua
--
-- The following dump features are available:
--
-- f Stack dump: function name, otherwise module:line. Default mode.
-- F Stack dump: ditto, but always prepend module.
-- l Stack dump: module:line.
-- <number> stack dump depth (callee < caller). Default: 1.
-- -<number> Inverse stack dump depth (caller > callee).
-- s Split stack dump after first stack level. Implies abs(depth) >= 2.
-- p Show full path for module names.
-- v Show VM states. Can be combined with stack dumps, e.g. vf or fv.
-- z Show zones. Can be combined with stack dumps, e.g. zf or fz.
-- r Show raw sample counts. Default: show percentages.
-- a Annotate excerpts from source code files.
-- A Annotate complete source code files.
-- G Produce raw output suitable for graphical tools (e.g. flame graphs).
-- m<number> Minimum sample percentage to be shown. Default: 3.
-- i<number> Sampling interval in milliseconds. Default: 10.
--
----------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local profile = require("jit.profile")
local vmdef = require("jit.vmdef")
local math = math
local pairs, ipairs, tonumber, floor = pairs, ipairs, tonumber, math.floor
local sort, format = table.sort, string.format
local stdout = io.stdout
local zone -- Load jit.zone module on demand.
-- Output file handle.
local out
------------------------------------------------------------------------------
local prof_ud
local prof_states, prof_split, prof_min, prof_raw, prof_fmt, prof_depth
local prof_ann, prof_count1, prof_count2, prof_samples
local map_vmmode = {
N = "Compiled",
I = "Interpreted",
C = "C code",
G = "Garbage Collector",
J = "JIT Compiler",
}
-- Profiler callback.
local function prof_cb(th, samples, vmmode)
prof_samples = prof_samples + samples
local key_stack, key_stack2, key_state
-- Collect keys for sample.
if prof_states then
if prof_states == "v" then
key_state = map_vmmode[vmmode] or vmmode
else
key_state = zone:get() or "(none)"
end
end
if prof_fmt then
key_stack = profile.dumpstack(th, prof_fmt, prof_depth)
key_stack = key_stack:gsub("%[builtin#(%d+)%]", function(x)
return vmdef.ffnames[tonumber(x)]
end)
if prof_split == 2 then
local k1, k2 = key_stack:match("(.-) [<>] (.*)")
if k2 then key_stack, key_stack2 = k1, k2 end
elseif prof_split == 3 then
key_stack2 = profile.dumpstack(th, "l", 1)
end
end
-- Order keys.
local k1, k2
if prof_split == 1 then
if key_state then
k1 = key_state
if key_stack then k2 = key_stack end
end
elseif key_stack then
k1 = key_stack
if key_stack2 then k2 = key_stack2 elseif key_state then k2 = key_state end
end
-- Coalesce samples in one or two levels.
if k1 then
local t1 = prof_count1
t1[k1] = (t1[k1] or 0) + samples
if k2 then
local t2 = prof_count2
local t3 = t2[k1]
if not t3 then t3 = {}; t2[k1] = t3 end
t3[k2] = (t3[k2] or 0) + samples
end
end
end
------------------------------------------------------------------------------
-- Show top N list.
local function prof_top(count1, count2, samples, indent)
local t, n = {}, 0
for k in pairs(count1) do
n = n + 1
t[n] = k
end
sort(t, function(a, b) return count1[a] > count1[b] end)
for i=1,n do
local k = t[i]
local v = count1[k]
local pct = floor(v*100/samples + 0.5)
if pct < prof_min then break end
if not prof_raw then
out:write(format("%s%2d%% %s\n", indent, pct, k))
elseif prof_raw == "r" then
out:write(format("%s%5d %s\n", indent, v, k))
else
out:write(format("%s %d\n", k, v))
end
if count2 then
local r = count2[k]
if r then
prof_top(r, nil, v, (prof_split == 3 or prof_split == 1) and " -- " or
(prof_depth < 0 and " -> " or " <- "))
end
end
end
end
-- Annotate source code
local function prof_annotate(count1, samples)
local files = {}
local ms = 0
for k, v in pairs(count1) do
local pct = floor(v*100/samples + 0.5)
ms = math.max(ms, v)
if pct >= prof_min then
local file, line = k:match("^(.*):(%d+)$")
if not file then file = k; line = 0 end
local fl = files[file]
if not fl then fl = {}; files[file] = fl; files[#files+1] = file end
line = tonumber(line)
fl[line] = prof_raw and v or pct
end
end
sort(files)
local fmtv, fmtn = " %3d%% | %s\n", " | %s\n"
if prof_raw then
local n = math.max(5, math.ceil(math.log10(ms)))
fmtv = "%"..n.."d | %s\n"
fmtn = (" "):rep(n).." | %s\n"
end
local ann = prof_ann
for _, file in ipairs(files) do
local f0 = file:byte()
if f0 == 40 or f0 == 91 then
out:write(format("\n====== %s ======\n[Cannot annotate non-file]\n", file))
break
end
local fp, err = io.open(file)
if not fp then
out:write(format("====== ERROR: %s: %s\n", file, err))
break
end
out:write(format("\n====== %s ======\n", file))
local fl = files[file]
local n, show = 1, false
if ann ~= 0 then
for i=1,ann do
if fl[i] then show = true; out:write("@@ 1 @@\n"); break end
end
end
for line in fp:lines() do
if line:byte() == 27 then
out:write("[Cannot annotate bytecode file]\n")
break
end
local v = fl[n]
if ann ~= 0 then
local v2 = fl[n+ann]
if show then
if v2 then show = n+ann elseif v then show = n
elseif show+ann < n then show = false end
elseif v2 then
show = n+ann
out:write(format("@@ %d @@\n", n))
end
if not show then goto next end
end
if v then
out:write(format(fmtv, v, line))
else
out:write(format(fmtn, line))
end
::next::
n = n + 1
end
fp:close()
end
end
------------------------------------------------------------------------------
-- Finish profiling and dump result.
local function prof_finish()
if prof_ud then
profile.stop()
local samples = prof_samples
if samples == 0 then
if prof_raw ~= true then out:write("[No samples collected]\n") end
return
end
if prof_ann then
prof_annotate(prof_count1, samples)
else
prof_top(prof_count1, prof_count2, samples, "")
end
prof_count1 = nil
prof_count2 = nil
prof_ud = nil
if out ~= stdout then out:close() end
end
end
-- Start profiling.
local function prof_start(mode)
local interval = ""
mode = mode:gsub("i%d*", function(s) interval = s; return "" end)
prof_min = 3
mode = mode:gsub("m(%d+)", function(s) prof_min = tonumber(s); return "" end)
prof_depth = 1
mode = mode:gsub("%-?%d+", function(s) prof_depth = tonumber(s); return "" end)
local m = {}
for c in mode:gmatch(".") do m[c] = c end
prof_states = m.z or m.v
if prof_states == "z" then zone = require("jit.zone") end
local scope = m.l or m.f or m.F or (prof_states and "" or "f")
local flags = (m.p or "")
prof_raw = m.r
if m.s then
prof_split = 2
if prof_depth == -1 or m["-"] then prof_depth = -2
elseif prof_depth == 1 then prof_depth = 2 end
elseif mode:find("[fF].*l") then
scope = "l"
prof_split = 3
else
prof_split = (scope == "" or mode:find("[zv].*[lfF]")) and 1 or 0
end
prof_ann = m.A and 0 or (m.a and 3)
if prof_ann then
scope = "l"
prof_fmt = "pl"
prof_split = 0
prof_depth = 1
elseif m.G and scope ~= "" then
prof_fmt = flags..scope.."Z;"
prof_depth = -100
prof_raw = true
prof_min = 0
elseif scope == "" then
prof_fmt = false
else
local sc = prof_split == 3 and m.f or m.F or scope
prof_fmt = flags..sc..(prof_depth >= 0 and "Z < " or "Z > ")
end
prof_count1 = {}
prof_count2 = {}
prof_samples = 0
profile.start(scope:lower()..interval, prof_cb)
prof_ud = newproxy(true)
getmetatable(prof_ud).__gc = prof_finish
end
------------------------------------------------------------------------------
local function start(mode, outfile)
if not outfile then outfile = os.getenv("LUAJIT_PROFILEFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stdout
end
prof_start(mode or "f")
end
-- Public module functions.
return {
start = start, -- For -j command line option.
stop = prof_finish
}
@@ -0,0 +1,170 @@
----------------------------------------------------------------------------
-- Verbose mode of the LuaJIT compiler.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module shows verbose information about the progress of the
-- JIT compiler. It prints one line for each generated trace. This module
-- is useful to see which code has been compiled or where the compiler
-- punts and falls back to the interpreter.
--
-- Example usage:
--
-- luajit -jv -e "for i=1,1000 do for j=1,1000 do end end"
-- luajit -jv=myapp.out myapp.lua
--
-- Default output is to stderr. To redirect the output to a file, pass a
-- filename as an argument (use '-' for stdout) or set the environment
-- variable LUAJIT_VERBOSEFILE. The file is overwritten every time the
-- module is started.
--
-- The output from the first example should look like this:
--
-- [TRACE 1 (command line):1 loop]
-- [TRACE 2 (1/3) (command line):1 -> 1]
--
-- The first number in each line is the internal trace number. Next are
-- the file name ('(command line)') and the line number (':1') where the
-- trace has started. Side traces also show the parent trace number and
-- the exit number where they are attached to in parentheses ('(1/3)').
-- An arrow at the end shows where the trace links to ('-> 1'), unless
-- it loops to itself.
--
-- In this case the inner loop gets hot and is traced first, generating
-- a root trace. Then the last exit from the 1st trace gets hot, too,
-- and triggers generation of the 2nd trace. The side trace follows the
-- path along the outer loop and *around* the inner loop, back to its
-- start, and then links to the 1st trace. Yes, this may seem unusual,
-- if you know how traditional compilers work. Trace compilers are full
-- of surprises like this -- have fun! :-)
--
-- Aborted traces are shown like this:
--
-- [TRACE --- foo.lua:44 -- leaving loop in root trace at foo:lua:50]
--
-- Don't worry -- trace aborts are quite common, even in programs which
-- can be fully compiled. The compiler may retry several times until it
-- finds a suitable trace.
--
-- Of course this doesn't work with features that are not-yet-implemented
-- (NYI error messages). The VM simply falls back to the interpreter. This
-- may not matter at all if the particular trace is not very high up in
-- the CPU usage profile. Oh, and the interpreter is quite fast, too.
--
-- Also check out the -jdump module, which prints all the gory details.
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local funcinfo, traceinfo = jutil.funcinfo, jutil.traceinfo
local type, format = type, string.format
local stdout, stderr = io.stdout, io.stderr
-- Active flag and output file handle.
local active, out
------------------------------------------------------------------------------
local startloc, startex
local function fmtfunc(func, pc)
local fi = funcinfo(func, pc)
if fi.loc then
return fi.loc
elseif fi.ffid then
return vmdef.ffnames[fi.ffid]
elseif fi.addr then
return format("C:%x", fi.addr)
else
return "(?)"
end
end
-- Format trace error message.
local function fmterr(err, info)
if type(err) == "number" then
if type(info) == "function" then info = fmtfunc(info) end
err = format(vmdef.traceerr[err], info)
end
return err
end
-- Dump trace states.
local function dump_trace(what, tr, func, pc, otr, oex)
if what == "start" then
startloc = fmtfunc(func, pc)
startex = otr and "("..otr.."/"..(oex == -1 and "stitch" or oex)..") " or ""
else
if what == "abort" then
local loc = fmtfunc(func, pc)
if loc ~= startloc then
out:write(format("[TRACE --- %s%s -- %s at %s]\n",
startex, startloc, fmterr(otr, oex), loc))
else
out:write(format("[TRACE --- %s%s -- %s]\n",
startex, startloc, fmterr(otr, oex)))
end
elseif what == "stop" then
local info = traceinfo(tr)
local link, ltype = info.link, info.linktype
if ltype == "interpreter" then
out:write(format("[TRACE %3s %s%s -- fallback to interpreter]\n",
tr, startex, startloc))
elseif ltype == "stitch" then
out:write(format("[TRACE %3s %s%s %s %s]\n",
tr, startex, startloc, ltype, fmtfunc(func, pc)))
elseif link == tr or link == 0 then
out:write(format("[TRACE %3s %s%s %s]\n",
tr, startex, startloc, ltype))
elseif ltype == "root" then
out:write(format("[TRACE %3s %s%s -> %d]\n",
tr, startex, startloc, link))
else
out:write(format("[TRACE %3s %s%s -> %d %s]\n",
tr, startex, startloc, link, ltype))
end
else
out:write(format("[TRACE %s]\n", what))
end
out:flush()
end
end
------------------------------------------------------------------------------
-- Detach dump handlers.
local function dumpoff()
if active then
active = false
jit.attach(dump_trace)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach dump handlers.
local function dumpon(outfile)
if active then dumpoff() end
if not outfile then outfile = os.getenv("LUAJIT_VERBOSEFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stderr
end
jit.attach(dump_trace, "trace")
active = true
end
-- Public module functions.
return {
on = dumpon,
off = dumpoff,
start = dumpon -- For -j command line option.
}
@@ -0,0 +1,395 @@
-- This is a generated file. DO NOT EDIT!
return {
bcnames = "ISLT ISGE ISLE ISGT ISEQV ISNEV ISEQS ISNES ISEQN ISNEN ISEQP ISNEP ISTC ISFC IST ISF ISTYPEISNUM MOV NOT UNM LEN ADDVN SUBVN MULVN DIVVN MODVN ADDNV SUBNV MULNV DIVNV MODNV ADDVV SUBVV MULVV DIVVV MODVV POW CAT KSTR KCDATAKSHORTKNUM KPRI KNIL UGET USETV USETS USETN USETP UCLO FNEW TNEW TDUP GGET GSET TGETV TGETS TGETB TGETR TSETV TSETS TSETB TSETM TSETR CALLM CALL CALLMTCALLT ITERC ITERN VARG ISNEXTRETM RET RET0 RET1 FORI JFORI FORL IFORL JFORL ITERL IITERLJITERLLOOP ILOOP JLOOP JMP FUNCF IFUNCFJFUNCFFUNCV IFUNCVJFUNCVFUNCC FUNCCW",
irnames = "LT GE LE GT ULT UGE ULE UGT EQ NE ABC RETF NOP BASE PVAL GCSTEPHIOP LOOP USE PHI RENAMEPROF KPRI KINT KGC KPTR KKPTR KNULL KNUM KINT64KSLOT BNOT BSWAP BAND BOR BXOR BSHL BSHR BSAR BROL BROR ADD SUB MUL DIV MOD POW NEG ABS LDEXP MIN MAX FPMATHADDOV SUBOV MULOV AREF HREFK HREF NEWREFUREFO UREFC FREF TMPREFSTRREFLREF ALOAD HLOAD ULOAD FLOAD XLOAD SLOAD VLOAD ALEN ASTOREHSTOREUSTOREFSTOREXSTORESNEW XSNEW TNEW TDUP CNEW CNEWI BUFHDRBUFPUTBUFSTRTBAR OBAR XBAR CONV TOBIT TOSTR STRTO CALLN CALLA CALLL CALLS CALLXSCARG ",
irfpm = { [0]="floor", "ceil", "trunc", "sqrt", "log", "log2", "other", },
irfield = { [0]="str.len", "func.env", "func.pc", "func.ffid", "thread.env", "tab.meta", "tab.array", "tab.node", "tab.asize", "tab.hmask", "tab.nomm", "udata.meta", "udata.udtype", "udata.file", "sbuf.w", "sbuf.e", "sbuf.b", "sbuf.l", "sbuf.ref", "sbuf.r", "cdata.ctypeid", "cdata.ptr", "cdata.int", "cdata.int64", "cdata.int64_4", },
ircall = {
[0]="lj_str_cmp",
"lj_str_find",
"lj_str_new",
"lj_strscan_num",
"lj_strfmt_int",
"lj_strfmt_num",
"lj_strfmt_char",
"lj_strfmt_putint",
"lj_strfmt_putnum",
"lj_strfmt_putquoted",
"lj_strfmt_putfxint",
"lj_strfmt_putfnum_int",
"lj_strfmt_putfnum_uint",
"lj_strfmt_putfnum",
"lj_strfmt_putfstr",
"lj_strfmt_putfchar",
"lj_buf_putmem",
"lj_buf_putstr",
"lj_buf_putchar",
"lj_buf_putstr_reverse",
"lj_buf_putstr_lower",
"lj_buf_putstr_upper",
"lj_buf_putstr_rep",
"lj_buf_puttab",
"lj_bufx_set",
"lj_bufx_more",
"lj_serialize_put",
"lj_serialize_get",
"lj_serialize_encode",
"lj_serialize_decode",
"lj_buf_tostr",
"lj_tab_new_ah",
"lj_tab_new1",
"lj_tab_dup",
"lj_tab_clear",
"lj_tab_newkey",
"lj_tab_keyindex",
"lj_vm_next",
"lj_tab_len",
"lj_tab_len_hint",
"lj_gc_step_jit",
"lj_gc_barrieruv",
"lj_mem_newgco",
"lj_prng_u64d",
"lj_vm_modi",
"log10",
"exp",
"sin",
"cos",
"tan",
"asin",
"acos",
"atan",
"sinh",
"cosh",
"tanh",
"fputc",
"fwrite",
"fflush",
"lj_vm_floor",
"lj_vm_ceil",
"lj_vm_trunc",
"sqrt",
"log",
"lj_vm_log2",
"lj_vm_powi",
"pow",
"atan2",
"ldexp",
"lj_vm_tobit",
"softfp_add",
"softfp_sub",
"softfp_mul",
"softfp_div",
"softfp_cmp",
"softfp_i2d",
"softfp_d2i",
"lj_vm_sfmin",
"lj_vm_sfmax",
"lj_vm_tointg",
"softfp_ui2d",
"softfp_f2d",
"softfp_d2ui",
"softfp_d2f",
"softfp_i2f",
"softfp_ui2f",
"softfp_f2i",
"softfp_f2ui",
"fp64_l2d",
"fp64_ul2d",
"fp64_l2f",
"fp64_ul2f",
"fp64_d2l",
"fp64_d2ul",
"fp64_f2l",
"fp64_f2ul",
"lj_carith_divi64",
"lj_carith_divu64",
"lj_carith_modi64",
"lj_carith_modu64",
"lj_carith_powi64",
"lj_carith_powu64",
"lj_cdata_newv",
"lj_cdata_setfin",
"strlen",
"memcpy",
"memset",
"lj_vm_errno",
"lj_carith_mul64",
"lj_carith_shl64",
"lj_carith_shr64",
"lj_carith_sar64",
"lj_carith_rol64",
"lj_carith_ror64",
},
traceerr = {
[0]="error thrown or hook called during recording",
"trace too short",
"trace too long",
"trace too deep",
"too many snapshots",
"blacklisted",
"retry recording",
"NYI: bytecode %d",
"leaving loop in root trace",
"inner loop in root trace",
"loop unroll limit reached",
"bad argument type",
"JIT compilation disabled for function",
"call unroll limit reached",
"down-recursion, restarting",
"NYI: unsupported variant of FastFunc %s",
"NYI: return to lower frame",
"store with nil or NaN key",
"missing metamethod",
"looping index lookup",
"NYI: mixed sparse/dense table",
"symbol not in cache",
"NYI: unsupported C type conversion",
"NYI: unsupported C function type",
"guard would always fail",
"too many PHIs",
"persistent type instability",
"failed to allocate mcode memory",
"machine code too long",
"hit mcode limit (retrying)",
"too many spill slots",
"inconsistent register allocation",
"NYI: cannot assemble IR instruction %d",
"NYI: PHI shuffling too complex",
"NYI: register coalescing too complex",
},
ffnames = {
[0]="Lua",
"C",
"assert",
"type",
"next",
"pairs",
"ipairs_aux",
"ipairs",
"getmetatable",
"setmetatable",
"getfenv",
"setfenv",
"rawget",
"rawset",
"rawequal",
"unpack",
"select",
"tonumber",
"tostring",
"error",
"pcall",
"xpcall",
"loadfile",
"load",
"loadstring",
"dofile",
"gcinfo",
"collectgarbage",
"newproxy",
"print",
"coroutine.status",
"coroutine.running",
"coroutine.isyieldable",
"coroutine.create",
"coroutine.yield",
"coroutine.resume",
"coroutine.wrap_aux",
"coroutine.wrap",
"math.abs",
"math.floor",
"math.ceil",
"math.sqrt",
"math.log10",
"math.exp",
"math.sin",
"math.cos",
"math.tan",
"math.asin",
"math.acos",
"math.atan",
"math.sinh",
"math.cosh",
"math.tanh",
"math.frexp",
"math.modf",
"math.log",
"math.atan2",
"math.pow",
"math.fmod",
"math.ldexp",
"math.min",
"math.max",
"math.random",
"math.randomseed",
"bit.tobit",
"bit.bnot",
"bit.bswap",
"bit.lshift",
"bit.rshift",
"bit.arshift",
"bit.rol",
"bit.ror",
"bit.band",
"bit.bor",
"bit.bxor",
"bit.tohex",
"string.byte",
"string.char",
"string.sub",
"string.rep",
"string.reverse",
"string.lower",
"string.upper",
"string.dump",
"string.find",
"string.match",
"string.gmatch_aux",
"string.gmatch",
"string.gsub",
"string.format",
"table.maxn",
"table.insert",
"table.concat",
"table.sort",
"table.new",
"table.clear",
"io.method.close",
"io.method.read",
"io.method.write",
"io.method.flush",
"io.method.seek",
"io.method.setvbuf",
"io.method.lines",
"io.method.__gc",
"io.method.__tostring",
"io.open",
"io.popen",
"io.tmpfile",
"io.close",
"io.read",
"io.write",
"io.flush",
"io.input",
"io.output",
"io.lines",
"io.type",
"os.execute",
"os.remove",
"os.rename",
"os.tmpname",
"os.getenv",
"os.exit",
"os.clock",
"os.date",
"os.time",
"os.difftime",
"os.setlocale",
"debug.getregistry",
"debug.getmetatable",
"debug.setmetatable",
"debug.getfenv",
"debug.setfenv",
"debug.getinfo",
"debug.getlocal",
"debug.setlocal",
"debug.getupvalue",
"debug.setupvalue",
"debug.upvalueid",
"debug.upvaluejoin",
"debug.sethook",
"debug.gethook",
"debug.debug",
"debug.traceback",
"jit.on",
"jit.off",
"jit.flush",
"jit.status",
"jit.security",
"jit.attach",
"jit.util.funcinfo",
"jit.util.funcbc",
"jit.util.funck",
"jit.util.funcuvname",
"jit.util.traceinfo",
"jit.util.traceir",
"jit.util.tracek",
"jit.util.tracesnap",
"jit.util.tracemc",
"jit.util.traceexitstub",
"jit.util.ircalladdr",
"jit.opt.start",
"jit.profile.start",
"jit.profile.stop",
"jit.profile.dumpstack",
"ffi.meta.__index",
"ffi.meta.__newindex",
"ffi.meta.__eq",
"ffi.meta.__len",
"ffi.meta.__lt",
"ffi.meta.__le",
"ffi.meta.__concat",
"ffi.meta.__call",
"ffi.meta.__add",
"ffi.meta.__sub",
"ffi.meta.__mul",
"ffi.meta.__div",
"ffi.meta.__mod",
"ffi.meta.__pow",
"ffi.meta.__unm",
"ffi.meta.__tostring",
"ffi.meta.__pairs",
"ffi.meta.__ipairs",
"ffi.clib.__index",
"ffi.clib.__newindex",
"ffi.clib.__gc",
"ffi.callback.free",
"ffi.callback.set",
"ffi.cdef",
"ffi.new",
"ffi.cast",
"ffi.typeof",
"ffi.typeinfo",
"ffi.istype",
"ffi.sizeof",
"ffi.alignof",
"ffi.offsetof",
"ffi.errno",
"ffi.string",
"ffi.copy",
"ffi.fill",
"ffi.abi",
"ffi.metatype",
"ffi.gc",
"ffi.load",
"buffer.method.free",
"buffer.method.reset",
"buffer.method.skip",
"buffer.method.set",
"buffer.method.put",
"buffer.method.putf",
"buffer.method.get",
"buffer.method.putcdata",
"buffer.method.reserve",
"buffer.method.commit",
"buffer.method.ref",
"buffer.method.encode",
"buffer.method.decode",
"buffer.method.__gc",
"buffer.method.__tostring",
"buffer.method.__len",
"buffer.new",
"buffer.encode",
"buffer.decode",
},
}
@@ -0,0 +1,45 @@
----------------------------------------------------------------------------
-- LuaJIT profiler zones.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module implements a simple hierarchical zone model.
--
-- Example usage:
--
-- local zone = require("jit.zone")
-- zone("AI")
-- ...
-- zone("A*")
-- ...
-- print(zone:get()) --> "A*"
-- ...
-- zone()
-- ...
-- print(zone:get()) --> "AI"
-- ...
-- zone()
--
----------------------------------------------------------------------------
local remove = table.remove
return setmetatable({
flush = function(t)
for i=#t,1,-1 do t[i] = nil end
end,
get = function(t)
return t[#t]
end
}, {
__call = function(t, zone)
if zone then
t[#t+1] = zone
else
return (assert(remove(t), "empty zone stack"))
end
end
})
@@ -0,0 +1,190 @@
----------------------------------------------------------------------------
-- LuaJIT bytecode listing module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module lists the bytecode of a Lua function. If it's loaded by -jbc
-- it hooks into the parser and lists all functions of a chunk as they
-- are parsed.
--
-- Example usage:
--
-- luajit -jbc -e 'local x=0; for i=1,1e6 do x=x+i end; print(x)'
-- luajit -jbc=- foo.lua
-- luajit -jbc=foo.list foo.lua
--
-- Default output is to stderr. To redirect the output to a file, pass a
-- filename as an argument (use '-' for stdout) or set the environment
-- variable LUAJIT_LISTFILE. The file is overwritten every time the module
-- is started.
--
-- This module can also be used programmatically:
--
-- local bc = require("jit.bc")
--
-- local function foo() print("hello") end
--
-- bc.dump(foo) --> -- BYTECODE -- [...]
-- print(bc.line(foo, 2)) --> 0002 KSTR 1 1 ; "hello"
--
-- local out = {
-- -- Do something with each line:
-- write = function(t, ...) io.write(...) end,
-- close = function(t) end,
-- flush = function(t) end,
-- }
-- bc.dump(foo, out)
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local bit = require("bit")
local sub, gsub, format = string.sub, string.gsub, string.format
local byte, band, shr = string.byte, bit.band, bit.rshift
local funcinfo, funcbc, funck = jutil.funcinfo, jutil.funcbc, jutil.funck
local funcuvname = jutil.funcuvname
local bcnames = vmdef.bcnames
local stdout, stderr = io.stdout, io.stderr
------------------------------------------------------------------------------
local function ctlsub(c)
if c == "\n" then return "\\n"
elseif c == "\r" then return "\\r"
elseif c == "\t" then return "\\t"
else return format("\\%03d", byte(c))
end
end
-- Return one bytecode line.
local function bcline(func, pc, prefix)
local ins, m = funcbc(func, pc)
if not ins then return end
local ma, mb, mc = band(m, 7), band(m, 15*8), band(m, 15*128)
local a = band(shr(ins, 8), 0xff)
local oidx = 6*band(ins, 0xff)
local op = sub(bcnames, oidx+1, oidx+6)
local s = format("%04d %s %-6s %3s ",
pc, prefix or " ", op, ma == 0 and "" or a)
local d = shr(ins, 16)
if mc == 13*128 then -- BCMjump
return format("%s=> %04d\n", s, pc+d-0x7fff)
end
if mb ~= 0 then
d = band(d, 0xff)
elseif mc == 0 then
return s.."\n"
end
local kc
if mc == 10*128 then -- BCMstr
kc = funck(func, -d-1)
kc = format(#kc > 40 and '"%.40s"~' or '"%s"', gsub(kc, "%c", ctlsub))
elseif mc == 9*128 then -- BCMnum
kc = funck(func, d)
if op == "TSETM " then kc = kc - 2^52 end
elseif mc == 12*128 then -- BCMfunc
local fi = funcinfo(funck(func, -d-1))
if fi.ffid then
kc = vmdef.ffnames[fi.ffid]
else
kc = fi.loc
end
elseif mc == 5*128 then -- BCMuv
kc = funcuvname(func, d)
end
if ma == 5 then -- BCMuv
local ka = funcuvname(func, a)
if kc then kc = ka.." ; "..kc else kc = ka end
end
if mb ~= 0 then
local b = shr(ins, 24)
if kc then return format("%s%3d %3d ; %s\n", s, b, d, kc) end
return format("%s%3d %3d\n", s, b, d)
end
if kc then return format("%s%3d ; %s\n", s, d, kc) end
if mc == 7*128 and d > 32767 then d = d - 65536 end -- BCMlits
return format("%s%3d\n", s, d)
end
-- Collect branch targets of a function.
local function bctargets(func)
local target = {}
for pc=1,1000000000 do
local ins, m = funcbc(func, pc)
if not ins then break end
if band(m, 15*128) == 13*128 then target[pc+shr(ins, 16)-0x7fff] = true end
end
return target
end
-- Dump bytecode instructions of a function.
local function bcdump(func, out, all)
if not out then out = stdout end
local fi = funcinfo(func)
if all and fi.children then
for n=-1,-1000000000,-1 do
local k = funck(func, n)
if not k then break end
if type(k) == "proto" then bcdump(k, out, true) end
end
end
out:write(format("-- BYTECODE -- %s-%d\n", fi.loc, fi.lastlinedefined))
local target = bctargets(func)
for pc=1,1000000000 do
local s = bcline(func, pc, target[pc] and "=>")
if not s then break end
out:write(s)
end
out:write("\n")
out:flush()
end
------------------------------------------------------------------------------
-- Active flag and output file handle.
local active, out
-- List handler.
local function h_list(func)
return bcdump(func, out)
end
-- Detach list handler.
local function bclistoff()
if active then
active = false
jit.attach(h_list)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach list handler.
local function bcliston(outfile)
if active then bclistoff() end
if not outfile then outfile = os.getenv("LUAJIT_LISTFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stderr
end
jit.attach(h_list, "bc")
active = true
end
-- Public module functions.
return {
line = bcline,
dump = bcdump,
targets = bctargets,
on = bcliston,
off = bclistoff,
start = bcliston -- For -j command line option.
}
@@ -0,0 +1,665 @@
----------------------------------------------------------------------------
-- LuaJIT module to save/list bytecode.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module saves or lists the bytecode for an input file.
-- It's run by the -b command line option.
--
------------------------------------------------------------------------------
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local bit = require("bit")
-- Symbol name prefix for LuaJIT bytecode.
local LJBC_PREFIX = "luaJIT_BC_"
local type, assert = type, assert
local format = string.format
local tremove, tconcat = table.remove, table.concat
------------------------------------------------------------------------------
local function usage()
io.stderr:write[[
Save LuaJIT bytecode: luajit -b[options] input output
-l Only list bytecode.
-s Strip debug info (default).
-g Keep debug info.
-n name Set module name (default: auto-detect from input name).
-t type Set output file type (default: auto-detect from output name).
-a arch Override architecture for object files (default: native).
-o os Override OS for object files (default: native).
-e chunk Use chunk string as input.
-- Stop handling options.
- Use stdin as input and/or stdout as output.
File types: c h obj o raw (default)
]]
os.exit(1)
end
local function check(ok, ...)
if ok then return ok, ... end
io.stderr:write("luajit: ", ...)
io.stderr:write("\n")
os.exit(1)
end
local function readfile(input)
if type(input) == "function" then return input end
if input == "-" then input = nil end
return check(loadfile(input))
end
local function savefile(name, mode)
if name == "-" then return io.stdout end
return check(io.open(name, mode))
end
------------------------------------------------------------------------------
local map_type = {
raw = "raw", c = "c", h = "h", o = "obj", obj = "obj",
}
local map_arch = {
x86 = { e = "le", b = 32, m = 3, p = 0x14c, },
x64 = { e = "le", b = 64, m = 62, p = 0x8664, },
arm = { e = "le", b = 32, m = 40, p = 0x1c0, },
arm64 = { e = "le", b = 64, m = 183, p = 0xaa64, },
arm64be = { e = "be", b = 64, m = 183, },
ppc = { e = "be", b = 32, m = 20, },
mips = { e = "be", b = 32, m = 8, f = 0x50001006, },
mipsel = { e = "le", b = 32, m = 8, f = 0x50001006, },
mips64 = { e = "be", b = 64, m = 8, f = 0x80000007, },
mips64el = { e = "le", b = 64, m = 8, f = 0x80000007, },
mips64r6 = { e = "be", b = 64, m = 8, f = 0xa0000407, },
mips64r6el = { e = "le", b = 64, m = 8, f = 0xa0000407, },
}
local map_os = {
linux = true, windows = true, osx = true, freebsd = true, netbsd = true,
openbsd = true, dragonfly = true, solaris = true,
}
local function checkarg(str, map, err)
str = str:lower()
local s = check(map[str], "unknown ", err)
return type(s) == "string" and s or str
end
local function detecttype(str)
local ext = str:lower():match("%.(%a+)$")
return map_type[ext] or "raw"
end
local function checkmodname(str)
check(str:match("^[%w_.%-]+$"), "bad module name")
return str:gsub("[%.%-]", "_")
end
local function detectmodname(str)
if type(str) == "string" then
local tail = str:match("[^/\\]+$")
if tail then str = tail end
local head = str:match("^(.*)%.[^.]*$")
if head then str = head end
str = str:match("^[%w_.%-]+")
else
str = nil
end
check(str, "cannot derive module name, use -n name")
return str:gsub("[%.%-]", "_")
end
------------------------------------------------------------------------------
local function bcsave_tail(fp, output, s)
local ok, err = fp:write(s)
if ok and output ~= "-" then ok, err = fp:close() end
check(ok, "cannot write ", output, ": ", err)
end
local function bcsave_raw(output, s)
local fp = savefile(output, "wb")
bcsave_tail(fp, output, s)
end
local function bcsave_c(ctx, output, s)
local fp = savefile(output, "w")
if ctx.type == "c" then
fp:write(format([[
#ifdef __cplusplus
extern "C"
#endif
#ifdef _WIN32
__declspec(dllexport)
#endif
const unsigned char %s%s[] = {
]], LJBC_PREFIX, ctx.modname))
else
fp:write(format([[
#define %s%s_SIZE %d
static const unsigned char %s%s[] = {
]], LJBC_PREFIX, ctx.modname, #s, LJBC_PREFIX, ctx.modname))
end
local t, n, m = {}, 0, 0
for i=1,#s do
local b = tostring(string.byte(s, i))
m = m + #b + 1
if m > 78 then
fp:write(tconcat(t, ",", 1, n), ",\n")
n, m = 0, #b + 1
end
n = n + 1
t[n] = b
end
bcsave_tail(fp, output, tconcat(t, ",", 1, n).."\n};\n")
end
local function bcsave_elfobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct {
uint8_t emagic[4], eclass, eendian, eversion, eosabi, eabiversion, epad[7];
uint16_t type, machine;
uint32_t version;
uint32_t entry, phofs, shofs;
uint32_t flags;
uint16_t ehsize, phentsize, phnum, shentsize, shnum, shstridx;
} ELF32header;
typedef struct {
uint8_t emagic[4], eclass, eendian, eversion, eosabi, eabiversion, epad[7];
uint16_t type, machine;
uint32_t version;
uint64_t entry, phofs, shofs;
uint32_t flags;
uint16_t ehsize, phentsize, phnum, shentsize, shnum, shstridx;
} ELF64header;
typedef struct {
uint32_t name, type, flags, addr, ofs, size, link, info, align, entsize;
} ELF32sectheader;
typedef struct {
uint32_t name, type;
uint64_t flags, addr, ofs, size;
uint32_t link, info;
uint64_t align, entsize;
} ELF64sectheader;
typedef struct {
uint32_t name, value, size;
uint8_t info, other;
uint16_t sectidx;
} ELF32symbol;
typedef struct {
uint32_t name;
uint8_t info, other;
uint16_t sectidx;
uint64_t value, size;
} ELF64symbol;
typedef struct {
ELF32header hdr;
ELF32sectheader sect[6];
ELF32symbol sym[2];
uint8_t space[4096];
} ELF32obj;
typedef struct {
ELF64header hdr;
ELF64sectheader sect[6];
ELF64symbol sym[2];
uint8_t space[4096];
} ELF64obj;
]]
local symname = LJBC_PREFIX..ctx.modname
local ai = assert(map_arch[ctx.arch])
local is64, isbe = ai.b == 64, ai.e == "be"
-- Handle different host/target endianess.
local function f32(x) return x end
local f16, fofs = f32, f32
if ffi.abi("be") ~= isbe then
f32 = bit.bswap
function f16(x) return bit.rshift(bit.bswap(x), 16) end
if is64 then
local two32 = ffi.cast("int64_t", 2^32)
function fofs(x) return bit.bswap(x)*two32 end
else
fofs = f32
end
end
-- Create ELF object and fill in header.
local o = ffi.new(is64 and "ELF64obj" or "ELF32obj")
local hdr = o.hdr
if ctx.os == "bsd" or ctx.os == "other" then -- Determine native hdr.eosabi.
local bf = assert(io.open("/bin/ls", "rb"))
local bs = bf:read(9)
bf:close()
ffi.copy(o, bs, 9)
check(hdr.emagic[0] == 127, "no support for writing native object files")
else
hdr.emagic = "\127ELF"
hdr.eosabi = ({ freebsd=9, netbsd=2, openbsd=12, solaris=6 })[ctx.os] or 0
end
hdr.eclass = is64 and 2 or 1
hdr.eendian = isbe and 2 or 1
hdr.eversion = 1
hdr.type = f16(1)
hdr.machine = f16(ai.m)
hdr.flags = f32(ai.f or 0)
hdr.version = f32(1)
hdr.shofs = fofs(ffi.offsetof(o, "sect"))
hdr.ehsize = f16(ffi.sizeof(hdr))
hdr.shentsize = f16(ffi.sizeof(o.sect[0]))
hdr.shnum = f16(6)
hdr.shstridx = f16(2)
-- Fill in sections and symbols.
local sofs, ofs = ffi.offsetof(o, "space"), 1
for i,name in ipairs{
".symtab", ".shstrtab", ".strtab", ".rodata", ".note.GNU-stack",
} do
local sect = o.sect[i]
sect.align = fofs(1)
sect.name = f32(ofs)
ffi.copy(o.space+ofs, name)
ofs = ofs + #name+1
end
o.sect[1].type = f32(2) -- .symtab
o.sect[1].link = f32(3)
o.sect[1].info = f32(1)
o.sect[1].align = fofs(8)
o.sect[1].ofs = fofs(ffi.offsetof(o, "sym"))
o.sect[1].entsize = fofs(ffi.sizeof(o.sym[0]))
o.sect[1].size = fofs(ffi.sizeof(o.sym))
o.sym[1].name = f32(1)
o.sym[1].sectidx = f16(4)
o.sym[1].size = fofs(#s)
o.sym[1].info = 17
o.sect[2].type = f32(3) -- .shstrtab
o.sect[2].ofs = fofs(sofs)
o.sect[2].size = fofs(ofs)
o.sect[3].type = f32(3) -- .strtab
o.sect[3].ofs = fofs(sofs + ofs)
o.sect[3].size = fofs(#symname+2)
ffi.copy(o.space+ofs+1, symname)
ofs = ofs + #symname + 2
o.sect[4].type = f32(1) -- .rodata
o.sect[4].flags = fofs(2)
o.sect[4].ofs = fofs(sofs + ofs)
o.sect[4].size = fofs(#s)
o.sect[5].type = f32(1) -- .note.GNU-stack
o.sect[5].ofs = fofs(sofs + ofs + #s)
-- Write ELF object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, ffi.sizeof(o)-4096+ofs))
bcsave_tail(fp, output, s)
end
local function bcsave_peobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct {
uint16_t arch, nsects;
uint32_t time, symtabofs, nsyms;
uint16_t opthdrsz, flags;
} PEheader;
typedef struct {
char name[8];
uint32_t vsize, vaddr, size, ofs, relocofs, lineofs;
uint16_t nreloc, nline;
uint32_t flags;
} PEsection;
typedef struct __attribute((packed)) {
union {
char name[8];
uint32_t nameref[2];
};
uint32_t value;
int16_t sect;
uint16_t type;
uint8_t scl, naux;
} PEsym;
typedef struct __attribute((packed)) {
uint32_t size;
uint16_t nreloc, nline;
uint32_t cksum;
uint16_t assoc;
uint8_t comdatsel, unused[3];
} PEsymaux;
typedef struct {
PEheader hdr;
PEsection sect[2];
// Must be an even number of symbol structs.
PEsym sym0;
PEsymaux sym0aux;
PEsym sym1;
PEsymaux sym1aux;
PEsym sym2;
PEsym sym3;
uint32_t strtabsize;
uint8_t space[4096];
} PEobj;
]]
local symname = LJBC_PREFIX..ctx.modname
local ai = assert(map_arch[ctx.arch])
local is64 = ai.b == 64
local symexport = " /EXPORT:"..symname..",DATA "
-- The file format is always little-endian. Swap if the host is big-endian.
local function f32(x) return x end
local f16 = f32
if ffi.abi("be") then
f32 = bit.bswap
function f16(x) return bit.rshift(bit.bswap(x), 16) end
end
-- Create PE object and fill in header.
local o = ffi.new("PEobj")
local hdr = o.hdr
hdr.arch = f16(assert(ai.p))
hdr.nsects = f16(2)
hdr.symtabofs = f32(ffi.offsetof(o, "sym0"))
hdr.nsyms = f32(6)
-- Fill in sections and symbols.
o.sect[0].name = ".drectve"
o.sect[0].size = f32(#symexport)
o.sect[0].flags = f32(0x00100a00)
o.sym0.sect = f16(1)
o.sym0.scl = 3
o.sym0.name = ".drectve"
o.sym0.naux = 1
o.sym0aux.size = f32(#symexport)
o.sect[1].name = ".rdata"
o.sect[1].size = f32(#s)
o.sect[1].flags = f32(0x40300040)
o.sym1.sect = f16(2)
o.sym1.scl = 3
o.sym1.name = ".rdata"
o.sym1.naux = 1
o.sym1aux.size = f32(#s)
o.sym2.sect = f16(2)
o.sym2.scl = 2
o.sym2.nameref[1] = f32(4)
o.sym3.sect = f16(-1)
o.sym3.scl = 2
o.sym3.value = f32(1)
o.sym3.name = "@feat.00" -- Mark as SafeSEH compliant.
ffi.copy(o.space, symname)
local ofs = #symname + 1
o.strtabsize = f32(ofs + 4)
o.sect[0].ofs = f32(ffi.offsetof(o, "space") + ofs)
ffi.copy(o.space + ofs, symexport)
ofs = ofs + #symexport
o.sect[1].ofs = f32(ffi.offsetof(o, "space") + ofs)
-- Write PE object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, ffi.sizeof(o)-4096+ofs))
bcsave_tail(fp, output, s)
end
local function bcsave_machobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct
{
uint32_t magic, cputype, cpusubtype, filetype, ncmds, sizeofcmds, flags;
} mach_header;
typedef struct
{
mach_header; uint32_t reserved;
} mach_header_64;
typedef struct {
uint32_t cmd, cmdsize;
char segname[16];
uint32_t vmaddr, vmsize, fileoff, filesize;
uint32_t maxprot, initprot, nsects, flags;
} mach_segment_command;
typedef struct {
uint32_t cmd, cmdsize;
char segname[16];
uint64_t vmaddr, vmsize, fileoff, filesize;
uint32_t maxprot, initprot, nsects, flags;
} mach_segment_command_64;
typedef struct {
char sectname[16], segname[16];
uint32_t addr, size;
uint32_t offset, align, reloff, nreloc, flags;
uint32_t reserved1, reserved2;
} mach_section;
typedef struct {
char sectname[16], segname[16];
uint64_t addr, size;
uint32_t offset, align, reloff, nreloc, flags;
uint32_t reserved1, reserved2, reserved3;
} mach_section_64;
typedef struct {
uint32_t cmd, cmdsize, symoff, nsyms, stroff, strsize;
} mach_symtab_command;
typedef struct {
int32_t strx;
uint8_t type, sect;
int16_t desc;
uint32_t value;
} mach_nlist;
typedef struct {
uint32_t strx;
uint8_t type, sect;
uint16_t desc;
uint64_t value;
} mach_nlist_64;
typedef struct
{
uint32_t magic, nfat_arch;
} mach_fat_header;
typedef struct
{
uint32_t cputype, cpusubtype, offset, size, align;
} mach_fat_arch;
typedef struct {
struct {
mach_header hdr;
mach_segment_command seg;
mach_section sec;
mach_symtab_command sym;
} arch[1];
mach_nlist sym_entry;
uint8_t space[4096];
} mach_obj;
typedef struct {
struct {
mach_header_64 hdr;
mach_segment_command_64 seg;
mach_section_64 sec;
mach_symtab_command sym;
} arch[1];
mach_nlist_64 sym_entry;
uint8_t space[4096];
} mach_obj_64;
typedef struct {
mach_fat_header fat;
mach_fat_arch fat_arch[2];
struct {
mach_header hdr;
mach_segment_command seg;
mach_section sec;
mach_symtab_command sym;
} arch[2];
mach_nlist sym_entry;
uint8_t space[4096];
} mach_fat_obj;
]]
local symname = '_'..LJBC_PREFIX..ctx.modname
local isfat, is64, align, mobj = false, false, 4, "mach_obj"
if ctx.arch == "x64" then
is64, align, mobj = true, 8, "mach_obj_64"
elseif ctx.arch == "arm" then
isfat, mobj = true, "mach_fat_obj"
elseif ctx.arch == "arm64" then
is64, align, isfat, mobj = true, 8, true, "mach_fat_obj"
else
check(ctx.arch == "x86", "unsupported architecture for OSX")
end
local function aligned(v, a) return bit.band(v+a-1, -a) end
local be32 = bit.bswap -- Mach-O FAT is BE, supported archs are LE.
-- Create Mach-O object and fill in header.
local o = ffi.new(mobj)
local mach_size = aligned(ffi.offsetof(o, "space")+#symname+2, align)
local cputype = ({ x86={7}, x64={0x01000007}, arm={7,12}, arm64={0x01000007,0x0100000c} })[ctx.arch]
local cpusubtype = ({ x86={3}, x64={3}, arm={3,9}, arm64={3,0} })[ctx.arch]
if isfat then
o.fat.magic = be32(0xcafebabe)
o.fat.nfat_arch = be32(#cpusubtype)
end
-- Fill in sections and symbols.
for i=0,#cpusubtype-1 do
local ofs = 0
if isfat then
local a = o.fat_arch[i]
a.cputype = be32(cputype[i+1])
a.cpusubtype = be32(cpusubtype[i+1])
-- Subsequent slices overlap each other to share data.
ofs = ffi.offsetof(o, "arch") + i*ffi.sizeof(o.arch[0])
a.offset = be32(ofs)
a.size = be32(mach_size-ofs+#s)
end
local a = o.arch[i]
a.hdr.magic = is64 and 0xfeedfacf or 0xfeedface
a.hdr.cputype = cputype[i+1]
a.hdr.cpusubtype = cpusubtype[i+1]
a.hdr.filetype = 1
a.hdr.ncmds = 2
a.hdr.sizeofcmds = ffi.sizeof(a.seg)+ffi.sizeof(a.sec)+ffi.sizeof(a.sym)
a.seg.cmd = is64 and 0x19 or 0x1
a.seg.cmdsize = ffi.sizeof(a.seg)+ffi.sizeof(a.sec)
a.seg.vmsize = #s
a.seg.fileoff = mach_size-ofs
a.seg.filesize = #s
a.seg.maxprot = 1
a.seg.initprot = 1
a.seg.nsects = 1
ffi.copy(a.sec.sectname, "__data")
ffi.copy(a.sec.segname, "__DATA")
a.sec.size = #s
a.sec.offset = mach_size-ofs
a.sym.cmd = 2
a.sym.cmdsize = ffi.sizeof(a.sym)
a.sym.symoff = ffi.offsetof(o, "sym_entry")-ofs
a.sym.nsyms = 1
a.sym.stroff = ffi.offsetof(o, "sym_entry")+ffi.sizeof(o.sym_entry)-ofs
a.sym.strsize = aligned(#symname+2, align)
end
o.sym_entry.type = 0xf
o.sym_entry.sect = 1
o.sym_entry.strx = 1
ffi.copy(o.space+1, symname)
-- Write Macho-O object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, mach_size))
bcsave_tail(fp, output, s)
end
local function bcsave_obj(ctx, output, s)
local ok, ffi = pcall(require, "ffi")
check(ok, "FFI library required to write this file type")
if ctx.os == "windows" then
return bcsave_peobj(ctx, output, s, ffi)
elseif ctx.os == "osx" then
return bcsave_machobj(ctx, output, s, ffi)
else
return bcsave_elfobj(ctx, output, s, ffi)
end
end
------------------------------------------------------------------------------
local function bclist(input, output)
local f = readfile(input)
require("jit.bc").dump(f, savefile(output, "w"), true)
end
local function bcsave(ctx, input, output)
local f = readfile(input)
local s = string.dump(f, ctx.strip)
local t = ctx.type
if not t then
t = detecttype(output)
ctx.type = t
end
if t == "raw" then
bcsave_raw(output, s)
else
if not ctx.modname then ctx.modname = detectmodname(input) end
if t == "obj" then
bcsave_obj(ctx, output, s)
else
bcsave_c(ctx, output, s)
end
end
end
local function docmd(...)
local arg = {...}
local n = 1
local list = false
local ctx = {
strip = true, arch = jit.arch, os = jit.os:lower(),
type = false, modname = false,
}
while n <= #arg do
local a = arg[n]
if type(a) == "string" and a:sub(1, 1) == "-" and a ~= "-" then
tremove(arg, n)
if a == "--" then break end
for m=2,#a do
local opt = a:sub(m, m)
if opt == "l" then
list = true
elseif opt == "s" then
ctx.strip = true
elseif opt == "g" then
ctx.strip = false
else
if arg[n] == nil or m ~= #a then usage() end
if opt == "e" then
if n ~= 1 then usage() end
arg[1] = check(loadstring(arg[1]))
elseif opt == "n" then
ctx.modname = checkmodname(tremove(arg, n))
elseif opt == "t" then
ctx.type = checkarg(tremove(arg, n), map_type, "file type")
elseif opt == "a" then
ctx.arch = checkarg(tremove(arg, n), map_arch, "architecture")
elseif opt == "o" then
ctx.os = checkarg(tremove(arg, n), map_os, "OS name")
else
usage()
end
end
end
else
n = n + 1
end
end
if list then
if #arg == 0 or #arg > 2 then usage() end
bclist(arg[1], arg[2] or "-")
else
if #arg ~= 2 then usage() end
bcsave(ctx, arg[1], arg[2])
end
end
------------------------------------------------------------------------------
-- Public module functions.
return {
start = docmd -- Process -b command line option.
}
@@ -0,0 +1,689 @@
----------------------------------------------------------------------------
-- LuaJIT ARM disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles most user-mode ARMv7 instructions
-- NYI: Advanced SIMD and VFP instructions.
------------------------------------------------------------------------------
local type = type
local sub, byte, format = string.sub, string.byte, string.format
local match, gmatch = string.match, string.gmatch
local concat = table.concat
local bit = require("bit")
local band, bor, ror, tohex = bit.band, bit.bor, bit.ror, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Opcode maps
------------------------------------------------------------------------------
local map_loadc = {
shift = 8, mask = 15,
[10] = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "vmovFmDN", "vstmFNdr",
_ = {
shift = 21, mask = 1,
[0] = "vstrFdl",
{ shift = 16, mask = 15, [13] = "vpushFdr", _ = "vstmdbFNdr", }
},
},
{
shift = 23, mask = 3,
[0] = "vmovFDNm",
{ shift = 16, mask = 15, [13] = "vpopFdr", _ = "vldmFNdr", },
_ = {
shift = 21, mask = 1,
[0] = "vldrFdl", "vldmdbFNdr",
},
},
},
[11] = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "vmovGmDN", "vstmGNdr",
_ = {
shift = 21, mask = 1,
[0] = "vstrGdl",
{ shift = 16, mask = 15, [13] = "vpushGdr", _ = "vstmdbGNdr", }
},
},
{
shift = 23, mask = 3,
[0] = "vmovGDNm",
{ shift = 16, mask = 15, [13] = "vpopGdr", _ = "vldmGNdr", },
_ = {
shift = 21, mask = 1,
[0] = "vldrGdl", "vldmdbGNdr",
},
},
},
_ = {
shift = 0, mask = 0 -- NYI ldc, mcrr, mrrc.
},
}
local map_vfps = {
shift = 6, mask = 0x2c001,
[0] = "vmlaF.dnm", "vmlsF.dnm",
[0x04000] = "vnmlsF.dnm", [0x04001] = "vnmlaF.dnm",
[0x08000] = "vmulF.dnm", [0x08001] = "vnmulF.dnm",
[0x0c000] = "vaddF.dnm", [0x0c001] = "vsubF.dnm",
[0x20000] = "vdivF.dnm",
[0x24000] = "vfnmsF.dnm", [0x24001] = "vfnmaF.dnm",
[0x28000] = "vfmaF.dnm", [0x28001] = "vfmsF.dnm",
[0x2c000] = "vmovF.dY",
[0x2c001] = {
shift = 7, mask = 0x1e01,
[0] = "vmovF.dm", "vabsF.dm",
[0x0200] = "vnegF.dm", [0x0201] = "vsqrtF.dm",
[0x0800] = "vcmpF.dm", [0x0801] = "vcmpeF.dm",
[0x0a00] = "vcmpzF.d", [0x0a01] = "vcmpzeF.d",
[0x0e01] = "vcvtG.dF.m",
[0x1000] = "vcvt.f32.u32Fdm", [0x1001] = "vcvt.f32.s32Fdm",
[0x1800] = "vcvtr.u32F.dm", [0x1801] = "vcvt.u32F.dm",
[0x1a00] = "vcvtr.s32F.dm", [0x1a01] = "vcvt.s32F.dm",
},
}
local map_vfpd = {
shift = 6, mask = 0x2c001,
[0] = "vmlaG.dnm", "vmlsG.dnm",
[0x04000] = "vnmlsG.dnm", [0x04001] = "vnmlaG.dnm",
[0x08000] = "vmulG.dnm", [0x08001] = "vnmulG.dnm",
[0x0c000] = "vaddG.dnm", [0x0c001] = "vsubG.dnm",
[0x20000] = "vdivG.dnm",
[0x24000] = "vfnmsG.dnm", [0x24001] = "vfnmaG.dnm",
[0x28000] = "vfmaG.dnm", [0x28001] = "vfmsG.dnm",
[0x2c000] = "vmovG.dY",
[0x2c001] = {
shift = 7, mask = 0x1e01,
[0] = "vmovG.dm", "vabsG.dm",
[0x0200] = "vnegG.dm", [0x0201] = "vsqrtG.dm",
[0x0800] = "vcmpG.dm", [0x0801] = "vcmpeG.dm",
[0x0a00] = "vcmpzG.d", [0x0a01] = "vcmpzeG.d",
[0x0e01] = "vcvtF.dG.m",
[0x1000] = "vcvt.f64.u32GdFm", [0x1001] = "vcvt.f64.s32GdFm",
[0x1800] = "vcvtr.u32FdG.m", [0x1801] = "vcvt.u32FdG.m",
[0x1a00] = "vcvtr.s32FdG.m", [0x1a01] = "vcvt.s32FdG.m",
},
}
local map_datac = {
shift = 24, mask = 1,
[0] = {
shift = 4, mask = 1,
[0] = {
shift = 8, mask = 15,
[10] = map_vfps,
[11] = map_vfpd,
-- NYI cdp, mcr, mrc.
},
{
shift = 8, mask = 15,
[10] = {
shift = 20, mask = 15,
[0] = "vmovFnD", "vmovFDn",
[14] = "vmsrD",
[15] = { shift = 12, mask = 15, [15] = "vmrs", _ = "vmrsD", },
},
},
},
"svcT",
}
local map_loadcu = {
shift = 0, mask = 0, -- NYI unconditional CP load/store.
}
local map_datacu = {
shift = 0, mask = 0, -- NYI unconditional CP data.
}
local map_simddata = {
shift = 0, mask = 0, -- NYI SIMD data.
}
local map_simdload = {
shift = 0, mask = 0, -- NYI SIMD load/store, preload.
}
local map_preload = {
shift = 0, mask = 0, -- NYI preload.
