Basic bytecode inspection for test1.lua

inspectbc.lua

@@ -0,0 +1,14 @@
+local readbc = require('readbc')
+local opcodes = require('opcodes')
+
+fp = io.open('tests/test1.luo')
+header = readbc.read_header(fp)
+proto = readbc.read_proto(fp)
+
+for i = 1, #proto.bcins do
+    local decoded = opcodes.decode(proto.bcins[i])
+    local def = opcodes.defs[decoded.id]
+    print(def.name)
+    print(decoded.a)
+    print(decoded.d)
+end

microLua.lua

@@ -2,4 +2,5 @@ local readbc = require('readbc')
 
 fp = io.open('tests/test1.luo')
 header = readbc.read_header(fp)
-print(header.name)
+proto = readbc.read_proto(fp)
+print(#proto.bcins)

opcodes.lua

@@ -0,0 +1,95 @@
+local bit = require('bit')
+
+local function opcode(id, name, a, b, c, metamethod)
+    return {
+        id=id,
+        name=name,
+        a=a,
+        b=b,
+        c=c,
+        methametod=metamethod,
+    }
+end
+
+local function gen_opcodes(tbl)
+    res = {}
+    for i = 1, #tbl do
+        res[tbl[i].id] = {
+            name = tbl[i].name,
+            a = tbl[i].a,
+            b = tbl[i].b,
+            c = tbl[i].c,
+            metamethod = tbl[i].metamethod,
+        }
+    end
+    return res
+end
+
+local Mode = {
+    none = 0,
+    dst = 1,
+    base = 2,
+    var = 3,
+    rbase = 4,
+    uv = 5,
+    lit = 6,
+    lits = 7,
+    pri = 8,
+    num = 9,
+    str = 10,
+    tab = 11,
+    func = 12,
+    jump = 13,
+    cdata = 14,
+    max = 15,
+    none = 15, -- same as max
+}
+
+local Metamethod = {
+    index = 0,
+    newindex = 1,
+    gc = 2,
+    mode = 3,
+    eq = 4,
+    len = 5,
+    lt = 6,
+    le = 7,
+    concat = 8,
+    call = 9,
+    add = 10,
+    sub = 11,
+    mul = 12,
+    div = 13,
+    mod = 14,
+    pow = 15,
+    unm = 16,
+    metatable = 17,
+    tostring = 18,
+}
+
+local Opcodes_defs = gen_opcodes {
+    opcode(37, 'POW', Mode.dst, Mode.none, Mode.lits, Metamethod.none),
+    opcode(41, 'KSHORT', Mode.dst, Mode.var, Mode.var, Metamethod.pow),
+    opcode(75, 'RET0', Mode.rbase, Mode.none, Mode.lit, Mode.none),
+}
+
+local function decode(ins)
+    -- TODO: endianess
+    local id = bit.band(ins, 0xff)
+    local a = bit.band(ins, 0xff00) / 0x100
+    local b = bit.band(ins, 0xff0000) / 0x10000
+    local c = bit.band(ins, 0xff000000) / 0x1000000
+    local d = bit.band(ins, 0xffff0000) / 0x10000
+    return {
+        id = id,
+        a = a,
+        b = b,
+        c = c,
+        d = d,
+    }
+end
+
+return {
+    defs = Opcodes_defs,
+    decode = decode,
+}

readbc.lua

@@ -5,13 +5,53 @@ local bit=require('bit')
 local HEADER_MAGIC = {27, string.byte('L'), string.byte('J')}
 local HEADER_FLAG_STRIP = 0x02
 
+local function read_u8(fp)
+    return string.byte(fp:read(1))
+end
+
+local function read_u16(fp)
+    -- TODO: endianess
+    local b1 = read_u8(fp)
+    local b2 = read_u8(fp)
+
+    return bit.bor(b2*0x100, b1)
+end
+
+local function read_u32(fp)
+    -- TODO: endianess
+    local h1 = read_u16(fp)
+    local h2 = read_u16(fp)
+
+    return bit.bor(h2*0x10000, h1)
+end
+
+local function read_u8n(fp, n)
+    return { string.byte(fp:read(n), 1, n) }
+end
+
+local function read_u16n(fp, n)
+    local res = {}
+    for i = 1, n do
+        res[i] = read_u16(fp)
+    end
+    return res
+end
+
+local function read_u32n(fp, n)
+    local res = {}
+    for i = 1, n do
+        res[i] = read_u32(fp)
+    end
+    return res
+end
+
 local function read_uleb128(fp)
     local last_byte = false
     local res = 0
     local shift = 0
     -- the bytes are read from the LSB to the MSB
     while not last_byte do
-        local uleb_byte = string.byte(fp:read(1))
+        local uleb_byte = read_u8(fp)
         res = bit.bor(res, bit.lshift(bit.band(uleb_byte, 0x7f), shift))
 
