wdc-utils/link.cpp

1769 lines
36 KiB
C++
Raw Normal View History

2017-01-10 19:16:02 +00:00
/*
* WDC to OMF Linker.
*
*
*/
#include <sysexits.h>
#include <stdlib.h>
#include <unistd.h>
#include <fcntl.h>
#include <err.h>
#include <assert.h>
#include <stdio.h>
2017-01-16 02:21:33 +00:00
#include <errno.h>
2017-01-16 03:02:30 +00:00
#include <string.h>
2017-01-12 01:17:06 +00:00
#include <cctype>
2017-01-10 19:16:02 +00:00
#include <string>
#include <vector>
#include <algorithm>
#include <unordered_map>
#include <unordered_set>
2017-01-16 02:21:33 +00:00
#include <map>
#include <set>
2017-01-10 19:16:02 +00:00
#include <array>
#include <utility>
#include <numeric>
#include <iterator>
#include "obj816.h"
#include "expression.h"
#include "omf.h"
2017-01-13 05:33:06 +00:00
#include "endian.h"
2017-01-10 19:16:02 +00:00
#ifndef O_BINARY
#define O_BINARY 0
#endif
struct {
bool _v = false;
bool _C = false;
bool _X = false;
bool _S = false;
std::string _o;
2017-01-16 02:21:33 +00:00
std::vector<std::string> _l;
std::vector<std::string> _L;
2017-01-10 19:16:02 +00:00
unsigned _errors = 0;
2017-01-11 19:05:12 +00:00
uint16_t _file_type;
uint32_t _aux_type;
2017-01-10 19:16:02 +00:00
} flags;
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>{});
}
#pragma pack(push, 1)
struct Header {
uint32_t magic; /* magic number for detection */
uint16_t version; /* version number of object format */
uint8_t filetype; /* file type, object or library */
};
#pragma pack(pop)
struct section {
std::string name;
uint8_t flags = 0;
uint32_t org = 0;
uint32_t size = 0;
unsigned number = -1;
std::vector<uint8_t> data;
std::vector<expression> expressions;
unsigned end_symbol = 0; // auto-generated _END_{name} symbol.
};
struct symbol {
std::string name;
uint8_t type = 0;
uint8_t flags = 0;
uint32_t offset = 0;
int section = -1;
};
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>
uint32_t read_32(T &iter) {
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;
}
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));
}
return sections;
}
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()) {
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);
symbols.emplace_back(std::move(s));
}
return symbols;
}
std::unordered_map<std::string, int> section_map;
std::vector<section> sections;
std::unordered_map<std::string, int> symbol_map;
std::vector<symbol> symbols;
2017-01-16 02:21:33 +00:00
std::set<std::string> undefined_symbols;
2017-01-10 19:16:02 +00:00
inline std::string parenthesize(const std::string &s) {
std::string tmp;
tmp.push_back('(');
tmp.append(s);
tmp.push_back(')');
return tmp;
}
inline void expr_error(bool fatal, const expression &e, const char *msg) {
warnx("%s:%04x %s", sections[e.section].name.c_str(), e.offset, msg);
// pretty-print the expression...
