Retro/pcjs
Watch
1
0
Fork
You've already forked pcjs
 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Disassembled Compaq BIOS can now be reassembled

Browse source
This commit is contained in:
Jeff Parsons 2015-04-06 16:02:27 -07:00 • committed by jeffpar
commit fd323fc9c7
4 changed files with 14396 additions and 15195 deletions
Download patch file Download diff file Expand all files Collapse all files

201
modules/textout/lib/textout.js
View file

@ -47,13 +47,14 @@ var str = require("../../shared/lib/strlib");
function TextOut()
{
this.fDebug = false;
this.fASCII = true;
this.sServerRoot = process.cwd();
this.asLines = [];
this.nTabWidth = 8;
this.sTarget = "; ";
}
TextOut.asTargetRefs = ["call", "jmp", "jz", "jnz", "jc", "jnc", "ja", "jna", "js", "jns", "jo", "jno", "jl", "jnl", "jg", "jng", "jpo", "jpe"];
TextOut.asTargetRefs = ["call", "jmp", "jz", "jnz", "jc", "jnc", "ja", "jna", "js", "jns", "jo", "jno", "jl", "jnl", "jg", "jng", "jpo", "jpe", "loop", "loope", "loopne", "jcxz", "jecxz"];
/*
* Class methods
@ -72,13 +73,15 @@ TextOut.asTargetRefs = ["call", "jmp", "jz", "jnz", "jc", "jnc", "ja", "jna", "j
* ---
* --nasm performs a variety of NASM-related processing, including:
*
* collapseRepeated(): looks for series of lines containing nothing more than a "db",
* "dw", or something equivalent, and collapses them into a single repetition.
* massageLines(): reorders basic elements of every line to make them "assemble-able"
*
* labelTargets(): looks for all JMP and CALL targets and labels them.
* collapseLines(): looks for series of lines containing nothing more than a "db",
* "dw", or something equivalent, and collapses them into a single repetition
*
* labelTargets(): looks for all JMP and CALL targets and labels them
*
* alignVertical(): looks for a predefined target string (eg, '; ') in the first line,
* and ensures that the same sequence in all subsequent lines starts at the same column.
* and ensures that the same sequence in all subsequent lines starts at the same column
*
* For now, all output is written to stdout only.
*
@ -106,7 +109,8 @@ TextOut.CLI = function()
text.loadFile(sFile, function(err) {
if (!err) {
if (argv['nasm']) {
text.collapseRepeated();
text.massageLines();
text.collapseLines();
text.labelTargets();
text.alignVertical();
}
@ -198,26 +202,131 @@ TextOut.prototype.setText = function(buf)
};
/**
* collapseRepeated()
* massageLines()
*
* @this {TextOut}
*/
TextOut.prototype.collapseRepeated = function()
TextOut.prototype.massageLines = function()
{
var iLineASCII = -1, sASCII = "";
for (var iLine = 0; iLine < this.asLines.length; iLine++) {
/*
* NDISASM sometimes inserts spurious lines containing a hyphen immediately followed by one or more hex bytes;
* this seems to happen whenever an instruction longer than 8 bytes is encountered, and it appears that these
* extra lines can be completely removed.
*/
var sLine = this.asLines[iLine];
if (sLine.match(/^\s*-[0-9A-F]+$/)) {
this.asLines.splice(iLine--, 1);
continue;
}
sLine = sLine.replace(/^([0-9A-F]+)\s+([0-9A-F]+)\s+(o32 |a32 |)([^\s]*) *(.*)$/, "\t$3$4\t$5\t\t\t; $1 $2");
var fASCII = false;
var match = sLine.match(/([0-9A-F]+)$/);
if (match) {
fASCII = this.fASCII;
var cBytes = 0, c;
var sBytes = match[1], sASCIILine = "";
for (var i = 0; i < sBytes.length; i += 2) {
c = str.parseInt(sBytes.substr(i, 2));
if (c != 0x0D && c != 0x0A && (c < 0x20 || c >= 0x7F)) {
c = 0x2E;
fASCII = false;
}
sASCIILine += String.fromCharCode(c);
cBytes++;
}
/*
* Don't interpret single-byte opcodes within the ASCII range as potential ASCII?