         if bit.band(uleb_byte, 0x80) == 0 then
@@ -22,8 +62,8 @@ local function read_uleb128(fp)
 end
 
 local function read_header(fp)
-    local magic = { string.byte(fp:read(3), 1, 3) }
+    local magic = read_u8n(fp, 3)
-    local version = string.byte(fp:read(1))
+    local version = read_u8(fp)
     local flags = read_uleb128(fp)
 
     local name
@@ -43,6 +83,57 @@ local function read_header(fp)
 end
 
 local function read_proto(fp)
+    local length = read_uleb128(fp)
+    -- if the the last read proto was the last one
+    -- we have reached the end marker, which is a zero
+    -- this overlaps with a uleb128 zero representation
+    if length == 0 then
+        return
+    end
+
+    -- read proto header
+    local flags = read_u8(fp)
+    local numparams = read_u8(fp)
+    local framesize = read_u8(fp)
+    local numuv = read_u8(fp)
+    local numkgc = read_uleb128(fp)
+    local numkn = read_uleb128(fp)
+    local numbc = read_uleb128(fp)
+
+    local debugLen = 0
+    local debugInfo = nil
+
+    if bit.band(flags, HEADER_FLAG_STRIP) ~= 0 then
+        debugLen = read_uleb128(fp)
+
+        -- TODO: is this correct?
+        if debugLen ~= 0 then
+            local firstLine = read_uleb128(fp)
+            local numLine = read_uleb128(fp)
+            debugInfo = {
+                firstLine = firstLine,
+                numLine = numLine,
+            }
+        end
+    end
+
+    local bcins = read_u32n(fp, numbc)
+    local uvdata = read_u16n(fp, numuv)
+    -- local kgc
+    -- local numkn
+
+    return {
+        flags = flags,
+        numparams = numparams,
+        framesize = framesize,
+        numuv = numuv,
+        numkgc = numkgc,
+        numkn = numkn,
+        numbc = numbc,
+        debugInfo = debugInfo,
+        bcins = bcins,
+        uvdata = uvdata,
+    }
 end
 
 return {

vm.lua

@@ -1,73 +1 @@
-local funcbc = require('jit.util').funcbc
 
-function opcode(id, name, a, b, c, metamethod)
-    return {
-        id=id,
-        name=name,
-        a=a,
-        b=b,
-        c=c,
-        methametod=metamethod,
-    }
-end
-
-function gen_opcodes(tbl)
-    length = #tbl
-    res = {}
-    for i=1,length do
-        res[i] = i+1
-    end
-    return res
-end
-
-Mode = {
-    none = 0,
-    dst = 1,
-    base = 2,
-    var = 3,,
-    rbase = 4,
-    uv = 5,
-    lit = 6,
-    lits = 7,
-    pri = 8,
-    num = 9,
-    str = 10,
-    tab = 11,
-    func = 12,
-    jump = 13,
-    cdata = 14,
-    max = 15,
-    none = 15, -- same as max
-}
-
-Metamethod = {
-    index = 0,
-    newindex = 1,
-    gc = 2,
-    mode = 3,
-    eq = 4,
-    len = 5,
-    lt = 6,
-    le = 7,
-    concat = 8,
-    call = 9,
-    add = 10,
-    sub = 11,
-    mul = 12,
-    div = 13,
-    mod = 14,
-    pow = 15,
-    unm = 16,
-    metatable = 17,
-    tostring = 18,
-}
-
-Opcodes = gen_opcodes {
-    opcode(37, 'POW', Mode.dst, Mode.none, Mode.lits, Metamethod.none),
-    opcode(41, 'KSHORT', Mode.dst, Mode.var, Mode.var, Metamethod.pow),
-}
-
-print(Opcodes[1])
-ins, m = funcbc(test1, 3) -- TODO: what is m?
-print(ins)
-print(m)