vt1200/dwt/loader/coff.py

#!/usr/bin/env python3
"""coff.py -- TI TMS34010 GSP COFF reader for the VT1200 loader work.

GSPLNK 2.00 emits the "TI COFF" described in the 1987 Assembler Tools guide,
Appendix A.  All section addresses and sizes in the section headers are in
BITS (the TMS34010 is bit-addressed); file offsets (scnptr/relptr/symptr) are
in BYTES.

Commands
    coff.py summary FILE...      dump header, optional header, sections,
                                symbols (with aux entries) and relocations
    coff.py raw ABS1 ABS2        extract each section from two absolute
                                outputs and report byte differences
    coff.py manifest FILE        print shell-ish KEY=VALUE lines for one file

The loader itself parses the same structures in assembly; this script is the
reference and the evidence generator.
"""
import struct
import sys

MAGIC = 0x0090
OPTHDR_MAGIC = 0x0108

RELOC_NAMES = {
    0x00: "R-ABS",
    0x10: "R-RELWORD",
    0x11: "R-RELLONG",
    0x18: "R-OCRLONG",
    0x19: "R-GSPPCR16",
}

SECTION_FLAGS = {
    0x20: "text",
    0x40: "data",
    0x80: "bss",
}


class CoffError(Exception):
    pass


def parse(path):
    d = open(path, "rb").read()
    if len(d) < 20:
        raise CoffError("%s: shorter than a COFF file header" % path)
    magic, nscns, timdat, symptr, nsyms, opthdr, flags = struct.unpack_from(
        "<HHIIIHH", d, 0)
    if magic != MAGIC:
        raise CoffError("%s: magic %#06x, expected %#06x"
                        % (path, magic, MAGIC))
    opt = None
    if opthdr:
        if 20 + opthdr > len(d):
            raise CoffError("%s: optional header past EOF" % path)
        # TI COFF optional header: magic, vstamp, textsize, datasize, bsssize,
        # entry, textstart, datastart (all sizes/addresses in bits).
        opt = struct.unpack_from("<HHIIIIII", d, 20)
    base = 20 + opthdr
    secs = []
    for i in range(nscns):
        off = base + i * 40
        if off + 40 > len(d):
            raise CoffError("%s: section header %d past EOF" % (path, i))
        raw = d[off:off + 40]
        name = raw[:8].rstrip(b"\0").decode("latin1")
        (paddr, vaddr, size, scnptr, relptr, lnnoptr, nreloc, nlnno,
         sflags) = struct.unpack_from("<IIIIIIHHI", d, off + 8)
        relocs = []
        for r in range(nreloc):
            ro = relptr + r * 10
            if ro + 10 > len(d):
                raise CoffError("%s: relocation %d past EOF" % (path, r))
            raddr, symidx, rtype = struct.unpack_from("<IIH", d, ro)
            relocs.append((raddr, symidx, rtype))
        secs.append(dict(index=i + 1, name=name, paddr=paddr, vaddr=vaddr,
                         size=size, scnptr=scnptr, relptr=relptr,
                         lnnoptr=lnnoptr, nreloc=nreloc, nlnno=nlnno,
                         flags=sflags, relocs=relocs, raw=raw))
    syms = []
    i = 0
    while i < nsyms:
        off = symptr + i * 18
        if off + 18 > len(d):
            raise CoffError("%s: symbol %d past EOF" % (path, i))
        e = d[off:off + 18]
        name = e[:8].rstrip(b"\0").decode("latin1", "replace")
        val, secno, stype, sclass, naux = struct.unpack_from("<IHhBB", e, 8)
        aux = d[off + 18:off + 18 + naux * 18]
        syms.append(dict(index=i, name=name, value=val, secno=secno,
                         type=stype, sclass=sclass, naux=naux, aux=aux))
        i += 1 + naux
    return dict(path=path, size=len(d), magic=magic, nscns=nscns,
                timdat=timdat, symptr=symptr, nsyms=nsyms, opthdr=opthdr,
                flags=flags, opt=opt, sections=secs, symbols=syms, data=d)