*
* if (cBytes == 1 && (c >= 0x40 && c <= 0x5F)) fASCII = false;
*/
sLine += " " + this.encodeASCII(sASCIILine);
}
this.asLines[iLine] = sLine;
if (fASCII) {
if (iLineASCII < 0) iLineASCII = iLine;
sASCII += sASCIILine;
continue;
}
if (iLineASCII >= 0) {
if (iLine - iLineASCII >= 4) {
/*
* This seems a better way of discriminating between single-byte opcodes and ASCII:
* output ASCII only when the number of ASCII bytes exceeds the number of opcode bytes.
*/
if (sASCII.length > iLine - iLineASCII + 1) {
this.asLines[iLineASCII] = "\tdb\t" + this.encodeASCII(sASCII);
this.asLines.splice(iLineASCII+1, iLine - iLineASCII - 1);
iLine = iLineASCII;
}
}
iLineASCII = -1; sASCII = "";
}
}
};
/**
* encodeASCII(s)
*
* @param s
* @return {string}
*/
TextOut.prototype.encodeASCII = function(s)
{
var sNew = "", fInQuotes = false;
for (var i = 0; i < s.length; i++) {
var c = s.charCodeAt(i);
if (c < 0x20 || c == 0x27) {
if (fInQuotes) {
sNew += "'";
fInQuotes = false;
}
if (sNew) sNew += ',';
sNew += str.toHexByte(c);
continue;
}
if (!fInQuotes) {
if (sNew) sNew += ',';
sNew += "'";
fInQuotes = true;
}
var ch = String.fromCharCode(c);
if (ch == "'") ch = "\\'";
sNew += ch;
}
if (fInQuotes) sNew += "'";
return sNew;
};
/**
* collapseLines()
*
* @this {TextOut}
*/
TextOut.prototype.collapseLines = function()
{
for (var i = 0; i < this.asLines.length; i++) {
var as = this.getLineParts(i);
var as = this.getLineParts(i, true);
if (!as) continue;
if (as[2] != "db" && as[2] != "dw") continue;
var cCombine = 0, asLast;
for (var j = i + 1; j < this.asLines.length; j++) {
var asNext = this.getLineParts(j);
var asNext = this.getLineParts(j, true);
if (!asNext) break;
if (as[2] != asNext[2] || as[2] != asNext[2]) break;
asLast = asNext;
cCombine++;
}
if (cCombine > 2) {
this.asLines[i] = "\n\ttimes\t" + (cCombine + 1) + ' ' + as[2] + ' ' + as[3] + "\t\t; " + as[4] + " - " + asLast[4] + '\n';
this.asLines[i] = "\n\ttimes\t" + (cCombine + 1) + ' ' + as[2] + ' ' + as[3] + "\t\t; " + as[4] + " - " + asLast[4];
this.asLines.splice(i + 1, cCombine);
}
}
@ -233,6 +342,7 @@ TextOut.prototype.labelTargets = function()
/*
* First pass: find all target references (eg, JMP and CALL instructions)
*/
var chPrefix = 'x';
var i, j, as, target, aTargets = [], aHardTargets = [];
for (i = 0; i < this.asLines.length; i++) {
@ -240,14 +350,25 @@ TextOut.prototype.labelTargets = function()
if (!as) continue;
var iTarget = TextOut.asTargetRefs.indexOf(as[2]);
if (iTarget < 0) continue;
target = str.parseInt(as[3]);
var sTarget = as[3];
var sShort = "short ";
if (sTarget.indexOf(sShort) !== 0) {
sShort = "";
} else {
sTarget = sTarget.substr(sShort.length);
}
target = str.parseInt(sTarget);
if (target == undefined) continue;
if (aTargets.indexOf(target) < 0) {
aTargets.push(target);
if (iTarget < 2) aHardTargets.push(target);
/*
* For now, we're classifying only "call" targets as "hard" targets, and thus worthy of extra whitespace
*/
if (iTarget < 1) aHardTargets.push(target);
}
this.asLines[i] = this.asLines[i].replace(as[3], 'l' + target.toString(16));
this.asLines[i] = this.asLines[i].replace(as[3], sShort + chPrefix + target.toString(16));
}
/*
* Second pass: label all targets
*/
@ -260,48 +381,41 @@ TextOut.prototype.labelTargets = function()
j = aTargets.indexOf(addr);
if (j >= 0) {
var fHard = (aHardTargets.indexOf(addr) >= 0);