}
local map_media = {
shift = 20, mask = 31,
[0] = false,
{ --01
shift = 5, mask = 7,
[0] = "sadd16DNM", "sasxDNM", "ssaxDNM", "ssub16DNM",
"sadd8DNM", false, false, "ssub8DNM",
},
{ --02
shift = 5, mask = 7,
[0] = "qadd16DNM", "qasxDNM", "qsaxDNM", "qsub16DNM",
"qadd8DNM", false, false, "qsub8DNM",
},
{ --03
shift = 5, mask = 7,
[0] = "shadd16DNM", "shasxDNM", "shsaxDNM", "shsub16DNM",
"shadd8DNM", false, false, "shsub8DNM",
},
false,
{ --05
shift = 5, mask = 7,
[0] = "uadd16DNM", "uasxDNM", "usaxDNM", "usub16DNM",
"uadd8DNM", false, false, "usub8DNM",
},
{ --06
shift = 5, mask = 7,
[0] = "uqadd16DNM", "uqasxDNM", "uqsaxDNM", "uqsub16DNM",
"uqadd8DNM", false, false, "uqsub8DNM",
},
{ --07
shift = 5, mask = 7,
[0] = "uhadd16DNM", "uhasxDNM", "uhsaxDNM", "uhsub16DNM",
"uhadd8DNM", false, false, "uhsub8DNM",
},
{ --08
shift = 5, mask = 7,
[0] = "pkhbtDNMU", false, "pkhtbDNMU",
{ shift = 16, mask = 15, [15] = "sxtb16DMU", _ = "sxtab16DNMU", },
"pkhbtDNMU", "selDNM", "pkhtbDNMU",
},
false,
{ --0a
shift = 5, mask = 7,
[0] = "ssatDxMu", "ssat16DxM", "ssatDxMu",
{ shift = 16, mask = 15, [15] = "sxtbDMU", _ = "sxtabDNMU", },
"ssatDxMu", false, "ssatDxMu",
},
{ --0b
shift = 5, mask = 7,
[0] = "ssatDxMu", "revDM", "ssatDxMu",
{ shift = 16, mask = 15, [15] = "sxthDMU", _ = "sxtahDNMU", },
"ssatDxMu", "rev16DM", "ssatDxMu",
},
{ --0c
shift = 5, mask = 7,
[3] = { shift = 16, mask = 15, [15] = "uxtb16DMU", _ = "uxtab16DNMU", },
},
false,
{ --0e
shift = 5, mask = 7,
[0] = "usatDwMu", "usat16DwM", "usatDwMu",
{ shift = 16, mask = 15, [15] = "uxtbDMU", _ = "uxtabDNMU", },
"usatDwMu", false, "usatDwMu",
},
{ --0f
shift = 5, mask = 7,
[0] = "usatDwMu", "rbitDM", "usatDwMu",
{ shift = 16, mask = 15, [15] = "uxthDMU", _ = "uxtahDNMU", },
"usatDwMu", "revshDM", "usatDwMu",
},
{ --10
shift = 12, mask = 15,
[15] = {
shift = 5, mask = 7,
"smuadNMS", "smuadxNMS", "smusdNMS", "smusdxNMS",
},
_ = {
shift = 5, mask = 7,
[0] = "smladNMSD", "smladxNMSD", "smlsdNMSD", "smlsdxNMSD",
},
},
false, false, false,
{ --14
shift = 5, mask = 7,
[0] = "smlaldDNMS", "smlaldxDNMS", "smlsldDNMS", "smlsldxDNMS",
},
{ --15
shift = 5, mask = 7,
[0] = { shift = 12, mask = 15, [15] = "smmulNMS", _ = "smmlaNMSD", },
{ shift = 12, mask = 15, [15] = "smmulrNMS", _ = "smmlarNMSD", },
false, false, false, false,
"smmlsNMSD", "smmlsrNMSD",
},
false, false,
{ --18
shift = 5, mask = 7,
[0] = { shift = 12, mask = 15, [15] = "usad8NMS", _ = "usada8NMSD", },
},
false,
{ --1a
shift = 5, mask = 3, [2] = "sbfxDMvw",
},
{ --1b
shift = 5, mask = 3, [2] = "sbfxDMvw",
},
{ --1c
shift = 5, mask = 3,
[0] = { shift = 0, mask = 15, [15] = "bfcDvX", _ = "bfiDMvX", },
},
{ --1d
shift = 5, mask = 3,
[0] = { shift = 0, mask = 15, [15] = "bfcDvX", _ = "bfiDMvX", },
},
{ --1e
shift = 5, mask = 3, [2] = "ubfxDMvw",
},
{ --1f
shift = 5, mask = 3, [2] = "ubfxDMvw",
},
}
local map_load = {
shift = 21, mask = 9,
{
shift = 20, mask = 5,
[0] = "strtDL", "ldrtDL", [4] = "strbtDL", [5] = "ldrbtDL",
},
_ = {
shift = 20, mask = 5,
[0] = "strDL", "ldrDL", [4] = "strbDL", [5] = "ldrbDL",
}
}
local map_load1 = {
shift = 4, mask = 1,
[0] = map_load, map_media,
}
local map_loadm = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "stmdaNR", "stmNR",
{ shift = 16, mask = 63, [45] = "pushR", _ = "stmdbNR", }, "stmibNR",
},
{
shift = 23, mask = 3,
[0] = "ldmdaNR", { shift = 16, mask = 63, [61] = "popR", _ = "ldmNR", },
"ldmdbNR", "ldmibNR",
},
}
local map_data = {
shift = 21, mask = 15,
[0] = "andDNPs", "eorDNPs", "subDNPs", "rsbDNPs",
"addDNPs", "adcDNPs", "sbcDNPs", "rscDNPs",
"tstNP", "teqNP", "cmpNP", "cmnNP",
"orrDNPs", "movDPs", "bicDNPs", "mvnDPs",
}
local map_mul = {
shift = 21, mask = 7,
[0] = "mulNMSs", "mlaNMSDs", "umaalDNMS", "mlsDNMS",
"umullDNMSs", "umlalDNMSs", "smullDNMSs", "smlalDNMSs",
}
local map_sync = {
shift = 20, mask = 15, -- NYI: brackets around N. R(D+1) for ldrexd/strexd.
[0] = "swpDMN", false, false, false,
"swpbDMN", false, false, false,
"strexDMN", "ldrexDN", "strexdDN", "ldrexdDN",
"strexbDMN", "ldrexbDN", "strexhDN", "ldrexhDN",
}
local map_mulh = {
shift = 21, mask = 3,
[0] = { shift = 5, mask = 3,
[0] = "smlabbNMSD", "smlatbNMSD", "smlabtNMSD", "smlattNMSD", },
{ shift = 5, mask = 3,
[0] = "smlawbNMSD", "smulwbNMS", "smlawtNMSD", "smulwtNMS", },
{ shift = 5, mask = 3,
[0] = "smlalbbDNMS", "smlaltbDNMS", "smlalbtDNMS", "smlalttDNMS", },
{ shift = 5, mask = 3,
[0] = "smulbbNMS", "smultbNMS", "smulbtNMS", "smulttNMS", },
}
local map_misc = {
shift = 4, mask = 7,
-- NYI: decode PSR bits of msr.
[0] = { shift = 21, mask = 1, [0] = "mrsD", "msrM", },
{ shift = 21, mask = 3, "bxM", false, "clzDM", },
{ shift = 21, mask = 3, "bxjM", },
{ shift = 21, mask = 3, "blxM", },
false,
{ shift = 21, mask = 3, [0] = "qaddDMN", "qsubDMN", "qdaddDMN", "qdsubDMN", },
false,
{ shift = 21, mask = 3, "bkptK", },
}
local map_datar = {
shift = 4, mask = 9,
[9] = {
shift = 5, mask = 3,
[0] = { shift = 24, mask = 1, [0] = map_mul, map_sync, },
{ shift = 20, mask = 1, [0] = "strhDL", "ldrhDL", },
{ shift = 20, mask = 1, [0] = "ldrdDL", "ldrsbDL", },
{ shift = 20, mask = 1, [0] = "strdDL", "ldrshDL", },
},
_ = {
shift = 20, mask = 25,
[16] = { shift = 7, mask = 1, [0] = map_misc, map_mulh, },
_ = {
shift = 0, mask = 0xffffffff,
[bor(0xe1a00000)] = "nop",
_ = map_data,
}
},
}
local map_datai = {
shift = 20, mask = 31, -- NYI: decode PSR bits of msr. Decode imm12.
[16] = "movwDW", [20] = "movtDW",
[18] = { shift = 0, mask = 0xf00ff, [0] = "nopv6", _ = "msrNW", },
[22] = "msrNW",
_ = map_data,
}
local map_branch = {
shift = 24, mask = 1,
[0] = "bB", "blB"
}
local map_condins = {
[0] = map_datar, map_datai, map_load, map_load1,
map_loadm, map_branch, map_loadc, map_datac
}
-- NYI: setend.
local map_uncondins = {
[0] = false, map_simddata, map_simdload, map_preload,
false, "blxB", map_loadcu, map_datacu,
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "sp", "lr", "pc",
}
local map_cond = {
[0] = "eq", "ne", "hs", "lo", "mi", "pl", "vs", "vc",
"hi", "ls", "ge", "lt", "gt", "le", "al",
}
local map_shift = { [0] = "lsl", "lsr", "asr", "ror", }
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then
extra = "\t->"..sym
elseif band(ctx.op, 0x0e000000) ~= 0x0a000000 then
extra = "\t; 0x"..tohex(ctx.rel)
end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-5s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-5s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
-- Format operand 2 of load/store opcodes.
local function fmtload(ctx, op, pos)
local base = map_gpr[band(rshift(op, 16), 15)]
local x, ofs
local ext = (band(op, 0x04000000) == 0)
if not ext and band(op, 0x02000000) == 0 then
ofs = band(op, 4095)
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
ofs = "#"..ofs
elseif ext and band(op, 0x00400000) ~= 0 then
ofs = band(op, 15) + band(rshift(op, 4), 0xf0)
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
ofs = "#"..ofs
else
ofs = map_gpr[band(op, 15)]
if ext or band(op, 0xfe0) == 0 then
elseif band(op, 0xfe0) == 0x60 then
ofs = format("%s, rrx", ofs)
else
local sh = band(rshift(op, 7), 31)
if sh == 0 then sh = 32 end
ofs = format("%s, %s #%d", ofs, map_shift[band(rshift(op, 5), 3)], sh)
end
if band(op, 0x00800000) == 0 then ofs = "-"..ofs end
end
if ofs == "#0" then
x = format("[%s]", base)
elseif band(op, 0x01000000) == 0 then
x = format("[%s], %s", base, ofs)
else
x = format("[%s, %s]", base, ofs)
end
if band(op, 0x01200000) == 0x01200000 then x = x.."!" end
return x
end
-- Format operand 2 of vector load/store opcodes.
local function fmtvload(ctx, op, pos)
local base = map_gpr[band(rshift(op, 16), 15)]
local ofs = band(op, 255)*4
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
if ofs == 0 then
return format("[%s]", base)
else
return format("[%s, #%d]", base, ofs)
end
end
local function fmtvr(op, vr, sh0, sh1)
if vr == "s" then
return format("s%d", 2*band(rshift(op, sh0), 15)+band(rshift(op, sh1), 1))
else
return format("d%d", band(rshift(op, sh0), 15)+band(rshift(op, sh1-4), 16))
end
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
local op = bor(lshift(b3, 24), lshift(b2, 16), lshift(b1, 8), b0)
local operands = {}
local suffix = ""
local last, name, pat
local vr
ctx.op = op
ctx.rel = nil
local cond = rshift(op, 28)
local opat
if cond == 15 then
opat = map_uncondins[band(rshift(op, 25), 7)]
else
if cond ~= 14 then suffix = map_cond[cond] end
opat = map_condins[band(rshift(op, 25), 7)]
end
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
opat = opat[band(rshift(op, opat.shift), opat.mask)] or opat._
end
name, pat = match(opat, "^([a-z0-9]*)(.*)")
if sub(pat, 1, 1) == "." then
local s2, p2 = match(pat, "^([a-z0-9.]*)(.*)")
suffix = suffix..s2
pat = p2
end
for p in gmatch(pat, ".") do
local x = nil
if p == "D" then
x = map_gpr[band(rshift(op, 12), 15)]
elseif p == "N" then
x = map_gpr[band(rshift(op, 16), 15)]
elseif p == "S" then
x = map_gpr[band(rshift(op, 8), 15)]
elseif p == "M" then
x = map_gpr[band(op, 15)]
elseif p == "d" then
x = fmtvr(op, vr, 12, 22)
elseif p == "n" then
x = fmtvr(op, vr, 16, 7)
elseif p == "m" then
x = fmtvr(op, vr, 0, 5)
elseif p == "P" then
if band(op, 0x02000000) ~= 0 then
x = ror(band(op, 255), 2*band(rshift(op, 8), 15))
else
x = map_gpr[band(op, 15)]
if band(op, 0xff0) ~= 0 then
operands[#operands+1] = x
local s = map_shift[band(rshift(op, 5), 3)]
local r = nil
if band(op, 0xf90) == 0 then
if s == "ror" then s = "rrx" else r = "#32" end
elseif band(op, 0x10) == 0 then
r = "#"..band(rshift(op, 7), 31)
else
r = map_gpr[band(rshift(op, 8), 15)]
end
if name == "mov" then name = s; x = r
elseif r then x = format("%s %s", s, r)
else x = s end
end
end
elseif p == "L" then
x = fmtload(ctx, op, pos)
elseif p == "l" then
x = fmtvload(ctx, op, pos)
elseif p == "B" then
local addr = ctx.addr + pos + 8 + arshift(lshift(op, 8), 6)
if cond == 15 then addr = addr + band(rshift(op, 23), 2) end
ctx.rel = addr
x = "0x"..tohex(addr)
elseif p == "F" then
vr = "s"
elseif p == "G" then
vr = "d"
elseif p == "." then
suffix = suffix..(vr == "s" and ".f32" or ".f64")
elseif p == "R" then
if band(op, 0x00200000) ~= 0 and #operands == 1 then
operands[1] = operands[1].."!"
end
local t = {}
for i=0,15 do
if band(rshift(op, i), 1) == 1 then t[#t+1] = map_gpr[i] end
end
x = "{"..concat(t, ", ").."}"
elseif p == "r" then
if band(op, 0x00200000) ~= 0 and #operands == 2 then
operands[1] = operands[1].."!"
end
local s = tonumber(sub(last, 2))
local n = band(op, 255)
if vr == "d" then n = rshift(n, 1) end
operands[#operands] = format("{%s-%s%d}", last, vr, s+n-1)
elseif p == "W" then
x = band(op, 0x0fff) + band(rshift(op, 4), 0xf000)
elseif p == "T" then
x = "#0x"..tohex(band(op, 0x00ffffff), 6)
elseif p == "U" then
x = band(rshift(op, 7), 31)
if x == 0 then x = nil end
elseif p == "u" then
x = band(rshift(op, 7), 31)
if band(op, 0x40) == 0 then
if x == 0 then x = nil else x = "lsl #"..x end
else
if x == 0 then x = "asr #32" else x = "asr #"..x end
end
elseif p == "v" then
x = band(rshift(op, 7), 31)
elseif p == "w" then
x = band(rshift(op, 16), 31)
elseif p == "x" then
x = band(rshift(op, 16), 31) + 1
elseif p == "X" then
x = band(rshift(op, 16), 31) - last + 1
elseif p == "Y" then
x = band(rshift(op, 12), 0xf0) + band(op, 0x0f)
elseif p == "K" then
x = "#0x"..tohex(band(rshift(op, 4), 0x0000fff0) + band(op, 15), 4)
elseif p == "s" then
if band(op, 0x00100000) ~= 0 then suffix = "s"..suffix end
else
assert(false)
end
if x then
last = x
if type(x) == "number" then x = "#"..x end
operands[#operands+1] = x
end
end
return putop(ctx, name..suffix, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
ctx.pos = ofs
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 16 then return map_gpr[r] end
return "d"..(r-16)
end
-- Public module functions.
return {
create = create,
disass = disass,
regname = regname
}
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,12 @@
----------------------------------------------------------------------------
-- LuaJIT ARM64BE disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- ARM64 instructions are always little-endian. So just forward to the
-- common ARM64 disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
return require((string.match(..., ".*%.") or "").."dis_arm64")
@@ -0,0 +1,694 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT/X license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles all standard MIPS32R1/R2 instructions.
-- Default mode is big-endian, but see: dis_mipsel.lua
------------------------------------------------------------------------------
local type = type
local byte, format = string.byte, string.format
local match, gmatch = string.match, string.gmatch
local concat = table.concat
local bit = require("bit")
local band, bor, tohex = bit.band, bit.bor, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Extended opcode maps common to all MIPS releases
------------------------------------------------------------------------------
local map_srl = { shift = 21, mask = 1, [0] = "srlDTA", "rotrDTA", }
local map_srlv = { shift = 6, mask = 1, [0] = "srlvDTS", "rotrvDTS", }
local map_cop0 = {
shift = 25, mask = 1,
[0] = {
shift = 21, mask = 15,
[0] = "mfc0TDW", [4] = "mtc0TDW",
[10] = "rdpgprDT",
[11] = { shift = 5, mask = 1, [0] = "diT0", "eiT0", },
[14] = "wrpgprDT",
}, {
shift = 0, mask = 63,
[1] = "tlbr", [2] = "tlbwi", [6] = "tlbwr", [8] = "tlbp",
[24] = "eret", [31] = "deret",
[32] = "wait",
},
}
------------------------------------------------------------------------------
-- Primary and extended opcode maps for MIPS R1-R5
------------------------------------------------------------------------------
local map_movci = { shift = 16, mask = 1, [0] = "movfDSC", "movtDSC", }
local map_special = {
shift = 0, mask = 63,
[0] = { shift = 0, mask = -1, [0] = "nop", _ = "sllDTA" },
map_movci, map_srl, "sraDTA",
"sllvDTS", false, map_srlv, "sravDTS",
"jrS", "jalrD1S", "movzDST", "movnDST",
"syscallY", "breakY", false, "sync",
"mfhiD", "mthiS", "mfloD", "mtloS",
"dsllvDST", false, "dsrlvDST", "dsravDST",
"multST", "multuST", "divST", "divuST",
"dmultST", "dmultuST", "ddivST", "ddivuST",
"addDST", "addu|moveDST0", "subDST", "subu|neguDS0T",
"andDST", "or|moveDST0", "xorDST", "nor|notDST0",
false, false, "sltDST", "sltuDST",
"daddDST", "dadduDST", "dsubDST", "dsubuDST",
"tgeSTZ", "tgeuSTZ", "tltSTZ", "tltuSTZ",
"teqSTZ", false, "tneSTZ", false,
"dsllDTA", false, "dsrlDTA", "dsraDTA",
"dsll32DTA", false, "dsrl32DTA", "dsra32DTA",
}
local map_special2 = {
shift = 0, mask = 63,
[0] = "maddST", "madduST", "mulDST", false,
"msubST", "msubuST",
[32] = "clzDS", [33] = "cloDS",
[63] = "sdbbpY",
}
local map_bshfl = {
shift = 6, mask = 31,
[2] = "wsbhDT",
[16] = "sebDT",
[24] = "sehDT",
}
local map_dbshfl = {
shift = 6, mask = 31,
[2] = "dsbhDT",
[5] = "dshdDT",
}
local map_special3 = {
shift = 0, mask = 63,
[0] = "extTSAK", [1] = "dextmTSAP", [3] = "dextTSAK",
[4] = "insTSAL", [6] = "dinsuTSEQ", [7] = "dinsTSAL",
[32] = map_bshfl, [36] = map_dbshfl, [59] = "rdhwrTD",
}
local map_regimm = {
shift = 16, mask = 31,
[0] = "bltzSB", "bgezSB", "bltzlSB", "bgezlSB",
false, false, false, false,
"tgeiSI", "tgeiuSI", "tltiSI", "tltiuSI",
"teqiSI", false, "tneiSI", false,
"bltzalSB", "bgezalSB", "bltzallSB", "bgezallSB",
false, false, false, false,
false, false, false, false,
false, false, false, "synciSO",
}
local map_cop1s = {
shift = 0, mask = 63,
[0] = "add.sFGH", "sub.sFGH", "mul.sFGH", "div.sFGH",
"sqrt.sFG", "abs.sFG", "mov.sFG", "neg.sFG",
"round.l.sFG", "trunc.l.sFG", "ceil.l.sFG", "floor.l.sFG",
"round.w.sFG", "trunc.w.sFG", "ceil.w.sFG", "floor.w.sFG",
false,
{ shift = 16, mask = 1, [0] = "movf.sFGC", "movt.sFGC" },
"movz.sFGT", "movn.sFGT",
false, "recip.sFG", "rsqrt.sFG", false,
false, false, false, false,
false, false, false, false,
false, "cvt.d.sFG", false, false,
"cvt.w.sFG", "cvt.l.sFG", "cvt.ps.sFGH", false,
false, false, false, false,
false, false, false, false,
"c.f.sVGH", "c.un.sVGH", "c.eq.sVGH", "c.ueq.sVGH",
"c.olt.sVGH", "c.ult.sVGH", "c.ole.sVGH", "c.ule.sVGH",
"c.sf.sVGH", "c.ngle.sVGH", "c.seq.sVGH", "c.ngl.sVGH",
"c.lt.sVGH", "c.nge.sVGH", "c.le.sVGH", "c.ngt.sVGH",
}
local map_cop1d = {
shift = 0, mask = 63,
[0] = "add.dFGH", "sub.dFGH", "mul.dFGH", "div.dFGH",
"sqrt.dFG", "abs.dFG", "mov.dFG", "neg.dFG",
"round.l.dFG", "trunc.l.dFG", "ceil.l.dFG", "floor.l.dFG",
"round.w.dFG", "trunc.w.dFG", "ceil.w.dFG", "floor.w.dFG",
false,
{ shift = 16, mask = 1, [0] = "movf.dFGC", "movt.dFGC" },
"movz.dFGT", "movn.dFGT",
false, "recip.dFG", "rsqrt.dFG", false,
false, false, false, false,
false, false, false, false,
"cvt.s.dFG", false, false, false,
"cvt.w.dFG", "cvt.l.dFG", false, false,
false, false, false, false,
false, false, false, false,
"c.f.dVGH", "c.un.dVGH", "c.eq.dVGH", "c.ueq.dVGH",
"c.olt.dVGH", "c.ult.dVGH", "c.ole.dVGH", "c.ule.dVGH",
"c.df.dVGH", "c.ngle.dVGH", "c.deq.dVGH", "c.ngl.dVGH",
"c.lt.dVGH", "c.nge.dVGH", "c.le.dVGH", "c.ngt.dVGH",
}
local map_cop1ps = {
shift = 0, mask = 63,
[0] = "add.psFGH", "sub.psFGH", "mul.psFGH", false,
false, "abs.psFG", "mov.psFG", "neg.psFG",
false, false, false, false,
false, false, false, false,
false,
{ shift = 16, mask = 1, [0] = "movf.psFGC", "movt.psFGC" },
"movz.psFGT", "movn.psFGT",
false, false, false, false,
false, false, false, false,
false, false, false, false,
"cvt.s.puFG", false, false, false,
false, false, false, false,
"cvt.s.plFG", false, false, false,
"pll.psFGH", "plu.psFGH", "pul.psFGH", "puu.psFGH",
"c.f.psVGH", "c.un.psVGH", "c.eq.psVGH", "c.ueq.psVGH",
"c.olt.psVGH", "c.ult.psVGH", "c.ole.psVGH", "c.ule.psVGH",
"c.psf.psVGH", "c.ngle.psVGH", "c.pseq.psVGH", "c.ngl.psVGH",
"c.lt.psVGH", "c.nge.psVGH", "c.le.psVGH", "c.ngt.psVGH",
}
local map_cop1w = {
shift = 0, mask = 63,
[32] = "cvt.s.wFG", [33] = "cvt.d.wFG",
}
local map_cop1l = {
shift = 0, mask = 63,
[32] = "cvt.s.lFG", [33] = "cvt.d.lFG",
}
local map_cop1bc = {
shift = 16, mask = 3,
[0] = "bc1fCB", "bc1tCB", "bc1flCB", "bc1tlCB",
}
local map_cop1 = {
shift = 21, mask = 31,
[0] = "mfc1TG", "dmfc1TG", "cfc1TG", "mfhc1TG",
"mtc1TG", "dmtc1TG", "ctc1TG", "mthc1TG",
map_cop1bc, false, false, false,
false, false, false, false,
map_cop1s, map_cop1d, false, false,
map_cop1w, map_cop1l, map_cop1ps,
}
local map_cop1x = {
shift = 0, mask = 63,
[0] = "lwxc1FSX", "ldxc1FSX", false, false,
false, "luxc1FSX", false, false,
"swxc1FSX", "sdxc1FSX", false, false,
false, "suxc1FSX", false, "prefxMSX",
false, false, false, false,
false, false, false, false,
false, false, false, false,
false, false, "alnv.psFGHS", false,
"madd.sFRGH", "madd.dFRGH", false, false,
false, false, "madd.psFRGH", false,
"msub.sFRGH", "msub.dFRGH", false, false,
false, false, "msub.psFRGH", false,
"nmadd.sFRGH", "nmadd.dFRGH", false, false,
false, false, "nmadd.psFRGH", false,
"nmsub.sFRGH", "nmsub.dFRGH", false, false,
false, false, "nmsub.psFRGH", false,
}
local map_pri = {
[0] = map_special, map_regimm, "jJ", "jalJ",
"beq|beqz|bST00B", "bne|bnezST0B", "blezSB", "bgtzSB",
"addiTSI", "addiu|liTS0I", "sltiTSI", "sltiuTSI",
"andiTSU", "ori|liTS0U", "xoriTSU", "luiTU",
map_cop0, map_cop1, false, map_cop1x,
"beql|beqzlST0B", "bnel|bnezlST0B", "blezlSB", "bgtzlSB",
"daddiTSI", "daddiuTSI", false, false,
map_special2, "jalxJ", false, map_special3,
"lbTSO", "lhTSO", "lwlTSO", "lwTSO",
"lbuTSO", "lhuTSO", "lwrTSO", false,
"sbTSO", "shTSO", "swlTSO", "swTSO",
false, false, "swrTSO", "cacheNSO",
"llTSO", "lwc1HSO", "lwc2TSO", "prefNSO",
false, "ldc1HSO", "ldc2TSO", "ldTSO",
"scTSO", "swc1HSO", "swc2TSO", false,
false, "sdc1HSO", "sdc2TSO", "sdTSO",
}
------------------------------------------------------------------------------
-- Primary and extended opcode maps for MIPS R6
------------------------------------------------------------------------------
local map_mul_r6 = { shift = 6, mask = 3, [2] = "mulDST", [3] = "muhDST" }
local map_mulu_r6 = { shift = 6, mask = 3, [2] = "muluDST", [3] = "muhuDST" }
local map_div_r6 = { shift = 6, mask = 3, [2] = "divDST", [3] = "modDST" }
local map_divu_r6 = { shift = 6, mask = 3, [2] = "divuDST", [3] = "moduDST" }
local map_dmul_r6 = { shift = 6, mask = 3, [2] = "dmulDST", [3] = "dmuhDST" }
local map_dmulu_r6 = { shift = 6, mask = 3, [2] = "dmuluDST", [3] = "dmuhuDST" }
local map_ddiv_r6 = { shift = 6, mask = 3, [2] = "ddivDST", [3] = "dmodDST" }
local map_ddivu_r6 = { shift = 6, mask = 3, [2] = "ddivuDST", [3] = "dmoduDST" }
local map_special_r6 = {
shift = 0, mask = 63,
[0] = { shift = 0, mask = -1, [0] = "nop", _ = "sllDTA" },
false, map_srl, "sraDTA",
"sllvDTS", false, map_srlv, "sravDTS",
"jrS", "jalrD1S", false, false,
"syscallY", "breakY", false, "sync",
"clzDS", "cloDS", "dclzDS", "dcloDS",
"dsllvDST", "dlsaDSTA", "dsrlvDST", "dsravDST",
map_mul_r6, map_mulu_r6, map_div_r6, map_divu_r6,
map_dmul_r6, map_dmulu_r6, map_ddiv_r6, map_ddivu_r6,
"addDST", "addu|moveDST0", "subDST", "subu|neguDS0T",
"andDST", "or|moveDST0", "xorDST", "nor|notDST0",
false, false, "sltDST", "sltuDST",
"daddDST", "dadduDST", "dsubDST", "dsubuDST",
"tgeSTZ", "tgeuSTZ", "tltSTZ", "tltuSTZ",
"teqSTZ", "seleqzDST", "tneSTZ", "selnezDST",
"dsllDTA", false, "dsrlDTA", "dsraDTA",
"dsll32DTA", false, "dsrl32DTA", "dsra32DTA",
}
local map_bshfl_r6 = {
shift = 9, mask = 3,
[1] = "alignDSTa",
_ = {
shift = 6, mask = 31,
[0] = "bitswapDT",
[2] = "wsbhDT",
[16] = "sebDT",
[24] = "sehDT",
}
}
local map_dbshfl_r6 = {
shift = 9, mask = 3,
[1] = "dalignDSTa",
_ = {
shift = 6, mask = 31,
[0] = "dbitswapDT",
[2] = "dsbhDT",
[5] = "dshdDT",
}
}
local map_special3_r6 = {
shift = 0, mask = 63,
[0] = "extTSAK", [1] = "dextmTSAP", [3] = "dextTSAK",
[4] = "insTSAL", [6] = "dinsuTSEQ", [7] = "dinsTSAL",
[32] = map_bshfl_r6, [36] = map_dbshfl_r6, [59] = "rdhwrTD",
}
local map_regimm_r6 = {
shift = 16, mask = 31,
[0] = "bltzSB", [1] = "bgezSB",
[6] = "dahiSI", [30] = "datiSI",
[23] = "sigrieI", [31] = "synciSO",
}
local map_pcrel_r6 = {
shift = 19, mask = 3,
[0] = "addiupcS2", "lwpcS2", "lwupcS2", {
shift = 18, mask = 1,
[0] = "ldpcS3", { shift = 16, mask = 3, [2] = "auipcSI", [3] = "aluipcSI" }
}
}
local map_cop1s_r6 = {
shift = 0, mask = 63,
[0] = "add.sFGH", "sub.sFGH", "mul.sFGH", "div.sFGH",
"sqrt.sFG", "abs.sFG", "mov.sFG", "neg.sFG",
"round.l.sFG", "trunc.l.sFG", "ceil.l.sFG", "floor.l.sFG",
"round.w.sFG", "trunc.w.sFG", "ceil.w.sFG", "floor.w.sFG",
"sel.sFGH", false, false, false,
"seleqz.sFGH", "recip.sFG", "rsqrt.sFG", "selnez.sFGH",
"maddf.sFGH", "msubf.sFGH", "rint.sFG", "class.sFG",
"min.sFGH", "mina.sFGH", "max.sFGH", "maxa.sFGH",
false, "cvt.d.sFG", false, false,
"cvt.w.sFG", "cvt.l.sFG",
}
local map_cop1d_r6 = {
shift = 0, mask = 63,
[0] = "add.dFGH", "sub.dFGH", "mul.dFGH", "div.dFGH",
"sqrt.dFG", "abs.dFG", "mov.dFG", "neg.dFG",
"round.l.dFG", "trunc.l.dFG", "ceil.l.dFG", "floor.l.dFG",
"round.w.dFG", "trunc.w.dFG", "ceil.w.dFG", "floor.w.dFG",
"sel.dFGH", false, false, false,
"seleqz.dFGH", "recip.dFG", "rsqrt.dFG", "selnez.dFGH",
"maddf.dFGH", "msubf.dFGH", "rint.dFG", "class.dFG",
"min.dFGH", "mina.dFGH", "max.dFGH", "maxa.dFGH",
"cvt.s.dFG", false, false, false,
"cvt.w.dFG", "cvt.l.dFG",
}
local map_cop1w_r6 = {
shift = 0, mask = 63,
[0] = "cmp.af.sFGH", "cmp.un.sFGH", "cmp.eq.sFGH", "cmp.ueq.sFGH",
"cmp.lt.sFGH", "cmp.ult.sFGH", "cmp.le.sFGH", "cmp.ule.sFGH",
"cmp.saf.sFGH", "cmp.sun.sFGH", "cmp.seq.sFGH", "cmp.sueq.sFGH",
"cmp.slt.sFGH", "cmp.sult.sFGH", "cmp.sle.sFGH", "cmp.sule.sFGH",
false, "cmp.or.sFGH", "cmp.une.sFGH", "cmp.ne.sFGH",
false, false, false, false,
false, "cmp.sor.sFGH", "cmp.sune.sFGH", "cmp.sne.sFGH",
false, false, false, false,
"cvt.s.wFG", "cvt.d.wFG",
}
local map_cop1l_r6 = {
shift = 0, mask = 63,
[0] = "cmp.af.dFGH", "cmp.un.dFGH", "cmp.eq.dFGH", "cmp.ueq.dFGH",
"cmp.lt.dFGH", "cmp.ult.dFGH", "cmp.le.dFGH", "cmp.ule.dFGH",
"cmp.saf.dFGH", "cmp.sun.dFGH", "cmp.seq.dFGH", "cmp.sueq.dFGH",
"cmp.slt.dFGH", "cmp.sult.dFGH", "cmp.sle.dFGH", "cmp.sule.dFGH",
false, "cmp.or.dFGH", "cmp.une.dFGH", "cmp.ne.dFGH",
false, false, false, false,
false, "cmp.sor.dFGH", "cmp.sune.dFGH", "cmp.sne.dFGH",
false, false, false, false,
"cvt.s.lFG", "cvt.d.lFG",
}
local map_cop1_r6 = {
shift = 21, mask = 31,
[0] = "mfc1TG", "dmfc1TG", "cfc1TG", "mfhc1TG",
"mtc1TG", "dmtc1TG", "ctc1TG", "mthc1TG",
false, "bc1eqzHB", false, false,
false, "bc1nezHB", false, false,
map_cop1s_r6, map_cop1d_r6, false, false,
map_cop1w_r6, map_cop1l_r6,
}
local function maprs_popTS(rs, rt)
if rt == 0 then return 0 elseif rs == 0 then return 1
elseif rs == rt then return 2 else return 3 end
end
local map_pop06_r6 = {
maprs = maprs_popTS, [0] = "blezSB", "blezalcTB", "bgezalcTB", "bgeucSTB"
}
local map_pop07_r6 = {
maprs = maprs_popTS, [0] = "bgtzSB", "bgtzalcTB", "bltzalcTB", "bltucSTB"
}
local map_pop26_r6 = {
maprs = maprs_popTS, "blezcTB", "bgezcTB", "bgecSTB"
}
local map_pop27_r6 = {
maprs = maprs_popTS, "bgtzcTB", "bltzcTB", "bltcSTB"
}
local function maprs_popS(rs, rt)
if rs == 0 then return 0 else return 1 end
end
local map_pop66_r6 = {
maprs = maprs_popS, [0] = "jicTI", "beqzcSb"
}
local map_pop76_r6 = {
maprs = maprs_popS, [0] = "jialcTI", "bnezcSb"
}
local function maprs_popST(rs, rt)
if rs >= rt then return 0 elseif rs == 0 then return 1 else return 2 end
end
local map_pop10_r6 = {
maprs = maprs_popST, [0] = "bovcSTB", "beqzalcTB", "beqcSTB"
}
local map_pop30_r6 = {
maprs = maprs_popST, [0] = "bnvcSTB", "bnezalcTB", "bnecSTB"
}
local map_pri_r6 = {
[0] = map_special_r6, map_regimm_r6, "jJ", "jalJ",
"beq|beqz|bST00B", "bne|bnezST0B", map_pop06_r6, map_pop07_r6,
map_pop10_r6, "addiu|liTS0I", "sltiTSI", "sltiuTSI",
"andiTSU", "ori|liTS0U", "xoriTSU", "aui|luiTS0U",
map_cop0, map_cop1_r6, false, false,
false, false, map_pop26_r6, map_pop27_r6,
map_pop30_r6, "daddiuTSI", false, false,
false, "dauiTSI", false, map_special3_r6,
"lbTSO", "lhTSO", false, "lwTSO",
"lbuTSO", "lhuTSO", false, false,
"sbTSO", "shTSO", false, "swTSO",
false, false, false, false,
false, "lwc1HSO", "bc#", false,
false, "ldc1HSO", map_pop66_r6, "ldTSO",
false, "swc1HSO", "balc#", map_pcrel_r6,
false, "sdc1HSO", map_pop76_r6, "sdTSO",
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
"r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
"r24", "r25", "r26", "r27", "r28", "sp", "r30", "ra",
}
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then extra = "\t->"..sym end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-7s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-7s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
local function get_be(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
return bor(lshift(b0, 24), lshift(b1, 16), lshift(b2, 8), b3)
end
local function get_le(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
return bor(lshift(b3, 24), lshift(b2, 16), lshift(b1, 8), b0)
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local op = ctx:get()
local operands = {}
local last = nil
ctx.op = op
ctx.rel = nil
local opat = ctx.map_pri[rshift(op, 26)]
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
if opat.maprs then
opat = opat[opat.maprs(band(rshift(op,21),31), band(rshift(op,16),31))]
else
opat = opat[band(rshift(op, opat.shift), opat.mask)] or opat._
end
end
local name, pat = match(opat, "^([a-z0-9_.]*)(.*)")
local altname, pat2 = match(pat, "|([a-z0-9_.|]*)(.*)")
if altname then pat = pat2 end
for p in gmatch(pat, ".") do
local x = nil
if p == "S" then
x = map_gpr[band(rshift(op, 21), 31)]
elseif p == "T" then
x = map_gpr[band(rshift(op, 16), 31)]
elseif p == "D" then
x = map_gpr[band(rshift(op, 11), 31)]
elseif p == "F" then
x = "f"..band(rshift(op, 6), 31)
elseif p == "G" then
x = "f"..band(rshift(op, 11), 31)
elseif p == "H" then
x = "f"..band(rshift(op, 16), 31)
elseif p == "R" then
x = "f"..band(rshift(op, 21), 31)
elseif p == "A" then
x = band(rshift(op, 6), 31)
elseif p == "a" then
x = band(rshift(op, 6), 7)
elseif p == "E" then
x = band(rshift(op, 6), 31) + 32
elseif p == "M" then
x = band(rshift(op, 11), 31)
elseif p == "N" then
x = band(rshift(op, 16), 31)
elseif p == "C" then
x = band(rshift(op, 18), 7)
if x == 0 then x = nil end
elseif p == "K" then
x = band(rshift(op, 11), 31) + 1
elseif p == "P" then
x = band(rshift(op, 11), 31) + 33
elseif p == "L" then
x = band(rshift(op, 11), 31) - last + 1
elseif p == "Q" then
x = band(rshift(op, 11), 31) - last + 33
elseif p == "I" then
x = arshift(lshift(op, 16), 16)
elseif p == "2" then
x = arshift(lshift(op, 13), 11)
elseif p == "3" then
x = arshift(lshift(op, 14), 11)
elseif p == "U" then
x = band(op, 0xffff)
elseif p == "O" then
local disp = arshift(lshift(op, 16), 16)
operands[#operands] = format("%d(%s)", disp, last)
elseif p == "X" then
local index = map_gpr[band(rshift(op, 16), 31)]
operands[#operands] = format("%s(%s)", index, last)
elseif p == "B" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 16), 14) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "b" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 11), 9) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "#" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 6), 4) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "J" then
local a = ctx.addr + ctx.pos
x = a - band(a, 0x0fffffff) + band(op, 0x03ffffff)*4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "V" then
x = band(rshift(op, 8), 7)
if x == 0 then x = nil end
elseif p == "W" then
x = band(op, 7)
if x == 0 then x = nil end
elseif p == "Y" then
x = band(rshift(op, 6), 0x000fffff)
if x == 0 then x = nil end
elseif p == "Z" then
x = band(rshift(op, 6), 1023)
if x == 0 then x = nil end
elseif p == "0" then
if last == "r0" or last == 0 then
local n = #operands
operands[n] = nil
last = operands[n-1]
if altname then
local a1, a2 = match(altname, "([^|]*)|(.*)")
if a1 then name, altname = a1, a2
else name = altname end
end
end
elseif p == "1" then
if last == "ra" then
operands[#operands] = nil
end
else
assert(false)
end
if x then operands[#operands+1] = x; last = x end
end
return putop(ctx, name, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
stop = stop - stop % 4
ctx.pos = ofs - ofs % 4
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
ctx.get = get_be
ctx.map_pri = map_pri
return ctx
end
local function create_el(code, addr, out)
local ctx = create(code, addr, out)
ctx.get = get_le
return ctx
end
local function create_r6(code, addr, out)
local ctx = create(code, addr, out)
ctx.map_pri = map_pri_r6
return ctx
end
local function create_r6_el(code, addr, out)
local ctx = create(code, addr, out)
ctx.get = get_le
ctx.map_pri = map_pri_r6
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
local function disass_el(code, addr, out)
create_el(code, addr, out):disass()
end
local function disass_r6(code, addr, out)
create_r6(code, addr, out):disass()
end
local function disass_r6_el(code, addr, out)
create_r6_el(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 32 then return map_gpr[r] end
return "f"..(r-32)
end
-- Public module functions.
return {
create = create,
create_el = create_el,
create_r6 = create_r6,
create_r6_el = create_r6_el,
disass = disass,
disass_el = disass_el,
disass_r6 = disass_r6,
disass_r6_el = disass_r6_el,
regname = regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the big-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create,
disass = dis_mips.disass,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64EL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_el,
disass = dis_mips.disass_el,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64R6 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the r6 big-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_r6,
disass = dis_mips.disass_r6,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64R6EL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the r6 little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_r6_el,
disass = dis_mips.disass_r6_el,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPSEL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_el,
disass = dis_mips.disass_el,
regname = dis_mips.regname
}
@@ -0,0 +1,591 @@
----------------------------------------------------------------------------
-- LuaJIT PPC disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT/X license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles all common, non-privileged 32/64 bit PowerPC instructions
-- plus the e500 SPE instructions and some Cell/Xenon extensions.
--
-- NYI: VMX, VMX128
------------------------------------------------------------------------------
local type = type
local byte, format = string.byte, string.format
local match, gmatch, gsub = string.match, string.gmatch, string.gsub
local concat = table.concat
local bit = require("bit")
local band, bor, tohex = bit.band, bit.bor, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Primary and extended opcode maps
------------------------------------------------------------------------------
local map_crops = {
shift = 1, mask = 1023,
[0] = "mcrfXX",
[33] = "crnor|crnotCCC=", [129] = "crandcCCC",
[193] = "crxor|crclrCCC%", [225] = "crnandCCC",
[257] = "crandCCC", [289] = "creqv|crsetCCC%",
[417] = "crorcCCC", [449] = "cror|crmoveCCC=",
[16] = "b_lrKB", [528] = "b_ctrKB",
[150] = "isync",
}
local map_rlwinm = setmetatable({
shift = 0, mask = -1,
},
{ __index = function(t, x)
local rot = band(rshift(x, 11), 31)
local mb = band(rshift(x, 6), 31)
local me = band(rshift(x, 1), 31)
if mb == 0 and me == 31-rot then
return "slwiRR~A."
elseif me == 31 and mb == 32-rot then
return "srwiRR~-A."
else
return "rlwinmRR~AAA."
end
end
})
local map_rld = {
shift = 2, mask = 7,
[0] = "rldiclRR~HM.", "rldicrRR~HM.", "rldicRR~HM.", "rldimiRR~HM.",
{
shift = 1, mask = 1,
[0] = "rldclRR~RM.", "rldcrRR~RM.",
},
}
local map_ext = setmetatable({
shift = 1, mask = 1023,
[0] = "cmp_YLRR", [32] = "cmpl_YLRR",
[4] = "twARR", [68] = "tdARR",
[8] = "subfcRRR.", [40] = "subfRRR.",
[104] = "negRR.", [136] = "subfeRRR.",
[200] = "subfzeRR.", [232] = "subfmeRR.",
[520] = "subfcoRRR.", [552] = "subfoRRR.",
[616] = "negoRR.", [648] = "subfeoRRR.",
[712] = "subfzeoRR.", [744] = "subfmeoRR.",
[9] = "mulhduRRR.", [73] = "mulhdRRR.", [233] = "mulldRRR.",
[457] = "divduRRR.", [489] = "divdRRR.",
[745] = "mulldoRRR.",
[969] = "divduoRRR.", [1001] = "divdoRRR.",
[10] = "addcRRR.", [138] = "addeRRR.",
[202] = "addzeRR.", [234] = "addmeRR.", [266] = "addRRR.",
[522] = "addcoRRR.", [650] = "addeoRRR.",
[714] = "addzeoRR.", [746] = "addmeoRR.", [778] = "addoRRR.",
[11] = "mulhwuRRR.", [75] = "mulhwRRR.", [235] = "mullwRRR.",
[459] = "divwuRRR.", [491] = "divwRRR.",
[747] = "mullwoRRR.",
[971] = "divwouRRR.", [1003] = "divwoRRR.",
[15] = "iselltRRR", [47] = "iselgtRRR", [79] = "iseleqRRR",
[144] = { shift = 20, mask = 1, [0] = "mtcrfRZ~", "mtocrfRZ~", },
[19] = { shift = 20, mask = 1, [0] = "mfcrR", "mfocrfRZ", },
[371] = { shift = 11, mask = 1023, [392] = "mftbR", [424] = "mftbuR", },
[339] = {
shift = 11, mask = 1023,
[32] = "mferR", [256] = "mflrR", [288] = "mfctrR", [16] = "mfspefscrR",
},
[467] = {
shift = 11, mask = 1023,
[32] = "mtxerR", [256] = "mtlrR", [288] = "mtctrR", [16] = "mtspefscrR",
},
[20] = "lwarxRR0R", [84] = "ldarxRR0R",
[21] = "ldxRR0R", [53] = "lduxRRR",
[149] = "stdxRR0R", [181] = "stduxRRR",
[341] = "lwaxRR0R", [373] = "lwauxRRR",
[23] = "lwzxRR0R", [55] = "lwzuxRRR",
[87] = "lbzxRR0R", [119] = "lbzuxRRR",
[151] = "stwxRR0R", [183] = "stwuxRRR",
[215] = "stbxRR0R", [247] = "stbuxRRR",
[279] = "lhzxRR0R", [311] = "lhzuxRRR",
[343] = "lhaxRR0R", [375] = "lhauxRRR",
[407] = "sthxRR0R", [439] = "sthuxRRR",
[54] = "dcbst-R0R", [86] = "dcbf-R0R",
[150] = "stwcxRR0R.", [214] = "stdcxRR0R.",
[246] = "dcbtst-R0R", [278] = "dcbt-R0R",
[310] = "eciwxRR0R", [438] = "ecowxRR0R",
[470] = "dcbi-RR",
[598] = {
shift = 21, mask = 3,
[0] = "sync", "lwsync", "ptesync",
},
[758] = "dcba-RR",
[854] = "eieio", [982] = "icbi-R0R", [1014] = "dcbz-R0R",
[26] = "cntlzwRR~", [58] = "cntlzdRR~",
[122] = "popcntbRR~",
[154] = "prtywRR~", [186] = "prtydRR~",
[28] = "andRR~R.", [60] = "andcRR~R.", [124] = "nor|notRR~R=.",
[284] = "eqvRR~R.", [316] = "xorRR~R.",
[412] = "orcRR~R.", [444] = "or|mrRR~R=.", [476] = "nandRR~R.",
[508] = "cmpbRR~R",
[512] = "mcrxrX",
[532] = "ldbrxRR0R", [660] = "stdbrxRR0R",
[533] = "lswxRR0R", [597] = "lswiRR0A",
[661] = "stswxRR0R", [725] = "stswiRR0A",
[534] = "lwbrxRR0R", [662] = "stwbrxRR0R",
[790] = "lhbrxRR0R", [918] = "sthbrxRR0R",
[535] = "lfsxFR0R", [567] = "lfsuxFRR",
[599] = "lfdxFR0R", [631] = "lfduxFRR",
[663] = "stfsxFR0R", [695] = "stfsuxFRR",
[727] = "stfdxFR0R", [759] = "stfduxFR0R",
[855] = "lfiwaxFR0R",
[983] = "stfiwxFR0R",
[24] = "slwRR~R.",
[27] = "sldRR~R.", [536] = "srwRR~R.",
[792] = "srawRR~R.", [824] = "srawiRR~A.",
[794] = "sradRR~R.", [826] = "sradiRR~H.", [827] = "sradiRR~H.",
[922] = "extshRR~.", [954] = "extsbRR~.", [986] = "extswRR~.",
[539] = "srdRR~R.",
},
{ __index = function(t, x)
if band(x, 31) == 15 then return "iselRRRC" end
end
})
local map_ld = {
shift = 0, mask = 3,
[0] = "ldRRE", "lduRRE", "lwaRRE",
}
local map_std = {
shift = 0, mask = 3,
[0] = "stdRRE", "stduRRE",
}
local map_fps = {
shift = 5, mask = 1,
{
shift = 1, mask = 15,
[0] = false, false, "fdivsFFF.", false,
"fsubsFFF.", "faddsFFF.", "fsqrtsF-F.", false,
"fresF-F.", "fmulsFF-F.", "frsqrtesF-F.", false,
"fmsubsFFFF~.", "fmaddsFFFF~.", "fnmsubsFFFF~.", "fnmaddsFFFF~.",
}
}
local map_fpd = {
shift = 5, mask = 1,
[0] = {
shift = 1, mask = 1023,
[0] = "fcmpuXFF", [32] = "fcmpoXFF", [64] = "mcrfsXX",
[38] = "mtfsb1A.", [70] = "mtfsb0A.", [134] = "mtfsfiA>>-A>",
[8] = "fcpsgnFFF.", [40] = "fnegF-F.", [72] = "fmrF-F.",
[136] = "fnabsF-F.", [264] = "fabsF-F.",
[12] = "frspF-F.",
[14] = "fctiwF-F.", [15] = "fctiwzF-F.",
[583] = "mffsF.", [711] = "mtfsfZF.",
[392] = "frinF-F.", [424] = "frizF-F.",
[456] = "fripF-F.", [488] = "frimF-F.",
[814] = "fctidF-F.", [815] = "fctidzF-F.", [846] = "fcfidF-F.",
},
{
shift = 1, mask = 15,
[0] = false, false, "fdivFFF.", false,
"fsubFFF.", "faddFFF.", "fsqrtF-F.", "fselFFFF~.",
"freF-F.", "fmulFF-F.", "frsqrteF-F.", false,
"fmsubFFFF~.", "fmaddFFFF~.", "fnmsubFFFF~.", "fnmaddFFFF~.",
}
}
local map_spe = {
shift = 0, mask = 2047,
[512] = "evaddwRRR", [514] = "evaddiwRAR~",
[516] = "evsubwRRR~", [518] = "evsubiwRAR~",
[520] = "evabsRR", [521] = "evnegRR",
[522] = "evextsbRR", [523] = "evextshRR", [524] = "evrndwRR",
[525] = "evcntlzwRR", [526] = "evcntlswRR",
[527] = "brincRRR",
[529] = "evandRRR", [530] = "evandcRRR", [534] = "evxorRRR",
[535] = "evor|evmrRRR=", [536] = "evnor|evnotRRR=",
[537] = "eveqvRRR", [539] = "evorcRRR", [542] = "evnandRRR",
[544] = "evsrwuRRR", [545] = "evsrwsRRR",
[546] = "evsrwiuRRA", [547] = "evsrwisRRA",
[548] = "evslwRRR", [550] = "evslwiRRA",
[552] = "evrlwRRR", [553] = "evsplatiRS",
[554] = "evrlwiRRA", [555] = "evsplatfiRS",
[556] = "evmergehiRRR", [557] = "evmergeloRRR",
[558] = "evmergehiloRRR", [559] = "evmergelohiRRR",
[560] = "evcmpgtuYRR", [561] = "evcmpgtsYRR",
[562] = "evcmpltuYRR", [563] = "evcmpltsYRR",
[564] = "evcmpeqYRR",
[632] = "evselRRR", [633] = "evselRRRW",
[634] = "evselRRRW", [635] = "evselRRRW",
[636] = "evselRRRW", [637] = "evselRRRW",
[638] = "evselRRRW", [639] = "evselRRRW",
[640] = "evfsaddRRR", [641] = "evfssubRRR",
[644] = "evfsabsRR", [645] = "evfsnabsRR", [646] = "evfsnegRR",
[648] = "evfsmulRRR", [649] = "evfsdivRRR",
[652] = "evfscmpgtYRR", [653] = "evfscmpltYRR", [654] = "evfscmpeqYRR",
[656] = "evfscfuiR-R", [657] = "evfscfsiR-R",
[658] = "evfscfufR-R", [659] = "evfscfsfR-R",
[660] = "evfsctuiR-R", [661] = "evfsctsiR-R",
[662] = "evfsctufR-R", [663] = "evfsctsfR-R",
[664] = "evfsctuizR-R", [666] = "evfsctsizR-R",
[668] = "evfststgtYRR", [669] = "evfststltYRR", [670] = "evfststeqYRR",
[704] = "efsaddRRR", [705] = "efssubRRR",
[708] = "efsabsRR", [709] = "efsnabsRR", [710] = "efsnegRR",
[712] = "efsmulRRR", [713] = "efsdivRRR",
[716] = "efscmpgtYRR", [717] = "efscmpltYRR", [718] = "efscmpeqYRR",
[719] = "efscfdR-R",
[720] = "efscfuiR-R", [721] = "efscfsiR-R",
[722] = "efscfufR-R", [723] = "efscfsfR-R",
[724] = "efsctuiR-R", [725] = "efsctsiR-R",
[726] = "efsctufR-R", [727] = "efsctsfR-R",
[728] = "efsctuizR-R", [730] = "efsctsizR-R",
[732] = "efststgtYRR", [733] = "efststltYRR", [734] = "efststeqYRR",
[736] = "efdaddRRR", [737] = "efdsubRRR",
[738] = "efdcfuidR-R", [739] = "efdcfsidR-R",
[740] = "efdabsRR", [741] = "efdnabsRR", [742] = "efdnegRR",
[744] = "efdmulRRR", [745] = "efddivRRR",
[746] = "efdctuidzR-R", [747] = "efdctsidzR-R",
[748] = "efdcmpgtYRR", [749] = "efdcmpltYRR", [750] = "efdcmpeqYRR",
[751] = "efdcfsR-R",
[752] = "efdcfuiR-R", [753] = "efdcfsiR-R",
[754] = "efdcfufR-R", [755] = "efdcfsfR-R",
[756] = "efdctuiR-R", [757] = "efdctsiR-R",
[758] = "efdctufR-R", [759] = "efdctsfR-R",
[760] = "efdctuizR-R", [762] = "efdctsizR-R",
[764] = "efdtstgtYRR", [765] = "efdtstltYRR", [766] = "efdtsteqYRR",
[768] = "evlddxRR0R", [769] = "evlddRR8",
[770] = "evldwxRR0R", [771] = "evldwRR8",
[772] = "evldhxRR0R", [773] = "evldhRR8",
[776] = "evlhhesplatxRR0R", [777] = "evlhhesplatRR2",
[780] = "evlhhousplatxRR0R", [781] = "evlhhousplatRR2",
[782] = "evlhhossplatxRR0R", [783] = "evlhhossplatRR2",
[784] = "evlwhexRR0R", [785] = "evlwheRR4",
[788] = "evlwhouxRR0R", [789] = "evlwhouRR4",
[790] = "evlwhosxRR0R", [791] = "evlwhosRR4",
[792] = "evlwwsplatxRR0R", [793] = "evlwwsplatRR4",
[796] = "evlwhsplatxRR0R", [797] = "evlwhsplatRR4",
[800] = "evstddxRR0R", [801] = "evstddRR8",
[802] = "evstdwxRR0R", [803] = "evstdwRR8",
[804] = "evstdhxRR0R", [805] = "evstdhRR8",
[816] = "evstwhexRR0R", [817] = "evstwheRR4",
[820] = "evstwhoxRR0R", [821] = "evstwhoRR4",
[824] = "evstwwexRR0R", [825] = "evstwweRR4",
[828] = "evstwwoxRR0R", [829] = "evstwwoRR4",
[1027] = "evmhessfRRR", [1031] = "evmhossfRRR", [1032] = "evmheumiRRR",
[1033] = "evmhesmiRRR", [1035] = "evmhesmfRRR", [1036] = "evmhoumiRRR",
[1037] = "evmhosmiRRR", [1039] = "evmhosmfRRR", [1059] = "evmhessfaRRR",
[1063] = "evmhossfaRRR", [1064] = "evmheumiaRRR", [1065] = "evmhesmiaRRR",
[1067] = "evmhesmfaRRR", [1068] = "evmhoumiaRRR", [1069] = "evmhosmiaRRR",
[1071] = "evmhosmfaRRR", [1095] = "evmwhssfRRR", [1096] = "evmwlumiRRR",
[1100] = "evmwhumiRRR", [1101] = "evmwhsmiRRR", [1103] = "evmwhsmfRRR",
[1107] = "evmwssfRRR", [1112] = "evmwumiRRR", [1113] = "evmwsmiRRR",
[1115] = "evmwsmfRRR", [1127] = "evmwhssfaRRR", [1128] = "evmwlumiaRRR",
[1132] = "evmwhumiaRRR", [1133] = "evmwhsmiaRRR", [1135] = "evmwhsmfaRRR",
[1139] = "evmwssfaRRR", [1144] = "evmwumiaRRR", [1145] = "evmwsmiaRRR",
[1147] = "evmwsmfaRRR",
[1216] = "evaddusiaawRR", [1217] = "evaddssiaawRR",
[1218] = "evsubfusiaawRR", [1219] = "evsubfssiaawRR",
[1220] = "evmraRR",
[1222] = "evdivwsRRR", [1223] = "evdivwuRRR",
[1224] = "evaddumiaawRR", [1225] = "evaddsmiaawRR",
[1226] = "evsubfumiaawRR", [1227] = "evsubfsmiaawRR",
[1280] = "evmheusiaawRRR", [1281] = "evmhessiaawRRR",
[1283] = "evmhessfaawRRR", [1284] = "evmhousiaawRRR",
[1285] = "evmhossiaawRRR", [1287] = "evmhossfaawRRR",
[1288] = "evmheumiaawRRR", [1289] = "evmhesmiaawRRR",
[1291] = "evmhesmfaawRRR", [1292] = "evmhoumiaawRRR",
[1293] = "evmhosmiaawRRR", [1295] = "evmhosmfaawRRR",
[1320] = "evmhegumiaaRRR", [1321] = "evmhegsmiaaRRR",
[1323] = "evmhegsmfaaRRR", [1324] = "evmhogumiaaRRR",
[1325] = "evmhogsmiaaRRR", [1327] = "evmhogsmfaaRRR",
[1344] = "evmwlusiaawRRR", [1345] = "evmwlssiaawRRR",
[1352] = "evmwlumiaawRRR", [1353] = "evmwlsmiaawRRR",
[1363] = "evmwssfaaRRR", [1368] = "evmwumiaaRRR",
[1369] = "evmwsmiaaRRR", [1371] = "evmwsmfaaRRR",
[1408] = "evmheusianwRRR", [1409] = "evmhessianwRRR",
[1411] = "evmhessfanwRRR", [1412] = "evmhousianwRRR",
[1413] = "evmhossianwRRR", [1415] = "evmhossfanwRRR",
[1416] = "evmheumianwRRR", [1417] = "evmhesmianwRRR",
[1419] = "evmhesmfanwRRR", [1420] = "evmhoumianwRRR",
[1421] = "evmhosmianwRRR", [1423] = "evmhosmfanwRRR",
[1448] = "evmhegumianRRR", [1449] = "evmhegsmianRRR",
[1451] = "evmhegsmfanRRR", [1452] = "evmhogumianRRR",
[1453] = "evmhogsmianRRR", [1455] = "evmhogsmfanRRR",
[1472] = "evmwlusianwRRR", [1473] = "evmwlssianwRRR",
[1480] = "evmwlumianwRRR", [1481] = "evmwlsmianwRRR",
[1491] = "evmwssfanRRR", [1496] = "evmwumianRRR",
[1497] = "evmwsmianRRR", [1499] = "evmwsmfanRRR",
}
local map_pri = {
[0] = false, false, "tdiARI", "twiARI",
map_spe, false, false, "mulliRRI",
"subficRRI", false, "cmpl_iYLRU", "cmp_iYLRI",
"addicRRI", "addic.RRI", "addi|liRR0I", "addis|lisRR0I",
"b_KBJ", "sc", "bKJ", map_crops,
"rlwimiRR~AAA.", map_rlwinm, false, "rlwnmRR~RAA.",
"oriNRR~U", "orisRR~U", "xoriRR~U", "xorisRR~U",
"andi.RR~U", "andis.RR~U", map_rld, map_ext,
"lwzRRD", "lwzuRRD", "lbzRRD", "lbzuRRD",
"stwRRD", "stwuRRD", "stbRRD", "stbuRRD",
"lhzRRD", "lhzuRRD", "lhaRRD", "lhauRRD",
"sthRRD", "sthuRRD", "lmwRRD", "stmwRRD",
"lfsFRD", "lfsuFRD", "lfdFRD", "lfduFRD",
"stfsFRD", "stfsuFRD", "stfdFRD", "stfduFRD",
false, false, map_ld, map_fps,
false, false, map_std, map_fpd,
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "sp", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
"r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
"r24", "r25", "r26", "r27", "r28", "r29", "r30", "r31",
}
local map_cond = { [0] = "lt", "gt", "eq", "so", "ge", "le", "ne", "ns", }
-- Format a condition bit.
