wdc-utils/dumpobj.cpp

666 lines
14 KiB
C++
Raw Normal View History

2016-12-24 19:58:55 +00:00
#include <string>
#include <err.h>
#include <unistd.h>
#include <sysexits.h>
#include <stdlib.h>
#include <fcntl.h>
#include <assert.h>
#include <stdio.h>
2017-01-03 18:09:10 +00:00
#include <string.h>
2016-12-24 19:58:55 +00:00
#include <type_traits>
#include <vector>
2016-12-25 07:42:43 +00:00
#include <algorithm>
#include <iterator>
2016-12-24 19:58:55 +00:00
#include "obj816.h"
#include "zrdz_disassembler.h"
2016-12-24 19:58:55 +00:00
2016-12-26 16:03:06 +00:00
#ifndef O_BINARY
#define O_BINARY 0
#endif
2017-01-01 17:34:45 +00:00
/* not documented */
#ifndef SEC_DATA
#define SEC_DATA 0x80
#endif
2017-01-04 02:23:45 +00:00
struct {
bool _S = false;
bool _g = false;
bool _n = false;
2017-01-04 02:23:45 +00:00
} flags;
2016-12-24 19:58:55 +00:00
enum class endian {
little = __ORDER_LITTLE_ENDIAN__,
big = __ORDER_BIG_ENDIAN__,
native = __BYTE_ORDER__
};
template<class T>
void swap_if(T &t, std::false_type) {}
void swap_if(uint8_t &, std::true_type) {}
void swap_if(uint16_t &value, std::true_type) {
value = __builtin_bswap16(value);
}
void swap_if(uint32_t &value, std::true_type) {
value = __builtin_bswap32(value);
}
void swap_if(uint64_t &value, std::true_type) {
value = __builtin_bswap64(value);
}
template<class T>
void le_to_host(T &value) {
swap_if(value, std::integral_constant<bool, endian::native == endian::big>{});
}
void usage() {
exit(EX_USAGE);
}
#pragma pack(push, 1)
struct Header {
uint32_t magic; /* magic number for detection */
uint16_t version; /* version number of object format */
2017-01-08 03:17:30 +00:00
uint8_t filetype; /* file type, object or library */
2016-12-24 19:58:55 +00:00
};
#pragma pack(pop)
template<class T>
uint8_t read_8(T &iter) {
uint8_t tmp = *iter;
++iter;
return tmp;
}
template<class T>
uint16_t read_16(T &iter) {
uint16_t tmp = 0;
tmp |= *iter << 0;
++iter;
tmp |= *iter << 8;
++iter;
return tmp;
}
template<class T>
2017-01-04 02:23:45 +00:00
uint32_t read_32(T &iter) {
2016-12-24 19:58:55 +00:00
uint32_t tmp = 0;
tmp |= *iter << 0;
++iter;
tmp |= *iter << 8;
++iter;
tmp |= *iter << 16;
++iter;
tmp |= *iter << 24;
++iter;
return tmp;
}
template<class T>
std::string read_cstring(T &iter) {
std::string s;
for(;;) {
uint8_t c = *iter;
++iter;
if (!c) break;
s.push_back(c);
}
return s;
}
template<class T>
std::string read_pstring(T &iter) {
std::string s;
unsigned size = *iter;
++iter;
s.reserve(size);
while (size--) {
uint8_t c = *iter;
++iter;
s.push_back(c);
}
return s;
}
2016-12-25 07:42:43 +00:00
2016-12-28 20:15:31 +00:00
std::vector<section> read_sections(const std::vector<uint8_t> &section_data) {
std::vector<section> sections;
auto iter = section_data.begin();
while (iter != section_data.end()) {
section s;
s.number = read_8(iter);
s.flags = read_8(iter);
s.size = read_32(iter);
s.org = read_32(iter);
if (!(s.flags & SEC_NONAME)) s.name = read_cstring(iter);