def summary(path):
    c = parse(path)
    print("== %s  (%d bytes)" % (path, c["size"]))
    print("   file hdr: magic=%#06x nscns=%d symptr=%#x nsyms=%d "
          "opthdr=%d flags=%#06x"
          % (c["magic"], c["nscns"], c["symptr"], c["nsyms"], c["opthdr"],
             c["flags"]))
    if c["opt"]:
        magic, vstamp, tsize, dsize, bsize, entry, tstart, dstart = c["opt"]
        print("   optional: magic=%#06x vstamp=%#06x text=%#x bits "
              "data=%#x bits bss=%#x bits entry=%#010x textstart=%#010x "
              "datastart=%#010x"
              % (magic, vstamp, tsize, dsize, bsize, entry, tstart, dstart))
        print("   optional: sizes in bytes text=%d data=%d bss=%d; "
              "entry in bits=%s"
              % (tsize // 8, dsize // 8, bsize // 8,
                 "0 (no -e global symbol)" if entry == 0 else hex(entry)))
    for s in c["sections"]:
        fl = ",".join(v for k, v in SECTION_FLAGS.items() if s["flags"] & k)
        print("   sec %d %-8s paddr=%010x vaddr=%010x size=%#x bits (%d B) "
              "scnptr=%#x nreloc=%d flags=%#x%s"
              % (s["index"], s["name"], s["paddr"], s["vaddr"], s["size"],
                 s["size"] // 8, s["scnptr"], s["nreloc"], s["flags"],
                 (" [" + fl + "]") if fl else ""))
        for raddr, symidx, rtype in s["relocs"]:
            print("       reloc at %#x sym=%d type=%#x %s"
                  % (raddr, symidx, rtype,
                     RELOC_NAMES.get(rtype, "?")))
    for sym in c["symbols"]:
        if sym["naux"]:
            print("   sym[%d] %-10s value=%010x sec=%d class=%d naux=%d"
                  % (sym["index"], sym["name"], sym["value"], sym["secno"],
                     sym["sclass"], sym["naux"]))
        else:
            print("   sym[%d] %-10s value=%010x sec=%d class=%d"
                  % (sym["index"], sym["name"], sym["value"], sym["secno"],
                     sym["sclass"]))
    print()


def section_bytes(c, name):
    for s in c["sections"]:
        if s["name"] == name:
            return s, c["data"][s["scnptr"]:s["scnptr"] + s["size"] // 8]
    return None, None


def raw(paths):
    if len(paths) != 2:
        raise SystemExit("coff.py raw: need exactly two files")
    a, b = parse(paths[0]), parse(paths[1])
    for name in (".text", ".data", ".bss"):
        sa, ba = section_bytes(a, name)
        sb, bb = section_bytes(b, name)
        if sa is None or sb is None:
            continue
        print("== %s: %d vs %d bytes" % (name, len(ba), len(bb)))
        n = min(len(ba), len(bb))
        bad = [i for i in range(n) if ba[i] != bb[i]]
        print("   differing bytes: %d" % (len(bad) + abs(len(ba) - len(bb))))
        for i in bad[:24]:
            print("     %#06x: %02x -> %02x" % (i, ba[i], bb[i]))
    print()


def manifest(path):
    c = parse(path)
    print("COFF=%s" % path)
    print("SIZE=%d" % c["size"])
    print("NSECTIONS=%d" % c["nscns"])
    if c["opt"]:
        _, _, tsize, dsize, bsize, entry, tstart, dstart = c["opt"]
        print("ENTRY=%#010x" % entry)
        print("TEXTSTART=%#010x" % tstart)
        print("DATASTART=%#010x" % dstart)
        print("TEXTSIZE_BITS=%d" % tsize)
        print("DATASIZE_BITS=%d" % dsize)
        print("BSSSIZE_BITS=%d" % bsize)
    for s in c["sections"]:
        key = s["name"].lstrip(".").upper() or ("SEC%d" % s["index"])
        print("%s_PADDR=%#010x" % (key, s["paddr"]))
        print("%s_SIZE_BITS=%d" % (key, s["size"]))
        print("%s_NRELOC=%d" % (key, s["nreloc"]))


def main(argv):
    if len(argv) < 3:
        raise SystemExit(__doc__)
    cmd = argv[1]
    if cmd == "summary":
        for p in argv[2:]:
            summary(p)
    elif cmd == "raw":
        raw(argv[2:])
    elif cmd == "manifest":
        manifest(argv[2])
    else:
        raise SystemExit(__doc__)


if __name__ == "__main__":
    main(sys.argv)