local function condfmt(cond)
if cond <= 3 then
return map_cond[band(cond, 3)]
else
return format("4*cr%d+%s", rshift(cond, 2), map_cond[band(cond, 3)])
end
end
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then extra = "\t->"..sym end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-7s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-7s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
local op = bor(lshift(b0, 24), lshift(b1, 16), lshift(b2, 8), b3)
local operands = {}
local last = nil
local rs = 21
ctx.op = op
ctx.rel = nil
local opat = map_pri[rshift(b0, 2)]
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
opat = opat[band(rshift(op, opat.shift), opat.mask)]
end
local name, pat = match(opat, "^([a-z0-9_.]*)(.*)")
local altname, pat2 = match(pat, "|([a-z0-9_.]*)(.*)")
if altname then pat = pat2 end
for p in gmatch(pat, ".") do
local x = nil
if p == "R" then
x = map_gpr[band(rshift(op, rs), 31)]
rs = rs - 5
elseif p == "F" then
x = "f"..band(rshift(op, rs), 31)
rs = rs - 5
elseif p == "A" then
x = band(rshift(op, rs), 31)
rs = rs - 5
elseif p == "S" then
x = arshift(lshift(op, 27-rs), 27)
rs = rs - 5
elseif p == "I" then
x = arshift(lshift(op, 16), 16)
elseif p == "U" then
x = band(op, 0xffff)
elseif p == "D" or p == "E" then
local disp = arshift(lshift(op, 16), 16)
if p == "E" then disp = band(disp, -4) end
if last == "r0" then last = "0" end
operands[#operands] = format("%d(%s)", disp, last)
elseif p >= "2" and p <= "8" then
local disp = band(rshift(op, rs), 31) * p
if last == "r0" then last = "0" end
operands[#operands] = format("%d(%s)", disp, last)
elseif p == "H" then
x = band(rshift(op, rs), 31) + lshift(band(op, 2), 4)
rs = rs - 5
elseif p == "M" then
x = band(rshift(op, rs), 31) + band(op, 0x20)
elseif p == "C" then
x = condfmt(band(rshift(op, rs), 31))
rs = rs - 5
elseif p == "B" then
local bo = rshift(op, 21)
local cond = band(rshift(op, 16), 31)
local cn = ""
rs = rs - 10
if band(bo, 4) == 0 then
cn = band(bo, 2) == 0 and "dnz" or "dz"
if band(bo, 0x10) == 0 then
cn = cn..(band(bo, 8) == 0 and "f" or "t")
end
if band(bo, 0x10) == 0 then x = condfmt(cond) end
name = name..(band(bo, 1) == band(rshift(op, 15), 1) and "-" or "+")
elseif band(bo, 0x10) == 0 then
cn = map_cond[band(cond, 3) + (band(bo, 8) == 0 and 4 or 0)]
if cond > 3 then x = "cr"..rshift(cond, 2) end
name = name..(band(bo, 1) == band(rshift(op, 15), 1) and "-" or "+")
end
name = gsub(name, "_", cn)
elseif p == "J" then
x = arshift(lshift(op, 27-rs), 29-rs)*4
if band(op, 2) == 0 then x = ctx.addr + pos + x end
ctx.rel = x
x = "0x"..tohex(x)
elseif p == "K" then
if band(op, 1) ~= 0 then name = name.."l" end
if band(op, 2) ~= 0 then name = name.."a" end
elseif p == "X" or p == "Y" then
x = band(rshift(op, rs+2), 7)
if x == 0 and p == "Y" then x = nil else x = "cr"..x end
rs = rs - 5
elseif p == "W" then
x = "cr"..band(op, 7)
elseif p == "Z" then
x = band(rshift(op, rs-4), 255)
rs = rs - 10
elseif p == ">" then
operands[#operands] = rshift(operands[#operands], 1)
elseif p == "0" then
if last == "r0" then
operands[#operands] = nil
if altname then name = altname end
end
elseif p == "L" then
name = gsub(name, "_", band(op, 0x00200000) ~= 0 and "d" or "w")
elseif p == "." then
if band(op, 1) == 1 then name = name.."." end
elseif p == "N" then
if op == 0x60000000 then name = "nop"; break end
elseif p == "~" then
local n = #operands
operands[n-1], operands[n] = operands[n], operands[n-1]
elseif p == "=" then
local n = #operands
if last == operands[n-1] then
operands[n] = nil
name = altname
end
elseif p == "%" then
local n = #operands
if last == operands[n-1] and last == operands[n-2] then
operands[n] = nil
operands[n-1] = nil
name = altname
end
elseif p == "-" then
rs = rs - 5
else
assert(false)
end
if x then operands[#operands+1] = x; last = x end
end
return putop(ctx, name, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
stop = stop - stop % 4
ctx.pos = ofs - ofs % 4
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 32 then return map_gpr[r] end
return "f"..(r-32)
end
-- Public module functions.
return {
create = create,
disass = disass,
regname = regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT x64 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the 64 bit functions from the combined
-- x86/x64 disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_x86 = require((string.match(..., ".*%.") or "").."dis_x86")
return {
create = dis_x86.create64,
disass = dis_x86.disass64,
regname = dis_x86.regname64
}
@@ -0,0 +1,953 @@
----------------------------------------------------------------------------
-- LuaJIT x86/x64 disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- Sending small code snippets to an external disassembler and mixing the
-- output with our own stuff was too fragile. So I had to bite the bullet
-- and write yet another x86 disassembler. Oh well ...
--
-- The output format is very similar to what ndisasm generates. But it has
-- been developed independently by looking at the opcode tables from the
-- Intel and AMD manuals. The supported instruction set is quite extensive
-- and reflects what a current generation Intel or AMD CPU implements in
-- 32 bit and 64 bit mode. Yes, this includes MMX, SSE, SSE2, SSE3, SSSE3,
-- SSE4.1, SSE4.2, SSE4a, AVX, AVX2 and even privileged and hypervisor
-- (VMX/SVM) instructions.
--
-- Notes:
-- * The (useless) a16 prefix, 3DNow and pre-586 opcodes are unsupported.
-- * No attempt at optimization has been made -- it's fast enough for my needs.
------------------------------------------------------------------------------
local type = type
local sub, byte, format = string.sub, string.byte, string.format
local match, gmatch, gsub = string.match, string.gmatch, string.gsub
local lower, rep = string.lower, string.rep
local bit = require("bit")
local tohex = bit.tohex
-- Map for 1st opcode byte in 32 bit mode. Ugly? Well ... read on.
local map_opc1_32 = {
--0x
[0]="addBmr","addVmr","addBrm","addVrm","addBai","addVai","push es","pop es",
"orBmr","orVmr","orBrm","orVrm","orBai","orVai","push cs","opc2*",
--1x
"adcBmr","adcVmr","adcBrm","adcVrm","adcBai","adcVai","push ss","pop ss",
"sbbBmr","sbbVmr","sbbBrm","sbbVrm","sbbBai","sbbVai","push ds","pop ds",
--2x
"andBmr","andVmr","andBrm","andVrm","andBai","andVai","es:seg","daa",
"subBmr","subVmr","subBrm","subVrm","subBai","subVai","cs:seg","das",
--3x
"xorBmr","xorVmr","xorBrm","xorVrm","xorBai","xorVai","ss:seg","aaa",
"cmpBmr","cmpVmr","cmpBrm","cmpVrm","cmpBai","cmpVai","ds:seg","aas",
--4x
"incVR","incVR","incVR","incVR","incVR","incVR","incVR","incVR",
"decVR","decVR","decVR","decVR","decVR","decVR","decVR","decVR",
--5x
"pushUR","pushUR","pushUR","pushUR","pushUR","pushUR","pushUR","pushUR",
"popUR","popUR","popUR","popUR","popUR","popUR","popUR","popUR",
--6x
"sz*pushaw,pusha","sz*popaw,popa","boundVrm","arplWmr",
"fs:seg","gs:seg","o16:","a16",
"pushUi","imulVrmi","pushBs","imulVrms",
"insb","insVS","outsb","outsVS",
--7x
"joBj","jnoBj","jbBj","jnbBj","jzBj","jnzBj","jbeBj","jaBj",
"jsBj","jnsBj","jpeBj","jpoBj","jlBj","jgeBj","jleBj","jgBj",
--8x
"arith!Bmi","arith!Vmi","arith!Bmi","arith!Vms",
"testBmr","testVmr","xchgBrm","xchgVrm",
"movBmr","movVmr","movBrm","movVrm",
"movVmg","leaVrm","movWgm","popUm",
--9x
"nop*xchgVaR|pause|xchgWaR|repne nop","xchgVaR","xchgVaR","xchgVaR",
"xchgVaR","xchgVaR","xchgVaR","xchgVaR",
"sz*cbw,cwde,cdqe","sz*cwd,cdq,cqo","call farViw","wait",
"sz*pushfw,pushf","sz*popfw,popf","sahf","lahf",
--Ax
"movBao","movVao","movBoa","movVoa",
"movsb","movsVS","cmpsb","cmpsVS",
"testBai","testVai","stosb","stosVS",
"lodsb","lodsVS","scasb","scasVS",
--Bx
"movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi",
"movVRI","movVRI","movVRI","movVRI","movVRI","movVRI","movVRI","movVRI",
--Cx
"shift!Bmu","shift!Vmu","retBw","ret","vex*3$lesVrm","vex*2$ldsVrm","movBmi","movVmi",
"enterBwu","leave","retfBw","retf","int3","intBu","into","iretVS",
--Dx
"shift!Bm1","shift!Vm1","shift!Bmc","shift!Vmc","aamBu","aadBu","salc","xlatb",
"fp*0","fp*1","fp*2","fp*3","fp*4","fp*5","fp*6","fp*7",
--Ex
"loopneBj","loopeBj","loopBj","sz*jcxzBj,jecxzBj,jrcxzBj",
"inBau","inVau","outBua","outVua",
"callVj","jmpVj","jmp farViw","jmpBj","inBad","inVad","outBda","outVda",
--Fx
"lock:","int1","repne:rep","rep:","hlt","cmc","testb!Bm","testv!Vm",
"clc","stc","cli","sti","cld","std","incb!Bm","incd!Vm",
}
assert(#map_opc1_32 == 255)
-- Map for 1st opcode byte in 64 bit mode (overrides only).
local map_opc1_64 = setmetatable({
[0x06]=false, [0x07]=false, [0x0e]=false,
[0x16]=false, [0x17]=false, [0x1e]=false, [0x1f]=false,
[0x27]=false, [0x2f]=false, [0x37]=false, [0x3f]=false,
[0x60]=false, [0x61]=false, [0x62]=false, [0x63]="movsxdVrDmt", [0x67]="a32:",
[0x40]="rex*", [0x41]="rex*b", [0x42]="rex*x", [0x43]="rex*xb",
[0x44]="rex*r", [0x45]="rex*rb", [0x46]="rex*rx", [0x47]="rex*rxb",
[0x48]="rex*w", [0x49]="rex*wb", [0x4a]="rex*wx", [0x4b]="rex*wxb",
[0x4c]="rex*wr", [0x4d]="rex*wrb", [0x4e]="rex*wrx", [0x4f]="rex*wrxb",
[0x82]=false, [0x9a]=false, [0xc4]="vex*3", [0xc5]="vex*2", [0xce]=false,
[0xd4]=false, [0xd5]=false, [0xd6]=false, [0xea]=false,
}, { __index = map_opc1_32 })
-- Map for 2nd opcode byte (0F xx). True CISC hell. Hey, I told you.
-- Prefix dependent MMX/SSE opcodes: (none)|rep|o16|repne, -|F3|66|F2
local map_opc2 = {
--0x
[0]="sldt!Dmp","sgdt!Ump","larVrm","lslVrm",nil,"syscall","clts","sysret",
"invd","wbinvd",nil,"ud1",nil,"$prefetch!Bm","femms","3dnowMrmu",
--1x
"movupsXrm|movssXrvm|movupdXrm|movsdXrvm",
"movupsXmr|movssXmvr|movupdXmr|movsdXmvr",
"movhlpsXrm$movlpsXrm|movsldupXrm|movlpdXrm|movddupXrm",
"movlpsXmr||movlpdXmr",
"unpcklpsXrvm||unpcklpdXrvm",
"unpckhpsXrvm||unpckhpdXrvm",
"movlhpsXrm$movhpsXrm|movshdupXrm|movhpdXrm",
"movhpsXmr||movhpdXmr",
"$prefetcht!Bm","hintnopVm","hintnopVm","hintnopVm",
"hintnopVm","hintnopVm","hintnopVm","hintnopVm",
--2x
"movUmx$","movUmy$","movUxm$","movUym$","movUmz$",nil,"movUzm$",nil,
"movapsXrm||movapdXrm",
"movapsXmr||movapdXmr",
"cvtpi2psXrMm|cvtsi2ssXrvVmt|cvtpi2pdXrMm|cvtsi2sdXrvVmt",
"movntpsXmr|movntssXmr|movntpdXmr|movntsdXmr",
"cvttps2piMrXm|cvttss2siVrXm|cvttpd2piMrXm|cvttsd2siVrXm",
"cvtps2piMrXm|cvtss2siVrXm|cvtpd2piMrXm|cvtsd2siVrXm",
"ucomissXrm||ucomisdXrm",
"comissXrm||comisdXrm",
--3x
"wrmsr","rdtsc","rdmsr","rdpmc","sysenter","sysexit",nil,"getsec",
"opc3*38",nil,"opc3*3a",nil,nil,nil,nil,nil,
--4x
"cmovoVrm","cmovnoVrm","cmovbVrm","cmovnbVrm",
"cmovzVrm","cmovnzVrm","cmovbeVrm","cmovaVrm",
"cmovsVrm","cmovnsVrm","cmovpeVrm","cmovpoVrm",
"cmovlVrm","cmovgeVrm","cmovleVrm","cmovgVrm",
--5x
"movmskpsVrXm$||movmskpdVrXm$","sqrtpsXrm|sqrtssXrm|sqrtpdXrm|sqrtsdXrm",
"rsqrtpsXrm|rsqrtssXrvm","rcppsXrm|rcpssXrvm",
"andpsXrvm||andpdXrvm","andnpsXrvm||andnpdXrvm",
"orpsXrvm||orpdXrvm","xorpsXrvm||xorpdXrvm",
"addpsXrvm|addssXrvm|addpdXrvm|addsdXrvm","mulpsXrvm|mulssXrvm|mulpdXrvm|mulsdXrvm",
"cvtps2pdXrm|cvtss2sdXrvm|cvtpd2psXrm|cvtsd2ssXrvm",
"cvtdq2psXrm|cvttps2dqXrm|cvtps2dqXrm",
"subpsXrvm|subssXrvm|subpdXrvm|subsdXrvm","minpsXrvm|minssXrvm|minpdXrvm|minsdXrvm",
"divpsXrvm|divssXrvm|divpdXrvm|divsdXrvm","maxpsXrvm|maxssXrvm|maxpdXrvm|maxsdXrvm",
--6x
"punpcklbwPrvm","punpcklwdPrvm","punpckldqPrvm","packsswbPrvm",
"pcmpgtbPrvm","pcmpgtwPrvm","pcmpgtdPrvm","packuswbPrvm",
"punpckhbwPrvm","punpckhwdPrvm","punpckhdqPrvm","packssdwPrvm",
"||punpcklqdqXrvm","||punpckhqdqXrvm",
"movPrVSm","movqMrm|movdquXrm|movdqaXrm",
--7x
"pshufwMrmu|pshufhwXrmu|pshufdXrmu|pshuflwXrmu","pshiftw!Pvmu",
"pshiftd!Pvmu","pshiftq!Mvmu||pshiftdq!Xvmu",
"pcmpeqbPrvm","pcmpeqwPrvm","pcmpeqdPrvm","emms*|",
"vmreadUmr||extrqXmuu$|insertqXrmuu$","vmwriteUrm||extrqXrm$|insertqXrm$",
nil,nil,
"||haddpdXrvm|haddpsXrvm","||hsubpdXrvm|hsubpsXrvm",
"movVSmMr|movqXrm|movVSmXr","movqMmr|movdquXmr|movdqaXmr",
--8x
"joVj","jnoVj","jbVj","jnbVj","jzVj","jnzVj","jbeVj","jaVj",
"jsVj","jnsVj","jpeVj","jpoVj","jlVj","jgeVj","jleVj","jgVj",
--9x
"setoBm","setnoBm","setbBm","setnbBm","setzBm","setnzBm","setbeBm","setaBm",
"setsBm","setnsBm","setpeBm","setpoBm","setlBm","setgeBm","setleBm","setgBm",
--Ax
"push fs","pop fs","cpuid","btVmr","shldVmru","shldVmrc",nil,nil,
"push gs","pop gs","rsm","btsVmr","shrdVmru","shrdVmrc","fxsave!Dmp","imulVrm",
--Bx
"cmpxchgBmr","cmpxchgVmr","$lssVrm","btrVmr",
"$lfsVrm","$lgsVrm","movzxVrBmt","movzxVrWmt",
"|popcntVrm","ud2Dp","bt!Vmu","btcVmr",
"bsfVrm","bsrVrm|lzcntVrm|bsrWrm","movsxVrBmt","movsxVrWmt",
--Cx
"xaddBmr","xaddVmr",
"cmppsXrvmu|cmpssXrvmu|cmppdXrvmu|cmpsdXrvmu","$movntiVmr|",
"pinsrwPrvWmu","pextrwDrPmu",
"shufpsXrvmu||shufpdXrvmu","$cmpxchg!Qmp",
"bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR",
--Dx
"||addsubpdXrvm|addsubpsXrvm","psrlwPrvm","psrldPrvm","psrlqPrvm",
"paddqPrvm","pmullwPrvm",
"|movq2dqXrMm|movqXmr|movdq2qMrXm$","pmovmskbVrMm||pmovmskbVrXm",
"psubusbPrvm","psubuswPrvm","pminubPrvm","pandPrvm",
"paddusbPrvm","padduswPrvm","pmaxubPrvm","pandnPrvm",
--Ex
"pavgbPrvm","psrawPrvm","psradPrvm","pavgwPrvm",
"pmulhuwPrvm","pmulhwPrvm",
"|cvtdq2pdXrm|cvttpd2dqXrm|cvtpd2dqXrm","$movntqMmr||$movntdqXmr",
"psubsbPrvm","psubswPrvm","pminswPrvm","porPrvm",
"paddsbPrvm","paddswPrvm","pmaxswPrvm","pxorPrvm",
--Fx
"|||lddquXrm","psllwPrvm","pslldPrvm","psllqPrvm",
"pmuludqPrvm","pmaddwdPrvm","psadbwPrvm","maskmovqMrm||maskmovdquXrm$",
"psubbPrvm","psubwPrvm","psubdPrvm","psubqPrvm",
"paddbPrvm","paddwPrvm","padddPrvm","ud",
}
assert(map_opc2[255] == "ud")
-- Map for three-byte opcodes. Can't wait for their next invention.
local map_opc3 = {
["38"] = { -- [66] 0f 38 xx
--0x
[0]="pshufbPrvm","phaddwPrvm","phadddPrvm","phaddswPrvm",
"pmaddubswPrvm","phsubwPrvm","phsubdPrvm","phsubswPrvm",
"psignbPrvm","psignwPrvm","psigndPrvm","pmulhrswPrvm",
"||permilpsXrvm","||permilpdXrvm",nil,nil,
--1x
"||pblendvbXrma",nil,nil,nil,
"||blendvpsXrma","||blendvpdXrma","||permpsXrvm","||ptestXrm",
"||broadcastssXrm","||broadcastsdXrm","||broadcastf128XrlXm",nil,
"pabsbPrm","pabswPrm","pabsdPrm",nil,
--2x
"||pmovsxbwXrm","||pmovsxbdXrm","||pmovsxbqXrm","||pmovsxwdXrm",
"||pmovsxwqXrm","||pmovsxdqXrm",nil,nil,
"||pmuldqXrvm","||pcmpeqqXrvm","||$movntdqaXrm","||packusdwXrvm",
"||maskmovpsXrvm","||maskmovpdXrvm","||maskmovpsXmvr","||maskmovpdXmvr",
--3x
"||pmovzxbwXrm","||pmovzxbdXrm","||pmovzxbqXrm","||pmovzxwdXrm",
"||pmovzxwqXrm","||pmovzxdqXrm","||permdXrvm","||pcmpgtqXrvm",
"||pminsbXrvm","||pminsdXrvm","||pminuwXrvm","||pminudXrvm",
"||pmaxsbXrvm","||pmaxsdXrvm","||pmaxuwXrvm","||pmaxudXrvm",
--4x
"||pmulddXrvm","||phminposuwXrm",nil,nil,
nil,"||psrlvVSXrvm","||psravdXrvm","||psllvVSXrvm",
--5x
[0x58] = "||pbroadcastdXrlXm",[0x59] = "||pbroadcastqXrlXm",
[0x5a] = "||broadcasti128XrlXm",
--7x
[0x78] = "||pbroadcastbXrlXm",[0x79] = "||pbroadcastwXrlXm",
--8x
[0x8c] = "||pmaskmovXrvVSm",
[0x8e] = "||pmaskmovVSmXvr",
--9x
[0x96] = "||fmaddsub132pHXrvm",[0x97] = "||fmsubadd132pHXrvm",
[0x98] = "||fmadd132pHXrvm",[0x99] = "||fmadd132sHXrvm",
[0x9a] = "||fmsub132pHXrvm",[0x9b] = "||fmsub132sHXrvm",
[0x9c] = "||fnmadd132pHXrvm",[0x9d] = "||fnmadd132sHXrvm",
[0x9e] = "||fnmsub132pHXrvm",[0x9f] = "||fnmsub132sHXrvm",
--Ax
[0xa6] = "||fmaddsub213pHXrvm",[0xa7] = "||fmsubadd213pHXrvm",
[0xa8] = "||fmadd213pHXrvm",[0xa9] = "||fmadd213sHXrvm",
[0xaa] = "||fmsub213pHXrvm",[0xab] = "||fmsub213sHXrvm",
[0xac] = "||fnmadd213pHXrvm",[0xad] = "||fnmadd213sHXrvm",
[0xae] = "||fnmsub213pHXrvm",[0xaf] = "||fnmsub213sHXrvm",
--Bx
[0xb6] = "||fmaddsub231pHXrvm",[0xb7] = "||fmsubadd231pHXrvm",
[0xb8] = "||fmadd231pHXrvm",[0xb9] = "||fmadd231sHXrvm",
[0xba] = "||fmsub231pHXrvm",[0xbb] = "||fmsub231sHXrvm",
[0xbc] = "||fnmadd231pHXrvm",[0xbd] = "||fnmadd231sHXrvm",
[0xbe] = "||fnmsub231pHXrvm",[0xbf] = "||fnmsub231sHXrvm",
--Dx
[0xdc] = "||aesencXrvm", [0xdd] = "||aesenclastXrvm",
[0xde] = "||aesdecXrvm", [0xdf] = "||aesdeclastXrvm",
--Fx
[0xf0] = "|||crc32TrBmt",[0xf1] = "|||crc32TrVmt",
[0xf7] = "| sarxVrmv| shlxVrmv| shrxVrmv",
},
["3a"] = { -- [66] 0f 3a xx
--0x
[0x00]="||permqXrmu","||permpdXrmu","||pblenddXrvmu",nil,
"||permilpsXrmu","||permilpdXrmu","||perm2f128Xrvmu",nil,
"||roundpsXrmu","||roundpdXrmu","||roundssXrvmu","||roundsdXrvmu",
"||blendpsXrvmu","||blendpdXrvmu","||pblendwXrvmu","palignrPrvmu",
--1x
nil,nil,nil,nil,
"||pextrbVmXru","||pextrwVmXru","||pextrVmSXru","||extractpsVmXru",
"||insertf128XrvlXmu","||extractf128XlXmYru",nil,nil,
nil,nil,nil,nil,
--2x
"||pinsrbXrvVmu","||insertpsXrvmu","||pinsrXrvVmuS",nil,
--3x
[0x38] = "||inserti128Xrvmu",[0x39] = "||extracti128XlXmYru",
--4x
[0x40] = "||dppsXrvmu",
[0x41] = "||dppdXrvmu",
[0x42] = "||mpsadbwXrvmu",
[0x44] = "||pclmulqdqXrvmu",
[0x46] = "||perm2i128Xrvmu",
[0x4a] = "||blendvpsXrvmb",[0x4b] = "||blendvpdXrvmb",
[0x4c] = "||pblendvbXrvmb",
--6x
[0x60] = "||pcmpestrmXrmu",[0x61] = "||pcmpestriXrmu",
[0x62] = "||pcmpistrmXrmu",[0x63] = "||pcmpistriXrmu",
[0xdf] = "||aeskeygenassistXrmu",
--Fx
[0xf0] = "||| rorxVrmu",
},
}
-- Map for VMX/SVM opcodes 0F 01 C0-FF (sgdt group with register operands).
local map_opcvm = {
[0xc1]="vmcall",[0xc2]="vmlaunch",[0xc3]="vmresume",[0xc4]="vmxoff",
[0xc8]="monitor",[0xc9]="mwait",
[0xd8]="vmrun",[0xd9]="vmmcall",[0xda]="vmload",[0xdb]="vmsave",
[0xdc]="stgi",[0xdd]="clgi",[0xde]="skinit",[0xdf]="invlpga",
[0xf8]="swapgs",[0xf9]="rdtscp",
}
-- Map for FP opcodes. And you thought stack machines are simple?
local map_opcfp = {
-- D8-DF 00-BF: opcodes with a memory operand.
-- D8
[0]="faddFm","fmulFm","fcomFm","fcompFm","fsubFm","fsubrFm","fdivFm","fdivrFm",
"fldFm",nil,"fstFm","fstpFm","fldenvVm","fldcwWm","fnstenvVm","fnstcwWm",
-- DA
"fiaddDm","fimulDm","ficomDm","ficompDm",
"fisubDm","fisubrDm","fidivDm","fidivrDm",
-- DB
"fildDm","fisttpDm","fistDm","fistpDm",nil,"fld twordFmp",nil,"fstp twordFmp",
-- DC
"faddGm","fmulGm","fcomGm","fcompGm","fsubGm","fsubrGm","fdivGm","fdivrGm",
-- DD
"fldGm","fisttpQm","fstGm","fstpGm","frstorDmp",nil,"fnsaveDmp","fnstswWm",
-- DE
"fiaddWm","fimulWm","ficomWm","ficompWm",
"fisubWm","fisubrWm","fidivWm","fidivrWm",
-- DF
"fildWm","fisttpWm","fistWm","fistpWm",
"fbld twordFmp","fildQm","fbstp twordFmp","fistpQm",
-- xx C0-FF: opcodes with a pseudo-register operand.
-- D8
"faddFf","fmulFf","fcomFf","fcompFf","fsubFf","fsubrFf","fdivFf","fdivrFf",
-- D9
"fldFf","fxchFf",{"fnop"},nil,
{"fchs","fabs",nil,nil,"ftst","fxam"},
{"fld1","fldl2t","fldl2e","fldpi","fldlg2","fldln2","fldz"},
{"f2xm1","fyl2x","fptan","fpatan","fxtract","fprem1","fdecstp","fincstp"},
{"fprem","fyl2xp1","fsqrt","fsincos","frndint","fscale","fsin","fcos"},
-- DA
"fcmovbFf","fcmoveFf","fcmovbeFf","fcmovuFf",nil,{nil,"fucompp"},nil,nil,
-- DB
"fcmovnbFf","fcmovneFf","fcmovnbeFf","fcmovnuFf",
{nil,nil,"fnclex","fninit"},"fucomiFf","fcomiFf",nil,
-- DC
"fadd toFf","fmul toFf",nil,nil,
"fsub toFf","fsubr toFf","fdivr toFf","fdiv toFf",
-- DD
"ffreeFf",nil,"fstFf","fstpFf","fucomFf","fucompFf",nil,nil,
-- DE
"faddpFf","fmulpFf",nil,{nil,"fcompp"},
"fsubrpFf","fsubpFf","fdivrpFf","fdivpFf",
-- DF
nil,nil,nil,nil,{"fnstsw ax"},"fucomipFf","fcomipFf",nil,
}
assert(map_opcfp[126] == "fcomipFf")
-- Map for opcode groups. The subkey is sp from the ModRM byte.
local map_opcgroup = {
arith = { "add", "or", "adc", "sbb", "and", "sub", "xor", "cmp" },
shift = { "rol", "ror", "rcl", "rcr", "shl", "shr", "sal", "sar" },
testb = { "testBmi", "testBmi", "not", "neg", "mul", "imul", "div", "idiv" },
testv = { "testVmi", "testVmi", "not", "neg", "mul", "imul", "div", "idiv" },
incb = { "inc", "dec" },
incd = { "inc", "dec", "callUmp", "$call farDmp",
"jmpUmp", "$jmp farDmp", "pushUm" },
sldt = { "sldt", "str", "lldt", "ltr", "verr", "verw" },
sgdt = { "vm*$sgdt", "vm*$sidt", "$lgdt", "vm*$lidt",
"smsw", nil, "lmsw", "vm*$invlpg" },
bt = { nil, nil, nil, nil, "bt", "bts", "btr", "btc" },
cmpxchg = { nil, "sz*,cmpxchg8bQmp,cmpxchg16bXmp", nil, nil,
nil, nil, "vmptrld|vmxon|vmclear", "vmptrst" },
pshiftw = { nil, nil, "psrlw", nil, "psraw", nil, "psllw" },
pshiftd = { nil, nil, "psrld", nil, "psrad", nil, "pslld" },
pshiftq = { nil, nil, "psrlq", nil, nil, nil, "psllq" },
pshiftdq = { nil, nil, "psrlq", "psrldq", nil, nil, "psllq", "pslldq" },
fxsave = { "$fxsave", "$fxrstor", "$ldmxcsr", "$stmxcsr",
nil, "lfenceDp$", "mfenceDp$", "sfenceDp$clflush" },
prefetch = { "prefetch", "prefetchw" },
prefetcht = { "prefetchnta", "prefetcht0", "prefetcht1", "prefetcht2" },
}
------------------------------------------------------------------------------
-- Maps for register names.
local map_regs = {
B = { "al", "cl", "dl", "bl", "ah", "ch", "dh", "bh",
"r8b", "r9b", "r10b", "r11b", "r12b", "r13b", "r14b", "r15b" },
B64 = { "al", "cl", "dl", "bl", "spl", "bpl", "sil", "dil",
"r8b", "r9b", "r10b", "r11b", "r12b", "r13b", "r14b", "r15b" },
W = { "ax", "cx", "dx", "bx", "sp", "bp", "si", "di",
"r8w", "r9w", "r10w", "r11w", "r12w", "r13w", "r14w", "r15w" },
D = { "eax", "ecx", "edx", "ebx", "esp", "ebp", "esi", "edi",
"r8d", "r9d", "r10d", "r11d", "r12d", "r13d", "r14d", "r15d" },
Q = { "rax", "rcx", "rdx", "rbx", "rsp", "rbp", "rsi", "rdi",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15" },
M = { "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
"mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7" }, -- No x64 ext!
X = { "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7",
"xmm8", "xmm9", "xmm10", "xmm11", "xmm12", "xmm13", "xmm14", "xmm15" },
Y = { "ymm0", "ymm1", "ymm2", "ymm3", "ymm4", "ymm5", "ymm6", "ymm7",
"ymm8", "ymm9", "ymm10", "ymm11", "ymm12", "ymm13", "ymm14", "ymm15" },
}
local map_segregs = { "es", "cs", "ss", "ds", "fs", "gs", "segr6", "segr7" }
-- Maps for size names.
local map_sz2n = {
B = 1, W = 2, D = 4, Q = 8, M = 8, X = 16, Y = 32,
}
local map_sz2prefix = {
B = "byte", W = "word", D = "dword",
Q = "qword",
M = "qword", X = "xword", Y = "yword",
F = "dword", G = "qword", -- No need for sizes/register names for these two.
}
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local code, pos, hex = ctx.code, ctx.pos, ""
local hmax = ctx.hexdump
if hmax > 0 then
for i=ctx.start,pos-1 do
hex = hex..format("%02X", byte(code, i, i))
end
if #hex > hmax then hex = sub(hex, 1, hmax)..". "
else hex = hex..rep(" ", hmax-#hex+2) end
end
if operands then text = text.." "..operands end
if ctx.o16 then text = "o16 "..text; ctx.o16 = false end
if ctx.a32 then text = "a32 "..text; ctx.a32 = false end
if ctx.rep then text = ctx.rep.." "..text; ctx.rep = false end
if ctx.rex then
local t = (ctx.rexw and "w" or "")..(ctx.rexr and "r" or "")..
(ctx.rexx and "x" or "")..(ctx.rexb and "b" or "")..
(ctx.vexl and "l" or "")
if ctx.vexv and ctx.vexv ~= 0 then t = t.."v"..ctx.vexv end
if t ~= "" then text = ctx.rex.."."..t.." "..gsub(text, "^ ", "")
elseif ctx.rex == "vex" then text = gsub("v"..text, "^v ", "") end
ctx.rexw = false; ctx.rexr = false; ctx.rexx = false; ctx.rexb = false
ctx.rex = false; ctx.vexl = false; ctx.vexv = false
end
if ctx.seg then
local text2, n = gsub(text, "%[", "["..ctx.seg..":")
if n == 0 then text = ctx.seg.." "..text else text = text2 end
ctx.seg = false
end
if ctx.lock then text = "lock "..text; ctx.lock = false end
local imm = ctx.imm
if imm then
local sym = ctx.symtab[imm]
if sym then text = text.."\t->"..sym end
end
ctx.out(format("%08x %s%s\n", ctx.addr+ctx.start, hex, text))
ctx.mrm = false
ctx.vexv = false
ctx.start = pos
ctx.imm = nil
end
-- Clear all prefix flags.
local function clearprefixes(ctx)
ctx.o16 = false; ctx.seg = false; ctx.lock = false; ctx.rep = false
ctx.rexw = false; ctx.rexr = false; ctx.rexx = false; ctx.rexb = false
ctx.rex = false; ctx.a32 = false; ctx.vexl = false
end
-- Fallback for incomplete opcodes at the end.
local function incomplete(ctx)
ctx.pos = ctx.stop+1
clearprefixes(ctx)
return putop(ctx, "(incomplete)")
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
clearprefixes(ctx)
return putop(ctx, "(unknown)")
end
-- Return an immediate of the specified size.
local function getimm(ctx, pos, n)
if pos+n-1 > ctx.stop then return incomplete(ctx) end
local code = ctx.code
if n == 1 then
local b1 = byte(code, pos, pos)
return b1
elseif n == 2 then
local b1, b2 = byte(code, pos, pos+1)
return b1+b2*256
else
local b1, b2, b3, b4 = byte(code, pos, pos+3)
local imm = b1+b2*256+b3*65536+b4*16777216
ctx.imm = imm
return imm
end
end
-- Process pattern string and generate the operands.
local function putpat(ctx, name, pat)
local operands, regs, sz, mode, sp, rm, sc, rx, sdisp
local code, pos, stop, vexl = ctx.code, ctx.pos, ctx.stop, ctx.vexl
-- Chars used: 1DFGHIMPQRSTUVWXYabcdfgijlmoprstuvwxyz
for p in gmatch(pat, ".") do
local x = nil
if p == "V" or p == "U" then
if ctx.rexw then sz = "Q"; ctx.rexw = false
elseif ctx.o16 then sz = "W"; ctx.o16 = false
elseif p == "U" and ctx.x64 then sz = "Q"
else sz = "D" end
regs = map_regs[sz]
elseif p == "T" then
if ctx.rexw then sz = "Q"; ctx.rexw = false else sz = "D" end
regs = map_regs[sz]
elseif p == "B" then
sz = "B"
regs = ctx.rex and map_regs.B64 or map_regs.B
elseif match(p, "[WDQMXYFG]") then
sz = p
if sz == "X" and vexl then sz = "Y"; ctx.vexl = false end
regs = map_regs[sz]
elseif p == "P" then
sz = ctx.o16 and "X" or "M"; ctx.o16 = false
if sz == "X" and vexl then sz = "Y"; ctx.vexl = false end
regs = map_regs[sz]
elseif p == "H" then
name = name..(ctx.rexw and "d" or "s")
ctx.rexw = false
elseif p == "S" then
name = name..lower(sz)
elseif p == "s" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = imm <= 127 and format("+0x%02x", imm)
or format("-0x%02x", 256-imm)
pos = pos+1
elseif p == "u" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = format("0x%02x", imm)
pos = pos+1
elseif p == "b" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = regs[imm/16+1]
pos = pos+1
elseif p == "w" then
local imm = getimm(ctx, pos, 2); if not imm then return end
x = format("0x%x", imm)
pos = pos+2
elseif p == "o" then -- [offset]
if ctx.x64 then
local imm1 = getimm(ctx, pos, 4); if not imm1 then return end
local imm2 = getimm(ctx, pos+4, 4); if not imm2 then return end
x = format("[0x%08x%08x]", imm2, imm1)
pos = pos+8
else
local imm = getimm(ctx, pos, 4); if not imm then return end
x = format("[0x%08x]", imm)
pos = pos+4
end
elseif p == "i" or p == "I" then
local n = map_sz2n[sz]
if n == 8 and ctx.x64 and p == "I" then
local imm1 = getimm(ctx, pos, 4); if not imm1 then return end
local imm2 = getimm(ctx, pos+4, 4); if not imm2 then return end
x = format("0x%08x%08x", imm2, imm1)
else
if n == 8 then n = 4 end
local imm = getimm(ctx, pos, n); if not imm then return end
if sz == "Q" and (imm < 0 or imm > 0x7fffffff) then
imm = (0xffffffff+1)-imm
x = format(imm > 65535 and "-0x%08x" or "-0x%x", imm)
else
x = format(imm > 65535 and "0x%08x" or "0x%x", imm)
end
end
pos = pos+n
elseif p == "j" then
local n = map_sz2n[sz]
if n == 8 then n = 4 end
local imm = getimm(ctx, pos, n); if not imm then return end
if sz == "B" and imm > 127 then imm = imm-256
elseif imm > 2147483647 then imm = imm-4294967296 end
pos = pos+n
imm = imm + pos + ctx.addr
if imm > 4294967295 and not ctx.x64 then imm = imm-4294967296 end
ctx.imm = imm
if sz == "W" then
x = format("word 0x%04x", imm%65536)
elseif ctx.x64 then
local lo = imm % 0x1000000
x = format("0x%02x%06x", (imm-lo) / 0x1000000, lo)
else
x = "0x"..tohex(imm)
end
elseif p == "R" then
local r = byte(code, pos-1, pos-1)%8
if ctx.rexb then r = r + 8; ctx.rexb = false end
x = regs[r+1]
elseif p == "a" then x = regs[1]
elseif p == "c" then x = "cl"
elseif p == "d" then x = "dx"
elseif p == "1" then x = "1"
else
if not mode then
mode = ctx.mrm
if not mode then
if pos > stop then return incomplete(ctx) end
mode = byte(code, pos, pos)
pos = pos+1
end
rm = mode%8; mode = (mode-rm)/8
sp = mode%8; mode = (mode-sp)/8
sdisp = ""
if mode < 3 then
if rm == 4 then
if pos > stop then return incomplete(ctx) end
sc = byte(code, pos, pos)
pos = pos+1
rm = sc%8; sc = (sc-rm)/8
rx = sc%8; sc = (sc-rx)/8
if ctx.rexx then rx = rx + 8; ctx.rexx = false end
if rx == 4 then rx = nil end
end
if mode > 0 or rm == 5 then
local dsz = mode
if dsz ~= 1 then dsz = 4 end
local disp = getimm(ctx, pos, dsz); if not disp then return end
if mode == 0 then rm = nil end
if rm or rx or (not sc and ctx.x64 and not ctx.a32) then
if dsz == 1 and disp > 127 then
sdisp = format("-0x%x", 256-disp)
elseif disp >= 0 and disp <= 0x7fffffff then
sdisp = format("+0x%x", disp)
else
sdisp = format("-0x%x", (0xffffffff+1)-disp)
end
else
sdisp = format(ctx.x64 and not ctx.a32 and
not (disp >= 0 and disp <= 0x7fffffff)
and "0xffffffff%08x" or "0x%08x", disp)
end
pos = pos+dsz
end
end
if rm and ctx.rexb then rm = rm + 8; ctx.rexb = false end
if ctx.rexr then sp = sp + 8; ctx.rexr = false end
end
if p == "m" then
if mode == 3 then x = regs[rm+1]
else
local aregs = ctx.a32 and map_regs.D or ctx.aregs
local srm, srx = "", ""
if rm then srm = aregs[rm+1]
elseif not sc and ctx.x64 and not ctx.a32 then srm = "rip" end
ctx.a32 = false
if rx then
if rm then srm = srm.."+" end
srx = aregs[rx+1]
if sc > 0 then srx = srx.."*"..(2^sc) end
end
x = format("[%s%s%s]", srm, srx, sdisp)
end
if mode < 3 and
(not match(pat, "[aRrgp]") or match(pat, "t")) then -- Yuck.
x = map_sz2prefix[sz].." "..x
end
elseif p == "r" then x = regs[sp+1]
elseif p == "g" then x = map_segregs[sp+1]
elseif p == "p" then -- Suppress prefix.
elseif p == "f" then x = "st"..rm
elseif p == "x" then
if sp == 0 and ctx.lock and not ctx.x64 then
x = "CR8"; ctx.lock = false
else
x = "CR"..sp
end
elseif p == "v" then
if ctx.vexv then
x = regs[ctx.vexv+1]; ctx.vexv = false
end
elseif p == "y" then x = "DR"..sp
elseif p == "z" then x = "TR"..sp
elseif p == "l" then vexl = false
elseif p == "t" then
else
error("bad pattern `"..pat.."'")
end
end
if x then operands = operands and operands..", "..x or x end
end
ctx.pos = pos
return putop(ctx, name, operands)
end
-- Forward declaration.
local map_act
-- Fetch and cache MRM byte.
local function getmrm(ctx)
local mrm = ctx.mrm
if not mrm then
local pos = ctx.pos
if pos > ctx.stop then return nil end
mrm = byte(ctx.code, pos, pos)
ctx.pos = pos+1
ctx.mrm = mrm
end
return mrm
end
-- Dispatch to handler depending on pattern.
local function dispatch(ctx, opat, patgrp)
if not opat then return unknown(ctx) end
if match(opat, "%|") then -- MMX/SSE variants depending on prefix.
local p
if ctx.rep then
p = ctx.rep=="rep" and "%|([^%|]*)" or "%|[^%|]*%|[^%|]*%|([^%|]*)"
ctx.rep = false
elseif ctx.o16 then p = "%|[^%|]*%|([^%|]*)"; ctx.o16 = false
else p = "^[^%|]*" end
opat = match(opat, p)
if not opat then return unknown(ctx) end
-- ctx.rep = false; ctx.o16 = false
--XXX fails for 66 f2 0f 38 f1 06 crc32 eax,WORD PTR [esi]
--XXX remove in branches?
end
if match(opat, "%$") then -- reg$mem variants.
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
opat = match(opat, mrm >= 192 and "^[^%$]*" or "%$(.*)")
if opat == "" then return unknown(ctx) end
end
if opat == "" then return unknown(ctx) end
local name, pat = match(opat, "^([a-z0-9 ]*)(.*)")
if pat == "" and patgrp then pat = patgrp end
return map_act[sub(pat, 1, 1)](ctx, name, pat)
end
-- Get a pattern from an opcode map and dispatch to handler.
local function dispatchmap(ctx, opcmap)
local pos = ctx.pos
local opat = opcmap[byte(ctx.code, pos, pos)]
pos = pos + 1
ctx.pos = pos
return dispatch(ctx, opat)
end
-- Map for action codes. The key is the first char after the name.
map_act = {
-- Simple opcodes without operands.
[""] = function(ctx, name, pat)
return putop(ctx, name)
end,
-- Operand size chars fall right through.
B = putpat, W = putpat, D = putpat, Q = putpat,
V = putpat, U = putpat, T = putpat,
M = putpat, X = putpat, P = putpat,
F = putpat, G = putpat, Y = putpat,
H = putpat,
-- Collect prefixes.
[":"] = function(ctx, name, pat)
ctx[pat == ":" and name or sub(pat, 2)] = name
if ctx.pos - ctx.start > 5 then return unknown(ctx) end -- Limit #prefixes.
end,
-- Chain to special handler specified by name.
["*"] = function(ctx, name, pat)
return map_act[name](ctx, name, sub(pat, 2))
end,
-- Use named subtable for opcode group.
["!"] = function(ctx, name, pat)
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
return dispatch(ctx, map_opcgroup[name][((mrm-(mrm%8))/8)%8+1], sub(pat, 2))
end,
-- o16,o32[,o64] variants.
sz = function(ctx, name, pat)
if ctx.o16 then ctx.o16 = false
else
pat = match(pat, ",(.*)")
if ctx.rexw then
local p = match(pat, ",(.*)")
if p then pat = p; ctx.rexw = false end
end
end
pat = match(pat, "^[^,]*")
return dispatch(ctx, pat)
end,
-- Two-byte opcode dispatch.
opc2 = function(ctx, name, pat)
return dispatchmap(ctx, map_opc2)
end,
-- Three-byte opcode dispatch.
opc3 = function(ctx, name, pat)
return dispatchmap(ctx, map_opc3[pat])
end,
-- VMX/SVM dispatch.
vm = function(ctx, name, pat)
return dispatch(ctx, map_opcvm[ctx.mrm])
end,
-- Floating point opcode dispatch.
fp = function(ctx, name, pat)
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
local rm = mrm%8
local idx = pat*8 + ((mrm-rm)/8)%8
if mrm >= 192 then idx = idx + 64 end
local opat = map_opcfp[idx]
if type(opat) == "table" then opat = opat[rm+1] end
return dispatch(ctx, opat)
end,
-- REX prefix.
rex = function(ctx, name, pat)
if ctx.rex then return unknown(ctx) end -- Only 1 REX or VEX prefix allowed.
for p in gmatch(pat, ".") do ctx["rex"..p] = true end
ctx.rex = "rex"
end,
-- VEX prefix.
vex = function(ctx, name, pat)
if ctx.rex then return unknown(ctx) end -- Only 1 REX or VEX prefix allowed.
ctx.rex = "vex"
local pos = ctx.pos
if ctx.mrm then
ctx.mrm = nil
pos = pos-1
end
local b = byte(ctx.code, pos, pos)
if not b then return incomplete(ctx) end
pos = pos+1
if b < 128 then ctx.rexr = true end
local m = 1
if pat == "3" then
m = b%32; b = (b-m)/32
local nb = b%2; b = (b-nb)/2
if nb == 0 then ctx.rexb = true end
local nx = b%2
if nx == 0 then ctx.rexx = true end
b = byte(ctx.code, pos, pos)
if not b then return incomplete(ctx) end
pos = pos+1
if b >= 128 then ctx.rexw = true end
end
ctx.pos = pos
local map
if m == 1 then map = map_opc2
elseif m == 2 then map = map_opc3["38"]
elseif m == 3 then map = map_opc3["3a"]
else return unknown(ctx) end
local p = b%4; b = (b-p)/4
if p == 1 then ctx.o16 = "o16"
elseif p == 2 then ctx.rep = "rep"
elseif p == 3 then ctx.rep = "repne" end
local l = b%2; b = (b-l)/2
if l ~= 0 then ctx.vexl = true end
ctx.vexv = (-1-b)%16
return dispatchmap(ctx, map)
end,
-- Special case for nop with REX prefix.
nop = function(ctx, name, pat)
return dispatch(ctx, ctx.rex and pat or "nop")
end,
-- Special case for 0F 77.
emms = function(ctx, name, pat)
if ctx.rex ~= "vex" then
return putop(ctx, "emms")
elseif ctx.vexl then
ctx.vexl = false
return putop(ctx, "zeroall")
else
return putop(ctx, "zeroupper")
end
end,
}
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
ofs = ofs + 1
ctx.start = ofs
ctx.pos = ofs
ctx.stop = stop
ctx.imm = nil
ctx.mrm = false
clearprefixes(ctx)
while ctx.pos <= stop do dispatchmap(ctx, ctx.map1) end
if ctx.pos ~= ctx.start then incomplete(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = (addr or 0) - 1
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 16
ctx.x64 = false
ctx.map1 = map_opc1_32
ctx.aregs = map_regs.D
return ctx
end
local function create64(code, addr, out)
local ctx = create(code, addr, out)
ctx.x64 = true
ctx.map1 = map_opc1_64
ctx.aregs = map_regs.Q
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
local function disass64(code, addr, out)
create64(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 8 then return map_regs.D[r+1] end
return map_regs.X[r-7]
end
local function regname64(r)
if r < 16 then return map_regs.Q[r+1] end
return map_regs.X[r-15]
end
-- Public module functions.
return {
create = create,
create64 = create64,
disass = disass,
disass64 = disass64,
regname = regname,
regname64 = regname64
}
@@ -0,0 +1,726 @@
----------------------------------------------------------------------------
-- LuaJIT compiler dump module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module can be used to debug the JIT compiler itself. It dumps the
-- code representations and structures used in various compiler stages.
--
-- Example usage:
--
-- luajit -jdump -e "local x=0; for i=1,1e6 do x=x+i end; print(x)"
-- luajit -jdump=im -e "for i=1,1000 do for j=1,1000 do end end" | less -R
-- luajit -jdump=is myapp.lua | less -R
-- luajit -jdump=-b myapp.lua
-- luajit -jdump=+aH,myapp.html myapp.lua
-- luajit -jdump=ixT,myapp.dump myapp.lua
--
-- The first argument specifies the dump mode. The second argument gives
-- the output file name. Default output is to stdout, unless the environment
-- variable LUAJIT_DUMPFILE is set. The file is overwritten every time the
-- module is started.
--
-- Different features can be turned on or off with the dump mode. If the
-- mode starts with a '+', the following features are added to the default
-- set of features; a '-' removes them. Otherwise the features are replaced.
--
-- The following dump features are available (* marks the default):
--
-- * t Print a line for each started, ended or aborted trace (see also -jv).
-- * b Dump the traced bytecode.
-- * i Dump the IR (intermediate representation).
-- r Augment the IR with register/stack slots.
-- s Dump the snapshot map.
-- * m Dump the generated machine code.
-- x Print each taken trace exit.
-- X Print each taken trace exit and the contents of all registers.
-- a Print the IR of aborted traces, too.
--
-- The output format can be set with the following characters:
--
-- T Plain text output.
-- A ANSI-colored text output
-- H Colorized HTML + CSS output.
--
-- The default output format is plain text. It's set to ANSI-colored text
-- if the COLORTERM variable is set. Note: this is independent of any output
-- redirection, which is actually considered a feature.