sections.emplace_back(std::move(s));
2016-12-28 20:15:31 +00:00
}
return sections;
}
2016-12-25 07:42:43 +00:00
std::vector<symbol> read_symbols(const std::vector<uint8_t> &symbol_data) {
std::vector<symbol> symbols;
auto iter = symbol_data.begin();
while (iter != symbol_data.end()) {
2016-12-28 20:15:31 +00:00
symbol s;
s.type = read_8(iter);
s.flags = read_8(iter);
s.section = read_8(iter);
s.offset = s.type == S_UND ? 0 : read_32(iter);
s.name = read_cstring(iter);
2016-12-25 07:42:43 +00:00
2016-12-28 20:15:31 +00:00
symbols.emplace_back(std::move(s));
2016-12-25 07:42:43 +00:00
}
return symbols;
}
2016-12-26 16:15:23 +00:00
bool dump_obj(const char *name, int fd)
2016-12-24 19:58:55 +00:00
{
Mod_head h;
ssize_t ok;
ok = read(fd, &h, sizeof(h));
2016-12-26 16:15:23 +00:00
if (ok == 0) return false;
2016-12-24 19:58:55 +00:00
if (ok != sizeof(h))
errx(EX_DATAERR, "%s is not an object file", name);
le_to_host(h.h_magic);
le_to_host(h.h_version);
le_to_host(h.h_filtyp);
le_to_host(h.h_namlen);
le_to_host(h.h_recsize);
le_to_host(h.h_secsize);
le_to_host(h.h_symsize);
le_to_host(h.h_optsize);
le_to_host(h.h_tot_secs);
le_to_host(h.h_num_secs);
le_to_host(h.h_num_syms);
assert(h.h_magic == MOD_MAGIC);
assert(h.h_version == 1);
assert(h.h_filtyp == 1);
// now read the name (h_namlen includes 0 terminator.)
std::vector<char> oname;
oname.resize(h.h_namlen);
ok = read(fd, oname.data(), h.h_namlen);
if (ok != h.h_namlen) errx(EX_DATAERR, "%s", name);
// records [until record_eof]
std::vector<uint8_t> data;
data.resize(h.h_recsize);
ok = read(fd, data.data(), h.h_recsize);
2016-12-25 06:02:07 +00:00
if (ok != h.h_recsize) errx(EX_DATAERR, "%s records truncated", name);
std::vector<uint8_t> section_data;
section_data.resize(h.h_secsize);
ok = read(fd, section_data.data(), h.h_secsize);
if (ok != h.h_secsize) errx(EX_DATAERR, "%s sections truncated", name);
std::vector<uint8_t> symbol_data;
symbol_data.resize(h.h_symsize);
ok = read(fd, symbol_data.data(), h.h_symsize);
if (ok != h.h_symsize) errx(EX_DATAERR, "%s symbols truncated", name);
zrdz_disassembler d(read_sections(section_data), read_symbols(symbol_data));
2016-12-24 19:58:55 +00:00
2016-12-28 20:15:31 +00:00
uint8_t op = REC_END;
2016-12-27 21:24:01 +00:00
unsigned line = 0;
d.front_matter(std::string(oname.data()));
2016-12-24 19:58:55 +00:00
auto iter = data.begin();
while (iter != data.end()) {
2016-12-25 07:42:43 +00:00
2016-12-24 19:58:55 +00:00
op = read_8(iter);
if (op == 0) break;
if (op < 0xf0) {
2016-12-25 17:35:04 +00:00
auto end = iter + op;
while (iter != end) {
2016-12-26 16:15:23 +00:00
d(*iter++);
2016-12-25 17:35:04 +00:00
}
2016-12-24 19:58:55 +00:00
continue;
}
switch(op) {
2016-12-27 02:13:58 +00:00
case REC_RELEXP:
2016-12-24 19:58:55 +00:00
case REC_EXPR:
{
2016-12-27 02:13:58 +00:00
// todo -- pass the relative flag to ()
// so it can verify it's appropriate for the opcode.