this.asLines[i] = (fHard || fPrevHard? "\nl" : 'l') + addr.toString(16) + ':' + this.asLines[i];
this.asLines[i] = (fHard || fPrevHard? '\n' : '') + chPrefix + addr.toString(16) + ':' + this.asLines[i];
aTargets.splice(j, 1);
} else {
if (fPrevHard) this.asLines[i] = '\n' + this.asLines[i];
}
fPrevHard = (as[2] == "jmp" || as[2] == "ret" || as[2] == "retf" || as[2] == "iret");
}
/*
* Third pass: for all targets that turned out to NOT be targets, fix all references
*/
var aMissingTargets = [];
var aRepairs = [];
if (aTargets.length) {
for (i = 0; i < this.asLines.length; i++) {
as = this.getLineParts(i);
if (!as) continue;
if (as[3].charAt(0) == 'l') {
if (as[3].charAt(0) == chPrefix) {
addr = str.parseInt(as[3].substr(1));
if (aTargets.indexOf(addr) >= 0) {
this.asLines[i] = this.asLines[i].replace(as[3], str.toHexWord(addr));
if (aMissingTargets.indexOf(addr) < 0) aMissingTargets.push(addr);
/*
* Instead of putting back the original target address, let's just convert the line to a "db"
*/
this.asLines[i] = as[1] + "\tdb\t";
for (j = 0; j < as[5].length; j += 2) {
this.asLines[i] += (j > 0? ',' : '') + "0x" + as[5].substr(j, 2);
}
// this.asLines[i] = this.asLines[i].replace(as[3], str.toHexWord(addr));
if (aRepairs.indexOf(addr) < 0) aRepairs.push(addr);
}
}
if (as[5]) {
var sASCII = "";
var sBytes = as[5];
for (j = 0; j < sBytes.length-1; j+=2) {
var k = str.parseInt(sBytes.substr(j, 2));
sASCII += (k < 0x20 || k >= 0x7f)? '.' : String.fromCharCode(k);
}
sBytes = " " + sBytes;
/*
* TODO: Determine why replace() fails when there's a trailing '$' in the replacement string;
* I'm working around it by appending a space.
*/
if (sASCII.slice(-1) == '$') sASCII += ' ';
this.asLines[i] = this.asLines[i].replace(sBytes, sBytes + " '" + sASCII + "'");
}
}
if (aTargets.length != aMissingTargets.length) {
console.log("; warning: " + aTargets.length + " unprocessed targets (" + aMissingTargets.length + " repaired):");
if (aTargets.length != aRepairs.length) {
console.log("; warning: " + aTargets.length + " unprocessed targets (" + aRepairs.length + " repaired):");
for (j = 0; j < aTargets.length; j++) {
if (aMissingTargets.indexOf(aTargets[j]) < 0) console.log(';\t' + str.toHexWord(aTargets[j]));
if (aRepairs.indexOf(aTargets[j]) < 0) console.log(';\t' + str.toHexWord(aTargets[j]));
}
}
}
@ -378,7 +492,7 @@ TextOut.prototype.findTarget = function(sSrc, fDebug)
};
/**
* getLineParts(iLine)
* getLineParts(iLine, fBogus)
*
* Returns the following array:
*
@ -387,17 +501,22 @@ TextOut.prototype.findTarget = function(sSrc, fDebug)
* asParts[3]: operand(s)
* asParts[4]: offset
* asParts[5]: byte sequence
* asParts[6]: ASCII sequence, if any
*
* or null if the line does not contain all of the above.
*
* @this {TextOut}
* @param {number} iLine
* @param {boolean} [fBogus]
* @return {Array.<string>}
*/
TextOut.prototype.getLineParts = function(iLine)
TextOut.prototype.getLineParts = function(iLine, fBogus)
{
var as = this.asLines[iLine].match(/^\n?([^\s:]+:|)\s*([^\s;]+)\s*([^;]*?)\s*;\s*([0-9A-F]+)\s*([0-9A-F]+)\s*$/);
if (as) {
var as = this.asLines[iLine].match(/^\n?([^\s:]+:|)\s*([^\s;]+)\s*([^;]*?)\s*;\s*([0-9A-F]+)\s*([0-9A-F]+)\s*([^\s]+|)$/);
if (as && fBogus) {
/*
* In most (but not all) cases, an "add [bx+si],al" instruction is bogus (TODO: Come up with an actual bogosity test)
*/
if (as[2] == "add" && as[3] == "[bx+si],al") {
as[2] = "dw";
as[3] = "0x0000";