--
-- You probably want to use less -R to enjoy viewing ANSI-colored text from
-- a pipe or a file. Add this to your ~/.bashrc: export LESS="-R"
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local funcinfo, funcbc = jutil.funcinfo, jutil.funcbc
local traceinfo, traceir, tracek = jutil.traceinfo, jutil.traceir, jutil.tracek
local tracemc, tracesnap = jutil.tracemc, jutil.tracesnap
local traceexitstub, ircalladdr = jutil.traceexitstub, jutil.ircalladdr
local bit = require("bit")
local band, shr, tohex = bit.band, bit.rshift, bit.tohex
local sub, gsub, format = string.sub, string.gsub, string.format
local byte, rep = string.byte, string.rep
local type, tostring = type, tostring
local stdout, stderr = io.stdout, io.stderr
-- Load other modules on-demand.
local bcline, disass
-- Active flag, output file handle and dump mode.
local active, out, dumpmode
------------------------------------------------------------------------------
local symtabmt = { __index = false }
local symtab = {}
local nexitsym = 0
-- Fill nested symbol table with per-trace exit stub addresses.
local function fillsymtab_tr(tr, nexit)
local t = {}
symtabmt.__index = t
if jit.arch:sub(1, 4) == "mips" then
t[traceexitstub(tr, 0)] = "exit"
return
end
for i=0,nexit-1 do
local addr = traceexitstub(tr, i)
if addr < 0 then addr = addr + 2^32 end
t[addr] = tostring(i)
end
local addr = traceexitstub(tr, nexit)
if addr then t[addr] = "stack_check" end
end
-- Fill symbol table with trace exit stub addresses.
local function fillsymtab(tr, nexit)
local t = symtab
if nexitsym == 0 then
local maskaddr = jit.arch == "arm" and -2
local ircall = vmdef.ircall
for i=0,#ircall do
local addr = ircalladdr(i)
if addr ~= 0 then
if maskaddr then addr = band(addr, maskaddr) end
if addr < 0 then addr = addr + 2^32 end
t[addr] = ircall[i]
end
end
end
if nexitsym == 1000000 then -- Per-trace exit stubs.
fillsymtab_tr(tr, nexit)
elseif nexit > nexitsym then -- Shared exit stubs.
for i=nexitsym,nexit-1 do
local addr = traceexitstub(i)
if addr == nil then -- Fall back to per-trace exit stubs.
fillsymtab_tr(tr, nexit)
setmetatable(symtab, symtabmt)
nexit = 1000000
break
end
if addr < 0 then addr = addr + 2^32 end
t[addr] = tostring(i)
end
nexitsym = nexit
end
return t
end
local function dumpwrite(s)
out:write(s)
end
-- Disassemble machine code.
local function dump_mcode(tr)
local info = traceinfo(tr)
if not info then return end
local mcode, addr, loop = tracemc(tr)
if not mcode then return end
if not disass then disass = require("jit.dis_"..jit.arch) end
if addr < 0 then addr = addr + 2^32 end
out:write("---- TRACE ", tr, " mcode ", #mcode, "\n")
local ctx = disass.create(mcode, addr, dumpwrite)
ctx.hexdump = 0
ctx.symtab = fillsymtab(tr, info.nexit)
if loop ~= 0 then
symtab[addr+loop] = "LOOP"
ctx:disass(0, loop)
out:write("->LOOP:\n")
ctx:disass(loop, #mcode-loop)
symtab[addr+loop] = nil
else
ctx:disass(0, #mcode)
end
end
------------------------------------------------------------------------------
local irtype_text = {
[0] = "nil",
"fal",
"tru",
"lud",
"str",
"p32",
"thr",
"pro",
"fun",
"p64",
"cdt",
"tab",
"udt",
"flt",
"num",
"i8 ",
"u8 ",
"i16",
"u16",
"int",
"u32",
"i64",
"u64",
"sfp",
}
local colortype_ansi = {
[0] = "%s",
"%s",
"%s",
"\027[36m%s\027[m",
"\027[32m%s\027[m",
"%s",
"\027[1m%s\027[m",
"%s",
"\027[1m%s\027[m",
"%s",
"\027[33m%s\027[m",
"\027[31m%s\027[m",
"\027[36m%s\027[m",
"\027[34m%s\027[m",
"\027[34m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
}
local function colorize_text(s)
return s
end
local function colorize_ansi(s, t, extra)
local out = format(colortype_ansi[t], s)
if extra then out = "\027[3m"..out end
return out
end
local irtype_ansi = setmetatable({},
{ __index = function(tab, t)
local s = colorize_ansi(irtype_text[t], t); tab[t] = s; return s; end })
local html_escape = { ["<"] = "&lt;", [">"] = "&gt;", ["&"] = "&amp;", }
local function colorize_html(s, t, extra)
s = gsub(s, "[<>&]", html_escape)
return format('<span class="irt_%s%s">%s</span>',
irtype_text[t], extra and " irt_extra" or "", s)
end
local irtype_html = setmetatable({},
{ __index = function(tab, t)
local s = colorize_html(irtype_text[t], t); tab[t] = s; return s; end })
local header_html = [[
<style type="text/css">
background { background: #ffffff; color: #000000; }
pre.ljdump {
font-size: 10pt;
background: #f0f4ff;
color: #000000;
border: 1px solid #bfcfff;
padding: 0.5em;
margin-left: 2em;
margin-right: 2em;
}
span.irt_str { color: #00a000; }
span.irt_thr, span.irt_fun { color: #404040; font-weight: bold; }
span.irt_tab { color: #c00000; }
span.irt_udt, span.irt_lud { color: #00c0c0; }
span.irt_num { color: #4040c0; }
span.irt_int, span.irt_i8, span.irt_u8, span.irt_i16, span.irt_u16 { color: #b040b0; }
span.irt_extra { font-style: italic; }
</style>
]]
local colorize, irtype
-- Lookup tables to convert some literals into names.
local litname = {
["SLOAD "] = setmetatable({}, { __index = function(t, mode)
local s = ""
if band(mode, 1) ~= 0 then s = s.."P" end
if band(mode, 2) ~= 0 then s = s.."F" end
if band(mode, 4) ~= 0 then s = s.."T" end
if band(mode, 8) ~= 0 then s = s.."C" end
if band(mode, 16) ~= 0 then s = s.."R" end
if band(mode, 32) ~= 0 then s = s.."I" end
if band(mode, 64) ~= 0 then s = s.."K" end
t[mode] = s
return s
end}),
["XLOAD "] = { [0] = "", "R", "V", "RV", "U", "RU", "VU", "RVU", },
["CONV "] = setmetatable({}, { __index = function(t, mode)
local s = irtype[band(mode, 31)]
s = irtype[band(shr(mode, 5), 31)].."."..s
if band(mode, 0x800) ~= 0 then s = s.." sext" end
local c = shr(mode, 12)
if c == 1 then s = s.." none"
elseif c == 2 then s = s.." index"
elseif c == 3 then s = s.." check" end
t[mode] = s
return s
end}),
["FLOAD "] = vmdef.irfield,
["FREF "] = vmdef.irfield,
["FPMATH"] = vmdef.irfpm,
["TMPREF"] = { [0] = "", "IN", "OUT", "INOUT", "", "", "OUT2", "INOUT2" },
["BUFHDR"] = { [0] = "RESET", "APPEND", "WRITE" },
["TOSTR "] = { [0] = "INT", "NUM", "CHAR" },
}
local function ctlsub(c)
if c == "\n" then return "\\n"
elseif c == "\r" then return "\\r"
elseif c == "\t" then return "\\t"
else return format("\\%03d", byte(c))
end
end
local function fmtfunc(func, pc)
local fi = funcinfo(func, pc)
if fi.loc then
return fi.loc
elseif fi.ffid then
return vmdef.ffnames[fi.ffid]
elseif fi.addr then
return format("C:%x", fi.addr)
else
return "(?)"
end
end
local function formatk(tr, idx, sn)
local k, t, slot = tracek(tr, idx)
local tn = type(k)
local s
if tn == "number" then
if t < 12 then
s = k == 0 and "NULL" or format("[0x%08x]", k)
elseif band(sn or 0, 0x30000) ~= 0 then
s = band(sn, 0x20000) ~= 0 and "contpc" or "ftsz"
elseif k == 2^52+2^51 then
s = "bias"
else
s = format(0 < k and k < 0x1p-1026 and "%+a" or "%+.14g", k)
end
elseif tn == "string" then
s = format(#k > 20 and '"%.20s"~' or '"%s"', gsub(k, "%c", ctlsub))
elseif tn == "function" then
s = fmtfunc(k)
elseif tn == "table" then
s = format("{%p}", k)
elseif tn == "userdata" then
if t == 12 then
s = format("userdata:%p", k)
else
s = format("[%p]", k)
if s == "[NULL]" then s = "NULL" end
end
elseif t == 21 then -- int64_t
s = sub(tostring(k), 1, -3)
if sub(s, 1, 1) ~= "-" then s = "+"..s end
elseif sn == 0x1057fff then -- SNAP(1, SNAP_FRAME | SNAP_NORESTORE, REF_NIL)
return "----" -- Special case for LJ_FR2 slot 1.
else
s = tostring(k) -- For primitives.
end
s = colorize(format("%-4s", s), t, band(sn or 0, 0x100000) ~= 0)
if slot then
s = format("%s @%d", s, slot)
end
return s
end
local function printsnap(tr, snap)
local n = 2
for s=0,snap[1]-1 do
local sn = snap[n]
if shr(sn, 24) == s then
n = n + 1
local ref = band(sn, 0xffff) - 0x8000 -- REF_BIAS
if ref < 0 then
out:write(formatk(tr, ref, sn))
elseif band(sn, 0x80000) ~= 0 then -- SNAP_SOFTFPNUM
out:write(colorize(format("%04d/%04d", ref, ref+1), 14))
else
local m, ot, op1, op2 = traceir(tr, ref)
out:write(colorize(format("%04d", ref), band(ot, 31), band(sn, 0x100000) ~= 0))
end
out:write(band(sn, 0x10000) == 0 and " " or "|") -- SNAP_FRAME
else
out:write("---- ")
end
end
out:write("]\n")
end
-- Dump snapshots (not interleaved with IR).
local function dump_snap(tr)
out:write("---- TRACE ", tr, " snapshots\n")
for i=0,1000000000 do
local snap = tracesnap(tr, i)
if not snap then break end
out:write(format("#%-3d %04d [ ", i, snap[0]))
printsnap(tr, snap)
end
end
-- Return a register name or stack slot for a rid/sp location.
local function ridsp_name(ridsp, ins)
if not disass then disass = require("jit.dis_"..jit.arch) end
local rid, slot = band(ridsp, 0xff), shr(ridsp, 8)
if rid == 253 or rid == 254 then
return (slot == 0 or slot == 255) and " {sink" or format(" {%04d", ins-slot)
end
if ridsp > 255 then return format("[%x]", slot*4) end
if rid < 128 then return disass.regname(rid) end
return ""
end
-- Dump CALL* function ref and return optional ctype.
local function dumpcallfunc(tr, ins)
local ctype
if ins > 0 then
local m, ot, op1, op2 = traceir(tr, ins)
if band(ot, 31) == 0 then -- nil type means CARG(func, ctype).
ins = op1
ctype = formatk(tr, op2)
end
end
if ins < 0 then
out:write(format("[0x%x](", tonumber((tracek(tr, ins)))))
else
out:write(format("%04d (", ins))
end
return ctype
end
-- Recursively gather CALL* args and dump them.
local function dumpcallargs(tr, ins)
if ins < 0 then
out:write(formatk(tr, ins))
else
local m, ot, op1, op2 = traceir(tr, ins)
local oidx = 6*shr(ot, 8)
local op = sub(vmdef.irnames, oidx+1, oidx+6)
if op == "CARG " then
dumpcallargs(tr, op1)
if op2 < 0 then
out:write(" ", formatk(tr, op2))
else
out:write(" ", format("%04d", op2))
end
else
out:write(format("%04d", ins))
end
end
end
-- Dump IR and interleaved snapshots.
local function dump_ir(tr, dumpsnap, dumpreg)
local info = traceinfo(tr)
if not info then return end
local nins = info.nins
out:write("---- TRACE ", tr, " IR\n")
local irnames = vmdef.irnames
local snapref = 65536
local snap, snapno
if dumpsnap then
snap = tracesnap(tr, 0)
snapref = snap[0]
snapno = 0
end
for ins=1,nins do
if ins >= snapref then
if dumpreg then
out:write(format(".... SNAP #%-3d [ ", snapno))
else
out:write(format(".... SNAP #%-3d [ ", snapno))
end
printsnap(tr, snap)
snapno = snapno + 1
snap = tracesnap(tr, snapno)
snapref = snap and snap[0] or 65536
end
local m, ot, op1, op2, ridsp = traceir(tr, ins)
local oidx, t = 6*shr(ot, 8), band(ot, 31)
local op = sub(irnames, oidx+1, oidx+6)
if op == "LOOP " then
if dumpreg then
out:write(format("%04d ------------ LOOP ------------\n", ins))
else
out:write(format("%04d ------ LOOP ------------\n", ins))
end
elseif op ~= "NOP " and op ~= "CARG " and
(dumpreg or op ~= "RENAME") then
local rid = band(ridsp, 255)
if dumpreg then
out:write(format("%04d %-6s", ins, ridsp_name(ridsp, ins)))
else
out:write(format("%04d ", ins))
end
out:write(format("%s%s %s %s ",
(rid == 254 or rid == 253) and "}" or
(band(ot, 128) == 0 and " " or ">"),
band(ot, 64) == 0 and " " or "+",
irtype[t], op))
local m1, m2 = band(m, 3), band(m, 3*4)
if sub(op, 1, 4) == "CALL" then
local ctype
if m2 == 1*4 then -- op2 == IRMlit
out:write(format("%-10s (", vmdef.ircall[op2]))
else
ctype = dumpcallfunc(tr, op2)
end
if op1 ~= -1 then dumpcallargs(tr, op1) end
out:write(")")
if ctype then out:write(" ctype ", ctype) end
elseif op == "CNEW " and op2 == -1 then
out:write(formatk(tr, op1))
elseif m1 ~= 3 then -- op1 != IRMnone
if op1 < 0 then
out:write(formatk(tr, op1))
else
out:write(format(m1 == 0 and "%04d" or "#%-3d", op1))
end
if m2 ~= 3*4 then -- op2 != IRMnone
if m2 == 1*4 then -- op2 == IRMlit
local litn = litname[op]
if litn and litn[op2] then
out:write(" ", litn[op2])
elseif op == "UREFO " or op == "UREFC " then
out:write(format(" #%-3d", shr(op2, 8)))
else
out:write(format(" #%-3d", op2))
end
elseif op2 < 0 then
out:write(" ", formatk(tr, op2))
else
out:write(format(" %04d", op2))
end
end
end
out:write("\n")
end
end
if snap then
if dumpreg then
out:write(format(".... SNAP #%-3d [ ", snapno))
else
out:write(format(".... SNAP #%-3d [ ", snapno))
end
printsnap(tr, snap)
end
end
------------------------------------------------------------------------------
local recprefix = ""
local recdepth = 0
-- Format trace error message.
local function fmterr(err, info)
if type(err) == "number" then
if type(info) == "function" then info = fmtfunc(info) end
err = format(vmdef.traceerr[err], info)
end
return err
end
-- Dump trace states.
local function dump_trace(what, tr, func, pc, otr, oex)
if what == "stop" or (what == "abort" and dumpmode.a) then
if dumpmode.i then dump_ir(tr, dumpmode.s, dumpmode.r and what == "stop")
elseif dumpmode.s then dump_snap(tr) end
if dumpmode.m then dump_mcode(tr) end
end
if what == "start" then
if dumpmode.H then out:write('<pre class="ljdump">\n') end
out:write("---- TRACE ", tr, " ", what)
if otr then out:write(" ", otr, "/", oex == -1 and "stitch" or oex) end
out:write(" ", fmtfunc(func, pc), "\n")
elseif what == "stop" or what == "abort" then
out:write("---- TRACE ", tr, " ", what)
if what == "abort" then
out:write(" ", fmtfunc(func, pc), " -- ", fmterr(otr, oex), "\n")
else
local info = traceinfo(tr)
local link, ltype = info.link, info.linktype
if link == tr or link == 0 then
out:write(" -> ", ltype, "\n")
elseif ltype == "root" then
out:write(" -> ", link, "\n")
else
out:write(" -> ", link, " ", ltype, "\n")
end
end
if dumpmode.H then out:write("</pre>\n\n") else out:write("\n") end
else
if what == "flush" then symtab, nexitsym = {}, 0 end
out:write("---- TRACE ", what, "\n\n")
end
out:flush()
end
-- Dump recorded bytecode.
local function dump_record(tr, func, pc, depth)
if depth ~= recdepth then
recdepth = depth
recprefix = rep(" .", depth)
end
local line
if pc >= 0 then
line = bcline(func, pc, recprefix)
if dumpmode.H then line = gsub(line, "[<>&]", html_escape) end
else
line = "0000 "..recprefix.." FUNCC \n"
end
if pc <= 0 then
out:write(sub(line, 1, -2), " ; ", fmtfunc(func), "\n")
else
out:write(line)
end
if pc >= 0 and band(funcbc(func, pc), 0xff) < 16 then -- ORDER BC
out:write(bcline(func, pc+1, recprefix)) -- Write JMP for cond.
end
end
------------------------------------------------------------------------------
local gpr64 = jit.arch:match("64")
local fprmips32 = jit.arch == "mips" or jit.arch == "mipsel"
-- Dump taken trace exits.
local function dump_texit(tr, ex, ngpr, nfpr, ...)
out:write("---- TRACE ", tr, " exit ", ex, "\n")
if dumpmode.X then
local regs = {...}
if gpr64 then
for i=1,ngpr do
out:write(format(" %016x", regs[i]))
if i % 4 == 0 then out:write("\n") end
end
else
for i=1,ngpr do
out:write(" ", tohex(regs[i]))
if i % 8 == 0 then out:write("\n") end
end
end
if fprmips32 then
for i=1,nfpr,2 do
out:write(format(" %+17.14g", regs[ngpr+i]))
if i % 8 == 7 then out:write("\n") end
end
else
for i=1,nfpr do
out:write(format(" %+17.14g", regs[ngpr+i]))
if i % 4 == 0 then out:write("\n") end
end
end
end
end
------------------------------------------------------------------------------
-- Detach dump handlers.
local function dumpoff()
if active then
active = false
jit.attach(dump_texit)
jit.attach(dump_record)
jit.attach(dump_trace)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach dump handlers.
local function dumpon(opt, outfile)
if active then dumpoff() end
local term = os.getenv("TERM")
local colormode = (term and term:match("color") or os.getenv("COLORTERM")) and "A" or "T"
if opt then
opt = gsub(opt, "[TAH]", function(mode) colormode = mode; return ""; end)
end
local m = { t=true, b=true, i=true, m=true, }
if opt and opt ~= "" then
local o = sub(opt, 1, 1)
if o ~= "+" and o ~= "-" then m = {} end
for i=1,#opt do m[sub(opt, i, i)] = (o ~= "-") end
end
dumpmode = m
if m.t or m.b or m.i or m.s or m.m then
jit.attach(dump_trace, "trace")
end
if m.b then
jit.attach(dump_record, "record")
if not bcline then bcline = require("jit.bc").line end
end
if m.x or m.X then
jit.attach(dump_texit, "texit")
end
if not outfile then outfile = os.getenv("LUAJIT_DUMPFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stdout
end
m[colormode] = true
if colormode == "A" then
colorize = colorize_ansi
irtype = irtype_ansi
elseif colormode == "H" then
colorize = colorize_html
irtype = irtype_html
out:write(header_html)
else
colorize = colorize_text
irtype = irtype_text
end
active = true
end
-- Public module functions.
return {
on = dumpon,
off = dumpoff,
start = dumpon -- For -j command line option.
}
@@ -0,0 +1,312 @@
----------------------------------------------------------------------------
-- LuaJIT profiler.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module is a simple command line interface to the built-in
-- low-overhead profiler of LuaJIT.
--
-- The lower-level API of the profiler is accessible via the "jit.profile"
-- module or the luaJIT_profile_* C API.
--
-- Example usage:
--
-- luajit -jp myapp.lua
-- luajit -jp=s myapp.lua
-- luajit -jp=-s myapp.lua
-- luajit -jp=vl myapp.lua
-- luajit -jp=G,profile.txt myapp.lua
--
-- The following dump features are available:
--
-- f Stack dump: function name, otherwise module:line. Default mode.
-- F Stack dump: ditto, but always prepend module.
-- l Stack dump: module:line.
-- <number> stack dump depth (callee < caller). Default: 1.
-- -<number> Inverse stack dump depth (caller > callee).
-- s Split stack dump after first stack level. Implies abs(depth) >= 2.
-- p Show full path for module names.
-- v Show VM states. Can be combined with stack dumps, e.g. vf or fv.
-- z Show zones. Can be combined with stack dumps, e.g. zf or fz.
-- r Show raw sample counts. Default: show percentages.
-- a Annotate excerpts from source code files.
-- A Annotate complete source code files.
-- G Produce raw output suitable for graphical tools (e.g. flame graphs).
-- m<number> Minimum sample percentage to be shown. Default: 3.
-- i<number> Sampling interval in milliseconds. Default: 10.
--
----------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local profile = require("jit.profile")
local vmdef = require("jit.vmdef")
local math = math
local pairs, ipairs, tonumber, floor = pairs, ipairs, tonumber, math.floor
local sort, format = table.sort, string.format
local stdout = io.stdout
local zone -- Load jit.zone module on demand.
-- Output file handle.
local out
------------------------------------------------------------------------------
local prof_ud
local prof_states, prof_split, prof_min, prof_raw, prof_fmt, prof_depth
local prof_ann, prof_count1, prof_count2, prof_samples
local map_vmmode = {
N = "Compiled",
I = "Interpreted",
C = "C code",
G = "Garbage Collector",
J = "JIT Compiler",
}
-- Profiler callback.
local function prof_cb(th, samples, vmmode)
prof_samples = prof_samples + samples
local key_stack, key_stack2, key_state
-- Collect keys for sample.
if prof_states then
if prof_states == "v" then
key_state = map_vmmode[vmmode] or vmmode
else
key_state = zone:get() or "(none)"
end
end
if prof_fmt then
key_stack = profile.dumpstack(th, prof_fmt, prof_depth)
key_stack = key_stack:gsub("%[builtin#(%d+)%]", function(x)
return vmdef.ffnames[tonumber(x)]
end)
if prof_split == 2 then
local k1, k2 = key_stack:match("(.-) [<>] (.*)")
if k2 then key_stack, key_stack2 = k1, k2 end
elseif prof_split == 3 then
key_stack2 = profile.dumpstack(th, "l", 1)
end
end
-- Order keys.
local k1, k2
if prof_split == 1 then
if key_state then
k1 = key_state
if key_stack then k2 = key_stack end
end
elseif key_stack then
k1 = key_stack
if key_stack2 then k2 = key_stack2 elseif key_state then k2 = key_state end
end
-- Coalesce samples in one or two levels.
if k1 then
local t1 = prof_count1
t1[k1] = (t1[k1] or 0) + samples
if k2 then
local t2 = prof_count2
local t3 = t2[k1]
if not t3 then t3 = {}; t2[k1] = t3 end
t3[k2] = (t3[k2] or 0) + samples
end
end
end
------------------------------------------------------------------------------
-- Show top N list.
local function prof_top(count1, count2, samples, indent)
local t, n = {}, 0
for k in pairs(count1) do
n = n + 1
t[n] = k
end
sort(t, function(a, b) return count1[a] > count1[b] end)
for i=1,n do
local k = t[i]
local v = count1[k]
local pct = floor(v*100/samples + 0.5)
if pct < prof_min then break end
if not prof_raw then
out:write(format("%s%2d%% %s\n", indent, pct, k))
elseif prof_raw == "r" then
out:write(format("%s%5d %s\n", indent, v, k))
else
out:write(format("%s %d\n", k, v))
end
if count2 then
local r = count2[k]
if r then
prof_top(r, nil, v, (prof_split == 3 or prof_split == 1) and " -- " or
(prof_depth < 0 and " -> " or " <- "))
end
end
end
end
-- Annotate source code
local function prof_annotate(count1, samples)
local files = {}
local ms = 0
for k, v in pairs(count1) do
local pct = floor(v*100/samples + 0.5)
ms = math.max(ms, v)
if pct >= prof_min then
local file, line = k:match("^(.*):(%d+)$")
if not file then file = k; line = 0 end
local fl = files[file]
if not fl then fl = {}; files[file] = fl; files[#files+1] = file end
line = tonumber(line)
fl[line] = prof_raw and v or pct
end
end
sort(files)
local fmtv, fmtn = " %3d%% | %s\n", " | %s\n"
if prof_raw then
local n = math.max(5, math.ceil(math.log10(ms)))
fmtv = "%"..n.."d | %s\n"
fmtn = (" "):rep(n).." | %s\n"
end
local ann = prof_ann
for _, file in ipairs(files) do
local f0 = file:byte()
if f0 == 40 or f0 == 91 then
out:write(format("\n====== %s ======\n[Cannot annotate non-file]\n", file))
break
end
local fp, err = io.open(file)
if not fp then
out:write(format("====== ERROR: %s: %s\n", file, err))
break
end
out:write(format("\n====== %s ======\n", file))
local fl = files[file]
local n, show = 1, false
if ann ~= 0 then
for i=1,ann do
if fl[i] then show = true; out:write("@@ 1 @@\n"); break end
end
end
for line in fp:lines() do
if line:byte() == 27 then
out:write("[Cannot annotate bytecode file]\n")
break
end
local v = fl[n]
if ann ~= 0 then
local v2 = fl[n+ann]
if show then
if v2 then show = n+ann elseif v then show = n
elseif show+ann < n then show = false end
elseif v2 then
show = n+ann
out:write(format("@@ %d @@\n", n))
end
if not show then goto next end
end
if v then
out:write(format(fmtv, v, line))
else
out:write(format(fmtn, line))
end
::next::
n = n + 1
end
fp:close()
end
end
------------------------------------------------------------------------------
-- Finish profiling and dump result.
local function prof_finish()
if prof_ud then
profile.stop()
local samples = prof_samples
if samples == 0 then
if prof_raw ~= true then out:write("[No samples collected]\n") end
return
end
if prof_ann then
prof_annotate(prof_count1, samples)
else
prof_top(prof_count1, prof_count2, samples, "")
end
prof_count1 = nil
prof_count2 = nil
prof_ud = nil
if out ~= stdout then out:close() end
end
end
-- Start profiling.
local function prof_start(mode)
local interval = ""
mode = mode:gsub("i%d*", function(s) interval = s; return "" end)
prof_min = 3
mode = mode:gsub("m(%d+)", function(s) prof_min = tonumber(s); return "" end)
prof_depth = 1
mode = mode:gsub("%-?%d+", function(s) prof_depth = tonumber(s); return "" end)
local m = {}
for c in mode:gmatch(".") do m[c] = c end
prof_states = m.z or m.v
if prof_states == "z" then zone = require("jit.zone") end
local scope = m.l or m.f or m.F or (prof_states and "" or "f")
local flags = (m.p or "")
prof_raw = m.r
if m.s then
prof_split = 2
if prof_depth == -1 or m["-"] then prof_depth = -2
elseif prof_depth == 1 then prof_depth = 2 end
elseif mode:find("[fF].*l") then
scope = "l"
prof_split = 3
else
prof_split = (scope == "" or mode:find("[zv].*[lfF]")) and 1 or 0
end
prof_ann = m.A and 0 or (m.a and 3)
if prof_ann then
scope = "l"
prof_fmt = "pl"
prof_split = 0
prof_depth = 1
elseif m.G and scope ~= "" then
prof_fmt = flags..scope.."Z;"
prof_depth = -100
prof_raw = true
prof_min = 0
elseif scope == "" then
prof_fmt = false
else
local sc = prof_split == 3 and m.f or m.F or scope
prof_fmt = flags..sc..(prof_depth >= 0 and "Z < " or "Z > ")
end
prof_count1 = {}
prof_count2 = {}
prof_samples = 0
profile.start(scope:lower()..interval, prof_cb)
prof_ud = newproxy(true)
getmetatable(prof_ud).__gc = prof_finish
end
------------------------------------------------------------------------------
local function start(mode, outfile)
if not outfile then outfile = os.getenv("LUAJIT_PROFILEFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stdout
end
prof_start(mode or "f")
end
-- Public module functions.
return {
start = start, -- For -j command line option.
stop = prof_finish
}
@@ -0,0 +1,170 @@
----------------------------------------------------------------------------
-- Verbose mode of the LuaJIT compiler.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module shows verbose information about the progress of the
-- JIT compiler. It prints one line for each generated trace. This module
-- is useful to see which code has been compiled or where the compiler
-- punts and falls back to the interpreter.
--
-- Example usage:
--
-- luajit -jv -e "for i=1,1000 do for j=1,1000 do end end"
-- luajit -jv=myapp.out myapp.lua
--
-- Default output is to stderr. To redirect the output to a file, pass a
-- filename as an argument (use '-' for stdout) or set the environment
-- variable LUAJIT_VERBOSEFILE. The file is overwritten every time the
-- module is started.
--
-- The output from the first example should look like this:
--
-- [TRACE 1 (command line):1 loop]
-- [TRACE 2 (1/3) (command line):1 -> 1]
--
-- The first number in each line is the internal trace number. Next are
-- the file name ('(command line)') and the line number (':1') where the
-- trace has started. Side traces also show the parent trace number and
-- the exit number where they are attached to in parentheses ('(1/3)').
-- An arrow at the end shows where the trace links to ('-> 1'), unless
-- it loops to itself.
--
-- In this case the inner loop gets hot and is traced first, generating
-- a root trace. Then the last exit from the 1st trace gets hot, too,
-- and triggers generation of the 2nd trace. The side trace follows the
-- path along the outer loop and *around* the inner loop, back to its
-- start, and then links to the 1st trace. Yes, this may seem unusual,
-- if you know how traditional compilers work. Trace compilers are full
-- of surprises like this -- have fun! :-)
--
-- Aborted traces are shown like this:
--
-- [TRACE --- foo.lua:44 -- leaving loop in root trace at foo:lua:50]
--
-- Don't worry -- trace aborts are quite common, even in programs which
-- can be fully compiled. The compiler may retry several times until it
-- finds a suitable trace.
--
-- Of course this doesn't work with features that are not-yet-implemented
-- (NYI error messages). The VM simply falls back to the interpreter. This
-- may not matter at all if the particular trace is not very high up in
-- the CPU usage profile. Oh, and the interpreter is quite fast, too.
--
-- Also check out the -jdump module, which prints all the gory details.
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local funcinfo, traceinfo = jutil.funcinfo, jutil.traceinfo
local type, format = type, string.format
local stdout, stderr = io.stdout, io.stderr
-- Active flag and output file handle.
local active, out
------------------------------------------------------------------------------
local startloc, startex
local function fmtfunc(func, pc)
local fi = funcinfo(func, pc)
if fi.loc then
return fi.loc
elseif fi.ffid then
return vmdef.ffnames[fi.ffid]
elseif fi.addr then
return format("C:%x", fi.addr)
else
return "(?)"
end
end
-- Format trace error message.
local function fmterr(err, info)
if type(err) == "number" then
if type(info) == "function" then info = fmtfunc(info) end
err = format(vmdef.traceerr[err], info)
end
return err
end
-- Dump trace states.
local function dump_trace(what, tr, func, pc, otr, oex)
if what == "start" then
startloc = fmtfunc(func, pc)
startex = otr and "("..otr.."/"..(oex == -1 and "stitch" or oex)..") " or ""
else
if what == "abort" then
local loc = fmtfunc(func, pc)
if loc ~= startloc then
out:write(format("[TRACE --- %s%s -- %s at %s]\n",
startex, startloc, fmterr(otr, oex), loc))
else
out:write(format("[TRACE --- %s%s -- %s]\n",
startex, startloc, fmterr(otr, oex)))
end
elseif what == "stop" then
local info = traceinfo(tr)
local link, ltype = info.link, info.linktype
if ltype == "interpreter" then
out:write(format("[TRACE %3s %s%s -- fallback to interpreter]\n",
tr, startex, startloc))
elseif ltype == "stitch" then
out:write(format("[TRACE %3s %s%s %s %s]\n",
tr, startex, startloc, ltype, fmtfunc(func, pc)))
elseif link == tr or link == 0 then
out:write(format("[TRACE %3s %s%s %s]\n",
tr, startex, startloc, ltype))
elseif ltype == "root" then
out:write(format("[TRACE %3s %s%s -> %d]\n",
tr, startex, startloc, link))
else
out:write(format("[TRACE %3s %s%s -> %d %s]\n",
tr, startex, startloc, link, ltype))
end
else
out:write(format("[TRACE %s]\n", what))
end
out:flush()
end
end
------------------------------------------------------------------------------
-- Detach dump handlers.
local function dumpoff()
if active then
active = false
jit.attach(dump_trace)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach dump handlers.
local function dumpon(outfile)
if active then dumpoff() end
if not outfile then outfile = os.getenv("LUAJIT_VERBOSEFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stderr
end
jit.attach(dump_trace, "trace")
active = true
end
-- Public module functions.
return {
on = dumpon,
off = dumpoff,
start = dumpon -- For -j command line option.
}
@@ -0,0 +1,395 @@
-- This is a generated file. DO NOT EDIT!
return {
bcnames = "ISLT ISGE ISLE ISGT ISEQV ISNEV ISEQS ISNES ISEQN ISNEN ISEQP ISNEP ISTC ISFC IST ISF ISTYPEISNUM MOV NOT UNM LEN ADDVN SUBVN MULVN DIVVN MODVN ADDNV SUBNV MULNV DIVNV MODNV ADDVV SUBVV MULVV DIVVV MODVV POW CAT KSTR KCDATAKSHORTKNUM KPRI KNIL UGET USETV USETS USETN USETP UCLO FNEW TNEW TDUP GGET GSET TGETV TGETS TGETB TGETR TSETV TSETS TSETB TSETM TSETR CALLM CALL CALLMTCALLT ITERC ITERN VARG ISNEXTRETM RET RET0 RET1 FORI JFORI FORL IFORL JFORL ITERL IITERLJITERLLOOP ILOOP JLOOP JMP FUNCF IFUNCFJFUNCFFUNCV IFUNCVJFUNCVFUNCC FUNCCW",
irnames = "LT GE LE GT ULT UGE ULE UGT EQ NE ABC RETF NOP BASE PVAL GCSTEPHIOP LOOP USE PHI RENAMEPROF KPRI KINT KGC KPTR KKPTR KNULL KNUM KINT64KSLOT BNOT BSWAP BAND BOR BXOR BSHL BSHR BSAR BROL BROR ADD SUB MUL DIV MOD POW NEG ABS LDEXP MIN MAX FPMATHADDOV SUBOV MULOV AREF HREFK HREF NEWREFUREFO UREFC FREF TMPREFSTRREFLREF ALOAD HLOAD ULOAD FLOAD XLOAD SLOAD VLOAD ALEN ASTOREHSTOREUSTOREFSTOREXSTORESNEW XSNEW TNEW TDUP CNEW CNEWI BUFHDRBUFPUTBUFSTRTBAR OBAR XBAR CONV TOBIT TOSTR STRTO CALLN CALLA CALLL CALLS CALLXSCARG ",
irfpm = { [0]="floor", "ceil", "trunc", "sqrt", "log", "log2", "other", },
irfield = { [0]="str.len", "func.env", "func.pc", "func.ffid", "thread.env", "tab.meta", "tab.array", "tab.node", "tab.asize", "tab.hmask", "tab.nomm", "udata.meta", "udata.udtype", "udata.file", "sbuf.w", "sbuf.e", "sbuf.b", "sbuf.l", "sbuf.ref", "sbuf.r", "cdata.ctypeid", "cdata.ptr", "cdata.int", "cdata.int64", "cdata.int64_4", },
ircall = {
[0]="lj_str_cmp",
"lj_str_find",
"lj_str_new",
"lj_strscan_num",
"lj_strfmt_int",
"lj_strfmt_num",
"lj_strfmt_char",
"lj_strfmt_putint",
"lj_strfmt_putnum",
"lj_strfmt_putquoted",
"lj_strfmt_putfxint",
"lj_strfmt_putfnum_int",
"lj_strfmt_putfnum_uint",
"lj_strfmt_putfnum",
"lj_strfmt_putfstr",
"lj_strfmt_putfchar",
"lj_buf_putmem",
"lj_buf_putstr",
"lj_buf_putchar",
"lj_buf_putstr_reverse",
"lj_buf_putstr_lower",
"lj_buf_putstr_upper",
"lj_buf_putstr_rep",
"lj_buf_puttab",
"lj_bufx_set",
"lj_bufx_more",
"lj_serialize_put",
"lj_serialize_get",
"lj_serialize_encode",
"lj_serialize_decode",
"lj_buf_tostr",
"lj_tab_new_ah",
"lj_tab_new1",
"lj_tab_dup",
"lj_tab_clear",
"lj_tab_newkey",
"lj_tab_keyindex",
"lj_vm_next",
"lj_tab_len",
"lj_tab_len_hint",
"lj_gc_step_jit",
"lj_gc_barrieruv",
"lj_mem_newgco",
"lj_prng_u64d",
"lj_vm_modi",
"log10",
"exp",
"sin",
"cos",
"tan",
"asin",
"acos",
"atan",
"sinh",
"cosh",
"tanh",
"fputc",
"fwrite",
"fflush",
"lj_vm_floor",
"lj_vm_ceil",
"lj_vm_trunc",
"sqrt",
"log",
"lj_vm_log2",
"lj_vm_powi",
"pow",
"atan2",
"ldexp",
"lj_vm_tobit",
"softfp_add",
"softfp_sub",
"softfp_mul",
"softfp_div",
"softfp_cmp",
"softfp_i2d",
"softfp_d2i",
"lj_vm_sfmin",
"lj_vm_sfmax",
"lj_vm_tointg",
"softfp_ui2d",
"softfp_f2d",
"softfp_d2ui",
"softfp_d2f",
"softfp_i2f",
"softfp_ui2f",
"softfp_f2i",
"softfp_f2ui",
"fp64_l2d",
"fp64_ul2d",
"fp64_l2f",
"fp64_ul2f",
"fp64_d2l",
"fp64_d2ul",
"fp64_f2l",
"fp64_f2ul",
"lj_carith_divi64",
"lj_carith_divu64",
"lj_carith_modi64",
"lj_carith_modu64",
"lj_carith_powi64",
"lj_carith_powu64",
"lj_cdata_newv",
"lj_cdata_setfin",
"strlen",
"memcpy",
"memset",
"lj_vm_errno",
"lj_carith_mul64",
"lj_carith_shl64",
"lj_carith_shr64",
"lj_carith_sar64",
"lj_carith_rol64",
"lj_carith_ror64",
},
traceerr = {
[0]="error thrown or hook called during recording",
"trace too short",
"trace too long",
"trace too deep",
"too many snapshots",
"blacklisted",
"retry recording",
"NYI: bytecode %d",
"leaving loop in root trace",
"inner loop in root trace",
"loop unroll limit reached",
"bad argument type",
"JIT compilation disabled for function",
"call unroll limit reached",
"down-recursion, restarting",
"NYI: unsupported variant of FastFunc %s",
"NYI: return to lower frame",
"store with nil or NaN key",
"missing metamethod",
"looping index lookup",
"NYI: mixed sparse/dense table",
"symbol not in cache",
"NYI: unsupported C type conversion",
"NYI: unsupported C function type",
"guard would always fail",
"too many PHIs",
"persistent type instability",
"failed to allocate mcode memory",
"machine code too long",
"hit mcode limit (retrying)",
"too many spill slots",
"inconsistent register allocation",
"NYI: cannot assemble IR instruction %d",
"NYI: PHI shuffling too complex",
"NYI: register coalescing too complex",
},
ffnames = {
[0]="Lua",
"C",
"assert",
"type",
"next",
"pairs",
"ipairs_aux",
"ipairs",
"getmetatable",
"setmetatable",
"getfenv",
"setfenv",
"rawget",
"rawset",
"rawequal",
"unpack",
"select",
"tonumber",
"tostring",
"error",
"pcall",
"xpcall",
"loadfile",
"load",
"loadstring",
"dofile",
"gcinfo",
"collectgarbage",
"newproxy",
"print",
"coroutine.status",
"coroutine.running",
"coroutine.isyieldable",
"coroutine.create",
"coroutine.yield",
"coroutine.resume",
"coroutine.wrap_aux",
"coroutine.wrap",
"math.abs",
"math.floor",
"math.ceil",
"math.sqrt",
"math.log10",
"math.exp",
"math.sin",
"math.cos",
"math.tan",
"math.asin",
"math.acos",
"math.atan",
"math.sinh",
"math.cosh",
"math.tanh",
"math.frexp",
"math.modf",
"math.log",
"math.atan2",
"math.pow",
"math.fmod",
"math.ldexp",
"math.min",
"math.max",
"math.random",
"math.randomseed",
"bit.tobit",
"bit.bnot",
"bit.bswap",
"bit.lshift",
"bit.rshift",
"bit.arshift",
"bit.rol",
"bit.ror",
"bit.band",
"bit.bor",
"bit.bxor",
"bit.tohex",
"string.byte",
"string.char",
"string.sub",
"string.rep",
"string.reverse",
"string.lower",
"string.upper",
"string.dump",
"string.find",
"string.match",
"string.gmatch_aux",
"string.gmatch",
"string.gsub",
"string.format",
"table.maxn",
"table.insert",
"table.concat",
"table.sort",
"table.new",
"table.clear",
"io.method.close",
"io.method.read",
"io.method.write",
"io.method.flush",
"io.method.seek",
"io.method.setvbuf",
"io.method.lines",
"io.method.__gc",
"io.method.__tostring",
"io.open",
"io.popen",
"io.tmpfile",
"io.close",
"io.read",
"io.write",
"io.flush",
"io.input",
"io.output",
"io.lines",
"io.type",
"os.execute",
"os.remove",
"os.rename",
"os.tmpname",
"os.getenv",
"os.exit",
"os.clock",
"os.date",
"os.time",
"os.difftime",
"os.setlocale",
"debug.getregistry",
"debug.getmetatable",
"debug.setmetatable",
"debug.getfenv",
"debug.setfenv",
"debug.getinfo",
"debug.getlocal",
"debug.setlocal",
"debug.getupvalue",
"debug.setupvalue",
"debug.upvalueid",
"debug.upvaluejoin",
"debug.sethook",
"debug.gethook",
"debug.debug",
"debug.traceback",
"jit.on",
"jit.off",
"jit.flush",
"jit.status",
"jit.security",
"jit.attach",
"jit.util.funcinfo",
"jit.util.funcbc",
"jit.util.funck",
"jit.util.funcuvname",
"jit.util.traceinfo",
"jit.util.traceir",
"jit.util.tracek",
"jit.util.tracesnap",
"jit.util.tracemc",
"jit.util.traceexitstub",
"jit.util.ircalladdr",
"jit.opt.start",
"jit.profile.start",
"jit.profile.stop",
"jit.profile.dumpstack",
"ffi.meta.__index",
"ffi.meta.__newindex",
"ffi.meta.__eq",
"ffi.meta.__len",
"ffi.meta.__lt",
"ffi.meta.__le",
"ffi.meta.__concat",
"ffi.meta.__call",
"ffi.meta.__add",
"ffi.meta.__sub",
"ffi.meta.__mul",
"ffi.meta.__div",
"ffi.meta.__mod",
"ffi.meta.__pow",
"ffi.meta.__unm",
"ffi.meta.__tostring",
"ffi.meta.__pairs",
"ffi.meta.__ipairs",
"ffi.clib.__index",
"ffi.clib.__newindex",
"ffi.clib.__gc",
"ffi.callback.free",
"ffi.callback.set",
"ffi.cdef",
"ffi.new",
"ffi.cast",
"ffi.typeof",
"ffi.typeinfo",
"ffi.istype",
"ffi.sizeof",
"ffi.alignof",
"ffi.offsetof",
"ffi.errno",
"ffi.string",
"ffi.copy",
"ffi.fill",
"ffi.abi",
"ffi.metatype",
"ffi.gc",
"ffi.load",
"buffer.method.free",
"buffer.method.reset",
"buffer.method.skip",
"buffer.method.set",
"buffer.method.put",
"buffer.method.putf",
"buffer.method.get",
"buffer.method.putcdata",
"buffer.method.reserve",
"buffer.method.commit",
"buffer.method.ref",
"buffer.method.encode",
"buffer.method.decode",
"buffer.method.__gc",
"buffer.method.__tostring",
"buffer.method.__len",
"buffer.new",
"buffer.encode",
"buffer.decode",
},
}
@@ -0,0 +1,45 @@
----------------------------------------------------------------------------
-- LuaJIT profiler zones.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module implements a simple hierarchical zone model.
--
-- Example usage:
--
-- local zone = require("jit.zone")
-- zone("AI")
-- ...
-- zone("A*")
-- ...
-- print(zone:get()) --> "A*"
-- ...
-- zone()
-- ...
-- print(zone:get()) --> "AI"
-- ...
-- zone()
--
----------------------------------------------------------------------------
local remove = table.remove
return setmetatable({
flush = function(t)
for i=#t,1,-1 do t[i] = nil end
end,
get = function(t)
return t[#t]
end
}, {
__call = function(t, zone)
if zone then
t[#t+1] = zone
else
return (assert(remove(t), "empty zone stack"))
end
end
})
Binary file not shown.
@@ -0,0 +1,190 @@
----------------------------------------------------------------------------
-- LuaJIT bytecode listing module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module lists the bytecode of a Lua function. If it's loaded by -jbc
-- it hooks into the parser and lists all functions of a chunk as they
-- are parsed.
--
-- Example usage:
--
-- luajit -jbc -e 'local x=0; for i=1,1e6 do x=x+i end; print(x)'
-- luajit -jbc=- foo.lua
-- luajit -jbc=foo.list foo.lua
--
-- Default output is to stderr. To redirect the output to a file, pass a
-- filename as an argument (use '-' for stdout) or set the environment
-- variable LUAJIT_LISTFILE. The file is overwritten every time the module
-- is started.
--
-- This module can also be used programmatically:
--
-- local bc = require("jit.bc")
--
-- local function foo() print("hello") end
--
-- bc.dump(foo) --> -- BYTECODE -- [...]
-- print(bc.line(foo, 2)) --> 0002 KSTR 1 1 ; "hello"
--
-- local out = {
-- -- Do something with each line:
-- write = function(t, ...) io.write(...) end,
-- close = function(t) end,
-- flush = function(t) end,
-- }
-- bc.dump(foo, out)
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local bit = require("bit")
local sub, gsub, format = string.sub, string.gsub, string.format
local byte, band, shr = string.byte, bit.band, bit.rshift
local funcinfo, funcbc, funck = jutil.funcinfo, jutil.funcbc, jutil.funck
local funcuvname = jutil.funcuvname
local bcnames = vmdef.bcnames
local stdout, stderr = io.stdout, io.stderr
------------------------------------------------------------------------------
local function ctlsub(c)
if c == "\n" then return "\\n"
elseif c == "\r" then return "\\r"
elseif c == "\t" then return "\\t"
else return format("\\%03d", byte(c))
end
end
-- Return one bytecode line.
local function bcline(func, pc, prefix)
local ins, m = funcbc(func, pc)
if not ins then return end
local ma, mb, mc = band(m, 7), band(m, 15*8), band(m, 15*128)
local a = band(shr(ins, 8), 0xff)
local oidx = 6*band(ins, 0xff)
local op = sub(bcnames, oidx+1, oidx+6)
local s = format("%04d %s %-6s %3s ",
pc, prefix or " ", op, ma == 0 and "" or a)
local d = shr(ins, 16)
if mc == 13*128 then -- BCMjump
return format("%s=> %04d\n", s, pc+d-0x7fff)
end
if mb ~= 0 then
d = band(d, 0xff)
elseif mc == 0 then
return s.."\n"
end
local kc
if mc == 10*128 then -- BCMstr
kc = funck(func, -d-1)
kc = format(#kc > 40 and '"%.40s"~' or '"%s"', gsub(kc, "%c", ctlsub))
elseif mc == 9*128 then -- BCMnum
kc = funck(func, d)
if op == "TSETM " then kc = kc - 2^52 end
elseif mc == 12*128 then -- BCMfunc
local fi = funcinfo(funck(func, -d-1))
if fi.ffid then
kc = vmdef.ffnames[fi.ffid]
else
kc = fi.loc
end
elseif mc == 5*128 then -- BCMuv
kc = funcuvname(func, d)
end
if ma == 5 then -- BCMuv
local ka = funcuvname(func, a)
if kc then kc = ka.." ; "..kc else kc = ka end
end
if mb ~= 0 then
local b = shr(ins, 24)
if kc then return format("%s%3d %3d ; %s\n", s, b, d, kc) end
return format("%s%3d %3d\n", s, b, d)
end
if kc then return format("%s%3d ; %s\n", s, d, kc) end
if mc == 7*128 and d > 32767 then d = d - 65536 end -- BCMlits
return format("%s%3d\n", s, d)
end
-- Collect branch targets of a function.
local function bctargets(func)
local target = {}
for pc=1,1000000000 do
local ins, m = funcbc(func, pc)
if not ins then break end
if band(m, 15*128) == 13*128 then target[pc+shr(ins, 16)-0x7fff] = true end
end
return target
end
-- Dump bytecode instructions of a function.
local function bcdump(func, out, all)
if not out then out = stdout end
local fi = funcinfo(func)
if all and fi.children then
for n=-1,-1000000000,-1 do
local k = funck(func, n)
if not k then break end
if type(k) == "proto" then bcdump(k, out, true) end
end
end
out:write(format("-- BYTECODE -- %s-%d\n", fi.loc, fi.lastlinedefined))
local target = bctargets(func)
for pc=1,1000000000 do
local s = bcline(func, pc, target[pc] and "=>")
if not s then break end
out:write(s)
end
out:write("\n")
out:flush()
end
------------------------------------------------------------------------------
-- Active flag and output file handle.
local active, out
-- List handler.
local function h_list(func)
return bcdump(func, out)
end
-- Detach list handler.
local function bclistoff()
if active then
active = false
jit.attach(h_list)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach list handler.
local function bcliston(outfile)
if active then bclistoff() end
if not outfile then outfile = os.getenv("LUAJIT_LISTFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stderr
end
jit.attach(h_list, "bc")
active = true
end
-- Public module functions.
return {
line = bcline,
dump = bcdump,
targets = bctargets,
on = bcliston,
off = bclistoff,
start = bcliston -- For -j command line option.
}
@@ -0,0 +1,665 @@
----------------------------------------------------------------------------
-- LuaJIT module to save/list bytecode.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module saves or lists the bytecode for an input file.
-- It's run by the -b command line option.
--
------------------------------------------------------------------------------
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local bit = require("bit")
-- Symbol name prefix for LuaJIT bytecode.
local LJBC_PREFIX = "luaJIT_BC_"
local type, assert = type, assert
local format = string.format
local tremove, tconcat = table.remove, table.concat
------------------------------------------------------------------------------
local function usage()
io.stderr:write[[
Save LuaJIT bytecode: luajit -b[options] input output
-l Only list bytecode.
-s Strip debug info (default).
-g Keep debug info.
-n name Set module name (default: auto-detect from input name).
-t type Set output file type (default: auto-detect from output name).
-a arch Override architecture for object files (default: native).
-o os Override OS for object files (default: native).
-e chunk Use chunk string as input.
-- Stop handling options.
- Use stdin as input and/or stdout as output.
File types: c h obj o raw (default)
]]
os.exit(1)
end
local function check(ok, ...)
if ok then return ok, ... end
io.stderr:write("luajit: ", ...)
io.stderr:write("\n")
os.exit(1)
end
local function readfile(input)
if type(input) == "function" then return input end
if input == "-" then input = nil end
return check(loadfile(input))
end
local function savefile(name, mode)
if name == "-" then return io.stdout end
return check(io.open(name, mode))
end
------------------------------------------------------------------------------
local map_type = {
raw = "raw", c = "c", h = "h", o = "obj", obj = "obj",
}
local map_arch = {
x86 = { e = "le", b = 32, m = 3, p = 0x14c, },
x64 = { e = "le", b = 64, m = 62, p = 0x8664, },
arm = { e = "le", b = 32, m = 40, p = 0x1c0, },
arm64 = { e = "le", b = 64, m = 183, p = 0xaa64, },
arm64be = { e = "be", b = 64, m = 183, },
ppc = { e = "be", b = 32, m = 20, },
mips = { e = "be", b = 32, m = 8, f = 0x50001006, },
mipsel = { e = "le", b = 32, m = 8, f = 0x50001006, },
mips64 = { e = "be", b = 64, m = 8, f = 0x80000007, },
mips64el = { e = "le", b = 64, m = 8, f = 0x80000007, },
mips64r6 = { e = "be", b = 64, m = 8, f = 0xa0000407, },
mips64r6el = { e = "le", b = 64, m = 8, f = 0xa0000407, },
}
local map_os = {
linux = true, windows = true, osx = true, freebsd = true, netbsd = true,
openbsd = true, dragonfly = true, solaris = true,
}
local function checkarg(str, map, err)
str = str:lower()
local s = check(map[str], "unknown ", err)
return type(s) == "string" and s or str
end
local function detecttype(str)
local ext = str:lower():match("%.(%a+)$")
return map_type[ext] or "raw"
end
local function checkmodname(str)
check(str:match("^[%w_.%-]+$"), "bad module name")
return str:gsub("[%.%-]", "_")
end
local function detectmodname(str)
if type(str) == "string" then
local tail = str:match("[^/\\]+$")
if tail then str = tail end
local head = str:match("^(.*)%.[^.]*$")
if head then str = head end
str = str:match("^[%w_.%-]+")
else
str = nil
end
check(str, "cannot derive module name, use -n name")
return str:gsub("[%.%-]", "_")
end
------------------------------------------------------------------------------
local function bcsave_tail(fp, output, s)
local ok, err = fp:write(s)
if ok and output ~= "-" then ok, err = fp:close() end
check(ok, "cannot write ", output, ": ", err)
end
local function bcsave_raw(output, s)
local fp = savefile(output, "wb")
bcsave_tail(fp, output, s)
end
local function bcsave_c(ctx, output, s)
local fp = savefile(output, "w")
if ctx.type == "c" then
fp:write(format([[
#ifdef __cplusplus
extern "C"
#endif
#ifdef _WIN32
__declspec(dllexport)
#endif
const unsigned char %s%s[] = {
]], LJBC_PREFIX, ctx.modname))
else
fp:write(format([[
#define %s%s_SIZE %d
static const unsigned char %s%s[] = {
]], LJBC_PREFIX, ctx.modname, #s, LJBC_PREFIX, ctx.modname))
end
local t, n, m = {}, 0, 0
for i=1,#s do
local b = tostring(string.byte(s, i))
m = m + #b + 1
if m > 78 then
fp:write(tconcat(t, ",", 1, n), ",\n")
n, m = 0, #b + 1
end
n = n + 1
t[n] = b
end
bcsave_tail(fp, output, tconcat(t, ",", 1, n).."\n};\n")
end
local function bcsave_elfobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct {
uint8_t emagic[4], eclass, eendian, eversion, eosabi, eabiversion, epad[7];
uint16_t type, machine;
uint32_t version;
uint32_t entry, phofs, shofs;
uint32_t flags;
uint16_t ehsize, phentsize, phnum, shentsize, shnum, shstridx;
} ELF32header;
typedef struct {
uint8_t emagic[4], eclass, eendian, eversion, eosabi, eabiversion, epad[7];
uint16_t type, machine;
uint32_t version;
uint64_t entry, phofs, shofs;
uint32_t flags;
uint16_t ehsize, phentsize, phnum, shentsize, shnum, shstridx;
} ELF64header;
typedef struct {
uint32_t name, type, flags, addr, ofs, size, link, info, align, entsize;
} ELF32sectheader;
typedef struct {
uint32_t name, type;
uint64_t flags, addr, ofs, size;
uint32_t link, info;
uint64_t align, entsize;
} ELF64sectheader;
typedef struct {
uint32_t name, value, size;
uint8_t info, other;
uint16_t sectidx;
} ELF32symbol;
typedef struct {
uint32_t name;
uint8_t info, other;
uint16_t sectidx;
uint64_t value, size;
} ELF64symbol;
typedef struct {
ELF32header hdr;
ELF32sectheader sect[6];
ELF32symbol sym[2];
uint8_t space[4096];
} ELF32obj;
typedef struct {
ELF64header hdr;
ELF64sectheader sect[6];
ELF64symbol sym[2];
uint8_t space[4096];
} ELF64obj;
]]
local symname = LJBC_PREFIX..ctx.modname
local ai = assert(map_arch[ctx.arch])
local is64, isbe = ai.b == 64, ai.e == "be"
-- Handle different host/target endianess.
local function f32(x) return x end
local f16, fofs = f32, f32
if ffi.abi("be") ~= isbe then
f32 = bit.bswap
function f16(x) return bit.rshift(bit.bswap(x), 16) end
if is64 then
local two32 = ffi.cast("int64_t", 2^32)
function fofs(x) return bit.bswap(x)*two32 end
else
fofs = f32
end
end
-- Create ELF object and fill in header.
local o = ffi.new(is64 and "ELF64obj" or "ELF32obj")
local hdr = o.hdr
if ctx.os == "bsd" or ctx.os == "other" then -- Determine native hdr.eosabi.
local bf = assert(io.open("/bin/ls", "rb"))
local bs = bf:read(9)
bf:close()
ffi.copy(o, bs, 9)
check(hdr.emagic[0] == 127, "no support for writing native object files")
else
hdr.emagic = "\127ELF"
hdr.eosabi = ({ freebsd=9, netbsd=2, openbsd=12, solaris=6 })[ctx.os] or 0
end
hdr.eclass = is64 and 2 or 1
hdr.eendian = isbe and 2 or 1
hdr.eversion = 1
hdr.type = f16(1)
hdr.machine = f16(ai.m)
hdr.flags = f32(ai.f or 0)
hdr.version = f32(1)
hdr.shofs = fofs(ffi.offsetof(o, "sect"))
hdr.ehsize = f16(ffi.sizeof(hdr))
hdr.shentsize = f16(ffi.sizeof(o.sect[0]))
hdr.shnum = f16(6)
hdr.shstridx = f16(2)
-- Fill in sections and symbols.
local sofs, ofs = ffi.offsetof(o, "space"), 1
for i,name in ipairs{
".symtab", ".shstrtab", ".strtab", ".rodata", ".note.GNU-stack",
} do
local sect = o.sect[i]
sect.align = fofs(1)
sect.name = f32(ofs)
ffi.copy(o.space+ofs, name)
ofs = ofs + #name+1
end
o.sect[1].type = f32(2) -- .symtab
o.sect[1].link = f32(3)
o.sect[1].info = f32(1)
o.sect[1].align = fofs(8)
o.sect[1].ofs = fofs(ffi.offsetof(o, "sym"))
o.sect[1].entsize = fofs(ffi.sizeof(o.sym[0]))
o.sect[1].size = fofs(ffi.sizeof(o.sym))
o.sym[1].name = f32(1)
o.sym[1].sectidx = f16(4)
o.sym[1].size = fofs(#s)
o.sym[1].info = 17
o.sect[2].type = f32(3) -- .shstrtab
o.sect[2].ofs = fofs(sofs)
o.sect[2].size = fofs(ofs)
o.sect[3].type = f32(3) -- .strtab
o.sect[3].ofs = fofs(sofs + ofs)
o.sect[3].size = fofs(#symname+2)
ffi.copy(o.space+ofs+1, symname)
ofs = ofs + #symname + 2
o.sect[4].type = f32(1) -- .rodata
o.sect[4].flags = fofs(2)
o.sect[4].ofs = fofs(sofs + ofs)
o.sect[4].size = fofs(#s)
o.sect[5].type = f32(1) -- .note.GNU-stack
o.sect[5].ofs = fofs(sofs + ofs + #s)
-- Write ELF object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, ffi.sizeof(o)-4096+ofs))
bcsave_tail(fp, output, s)
end
local function bcsave_peobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct {
uint16_t arch, nsects;
uint32_t time, symtabofs, nsyms;
uint16_t opthdrsz, flags;
} PEheader;
typedef struct {
char name[8];
uint32_t vsize, vaddr, size, ofs, relocofs, lineofs;
uint16_t nreloc, nline;
uint32_t flags;
} PEsection;
typedef struct __attribute((packed)) {
union {
char name[8];
uint32_t nameref[2];
};
uint32_t value;
int16_t sect;
uint16_t type;
uint8_t scl, naux;
} PEsym;
typedef struct __attribute((packed)) {
uint32_t size;
uint16_t nreloc, nline;
uint32_t cksum;
uint16_t assoc;
uint8_t comdatsel, unused[3];
} PEsymaux;
typedef struct {
PEheader hdr;
PEsection sect[2];
// Must be an even number of symbol structs.