// todo -- move all this stuff to a separate function.
uint8_t size = read_8(iter);
2016-12-24 19:58:55 +00:00
2016-12-25 06:02:07 +00:00
std::vector<std::string> stack;
// todo -- need to keep operation for precedence?
// this ignores all precedence...
2016-12-24 19:58:55 +00:00
for(;;) {
2016-12-27 02:13:58 +00:00
uint8_t op = read_8(iter);
2016-12-24 19:58:55 +00:00
if (op == OP_END) break;
switch (op) {
case OP_LOC: {
2016-12-25 06:02:07 +00:00
uint8_t section = read_8(iter);
uint32_t offset = read_32(iter);
2016-12-25 07:42:43 +00:00
std::string name;
if (flags._n) {
name = d.section_name(section) + "+" + d.to_x(offset, 4, '$');
} else {
name = d.location_name(section, offset);
}
stack.emplace_back(std::move(name));
2016-12-24 19:58:55 +00:00
break;
2016-12-25 06:02:07 +00:00
}
2016-12-24 19:58:55 +00:00
case OP_VAL:
stack.push_back(d.to_x(read_32(iter), 4, '$'));
2016-12-24 19:58:55 +00:00
break;
2016-12-25 06:02:07 +00:00
2016-12-25 07:42:43 +00:00
case OP_SYM: {
uint16_t symbol = read_16(iter);
stack.emplace_back(d.symbol_name(symbol));
2016-12-25 06:02:07 +00:00
break;
2016-12-25 07:42:43 +00:00
}
2016-12-25 06:02:07 +00:00
2016-12-27 02:13:58 +00:00
// unary operatos
case OP_NOT:
case OP_NEG:
case OP_FLP: {
static const char *ops[] = {
2016-12-28 20:15:31 +00:00
".NOT.", "-", "\\"
2016-12-27 02:13:58 +00:00
};
if (stack.empty()) errx(EX_DATAERR, "%s : stack underflow error", name);
std::string a = std::move(stack.back()); stack.pop_back();
std::string b(ops[op-10]);
stack.emplace_back(b + a);
break;
}
// binary operators
2016-12-25 06:02:07 +00:00
case OP_SHR:
case OP_SHL:
case OP_ADD:
case OP_SUB: {
static const char *ops[] = {
2016-12-27 02:13:58 +00:00
"**", "*", "/", ".MOD.", ">>", "<<", "+", "-", "&", "|", "^", "=", ">", "<"
2016-12-25 06:02:07 +00:00
};
if (stack.size() < 2) errx(EX_DATAERR, "%s : stack underflow error", name);
std::string a = std::move(stack.back()); stack.pop_back();
std::string b = std::move(stack.back()); stack.pop_back();
stack.emplace_back(b + ops[op-20] + a);
2016-12-24 19:58:55 +00:00
break;
2016-12-25 06:02:07 +00:00
}
2016-12-24 19:58:55 +00:00
default:
errx(EX_DATAERR, "%s: unknown expression opcode %02x", name, op);
}
}
2016-12-25 06:02:07 +00:00
if (stack.size() != 1) errx(EX_DATAERR, "%s stack overflow error.", name);
2016-12-27 02:13:58 +00:00
d(stack.front(), size);
2016-12-24 19:58:55 +00:00
}
break;
case REC_DEBUG:
{
2016-12-25 05:04:17 +00:00
d.flush();
2016-12-24 19:58:55 +00:00
uint16_t size = read_16(iter);
2016-12-27 21:24:01 +00:00
auto end = iter + size;
while (iter < end) {
2016-12-24 19:58:55 +00:00
uint8_t op = read_8(iter);
switch(op) {
case D_LONGA_ON:
2016-12-25 05:04:17 +00:00