PEsym sym0;
PEsymaux sym0aux;
PEsym sym1;
PEsymaux sym1aux;
PEsym sym2;
PEsym sym3;
uint32_t strtabsize;
uint8_t space[4096];
} PEobj;
]]
local symname = LJBC_PREFIX..ctx.modname
local ai = assert(map_arch[ctx.arch])
local is64 = ai.b == 64
local symexport = " /EXPORT:"..symname..",DATA "
-- The file format is always little-endian. Swap if the host is big-endian.
local function f32(x) return x end
local f16 = f32
if ffi.abi("be") then
f32 = bit.bswap
function f16(x) return bit.rshift(bit.bswap(x), 16) end
end
-- Create PE object and fill in header.
local o = ffi.new("PEobj")
local hdr = o.hdr
hdr.arch = f16(assert(ai.p))
hdr.nsects = f16(2)
hdr.symtabofs = f32(ffi.offsetof(o, "sym0"))
hdr.nsyms = f32(6)
-- Fill in sections and symbols.
o.sect[0].name = ".drectve"
o.sect[0].size = f32(#symexport)
o.sect[0].flags = f32(0x00100a00)
o.sym0.sect = f16(1)
o.sym0.scl = 3
o.sym0.name = ".drectve"
o.sym0.naux = 1
o.sym0aux.size = f32(#symexport)
o.sect[1].name = ".rdata"
o.sect[1].size = f32(#s)
o.sect[1].flags = f32(0x40300040)
o.sym1.sect = f16(2)
o.sym1.scl = 3
o.sym1.name = ".rdata"
o.sym1.naux = 1
o.sym1aux.size = f32(#s)
o.sym2.sect = f16(2)
o.sym2.scl = 2
o.sym2.nameref[1] = f32(4)
o.sym3.sect = f16(-1)
o.sym3.scl = 2
o.sym3.value = f32(1)
o.sym3.name = "@feat.00" -- Mark as SafeSEH compliant.
ffi.copy(o.space, symname)
local ofs = #symname + 1
o.strtabsize = f32(ofs + 4)
o.sect[0].ofs = f32(ffi.offsetof(o, "space") + ofs)
ffi.copy(o.space + ofs, symexport)
ofs = ofs + #symexport
o.sect[1].ofs = f32(ffi.offsetof(o, "space") + ofs)
-- Write PE object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, ffi.sizeof(o)-4096+ofs))
bcsave_tail(fp, output, s)
end
local function bcsave_machobj(ctx, output, s, ffi)
ffi.cdef[[
typedef struct
{
uint32_t magic, cputype, cpusubtype, filetype, ncmds, sizeofcmds, flags;
} mach_header;
typedef struct
{
mach_header; uint32_t reserved;
} mach_header_64;
typedef struct {
uint32_t cmd, cmdsize;
char segname[16];
uint32_t vmaddr, vmsize, fileoff, filesize;
uint32_t maxprot, initprot, nsects, flags;
} mach_segment_command;
typedef struct {
uint32_t cmd, cmdsize;
char segname[16];
uint64_t vmaddr, vmsize, fileoff, filesize;
uint32_t maxprot, initprot, nsects, flags;
} mach_segment_command_64;
typedef struct {
char sectname[16], segname[16];
uint32_t addr, size;
uint32_t offset, align, reloff, nreloc, flags;
uint32_t reserved1, reserved2;
} mach_section;
typedef struct {
char sectname[16], segname[16];
uint64_t addr, size;
uint32_t offset, align, reloff, nreloc, flags;
uint32_t reserved1, reserved2, reserved3;
} mach_section_64;
typedef struct {
uint32_t cmd, cmdsize, symoff, nsyms, stroff, strsize;
} mach_symtab_command;
typedef struct {
int32_t strx;
uint8_t type, sect;
int16_t desc;
uint32_t value;
} mach_nlist;
typedef struct {
uint32_t strx;
uint8_t type, sect;
uint16_t desc;
uint64_t value;
} mach_nlist_64;
typedef struct
{
uint32_t magic, nfat_arch;
} mach_fat_header;
typedef struct
{
uint32_t cputype, cpusubtype, offset, size, align;
} mach_fat_arch;
typedef struct {
struct {
mach_header hdr;
mach_segment_command seg;
mach_section sec;
mach_symtab_command sym;
} arch[1];
mach_nlist sym_entry;
uint8_t space[4096];
} mach_obj;
typedef struct {
struct {
mach_header_64 hdr;
mach_segment_command_64 seg;
mach_section_64 sec;
mach_symtab_command sym;
} arch[1];
mach_nlist_64 sym_entry;
uint8_t space[4096];
} mach_obj_64;
typedef struct {
mach_fat_header fat;
mach_fat_arch fat_arch[2];
struct {
mach_header hdr;
mach_segment_command seg;
mach_section sec;
mach_symtab_command sym;
} arch[2];
mach_nlist sym_entry;
uint8_t space[4096];
} mach_fat_obj;
]]
local symname = '_'..LJBC_PREFIX..ctx.modname
local isfat, is64, align, mobj = false, false, 4, "mach_obj"
if ctx.arch == "x64" then
is64, align, mobj = true, 8, "mach_obj_64"
elseif ctx.arch == "arm" then
isfat, mobj = true, "mach_fat_obj"
elseif ctx.arch == "arm64" then
is64, align, isfat, mobj = true, 8, true, "mach_fat_obj"
else
check(ctx.arch == "x86", "unsupported architecture for OSX")
end
local function aligned(v, a) return bit.band(v+a-1, -a) end
local be32 = bit.bswap -- Mach-O FAT is BE, supported archs are LE.
-- Create Mach-O object and fill in header.
local o = ffi.new(mobj)
local mach_size = aligned(ffi.offsetof(o, "space")+#symname+2, align)
local cputype = ({ x86={7}, x64={0x01000007}, arm={7,12}, arm64={0x01000007,0x0100000c} })[ctx.arch]
local cpusubtype = ({ x86={3}, x64={3}, arm={3,9}, arm64={3,0} })[ctx.arch]
if isfat then
o.fat.magic = be32(0xcafebabe)
o.fat.nfat_arch = be32(#cpusubtype)
end
-- Fill in sections and symbols.
for i=0,#cpusubtype-1 do
local ofs = 0
if isfat then
local a = o.fat_arch[i]
a.cputype = be32(cputype[i+1])
a.cpusubtype = be32(cpusubtype[i+1])
-- Subsequent slices overlap each other to share data.
ofs = ffi.offsetof(o, "arch") + i*ffi.sizeof(o.arch[0])
a.offset = be32(ofs)
a.size = be32(mach_size-ofs+#s)
end
local a = o.arch[i]
a.hdr.magic = is64 and 0xfeedfacf or 0xfeedface
a.hdr.cputype = cputype[i+1]
a.hdr.cpusubtype = cpusubtype[i+1]
a.hdr.filetype = 1
a.hdr.ncmds = 2
a.hdr.sizeofcmds = ffi.sizeof(a.seg)+ffi.sizeof(a.sec)+ffi.sizeof(a.sym)
a.seg.cmd = is64 and 0x19 or 0x1
a.seg.cmdsize = ffi.sizeof(a.seg)+ffi.sizeof(a.sec)
a.seg.vmsize = #s
a.seg.fileoff = mach_size-ofs
a.seg.filesize = #s
a.seg.maxprot = 1
a.seg.initprot = 1
a.seg.nsects = 1
ffi.copy(a.sec.sectname, "__data")
ffi.copy(a.sec.segname, "__DATA")
a.sec.size = #s
a.sec.offset = mach_size-ofs
a.sym.cmd = 2
a.sym.cmdsize = ffi.sizeof(a.sym)
a.sym.symoff = ffi.offsetof(o, "sym_entry")-ofs
a.sym.nsyms = 1
a.sym.stroff = ffi.offsetof(o, "sym_entry")+ffi.sizeof(o.sym_entry)-ofs
a.sym.strsize = aligned(#symname+2, align)
end
o.sym_entry.type = 0xf
o.sym_entry.sect = 1
o.sym_entry.strx = 1
ffi.copy(o.space+1, symname)
-- Write Macho-O object file.
local fp = savefile(output, "wb")
fp:write(ffi.string(o, mach_size))
bcsave_tail(fp, output, s)
end
local function bcsave_obj(ctx, output, s)
local ok, ffi = pcall(require, "ffi")
check(ok, "FFI library required to write this file type")
if ctx.os == "windows" then
return bcsave_peobj(ctx, output, s, ffi)
elseif ctx.os == "osx" then
return bcsave_machobj(ctx, output, s, ffi)
else
return bcsave_elfobj(ctx, output, s, ffi)
end
end
------------------------------------------------------------------------------
local function bclist(input, output)
local f = readfile(input)
require("jit.bc").dump(f, savefile(output, "w"), true)
end
local function bcsave(ctx, input, output)
local f = readfile(input)
local s = string.dump(f, ctx.strip)
local t = ctx.type
if not t then
t = detecttype(output)
ctx.type = t
end
if t == "raw" then
bcsave_raw(output, s)
else
if not ctx.modname then ctx.modname = detectmodname(input) end
if t == "obj" then
bcsave_obj(ctx, output, s)
else
bcsave_c(ctx, output, s)
end
end
end
local function docmd(...)
local arg = {...}
local n = 1
local list = false
local ctx = {
strip = true, arch = jit.arch, os = jit.os:lower(),
type = false, modname = false,
}
while n <= #arg do
local a = arg[n]
if type(a) == "string" and a:sub(1, 1) == "-" and a ~= "-" then
tremove(arg, n)
if a == "--" then break end
for m=2,#a do
local opt = a:sub(m, m)
if opt == "l" then
list = true
elseif opt == "s" then
ctx.strip = true
elseif opt == "g" then
ctx.strip = false
else
if arg[n] == nil or m ~= #a then usage() end
if opt == "e" then
if n ~= 1 then usage() end
arg[1] = check(loadstring(arg[1]))
elseif opt == "n" then
ctx.modname = checkmodname(tremove(arg, n))
elseif opt == "t" then
ctx.type = checkarg(tremove(arg, n), map_type, "file type")
elseif opt == "a" then
ctx.arch = checkarg(tremove(arg, n), map_arch, "architecture")
elseif opt == "o" then
ctx.os = checkarg(tremove(arg, n), map_os, "OS name")
else
usage()
end
end
end
else
n = n + 1
end
end
if list then
if #arg == 0 or #arg > 2 then usage() end
bclist(arg[1], arg[2] or "-")
else
if #arg ~= 2 then usage() end
bcsave(ctx, arg[1], arg[2])
end
end
------------------------------------------------------------------------------
-- Public module functions.
return {
start = docmd -- Process -b command line option.
}
@@ -0,0 +1,689 @@
----------------------------------------------------------------------------
-- LuaJIT ARM disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles most user-mode ARMv7 instructions
-- NYI: Advanced SIMD and VFP instructions.
------------------------------------------------------------------------------
local type = type
local sub, byte, format = string.sub, string.byte, string.format
local match, gmatch = string.match, string.gmatch
local concat = table.concat
local bit = require("bit")
local band, bor, ror, tohex = bit.band, bit.bor, bit.ror, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Opcode maps
------------------------------------------------------------------------------
local map_loadc = {
shift = 8, mask = 15,
[10] = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "vmovFmDN", "vstmFNdr",
_ = {
shift = 21, mask = 1,
[0] = "vstrFdl",
{ shift = 16, mask = 15, [13] = "vpushFdr", _ = "vstmdbFNdr", }
},
},
{
shift = 23, mask = 3,
[0] = "vmovFDNm",
{ shift = 16, mask = 15, [13] = "vpopFdr", _ = "vldmFNdr", },
_ = {
shift = 21, mask = 1,
[0] = "vldrFdl", "vldmdbFNdr",
},
},
},
[11] = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "vmovGmDN", "vstmGNdr",
_ = {
shift = 21, mask = 1,
[0] = "vstrGdl",
{ shift = 16, mask = 15, [13] = "vpushGdr", _ = "vstmdbGNdr", }
},
},
{
shift = 23, mask = 3,
[0] = "vmovGDNm",
{ shift = 16, mask = 15, [13] = "vpopGdr", _ = "vldmGNdr", },
_ = {
shift = 21, mask = 1,
[0] = "vldrGdl", "vldmdbGNdr",
},
},
},
_ = {
shift = 0, mask = 0 -- NYI ldc, mcrr, mrrc.
},
}
local map_vfps = {
shift = 6, mask = 0x2c001,
[0] = "vmlaF.dnm", "vmlsF.dnm",
[0x04000] = "vnmlsF.dnm", [0x04001] = "vnmlaF.dnm",
[0x08000] = "vmulF.dnm", [0x08001] = "vnmulF.dnm",
[0x0c000] = "vaddF.dnm", [0x0c001] = "vsubF.dnm",
[0x20000] = "vdivF.dnm",
[0x24000] = "vfnmsF.dnm", [0x24001] = "vfnmaF.dnm",
[0x28000] = "vfmaF.dnm", [0x28001] = "vfmsF.dnm",
[0x2c000] = "vmovF.dY",
[0x2c001] = {
shift = 7, mask = 0x1e01,
[0] = "vmovF.dm", "vabsF.dm",
[0x0200] = "vnegF.dm", [0x0201] = "vsqrtF.dm",
[0x0800] = "vcmpF.dm", [0x0801] = "vcmpeF.dm",
[0x0a00] = "vcmpzF.d", [0x0a01] = "vcmpzeF.d",
[0x0e01] = "vcvtG.dF.m",
[0x1000] = "vcvt.f32.u32Fdm", [0x1001] = "vcvt.f32.s32Fdm",
[0x1800] = "vcvtr.u32F.dm", [0x1801] = "vcvt.u32F.dm",
[0x1a00] = "vcvtr.s32F.dm", [0x1a01] = "vcvt.s32F.dm",
},
}
local map_vfpd = {
shift = 6, mask = 0x2c001,
[0] = "vmlaG.dnm", "vmlsG.dnm",
[0x04000] = "vnmlsG.dnm", [0x04001] = "vnmlaG.dnm",
[0x08000] = "vmulG.dnm", [0x08001] = "vnmulG.dnm",
[0x0c000] = "vaddG.dnm", [0x0c001] = "vsubG.dnm",
[0x20000] = "vdivG.dnm",
[0x24000] = "vfnmsG.dnm", [0x24001] = "vfnmaG.dnm",
[0x28000] = "vfmaG.dnm", [0x28001] = "vfmsG.dnm",
[0x2c000] = "vmovG.dY",
[0x2c001] = {
shift = 7, mask = 0x1e01,
[0] = "vmovG.dm", "vabsG.dm",
[0x0200] = "vnegG.dm", [0x0201] = "vsqrtG.dm",
[0x0800] = "vcmpG.dm", [0x0801] = "vcmpeG.dm",
[0x0a00] = "vcmpzG.d", [0x0a01] = "vcmpzeG.d",
[0x0e01] = "vcvtF.dG.m",
[0x1000] = "vcvt.f64.u32GdFm", [0x1001] = "vcvt.f64.s32GdFm",
[0x1800] = "vcvtr.u32FdG.m", [0x1801] = "vcvt.u32FdG.m",
[0x1a00] = "vcvtr.s32FdG.m", [0x1a01] = "vcvt.s32FdG.m",
},
}
local map_datac = {
shift = 24, mask = 1,
[0] = {
shift = 4, mask = 1,
[0] = {
shift = 8, mask = 15,
[10] = map_vfps,
[11] = map_vfpd,
-- NYI cdp, mcr, mrc.
},
{
shift = 8, mask = 15,
[10] = {
shift = 20, mask = 15,
[0] = "vmovFnD", "vmovFDn",
[14] = "vmsrD",
[15] = { shift = 12, mask = 15, [15] = "vmrs", _ = "vmrsD", },
},
},
},
"svcT",
}
local map_loadcu = {
shift = 0, mask = 0, -- NYI unconditional CP load/store.
}
local map_datacu = {
shift = 0, mask = 0, -- NYI unconditional CP data.
}
local map_simddata = {
shift = 0, mask = 0, -- NYI SIMD data.
}
local map_simdload = {
shift = 0, mask = 0, -- NYI SIMD load/store, preload.
}
local map_preload = {
shift = 0, mask = 0, -- NYI preload.
}
local map_media = {
shift = 20, mask = 31,
[0] = false,
{ --01
shift = 5, mask = 7,
[0] = "sadd16DNM", "sasxDNM", "ssaxDNM", "ssub16DNM",
"sadd8DNM", false, false, "ssub8DNM",
},
{ --02
shift = 5, mask = 7,
[0] = "qadd16DNM", "qasxDNM", "qsaxDNM", "qsub16DNM",
"qadd8DNM", false, false, "qsub8DNM",
},
{ --03
shift = 5, mask = 7,
[0] = "shadd16DNM", "shasxDNM", "shsaxDNM", "shsub16DNM",
"shadd8DNM", false, false, "shsub8DNM",
},
false,
{ --05
shift = 5, mask = 7,
[0] = "uadd16DNM", "uasxDNM", "usaxDNM", "usub16DNM",
"uadd8DNM", false, false, "usub8DNM",
},
{ --06
shift = 5, mask = 7,
[0] = "uqadd16DNM", "uqasxDNM", "uqsaxDNM", "uqsub16DNM",
"uqadd8DNM", false, false, "uqsub8DNM",
},
{ --07
shift = 5, mask = 7,
[0] = "uhadd16DNM", "uhasxDNM", "uhsaxDNM", "uhsub16DNM",
"uhadd8DNM", false, false, "uhsub8DNM",
},
{ --08
shift = 5, mask = 7,
[0] = "pkhbtDNMU", false, "pkhtbDNMU",
{ shift = 16, mask = 15, [15] = "sxtb16DMU", _ = "sxtab16DNMU", },
"pkhbtDNMU", "selDNM", "pkhtbDNMU",
},
false,
{ --0a
shift = 5, mask = 7,
[0] = "ssatDxMu", "ssat16DxM", "ssatDxMu",
{ shift = 16, mask = 15, [15] = "sxtbDMU", _ = "sxtabDNMU", },
"ssatDxMu", false, "ssatDxMu",
},
{ --0b
shift = 5, mask = 7,
[0] = "ssatDxMu", "revDM", "ssatDxMu",
{ shift = 16, mask = 15, [15] = "sxthDMU", _ = "sxtahDNMU", },
"ssatDxMu", "rev16DM", "ssatDxMu",
},
{ --0c
shift = 5, mask = 7,
[3] = { shift = 16, mask = 15, [15] = "uxtb16DMU", _ = "uxtab16DNMU", },
},
false,
{ --0e
shift = 5, mask = 7,
[0] = "usatDwMu", "usat16DwM", "usatDwMu",
{ shift = 16, mask = 15, [15] = "uxtbDMU", _ = "uxtabDNMU", },
"usatDwMu", false, "usatDwMu",
},
{ --0f
shift = 5, mask = 7,
[0] = "usatDwMu", "rbitDM", "usatDwMu",
{ shift = 16, mask = 15, [15] = "uxthDMU", _ = "uxtahDNMU", },
"usatDwMu", "revshDM", "usatDwMu",
},
{ --10
shift = 12, mask = 15,
[15] = {
shift = 5, mask = 7,
"smuadNMS", "smuadxNMS", "smusdNMS", "smusdxNMS",
},
_ = {
shift = 5, mask = 7,
[0] = "smladNMSD", "smladxNMSD", "smlsdNMSD", "smlsdxNMSD",
},
},
false, false, false,
{ --14
shift = 5, mask = 7,
[0] = "smlaldDNMS", "smlaldxDNMS", "smlsldDNMS", "smlsldxDNMS",
},
{ --15
shift = 5, mask = 7,
[0] = { shift = 12, mask = 15, [15] = "smmulNMS", _ = "smmlaNMSD", },
{ shift = 12, mask = 15, [15] = "smmulrNMS", _ = "smmlarNMSD", },
false, false, false, false,
"smmlsNMSD", "smmlsrNMSD",
},
false, false,
{ --18
shift = 5, mask = 7,
[0] = { shift = 12, mask = 15, [15] = "usad8NMS", _ = "usada8NMSD", },
},
false,
{ --1a
shift = 5, mask = 3, [2] = "sbfxDMvw",
},
{ --1b
shift = 5, mask = 3, [2] = "sbfxDMvw",
},
{ --1c
shift = 5, mask = 3,
[0] = { shift = 0, mask = 15, [15] = "bfcDvX", _ = "bfiDMvX", },
},
{ --1d
shift = 5, mask = 3,
[0] = { shift = 0, mask = 15, [15] = "bfcDvX", _ = "bfiDMvX", },
},
{ --1e
shift = 5, mask = 3, [2] = "ubfxDMvw",
},
{ --1f
shift = 5, mask = 3, [2] = "ubfxDMvw",
},
}
local map_load = {
shift = 21, mask = 9,
{
shift = 20, mask = 5,
[0] = "strtDL", "ldrtDL", [4] = "strbtDL", [5] = "ldrbtDL",
},
_ = {
shift = 20, mask = 5,
[0] = "strDL", "ldrDL", [4] = "strbDL", [5] = "ldrbDL",
}
}
local map_load1 = {
shift = 4, mask = 1,
[0] = map_load, map_media,
}
local map_loadm = {
shift = 20, mask = 1,
[0] = {
shift = 23, mask = 3,
[0] = "stmdaNR", "stmNR",
{ shift = 16, mask = 63, [45] = "pushR", _ = "stmdbNR", }, "stmibNR",
},
{
shift = 23, mask = 3,
[0] = "ldmdaNR", { shift = 16, mask = 63, [61] = "popR", _ = "ldmNR", },
"ldmdbNR", "ldmibNR",
},
}
local map_data = {
shift = 21, mask = 15,
[0] = "andDNPs", "eorDNPs", "subDNPs", "rsbDNPs",
"addDNPs", "adcDNPs", "sbcDNPs", "rscDNPs",
"tstNP", "teqNP", "cmpNP", "cmnNP",
"orrDNPs", "movDPs", "bicDNPs", "mvnDPs",
}
local map_mul = {
shift = 21, mask = 7,
[0] = "mulNMSs", "mlaNMSDs", "umaalDNMS", "mlsDNMS",
"umullDNMSs", "umlalDNMSs", "smullDNMSs", "smlalDNMSs",
}
local map_sync = {
shift = 20, mask = 15, -- NYI: brackets around N. R(D+1) for ldrexd/strexd.
[0] = "swpDMN", false, false, false,
"swpbDMN", false, false, false,
"strexDMN", "ldrexDN", "strexdDN", "ldrexdDN",
"strexbDMN", "ldrexbDN", "strexhDN", "ldrexhDN",
}
local map_mulh = {
shift = 21, mask = 3,
[0] = { shift = 5, mask = 3,
[0] = "smlabbNMSD", "smlatbNMSD", "smlabtNMSD", "smlattNMSD", },
{ shift = 5, mask = 3,
[0] = "smlawbNMSD", "smulwbNMS", "smlawtNMSD", "smulwtNMS", },
{ shift = 5, mask = 3,
[0] = "smlalbbDNMS", "smlaltbDNMS", "smlalbtDNMS", "smlalttDNMS", },
{ shift = 5, mask = 3,
[0] = "smulbbNMS", "smultbNMS", "smulbtNMS", "smulttNMS", },
}
local map_misc = {
shift = 4, mask = 7,
-- NYI: decode PSR bits of msr.
[0] = { shift = 21, mask = 1, [0] = "mrsD", "msrM", },
{ shift = 21, mask = 3, "bxM", false, "clzDM", },
{ shift = 21, mask = 3, "bxjM", },
{ shift = 21, mask = 3, "blxM", },
false,
{ shift = 21, mask = 3, [0] = "qaddDMN", "qsubDMN", "qdaddDMN", "qdsubDMN", },
false,
{ shift = 21, mask = 3, "bkptK", },
}
local map_datar = {
shift = 4, mask = 9,
[9] = {
shift = 5, mask = 3,
[0] = { shift = 24, mask = 1, [0] = map_mul, map_sync, },
{ shift = 20, mask = 1, [0] = "strhDL", "ldrhDL", },
{ shift = 20, mask = 1, [0] = "ldrdDL", "ldrsbDL", },
{ shift = 20, mask = 1, [0] = "strdDL", "ldrshDL", },
},
_ = {
shift = 20, mask = 25,
[16] = { shift = 7, mask = 1, [0] = map_misc, map_mulh, },
_ = {
shift = 0, mask = 0xffffffff,
[bor(0xe1a00000)] = "nop",
_ = map_data,
}
},
}
local map_datai = {
shift = 20, mask = 31, -- NYI: decode PSR bits of msr. Decode imm12.
[16] = "movwDW", [20] = "movtDW",
[18] = { shift = 0, mask = 0xf00ff, [0] = "nopv6", _ = "msrNW", },
[22] = "msrNW",
_ = map_data,
}
local map_branch = {
shift = 24, mask = 1,
[0] = "bB", "blB"
}
local map_condins = {
[0] = map_datar, map_datai, map_load, map_load1,
map_loadm, map_branch, map_loadc, map_datac
}
-- NYI: setend.
local map_uncondins = {
[0] = false, map_simddata, map_simdload, map_preload,
false, "blxB", map_loadcu, map_datacu,
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "sp", "lr", "pc",
}
local map_cond = {
[0] = "eq", "ne", "hs", "lo", "mi", "pl", "vs", "vc",
"hi", "ls", "ge", "lt", "gt", "le", "al",
}
local map_shift = { [0] = "lsl", "lsr", "asr", "ror", }
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then
extra = "\t->"..sym
elseif band(ctx.op, 0x0e000000) ~= 0x0a000000 then
extra = "\t; 0x"..tohex(ctx.rel)
end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-5s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-5s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
-- Format operand 2 of load/store opcodes.
local function fmtload(ctx, op, pos)
local base = map_gpr[band(rshift(op, 16), 15)]
local x, ofs
local ext = (band(op, 0x04000000) == 0)
if not ext and band(op, 0x02000000) == 0 then
ofs = band(op, 4095)
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
ofs = "#"..ofs
elseif ext and band(op, 0x00400000) ~= 0 then
ofs = band(op, 15) + band(rshift(op, 4), 0xf0)
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
ofs = "#"..ofs
else
ofs = map_gpr[band(op, 15)]
if ext or band(op, 0xfe0) == 0 then
elseif band(op, 0xfe0) == 0x60 then
ofs = format("%s, rrx", ofs)
else
local sh = band(rshift(op, 7), 31)
if sh == 0 then sh = 32 end
ofs = format("%s, %s #%d", ofs, map_shift[band(rshift(op, 5), 3)], sh)
end
if band(op, 0x00800000) == 0 then ofs = "-"..ofs end
end
if ofs == "#0" then
x = format("[%s]", base)
elseif band(op, 0x01000000) == 0 then
x = format("[%s], %s", base, ofs)
else
x = format("[%s, %s]", base, ofs)
end
if band(op, 0x01200000) == 0x01200000 then x = x.."!" end
return x
end
-- Format operand 2 of vector load/store opcodes.
local function fmtvload(ctx, op, pos)
local base = map_gpr[band(rshift(op, 16), 15)]
local ofs = band(op, 255)*4
if band(op, 0x00800000) == 0 then ofs = -ofs end
if base == "pc" then ctx.rel = ctx.addr + pos + 8 + ofs end
if ofs == 0 then
return format("[%s]", base)
else
return format("[%s, #%d]", base, ofs)
end
end
local function fmtvr(op, vr, sh0, sh1)
if vr == "s" then
return format("s%d", 2*band(rshift(op, sh0), 15)+band(rshift(op, sh1), 1))
else
return format("d%d", band(rshift(op, sh0), 15)+band(rshift(op, sh1-4), 16))
end
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
local op = bor(lshift(b3, 24), lshift(b2, 16), lshift(b1, 8), b0)
local operands = {}
local suffix = ""
local last, name, pat
local vr
ctx.op = op
ctx.rel = nil
local cond = rshift(op, 28)
local opat
if cond == 15 then
opat = map_uncondins[band(rshift(op, 25), 7)]
else
if cond ~= 14 then suffix = map_cond[cond] end
opat = map_condins[band(rshift(op, 25), 7)]
end
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
opat = opat[band(rshift(op, opat.shift), opat.mask)] or opat._
end
name, pat = match(opat, "^([a-z0-9]*)(.*)")
if sub(pat, 1, 1) == "." then
local s2, p2 = match(pat, "^([a-z0-9.]*)(.*)")
suffix = suffix..s2
pat = p2
end
for p in gmatch(pat, ".") do
local x = nil
if p == "D" then
x = map_gpr[band(rshift(op, 12), 15)]
elseif p == "N" then
x = map_gpr[band(rshift(op, 16), 15)]
elseif p == "S" then
x = map_gpr[band(rshift(op, 8), 15)]
elseif p == "M" then
x = map_gpr[band(op, 15)]
elseif p == "d" then
x = fmtvr(op, vr, 12, 22)
elseif p == "n" then
x = fmtvr(op, vr, 16, 7)
elseif p == "m" then
x = fmtvr(op, vr, 0, 5)
elseif p == "P" then
if band(op, 0x02000000) ~= 0 then
x = ror(band(op, 255), 2*band(rshift(op, 8), 15))
else
x = map_gpr[band(op, 15)]
if band(op, 0xff0) ~= 0 then
operands[#operands+1] = x
local s = map_shift[band(rshift(op, 5), 3)]
local r = nil
if band(op, 0xf90) == 0 then
if s == "ror" then s = "rrx" else r = "#32" end
elseif band(op, 0x10) == 0 then
r = "#"..band(rshift(op, 7), 31)
else
r = map_gpr[band(rshift(op, 8), 15)]
end
if name == "mov" then name = s; x = r
elseif r then x = format("%s %s", s, r)
else x = s end
end
end
elseif p == "L" then
x = fmtload(ctx, op, pos)
elseif p == "l" then
x = fmtvload(ctx, op, pos)
elseif p == "B" then
local addr = ctx.addr + pos + 8 + arshift(lshift(op, 8), 6)
if cond == 15 then addr = addr + band(rshift(op, 23), 2) end
ctx.rel = addr
x = "0x"..tohex(addr)
elseif p == "F" then
vr = "s"
elseif p == "G" then
vr = "d"
elseif p == "." then
suffix = suffix..(vr == "s" and ".f32" or ".f64")
elseif p == "R" then
if band(op, 0x00200000) ~= 0 and #operands == 1 then
operands[1] = operands[1].."!"
end
local t = {}
for i=0,15 do
if band(rshift(op, i), 1) == 1 then t[#t+1] = map_gpr[i] end
end
x = "{"..concat(t, ", ").."}"
elseif p == "r" then
if band(op, 0x00200000) ~= 0 and #operands == 2 then
operands[1] = operands[1].."!"
end
local s = tonumber(sub(last, 2))
local n = band(op, 255)
if vr == "d" then n = rshift(n, 1) end
operands[#operands] = format("{%s-%s%d}", last, vr, s+n-1)
elseif p == "W" then
x = band(op, 0x0fff) + band(rshift(op, 4), 0xf000)
elseif p == "T" then
x = "#0x"..tohex(band(op, 0x00ffffff), 6)
elseif p == "U" then
x = band(rshift(op, 7), 31)
if x == 0 then x = nil end
elseif p == "u" then
x = band(rshift(op, 7), 31)
if band(op, 0x40) == 0 then
if x == 0 then x = nil else x = "lsl #"..x end
else
if x == 0 then x = "asr #32" else x = "asr #"..x end
end
elseif p == "v" then
x = band(rshift(op, 7), 31)
elseif p == "w" then
x = band(rshift(op, 16), 31)
elseif p == "x" then
x = band(rshift(op, 16), 31) + 1
elseif p == "X" then
x = band(rshift(op, 16), 31) - last + 1
elseif p == "Y" then
x = band(rshift(op, 12), 0xf0) + band(op, 0x0f)
elseif p == "K" then
x = "#0x"..tohex(band(rshift(op, 4), 0x0000fff0) + band(op, 15), 4)
elseif p == "s" then
if band(op, 0x00100000) ~= 0 then suffix = "s"..suffix end
else
assert(false)
end
if x then
last = x
if type(x) == "number" then x = "#"..x end
operands[#operands+1] = x
end
end
return putop(ctx, name..suffix, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
ctx.pos = ofs
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 16 then return map_gpr[r] end
return "d"..(r-16)
end
-- Public module functions.
return {
create = create,
disass = disass,
regname = regname
}
File diff suppressed because it is too large Load Diff
@@ -0,0 +1,12 @@
----------------------------------------------------------------------------
-- LuaJIT ARM64BE disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- ARM64 instructions are always little-endian. So just forward to the
-- common ARM64 disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
return require((string.match(..., ".*%.") or "").."dis_arm64")
@@ -0,0 +1,694 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT/X license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles all standard MIPS32R1/R2 instructions.
-- Default mode is big-endian, but see: dis_mipsel.lua
------------------------------------------------------------------------------
local type = type
local byte, format = string.byte, string.format
local match, gmatch = string.match, string.gmatch
local concat = table.concat
local bit = require("bit")
local band, bor, tohex = bit.band, bit.bor, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Extended opcode maps common to all MIPS releases
------------------------------------------------------------------------------
local map_srl = { shift = 21, mask = 1, [0] = "srlDTA", "rotrDTA", }
local map_srlv = { shift = 6, mask = 1, [0] = "srlvDTS", "rotrvDTS", }
local map_cop0 = {
shift = 25, mask = 1,
[0] = {
shift = 21, mask = 15,
[0] = "mfc0TDW", [4] = "mtc0TDW",
[10] = "rdpgprDT",
[11] = { shift = 5, mask = 1, [0] = "diT0", "eiT0", },
[14] = "wrpgprDT",
}, {
shift = 0, mask = 63,
[1] = "tlbr", [2] = "tlbwi", [6] = "tlbwr", [8] = "tlbp",
[24] = "eret", [31] = "deret",
[32] = "wait",
},
}
------------------------------------------------------------------------------
-- Primary and extended opcode maps for MIPS R1-R5
------------------------------------------------------------------------------
local map_movci = { shift = 16, mask = 1, [0] = "movfDSC", "movtDSC", }
local map_special = {
shift = 0, mask = 63,
[0] = { shift = 0, mask = -1, [0] = "nop", _ = "sllDTA" },
map_movci, map_srl, "sraDTA",
"sllvDTS", false, map_srlv, "sravDTS",
"jrS", "jalrD1S", "movzDST", "movnDST",
"syscallY", "breakY", false, "sync",
"mfhiD", "mthiS", "mfloD", "mtloS",
"dsllvDST", false, "dsrlvDST", "dsravDST",
"multST", "multuST", "divST", "divuST",
"dmultST", "dmultuST", "ddivST", "ddivuST",
"addDST", "addu|moveDST0", "subDST", "subu|neguDS0T",
"andDST", "or|moveDST0", "xorDST", "nor|notDST0",
false, false, "sltDST", "sltuDST",
"daddDST", "dadduDST", "dsubDST", "dsubuDST",
"tgeSTZ", "tgeuSTZ", "tltSTZ", "tltuSTZ",
"teqSTZ", false, "tneSTZ", false,
"dsllDTA", false, "dsrlDTA", "dsraDTA",
"dsll32DTA", false, "dsrl32DTA", "dsra32DTA",
}
local map_special2 = {
shift = 0, mask = 63,
[0] = "maddST", "madduST", "mulDST", false,
"msubST", "msubuST",
[32] = "clzDS", [33] = "cloDS",
[63] = "sdbbpY",
}
local map_bshfl = {
shift = 6, mask = 31,
[2] = "wsbhDT",
[16] = "sebDT",
[24] = "sehDT",
}
local map_dbshfl = {
shift = 6, mask = 31,
[2] = "dsbhDT",
[5] = "dshdDT",
}
local map_special3 = {
shift = 0, mask = 63,
[0] = "extTSAK", [1] = "dextmTSAP", [3] = "dextTSAK",
[4] = "insTSAL", [6] = "dinsuTSEQ", [7] = "dinsTSAL",
[32] = map_bshfl, [36] = map_dbshfl, [59] = "rdhwrTD",
}
local map_regimm = {
shift = 16, mask = 31,
[0] = "bltzSB", "bgezSB", "bltzlSB", "bgezlSB",
false, false, false, false,
"tgeiSI", "tgeiuSI", "tltiSI", "tltiuSI",
"teqiSI", false, "tneiSI", false,
"bltzalSB", "bgezalSB", "bltzallSB", "bgezallSB",
false, false, false, false,
false, false, false, false,
false, false, false, "synciSO",
}
local map_cop1s = {
shift = 0, mask = 63,
[0] = "add.sFGH", "sub.sFGH", "mul.sFGH", "div.sFGH",
"sqrt.sFG", "abs.sFG", "mov.sFG", "neg.sFG",
"round.l.sFG", "trunc.l.sFG", "ceil.l.sFG", "floor.l.sFG",
"round.w.sFG", "trunc.w.sFG", "ceil.w.sFG", "floor.w.sFG",
false,
{ shift = 16, mask = 1, [0] = "movf.sFGC", "movt.sFGC" },
"movz.sFGT", "movn.sFGT",
false, "recip.sFG", "rsqrt.sFG", false,
false, false, false, false,
false, false, false, false,
false, "cvt.d.sFG", false, false,
"cvt.w.sFG", "cvt.l.sFG", "cvt.ps.sFGH", false,
false, false, false, false,
false, false, false, false,
"c.f.sVGH", "c.un.sVGH", "c.eq.sVGH", "c.ueq.sVGH",
"c.olt.sVGH", "c.ult.sVGH", "c.ole.sVGH", "c.ule.sVGH",
"c.sf.sVGH", "c.ngle.sVGH", "c.seq.sVGH", "c.ngl.sVGH",
"c.lt.sVGH", "c.nge.sVGH", "c.le.sVGH", "c.ngt.sVGH",
}
local map_cop1d = {
shift = 0, mask = 63,
[0] = "add.dFGH", "sub.dFGH", "mul.dFGH", "div.dFGH",
"sqrt.dFG", "abs.dFG", "mov.dFG", "neg.dFG",
"round.l.dFG", "trunc.l.dFG", "ceil.l.dFG", "floor.l.dFG",
"round.w.dFG", "trunc.w.dFG", "ceil.w.dFG", "floor.w.dFG",
false,
{ shift = 16, mask = 1, [0] = "movf.dFGC", "movt.dFGC" },
"movz.dFGT", "movn.dFGT",
false, "recip.dFG", "rsqrt.dFG", false,
false, false, false, false,
false, false, false, false,
"cvt.s.dFG", false, false, false,
"cvt.w.dFG", "cvt.l.dFG", false, false,
false, false, false, false,
false, false, false, false,
"c.f.dVGH", "c.un.dVGH", "c.eq.dVGH", "c.ueq.dVGH",
"c.olt.dVGH", "c.ult.dVGH", "c.ole.dVGH", "c.ule.dVGH",
"c.df.dVGH", "c.ngle.dVGH", "c.deq.dVGH", "c.ngl.dVGH",
"c.lt.dVGH", "c.nge.dVGH", "c.le.dVGH", "c.ngt.dVGH",
}
local map_cop1ps = {
shift = 0, mask = 63,
[0] = "add.psFGH", "sub.psFGH", "mul.psFGH", false,
false, "abs.psFG", "mov.psFG", "neg.psFG",
false, false, false, false,
false, false, false, false,
false,
{ shift = 16, mask = 1, [0] = "movf.psFGC", "movt.psFGC" },
"movz.psFGT", "movn.psFGT",
false, false, false, false,
false, false, false, false,
false, false, false, false,
"cvt.s.puFG", false, false, false,
false, false, false, false,
"cvt.s.plFG", false, false, false,
"pll.psFGH", "plu.psFGH", "pul.psFGH", "puu.psFGH",
"c.f.psVGH", "c.un.psVGH", "c.eq.psVGH", "c.ueq.psVGH",
"c.olt.psVGH", "c.ult.psVGH", "c.ole.psVGH", "c.ule.psVGH",
"c.psf.psVGH", "c.ngle.psVGH", "c.pseq.psVGH", "c.ngl.psVGH",
"c.lt.psVGH", "c.nge.psVGH", "c.le.psVGH", "c.ngt.psVGH",
}
local map_cop1w = {
shift = 0, mask = 63,
[32] = "cvt.s.wFG", [33] = "cvt.d.wFG",
}
local map_cop1l = {
shift = 0, mask = 63,
[32] = "cvt.s.lFG", [33] = "cvt.d.lFG",
}
local map_cop1bc = {
shift = 16, mask = 3,
[0] = "bc1fCB", "bc1tCB", "bc1flCB", "bc1tlCB",
}
local map_cop1 = {
shift = 21, mask = 31,
[0] = "mfc1TG", "dmfc1TG", "cfc1TG", "mfhc1TG",
"mtc1TG", "dmtc1TG", "ctc1TG", "mthc1TG",
map_cop1bc, false, false, false,
false, false, false, false,
map_cop1s, map_cop1d, false, false,
map_cop1w, map_cop1l, map_cop1ps,
}
local map_cop1x = {
shift = 0, mask = 63,
[0] = "lwxc1FSX", "ldxc1FSX", false, false,
false, "luxc1FSX", false, false,
"swxc1FSX", "sdxc1FSX", false, false,
false, "suxc1FSX", false, "prefxMSX",
false, false, false, false,
false, false, false, false,
false, false, false, false,
false, false, "alnv.psFGHS", false,
"madd.sFRGH", "madd.dFRGH", false, false,
false, false, "madd.psFRGH", false,
"msub.sFRGH", "msub.dFRGH", false, false,
false, false, "msub.psFRGH", false,
"nmadd.sFRGH", "nmadd.dFRGH", false, false,
false, false, "nmadd.psFRGH", false,
"nmsub.sFRGH", "nmsub.dFRGH", false, false,
false, false, "nmsub.psFRGH", false,
}
local map_pri = {
[0] = map_special, map_regimm, "jJ", "jalJ",
"beq|beqz|bST00B", "bne|bnezST0B", "blezSB", "bgtzSB",
"addiTSI", "addiu|liTS0I", "sltiTSI", "sltiuTSI",
"andiTSU", "ori|liTS0U", "xoriTSU", "luiTU",
map_cop0, map_cop1, false, map_cop1x,
"beql|beqzlST0B", "bnel|bnezlST0B", "blezlSB", "bgtzlSB",
"daddiTSI", "daddiuTSI", false, false,
map_special2, "jalxJ", false, map_special3,
"lbTSO", "lhTSO", "lwlTSO", "lwTSO",
"lbuTSO", "lhuTSO", "lwrTSO", false,
"sbTSO", "shTSO", "swlTSO", "swTSO",
false, false, "swrTSO", "cacheNSO",
"llTSO", "lwc1HSO", "lwc2TSO", "prefNSO",
false, "ldc1HSO", "ldc2TSO", "ldTSO",
"scTSO", "swc1HSO", "swc2TSO", false,
false, "sdc1HSO", "sdc2TSO", "sdTSO",
}
------------------------------------------------------------------------------
-- Primary and extended opcode maps for MIPS R6
------------------------------------------------------------------------------
local map_mul_r6 = { shift = 6, mask = 3, [2] = "mulDST", [3] = "muhDST" }
local map_mulu_r6 = { shift = 6, mask = 3, [2] = "muluDST", [3] = "muhuDST" }
local map_div_r6 = { shift = 6, mask = 3, [2] = "divDST", [3] = "modDST" }
local map_divu_r6 = { shift = 6, mask = 3, [2] = "divuDST", [3] = "moduDST" }
local map_dmul_r6 = { shift = 6, mask = 3, [2] = "dmulDST", [3] = "dmuhDST" }
local map_dmulu_r6 = { shift = 6, mask = 3, [2] = "dmuluDST", [3] = "dmuhuDST" }
local map_ddiv_r6 = { shift = 6, mask = 3, [2] = "ddivDST", [3] = "dmodDST" }
local map_ddivu_r6 = { shift = 6, mask = 3, [2] = "ddivuDST", [3] = "dmoduDST" }
local map_special_r6 = {
shift = 0, mask = 63,
[0] = { shift = 0, mask = -1, [0] = "nop", _ = "sllDTA" },
false, map_srl, "sraDTA",
"sllvDTS", false, map_srlv, "sravDTS",
"jrS", "jalrD1S", false, false,
"syscallY", "breakY", false, "sync",
"clzDS", "cloDS", "dclzDS", "dcloDS",
"dsllvDST", "dlsaDSTA", "dsrlvDST", "dsravDST",
map_mul_r6, map_mulu_r6, map_div_r6, map_divu_r6,
map_dmul_r6, map_dmulu_r6, map_ddiv_r6, map_ddivu_r6,
"addDST", "addu|moveDST0", "subDST", "subu|neguDS0T",
"andDST", "or|moveDST0", "xorDST", "nor|notDST0",
false, false, "sltDST", "sltuDST",
"daddDST", "dadduDST", "dsubDST", "dsubuDST",
"tgeSTZ", "tgeuSTZ", "tltSTZ", "tltuSTZ",
"teqSTZ", "seleqzDST", "tneSTZ", "selnezDST",
"dsllDTA", false, "dsrlDTA", "dsraDTA",
"dsll32DTA", false, "dsrl32DTA", "dsra32DTA",
}
local map_bshfl_r6 = {
shift = 9, mask = 3,
[1] = "alignDSTa",
_ = {
shift = 6, mask = 31,
[0] = "bitswapDT",
[2] = "wsbhDT",
[16] = "sebDT",
[24] = "sehDT",
}
}
local map_dbshfl_r6 = {
shift = 9, mask = 3,
[1] = "dalignDSTa",
_ = {
shift = 6, mask = 31,
[0] = "dbitswapDT",
[2] = "dsbhDT",
[5] = "dshdDT",
}
}
local map_special3_r6 = {
shift = 0, mask = 63,
[0] = "extTSAK", [1] = "dextmTSAP", [3] = "dextTSAK",
[4] = "insTSAL", [6] = "dinsuTSEQ", [7] = "dinsTSAL",
[32] = map_bshfl_r6, [36] = map_dbshfl_r6, [59] = "rdhwrTD",
}
local map_regimm_r6 = {
shift = 16, mask = 31,
[0] = "bltzSB", [1] = "bgezSB",
[6] = "dahiSI", [30] = "datiSI",
[23] = "sigrieI", [31] = "synciSO",
}
local map_pcrel_r6 = {
shift = 19, mask = 3,
[0] = "addiupcS2", "lwpcS2", "lwupcS2", {
shift = 18, mask = 1,
[0] = "ldpcS3", { shift = 16, mask = 3, [2] = "auipcSI", [3] = "aluipcSI" }
}
}
local map_cop1s_r6 = {
shift = 0, mask = 63,
[0] = "add.sFGH", "sub.sFGH", "mul.sFGH", "div.sFGH",
"sqrt.sFG", "abs.sFG", "mov.sFG", "neg.sFG",
"round.l.sFG", "trunc.l.sFG", "ceil.l.sFG", "floor.l.sFG",
"round.w.sFG", "trunc.w.sFG", "ceil.w.sFG", "floor.w.sFG",
"sel.sFGH", false, false, false,
"seleqz.sFGH", "recip.sFG", "rsqrt.sFG", "selnez.sFGH",
"maddf.sFGH", "msubf.sFGH", "rint.sFG", "class.sFG",
"min.sFGH", "mina.sFGH", "max.sFGH", "maxa.sFGH",
false, "cvt.d.sFG", false, false,
"cvt.w.sFG", "cvt.l.sFG",
}
local map_cop1d_r6 = {
shift = 0, mask = 63,
[0] = "add.dFGH", "sub.dFGH", "mul.dFGH", "div.dFGH",
"sqrt.dFG", "abs.dFG", "mov.dFG", "neg.dFG",
"round.l.dFG", "trunc.l.dFG", "ceil.l.dFG", "floor.l.dFG",
"round.w.dFG", "trunc.w.dFG", "ceil.w.dFG", "floor.w.dFG",
"sel.dFGH", false, false, false,
"seleqz.dFGH", "recip.dFG", "rsqrt.dFG", "selnez.dFGH",
"maddf.dFGH", "msubf.dFGH", "rint.dFG", "class.dFG",
"min.dFGH", "mina.dFGH", "max.dFGH", "maxa.dFGH",
"cvt.s.dFG", false, false, false,
"cvt.w.dFG", "cvt.l.dFG",
}
local map_cop1w_r6 = {
shift = 0, mask = 63,
[0] = "cmp.af.sFGH", "cmp.un.sFGH", "cmp.eq.sFGH", "cmp.ueq.sFGH",
"cmp.lt.sFGH", "cmp.ult.sFGH", "cmp.le.sFGH", "cmp.ule.sFGH",
"cmp.saf.sFGH", "cmp.sun.sFGH", "cmp.seq.sFGH", "cmp.sueq.sFGH",
"cmp.slt.sFGH", "cmp.sult.sFGH", "cmp.sle.sFGH", "cmp.sule.sFGH",
false, "cmp.or.sFGH", "cmp.une.sFGH", "cmp.ne.sFGH",
false, false, false, false,
false, "cmp.sor.sFGH", "cmp.sune.sFGH", "cmp.sne.sFGH",
false, false, false, false,
"cvt.s.wFG", "cvt.d.wFG",
}
local map_cop1l_r6 = {
shift = 0, mask = 63,
[0] = "cmp.af.dFGH", "cmp.un.dFGH", "cmp.eq.dFGH", "cmp.ueq.dFGH",
"cmp.lt.dFGH", "cmp.ult.dFGH", "cmp.le.dFGH", "cmp.ule.dFGH",
"cmp.saf.dFGH", "cmp.sun.dFGH", "cmp.seq.dFGH", "cmp.sueq.dFGH",
"cmp.slt.dFGH", "cmp.sult.dFGH", "cmp.sle.dFGH", "cmp.sule.dFGH",
false, "cmp.or.dFGH", "cmp.une.dFGH", "cmp.ne.dFGH",
false, false, false, false,
false, "cmp.sor.dFGH", "cmp.sune.dFGH", "cmp.sne.dFGH",
false, false, false, false,
"cvt.s.lFG", "cvt.d.lFG",
}
local map_cop1_r6 = {
shift = 21, mask = 31,
[0] = "mfc1TG", "dmfc1TG", "cfc1TG", "mfhc1TG",
"mtc1TG", "dmtc1TG", "ctc1TG", "mthc1TG",
false, "bc1eqzHB", false, false,
false, "bc1nezHB", false, false,
map_cop1s_r6, map_cop1d_r6, false, false,
map_cop1w_r6, map_cop1l_r6,
}
local function maprs_popTS(rs, rt)
if rt == 0 then return 0 elseif rs == 0 then return 1
elseif rs == rt then return 2 else return 3 end
end
local map_pop06_r6 = {
maprs = maprs_popTS, [0] = "blezSB", "blezalcTB", "bgezalcTB", "bgeucSTB"
}
local map_pop07_r6 = {
maprs = maprs_popTS, [0] = "bgtzSB", "bgtzalcTB", "bltzalcTB", "bltucSTB"
}
local map_pop26_r6 = {
maprs = maprs_popTS, "blezcTB", "bgezcTB", "bgecSTB"
}
local map_pop27_r6 = {
maprs = maprs_popTS, "bgtzcTB", "bltzcTB", "bltcSTB"
}
local function maprs_popS(rs, rt)
if rs == 0 then return 0 else return 1 end
end
local map_pop66_r6 = {
maprs = maprs_popS, [0] = "jicTI", "beqzcSb"
}
local map_pop76_r6 = {
maprs = maprs_popS, [0] = "jialcTI", "bnezcSb"
}
local function maprs_popST(rs, rt)
if rs >= rt then return 0 elseif rs == 0 then return 1 else return 2 end
end
local map_pop10_r6 = {
maprs = maprs_popST, [0] = "bovcSTB", "beqzalcTB", "beqcSTB"
}
local map_pop30_r6 = {
maprs = maprs_popST, [0] = "bnvcSTB", "bnezalcTB", "bnecSTB"
}
local map_pri_r6 = {
[0] = map_special_r6, map_regimm_r6, "jJ", "jalJ",
"beq|beqz|bST00B", "bne|bnezST0B", map_pop06_r6, map_pop07_r6,
map_pop10_r6, "addiu|liTS0I", "sltiTSI", "sltiuTSI",
"andiTSU", "ori|liTS0U", "xoriTSU", "aui|luiTS0U",
map_cop0, map_cop1_r6, false, false,
false, false, map_pop26_r6, map_pop27_r6,
map_pop30_r6, "daddiuTSI", false, false,
false, "dauiTSI", false, map_special3_r6,
"lbTSO", "lhTSO", false, "lwTSO",
"lbuTSO", "lhuTSO", false, false,
"sbTSO", "shTSO", false, "swTSO",
false, false, false, false,
false, "lwc1HSO", "bc#", false,
false, "ldc1HSO", map_pop66_r6, "ldTSO",
false, "swc1HSO", "balc#", map_pcrel_r6,
false, "sdc1HSO", map_pop76_r6, "sdTSO",
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "r1", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
"r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
"r24", "r25", "r26", "r27", "r28", "sp", "r30", "ra",
}
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then extra = "\t->"..sym end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-7s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-7s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
local function get_be(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
return bor(lshift(b0, 24), lshift(b1, 16), lshift(b2, 8), b3)
end
local function get_le(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
return bor(lshift(b3, 24), lshift(b2, 16), lshift(b1, 8), b0)
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local op = ctx:get()
local operands = {}
local last = nil
ctx.op = op
ctx.rel = nil
local opat = ctx.map_pri[rshift(op, 26)]
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
if opat.maprs then
opat = opat[opat.maprs(band(rshift(op,21),31), band(rshift(op,16),31))]
else
opat = opat[band(rshift(op, opat.shift), opat.mask)] or opat._
end
end
local name, pat = match(opat, "^([a-z0-9_.]*)(.*)")
local altname, pat2 = match(pat, "|([a-z0-9_.|]*)(.*)")
if altname then pat = pat2 end
for p in gmatch(pat, ".") do
local x = nil
if p == "S" then
x = map_gpr[band(rshift(op, 21), 31)]
elseif p == "T" then
x = map_gpr[band(rshift(op, 16), 31)]
elseif p == "D" then
x = map_gpr[band(rshift(op, 11), 31)]
elseif p == "F" then
x = "f"..band(rshift(op, 6), 31)
elseif p == "G" then
x = "f"..band(rshift(op, 11), 31)
elseif p == "H" then
x = "f"..band(rshift(op, 16), 31)
elseif p == "R" then
x = "f"..band(rshift(op, 21), 31)
elseif p == "A" then
x = band(rshift(op, 6), 31)
elseif p == "a" then
x = band(rshift(op, 6), 7)
elseif p == "E" then
x = band(rshift(op, 6), 31) + 32
elseif p == "M" then
x = band(rshift(op, 11), 31)
elseif p == "N" then
x = band(rshift(op, 16), 31)
elseif p == "C" then
x = band(rshift(op, 18), 7)
if x == 0 then x = nil end
elseif p == "K" then
x = band(rshift(op, 11), 31) + 1
elseif p == "P" then
x = band(rshift(op, 11), 31) + 33
elseif p == "L" then
x = band(rshift(op, 11), 31) - last + 1
elseif p == "Q" then
x = band(rshift(op, 11), 31) - last + 33
elseif p == "I" then
x = arshift(lshift(op, 16), 16)
elseif p == "2" then
x = arshift(lshift(op, 13), 11)
elseif p == "3" then
x = arshift(lshift(op, 14), 11)
elseif p == "U" then
x = band(op, 0xffff)
elseif p == "O" then
local disp = arshift(lshift(op, 16), 16)
operands[#operands] = format("%d(%s)", disp, last)
elseif p == "X" then
local index = map_gpr[band(rshift(op, 16), 31)]
operands[#operands] = format("%s(%s)", index, last)
elseif p == "B" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 16), 14) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "b" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 11), 9) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "#" then
x = ctx.addr + ctx.pos + arshift(lshift(op, 6), 4) + 4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "J" then
local a = ctx.addr + ctx.pos
x = a - band(a, 0x0fffffff) + band(op, 0x03ffffff)*4
ctx.rel = x
x = format("0x%08x", x)
elseif p == "V" then
x = band(rshift(op, 8), 7)
if x == 0 then x = nil end
elseif p == "W" then
x = band(op, 7)
if x == 0 then x = nil end
elseif p == "Y" then
x = band(rshift(op, 6), 0x000fffff)
if x == 0 then x = nil end
elseif p == "Z" then
x = band(rshift(op, 6), 1023)
if x == 0 then x = nil end
elseif p == "0" then
if last == "r0" or last == 0 then
local n = #operands
operands[n] = nil
last = operands[n-1]
if altname then
local a1, a2 = match(altname, "([^|]*)|(.*)")
if a1 then name, altname = a1, a2
else name = altname end
end
end
elseif p == "1" then
if last == "ra" then
operands[#operands] = nil
end
else
assert(false)
end
if x then operands[#operands+1] = x; last = x end
end
return putop(ctx, name, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
stop = stop - stop % 4
ctx.pos = ofs - ofs % 4
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
ctx.get = get_be
ctx.map_pri = map_pri
return ctx
end
local function create_el(code, addr, out)
local ctx = create(code, addr, out)
ctx.get = get_le
return ctx
end
local function create_r6(code, addr, out)
local ctx = create(code, addr, out)
ctx.map_pri = map_pri_r6
return ctx
end
local function create_r6_el(code, addr, out)
local ctx = create(code, addr, out)
ctx.get = get_le
ctx.map_pri = map_pri_r6
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
local function disass_el(code, addr, out)
create_el(code, addr, out):disass()
end
local function disass_r6(code, addr, out)
create_r6(code, addr, out):disass()
end
local function disass_r6_el(code, addr, out)
create_r6_el(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 32 then return map_gpr[r] end
return "f"..(r-32)
end
-- Public module functions.
return {
create = create,
create_el = create_el,
create_r6 = create_r6,
create_r6_el = create_r6_el,
disass = disass,
disass_el = disass_el,
disass_r6 = disass_r6,
disass_r6_el = disass_r6_el,
regname = regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the big-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create,
disass = dis_mips.disass,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64EL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_el,
disass = dis_mips.disass_el,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64R6 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the r6 big-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_r6,
disass = dis_mips.disass_r6,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPS64R6EL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the r6 little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_r6_el,
disass = dis_mips.disass_r6_el,
regname = dis_mips.regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT MIPSEL disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the little-endian functions from the
-- MIPS disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
return {
create = dis_mips.create_el,
disass = dis_mips.disass_el,
regname = dis_mips.regname
}
@@ -0,0 +1,591 @@
----------------------------------------------------------------------------
-- LuaJIT PPC disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT/X license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- It disassembles all common, non-privileged 32/64 bit PowerPC instructions
-- plus the e500 SPE instructions and some Cell/Xenon extensions.