d.set_m(true);
d.emit("", "longa", "on");
2016-12-24 19:58:55 +00:00
break;
case D_LONGA_OFF:
2016-12-25 05:04:17 +00:00
d.set_m(false);
d.emit("", "longa", "off");
2016-12-24 19:58:55 +00:00
break;
case D_LONGI_ON:
2016-12-25 05:04:17 +00:00
d.set_x(true);
d.emit("", "longi", "on");
2016-12-24 19:58:55 +00:00
break;
case D_LONGI_OFF:
2016-12-25 05:04:17 +00:00
d.set_x(false);
d.emit("", "longi", "off");
2016-12-24 19:58:55 +00:00
break;
case D_C_FILE: {
2017-01-05 17:25:09 +00:00
std::string file = read_cstring(iter);
2016-12-27 21:24:01 +00:00
line = read_16(iter);
2017-01-02 19:14:51 +00:00
std::string tmp = file + ", " + std::to_string(line);
d.emit("", ".file", tmp);
2016-12-24 19:58:55 +00:00
break;
}
case D_C_LINE: {
2016-12-27 21:24:01 +00:00
line = read_16(iter);
d.emit("",".line", std::to_string(line));
2016-12-24 19:58:55 +00:00
break;
}
case D_C_BLOCK: {
uint16_t block = read_16(iter);
2017-01-05 19:14:22 +00:00
d.emit("",".block", std::to_string(block));
2016-12-24 19:58:55 +00:00
break;
}
2016-12-31 02:59:15 +00:00
case D_C_ENDBLOCK: {
2017-01-05 19:14:22 +00:00
uint16_t block = read_16(iter);
d.emit("",".endblock", std::to_string(block));
2016-12-31 02:59:15 +00:00
break;
}
2016-12-26 16:03:06 +00:00
case D_C_FUNC: {
uint16_t arg = read_16(iter);
2017-01-05 19:14:22 +00:00
d.emit("",".function", std::to_string(arg));
2016-12-26 16:03:06 +00:00
break;
}
2016-12-31 02:59:15 +00:00
case D_C_ENDFUNC: {
uint16_t line = read_16(iter);
uint16_t local_offset = read_16(iter);
uint16_t arg_offset = read_16(iter);
2017-01-02 19:14:51 +00:00
std::string tmp;
tmp = std::to_string(line) + ", "
+ std::to_string(local_offset) + ", "
+ std::to_string(arg_offset);
d.emit("",".endfunc", tmp);
2016-12-31 02:59:15 +00:00
break;
}
2016-12-31 20:31:05 +00:00
// etag? reserved for enums but not actually used?
case D_C_STAG:
case D_C_ETAG:
case D_C_UTAG: {
const char *kOpNames[] = { ".stag", ".etag", ".utag" };
const char *opname = kOpNames[op - D_C_STAG];
std::string name = read_cstring(iter);
uint16_t size = read_16(iter);
uint16_t tag = read_16(iter);
2017-01-02 19:14:51 +00:00
std::string tmp;
tmp = name + ", " + std::to_string(size) + ", " + std::to_string(tag);
d.emit("", opname, tmp);
2016-12-31 20:31:05 +00:00
break;
}
case D_C_EOS: {
d.emit("", ".eos");
2016-12-31 20:31:05 +00:00
break;
}
case D_C_MEMBER:
2016-12-24 19:58:55 +00:00
case D_C_SYM: {
2016-12-31 02:59:15 +00:00
// warning - i don't fully understand this one..
std::string name = read_cstring(iter);
uint8_t version = read_8(iter); //???
uint32_t value;
if (version == 0) value = read_16(iter); // symbol
if (version == 1) value = read_32(iter); // numeric value.