--
-- NYI: VMX, VMX128
------------------------------------------------------------------------------
local type = type
local byte, format = string.byte, string.format
local match, gmatch, gsub = string.match, string.gmatch, string.gsub
local concat = table.concat
local bit = require("bit")
local band, bor, tohex = bit.band, bit.bor, bit.tohex
local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
------------------------------------------------------------------------------
-- Primary and extended opcode maps
------------------------------------------------------------------------------
local map_crops = {
shift = 1, mask = 1023,
[0] = "mcrfXX",
[33] = "crnor|crnotCCC=", [129] = "crandcCCC",
[193] = "crxor|crclrCCC%", [225] = "crnandCCC",
[257] = "crandCCC", [289] = "creqv|crsetCCC%",
[417] = "crorcCCC", [449] = "cror|crmoveCCC=",
[16] = "b_lrKB", [528] = "b_ctrKB",
[150] = "isync",
}
local map_rlwinm = setmetatable({
shift = 0, mask = -1,
},
{ __index = function(t, x)
local rot = band(rshift(x, 11), 31)
local mb = band(rshift(x, 6), 31)
local me = band(rshift(x, 1), 31)
if mb == 0 and me == 31-rot then
return "slwiRR~A."
elseif me == 31 and mb == 32-rot then
return "srwiRR~-A."
else
return "rlwinmRR~AAA."
end
end
})
local map_rld = {
shift = 2, mask = 7,
[0] = "rldiclRR~HM.", "rldicrRR~HM.", "rldicRR~HM.", "rldimiRR~HM.",
{
shift = 1, mask = 1,
[0] = "rldclRR~RM.", "rldcrRR~RM.",
},
}
local map_ext = setmetatable({
shift = 1, mask = 1023,
[0] = "cmp_YLRR", [32] = "cmpl_YLRR",
[4] = "twARR", [68] = "tdARR",
[8] = "subfcRRR.", [40] = "subfRRR.",
[104] = "negRR.", [136] = "subfeRRR.",
[200] = "subfzeRR.", [232] = "subfmeRR.",
[520] = "subfcoRRR.", [552] = "subfoRRR.",
[616] = "negoRR.", [648] = "subfeoRRR.",
[712] = "subfzeoRR.", [744] = "subfmeoRR.",
[9] = "mulhduRRR.", [73] = "mulhdRRR.", [233] = "mulldRRR.",
[457] = "divduRRR.", [489] = "divdRRR.",
[745] = "mulldoRRR.",
[969] = "divduoRRR.", [1001] = "divdoRRR.",
[10] = "addcRRR.", [138] = "addeRRR.",
[202] = "addzeRR.", [234] = "addmeRR.", [266] = "addRRR.",
[522] = "addcoRRR.", [650] = "addeoRRR.",
[714] = "addzeoRR.", [746] = "addmeoRR.", [778] = "addoRRR.",
[11] = "mulhwuRRR.", [75] = "mulhwRRR.", [235] = "mullwRRR.",
[459] = "divwuRRR.", [491] = "divwRRR.",
[747] = "mullwoRRR.",
[971] = "divwouRRR.", [1003] = "divwoRRR.",
[15] = "iselltRRR", [47] = "iselgtRRR", [79] = "iseleqRRR",
[144] = { shift = 20, mask = 1, [0] = "mtcrfRZ~", "mtocrfRZ~", },
[19] = { shift = 20, mask = 1, [0] = "mfcrR", "mfocrfRZ", },
[371] = { shift = 11, mask = 1023, [392] = "mftbR", [424] = "mftbuR", },
[339] = {
shift = 11, mask = 1023,
[32] = "mferR", [256] = "mflrR", [288] = "mfctrR", [16] = "mfspefscrR",
},
[467] = {
shift = 11, mask = 1023,
[32] = "mtxerR", [256] = "mtlrR", [288] = "mtctrR", [16] = "mtspefscrR",
},
[20] = "lwarxRR0R", [84] = "ldarxRR0R",
[21] = "ldxRR0R", [53] = "lduxRRR",
[149] = "stdxRR0R", [181] = "stduxRRR",
[341] = "lwaxRR0R", [373] = "lwauxRRR",
[23] = "lwzxRR0R", [55] = "lwzuxRRR",
[87] = "lbzxRR0R", [119] = "lbzuxRRR",
[151] = "stwxRR0R", [183] = "stwuxRRR",
[215] = "stbxRR0R", [247] = "stbuxRRR",
[279] = "lhzxRR0R", [311] = "lhzuxRRR",
[343] = "lhaxRR0R", [375] = "lhauxRRR",
[407] = "sthxRR0R", [439] = "sthuxRRR",
[54] = "dcbst-R0R", [86] = "dcbf-R0R",
[150] = "stwcxRR0R.", [214] = "stdcxRR0R.",
[246] = "dcbtst-R0R", [278] = "dcbt-R0R",
[310] = "eciwxRR0R", [438] = "ecowxRR0R",
[470] = "dcbi-RR",
[598] = {
shift = 21, mask = 3,
[0] = "sync", "lwsync", "ptesync",
},
[758] = "dcba-RR",
[854] = "eieio", [982] = "icbi-R0R", [1014] = "dcbz-R0R",
[26] = "cntlzwRR~", [58] = "cntlzdRR~",
[122] = "popcntbRR~",
[154] = "prtywRR~", [186] = "prtydRR~",
[28] = "andRR~R.", [60] = "andcRR~R.", [124] = "nor|notRR~R=.",
[284] = "eqvRR~R.", [316] = "xorRR~R.",
[412] = "orcRR~R.", [444] = "or|mrRR~R=.", [476] = "nandRR~R.",
[508] = "cmpbRR~R",
[512] = "mcrxrX",
[532] = "ldbrxRR0R", [660] = "stdbrxRR0R",
[533] = "lswxRR0R", [597] = "lswiRR0A",
[661] = "stswxRR0R", [725] = "stswiRR0A",
[534] = "lwbrxRR0R", [662] = "stwbrxRR0R",
[790] = "lhbrxRR0R", [918] = "sthbrxRR0R",
[535] = "lfsxFR0R", [567] = "lfsuxFRR",
[599] = "lfdxFR0R", [631] = "lfduxFRR",
[663] = "stfsxFR0R", [695] = "stfsuxFRR",
[727] = "stfdxFR0R", [759] = "stfduxFR0R",
[855] = "lfiwaxFR0R",
[983] = "stfiwxFR0R",
[24] = "slwRR~R.",
[27] = "sldRR~R.", [536] = "srwRR~R.",
[792] = "srawRR~R.", [824] = "srawiRR~A.",
[794] = "sradRR~R.", [826] = "sradiRR~H.", [827] = "sradiRR~H.",
[922] = "extshRR~.", [954] = "extsbRR~.", [986] = "extswRR~.",
[539] = "srdRR~R.",
},
{ __index = function(t, x)
if band(x, 31) == 15 then return "iselRRRC" end
end
})
local map_ld = {
shift = 0, mask = 3,
[0] = "ldRRE", "lduRRE", "lwaRRE",
}
local map_std = {
shift = 0, mask = 3,
[0] = "stdRRE", "stduRRE",
}
local map_fps = {
shift = 5, mask = 1,
{
shift = 1, mask = 15,
[0] = false, false, "fdivsFFF.", false,
"fsubsFFF.", "faddsFFF.", "fsqrtsF-F.", false,
"fresF-F.", "fmulsFF-F.", "frsqrtesF-F.", false,
"fmsubsFFFF~.", "fmaddsFFFF~.", "fnmsubsFFFF~.", "fnmaddsFFFF~.",
}
}
local map_fpd = {
shift = 5, mask = 1,
[0] = {
shift = 1, mask = 1023,
[0] = "fcmpuXFF", [32] = "fcmpoXFF", [64] = "mcrfsXX",
[38] = "mtfsb1A.", [70] = "mtfsb0A.", [134] = "mtfsfiA>>-A>",
[8] = "fcpsgnFFF.", [40] = "fnegF-F.", [72] = "fmrF-F.",
[136] = "fnabsF-F.", [264] = "fabsF-F.",
[12] = "frspF-F.",
[14] = "fctiwF-F.", [15] = "fctiwzF-F.",
[583] = "mffsF.", [711] = "mtfsfZF.",
[392] = "frinF-F.", [424] = "frizF-F.",
[456] = "fripF-F.", [488] = "frimF-F.",
[814] = "fctidF-F.", [815] = "fctidzF-F.", [846] = "fcfidF-F.",
},
{
shift = 1, mask = 15,
[0] = false, false, "fdivFFF.", false,
"fsubFFF.", "faddFFF.", "fsqrtF-F.", "fselFFFF~.",
"freF-F.", "fmulFF-F.", "frsqrteF-F.", false,
"fmsubFFFF~.", "fmaddFFFF~.", "fnmsubFFFF~.", "fnmaddFFFF~.",
}
}
local map_spe = {
shift = 0, mask = 2047,
[512] = "evaddwRRR", [514] = "evaddiwRAR~",
[516] = "evsubwRRR~", [518] = "evsubiwRAR~",
[520] = "evabsRR", [521] = "evnegRR",
[522] = "evextsbRR", [523] = "evextshRR", [524] = "evrndwRR",
[525] = "evcntlzwRR", [526] = "evcntlswRR",
[527] = "brincRRR",
[529] = "evandRRR", [530] = "evandcRRR", [534] = "evxorRRR",
[535] = "evor|evmrRRR=", [536] = "evnor|evnotRRR=",
[537] = "eveqvRRR", [539] = "evorcRRR", [542] = "evnandRRR",
[544] = "evsrwuRRR", [545] = "evsrwsRRR",
[546] = "evsrwiuRRA", [547] = "evsrwisRRA",
[548] = "evslwRRR", [550] = "evslwiRRA",
[552] = "evrlwRRR", [553] = "evsplatiRS",
[554] = "evrlwiRRA", [555] = "evsplatfiRS",
[556] = "evmergehiRRR", [557] = "evmergeloRRR",
[558] = "evmergehiloRRR", [559] = "evmergelohiRRR",
[560] = "evcmpgtuYRR", [561] = "evcmpgtsYRR",
[562] = "evcmpltuYRR", [563] = "evcmpltsYRR",
[564] = "evcmpeqYRR",
[632] = "evselRRR", [633] = "evselRRRW",
[634] = "evselRRRW", [635] = "evselRRRW",
[636] = "evselRRRW", [637] = "evselRRRW",
[638] = "evselRRRW", [639] = "evselRRRW",
[640] = "evfsaddRRR", [641] = "evfssubRRR",
[644] = "evfsabsRR", [645] = "evfsnabsRR", [646] = "evfsnegRR",
[648] = "evfsmulRRR", [649] = "evfsdivRRR",
[652] = "evfscmpgtYRR", [653] = "evfscmpltYRR", [654] = "evfscmpeqYRR",
[656] = "evfscfuiR-R", [657] = "evfscfsiR-R",
[658] = "evfscfufR-R", [659] = "evfscfsfR-R",
[660] = "evfsctuiR-R", [661] = "evfsctsiR-R",
[662] = "evfsctufR-R", [663] = "evfsctsfR-R",
[664] = "evfsctuizR-R", [666] = "evfsctsizR-R",
[668] = "evfststgtYRR", [669] = "evfststltYRR", [670] = "evfststeqYRR",
[704] = "efsaddRRR", [705] = "efssubRRR",
[708] = "efsabsRR", [709] = "efsnabsRR", [710] = "efsnegRR",
[712] = "efsmulRRR", [713] = "efsdivRRR",
[716] = "efscmpgtYRR", [717] = "efscmpltYRR", [718] = "efscmpeqYRR",
[719] = "efscfdR-R",
[720] = "efscfuiR-R", [721] = "efscfsiR-R",
[722] = "efscfufR-R", [723] = "efscfsfR-R",
[724] = "efsctuiR-R", [725] = "efsctsiR-R",
[726] = "efsctufR-R", [727] = "efsctsfR-R",
[728] = "efsctuizR-R", [730] = "efsctsizR-R",
[732] = "efststgtYRR", [733] = "efststltYRR", [734] = "efststeqYRR",
[736] = "efdaddRRR", [737] = "efdsubRRR",
[738] = "efdcfuidR-R", [739] = "efdcfsidR-R",
[740] = "efdabsRR", [741] = "efdnabsRR", [742] = "efdnegRR",
[744] = "efdmulRRR", [745] = "efddivRRR",
[746] = "efdctuidzR-R", [747] = "efdctsidzR-R",
[748] = "efdcmpgtYRR", [749] = "efdcmpltYRR", [750] = "efdcmpeqYRR",
[751] = "efdcfsR-R",
[752] = "efdcfuiR-R", [753] = "efdcfsiR-R",
[754] = "efdcfufR-R", [755] = "efdcfsfR-R",
[756] = "efdctuiR-R", [757] = "efdctsiR-R",
[758] = "efdctufR-R", [759] = "efdctsfR-R",
[760] = "efdctuizR-R", [762] = "efdctsizR-R",
[764] = "efdtstgtYRR", [765] = "efdtstltYRR", [766] = "efdtsteqYRR",
[768] = "evlddxRR0R", [769] = "evlddRR8",
[770] = "evldwxRR0R", [771] = "evldwRR8",
[772] = "evldhxRR0R", [773] = "evldhRR8",
[776] = "evlhhesplatxRR0R", [777] = "evlhhesplatRR2",
[780] = "evlhhousplatxRR0R", [781] = "evlhhousplatRR2",
[782] = "evlhhossplatxRR0R", [783] = "evlhhossplatRR2",
[784] = "evlwhexRR0R", [785] = "evlwheRR4",
[788] = "evlwhouxRR0R", [789] = "evlwhouRR4",
[790] = "evlwhosxRR0R", [791] = "evlwhosRR4",
[792] = "evlwwsplatxRR0R", [793] = "evlwwsplatRR4",
[796] = "evlwhsplatxRR0R", [797] = "evlwhsplatRR4",
[800] = "evstddxRR0R", [801] = "evstddRR8",
[802] = "evstdwxRR0R", [803] = "evstdwRR8",
[804] = "evstdhxRR0R", [805] = "evstdhRR8",
[816] = "evstwhexRR0R", [817] = "evstwheRR4",
[820] = "evstwhoxRR0R", [821] = "evstwhoRR4",
[824] = "evstwwexRR0R", [825] = "evstwweRR4",
[828] = "evstwwoxRR0R", [829] = "evstwwoRR4",
[1027] = "evmhessfRRR", [1031] = "evmhossfRRR", [1032] = "evmheumiRRR",
[1033] = "evmhesmiRRR", [1035] = "evmhesmfRRR", [1036] = "evmhoumiRRR",
[1037] = "evmhosmiRRR", [1039] = "evmhosmfRRR", [1059] = "evmhessfaRRR",
[1063] = "evmhossfaRRR", [1064] = "evmheumiaRRR", [1065] = "evmhesmiaRRR",
[1067] = "evmhesmfaRRR", [1068] = "evmhoumiaRRR", [1069] = "evmhosmiaRRR",
[1071] = "evmhosmfaRRR", [1095] = "evmwhssfRRR", [1096] = "evmwlumiRRR",
[1100] = "evmwhumiRRR", [1101] = "evmwhsmiRRR", [1103] = "evmwhsmfRRR",
[1107] = "evmwssfRRR", [1112] = "evmwumiRRR", [1113] = "evmwsmiRRR",
[1115] = "evmwsmfRRR", [1127] = "evmwhssfaRRR", [1128] = "evmwlumiaRRR",
[1132] = "evmwhumiaRRR", [1133] = "evmwhsmiaRRR", [1135] = "evmwhsmfaRRR",
[1139] = "evmwssfaRRR", [1144] = "evmwumiaRRR", [1145] = "evmwsmiaRRR",
[1147] = "evmwsmfaRRR",
[1216] = "evaddusiaawRR", [1217] = "evaddssiaawRR",
[1218] = "evsubfusiaawRR", [1219] = "evsubfssiaawRR",
[1220] = "evmraRR",
[1222] = "evdivwsRRR", [1223] = "evdivwuRRR",
[1224] = "evaddumiaawRR", [1225] = "evaddsmiaawRR",
[1226] = "evsubfumiaawRR", [1227] = "evsubfsmiaawRR",
[1280] = "evmheusiaawRRR", [1281] = "evmhessiaawRRR",
[1283] = "evmhessfaawRRR", [1284] = "evmhousiaawRRR",
[1285] = "evmhossiaawRRR", [1287] = "evmhossfaawRRR",
[1288] = "evmheumiaawRRR", [1289] = "evmhesmiaawRRR",
[1291] = "evmhesmfaawRRR", [1292] = "evmhoumiaawRRR",
[1293] = "evmhosmiaawRRR", [1295] = "evmhosmfaawRRR",
[1320] = "evmhegumiaaRRR", [1321] = "evmhegsmiaaRRR",
[1323] = "evmhegsmfaaRRR", [1324] = "evmhogumiaaRRR",
[1325] = "evmhogsmiaaRRR", [1327] = "evmhogsmfaaRRR",
[1344] = "evmwlusiaawRRR", [1345] = "evmwlssiaawRRR",
[1352] = "evmwlumiaawRRR", [1353] = "evmwlsmiaawRRR",
[1363] = "evmwssfaaRRR", [1368] = "evmwumiaaRRR",
[1369] = "evmwsmiaaRRR", [1371] = "evmwsmfaaRRR",
[1408] = "evmheusianwRRR", [1409] = "evmhessianwRRR",
[1411] = "evmhessfanwRRR", [1412] = "evmhousianwRRR",
[1413] = "evmhossianwRRR", [1415] = "evmhossfanwRRR",
[1416] = "evmheumianwRRR", [1417] = "evmhesmianwRRR",
[1419] = "evmhesmfanwRRR", [1420] = "evmhoumianwRRR",
[1421] = "evmhosmianwRRR", [1423] = "evmhosmfanwRRR",
[1448] = "evmhegumianRRR", [1449] = "evmhegsmianRRR",
[1451] = "evmhegsmfanRRR", [1452] = "evmhogumianRRR",
[1453] = "evmhogsmianRRR", [1455] = "evmhogsmfanRRR",
[1472] = "evmwlusianwRRR", [1473] = "evmwlssianwRRR",
[1480] = "evmwlumianwRRR", [1481] = "evmwlsmianwRRR",
[1491] = "evmwssfanRRR", [1496] = "evmwumianRRR",
[1497] = "evmwsmianRRR", [1499] = "evmwsmfanRRR",
}
local map_pri = {
[0] = false, false, "tdiARI", "twiARI",
map_spe, false, false, "mulliRRI",
"subficRRI", false, "cmpl_iYLRU", "cmp_iYLRI",
"addicRRI", "addic.RRI", "addi|liRR0I", "addis|lisRR0I",
"b_KBJ", "sc", "bKJ", map_crops,
"rlwimiRR~AAA.", map_rlwinm, false, "rlwnmRR~RAA.",
"oriNRR~U", "orisRR~U", "xoriRR~U", "xorisRR~U",
"andi.RR~U", "andis.RR~U", map_rld, map_ext,
"lwzRRD", "lwzuRRD", "lbzRRD", "lbzuRRD",
"stwRRD", "stwuRRD", "stbRRD", "stbuRRD",
"lhzRRD", "lhzuRRD", "lhaRRD", "lhauRRD",
"sthRRD", "sthuRRD", "lmwRRD", "stmwRRD",
"lfsFRD", "lfsuFRD", "lfdFRD", "lfduFRD",
"stfsFRD", "stfsuFRD", "stfdFRD", "stfduFRD",
false, false, map_ld, map_fps,
false, false, map_std, map_fpd,
}
------------------------------------------------------------------------------
local map_gpr = {
[0] = "r0", "sp", "r2", "r3", "r4", "r5", "r6", "r7",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15",
"r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23",
"r24", "r25", "r26", "r27", "r28", "r29", "r30", "r31",
}
local map_cond = { [0] = "lt", "gt", "eq", "so", "ge", "le", "ne", "ns", }
-- Format a condition bit.
local function condfmt(cond)
if cond <= 3 then
return map_cond[band(cond, 3)]
else
return format("4*cr%d+%s", rshift(cond, 2), map_cond[band(cond, 3)])
end
end
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local pos = ctx.pos
local extra = ""
if ctx.rel then
local sym = ctx.symtab[ctx.rel]
if sym then extra = "\t->"..sym end
end
if ctx.hexdump > 0 then
ctx.out(format("%08x %s %-7s %s%s\n",
ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
else
ctx.out(format("%08x %-7s %s%s\n",
ctx.addr+pos, text, concat(operands, ", "), extra))
end
ctx.pos = pos + 4
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
end
-- Disassemble a single instruction.
local function disass_ins(ctx)
local pos = ctx.pos
local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
local op = bor(lshift(b0, 24), lshift(b1, 16), lshift(b2, 8), b3)
local operands = {}
local last = nil
local rs = 21
ctx.op = op
ctx.rel = nil
local opat = map_pri[rshift(b0, 2)]
while type(opat) ~= "string" do
if not opat then return unknown(ctx) end
opat = opat[band(rshift(op, opat.shift), opat.mask)]
end
local name, pat = match(opat, "^([a-z0-9_.]*)(.*)")
local altname, pat2 = match(pat, "|([a-z0-9_.]*)(.*)")
if altname then pat = pat2 end
for p in gmatch(pat, ".") do
local x = nil
if p == "R" then
x = map_gpr[band(rshift(op, rs), 31)]
rs = rs - 5
elseif p == "F" then
x = "f"..band(rshift(op, rs), 31)
rs = rs - 5
elseif p == "A" then
x = band(rshift(op, rs), 31)
rs = rs - 5
elseif p == "S" then
x = arshift(lshift(op, 27-rs), 27)
rs = rs - 5
elseif p == "I" then
x = arshift(lshift(op, 16), 16)
elseif p == "U" then
x = band(op, 0xffff)
elseif p == "D" or p == "E" then
local disp = arshift(lshift(op, 16), 16)
if p == "E" then disp = band(disp, -4) end
if last == "r0" then last = "0" end
operands[#operands] = format("%d(%s)", disp, last)
elseif p >= "2" and p <= "8" then
local disp = band(rshift(op, rs), 31) * p
if last == "r0" then last = "0" end
operands[#operands] = format("%d(%s)", disp, last)
elseif p == "H" then
x = band(rshift(op, rs), 31) + lshift(band(op, 2), 4)
rs = rs - 5
elseif p == "M" then
x = band(rshift(op, rs), 31) + band(op, 0x20)
elseif p == "C" then
x = condfmt(band(rshift(op, rs), 31))
rs = rs - 5
elseif p == "B" then
local bo = rshift(op, 21)
local cond = band(rshift(op, 16), 31)
local cn = ""
rs = rs - 10
if band(bo, 4) == 0 then
cn = band(bo, 2) == 0 and "dnz" or "dz"
if band(bo, 0x10) == 0 then
cn = cn..(band(bo, 8) == 0 and "f" or "t")
end
if band(bo, 0x10) == 0 then x = condfmt(cond) end
name = name..(band(bo, 1) == band(rshift(op, 15), 1) and "-" or "+")
elseif band(bo, 0x10) == 0 then
cn = map_cond[band(cond, 3) + (band(bo, 8) == 0 and 4 or 0)]
if cond > 3 then x = "cr"..rshift(cond, 2) end
name = name..(band(bo, 1) == band(rshift(op, 15), 1) and "-" or "+")
end
name = gsub(name, "_", cn)
elseif p == "J" then
x = arshift(lshift(op, 27-rs), 29-rs)*4
if band(op, 2) == 0 then x = ctx.addr + pos + x end
ctx.rel = x
x = "0x"..tohex(x)
elseif p == "K" then
if band(op, 1) ~= 0 then name = name.."l" end
if band(op, 2) ~= 0 then name = name.."a" end
elseif p == "X" or p == "Y" then
x = band(rshift(op, rs+2), 7)
if x == 0 and p == "Y" then x = nil else x = "cr"..x end
rs = rs - 5
elseif p == "W" then
x = "cr"..band(op, 7)
elseif p == "Z" then
x = band(rshift(op, rs-4), 255)
rs = rs - 10
elseif p == ">" then
operands[#operands] = rshift(operands[#operands], 1)
elseif p == "0" then
if last == "r0" then
operands[#operands] = nil
if altname then name = altname end
end
elseif p == "L" then
name = gsub(name, "_", band(op, 0x00200000) ~= 0 and "d" or "w")
elseif p == "." then
if band(op, 1) == 1 then name = name.."." end
elseif p == "N" then
if op == 0x60000000 then name = "nop"; break end
elseif p == "~" then
local n = #operands
operands[n-1], operands[n] = operands[n], operands[n-1]
elseif p == "=" then
local n = #operands
if last == operands[n-1] then
operands[n] = nil
name = altname
end
elseif p == "%" then
local n = #operands
if last == operands[n-1] and last == operands[n-2] then
operands[n] = nil
operands[n-1] = nil
name = altname
end
elseif p == "-" then
rs = rs - 5
else
assert(false)
end
if x then operands[#operands+1] = x; last = x end
end
return putop(ctx, name, operands)
end
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
stop = stop - stop % 4
ctx.pos = ofs - ofs % 4
ctx.rel = nil
while ctx.pos < stop do disass_ins(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = addr or 0
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 8
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 32 then return map_gpr[r] end
return "f"..(r-32)
end
-- Public module functions.
return {
create = create,
disass = disass,
regname = regname
}
@@ -0,0 +1,17 @@
----------------------------------------------------------------------------
-- LuaJIT x64 disassembler wrapper module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This module just exports the 64 bit functions from the combined
-- x86/x64 disassembler module. All the interesting stuff is there.
------------------------------------------------------------------------------
local dis_x86 = require((string.match(..., ".*%.") or "").."dis_x86")
return {
create = dis_x86.create64,
disass = dis_x86.disass64,
regname = dis_x86.regname64
}
@@ -0,0 +1,953 @@
----------------------------------------------------------------------------
-- LuaJIT x86/x64 disassembler module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
-- This is a helper module used by the LuaJIT machine code dumper module.
--
-- Sending small code snippets to an external disassembler and mixing the
-- output with our own stuff was too fragile. So I had to bite the bullet
-- and write yet another x86 disassembler. Oh well ...
--
-- The output format is very similar to what ndisasm generates. But it has
-- been developed independently by looking at the opcode tables from the
-- Intel and AMD manuals. The supported instruction set is quite extensive
-- and reflects what a current generation Intel or AMD CPU implements in
-- 32 bit and 64 bit mode. Yes, this includes MMX, SSE, SSE2, SSE3, SSSE3,
-- SSE4.1, SSE4.2, SSE4a, AVX, AVX2 and even privileged and hypervisor
-- (VMX/SVM) instructions.
--
-- Notes:
-- * The (useless) a16 prefix, 3DNow and pre-586 opcodes are unsupported.
-- * No attempt at optimization has been made -- it's fast enough for my needs.
------------------------------------------------------------------------------
local type = type
local sub, byte, format = string.sub, string.byte, string.format
local match, gmatch, gsub = string.match, string.gmatch, string.gsub
local lower, rep = string.lower, string.rep
local bit = require("bit")
local tohex = bit.tohex
-- Map for 1st opcode byte in 32 bit mode. Ugly? Well ... read on.
local map_opc1_32 = {
--0x
[0]="addBmr","addVmr","addBrm","addVrm","addBai","addVai","push es","pop es",
"orBmr","orVmr","orBrm","orVrm","orBai","orVai","push cs","opc2*",
--1x
"adcBmr","adcVmr","adcBrm","adcVrm","adcBai","adcVai","push ss","pop ss",
"sbbBmr","sbbVmr","sbbBrm","sbbVrm","sbbBai","sbbVai","push ds","pop ds",
--2x
"andBmr","andVmr","andBrm","andVrm","andBai","andVai","es:seg","daa",
"subBmr","subVmr","subBrm","subVrm","subBai","subVai","cs:seg","das",
--3x
"xorBmr","xorVmr","xorBrm","xorVrm","xorBai","xorVai","ss:seg","aaa",
"cmpBmr","cmpVmr","cmpBrm","cmpVrm","cmpBai","cmpVai","ds:seg","aas",
--4x
"incVR","incVR","incVR","incVR","incVR","incVR","incVR","incVR",
"decVR","decVR","decVR","decVR","decVR","decVR","decVR","decVR",
--5x
"pushUR","pushUR","pushUR","pushUR","pushUR","pushUR","pushUR","pushUR",
"popUR","popUR","popUR","popUR","popUR","popUR","popUR","popUR",
--6x
"sz*pushaw,pusha","sz*popaw,popa","boundVrm","arplWmr",
"fs:seg","gs:seg","o16:","a16",
"pushUi","imulVrmi","pushBs","imulVrms",
"insb","insVS","outsb","outsVS",
--7x
"joBj","jnoBj","jbBj","jnbBj","jzBj","jnzBj","jbeBj","jaBj",
"jsBj","jnsBj","jpeBj","jpoBj","jlBj","jgeBj","jleBj","jgBj",
--8x
"arith!Bmi","arith!Vmi","arith!Bmi","arith!Vms",
"testBmr","testVmr","xchgBrm","xchgVrm",
"movBmr","movVmr","movBrm","movVrm",
"movVmg","leaVrm","movWgm","popUm",
--9x
"nop*xchgVaR|pause|xchgWaR|repne nop","xchgVaR","xchgVaR","xchgVaR",
"xchgVaR","xchgVaR","xchgVaR","xchgVaR",
"sz*cbw,cwde,cdqe","sz*cwd,cdq,cqo","call farViw","wait",
"sz*pushfw,pushf","sz*popfw,popf","sahf","lahf",
--Ax
"movBao","movVao","movBoa","movVoa",
"movsb","movsVS","cmpsb","cmpsVS",
"testBai","testVai","stosb","stosVS",
"lodsb","lodsVS","scasb","scasVS",
--Bx
"movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi","movBRi",
"movVRI","movVRI","movVRI","movVRI","movVRI","movVRI","movVRI","movVRI",
--Cx
"shift!Bmu","shift!Vmu","retBw","ret","vex*3$lesVrm","vex*2$ldsVrm","movBmi","movVmi",
"enterBwu","leave","retfBw","retf","int3","intBu","into","iretVS",
--Dx
"shift!Bm1","shift!Vm1","shift!Bmc","shift!Vmc","aamBu","aadBu","salc","xlatb",
"fp*0","fp*1","fp*2","fp*3","fp*4","fp*5","fp*6","fp*7",
--Ex
"loopneBj","loopeBj","loopBj","sz*jcxzBj,jecxzBj,jrcxzBj",
"inBau","inVau","outBua","outVua",
"callVj","jmpVj","jmp farViw","jmpBj","inBad","inVad","outBda","outVda",
--Fx
"lock:","int1","repne:rep","rep:","hlt","cmc","testb!Bm","testv!Vm",
"clc","stc","cli","sti","cld","std","incb!Bm","incd!Vm",
}
assert(#map_opc1_32 == 255)
-- Map for 1st opcode byte in 64 bit mode (overrides only).
local map_opc1_64 = setmetatable({
[0x06]=false, [0x07]=false, [0x0e]=false,
[0x16]=false, [0x17]=false, [0x1e]=false, [0x1f]=false,
[0x27]=false, [0x2f]=false, [0x37]=false, [0x3f]=false,
[0x60]=false, [0x61]=false, [0x62]=false, [0x63]="movsxdVrDmt", [0x67]="a32:",
[0x40]="rex*", [0x41]="rex*b", [0x42]="rex*x", [0x43]="rex*xb",
[0x44]="rex*r", [0x45]="rex*rb", [0x46]="rex*rx", [0x47]="rex*rxb",
[0x48]="rex*w", [0x49]="rex*wb", [0x4a]="rex*wx", [0x4b]="rex*wxb",
[0x4c]="rex*wr", [0x4d]="rex*wrb", [0x4e]="rex*wrx", [0x4f]="rex*wrxb",
[0x82]=false, [0x9a]=false, [0xc4]="vex*3", [0xc5]="vex*2", [0xce]=false,
[0xd4]=false, [0xd5]=false, [0xd6]=false, [0xea]=false,
}, { __index = map_opc1_32 })
-- Map for 2nd opcode byte (0F xx). True CISC hell. Hey, I told you.
-- Prefix dependent MMX/SSE opcodes: (none)|rep|o16|repne, -|F3|66|F2
local map_opc2 = {
--0x
[0]="sldt!Dmp","sgdt!Ump","larVrm","lslVrm",nil,"syscall","clts","sysret",
"invd","wbinvd",nil,"ud1",nil,"$prefetch!Bm","femms","3dnowMrmu",
--1x
"movupsXrm|movssXrvm|movupdXrm|movsdXrvm",
"movupsXmr|movssXmvr|movupdXmr|movsdXmvr",
"movhlpsXrm$movlpsXrm|movsldupXrm|movlpdXrm|movddupXrm",
"movlpsXmr||movlpdXmr",
"unpcklpsXrvm||unpcklpdXrvm",
"unpckhpsXrvm||unpckhpdXrvm",
"movlhpsXrm$movhpsXrm|movshdupXrm|movhpdXrm",
"movhpsXmr||movhpdXmr",
"$prefetcht!Bm","hintnopVm","hintnopVm","hintnopVm",
"hintnopVm","hintnopVm","hintnopVm","hintnopVm",
--2x
"movUmx$","movUmy$","movUxm$","movUym$","movUmz$",nil,"movUzm$",nil,
"movapsXrm||movapdXrm",
"movapsXmr||movapdXmr",
"cvtpi2psXrMm|cvtsi2ssXrvVmt|cvtpi2pdXrMm|cvtsi2sdXrvVmt",
"movntpsXmr|movntssXmr|movntpdXmr|movntsdXmr",
"cvttps2piMrXm|cvttss2siVrXm|cvttpd2piMrXm|cvttsd2siVrXm",
"cvtps2piMrXm|cvtss2siVrXm|cvtpd2piMrXm|cvtsd2siVrXm",
"ucomissXrm||ucomisdXrm",
"comissXrm||comisdXrm",
--3x
"wrmsr","rdtsc","rdmsr","rdpmc","sysenter","sysexit",nil,"getsec",
"opc3*38",nil,"opc3*3a",nil,nil,nil,nil,nil,
--4x
"cmovoVrm","cmovnoVrm","cmovbVrm","cmovnbVrm",
"cmovzVrm","cmovnzVrm","cmovbeVrm","cmovaVrm",
"cmovsVrm","cmovnsVrm","cmovpeVrm","cmovpoVrm",
"cmovlVrm","cmovgeVrm","cmovleVrm","cmovgVrm",
--5x
"movmskpsVrXm$||movmskpdVrXm$","sqrtpsXrm|sqrtssXrm|sqrtpdXrm|sqrtsdXrm",
"rsqrtpsXrm|rsqrtssXrvm","rcppsXrm|rcpssXrvm",
"andpsXrvm||andpdXrvm","andnpsXrvm||andnpdXrvm",
"orpsXrvm||orpdXrvm","xorpsXrvm||xorpdXrvm",
"addpsXrvm|addssXrvm|addpdXrvm|addsdXrvm","mulpsXrvm|mulssXrvm|mulpdXrvm|mulsdXrvm",
"cvtps2pdXrm|cvtss2sdXrvm|cvtpd2psXrm|cvtsd2ssXrvm",
"cvtdq2psXrm|cvttps2dqXrm|cvtps2dqXrm",
"subpsXrvm|subssXrvm|subpdXrvm|subsdXrvm","minpsXrvm|minssXrvm|minpdXrvm|minsdXrvm",
"divpsXrvm|divssXrvm|divpdXrvm|divsdXrvm","maxpsXrvm|maxssXrvm|maxpdXrvm|maxsdXrvm",
--6x
"punpcklbwPrvm","punpcklwdPrvm","punpckldqPrvm","packsswbPrvm",
"pcmpgtbPrvm","pcmpgtwPrvm","pcmpgtdPrvm","packuswbPrvm",
"punpckhbwPrvm","punpckhwdPrvm","punpckhdqPrvm","packssdwPrvm",
"||punpcklqdqXrvm","||punpckhqdqXrvm",
"movPrVSm","movqMrm|movdquXrm|movdqaXrm",
--7x
"pshufwMrmu|pshufhwXrmu|pshufdXrmu|pshuflwXrmu","pshiftw!Pvmu",
"pshiftd!Pvmu","pshiftq!Mvmu||pshiftdq!Xvmu",
"pcmpeqbPrvm","pcmpeqwPrvm","pcmpeqdPrvm","emms*|",
"vmreadUmr||extrqXmuu$|insertqXrmuu$","vmwriteUrm||extrqXrm$|insertqXrm$",
nil,nil,
"||haddpdXrvm|haddpsXrvm","||hsubpdXrvm|hsubpsXrvm",
"movVSmMr|movqXrm|movVSmXr","movqMmr|movdquXmr|movdqaXmr",
--8x
"joVj","jnoVj","jbVj","jnbVj","jzVj","jnzVj","jbeVj","jaVj",
"jsVj","jnsVj","jpeVj","jpoVj","jlVj","jgeVj","jleVj","jgVj",
--9x
"setoBm","setnoBm","setbBm","setnbBm","setzBm","setnzBm","setbeBm","setaBm",
"setsBm","setnsBm","setpeBm","setpoBm","setlBm","setgeBm","setleBm","setgBm",
--Ax
"push fs","pop fs","cpuid","btVmr","shldVmru","shldVmrc",nil,nil,
"push gs","pop gs","rsm","btsVmr","shrdVmru","shrdVmrc","fxsave!Dmp","imulVrm",
--Bx
"cmpxchgBmr","cmpxchgVmr","$lssVrm","btrVmr",
"$lfsVrm","$lgsVrm","movzxVrBmt","movzxVrWmt",
"|popcntVrm","ud2Dp","bt!Vmu","btcVmr",
"bsfVrm","bsrVrm|lzcntVrm|bsrWrm","movsxVrBmt","movsxVrWmt",
--Cx
"xaddBmr","xaddVmr",
"cmppsXrvmu|cmpssXrvmu|cmppdXrvmu|cmpsdXrvmu","$movntiVmr|",
"pinsrwPrvWmu","pextrwDrPmu",
"shufpsXrvmu||shufpdXrvmu","$cmpxchg!Qmp",
"bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR","bswapVR",
--Dx
"||addsubpdXrvm|addsubpsXrvm","psrlwPrvm","psrldPrvm","psrlqPrvm",
"paddqPrvm","pmullwPrvm",
"|movq2dqXrMm|movqXmr|movdq2qMrXm$","pmovmskbVrMm||pmovmskbVrXm",
"psubusbPrvm","psubuswPrvm","pminubPrvm","pandPrvm",
"paddusbPrvm","padduswPrvm","pmaxubPrvm","pandnPrvm",
--Ex
"pavgbPrvm","psrawPrvm","psradPrvm","pavgwPrvm",
"pmulhuwPrvm","pmulhwPrvm",
"|cvtdq2pdXrm|cvttpd2dqXrm|cvtpd2dqXrm","$movntqMmr||$movntdqXmr",
"psubsbPrvm","psubswPrvm","pminswPrvm","porPrvm",
"paddsbPrvm","paddswPrvm","pmaxswPrvm","pxorPrvm",
--Fx
"|||lddquXrm","psllwPrvm","pslldPrvm","psllqPrvm",
"pmuludqPrvm","pmaddwdPrvm","psadbwPrvm","maskmovqMrm||maskmovdquXrm$",
"psubbPrvm","psubwPrvm","psubdPrvm","psubqPrvm",
"paddbPrvm","paddwPrvm","padddPrvm","ud",
}
assert(map_opc2[255] == "ud")
-- Map for three-byte opcodes. Can't wait for their next invention.
local map_opc3 = {
["38"] = { -- [66] 0f 38 xx
--0x
[0]="pshufbPrvm","phaddwPrvm","phadddPrvm","phaddswPrvm",
"pmaddubswPrvm","phsubwPrvm","phsubdPrvm","phsubswPrvm",
"psignbPrvm","psignwPrvm","psigndPrvm","pmulhrswPrvm",
"||permilpsXrvm","||permilpdXrvm",nil,nil,
--1x
"||pblendvbXrma",nil,nil,nil,
"||blendvpsXrma","||blendvpdXrma","||permpsXrvm","||ptestXrm",
"||broadcastssXrm","||broadcastsdXrm","||broadcastf128XrlXm",nil,
"pabsbPrm","pabswPrm","pabsdPrm",nil,
--2x
"||pmovsxbwXrm","||pmovsxbdXrm","||pmovsxbqXrm","||pmovsxwdXrm",
"||pmovsxwqXrm","||pmovsxdqXrm",nil,nil,
"||pmuldqXrvm","||pcmpeqqXrvm","||$movntdqaXrm","||packusdwXrvm",
"||maskmovpsXrvm","||maskmovpdXrvm","||maskmovpsXmvr","||maskmovpdXmvr",
--3x
"||pmovzxbwXrm","||pmovzxbdXrm","||pmovzxbqXrm","||pmovzxwdXrm",
"||pmovzxwqXrm","||pmovzxdqXrm","||permdXrvm","||pcmpgtqXrvm",
"||pminsbXrvm","||pminsdXrvm","||pminuwXrvm","||pminudXrvm",
"||pmaxsbXrvm","||pmaxsdXrvm","||pmaxuwXrvm","||pmaxudXrvm",
--4x
"||pmulddXrvm","||phminposuwXrm",nil,nil,
nil,"||psrlvVSXrvm","||psravdXrvm","||psllvVSXrvm",
--5x
[0x58] = "||pbroadcastdXrlXm",[0x59] = "||pbroadcastqXrlXm",
[0x5a] = "||broadcasti128XrlXm",
--7x
[0x78] = "||pbroadcastbXrlXm",[0x79] = "||pbroadcastwXrlXm",
--8x
[0x8c] = "||pmaskmovXrvVSm",
[0x8e] = "||pmaskmovVSmXvr",
--9x
[0x96] = "||fmaddsub132pHXrvm",[0x97] = "||fmsubadd132pHXrvm",
[0x98] = "||fmadd132pHXrvm",[0x99] = "||fmadd132sHXrvm",
[0x9a] = "||fmsub132pHXrvm",[0x9b] = "||fmsub132sHXrvm",
[0x9c] = "||fnmadd132pHXrvm",[0x9d] = "||fnmadd132sHXrvm",
[0x9e] = "||fnmsub132pHXrvm",[0x9f] = "||fnmsub132sHXrvm",
--Ax
[0xa6] = "||fmaddsub213pHXrvm",[0xa7] = "||fmsubadd213pHXrvm",
[0xa8] = "||fmadd213pHXrvm",[0xa9] = "||fmadd213sHXrvm",
[0xaa] = "||fmsub213pHXrvm",[0xab] = "||fmsub213sHXrvm",
[0xac] = "||fnmadd213pHXrvm",[0xad] = "||fnmadd213sHXrvm",
[0xae] = "||fnmsub213pHXrvm",[0xaf] = "||fnmsub213sHXrvm",
--Bx
[0xb6] = "||fmaddsub231pHXrvm",[0xb7] = "||fmsubadd231pHXrvm",
[0xb8] = "||fmadd231pHXrvm",[0xb9] = "||fmadd231sHXrvm",
[0xba] = "||fmsub231pHXrvm",[0xbb] = "||fmsub231sHXrvm",
[0xbc] = "||fnmadd231pHXrvm",[0xbd] = "||fnmadd231sHXrvm",
[0xbe] = "||fnmsub231pHXrvm",[0xbf] = "||fnmsub231sHXrvm",
--Dx
[0xdc] = "||aesencXrvm", [0xdd] = "||aesenclastXrvm",
[0xde] = "||aesdecXrvm", [0xdf] = "||aesdeclastXrvm",
--Fx
[0xf0] = "|||crc32TrBmt",[0xf1] = "|||crc32TrVmt",
[0xf7] = "| sarxVrmv| shlxVrmv| shrxVrmv",
},
["3a"] = { -- [66] 0f 3a xx
--0x
[0x00]="||permqXrmu","||permpdXrmu","||pblenddXrvmu",nil,
"||permilpsXrmu","||permilpdXrmu","||perm2f128Xrvmu",nil,
"||roundpsXrmu","||roundpdXrmu","||roundssXrvmu","||roundsdXrvmu",
"||blendpsXrvmu","||blendpdXrvmu","||pblendwXrvmu","palignrPrvmu",
--1x
nil,nil,nil,nil,
"||pextrbVmXru","||pextrwVmXru","||pextrVmSXru","||extractpsVmXru",
"||insertf128XrvlXmu","||extractf128XlXmYru",nil,nil,
nil,nil,nil,nil,
--2x
"||pinsrbXrvVmu","||insertpsXrvmu","||pinsrXrvVmuS",nil,
--3x
[0x38] = "||inserti128Xrvmu",[0x39] = "||extracti128XlXmYru",
--4x
[0x40] = "||dppsXrvmu",
[0x41] = "||dppdXrvmu",
[0x42] = "||mpsadbwXrvmu",
[0x44] = "||pclmulqdqXrvmu",
[0x46] = "||perm2i128Xrvmu",
[0x4a] = "||blendvpsXrvmb",[0x4b] = "||blendvpdXrvmb",
[0x4c] = "||pblendvbXrvmb",
--6x
[0x60] = "||pcmpestrmXrmu",[0x61] = "||pcmpestriXrmu",
[0x62] = "||pcmpistrmXrmu",[0x63] = "||pcmpistriXrmu",
[0xdf] = "||aeskeygenassistXrmu",
--Fx
[0xf0] = "||| rorxVrmu",
},
}
-- Map for VMX/SVM opcodes 0F 01 C0-FF (sgdt group with register operands).
local map_opcvm = {
[0xc1]="vmcall",[0xc2]="vmlaunch",[0xc3]="vmresume",[0xc4]="vmxoff",
[0xc8]="monitor",[0xc9]="mwait",
[0xd8]="vmrun",[0xd9]="vmmcall",[0xda]="vmload",[0xdb]="vmsave",
[0xdc]="stgi",[0xdd]="clgi",[0xde]="skinit",[0xdf]="invlpga",
[0xf8]="swapgs",[0xf9]="rdtscp",
}
-- Map for FP opcodes. And you thought stack machines are simple?
local map_opcfp = {
-- D8-DF 00-BF: opcodes with a memory operand.
-- D8
[0]="faddFm","fmulFm","fcomFm","fcompFm","fsubFm","fsubrFm","fdivFm","fdivrFm",
"fldFm",nil,"fstFm","fstpFm","fldenvVm","fldcwWm","fnstenvVm","fnstcwWm",
-- DA
"fiaddDm","fimulDm","ficomDm","ficompDm",
"fisubDm","fisubrDm","fidivDm","fidivrDm",
-- DB
"fildDm","fisttpDm","fistDm","fistpDm",nil,"fld twordFmp",nil,"fstp twordFmp",
-- DC
"faddGm","fmulGm","fcomGm","fcompGm","fsubGm","fsubrGm","fdivGm","fdivrGm",
-- DD
"fldGm","fisttpQm","fstGm","fstpGm","frstorDmp",nil,"fnsaveDmp","fnstswWm",
-- DE
"fiaddWm","fimulWm","ficomWm","ficompWm",
"fisubWm","fisubrWm","fidivWm","fidivrWm",
-- DF
"fildWm","fisttpWm","fistWm","fistpWm",
"fbld twordFmp","fildQm","fbstp twordFmp","fistpQm",
-- xx C0-FF: opcodes with a pseudo-register operand.
-- D8
"faddFf","fmulFf","fcomFf","fcompFf","fsubFf","fsubrFf","fdivFf","fdivrFf",
-- D9
"fldFf","fxchFf",{"fnop"},nil,
{"fchs","fabs",nil,nil,"ftst","fxam"},
{"fld1","fldl2t","fldl2e","fldpi","fldlg2","fldln2","fldz"},
{"f2xm1","fyl2x","fptan","fpatan","fxtract","fprem1","fdecstp","fincstp"},
{"fprem","fyl2xp1","fsqrt","fsincos","frndint","fscale","fsin","fcos"},
-- DA
"fcmovbFf","fcmoveFf","fcmovbeFf","fcmovuFf",nil,{nil,"fucompp"},nil,nil,
-- DB
"fcmovnbFf","fcmovneFf","fcmovnbeFf","fcmovnuFf",
{nil,nil,"fnclex","fninit"},"fucomiFf","fcomiFf",nil,
-- DC
"fadd toFf","fmul toFf",nil,nil,
"fsub toFf","fsubr toFf","fdivr toFf","fdiv toFf",
-- DD
"ffreeFf",nil,"fstFf","fstpFf","fucomFf","fucompFf",nil,nil,
-- DE
"faddpFf","fmulpFf",nil,{nil,"fcompp"},
"fsubrpFf","fsubpFf","fdivrpFf","fdivpFf",
-- DF
nil,nil,nil,nil,{"fnstsw ax"},"fucomipFf","fcomipFf",nil,
}
assert(map_opcfp[126] == "fcomipFf")
-- Map for opcode groups. The subkey is sp from the ModRM byte.
local map_opcgroup = {
arith = { "add", "or", "adc", "sbb", "and", "sub", "xor", "cmp" },
shift = { "rol", "ror", "rcl", "rcr", "shl", "shr", "sal", "sar" },
testb = { "testBmi", "testBmi", "not", "neg", "mul", "imul", "div", "idiv" },
testv = { "testVmi", "testVmi", "not", "neg", "mul", "imul", "div", "idiv" },
incb = { "inc", "dec" },
incd = { "inc", "dec", "callUmp", "$call farDmp",
"jmpUmp", "$jmp farDmp", "pushUm" },
sldt = { "sldt", "str", "lldt", "ltr", "verr", "verw" },
sgdt = { "vm*$sgdt", "vm*$sidt", "$lgdt", "vm*$lidt",
"smsw", nil, "lmsw", "vm*$invlpg" },
bt = { nil, nil, nil, nil, "bt", "bts", "btr", "btc" },
cmpxchg = { nil, "sz*,cmpxchg8bQmp,cmpxchg16bXmp", nil, nil,
nil, nil, "vmptrld|vmxon|vmclear", "vmptrst" },
pshiftw = { nil, nil, "psrlw", nil, "psraw", nil, "psllw" },
pshiftd = { nil, nil, "psrld", nil, "psrad", nil, "pslld" },
pshiftq = { nil, nil, "psrlq", nil, nil, nil, "psllq" },
pshiftdq = { nil, nil, "psrlq", "psrldq", nil, nil, "psllq", "pslldq" },
fxsave = { "$fxsave", "$fxrstor", "$ldmxcsr", "$stmxcsr",
nil, "lfenceDp$", "mfenceDp$", "sfenceDp$clflush" },
prefetch = { "prefetch", "prefetchw" },
prefetcht = { "prefetchnta", "prefetcht0", "prefetcht1", "prefetcht2" },
}
------------------------------------------------------------------------------
-- Maps for register names.
local map_regs = {
B = { "al", "cl", "dl", "bl", "ah", "ch", "dh", "bh",
"r8b", "r9b", "r10b", "r11b", "r12b", "r13b", "r14b", "r15b" },
B64 = { "al", "cl", "dl", "bl", "spl", "bpl", "sil", "dil",
"r8b", "r9b", "r10b", "r11b", "r12b", "r13b", "r14b", "r15b" },
W = { "ax", "cx", "dx", "bx", "sp", "bp", "si", "di",
"r8w", "r9w", "r10w", "r11w", "r12w", "r13w", "r14w", "r15w" },
D = { "eax", "ecx", "edx", "ebx", "esp", "ebp", "esi", "edi",
"r8d", "r9d", "r10d", "r11d", "r12d", "r13d", "r14d", "r15d" },
Q = { "rax", "rcx", "rdx", "rbx", "rsp", "rbp", "rsi", "rdi",
"r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15" },
M = { "mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7",
"mm0", "mm1", "mm2", "mm3", "mm4", "mm5", "mm6", "mm7" }, -- No x64 ext!
X = { "xmm0", "xmm1", "xmm2", "xmm3", "xmm4", "xmm5", "xmm6", "xmm7",
"xmm8", "xmm9", "xmm10", "xmm11", "xmm12", "xmm13", "xmm14", "xmm15" },
Y = { "ymm0", "ymm1", "ymm2", "ymm3", "ymm4", "ymm5", "ymm6", "ymm7",
"ymm8", "ymm9", "ymm10", "ymm11", "ymm12", "ymm13", "ymm14", "ymm15" },
}
local map_segregs = { "es", "cs", "ss", "ds", "fs", "gs", "segr6", "segr7" }
-- Maps for size names.
local map_sz2n = {
B = 1, W = 2, D = 4, Q = 8, M = 8, X = 16, Y = 32,
}
local map_sz2prefix = {
B = "byte", W = "word", D = "dword",
Q = "qword",
M = "qword", X = "xword", Y = "yword",
F = "dword", G = "qword", -- No need for sizes/register names for these two.
}
------------------------------------------------------------------------------
-- Output a nicely formatted line with an opcode and operands.
local function putop(ctx, text, operands)
local code, pos, hex = ctx.code, ctx.pos, ""
local hmax = ctx.hexdump
if hmax > 0 then
for i=ctx.start,pos-1 do
hex = hex..format("%02X", byte(code, i, i))
end
if #hex > hmax then hex = sub(hex, 1, hmax)..". "
else hex = hex..rep(" ", hmax-#hex+2) end
end
if operands then text = text.." "..operands end
if ctx.o16 then text = "o16 "..text; ctx.o16 = false end
if ctx.a32 then text = "a32 "..text; ctx.a32 = false end
if ctx.rep then text = ctx.rep.." "..text; ctx.rep = false end
if ctx.rex then
local t = (ctx.rexw and "w" or "")..(ctx.rexr and "r" or "")..
(ctx.rexx and "x" or "")..(ctx.rexb and "b" or "")..