assert(version == 0 || version == 1);
uint32_t type = read_32(iter);
uint8_t klass = read_8(iter);
uint16_t size = read_16(iter);
2016-12-31 20:31:05 +00:00
const char *opname = ".sym";
if (op == D_C_MEMBER) opname = ".member";
2017-01-02 19:14:51 +00:00
std::string attr;
2016-12-31 02:59:15 +00:00
if (version == 0) {
std::string svalue;
svalue = d.symbol_name(value);
2017-01-02 19:14:51 +00:00
attr = name + ", " + svalue;
2016-12-31 02:59:15 +00:00
}
2017-01-02 19:14:51 +00:00
if (version == 1) {
attr = name + ", " + std::to_string(value);
}
attr += ", " + std::to_string(type);
attr += ", " + std::to_string(klass);
attr += ", " + std::to_string(size);
2016-12-31 02:59:15 +00:00
/*
* type bits 1 ... 5 are T_xxxx
* then 3 bits of DT_xxx (repeatedly)
*
2016-12-31 21:31:16 +00:00
* eg, char ** = (DT_PTR << 11) + (DT_PTR << 8) + T_CHAR
*/
int t = type & 0x1f;
if ((t == T_STRUCT) || (t == T_UNION)) {
uint16_t tag = read_16(iter);
2017-01-02 19:14:51 +00:00
attr += ", " + std::to_string(tag);
}
// need to do it until t == 0 for
// multidimensional arrays.
for ( t = type >> 5; t; t >>= 3) {
if ((t & 0x07) == DT_ARY) {
uint16_t dim = read_16(iter);
2017-01-02 19:14:51 +00:00
attr += ", " + std::to_string(dim);
}
}
d.emit("", opname, attr);
2017-01-02 19:14:51 +00:00
2016-12-24 19:58:55 +00:00
break;
}
default:
2016-12-26 16:03:06 +00:00
errx(EX_DATAERR, "%s: unknown debug opcode %02x (%d)", name, op, op);
2016-12-24 19:58:55 +00:00
break;
}
}
}
break;
case REC_SECT: {
2016-12-25 07:42:43 +00:00
uint8_t sec = read_8(iter);
d.set_section(sec);
2016-12-24 19:58:55 +00:00
break;
}
2016-12-25 05:04:17 +00:00
2016-12-27 02:13:58 +00:00
case REC_ORG: {
2016-12-25 05:04:17 +00:00
d.flush();
uint32_t org = read_32(iter);
d.emit("", ".org", d.to_x(org, 4, '$'));
2016-12-25 07:42:43 +00:00
d.set_pc(org);
break;
}
2016-12-26 16:15:23 +00:00
case REC_SPACE: {
2016-12-27 02:13:58 +00:00
uint16_t count = read_16(iter);
2017-01-05 17:25:09 +00:00
d.space(count);
2016-12-26 16:15:23 +00:00
break;
}
2016-12-27 02:13:58 +00:00
case REC_LINE: {
2016-12-27 21:24:01 +00:00
// bump line counter, no argument.
2016-12-27 02:13:58 +00:00
d.flush();
2016-12-27 21:24:01 +00:00
++line;
2016-12-27 02:13:58 +00:00
break;
}
2016-12-24 19:58:55 +00:00
default:
2016-12-25 05:04:17 +00:00
d.flush();
2016-12-24 19:58:55 +00:00
errx(EX_DATAERR, "%s: unknown opcode %02x", name, op);
}
}
2016-12-25 17:35:04 +00:00
2017-01-05 19:14:22 +00:00
unsigned f = 0;
if (flags._S) f |= 0x01;
d.back_matter(f);
2016-12-24 19:58:55 +00:00
if (iter != data.end() || op != REC_END) errx(EX_DATAERR, "%s records ended early", name);
2016-12-24 19:58:55 +00:00
2017-01-04 18:35:14 +00:00
2017-01-04 02:23:45 +00:00
2016-12-26 16:15:23 +00:00
return true;
2016-12-24 19:58:55 +00:00
}
void dump_lib(const char *name, int fd)
{
Lib_head h;
ssize_t ok;
ok = read(fd, &h, sizeof(h));
if (ok != sizeof(h))
errx(EX_DATAERR, "%s is not an object file", name);
le_to_host(h.l_magic);
le_to_host(h.l_version);
le_to_host(h.l_filtyp);
le_to_host(h.l_modstart);
le_to_host(h.l_numsyms);
le_to_host(h.l_symsize);
le_to_host(h.l_numfiles);
assert(h.l_magic == MOD_MAGIC);
assert(h.l_version == 1);
assert(h.l_filtyp == 2);
2016-12-25 06:02:07 +00:00
printf("modstart : $%04x\n", h.l_modstart);
printf("number symbols: $%04x\n", h.l_numsyms);
printf("number files : $%04x\n", h.l_numfiles);
2016-12-24 19:58:55 +00:00
printf("\n");
std::vector<uint8_t> data;
long count = h.l_modstart - sizeof(h);
if (count < 0) errx(EX_DATAERR, "%s", name);
data.reserve(count);
ok = read(fd, data.data(), count);
if (ok != count) errx(EX_DATAERR, "%s truncated", name);
// files
auto iter = data.begin();
for (int i = 0; i < h.l_numfiles; ++i) {
uint16_t file_number = read_16(iter);
std::string s = read_pstring(iter);
2016-12-25 06:02:07 +00:00
printf("$%02x %s\n", file_number, s.c_str());
2016-12-24 19:58:55 +00:00
}
printf("\n");
// symbols
auto name_iter = iter + h.l_numsyms * 8;
for (int i = 0; i < h.l_numsyms; ++i) {
uint16_t name_offset = read_16(iter);
uint16_t file_number = read_16(iter);
uint32_t offset = read_32(iter);
std::string name = read_pstring(name_iter);
2016-12-25 06:02:07 +00:00
printf("symbol : $%04x %s\n", i, name.c_str());
2016-12-24 19:58:55 +00:00
//printf("name offset: %02x\n", name_offset);
2016-12-25 06:02:07 +00:00
printf("file_number : $%02x\n", file_number);
printf("module offset: $%04x\n", offset);
2016-12-24 19:58:55 +00:00
}
printf("\n");
}
void dump(const char *name) {
Header h;
int fd;
ssize_t ok;
2016-12-26 16:03:06 +00:00
fd = open(name, O_RDONLY | O_BINARY);
2016-12-24 19:58:55 +00:00
if (fd < 0) err(EX_NOINPUT, "Unable to open %s", name);
ok = read(fd, &h, sizeof(h));
if (ok != sizeof(h))
errx(EX_DATAERR, "%s is not an object file", name);
le_to_host(h.magic);
le_to_host(h.version);
le_to_host(h.filetype);
2016-12-24 19:58:55 +00:00
if (h.magic != MOD_MAGIC || h.version != 1 || h.filetype > 2)
2016-12-24 19:58:55 +00:00
errx(EX_DATAERR, "%s is not an object file", name);
lseek(fd, 0, SEEK_SET);
if (h.filetype == 2) dump_lib(name, fd);
2016-12-26 16:15:23 +00:00
// files may contain multiple modules.
while (dump_obj(name, fd)) /* ... */;
2016-12-24 19:58:55 +00:00
close(fd);
}
int main(int argc, char **argv) {
int c;
while ((c = getopt(argc, argv, "Sgn")) != -1) {
2017-01-04 02:23:45 +00:00
switch(c) {
case 'S': flags._S = true; break;
case 'g': flags._g = true; break;
case 'n': flags._n = true; break;
2017-01-04 02:23:45 +00:00
default: exit(EX_USAGE); break;
}
2016-12-24 19:58:55 +00:00
}
argv += optind;
argc -= optind;
if (argc == 0) usage();
for (int i = 0; i < argc; ++i) {
dump(argv[i]);
}
return 0;
2017-01-02 19:14:51 +00:00
}