(ctx.vexl and "l" or "")
if ctx.vexv and ctx.vexv ~= 0 then t = t.."v"..ctx.vexv end
if t ~= "" then text = ctx.rex.."."..t.." "..gsub(text, "^ ", "")
elseif ctx.rex == "vex" then text = gsub("v"..text, "^v ", "") end
ctx.rexw = false; ctx.rexr = false; ctx.rexx = false; ctx.rexb = false
ctx.rex = false; ctx.vexl = false; ctx.vexv = false
end
if ctx.seg then
local text2, n = gsub(text, "%[", "["..ctx.seg..":")
if n == 0 then text = ctx.seg.." "..text else text = text2 end
ctx.seg = false
end
if ctx.lock then text = "lock "..text; ctx.lock = false end
local imm = ctx.imm
if imm then
local sym = ctx.symtab[imm]
if sym then text = text.."\t->"..sym end
end
ctx.out(format("%08x %s%s\n", ctx.addr+ctx.start, hex, text))
ctx.mrm = false
ctx.vexv = false
ctx.start = pos
ctx.imm = nil
end
-- Clear all prefix flags.
local function clearprefixes(ctx)
ctx.o16 = false; ctx.seg = false; ctx.lock = false; ctx.rep = false
ctx.rexw = false; ctx.rexr = false; ctx.rexx = false; ctx.rexb = false
ctx.rex = false; ctx.a32 = false; ctx.vexl = false
end
-- Fallback for incomplete opcodes at the end.
local function incomplete(ctx)
ctx.pos = ctx.stop+1
clearprefixes(ctx)
return putop(ctx, "(incomplete)")
end
-- Fallback for unknown opcodes.
local function unknown(ctx)
clearprefixes(ctx)
return putop(ctx, "(unknown)")
end
-- Return an immediate of the specified size.
local function getimm(ctx, pos, n)
if pos+n-1 > ctx.stop then return incomplete(ctx) end
local code = ctx.code
if n == 1 then
local b1 = byte(code, pos, pos)
return b1
elseif n == 2 then
local b1, b2 = byte(code, pos, pos+1)
return b1+b2*256
else
local b1, b2, b3, b4 = byte(code, pos, pos+3)
local imm = b1+b2*256+b3*65536+b4*16777216
ctx.imm = imm
return imm
end
end
-- Process pattern string and generate the operands.
local function putpat(ctx, name, pat)
local operands, regs, sz, mode, sp, rm, sc, rx, sdisp
local code, pos, stop, vexl = ctx.code, ctx.pos, ctx.stop, ctx.vexl
-- Chars used: 1DFGHIMPQRSTUVWXYabcdfgijlmoprstuvwxyz
for p in gmatch(pat, ".") do
local x = nil
if p == "V" or p == "U" then
if ctx.rexw then sz = "Q"; ctx.rexw = false
elseif ctx.o16 then sz = "W"; ctx.o16 = false
elseif p == "U" and ctx.x64 then sz = "Q"
else sz = "D" end
regs = map_regs[sz]
elseif p == "T" then
if ctx.rexw then sz = "Q"; ctx.rexw = false else sz = "D" end
regs = map_regs[sz]
elseif p == "B" then
sz = "B"
regs = ctx.rex and map_regs.B64 or map_regs.B
elseif match(p, "[WDQMXYFG]") then
sz = p
if sz == "X" and vexl then sz = "Y"; ctx.vexl = false end
regs = map_regs[sz]
elseif p == "P" then
sz = ctx.o16 and "X" or "M"; ctx.o16 = false
if sz == "X" and vexl then sz = "Y"; ctx.vexl = false end
regs = map_regs[sz]
elseif p == "H" then
name = name..(ctx.rexw and "d" or "s")
ctx.rexw = false
elseif p == "S" then
name = name..lower(sz)
elseif p == "s" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = imm <= 127 and format("+0x%02x", imm)
or format("-0x%02x", 256-imm)
pos = pos+1
elseif p == "u" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = format("0x%02x", imm)
pos = pos+1
elseif p == "b" then
local imm = getimm(ctx, pos, 1); if not imm then return end
x = regs[imm/16+1]
pos = pos+1
elseif p == "w" then
local imm = getimm(ctx, pos, 2); if not imm then return end
x = format("0x%x", imm)
pos = pos+2
elseif p == "o" then -- [offset]
if ctx.x64 then
local imm1 = getimm(ctx, pos, 4); if not imm1 then return end
local imm2 = getimm(ctx, pos+4, 4); if not imm2 then return end
x = format("[0x%08x%08x]", imm2, imm1)
pos = pos+8
else
local imm = getimm(ctx, pos, 4); if not imm then return end
x = format("[0x%08x]", imm)
pos = pos+4
end
elseif p == "i" or p == "I" then
local n = map_sz2n[sz]
if n == 8 and ctx.x64 and p == "I" then
local imm1 = getimm(ctx, pos, 4); if not imm1 then return end
local imm2 = getimm(ctx, pos+4, 4); if not imm2 then return end
x = format("0x%08x%08x", imm2, imm1)
else
if n == 8 then n = 4 end
local imm = getimm(ctx, pos, n); if not imm then return end
if sz == "Q" and (imm < 0 or imm > 0x7fffffff) then
imm = (0xffffffff+1)-imm
x = format(imm > 65535 and "-0x%08x" or "-0x%x", imm)
else
x = format(imm > 65535 and "0x%08x" or "0x%x", imm)
end
end
pos = pos+n
elseif p == "j" then
local n = map_sz2n[sz]
if n == 8 then n = 4 end
local imm = getimm(ctx, pos, n); if not imm then return end
if sz == "B" and imm > 127 then imm = imm-256
elseif imm > 2147483647 then imm = imm-4294967296 end
pos = pos+n
imm = imm + pos + ctx.addr
if imm > 4294967295 and not ctx.x64 then imm = imm-4294967296 end
ctx.imm = imm
if sz == "W" then
x = format("word 0x%04x", imm%65536)
elseif ctx.x64 then
local lo = imm % 0x1000000
x = format("0x%02x%06x", (imm-lo) / 0x1000000, lo)
else
x = "0x"..tohex(imm)
end
elseif p == "R" then
local r = byte(code, pos-1, pos-1)%8
if ctx.rexb then r = r + 8; ctx.rexb = false end
x = regs[r+1]
elseif p == "a" then x = regs[1]
elseif p == "c" then x = "cl"
elseif p == "d" then x = "dx"
elseif p == "1" then x = "1"
else
if not mode then
mode = ctx.mrm
if not mode then
if pos > stop then return incomplete(ctx) end
mode = byte(code, pos, pos)
pos = pos+1
end
rm = mode%8; mode = (mode-rm)/8
sp = mode%8; mode = (mode-sp)/8
sdisp = ""
if mode < 3 then
if rm == 4 then
if pos > stop then return incomplete(ctx) end
sc = byte(code, pos, pos)
pos = pos+1
rm = sc%8; sc = (sc-rm)/8
rx = sc%8; sc = (sc-rx)/8
if ctx.rexx then rx = rx + 8; ctx.rexx = false end
if rx == 4 then rx = nil end
end
if mode > 0 or rm == 5 then
local dsz = mode
if dsz ~= 1 then dsz = 4 end
local disp = getimm(ctx, pos, dsz); if not disp then return end
if mode == 0 then rm = nil end
if rm or rx or (not sc and ctx.x64 and not ctx.a32) then
if dsz == 1 and disp > 127 then
sdisp = format("-0x%x", 256-disp)
elseif disp >= 0 and disp <= 0x7fffffff then
sdisp = format("+0x%x", disp)
else
sdisp = format("-0x%x", (0xffffffff+1)-disp)
end
else
sdisp = format(ctx.x64 and not ctx.a32 and
not (disp >= 0 and disp <= 0x7fffffff)
and "0xffffffff%08x" or "0x%08x", disp)
end
pos = pos+dsz
end
end
if rm and ctx.rexb then rm = rm + 8; ctx.rexb = false end
if ctx.rexr then sp = sp + 8; ctx.rexr = false end
end
if p == "m" then
if mode == 3 then x = regs[rm+1]
else
local aregs = ctx.a32 and map_regs.D or ctx.aregs
local srm, srx = "", ""
if rm then srm = aregs[rm+1]
elseif not sc and ctx.x64 and not ctx.a32 then srm = "rip" end
ctx.a32 = false
if rx then
if rm then srm = srm.."+" end
srx = aregs[rx+1]
if sc > 0 then srx = srx.."*"..(2^sc) end
end
x = format("[%s%s%s]", srm, srx, sdisp)
end
if mode < 3 and
(not match(pat, "[aRrgp]") or match(pat, "t")) then -- Yuck.
x = map_sz2prefix[sz].." "..x
end
elseif p == "r" then x = regs[sp+1]
elseif p == "g" then x = map_segregs[sp+1]
elseif p == "p" then -- Suppress prefix.
elseif p == "f" then x = "st"..rm
elseif p == "x" then
if sp == 0 and ctx.lock and not ctx.x64 then
x = "CR8"; ctx.lock = false
else
x = "CR"..sp
end
elseif p == "v" then
if ctx.vexv then
x = regs[ctx.vexv+1]; ctx.vexv = false
end
elseif p == "y" then x = "DR"..sp
elseif p == "z" then x = "TR"..sp
elseif p == "l" then vexl = false
elseif p == "t" then
else
error("bad pattern `"..pat.."'")
end
end
if x then operands = operands and operands..", "..x or x end
end
ctx.pos = pos
return putop(ctx, name, operands)
end
-- Forward declaration.
local map_act
-- Fetch and cache MRM byte.
local function getmrm(ctx)
local mrm = ctx.mrm
if not mrm then
local pos = ctx.pos
if pos > ctx.stop then return nil end
mrm = byte(ctx.code, pos, pos)
ctx.pos = pos+1
ctx.mrm = mrm
end
return mrm
end
-- Dispatch to handler depending on pattern.
local function dispatch(ctx, opat, patgrp)
if not opat then return unknown(ctx) end
if match(opat, "%|") then -- MMX/SSE variants depending on prefix.
local p
if ctx.rep then
p = ctx.rep=="rep" and "%|([^%|]*)" or "%|[^%|]*%|[^%|]*%|([^%|]*)"
ctx.rep = false
elseif ctx.o16 then p = "%|[^%|]*%|([^%|]*)"; ctx.o16 = false
else p = "^[^%|]*" end
opat = match(opat, p)
if not opat then return unknown(ctx) end
-- ctx.rep = false; ctx.o16 = false
--XXX fails for 66 f2 0f 38 f1 06 crc32 eax,WORD PTR [esi]
--XXX remove in branches?
end
if match(opat, "%$") then -- reg$mem variants.
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
opat = match(opat, mrm >= 192 and "^[^%$]*" or "%$(.*)")
if opat == "" then return unknown(ctx) end
end
if opat == "" then return unknown(ctx) end
local name, pat = match(opat, "^([a-z0-9 ]*)(.*)")
if pat == "" and patgrp then pat = patgrp end
return map_act[sub(pat, 1, 1)](ctx, name, pat)
end
-- Get a pattern from an opcode map and dispatch to handler.
local function dispatchmap(ctx, opcmap)
local pos = ctx.pos
local opat = opcmap[byte(ctx.code, pos, pos)]
pos = pos + 1
ctx.pos = pos
return dispatch(ctx, opat)
end
-- Map for action codes. The key is the first char after the name.
map_act = {
-- Simple opcodes without operands.
[""] = function(ctx, name, pat)
return putop(ctx, name)
end,
-- Operand size chars fall right through.
B = putpat, W = putpat, D = putpat, Q = putpat,
V = putpat, U = putpat, T = putpat,
M = putpat, X = putpat, P = putpat,
F = putpat, G = putpat, Y = putpat,
H = putpat,
-- Collect prefixes.
[":"] = function(ctx, name, pat)
ctx[pat == ":" and name or sub(pat, 2)] = name
if ctx.pos - ctx.start > 5 then return unknown(ctx) end -- Limit #prefixes.
end,
-- Chain to special handler specified by name.
["*"] = function(ctx, name, pat)
return map_act[name](ctx, name, sub(pat, 2))
end,
-- Use named subtable for opcode group.
["!"] = function(ctx, name, pat)
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
return dispatch(ctx, map_opcgroup[name][((mrm-(mrm%8))/8)%8+1], sub(pat, 2))
end,
-- o16,o32[,o64] variants.
sz = function(ctx, name, pat)
if ctx.o16 then ctx.o16 = false
else
pat = match(pat, ",(.*)")
if ctx.rexw then
local p = match(pat, ",(.*)")
if p then pat = p; ctx.rexw = false end
end
end
pat = match(pat, "^[^,]*")
return dispatch(ctx, pat)
end,
-- Two-byte opcode dispatch.
opc2 = function(ctx, name, pat)
return dispatchmap(ctx, map_opc2)
end,
-- Three-byte opcode dispatch.
opc3 = function(ctx, name, pat)
return dispatchmap(ctx, map_opc3[pat])
end,
-- VMX/SVM dispatch.
vm = function(ctx, name, pat)
return dispatch(ctx, map_opcvm[ctx.mrm])
end,
-- Floating point opcode dispatch.
fp = function(ctx, name, pat)
local mrm = getmrm(ctx); if not mrm then return incomplete(ctx) end
local rm = mrm%8
local idx = pat*8 + ((mrm-rm)/8)%8
if mrm >= 192 then idx = idx + 64 end
local opat = map_opcfp[idx]
if type(opat) == "table" then opat = opat[rm+1] end
return dispatch(ctx, opat)
end,
-- REX prefix.
rex = function(ctx, name, pat)
if ctx.rex then return unknown(ctx) end -- Only 1 REX or VEX prefix allowed.
for p in gmatch(pat, ".") do ctx["rex"..p] = true end
ctx.rex = "rex"
end,
-- VEX prefix.
vex = function(ctx, name, pat)
if ctx.rex then return unknown(ctx) end -- Only 1 REX or VEX prefix allowed.
ctx.rex = "vex"
local pos = ctx.pos
if ctx.mrm then
ctx.mrm = nil
pos = pos-1
end
local b = byte(ctx.code, pos, pos)
if not b then return incomplete(ctx) end
pos = pos+1
if b < 128 then ctx.rexr = true end
local m = 1
if pat == "3" then
m = b%32; b = (b-m)/32
local nb = b%2; b = (b-nb)/2
if nb == 0 then ctx.rexb = true end
local nx = b%2
if nx == 0 then ctx.rexx = true end
b = byte(ctx.code, pos, pos)
if not b then return incomplete(ctx) end
pos = pos+1
if b >= 128 then ctx.rexw = true end
end
ctx.pos = pos
local map
if m == 1 then map = map_opc2
elseif m == 2 then map = map_opc3["38"]
elseif m == 3 then map = map_opc3["3a"]
else return unknown(ctx) end
local p = b%4; b = (b-p)/4
if p == 1 then ctx.o16 = "o16"
elseif p == 2 then ctx.rep = "rep"
elseif p == 3 then ctx.rep = "repne" end
local l = b%2; b = (b-l)/2
if l ~= 0 then ctx.vexl = true end
ctx.vexv = (-1-b)%16
return dispatchmap(ctx, map)
end,
-- Special case for nop with REX prefix.
nop = function(ctx, name, pat)
return dispatch(ctx, ctx.rex and pat or "nop")
end,
-- Special case for 0F 77.
emms = function(ctx, name, pat)
if ctx.rex ~= "vex" then
return putop(ctx, "emms")
elseif ctx.vexl then
ctx.vexl = false
return putop(ctx, "zeroall")
else
return putop(ctx, "zeroupper")
end
end,
}
------------------------------------------------------------------------------
-- Disassemble a block of code.
local function disass_block(ctx, ofs, len)
if not ofs then ofs = 0 end
local stop = len and ofs+len or #ctx.code
ofs = ofs + 1
ctx.start = ofs
ctx.pos = ofs
ctx.stop = stop
ctx.imm = nil
ctx.mrm = false
clearprefixes(ctx)
while ctx.pos <= stop do dispatchmap(ctx, ctx.map1) end
if ctx.pos ~= ctx.start then incomplete(ctx) end
end
-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
local function create(code, addr, out)
local ctx = {}
ctx.code = code
ctx.addr = (addr or 0) - 1
ctx.out = out or io.write
ctx.symtab = {}
ctx.disass = disass_block
ctx.hexdump = 16
ctx.x64 = false
ctx.map1 = map_opc1_32
ctx.aregs = map_regs.D
return ctx
end
local function create64(code, addr, out)
local ctx = create(code, addr, out)
ctx.x64 = true
ctx.map1 = map_opc1_64
ctx.aregs = map_regs.Q
return ctx
end
-- Simple API: disassemble code (a string) at address and output via out.
local function disass(code, addr, out)
create(code, addr, out):disass()
end
local function disass64(code, addr, out)
create64(code, addr, out):disass()
end
-- Return register name for RID.
local function regname(r)
if r < 8 then return map_regs.D[r+1] end
return map_regs.X[r-7]
end
local function regname64(r)
if r < 16 then return map_regs.Q[r+1] end
return map_regs.X[r-15]
end
-- Public module functions.
return {
create = create,
create64 = create64,
disass = disass,
disass64 = disass64,
regname = regname,
regname64 = regname64
}
@@ -0,0 +1,726 @@
----------------------------------------------------------------------------
-- LuaJIT compiler dump module.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module can be used to debug the JIT compiler itself. It dumps the
-- code representations and structures used in various compiler stages.
--
-- Example usage:
--
-- luajit -jdump -e "local x=0; for i=1,1e6 do x=x+i end; print(x)"
-- luajit -jdump=im -e "for i=1,1000 do for j=1,1000 do end end" | less -R
-- luajit -jdump=is myapp.lua | less -R
-- luajit -jdump=-b myapp.lua
-- luajit -jdump=+aH,myapp.html myapp.lua
-- luajit -jdump=ixT,myapp.dump myapp.lua
--
-- The first argument specifies the dump mode. The second argument gives
-- the output file name. Default output is to stdout, unless the environment
-- variable LUAJIT_DUMPFILE is set. The file is overwritten every time the
-- module is started.
--
-- Different features can be turned on or off with the dump mode. If the
-- mode starts with a '+', the following features are added to the default
-- set of features; a '-' removes them. Otherwise the features are replaced.
--
-- The following dump features are available (* marks the default):
--
-- * t Print a line for each started, ended or aborted trace (see also -jv).
-- * b Dump the traced bytecode.
-- * i Dump the IR (intermediate representation).
-- r Augment the IR with register/stack slots.
-- s Dump the snapshot map.
-- * m Dump the generated machine code.
-- x Print each taken trace exit.
-- X Print each taken trace exit and the contents of all registers.
-- a Print the IR of aborted traces, too.
--
-- The output format can be set with the following characters:
--
-- T Plain text output.
-- A ANSI-colored text output
-- H Colorized HTML + CSS output.
--
-- The default output format is plain text. It's set to ANSI-colored text
-- if the COLORTERM variable is set. Note: this is independent of any output
-- redirection, which is actually considered a feature.
--
-- You probably want to use less -R to enjoy viewing ANSI-colored text from
-- a pipe or a file. Add this to your ~/.bashrc: export LESS="-R"
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local funcinfo, funcbc = jutil.funcinfo, jutil.funcbc
local traceinfo, traceir, tracek = jutil.traceinfo, jutil.traceir, jutil.tracek
local tracemc, tracesnap = jutil.tracemc, jutil.tracesnap
local traceexitstub, ircalladdr = jutil.traceexitstub, jutil.ircalladdr
local bit = require("bit")
local band, shr, tohex = bit.band, bit.rshift, bit.tohex
local sub, gsub, format = string.sub, string.gsub, string.format
local byte, rep = string.byte, string.rep
local type, tostring = type, tostring
local stdout, stderr = io.stdout, io.stderr
-- Load other modules on-demand.
local bcline, disass
-- Active flag, output file handle and dump mode.
local active, out, dumpmode
------------------------------------------------------------------------------
local symtabmt = { __index = false }
local symtab = {}
local nexitsym = 0
-- Fill nested symbol table with per-trace exit stub addresses.
local function fillsymtab_tr(tr, nexit)
local t = {}
symtabmt.__index = t
if jit.arch:sub(1, 4) == "mips" then
t[traceexitstub(tr, 0)] = "exit"
return
end
for i=0,nexit-1 do
local addr = traceexitstub(tr, i)
if addr < 0 then addr = addr + 2^32 end
t[addr] = tostring(i)
end
local addr = traceexitstub(tr, nexit)
if addr then t[addr] = "stack_check" end
end
-- Fill symbol table with trace exit stub addresses.
local function fillsymtab(tr, nexit)
local t = symtab
if nexitsym == 0 then
local maskaddr = jit.arch == "arm" and -2
local ircall = vmdef.ircall
for i=0,#ircall do
local addr = ircalladdr(i)
if addr ~= 0 then
if maskaddr then addr = band(addr, maskaddr) end
if addr < 0 then addr = addr + 2^32 end
t[addr] = ircall[i]
end
end
end
if nexitsym == 1000000 then -- Per-trace exit stubs.
fillsymtab_tr(tr, nexit)
elseif nexit > nexitsym then -- Shared exit stubs.
for i=nexitsym,nexit-1 do
local addr = traceexitstub(i)
if addr == nil then -- Fall back to per-trace exit stubs.
fillsymtab_tr(tr, nexit)
setmetatable(symtab, symtabmt)
nexit = 1000000
break
end
if addr < 0 then addr = addr + 2^32 end
t[addr] = tostring(i)
end
nexitsym = nexit
end
return t
end
local function dumpwrite(s)
out:write(s)
end
-- Disassemble machine code.
local function dump_mcode(tr)
local info = traceinfo(tr)
if not info then return end
local mcode, addr, loop = tracemc(tr)
if not mcode then return end
if not disass then disass = require("jit.dis_"..jit.arch) end
if addr < 0 then addr = addr + 2^32 end
out:write("---- TRACE ", tr, " mcode ", #mcode, "\n")
local ctx = disass.create(mcode, addr, dumpwrite)
ctx.hexdump = 0
ctx.symtab = fillsymtab(tr, info.nexit)
if loop ~= 0 then
symtab[addr+loop] = "LOOP"
ctx:disass(0, loop)
out:write("->LOOP:\n")
ctx:disass(loop, #mcode-loop)
symtab[addr+loop] = nil
else
ctx:disass(0, #mcode)
end
end
------------------------------------------------------------------------------
local irtype_text = {
[0] = "nil",
"fal",
"tru",
"lud",
"str",
"p32",
"thr",
"pro",
"fun",
"p64",
"cdt",
"tab",
"udt",
"flt",
"num",
"i8 ",
"u8 ",
"i16",
"u16",
"int",
"u32",
"i64",
"u64",
"sfp",
}
local colortype_ansi = {
[0] = "%s",
"%s",
"%s",
"\027[36m%s\027[m",
"\027[32m%s\027[m",
"%s",
"\027[1m%s\027[m",
"%s",
"\027[1m%s\027[m",
"%s",
"\027[33m%s\027[m",
"\027[31m%s\027[m",
"\027[36m%s\027[m",
"\027[34m%s\027[m",
"\027[34m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
"\027[35m%s\027[m",
}
local function colorize_text(s)
return s
end
local function colorize_ansi(s, t, extra)
local out = format(colortype_ansi[t], s)
if extra then out = "\027[3m"..out end
return out
end
local irtype_ansi = setmetatable({},
{ __index = function(tab, t)
local s = colorize_ansi(irtype_text[t], t); tab[t] = s; return s; end })
local html_escape = { ["<"] = "&lt;", [">"] = "&gt;", ["&"] = "&amp;", }
local function colorize_html(s, t, extra)
s = gsub(s, "[<>&]", html_escape)
return format('<span class="irt_%s%s">%s</span>',
irtype_text[t], extra and " irt_extra" or "", s)
end
local irtype_html = setmetatable({},
{ __index = function(tab, t)
local s = colorize_html(irtype_text[t], t); tab[t] = s; return s; end })
local header_html = [[
<style type="text/css">
background { background: #ffffff; color: #000000; }
pre.ljdump {
font-size: 10pt;
background: #f0f4ff;
color: #000000;
border: 1px solid #bfcfff;
padding: 0.5em;
margin-left: 2em;
margin-right: 2em;
}
span.irt_str { color: #00a000; }
span.irt_thr, span.irt_fun { color: #404040; font-weight: bold; }
span.irt_tab { color: #c00000; }
span.irt_udt, span.irt_lud { color: #00c0c0; }
span.irt_num { color: #4040c0; }
span.irt_int, span.irt_i8, span.irt_u8, span.irt_i16, span.irt_u16 { color: #b040b0; }
span.irt_extra { font-style: italic; }
</style>
]]
local colorize, irtype
-- Lookup tables to convert some literals into names.
local litname = {
["SLOAD "] = setmetatable({}, { __index = function(t, mode)
local s = ""
if band(mode, 1) ~= 0 then s = s.."P" end
if band(mode, 2) ~= 0 then s = s.."F" end
if band(mode, 4) ~= 0 then s = s.."T" end
if band(mode, 8) ~= 0 then s = s.."C" end
if band(mode, 16) ~= 0 then s = s.."R" end
if band(mode, 32) ~= 0 then s = s.."I" end
if band(mode, 64) ~= 0 then s = s.."K" end
t[mode] = s
return s
end}),
["XLOAD "] = { [0] = "", "R", "V", "RV", "U", "RU", "VU", "RVU", },
["CONV "] = setmetatable({}, { __index = function(t, mode)
local s = irtype[band(mode, 31)]
s = irtype[band(shr(mode, 5), 31)].."."..s
if band(mode, 0x800) ~= 0 then s = s.." sext" end
local c = shr(mode, 12)
if c == 1 then s = s.." none"
elseif c == 2 then s = s.." index"
elseif c == 3 then s = s.." check" end
t[mode] = s
return s
end}),
["FLOAD "] = vmdef.irfield,
["FREF "] = vmdef.irfield,
["FPMATH"] = vmdef.irfpm,
["TMPREF"] = { [0] = "", "IN", "OUT", "INOUT", "", "", "OUT2", "INOUT2" },
["BUFHDR"] = { [0] = "RESET", "APPEND", "WRITE" },
["TOSTR "] = { [0] = "INT", "NUM", "CHAR" },
}
local function ctlsub(c)
if c == "\n" then return "\\n"
elseif c == "\r" then return "\\r"
elseif c == "\t" then return "\\t"
else return format("\\%03d", byte(c))
end
end
local function fmtfunc(func, pc)
local fi = funcinfo(func, pc)
if fi.loc then
return fi.loc
elseif fi.ffid then
return vmdef.ffnames[fi.ffid]
elseif fi.addr then
return format("C:%x", fi.addr)
else
return "(?)"
end
end
local function formatk(tr, idx, sn)
local k, t, slot = tracek(tr, idx)
local tn = type(k)
local s
if tn == "number" then
if t < 12 then
s = k == 0 and "NULL" or format("[0x%08x]", k)
elseif band(sn or 0, 0x30000) ~= 0 then
s = band(sn, 0x20000) ~= 0 and "contpc" or "ftsz"
elseif k == 2^52+2^51 then
s = "bias"
else
s = format(0 < k and k < 0x1p-1026 and "%+a" or "%+.14g", k)
end
elseif tn == "string" then
s = format(#k > 20 and '"%.20s"~' or '"%s"', gsub(k, "%c", ctlsub))
elseif tn == "function" then
s = fmtfunc(k)
elseif tn == "table" then
s = format("{%p}", k)
elseif tn == "userdata" then
if t == 12 then
s = format("userdata:%p", k)
else
s = format("[%p]", k)
if s == "[NULL]" then s = "NULL" end
end
elseif t == 21 then -- int64_t
s = sub(tostring(k), 1, -3)
if sub(s, 1, 1) ~= "-" then s = "+"..s end
elseif sn == 0x1057fff then -- SNAP(1, SNAP_FRAME | SNAP_NORESTORE, REF_NIL)
return "----" -- Special case for LJ_FR2 slot 1.
else
s = tostring(k) -- For primitives.
end
s = colorize(format("%-4s", s), t, band(sn or 0, 0x100000) ~= 0)
if slot then
s = format("%s @%d", s, slot)
end
return s
end
local function printsnap(tr, snap)
local n = 2
for s=0,snap[1]-1 do
local sn = snap[n]
if shr(sn, 24) == s then
n = n + 1
local ref = band(sn, 0xffff) - 0x8000 -- REF_BIAS
if ref < 0 then
out:write(formatk(tr, ref, sn))
elseif band(sn, 0x80000) ~= 0 then -- SNAP_SOFTFPNUM
out:write(colorize(format("%04d/%04d", ref, ref+1), 14))
else
local m, ot, op1, op2 = traceir(tr, ref)
out:write(colorize(format("%04d", ref), band(ot, 31), band(sn, 0x100000) ~= 0))
end
out:write(band(sn, 0x10000) == 0 and " " or "|") -- SNAP_FRAME
else
out:write("---- ")
end
end
out:write("]\n")
end
-- Dump snapshots (not interleaved with IR).
local function dump_snap(tr)
out:write("---- TRACE ", tr, " snapshots\n")
for i=0,1000000000 do
local snap = tracesnap(tr, i)
if not snap then break end
out:write(format("#%-3d %04d [ ", i, snap[0]))
printsnap(tr, snap)
end
end
-- Return a register name or stack slot for a rid/sp location.
local function ridsp_name(ridsp, ins)
if not disass then disass = require("jit.dis_"..jit.arch) end
local rid, slot = band(ridsp, 0xff), shr(ridsp, 8)
if rid == 253 or rid == 254 then
return (slot == 0 or slot == 255) and " {sink" or format(" {%04d", ins-slot)
end
if ridsp > 255 then return format("[%x]", slot*4) end
if rid < 128 then return disass.regname(rid) end
return ""
end
-- Dump CALL* function ref and return optional ctype.
local function dumpcallfunc(tr, ins)
local ctype
if ins > 0 then
local m, ot, op1, op2 = traceir(tr, ins)
if band(ot, 31) == 0 then -- nil type means CARG(func, ctype).
ins = op1
ctype = formatk(tr, op2)
end
end
if ins < 0 then
out:write(format("[0x%x](", tonumber((tracek(tr, ins)))))
else
out:write(format("%04d (", ins))
end
return ctype
end
-- Recursively gather CALL* args and dump them.
local function dumpcallargs(tr, ins)
if ins < 0 then
out:write(formatk(tr, ins))
else
local m, ot, op1, op2 = traceir(tr, ins)
local oidx = 6*shr(ot, 8)
local op = sub(vmdef.irnames, oidx+1, oidx+6)
if op == "CARG " then
dumpcallargs(tr, op1)
if op2 < 0 then
out:write(" ", formatk(tr, op2))
else
out:write(" ", format("%04d", op2))
end
else
out:write(format("%04d", ins))
end
end
end
-- Dump IR and interleaved snapshots.
local function dump_ir(tr, dumpsnap, dumpreg)
local info = traceinfo(tr)
if not info then return end
local nins = info.nins
out:write("---- TRACE ", tr, " IR\n")
local irnames = vmdef.irnames
local snapref = 65536
local snap, snapno
if dumpsnap then
snap = tracesnap(tr, 0)
snapref = snap[0]
snapno = 0
end
for ins=1,nins do
if ins >= snapref then
if dumpreg then
out:write(format(".... SNAP #%-3d [ ", snapno))
else
out:write(format(".... SNAP #%-3d [ ", snapno))
end
printsnap(tr, snap)
snapno = snapno + 1
snap = tracesnap(tr, snapno)
snapref = snap and snap[0] or 65536
end
local m, ot, op1, op2, ridsp = traceir(tr, ins)
local oidx, t = 6*shr(ot, 8), band(ot, 31)
local op = sub(irnames, oidx+1, oidx+6)
if op == "LOOP " then
if dumpreg then
out:write(format("%04d ------------ LOOP ------------\n", ins))
else
out:write(format("%04d ------ LOOP ------------\n", ins))
end
elseif op ~= "NOP " and op ~= "CARG " and
(dumpreg or op ~= "RENAME") then
local rid = band(ridsp, 255)
if dumpreg then
out:write(format("%04d %-6s", ins, ridsp_name(ridsp, ins)))
else
out:write(format("%04d ", ins))
end
out:write(format("%s%s %s %s ",
(rid == 254 or rid == 253) and "}" or
(band(ot, 128) == 0 and " " or ">"),
band(ot, 64) == 0 and " " or "+",
irtype[t], op))
local m1, m2 = band(m, 3), band(m, 3*4)
if sub(op, 1, 4) == "CALL" then
local ctype
if m2 == 1*4 then -- op2 == IRMlit
out:write(format("%-10s (", vmdef.ircall[op2]))
else
ctype = dumpcallfunc(tr, op2)
end
if op1 ~= -1 then dumpcallargs(tr, op1) end
out:write(")")
if ctype then out:write(" ctype ", ctype) end
elseif op == "CNEW " and op2 == -1 then
out:write(formatk(tr, op1))
elseif m1 ~= 3 then -- op1 != IRMnone
if op1 < 0 then
out:write(formatk(tr, op1))
else
out:write(format(m1 == 0 and "%04d" or "#%-3d", op1))
end
if m2 ~= 3*4 then -- op2 != IRMnone
if m2 == 1*4 then -- op2 == IRMlit
local litn = litname[op]
if litn and litn[op2] then
out:write(" ", litn[op2])
elseif op == "UREFO " or op == "UREFC " then
out:write(format(" #%-3d", shr(op2, 8)))
else
out:write(format(" #%-3d", op2))
end
elseif op2 < 0 then
out:write(" ", formatk(tr, op2))
else
out:write(format(" %04d", op2))
end
end
end
out:write("\n")
end
end
if snap then
if dumpreg then
out:write(format(".... SNAP #%-3d [ ", snapno))
else
out:write(format(".... SNAP #%-3d [ ", snapno))
end
printsnap(tr, snap)
end
end
------------------------------------------------------------------------------
local recprefix = ""
local recdepth = 0
-- Format trace error message.
local function fmterr(err, info)
if type(err) == "number" then
if type(info) == "function" then info = fmtfunc(info) end
err = format(vmdef.traceerr[err], info)
end
return err
end
-- Dump trace states.
local function dump_trace(what, tr, func, pc, otr, oex)
if what == "stop" or (what == "abort" and dumpmode.a) then
if dumpmode.i then dump_ir(tr, dumpmode.s, dumpmode.r and what == "stop")
elseif dumpmode.s then dump_snap(tr) end
if dumpmode.m then dump_mcode(tr) end
end
if what == "start" then
if dumpmode.H then out:write('<pre class="ljdump">\n') end
out:write("---- TRACE ", tr, " ", what)
if otr then out:write(" ", otr, "/", oex == -1 and "stitch" or oex) end
out:write(" ", fmtfunc(func, pc), "\n")
elseif what == "stop" or what == "abort" then
out:write("---- TRACE ", tr, " ", what)
if what == "abort" then
out:write(" ", fmtfunc(func, pc), " -- ", fmterr(otr, oex), "\n")
else
local info = traceinfo(tr)
local link, ltype = info.link, info.linktype
if link == tr or link == 0 then
out:write(" -> ", ltype, "\n")
elseif ltype == "root" then
out:write(" -> ", link, "\n")
else
out:write(" -> ", link, " ", ltype, "\n")
end
end
if dumpmode.H then out:write("</pre>\n\n") else out:write("\n") end
else
if what == "flush" then symtab, nexitsym = {}, 0 end
out:write("---- TRACE ", what, "\n\n")
end
out:flush()
end
-- Dump recorded bytecode.
local function dump_record(tr, func, pc, depth)
if depth ~= recdepth then
recdepth = depth
recprefix = rep(" .", depth)
end
local line
if pc >= 0 then
line = bcline(func, pc, recprefix)
if dumpmode.H then line = gsub(line, "[<>&]", html_escape) end
else
line = "0000 "..recprefix.." FUNCC \n"
end
if pc <= 0 then
out:write(sub(line, 1, -2), " ; ", fmtfunc(func), "\n")
else
out:write(line)
end
if pc >= 0 and band(funcbc(func, pc), 0xff) < 16 then -- ORDER BC
out:write(bcline(func, pc+1, recprefix)) -- Write JMP for cond.
end
end
------------------------------------------------------------------------------
local gpr64 = jit.arch:match("64")
local fprmips32 = jit.arch == "mips" or jit.arch == "mipsel"
-- Dump taken trace exits.
local function dump_texit(tr, ex, ngpr, nfpr, ...)
out:write("---- TRACE ", tr, " exit ", ex, "\n")
if dumpmode.X then
local regs = {...}
if gpr64 then
for i=1,ngpr do
out:write(format(" %016x", regs[i]))
if i % 4 == 0 then out:write("\n") end
end
else
for i=1,ngpr do
out:write(" ", tohex(regs[i]))
if i % 8 == 0 then out:write("\n") end
end
end
if fprmips32 then
for i=1,nfpr,2 do
out:write(format(" %+17.14g", regs[ngpr+i]))
if i % 8 == 7 then out:write("\n") end
end
else
for i=1,nfpr do
out:write(format(" %+17.14g", regs[ngpr+i]))
if i % 4 == 0 then out:write("\n") end
end
end
end
end
------------------------------------------------------------------------------
-- Detach dump handlers.
local function dumpoff()
if active then
active = false
jit.attach(dump_texit)
jit.attach(dump_record)
jit.attach(dump_trace)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach dump handlers.
local function dumpon(opt, outfile)
if active then dumpoff() end
local term = os.getenv("TERM")
local colormode = (term and term:match("color") or os.getenv("COLORTERM")) and "A" or "T"
if opt then
opt = gsub(opt, "[TAH]", function(mode) colormode = mode; return ""; end)
end
local m = { t=true, b=true, i=true, m=true, }
if opt and opt ~= "" then
local o = sub(opt, 1, 1)
if o ~= "+" and o ~= "-" then m = {} end
for i=1,#opt do m[sub(opt, i, i)] = (o ~= "-") end
end
dumpmode = m
if m.t or m.b or m.i or m.s or m.m then
jit.attach(dump_trace, "trace")
end
if m.b then
jit.attach(dump_record, "record")
if not bcline then bcline = require("jit.bc").line end
end
if m.x or m.X then
jit.attach(dump_texit, "texit")
end
if not outfile then outfile = os.getenv("LUAJIT_DUMPFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stdout
end
m[colormode] = true
if colormode == "A" then
colorize = colorize_ansi
irtype = irtype_ansi
elseif colormode == "H" then
colorize = colorize_html
irtype = irtype_html
out:write(header_html)
else
colorize = colorize_text
irtype = irtype_text
end
active = true
end
-- Public module functions.
return {
on = dumpon,
off = dumpoff,
start = dumpon -- For -j command line option.
}
@@ -0,0 +1,312 @@
----------------------------------------------------------------------------
-- LuaJIT profiler.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module is a simple command line interface to the built-in
-- low-overhead profiler of LuaJIT.
--
-- The lower-level API of the profiler is accessible via the "jit.profile"
-- module or the luaJIT_profile_* C API.
--
-- Example usage:
--
-- luajit -jp myapp.lua
-- luajit -jp=s myapp.lua
-- luajit -jp=-s myapp.lua
-- luajit -jp=vl myapp.lua
-- luajit -jp=G,profile.txt myapp.lua
--
-- The following dump features are available:
--
-- f Stack dump: function name, otherwise module:line. Default mode.
-- F Stack dump: ditto, but always prepend module.
-- l Stack dump: module:line.
-- <number> stack dump depth (callee < caller). Default: 1.
-- -<number> Inverse stack dump depth (caller > callee).
-- s Split stack dump after first stack level. Implies abs(depth) >= 2.
-- p Show full path for module names.
-- v Show VM states. Can be combined with stack dumps, e.g. vf or fv.
-- z Show zones. Can be combined with stack dumps, e.g. zf or fz.
-- r Show raw sample counts. Default: show percentages.
-- a Annotate excerpts from source code files.
-- A Annotate complete source code files.
-- G Produce raw output suitable for graphical tools (e.g. flame graphs).
-- m<number> Minimum sample percentage to be shown. Default: 3.
-- i<number> Sampling interval in milliseconds. Default: 10.
--
----------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local profile = require("jit.profile")
local vmdef = require("jit.vmdef")
local math = math
local pairs, ipairs, tonumber, floor = pairs, ipairs, tonumber, math.floor
local sort, format = table.sort, string.format
local stdout = io.stdout
local zone -- Load jit.zone module on demand.
-- Output file handle.
local out
------------------------------------------------------------------------------
local prof_ud
local prof_states, prof_split, prof_min, prof_raw, prof_fmt, prof_depth
local prof_ann, prof_count1, prof_count2, prof_samples
local map_vmmode = {
N = "Compiled",
I = "Interpreted",
C = "C code",
G = "Garbage Collector",
J = "JIT Compiler",
}
-- Profiler callback.
local function prof_cb(th, samples, vmmode)
prof_samples = prof_samples + samples
local key_stack, key_stack2, key_state
-- Collect keys for sample.
if prof_states then
if prof_states == "v" then
key_state = map_vmmode[vmmode] or vmmode
else
key_state = zone:get() or "(none)"
end
end
if prof_fmt then
key_stack = profile.dumpstack(th, prof_fmt, prof_depth)
key_stack = key_stack:gsub("%[builtin#(%d+)%]", function(x)
return vmdef.ffnames[tonumber(x)]
end)
if prof_split == 2 then
local k1, k2 = key_stack:match("(.-) [<>] (.*)")
if k2 then key_stack, key_stack2 = k1, k2 end
elseif prof_split == 3 then
key_stack2 = profile.dumpstack(th, "l", 1)
end
end
-- Order keys.
local k1, k2
if prof_split == 1 then
if key_state then
k1 = key_state
if key_stack then k2 = key_stack end
end
elseif key_stack then
k1 = key_stack
if key_stack2 then k2 = key_stack2 elseif key_state then k2 = key_state end
end
-- Coalesce samples in one or two levels.
if k1 then
local t1 = prof_count1
t1[k1] = (t1[k1] or 0) + samples
if k2 then
local t2 = prof_count2
local t3 = t2[k1]
if not t3 then t3 = {}; t2[k1] = t3 end
t3[k2] = (t3[k2] or 0) + samples
end
end
end
------------------------------------------------------------------------------
-- Show top N list.
local function prof_top(count1, count2, samples, indent)
local t, n = {}, 0
for k in pairs(count1) do
n = n + 1
t[n] = k
end
sort(t, function(a, b) return count1[a] > count1[b] end)
for i=1,n do
local k = t[i]
local v = count1[k]
local pct = floor(v*100/samples + 0.5)
if pct < prof_min then break end
if not prof_raw then
out:write(format("%s%2d%% %s\n", indent, pct, k))
elseif prof_raw == "r" then
out:write(format("%s%5d %s\n", indent, v, k))
else
out:write(format("%s %d\n", k, v))
end
if count2 then
local r = count2[k]
if r then
prof_top(r, nil, v, (prof_split == 3 or prof_split == 1) and " -- " or
(prof_depth < 0 and " -> " or " <- "))
end
end
end
end
-- Annotate source code
local function prof_annotate(count1, samples)
local files = {}
local ms = 0
for k, v in pairs(count1) do
local pct = floor(v*100/samples + 0.5)
ms = math.max(ms, v)
if pct >= prof_min then
local file, line = k:match("^(.*):(%d+)$")
if not file then file = k; line = 0 end
local fl = files[file]
if not fl then fl = {}; files[file] = fl; files[#files+1] = file end
line = tonumber(line)
fl[line] = prof_raw and v or pct
end
end
sort(files)
local fmtv, fmtn = " %3d%% | %s\n", " | %s\n"
if prof_raw then
local n = math.max(5, math.ceil(math.log10(ms)))
fmtv = "%"..n.."d | %s\n"
fmtn = (" "):rep(n).." | %s\n"
end
local ann = prof_ann
for _, file in ipairs(files) do
local f0 = file:byte()
if f0 == 40 or f0 == 91 then
out:write(format("\n====== %s ======\n[Cannot annotate non-file]\n", file))
break
end
local fp, err = io.open(file)
if not fp then
out:write(format("====== ERROR: %s: %s\n", file, err))
break
end
out:write(format("\n====== %s ======\n", file))
local fl = files[file]
local n, show = 1, false
if ann ~= 0 then
for i=1,ann do
if fl[i] then show = true; out:write("@@ 1 @@\n"); break end
end
end
for line in fp:lines() do
if line:byte() == 27 then
out:write("[Cannot annotate bytecode file]\n")
break
end
local v = fl[n]
if ann ~= 0 then
local v2 = fl[n+ann]
if show then
if v2 then show = n+ann elseif v then show = n
elseif show+ann < n then show = false end
elseif v2 then
show = n+ann
out:write(format("@@ %d @@\n", n))
end
if not show then goto next end
end
if v then
out:write(format(fmtv, v, line))
else
out:write(format(fmtn, line))
end
::next::
n = n + 1
end
fp:close()
end
end
------------------------------------------------------------------------------
-- Finish profiling and dump result.
local function prof_finish()
if prof_ud then
profile.stop()
local samples = prof_samples
if samples == 0 then
if prof_raw ~= true then out:write("[No samples collected]\n") end
return
end
if prof_ann then
prof_annotate(prof_count1, samples)
else
prof_top(prof_count1, prof_count2, samples, "")
end
prof_count1 = nil
prof_count2 = nil
prof_ud = nil
if out ~= stdout then out:close() end
end
end
-- Start profiling.
local function prof_start(mode)
local interval = ""
mode = mode:gsub("i%d*", function(s) interval = s; return "" end)
prof_min = 3
mode = mode:gsub("m(%d+)", function(s) prof_min = tonumber(s); return "" end)
prof_depth = 1
mode = mode:gsub("%-?%d+", function(s) prof_depth = tonumber(s); return "" end)
local m = {}
for c in mode:gmatch(".") do m[c] = c end
prof_states = m.z or m.v
if prof_states == "z" then zone = require("jit.zone") end
local scope = m.l or m.f or m.F or (prof_states and "" or "f")
local flags = (m.p or "")
prof_raw = m.r
if m.s then
prof_split = 2
if prof_depth == -1 or m["-"] then prof_depth = -2
elseif prof_depth == 1 then prof_depth = 2 end
elseif mode:find("[fF].*l") then
scope = "l"
prof_split = 3
else
prof_split = (scope == "" or mode:find("[zv].*[lfF]")) and 1 or 0
end
prof_ann = m.A and 0 or (m.a and 3)
if prof_ann then
scope = "l"
prof_fmt = "pl"
prof_split = 0
prof_depth = 1
elseif m.G and scope ~= "" then
prof_fmt = flags..scope.."Z;"
prof_depth = -100
prof_raw = true
prof_min = 0
elseif scope == "" then
prof_fmt = false
else
local sc = prof_split == 3 and m.f or m.F or scope
prof_fmt = flags..sc..(prof_depth >= 0 and "Z < " or "Z > ")
end
prof_count1 = {}
prof_count2 = {}
prof_samples = 0
profile.start(scope:lower()..interval, prof_cb)
prof_ud = newproxy(true)
getmetatable(prof_ud).__gc = prof_finish
end
------------------------------------------------------------------------------
local function start(mode, outfile)
if not outfile then outfile = os.getenv("LUAJIT_PROFILEFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stdout
end
prof_start(mode or "f")
end
-- Public module functions.
return {
start = start, -- For -j command line option.
stop = prof_finish
}
@@ -0,0 +1,170 @@
----------------------------------------------------------------------------
-- Verbose mode of the LuaJIT compiler.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module shows verbose information about the progress of the
-- JIT compiler. It prints one line for each generated trace. This module
-- is useful to see which code has been compiled or where the compiler
-- punts and falls back to the interpreter.
--
-- Example usage:
--
-- luajit -jv -e "for i=1,1000 do for j=1,1000 do end end"
-- luajit -jv=myapp.out myapp.lua
--
-- Default output is to stderr. To redirect the output to a file, pass a
-- filename as an argument (use '-' for stdout) or set the environment
-- variable LUAJIT_VERBOSEFILE. The file is overwritten every time the
-- module is started.
--
-- The output from the first example should look like this:
--
-- [TRACE 1 (command line):1 loop]
-- [TRACE 2 (1/3) (command line):1 -> 1]
--
-- The first number in each line is the internal trace number. Next are
-- the file name ('(command line)') and the line number (':1') where the
-- trace has started. Side traces also show the parent trace number and
-- the exit number where they are attached to in parentheses ('(1/3)').
-- An arrow at the end shows where the trace links to ('-> 1'), unless
-- it loops to itself.
--
-- In this case the inner loop gets hot and is traced first, generating
-- a root trace. Then the last exit from the 1st trace gets hot, too,
-- and triggers generation of the 2nd trace. The side trace follows the
-- path along the outer loop and *around* the inner loop, back to its
-- start, and then links to the 1st trace. Yes, this may seem unusual,
-- if you know how traditional compilers work. Trace compilers are full
-- of surprises like this -- have fun! :-)
--
-- Aborted traces are shown like this:
--
-- [TRACE --- foo.lua:44 -- leaving loop in root trace at foo:lua:50]
--
-- Don't worry -- trace aborts are quite common, even in programs which
-- can be fully compiled. The compiler may retry several times until it
-- finds a suitable trace.
--
-- Of course this doesn't work with features that are not-yet-implemented
-- (NYI error messages). The VM simply falls back to the interpreter. This
-- may not matter at all if the particular trace is not very high up in
-- the CPU usage profile. Oh, and the interpreter is quite fast, too.
--
-- Also check out the -jdump module, which prints all the gory details.
--
------------------------------------------------------------------------------
-- Cache some library functions and objects.
local jit = require("jit")
assert(jit.version_num == 20100, "LuaJIT core/library version mismatch")
local jutil = require("jit.util")
local vmdef = require("jit.vmdef")
local funcinfo, traceinfo = jutil.funcinfo, jutil.traceinfo
local type, format = type, string.format
local stdout, stderr = io.stdout, io.stderr
-- Active flag and output file handle.
local active, out
------------------------------------------------------------------------------
local startloc, startex
local function fmtfunc(func, pc)
local fi = funcinfo(func, pc)
if fi.loc then
return fi.loc
elseif fi.ffid then
return vmdef.ffnames[fi.ffid]
elseif fi.addr then
return format("C:%x", fi.addr)
else
return "(?)"
end
end
-- Format trace error message.
local function fmterr(err, info)
if type(err) == "number" then
if type(info) == "function" then info = fmtfunc(info) end
err = format(vmdef.traceerr[err], info)
end
return err
end
-- Dump trace states.
local function dump_trace(what, tr, func, pc, otr, oex)
if what == "start" then
startloc = fmtfunc(func, pc)
startex = otr and "("..otr.."/"..(oex == -1 and "stitch" or oex)..") " or ""
else
if what == "abort" then
local loc = fmtfunc(func, pc)
if loc ~= startloc then
out:write(format("[TRACE --- %s%s -- %s at %s]\n",
startex, startloc, fmterr(otr, oex), loc))
else
out:write(format("[TRACE --- %s%s -- %s]\n",
startex, startloc, fmterr(otr, oex)))
end
elseif what == "stop" then
local info = traceinfo(tr)
local link, ltype = info.link, info.linktype
if ltype == "interpreter" then
out:write(format("[TRACE %3s %s%s -- fallback to interpreter]\n",
tr, startex, startloc))
elseif ltype == "stitch" then
out:write(format("[TRACE %3s %s%s %s %s]\n",
tr, startex, startloc, ltype, fmtfunc(func, pc)))
elseif link == tr or link == 0 then
out:write(format("[TRACE %3s %s%s %s]\n",
tr, startex, startloc, ltype))
elseif ltype == "root" then
out:write(format("[TRACE %3s %s%s -> %d]\n",
tr, startex, startloc, link))
else
out:write(format("[TRACE %3s %s%s -> %d %s]\n",
tr, startex, startloc, link, ltype))
end
else
out:write(format("[TRACE %s]\n", what))
end
out:flush()
end
end
------------------------------------------------------------------------------
-- Detach dump handlers.
local function dumpoff()
if active then
active = false
jit.attach(dump_trace)
if out and out ~= stdout and out ~= stderr then out:close() end
out = nil
end
end
-- Open the output file and attach dump handlers.
local function dumpon(outfile)
if active then dumpoff() end
if not outfile then outfile = os.getenv("LUAJIT_VERBOSEFILE") end
if outfile then
out = outfile == "-" and stdout or assert(io.open(outfile, "w"))
else
out = stderr
end
jit.attach(dump_trace, "trace")
active = true
end
-- Public module functions.
return {
on = dumpon,
off = dumpoff,
start = dumpon -- For -j command line option.
}
@@ -0,0 +1,395 @@
-- This is a generated file. DO NOT EDIT!
return {
bcnames = "ISLT ISGE ISLE ISGT ISEQV ISNEV ISEQS ISNES ISEQN ISNEN ISEQP ISNEP ISTC ISFC IST ISF ISTYPEISNUM MOV NOT UNM LEN ADDVN SUBVN MULVN DIVVN MODVN ADDNV SUBNV MULNV DIVNV MODNV ADDVV SUBVV MULVV DIVVV MODVV POW CAT KSTR KCDATAKSHORTKNUM KPRI KNIL UGET USETV USETS USETN USETP UCLO FNEW TNEW TDUP GGET GSET TGETV TGETS TGETB TGETR TSETV TSETS TSETB TSETM TSETR CALLM CALL CALLMTCALLT ITERC ITERN VARG ISNEXTRETM RET RET0 RET1 FORI JFORI FORL IFORL JFORL ITERL IITERLJITERLLOOP ILOOP JLOOP JMP FUNCF IFUNCFJFUNCFFUNCV IFUNCVJFUNCVFUNCC FUNCCW",
irnames = "LT GE LE GT ULT UGE ULE UGT EQ NE ABC RETF NOP BASE PVAL GCSTEPHIOP LOOP USE PHI RENAMEPROF KPRI KINT KGC KPTR KKPTR KNULL KNUM KINT64KSLOT BNOT BSWAP BAND BOR BXOR BSHL BSHR BSAR BROL BROR ADD SUB MUL DIV MOD POW NEG ABS LDEXP MIN MAX FPMATHADDOV SUBOV MULOV AREF HREFK HREF NEWREFUREFO UREFC FREF TMPREFSTRREFLREF ALOAD HLOAD ULOAD FLOAD XLOAD SLOAD VLOAD ALEN ASTOREHSTOREUSTOREFSTOREXSTORESNEW XSNEW TNEW TDUP CNEW CNEWI BUFHDRBUFPUTBUFSTRTBAR OBAR XBAR CONV TOBIT TOSTR STRTO CALLN CALLA CALLL CALLS CALLXSCARG ",
irfpm = { [0]="floor", "ceil", "trunc", "sqrt", "log", "log2", "other", },
irfield = { [0]="str.len", "func.env", "func.pc", "func.ffid", "thread.env", "tab.meta", "tab.array", "tab.node", "tab.asize", "tab.hmask", "tab.nomm", "udata.meta", "udata.udtype", "udata.file", "sbuf.w", "sbuf.e", "sbuf.b", "sbuf.l", "sbuf.ref", "sbuf.r", "cdata.ctypeid", "cdata.ptr", "cdata.int", "cdata.int64", "cdata.int64_4", },
ircall = {
[0]="lj_str_cmp",
"lj_str_find",
"lj_str_new",
"lj_strscan_num",
"lj_strfmt_int",
"lj_strfmt_num",
"lj_strfmt_char",
"lj_strfmt_putint",
"lj_strfmt_putnum",
"lj_strfmt_putquoted",
"lj_strfmt_putfxint",
"lj_strfmt_putfnum_int",
"lj_strfmt_putfnum_uint",
"lj_strfmt_putfnum",
"lj_strfmt_putfstr",
"lj_strfmt_putfchar",
"lj_buf_putmem",
"lj_buf_putstr",
"lj_buf_putchar",
"lj_buf_putstr_reverse",
"lj_buf_putstr_lower",
"lj_buf_putstr_upper",
"lj_buf_putstr_rep",
"lj_buf_puttab",
"lj_bufx_set",
"lj_bufx_more",
"lj_serialize_put",
"lj_serialize_get",
"lj_serialize_encode",
"lj_serialize_decode",
"lj_buf_tostr",
"lj_tab_new_ah",
"lj_tab_new1",
"lj_tab_dup",
"lj_tab_clear",
"lj_tab_newkey",
"lj_tab_keyindex",
"lj_vm_next",
"lj_tab_len",
"lj_tab_len_hint",
"lj_gc_step_jit",
"lj_gc_barrieruv",
"lj_mem_newgco",
"lj_prng_u64d",
"lj_vm_modi",
"log10",
"exp",
"sin",
"cos",
"tan",
"asin",
"acos",
"atan",
"sinh",
"cosh",
"tanh",
"fputc",
"fwrite",
"fflush",
"lj_vm_floor",
"lj_vm_ceil",
"lj_vm_trunc",
"sqrt",
"log",
"lj_vm_log2",
"lj_vm_powi",
"pow",
"atan2",
"ldexp",
"lj_vm_tobit",
"softfp_add",
"softfp_sub",
"softfp_mul",
"softfp_div",
"softfp_cmp",
"softfp_i2d",
"softfp_d2i",
"lj_vm_sfmin",
"lj_vm_sfmax",
"lj_vm_tointg",
"softfp_ui2d",
"softfp_f2d",
"softfp_d2ui",
"softfp_d2f",
"softfp_i2f",
"softfp_ui2f",
"softfp_f2i",
"softfp_f2ui",
"fp64_l2d",
"fp64_ul2d",
"fp64_l2f",
"fp64_ul2f",
"fp64_d2l",
"fp64_d2ul",
"fp64_f2l",
"fp64_f2ul",
"lj_carith_divi64",
"lj_carith_divu64",
"lj_carith_modi64",
"lj_carith_modu64",
"lj_carith_powi64",
"lj_carith_powu64",
"lj_cdata_newv",
"lj_cdata_setfin",
"strlen",
"memcpy",
"memset",
"lj_vm_errno",
"lj_carith_mul64",
"lj_carith_shl64",
"lj_carith_shr64",
"lj_carith_sar64",
"lj_carith_rol64",
"lj_carith_ror64",
},
traceerr = {
[0]="error thrown or hook called during recording",
"trace too short",
"trace too long",
"trace too deep",
"too many snapshots",
"blacklisted",
"retry recording",
"NYI: bytecode %d",
"leaving loop in root trace",
"inner loop in root trace",
"loop unroll limit reached",
"bad argument type",
"JIT compilation disabled for function",
"call unroll limit reached",
"down-recursion, restarting",
"NYI: unsupported variant of FastFunc %s",
"NYI: return to lower frame",
"store with nil or NaN key",
"missing metamethod",
"looping index lookup",
"NYI: mixed sparse/dense table",
"symbol not in cache",
"NYI: unsupported C type conversion",
"NYI: unsupported C function type",
"guard would always fail",
"too many PHIs",
"persistent type instability",
"failed to allocate mcode memory",
"machine code too long",
"hit mcode limit (retrying)",
"too many spill slots",
"inconsistent register allocation",
"NYI: cannot assemble IR instruction %d",
"NYI: PHI shuffling too complex",
"NYI: register coalescing too complex",
},
ffnames = {
[0]="Lua",
"C",
"assert",
"type",
"next",
"pairs",
"ipairs_aux",
"ipairs",
"getmetatable",
"setmetatable",
"getfenv",
"setfenv",
"rawget",
"rawset",
"rawequal",
"unpack",
"select",
"tonumber",
"tostring",
"error",
"pcall",
"xpcall",
"loadfile",
"load",
"loadstring",
"dofile",
"gcinfo",
"collectgarbage",
"newproxy",
"print",
"coroutine.status",
"coroutine.running",
"coroutine.isyieldable",
"coroutine.create",
"coroutine.yield",
"coroutine.resume",
"coroutine.wrap_aux",
"coroutine.wrap",
"math.abs",
"math.floor",
"math.ceil",
"math.sqrt",
"math.log10",
"math.exp",
"math.sin",
"math.cos",
"math.tan",
"math.asin",
"math.acos",
"math.atan",
"math.sinh",
"math.cosh",
"math.tanh",
"math.frexp",
"math.modf",
"math.log",
"math.atan2",
"math.pow",
"math.fmod",
"math.ldexp",
"math.min",
"math.max",
"math.random",
"math.randomseed",
"bit.tobit",
"bit.bnot",
"bit.bswap",
"bit.lshift",
"bit.rshift",
"bit.arshift",
"bit.rol",
"bit.ror",
"bit.band",
"bit.bor",
"bit.bxor",
"bit.tohex",
"string.byte",
"string.char",
"string.sub",
"string.rep",
"string.reverse",
"string.lower",
"string.upper",
"string.dump",
"string.find",
"string.match",
"string.gmatch_aux",
"string.gmatch",
"string.gsub",
"string.format",
"table.maxn",
"table.insert",
"table.concat",
"table.sort",
"table.new",
"table.clear",
"io.method.close",
"io.method.read",
"io.method.write",
"io.method.flush",
"io.method.seek",
"io.method.setvbuf",
"io.method.lines",
"io.method.__gc",
"io.method.__tostring",
"io.open",
"io.popen",
"io.tmpfile",
"io.close",
"io.read",
"io.write",
"io.flush",
"io.input",
"io.output",
"io.lines",
"io.type",
"os.execute",
"os.remove",
"os.rename",
"os.tmpname",
"os.getenv",
"os.exit",
"os.clock",
"os.date",
"os.time",
"os.difftime",
"os.setlocale",
"debug.getregistry",
"debug.getmetatable",
"debug.setmetatable",
"debug.getfenv",
"debug.setfenv",
"debug.getinfo",
"debug.getlocal",
"debug.setlocal",
"debug.getupvalue",
"debug.setupvalue",
"debug.upvalueid",
"debug.upvaluejoin",
"debug.sethook",
"debug.gethook",
"debug.debug",
"debug.traceback",
"jit.on",
"jit.off",
"jit.flush",
"jit.status",
"jit.security",
"jit.attach",
"jit.util.funcinfo",
"jit.util.funcbc",
"jit.util.funck",
"jit.util.funcuvname",
"jit.util.traceinfo",
"jit.util.traceir",
"jit.util.tracek",
"jit.util.tracesnap",
"jit.util.tracemc",
"jit.util.traceexitstub",
"jit.util.ircalladdr",
"jit.opt.start",
"jit.profile.start",
"jit.profile.stop",
"jit.profile.dumpstack",
"ffi.meta.__index",
"ffi.meta.__newindex",
"ffi.meta.__eq",
"ffi.meta.__len",
"ffi.meta.__lt",
"ffi.meta.__le",
"ffi.meta.__concat",
"ffi.meta.__call",
"ffi.meta.__add",
"ffi.meta.__sub",
"ffi.meta.__mul",
"ffi.meta.__div",
"ffi.meta.__mod",
"ffi.meta.__pow",
"ffi.meta.__unm",
"ffi.meta.__tostring",
"ffi.meta.__pairs",
"ffi.meta.__ipairs",
"ffi.clib.__index",
"ffi.clib.__newindex",
"ffi.clib.__gc",
"ffi.callback.free",
"ffi.callback.set",
"ffi.cdef",
"ffi.new",
"ffi.cast",
"ffi.typeof",
"ffi.typeinfo",
"ffi.istype",
"ffi.sizeof",
"ffi.alignof",
"ffi.offsetof",
"ffi.errno",
"ffi.string",
"ffi.copy",
"ffi.fill",
"ffi.abi",
"ffi.metatype",
"ffi.gc",
"ffi.load",
"buffer.method.free",
"buffer.method.reset",
"buffer.method.skip",
"buffer.method.set",
"buffer.method.put",
"buffer.method.putf",
"buffer.method.get",
"buffer.method.putcdata",
"buffer.method.reserve",
"buffer.method.commit",
"buffer.method.ref",
"buffer.method.encode",
"buffer.method.decode",
"buffer.method.__gc",
"buffer.method.__tostring",
"buffer.method.__len",
"buffer.new",
"buffer.encode",
"buffer.decode",
},
}
@@ -0,0 +1,45 @@
----------------------------------------------------------------------------
-- LuaJIT profiler zones.
--
-- Copyright (C) 2005-2021 Mike Pall. All rights reserved.
-- Released under the MIT license. See Copyright Notice in luajit.h
----------------------------------------------------------------------------
--
-- This module implements a simple hierarchical zone model.
--
-- Example usage:
--
-- local zone = require("jit.zone")
-- zone("AI")
-- ...
-- zone("A*")
-- ...
-- print(zone:get()) --> "A*"
-- ...
-- zone()
-- ...
-- print(zone:get()) --> "AI"
-- ...
-- zone()
--
----------------------------------------------------------------------------
local remove = table.remove
return setmetatable({
flush = function(t)
for i=#t,1,-1 do t[i] = nil end
end,
get = function(t)
return t[#t]
end
}, {
__call = function(t, zone)
if zone then
t[#t+1] = zone
else
return (assert(remove(t), "empty zone stack"))
end
end
})
Binary file not shown.
@@ -28,6 +28,8 @@ wsl make HOST_LUA=luajit.exe "HOST_CC=clang.exe -m32" HOST_CFLAGS=-D_CRT_SECURE_
if not "%ERRORLEVEL%" == "0" goto :error
copy src\libluajit.a android\armeabi-v7a\libluajit.a
if not "%ERRORLEVEL%" == "0" goto :error
xcopy src\jit android\armeabi-v7a\jit /I
del android\armeabi-v7a\jit\.gitignore
:x86
rem x86
@@ -48,6 +50,8 @@ wsl make HOST_LUA=luajit.exe "HOST_CC=clang.exe %4" HOST_CFLAGS=-D_CRT_SECURE_NO
if not "%ERRORLEVEL%" == "0" goto :error
copy src\libluajit.a android\%1\libluajit.a
if not "%ERRORLEVEL%" == "0" goto :error
xcopy src\jit android\%1\jit /I
del android\%1\jit\.gitignore
goto :done
:error
+4 -2
View File
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>Contact</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -37,6 +37,8 @@
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
+693
View File
@@ -0,0 +1,693 @@
<!DOCTYPE html>
<html>
<head>
<title>String Buffer Library</title>
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
<link rel="stylesheet" type="text/css" href="bluequad-print.css" media="print">
<style type="text/css">
.lib {
vertical-align: middle;
margin-left: 5px;
padding: 0 5px;
font-size: 60%;
border-radius: 5px;
background: #c5d5ff;
color: #000;
}
</style>
</head>
<body>
<div id="site">
<a href="https://luajit.org"><span>Lua<span id="logo">JIT</span></span></a>
</div>
<div id="head">
<h1>String Buffer Library</h1>
</div>
<div id="nav">
<ul><li>
<a href="luajit.html">LuaJIT</a>
<ul><li>
<a href="https://luajit.org/download.html">Download <span class="ext">&raquo;</span></a>
</li><li>
<a href="install.html">Installation</a>
</li><li>
<a href="running.html">Running</a>
</li></ul>
</li><li>
<a href="extensions.html">Extensions</a>
<ul><li>
<a href="ext_ffi.html">FFI Library</a>
<ul><li>
<a href="ext_ffi_tutorial.html">FFI Tutorial</a>
</li><li>
<a href="ext_ffi_api.html">ffi.* API</a>
</li><li>
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a class="current" href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
</li><li>
<a href="ext_profiler.html">Profiler</a>
</li></ul>
</li><li>
<a href="status.html">Status</a>
</li><li>
<a href="faq.html">FAQ</a>
</li><li>
<a href="http://wiki.luajit.org/">Wiki <span class="ext">&raquo;</span></a>
</li><li>
<a href="https://luajit.org/list.html">Mailing List <span class="ext">&raquo;</span></a>
</li></ul>
</div>
<div id="main">
<p>
The string buffer library allows <b>high-performance manipulation of
string-like data</b>.
</p>
<p>
Unlike Lua strings, which are constants, string buffers are
<b>mutable</b> sequences of 8-bit (binary-transparent) characters. Data
can be stored, formatted and encoded into a string buffer and later
converted, extracted or decoded.
</p>
<p>
The convenient string buffer API simplifies common string manipulation
tasks, that would otherwise require creating many intermediate strings.
String buffers improve performance by eliminating redundant memory
copies, object creation, string interning and garbage collection
overhead. In conjunction with the FFI library, they allow zero-copy
operations.
</p>
<p>
The string buffer libary also includes a high-performance
<a href="serialize">serializer</a> for Lua objects.
</p>
<h2 id="wip" style="color:#ff0000">Work in Progress</h2>
<p>
<b style="color:#ff0000">This library is a work in progress. More
functionality will be added soon.</b>
</p>
<h2 id="use">Using the String Buffer Library</h2>
<p>
The string buffer library is built into LuaJIT by default, but it's not
loaded by default. Add this to the start of every Lua file that needs
one of its functions:
</p>
<pre class="code">
local buffer = require("string.buffer")
</pre>
<p>
The convention for the syntax shown on this page is that <tt>buffer</tt>
refers to the buffer library and <tt>buf</tt> refers to an individual
buffer object.
</p>
<p>
Please note the difference between a Lua function call, e.g.
<tt>buffer.new()</tt> (with a dot) and a Lua method call, e.g.
<tt>buf:reset()</tt> (with a colon).
</p>
<h3 id="buffer_object">Buffer Objects</h3>
<p>
A buffer object is a garbage-collected Lua object. After creation with
<tt>buffer.new()</tt>, it can (and should) be reused for many operations.
When the last reference to a buffer object is gone, it will eventually
be freed by the garbage collector, along with the allocated buffer
space.
</p>
<p>
Buffers operate like a FIFO (first-in first-out) data structure. Data
can be appended (written) to the end of the buffer and consumed (read)
from the front of the buffer. These operations may be freely mixed.
</p>
<p>
The buffer space that holds the characters is managed automatically
&mdash; it grows as needed and already consumed space is recycled. Use
<tt>buffer.new(size)</tt> and <tt>buf:free()</tt>, if you need more
control.
</p>
<p>
The maximum size of a single buffer is the same as the maximum size of a
Lua string, which is slightly below two gigabytes. For huge data sizes,
neither strings nor buffers are the right data structure &mdash; use the
FFI library to directly map memory or files up to the virtual memory
limit of your OS.
</p>
<h3 id="buffer_overview">Buffer Method Overview</h3>
<ul>
<li>
The <tt>buf:put*()</tt>-like methods append (write) characters to the
end of the buffer.
</li>
<li>
The <tt>buf:get*()</tt>-like methods consume (read) characters from the
front of the buffer.
</li>
<li>
Other methods, like <tt>buf:tostring()</tt> only read the buffer
contents, but don't change the buffer.
</li>
<li>
The <tt>buf:set()</tt> method allows zero-copy consumption of a string
or an FFI cdata object as a buffer.
</li>
<li>
The FFI-specific methods allow zero-copy read/write-style operations or
modifying the buffer contents in-place. Please check the
<a href="#ffi_caveats">FFI caveats</a> below, too.
</li>
<li>
Methods that don't need to return anything specific, return the buffer
object itself as a convenience. This allows method chaining, e.g.:
<tt>buf:reset():encode(obj)</tt> or <tt>buf:skip(len):get()</tt>
</li>
</ul>
<h2 id="create">Buffer Creation and Management</h2>
<h3 id="buffer_new"><tt>local buf = buffer.new([size [,options]])<br>
local buf = buffer.new([options])</tt></h3>
<p>
Creates a new buffer object.
</p>
<p>
The optional <tt>size</tt> argument ensures a minimum initial buffer
size. This is strictly an optimization when the required buffer size is
known beforehand. The buffer space will grow as needed, in any case.
</p>
<p>
The optional table <tt>options</tt> sets various
<a href="#serialize_options">serialization options</a>.
</p>
<h3 id="buffer_reset"><tt>buf = buf:reset()</tt></h3>
<p>
Reset (empty) the buffer. The allocated buffer space is not freed and
may be reused.
</p>
<h3 id="buffer_free"><tt>buf = buf:free()</tt></h3>
<p>
The buffer space of the buffer object is freed. The object itself
remains intact, empty and may be reused.
</p>
<p>
Note: you normally don't need to use this method. The garbage collector
automatically frees the buffer space, when the buffer object is
collected. Use this method, if you need to free the associated memory
immediately.
</p>
<h2 id="write">Buffer Writers</h2>
<h3 id="buffer_put"><tt>buf = buf:put([str|num|obj] [,…])</tt></h3>
<p>
Appends a string <tt>str</tt>, a number <tt>num</tt> or any object
<tt>obj</tt> with a <tt>__tostring</tt> metamethod to the buffer.
Multiple arguments are appended in the given order.
</p>
<p>
Appending a buffer to a buffer is possible and short-circuited
internally. But it still involves a copy. Better combine the buffer
writes to use a single buffer.
</p>
<h3 id="buffer_putf"><tt>buf = buf:putf(format, …)</tt></h3>
<p>
Appends the formatted arguments to the buffer. The <tt>format</tt>
string supports the same options as <tt>string.format()</tt>.
</p>
<h3 id="buffer_putcdata"><tt>buf = buf:putcdata(cdata, len)</tt><span class="lib">FFI</span></h3>
<p>
Appends the given <tt>len</tt> number of bytes from the memory pointed
to by the FFI <tt>cdata</tt> object to the buffer. The object needs to
be convertible to a (constant) pointer.
</p>
<h3 id="buffer_set"><tt>buf = buf:set(str)<br>
buf = buf:set(cdata, len)</tt><span class="lib">FFI</span></h3>
<p>
This method allows zero-copy consumption of a string or an FFI cdata
object as a buffer. It stores a reference to the passed string
<tt>str</tt> or the FFI <tt>cdata</tt> object in the buffer. Any buffer
space originally allocated is freed. This is <i>not</i> an append
operation, unlike the <tt>buf:put*()</tt> methods.
</p>
<p>
After calling this method, the buffer behaves as if
<tt>buf:free():put(str)</tt> or <tt>buf:free():put(cdata,&nbsp;len)</tt>
had been called. However, the data is only referenced and not copied, as
long as the buffer is only consumed.
</p>
<p>
In case the buffer is written to later on, the referenced data is copied
and the object reference is removed (copy-on-write semantics).
</p>
<p>
The stored reference is an anchor for the garbage collector and keeps the
originally passed string or FFI cdata object alive.
</p>
<h3 id="buffer_reserve"><tt>ptr, len = buf:reserve(size)</tt><span class="lib">FFI</span><br>
<tt>buf = buf:commit(used)</tt><span class="lib">FFI</span></h3>
<p>
The <tt>reserve</tt> method reserves at least <tt>size</tt> bytes of
write space in the buffer. It returns an <tt>uint8_t&nbsp;*</tt> FFI
cdata pointer <tt>ptr</tt> that points to this space.
</p>
<p>
The available length in bytes is returned in <tt>len</tt>. This is at
least <tt>size</tt> bytes, but may be more to facilitate efficient
buffer growth. You can either make use of the additional space or ignore
<tt>len</tt> and only use <tt>size</tt> bytes.
</p>
<p>
The <tt>commit</tt> method appends the <tt>used</tt> bytes of the
previously returned write space to the buffer data.
</p>
<p>
This pair of methods allows zero-copy use of C read-style APIs:
</p>
<pre class="code">
local MIN_SIZE = 65536
repeat
local ptr, len = buf:reserve(MIN_SIZE)
local n = C.read(fd, ptr, len)
if n == 0 then break end -- EOF.
if n &lt; 0 then error("read error") end
buf:commit(n)
until false
</pre>
<p>
The reserved write space is <i>not</i> initialized. At least the
<tt>used</tt> bytes <b>must</b> be written to before calling the
<tt>commit</tt> method. There's no need to call the <tt>commit</tt>
method, if nothing is added to the buffer (e.g. on error).
</p>
<h2 id="read">Buffer Readers</h2>
<h3 id="buffer_length"><tt>len = #buf</tt></h3>
<p>
Returns the current length of the buffer data in bytes.
</p>
<h3 id="buffer_concat"><tt>res = str|num|buf .. str|num|buf […]</tt></h3>
<p>
The Lua concatenation operator <tt>..</tt> also accepts buffers, just
like strings or numbers. It always returns a string and not a buffer.
</p>
<p>
Note that although this is supported for convenience, this thwarts one
of the main reasons to use buffers, which is to avoid string
allocations. Rewrite it with <tt>buf:put()</tt> and <tt>buf:get()</tt>.
</p>
<p>
Mixing this with unrelated objects that have a <tt>__concat</tt>
metamethod may not work, since these probably only expect strings.
</p>
<h3 id="buffer_skip"><tt>buf = buf:skip(len)</tt></h3>
<p>
Skips (consumes) <tt>len</tt> bytes from the buffer up to the current
length of the buffer data.
</p>
<h3 id="buffer_get"><tt>str, … = buf:get([len|nil] [,…])</tt></h3>
<p>
Consumes the buffer data and returns one or more strings. If called
without arguments, the whole buffer data is consumed. If called with a
number, up to <tt>len</tt> bytes are consumed. A <tt>nil</tt> argument
consumes the remaining buffer space (this only makes sense as the last
argument). Multiple arguments consume the buffer data in the given
order.
</p>
<p>
Note: a zero length or no remaining buffer data returns an empty string
and not <tt>nil</tt>.
</p>
<h3 id="buffer_tostring"><tt>str = buf:tostring()<br>
str = tostring(buf)</tt></h3>
<p>
Creates a string from the buffer data, but doesn't consume it. The
buffer remains unchanged.
</p>
<p>
Buffer objects also define a <tt>__tostring</tt> metamethod. This means
buffers can be passed to the global <tt>tostring()</tt> function and
many other functions that accept this in place of strings. The important
internal uses in functions like <tt>io.write()</tt> are short-circuited
to avoid the creation of an intermediate string object.
</p>
<h3 id="buffer_ref"><tt>ptr, len = buf:ref()</tt><span class="lib">FFI</span></h3>
<p>
Returns an <tt>uint8_t&nbsp;*</tt> FFI cdata pointer <tt>ptr</tt> that
points to the buffer data. The length of the buffer data in bytes is
returned in <tt>len</tt>.
</p>
<p>
The returned pointer can be directly passed to C functions that expect a
buffer and a length. You can also do bytewise reads
(<tt>local&nbsp;x&nbsp;=&nbsp;ptr[i]</tt>) or writes
(<tt>ptr[i]&nbsp;=&nbsp;0x40</tt>) of the buffer data.
</p>
<p>
In conjunction with the <tt>skip</tt> method, this allows zero-copy use
of C write-style APIs:
</p>
<pre class="code">
repeat
local ptr, len = buf:ref()
if len == 0 then break end
local n = C.write(fd, ptr, len)
if n &lt; 0 then error("write error") end
buf:skip(n)
until n >= len
</pre>
<p>
Unlike Lua strings, buffer data is <i>not</i> implicitly
zero-terminated. It's not safe to pass <tt>ptr</tt> to C functions that
expect zero-terminated strings. If you're not using <tt>len</tt>, then
you're doing something wrong.
</p>
<h2 id="serialize">Serialization of Lua Objects</h2>
<p>
The following functions and methods allow <b>high-speed serialization</b>
(encoding) of a Lua object into a string and decoding it back to a Lua
object. This allows convenient storage and transport of <b>structured
data</b>.
</p>
<p>
The encoded data is in an <a href="#serialize_format">internal binary
format</a>. The data can be stored in files, binary-transparent
databases or transmitted to other LuaJIT instances across threads,
processes or networks.
</p>
<p>
Encoding speed can reach up to 1 Gigabyte/second on a modern desktop- or
server-class system, even when serializing many small objects. Decoding
speed is mostly constrained by object creation cost.
</p>
<p>
The serializer handles most Lua types, common FFI number types and
nested structures. Functions, thread objects, other FFI cdata and full
userdata cannot be serialized (yet).
</p>
<p>
The encoder serializes nested structures as trees. Multiple references
to a single object will be stored separately and create distinct objects
after decoding. Circular references cause an error.
</p>
<h3 id="serialize_methods">Serialization Functions and Methods</h3>
<h3 id="buffer_encode"><tt>str = buffer.encode(obj)<br>
buf = buf:encode(obj)</tt></h3>
<p>
Serializes (encodes) the Lua object <tt>obj</tt>. The stand-alone
function returns a string <tt>str</tt>. The buffer method appends the
encoding to the buffer.
</p>
<p>
<tt>obj</tt> can be any of the supported Lua types &mdash; it doesn't
need to be a Lua table.
</p>
<p>
This function may throw an error when attempting to serialize
unsupported object types, circular references or deeply nested tables.
</p>
<h3 id="buffer_decode"><tt>obj = buffer.decode(str)<br>
obj = buf:decode()</tt></h3>
<p>
The stand-alone function de-serializes (decodes) the string
<tt>str</tt>, the buffer method de-serializes one object from the
buffer. Both return a Lua object <tt>obj</tt>.
</p>
<p>
The returned object may be any of the supported Lua types &mdash;
even <tt>nil</tt>.
</p>
<p>
This function may throw an error when fed with malformed or incomplete
encoded data. The stand-alone function throws when there's left-over
data after decoding a single top-level object. The buffer method leaves
any left-over data in the buffer.
</p>
<h3 id="serialize_options">Serialization Options</h3>
<p>
The <tt>options</tt> table passed to <tt>buffer.new()</tt> may contain
the following members (all optional):
</p>
<ul>
<li>
<tt>dict</tt> is a Lua table holding a <b>dictionary of strings</b> that
commonly occur as table keys of objects you are serializing. These keys
are compactly encoded as indexes during serialization. A well chosen
dictionary saves space and improves serialization performance.
</li>
<li>
<tt>metatable</tt> is a Lua table holding a <b>dictionary of metatables</b>
for the table objects you are serializing.
</li>
</ul>
<p>
<tt>dict</tt> needs to be an array of strings and <tt>metatable</tt> needs
to be an array of tables. Both starting at index 1 and without holes (no
<tt>nil</tt> inbetween). The tables are anchored in the buffer object and
internally modified into a two-way index (don't do this yourself, just pass
a plain array). The tables must not be modified after they have been passed
to <tt>buffer.new()</tt>.
</p>
<p>
The <tt>dict</tt> and <tt>metatable</tt> tables used by the encoder and
decoder must be the same. Put the most common entries at the front. Extend
at the end to ensure backwards-compatibility &mdash; older encodings can
then still be read. You may also set some indexes to <tt>false</tt> to
explicitly drop backwards-compatibility. Old encodings that use these
indexes will throw an error when decoded.
</p>
<p>
Metatables that are not found in the <tt>metatable</tt> dictionary are
ignored when encoding. Decoding returns a table with a <tt>nil</tt>
metatable.
</p>
<p>
Note: parsing and preparation of the options table is somewhat
expensive. Create a buffer object only once and recycle it for multiple
uses. Avoid mixing encoder and decoder buffers, since the
<tt>buf:set()</tt> method frees the already allocated buffer space:
</p>
<pre class="code">
local options = {
dict = { "commonly", "used", "string", "keys" },
}
local buf_enc = buffer.new(options)
local buf_dec = buffer.new(options)
local function encode(obj)
return buf_enc:reset():encode(obj):get()
end
local function decode(str)
return buf_dec:set(str):decode()
end
</pre>
<h3 id="serialize_stream">Streaming Serialization</h3>
<p>
In some contexts, it's desirable to do piecewise serialization of large
datasets, also known as <i>streaming</i>.
</p>
<p>
This serialization format can be safely concatenated and supports streaming.
Multiple encodings can simply be appended to a buffer and later decoded
individually:
</p>
<pre class="code">
local buf = buffer.new()
buf:encode(obj1)
buf:encode(obj2)
local copy1 = buf:decode()
local copy2 = buf:decode()
</pre>
<p>
Here's how to iterate over a stream:
</p>
<pre class="code">
while #buf ~= 0 do
local obj = buf:decode()
-- Do something with obj.
end
</pre>
<p>
Since the serialization format doesn't prepend a length to its encoding,
network applications may need to transmit the length, too.
</p>
<h3 id="serialize_format">Serialization Format Specification</h3>
<p>
This serialization format is designed for <b>internal use</b> by LuaJIT
applications. Serialized data is upwards-compatible and portable across
all supported LuaJIT platforms.
</p>
<p>
It's an <b>8-bit binary format</b> and not human-readable. It uses e.g.
embedded zeroes and stores embedded Lua string objects unmodified, which
are 8-bit-clean, too. Encoded data can be safely concatenated for
streaming and later decoded one top-level object at a time.
</p>
<p>
The encoding is reasonably compact, but tuned for maximum performance,
not for minimum space usage. It compresses well with any of the common
byte-oriented data compression algorithms.
</p>
<p>
Although documented here for reference, this format is explicitly
<b>not</b> intended to be a 'public standard' for structured data
interchange across computer languages (like JSON or MessagePack). Please
do not use it as such.
</p>
<p>
The specification is given below as a context-free grammar with a
top-level <tt>object</tt> as the starting point. Alternatives are
separated by the <tt>|</tt> symbol and <tt>*</tt> indicates repeats.
Grouping is implicit or indicated by <tt>{…}</tt>. Terminals are
either plain hex numbers, encoded as bytes, or have a <tt>.format</tt>
suffix.
</p>
<pre>
object → nil | false | true
| null | lightud32 | lightud64
| int | num | tab | tab_mt
| int64 | uint64 | complex
| string
nil → 0x00
false → 0x01
true → 0x02
null → 0x03 // NULL lightuserdata
lightud32 → 0x04 data.I // 32 bit lightuserdata
lightud64 → 0x05 data.L // 64 bit lightuserdata
int → 0x06 int.I // int32_t
num → 0x07 double.L
tab → 0x08 // Empty table
| 0x09 h.U h*{object object} // Key/value hash
| 0x0a a.U a*object // 0-based array
| 0x0b a.U a*object h.U h*{object object} // Mixed
| 0x0c a.U (a-1)*object // 1-based array
| 0x0d a.U (a-1)*object h.U h*{object object} // Mixed
tab_mt → 0x0e (index-1).U tab // Metatable dict entry
int64 → 0x10 int.L // FFI int64_t
uint64 → 0x11 uint.L // FFI uint64_t
complex → 0x12 re.L im.L // FFI complex
string → (0x20+len).U len*char.B
| 0x0f (index-1).U // String dict entry
.B = 8 bit
.I = 32 bit little-endian
.L = 64 bit little-endian
.U = prefix-encoded 32 bit unsigned number n:
0x00..0xdf → n.B
0xe0..0x1fdf → (0xe0|(((n-0xe0)>>8)&0x1f)).B ((n-0xe0)&0xff).B
0x1fe0.. → 0xff n.I
</pre>
<h2 id="error">Error handling</h2>
<p>
Many of the buffer methods can throw an error. Out-of-memory or usage
errors are best caught with an outer wrapper for larger parts of code.
There's not much one can do after that, anyway.
</p>
<p>
OTOH you may want to catch some errors individually. Buffer methods need
to receive the buffer object as the first argument. The Lua colon-syntax
<tt>obj:method()</tt> does that implicitly. But to wrap a method with
<tt>pcall()</tt>, the arguments need to be passed like this:
</p>
<pre class="code">
local ok, err = pcall(buf.encode, buf, obj)
if not ok then
-- Handle error in err.
end
</pre>
<h2 id="ffi_caveats">FFI caveats</h2>
<p>
The string buffer library has been designed to work well together with
the FFI library. But due to the low-level nature of the FFI library,
some care needs to be taken:
</p>
<p>
First, please remember that FFI pointers are zero-indexed. The space
returned by <tt>buf:reserve()</tt> and <tt>buf:ref()</tt> starts at the
returned pointer and ends before <tt>len</tt> bytes after that.
</p>
<p>
I.e. the first valid index is <tt>ptr[0]</tt> and the last valid index
is <tt>ptr[len-1]</tt>. If the returned length is zero, there's no valid
index at all. The returned pointer may even be <tt>NULL</tt>.
</p>
<p>
The space pointed to by the returned pointer is only valid as long as
the buffer is not modified in any way (neither append, nor consume, nor
reset, etc.). The pointer is also not a GC anchor for the buffer object
itself.
</p>
<p>
Buffer data is only guaranteed to be byte-aligned. Casting the returned
pointer to a data type with higher alignment may cause unaligned
accesses. It depends on the CPU architecture whether this is allowed or
not (it's always OK on x86/x64 and mostly OK on other modern
architectures).
</p>
<p>
FFI pointers or references do not count as GC anchors for an underlying
object. E.g. an <tt>array</tt> allocated with <tt>ffi.new()</tt> is
anchored by <tt>buf:set(array,&nbsp;len)</tt>, but not by
<tt>buf:set(array+offset,&nbsp;len)</tt>. The addition of the offset
creates a new pointer, even when the offset is zero. In this case, you
need to make sure there's still a reference to the original array as
long as its contents are in use by the buffer.
</p>
<p>
Even though each LuaJIT VM instance is single-threaded (but you can
create multiple VMs), FFI data structures can be accessed concurrently.
Be careful when reading/writing FFI cdata from/to buffers to avoid
concurrent accesses or modifications. In particular, the memory
referenced by <tt>buf:set(cdata,&nbsp;len)</tt> must not be modified
while buffer readers are working on it. Shared, but read-only memory
mappings of files are OK, but only if the file does not change.
</p>
<br class="flush">
</div>
<div id="foot">
<hr class="hide">
Copyright &copy; 2005-2021
<span class="noprint">
&middot;
<a href="contact.html">Contact</a>
</span>
</div>
</body>
</html>
+4 -2
View File
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>Lua/C API Extensions</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -37,6 +37,8 @@
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a class="current" href="ext_c_api.html">Lua/C API</a>
+4 -2
View File
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>FFI Library</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -37,6 +37,8 @@
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
+4 -2
View File
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>ffi.* API Functions</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -42,6 +42,8 @@ td.abiparam { font-weight: bold; width: 6em; }
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>FFI Semantics</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -42,6 +42,8 @@ td.convop { font-style: italic; width: 40%; }
<a class="current" href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>FFI Tutorial</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -44,6 +44,8 @@ td.idiomlua b { font-weight: normal; color: #2142bf; }
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
+4 -2
View File
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>jit.* Library</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -37,6 +37,8 @@
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a class="current" href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>Profiler</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -37,6 +37,8 @@
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
+13 -19
View File
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>Extensions</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -54,6 +54,8 @@ td.excinterop {
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
@@ -392,29 +394,19 @@ the toolchain used to compile LuaJIT:
<td class="excinterop">Interoperability</td>
</tr>
<tr class="odd separate">
<td class="excplatform">POSIX/x64, DWARF2 unwinding</td>
<td class="exccompiler">GCC 4.3+, Clang</td>
<td class="excplatform">External frame unwinding</td>
<td class="exccompiler">GCC, Clang, MSVC</td>
<td class="excinterop"><b style="color: #00a000;">Full</b></td>
</tr>
<tr class="even">
<td class="excplatform">ARM <tt>-DLUAJIT_UNWIND_EXTERNAL</tt></td>
<td class="exccompiler">GCC, Clang</td>
<td class="excinterop"><b style="color: #00a000;">Full</b></td>
</tr>
<tr class="odd">
<td class="excplatform">Other platforms, DWARF2 unwinding</td>
<td class="excplatform">Internal frame unwinding + DWARF2</td>
<td class="exccompiler">GCC, Clang</td>
<td class="excinterop"><b style="color: #c06000;">Limited</b></td>
</tr>
<tr class="even">
<td class="excplatform">Windows/x64</td>
<td class="exccompiler">MSVC</td>
<td class="excinterop"><b style="color: #00a000;">Full</b></td>
</tr>
<tr class="odd">
<td class="excplatform">Windows/x86</td>
<td class="exccompiler">Any</td>
<td class="excinterop"><b style="color: #00a000;">Full</b></td>
<td class="excplatform">Windows 64 bit</td>
<td class="exccompiler">non-MSVC</td>
<td class="excinterop"><b style="color: #c06000;">Limited</b></td>
</tr>
<tr class="even">
<td class="excplatform">Other platforms</td>
@@ -435,7 +427,9 @@ the toolchain used to compile LuaJIT:
on the C&nbsp;stack. The contents of the C++&nbsp;exception object
pass through unmodified.</li>
<li>Lua errors can be caught on the C++ side with <tt>catch(...)</tt>.
The corresponding Lua error message can be retrieved from the Lua stack.</li>
The corresponding Lua error message can be retrieved from the Lua stack.<br>
For MSVC for Windows 64 bit this requires compilation of your C++ code
with <tt>/EHa</tt>.</li>
<li>Throwing Lua errors across C++ frames is safe. C++ destructors
will be called.</li>
</ul>
+19 -5
View File
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>Frequently Asked Questions (FAQ)</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -40,6 +40,8 @@ dd { margin-left: 1.5em; }
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
@@ -74,7 +76,7 @@ about LuaJIT.</li>
and links to books and papers about Lua.</li>
<li>The community-managed <a href="http://lua-users.org/wiki/"><span class="ext">&raquo;</span>&nbsp;Lua Wiki</a>
has information about diverse topics.</li>
</ul>
</ul></dd>
</dl>
<dl id="tech">
@@ -119,8 +121,7 @@ Consider testing your application with older versions, too.<br>
Similarly, the Borland/Delphi runtime modifies the FPU control word and
enables FP exceptions. Of course this violates the Windows ABI, too.
Please check the Delphi docs for the Set8087CW method.
Please check the Delphi docs for the Set8087CW method.</dd>
</dl>
<dl id="ctrlc">
@@ -133,6 +134,19 @@ You have to press Ctrl-C twice to get stop your program. That's similar
to when it's stuck running inside a C function under the Lua interpreter.</dd>
</dl>
<dl id="order">
<dt>Q: Table iteration with <tt>pairs()</tt> does not result in the same order?</dt>
<dd>The order of table iteration is explicitly <b>undefined</b> by
the Lua language standard.<br>
Different Lua implementations or versions may use different orders for
otherwise identical tables. Different ways of constructing a table may
result in different orders, too.<br>
Due to improved VM security, LuaJIT 2.1 may even use a different order
on separate VM invocations or when string keys are newly interned.<br><br>
If your program relies on a deterministic order, it has a bug. Rewrite it,
so it doesn't rely on the key order. Or sort the table keys, if you must.</dd>
</dl>
<dl id="sandbox">
<dt>Q: Can Lua code be safely sandboxed?</dt>
<dd>
+19 -10
View File
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>Installation</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -65,6 +65,8 @@ td.compatno {
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
@@ -186,14 +188,18 @@ Debian/Ubuntu, install <tt>libc6-dev</tt> with the package manager.
</p>
<p>
The recommended way to fetch the latest version is to do a pull from
the git repository. Alternatively download the latest source package of
LuaJIT (pick the .tar.gz). Move it to a directory of your choice,
open a terminal window and change to this directory. Now unpack the archive
and change to the newly created directory:
the git repository.
</p>
<p>
Alternatively download the latest source package of LuaJIT (pick the .tar.gz).
Move it to a directory of your choice, open a terminal window and change
to this directory. Now unpack the archive and change to the newly created
directory (replace XX.YY.ZZ with the version you downloaded):
</p>
<pre class="code">
tar zxf LuaJIT-2.1.0-beta3.tar.gz
cd LuaJIT-2.1.0-beta3</pre>
tar zxf LuaJIT-XX.YY.ZZ.tar.gz
cd LuaJIT-XX.YY.ZZ
</pre>
<h3>Building LuaJIT</h3>
<p>
The supplied Makefiles try to auto-detect the settings needed for your
@@ -217,9 +223,12 @@ You can add an extra prefix to the search paths by appending the
make PREFIX=/home/myself/lj2
</pre>
<p>
Note for macOS: you must set the <tt>MACOSX_DEPLOYMENT_TARGET</tt>
environment variable to a value supported by your toolchain.
Note for macOS: you <b>must</b> set the <tt>MACOSX_DEPLOYMENT_TARGET</tt>
environment variable to a value supported by your toolchain:
</p>
<pre class="code">
MACOSX_DEPLOYMENT_TARGET=XX.YY make
</pre>
<h3>Installing LuaJIT</h3>
<p>
The top-level Makefile installs LuaJIT by default under
+4 -2
View File
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>LuaJIT</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -122,6 +122,8 @@ table.feature small {
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
+6 -4
View File
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>Running LuaJIT</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -59,6 +59,8 @@ td.param_default {
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
@@ -183,8 +185,8 @@ itself. For a description of their options and output format, please
read the comment block at the start of their source.
They can be found in the <tt>lib</tt> directory of the source
distribution or installed under the <tt>jit</tt> directory. By default
this is <tt>/usr/local/share/luajit-2.1.0-beta3/jit</tt> on POSIX
systems.
this is <tt>/usr/local/share/luajit-XX.YY.ZZ>/jit</tt> on POSIX
systems (replace XX.YY.ZZ by the installed version).
</p>
<h3 id="opt_O"><tt>-O[level]</tt><br>
+4 -8
View File
@@ -1,8 +1,8 @@
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<!DOCTYPE html>
<html>
<head>
<title>Status</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta charset="utf-8">
<meta name="Copyright" content="Copyright (C) 2005-2021">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
@@ -40,6 +40,8 @@ ul li { padding-bottom: 0.3em; }
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_buffer.html">String Buffers</a>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
@@ -90,12 +92,6 @@ The Lua <b>debug API</b> is missing a couple of features (return
hooks for non-Lua functions) and shows slightly different behavior
in LuaJIT (no per-coroutine hooks, no tail call counting).
</li>
<li>
Currently some <b>out-of-memory</b> errors from <b>on-trace code</b> are not
handled correctly. The error may fall through an on-trace
<tt>pcall</tt> or it may be passed on to the function set with
<tt>lua_atpanic</tt> on x64.
</li>
</ul>
<br class="flush">
</div>
+5 -2
View File
@@ -294,7 +294,7 @@ int dasm_link(Dst_DECL, size_t *szp)
{ /* Handle globals not defined in this translation unit. */
int idx;
for (idx = 20; idx*sizeof(int) < D->lgsize; idx++) {
for (idx = 10; idx*sizeof(int) < D->lgsize; idx++) {
int n = D->lglabels[idx];
/* Undefined label: Collapse rel chain and replace with marker (< 0). */
while (n > 0) { int *pb = DASM_POS2PTR(D, n); n = *pb; *pb = -idx; }
@@ -371,7 +371,10 @@ int dasm_encode(Dst_DECL, void *buffer)
ins &= 255; while ((((char *)cp - base) & ins)) *cp++ = 0xe1a00000;
break;
case DASM_REL_LG:
CK(n >= 0, UNDEF_LG);
if (n < 0) {
n = (int)((ptrdiff_t)D->globals[-n] - (ptrdiff_t)cp - 4);
goto patchrel;
}
/* fallthrough */
case DASM_REL_PC:
CK(n >= 0, UNDEF_PC);
+3 -3
View File
@@ -9,9 +9,9 @@
local _info = {
arch = "arm",
description = "DynASM ARM module",
version = "1.4.0",
vernum = 10400,
release = "2015-10-18",
version = "1.5.0",
vernum = 10500,
release = "2021-05-02",
author = "Mike Pall",
license = "MIT",
}
+50 -8
View File
@@ -21,8 +21,9 @@ enum {
/* The following actions need a buffer position. */
DASM_ALIGN, DASM_REL_LG, DASM_LABEL_LG,
/* The following actions also have an argument. */
DASM_REL_PC, DASM_LABEL_PC,
DASM_REL_PC, DASM_LABEL_PC, DASM_REL_A,
DASM_IMM, DASM_IMM6, DASM_IMM12, DASM_IMM13W, DASM_IMM13X, DASM_IMML,
DASM_IMMV, DASM_VREG,
DASM__MAX
};
@@ -39,6 +40,7 @@ enum {
#define DASM_S_RANGE_LG 0x13000000
#define DASM_S_RANGE_PC 0x14000000
#define DASM_S_RANGE_REL 0x15000000
#define DASM_S_RANGE_VREG 0x16000000
#define DASM_S_UNDEF_LG 0x21000000
#define DASM_S_UNDEF_PC 0x22000000
@@ -247,7 +249,7 @@ void dasm_put(Dst_DECL, int start, ...)
n = (ins & 255); CK(n < D->maxsection, RANGE_SEC);
D->section = &D->sections[n]; goto stop;
case DASM_ESC: p++; ofs += 4; break;
case DASM_REL_EXT: break;
case DASM_REL_EXT: if ((ins & 0x8000)) ofs += 8; break;
case DASM_ALIGN: ofs += (ins & 255); b[pos++] = ofs; break;
case DASM_REL_LG:
n = (ins & 2047) - 10; pl = D->lglabels + n;
@@ -268,6 +270,11 @@ void dasm_put(Dst_DECL, int start, ...)
*pl = pos;
}
pos++;
if ((ins & 0x8000)) ofs += 8;
break;
case DASM_REL_A:
b[pos++] = n;
b[pos++] = va_arg(ap, int);
break;
case DASM_LABEL_LG:
pl = D->lglabels + (ins & 2047) - 10; CKPL(lg, LG); goto putlabel;
@@ -312,13 +319,21 @@ void dasm_put(Dst_DECL, int start, ...)
}
case DASM_IMML: {
#ifdef DASM_CHECKS
int scale = (p[-2] >> 30);
int scale = (ins & 3);
CK((!(n & ((1<<scale)-1)) && (unsigned int)(n>>scale) < 4096) ||
(unsigned int)(n+256) < 512, RANGE_I);
#endif
b[pos++] = n;
break;
}
case DASM_IMMV:
ofs += 4;
b[pos++] = n;
break;
case DASM_VREG:
CK(n < 32, RANGE_VREG);
b[pos++] = n;
break;
}
}
}
@@ -348,7 +363,7 @@ int dasm_link(Dst_DECL, size_t *szp)
{ /* Handle globals not defined in this translation unit. */
int idx;
for (idx = 20; idx*sizeof(int) < D->lgsize; idx++) {
for (idx = 10; idx*sizeof(int) < D->lgsize; idx++) {
int n = D->lglabels[idx];
/* Undefined label: Collapse rel chain and replace with marker (< 0). */
while (n > 0) { int *pb = DASM_POS2PTR(D, n); n = *pb; *pb = -idx; }
@@ -375,8 +390,8 @@ int dasm_link(Dst_DECL, size_t *szp)
case DASM_REL_LG: case DASM_REL_PC: pos++; break;
case DASM_LABEL_LG: case DASM_LABEL_PC: b[pos++] += ofs; break;
case DASM_IMM: case DASM_IMM6: case DASM_IMM12: case DASM_IMM13W:
case DASM_IMML: pos++; break;
case DASM_IMM13X: pos += 2; break;
case DASM_IMML: case DASM_IMMV: case DASM_VREG: pos++; break;
case DASM_IMM13X: case DASM_REL_A: pos += 2; break;
}
}
stop: (void)0;
@@ -426,7 +441,12 @@ int dasm_encode(Dst_DECL, void *buffer)
ins &= 255; while ((((char *)cp - base) & ins)) *cp++ = 0xe1a00000;
break;
case DASM_REL_LG:
CK(n >= 0, UNDEF_LG);
if (n < 0) {
ptrdiff_t na = (ptrdiff_t)D->globals[-n] - (ptrdiff_t)cp + 4;
n = (int)na;
CK((ptrdiff_t)n == na, RANGE_REL);
goto patchrel;
}
/* fallthrough */
case DASM_REL_PC:
CK(n >= 0, UNDEF_PC);
@@ -446,8 +466,24 @@ int dasm_encode(Dst_DECL, void *buffer)
} else if ((ins & 0x1000)) { /* TBZ, TBNZ */
CK((n & 3) == 0 && ((n+0x00008000) >> 16) == 0, RANGE_REL);
cp[-1] |= ((n << 3) & 0x0007ffe0);
} else if ((ins & 0x8000)) { /* absolute */
cp[0] = (unsigned int)((ptrdiff_t)cp - 4 + n);
cp[1] = (unsigned int)(((ptrdiff_t)cp - 4 + n) >> 32);
cp += 2;
}
break;
case DASM_REL_A: {
ptrdiff_t na = (((ptrdiff_t)(*b++) << 32) | (unsigned int)n);
if ((ins & 0x3000) == 0x3000) { /* ADRP */
ins &= ~0x1000;
na = (na >> 12) - (((ptrdiff_t)cp - 4) >> 12);
} else {
na = na - (ptrdiff_t)cp + 4;
}
n = (int)na;
CK((ptrdiff_t)n == na, RANGE_REL);
goto patchrel;
}
case DASM_LABEL_LG:
ins &= 2047; if (ins >= 20) D->globals[ins-10] = (void *)(base + n);
break;
@@ -468,11 +504,17 @@ int dasm_encode(Dst_DECL, void *buffer)
cp[-1] |= (dasm_imm13(n, *b++) << 10);
break;
case DASM_IMML: {
int scale = (p[-2] >> 30);
int scale = (ins & 3);
cp[-1] |= (!(n & ((1<<scale)-1)) && (unsigned int)(n>>scale) < 4096) ?
((n << (10-scale)) | 0x01000000) : ((n & 511) << 12);
break;
}
case DASM_IMMV:
*cp++ = n;
break;
case DASM_VREG:
cp[-1] |= (n & 0x1f) << (ins & 0x1f);
break;
default: *cp++ = ins; break;
}
}
+94 -41
View File
@@ -9,9 +9,9 @@
local _info = {
arch = "arm",
description = "DynASM ARM64 module",
version = "1.4.0",
vernum = 10400,
release = "2015-10-18",
version = "1.5.0",
vernum = 10500,
release = "2021-05-02",
author = "Mike Pall",
license = "MIT",
}
@@ -23,12 +23,12 @@ local _M = { _info = _info }
local type, tonumber, pairs, ipairs = type, tonumber, pairs, ipairs
local assert, setmetatable, rawget = assert, setmetatable, rawget
local _s = string
local sub, format, byte, char = _s.sub, _s.format, _s.byte, _s.char
local format, byte, char = _s.format, _s.byte, _s.char
local match, gmatch, gsub = _s.match, _s.gmatch, _s.gsub
local concat, sort, insert = table.concat, table.sort, table.insert
local bit = bit or require("bit")
local band, shl, shr, sar = bit.band, bit.lshift, bit.rshift, bit.arshift
local ror, tohex = bit.ror, bit.tohex
local ror, tohex, tobit = bit.ror, bit.tohex, bit.tobit
-- Inherited tables and callbacks.
local g_opt, g_arch
@@ -39,7 +39,9 @@ local wline, werror, wfatal, wwarn
local action_names = {
"STOP", "SECTION", "ESC", "REL_EXT",
"ALIGN", "REL_LG", "LABEL_LG",
"REL_PC", "LABEL_PC", "IMM", "IMM6", "IMM12", "IMM13W", "IMM13X", "IMML",
"REL_PC", "LABEL_PC", "REL_A",
"IMM", "IMM6", "IMM12", "IMM13W", "IMM13X", "IMML", "IMMV",
"VREG",
}
-- Maximum number of section buffer positions for dasm_put().
@@ -246,9 +248,12 @@ local map_cond = {
local parse_reg_type
local function parse_reg(expr)
local function parse_reg(expr, shift)
if not expr then werror("expected register name") end
local tname, ovreg = match(expr, "^([%w_]+):(@?%l%d+)$")
if not tname then
tname, ovreg = match(expr, "^([%w_]+):(R[xwqdshb]%b())$")
end
local tp = map_type[tname or expr]
if tp then
local reg = ovreg or tp.reg
@@ -266,18 +271,28 @@ local function parse_reg(expr)
elseif parse_reg_type ~= rt then
werror("register size mismatch")
end
return r, tp
return shl(r, shift), tp
end
end
local vrt, vreg = match(expr, "^R([xwqdshb])(%b())$")
if vreg then
if not parse_reg_type then
parse_reg_type = vrt
elseif parse_reg_type ~= vrt then
werror("register size mismatch")
end
if shift then waction("VREG", shift, vreg) end
return 0
end
werror("bad register name `"..expr.."'")
end
local function parse_reg_base(expr)
if expr == "sp" then return 0x3e0 end
local base, tp = parse_reg(expr)
local base, tp = parse_reg(expr, 5)
if parse_reg_type ~= "x" then werror("bad register type") end
parse_reg_type = false
return shl(base, 5), tp
return base, tp
end
local parse_ctx = {}
@@ -297,7 +312,7 @@ local function parse_number(n)
local code = loadenv("return "..n)
if code then
local ok, y = pcall(code)
if ok then return y end
if ok and type(y) == "number" then return y end
end
return nil
end
@@ -403,7 +418,7 @@ local function parse_imm_load(imm, scale)
end
werror("out of range immediate `"..imm.."'")
else
waction("IMML", 0, imm)
waction("IMML", scale, imm)
return 0
end
end
@@ -462,6 +477,7 @@ end
local function parse_load(params, nparams, n, op)
if params[n+2] then werror("too many operands") end
local scale = shr(op, 30)
local pn, p2 = params[n], params[n+1]
local p1, wb = match(pn, "^%[%s*(.-)%s*%](!?)$")
if not p1 then
@@ -470,14 +486,13 @@ local function parse_load(params, nparams, n, op)
if reg and tailr ~= "" then
local base, tp = parse_reg_base(reg)
if tp then
waction("IMML", 0, format(tp.ctypefmt, tailr))
waction("IMML", scale, format(tp.ctypefmt, tailr))
return op + base
end
end
end
werror("expected address operand")
end
local scale = shr(op, 30)
if p2 then
if wb == "!" then werror("bad use of '!'") end
op = op + parse_reg_base(p1) + parse_imm(p2, 9, 12, 0, true) + 0x400
@@ -494,7 +509,7 @@ local function parse_load(params, nparams, n, op)
op = op + parse_imm_load(imm, scale)
else
local p2b, p3b, p3s = match(p2a, "^,%s*([^,%s]*)%s*,?%s*(%S*)%s*(.*)$")
op = op + shl(parse_reg(p2b), 16) + 0x00200800
op = op + parse_reg(p2b, 16) + 0x00200800
if parse_reg_type ~= "x" and parse_reg_type ~= "w" then
werror("bad index register type")
end
@@ -561,14 +576,14 @@ local function parse_load_pair(params, nparams, n, op)
end
local function parse_label(label, def)
local prefix = sub(label, 1, 2)
local prefix = label:sub(1, 2)
-- =>label (pc label reference)
if prefix == "=>" then
return "PC", 0, sub(label, 3)
return "PC", 0, label:sub(3)
end
-- ->name (global label reference)
if prefix == "->" then
return "LG", map_global[sub(label, 3)]
return "LG", map_global[label:sub(3)]
end
if def then
-- [1-9] (local label definition)
@@ -586,8 +601,11 @@ local function parse_label(label, def)
if extname then
return "EXT", map_extern[extname]
end
-- &expr (pointer)
if label:sub(1, 1) == "&" then
return "A", 0, format("(ptrdiff_t)(%s)", label:sub(2))
end
end
werror("bad label `"..label.."'")
end
local function branch_type(op)
@@ -620,24 +638,24 @@ local function alias_bfx(p)
end
local function alias_bfiz(p)
parse_reg(p[1])
parse_reg(p[1], 0)
if parse_reg_type == "w" then
p[3] = "#-("..p[3]:sub(2)..")%32"
p[3] = "#(32-("..p[3]:sub(2).."))%32"
p[4] = "#("..p[4]:sub(2)..")-1"
else
p[3] = "#-("..p[3]:sub(2)..")%64"
p[3] = "#(64-("..p[3]:sub(2).."))%64"
p[4] = "#("..p[4]:sub(2)..")-1"
end
end
local alias_lslimm = op_alias("ubfm_4", function(p)
parse_reg(p[1])
parse_reg(p[1], 0)
local sh = p[3]:sub(2)
if parse_reg_type == "w" then
p[3] = "#-("..sh..")%32"
p[3] = "#(32-("..sh.."))%32"
p[4] = "#31-("..sh..")"
else
p[3] = "#-("..sh..")%64"
p[3] = "#(64-("..sh.."))%64"
p[4] = "#63-("..sh..")"
end
end)
@@ -881,25 +899,25 @@ end
-- Handle opcodes defined with template strings.
local function parse_template(params, template, nparams, pos)
local op = tonumber(sub(template, 1, 8), 16)
local op = tonumber(template:sub(1, 8), 16)
local n = 1
local rtt = {}
parse_reg_type = false
-- Process each character.
for p in gmatch(sub(template, 9), ".") do
for p in gmatch(template:sub(9), ".") do
local q = params[n]
if p == "D" then
op = op + parse_reg(q); n = n + 1
op = op + parse_reg(q, 0); n = n + 1
elseif p == "N" then
op = op + shl(parse_reg(q), 5); n = n + 1
op = op + parse_reg(q, 5); n = n + 1
elseif p == "M" then
op = op + shl(parse_reg(q), 16); n = n + 1
op = op + parse_reg(q, 16); n = n + 1
elseif p == "A" then
op = op + shl(parse_reg(q), 10); n = n + 1
op = op + parse_reg(q, 10); n = n + 1
elseif p == "m" then
op = op + shl(parse_reg(params[n-1]), 16)
op = op + parse_reg(params[n-1], 16)
elseif p == "p" then
if q == "sp" then params[n] = "@x31" end
@@ -930,8 +948,14 @@ local function parse_template(params, template, nparams, pos)
elseif p == "B" then
local mode, v, s = parse_label(q, false); n = n + 1
if not mode then werror("bad label `"..q.."'") end
local m = branch_type(op)
waction("REL_"..mode, v+m, s, 1)
if mode == "A" then
waction("REL_"..mode, v+m, format("(unsigned int)(%s)", s))
actargs[#actargs+1] = format("(unsigned int)((%s)>>32)", s)
else
waction("REL_"..mode, v+m, s, 1)
end
elseif p == "I" then
op = op + parse_imm12(q); n = n + 1
@@ -977,8 +1001,8 @@ function op_template(params, template, nparams)
if not params then return template:gsub("%x%x%x%x%x%x%x%x", "") end
-- Limit number of section buffer positions used by a single dasm_put().
-- A single opcode needs a maximum of 3 positions.
if secpos+3 > maxsecpos then wflush() end
-- A single opcode needs a maximum of 4 positions.
if secpos+4 > maxsecpos then wflush() end
local pos = wpos()
local lpos, apos, spos = #actlist, #actargs, secpos
@@ -990,9 +1014,11 @@ function op_template(params, template, nparams)
actlist[lpos+1] = nil
actlist[lpos+2] = nil
actlist[lpos+3] = nil
actlist[lpos+4] = nil
actargs[apos+1] = nil
actargs[apos+2] = nil
actargs[apos+3] = nil
actargs[apos+4] = nil
end
error(err, 0)
end
@@ -1036,23 +1062,50 @@ map_op[".label_1"] = function(params)
if not params then return "[1-9] | ->global | =>pcexpr" end
if secpos+1 > maxsecpos then wflush() end
local mode, n, s = parse_label(params[1], true)
if mode == "EXT" then werror("bad label definition") end
if not mode or mode == "EXT" then werror("bad label definition") end
waction("LABEL_"..mode, n, s, 1)
end
------------------------------------------------------------------------------
-- Pseudo-opcodes for data storage.
map_op[".long_*"] = function(params)
local function op_data(params)
if not params then return "imm..." end
local sz = params.op == ".long" and 4 or 8
for _,p in ipairs(params) do
local n = tonumber(p)
if not n then werror("bad immediate `"..p.."'") end
if n < 0 then n = n + 2^32 end
wputw(n)
local imm = parse_number(p)
if imm then
local n = tobit(imm)
if n == imm or (n < 0 and n + 2^32 == imm) then
wputw(n < 0 and n + 2^32 or n)
if sz == 8 then
wputw(imm < 0 and 0xffffffff or 0)
end
elseif sz == 4 then
werror("bad immediate `"..p.."'")
else
imm = nil
end
end
if not imm then
local mode, v, s = parse_label(p, false)
if sz == 4 then
if mode then werror("label does not fit into .long") end
waction("IMMV", 0, p)
elseif mode and mode ~= "A" then
waction("REL_"..mode, v+0x8000, s, 1)
else
if mode == "A" then p = s end
waction("IMMV", 0, format("(unsigned int)(%s)", p))
waction("IMMV", 0, format("(unsigned int)((unsigned long long)(%s)>>32)", p))
end
end
if secpos+2 > maxsecpos then wflush() end
end
end
map_op[".long_*"] = op_data
map_op[".quad_*"] = op_data
map_op[".addr_*"] = op_data
-- Alignment pseudo-opcode.
map_op[".align_1"] = function(params)

Some files were not shown because too many files have changed in this diff Show More