bool underflow = false;
struct pair { std::string name; int precedence = 0; };
std::vector<pair> stack;
int p;
for (auto &x : e.stack) {
auto tag = x.tag;
switch(tag) {
case OP_VAL: {
p = 0;
char buffer[6];
snprintf(buffer, sizeof(buffer), "$%02x", x.value);
stack.emplace_back(pair{buffer, p});
break;
}
case OP_SYM: {
p = 0;
stack.emplace_back(pair{symbols[x.section].name, p});
break;
}
case OP_LOC: {
p = 0;
std::string tmp = sections[x.section].name;
if (x.value) {
char buffer[6];
snprintf(buffer, sizeof(buffer), "$%02x", x.value);
tmp.push_back('+');
tmp.append(buffer);
p = 3;
}
stack.emplace_back(pair{tmp, p});
break;
}
case OP_NOT:
case OP_NEG:
case OP_FLP: {
static const std::string ops[] = {
".NOT.", "-", "\\"
};
p = 0;
if (stack.empty()) {
underflow = true;
} else {
auto &back = stack.back();
if (p < back.precedence) {
back.name = parenthesize(back.name);
}
back.name = ops[tag - OP_UNA] + back.name;
back.precedence = p;
}
break;
}
case OP_EXP:
case OP_MUL:
case OP_DIV:
case OP_MOD:
case OP_SHR:
case OP_SHL:
case OP_ADD:
case OP_SUB:
case OP_AND:
case OP_OR:
case OP_XOR:
case OP_EQ:
case OP_GT:
case OP_LT:
case OP_UGT:
case OP_ULT: {
static const pair ops[] = {
{ "**", 1 },
{ "*", 2 },
{ "/", 2 },
{ ".MOD.", 2 },
{ ">>", 2 },
{ "<<", 2 },
{ "+", 3 },
{ "-", 3 },
{ "&", 4 },
{ "|", 5 },
{ "^", 5 },
{ "=", 6 },
{ ">", 6},
{ "<", 6},
{ ".UGT.", 6},
{ ".ULT.", 6 }
};
p = ops[tag - OP_BIN].precedence;
if (stack.size() < 2) {
underflow = true;
} else {
pair b = std::move(stack.back());
stack.pop_back();
pair &a = stack.back();
if (p < b.precedence) b.name = parenthesize(b.name);
if (p < a.precedence) a.name = parenthesize(a.name);
a.name += ops[tag - OP_BIN].name + b.name;
a.precedence = p;
}
break;
}
default:
fprintf(stderr, "Unrecognized expression op %02x\n", tag);
break;
}
}
if (stack.size() == 1) {
fprintf(stderr, "Expression: %s\n", stack.front().name.c_str());
} else if (stack.empty() || underflow) {
fprintf(stderr, "Expression underflow error.\n");
fatal = true;
} else {
fprintf(stderr, "Expression overflow error.\n");
fatal = true;
}
2017-01-10 19:16:02 +00:00
if (fatal) flags._errors++;
}
2017-01-10 19:16:02 +00:00
/*
* replace undefined symbols (if possible) and simplify expressions.
*
*/
void simplify() {
for (auto &s : sections) {
for (auto &e : s.expressions) {
bool delta = false;
// first check for undefined symbols.
if (e.undefined) {
e.undefined = false;
for (auto &t : e.stack) {
if (t.tag == OP_SYM) {
const auto &ss = symbols[t.section];
switch(ss.type & 0x0f) {
case S_UND:
e.undefined = true;
break;
case S_REL:
t = expr{OP_LOC, ss.offset, (uint32_t)ss.section};
delta = true;
break;
case S_ABS:
t = expr{OP_VAL, (uint32_t)ss.offset};
delta = true;
break;
}
}
}
}
if (e.stack.size() > 1) simplify_expression(e);
}
}
}
/*
* read and process all sections...
* if section > 5, remap based on name.
*
*/
std::string &upper_case(std::string &s) {
2017-01-21 03:47:45 +00:00
std::transform(s.begin(), s.end(), s.begin(), toupper);
return s;
}
2017-01-16 02:21:33 +00:00
void one_module(const std::vector<uint8_t> &data,
const std::vector<uint8_t> &section_data,
const std::vector<uint8_t> &symbol_data,
std::set<std::string> *local_undefined = nullptr) {
2017-01-10 19:16:02 +00:00
std::array<int, 256> remap_section;
std::fill(remap_section.begin(), remap_section.end(), -1);
remap_section[SECT_PAGE0] = SECT_PAGE0;
remap_section[SECT_CODE] = SECT_CODE;
remap_section[SECT_KDATA] = SECT_KDATA;
remap_section[SECT_DATA] = SECT_DATA;
remap_section[SECT_UDATA] = SECT_UDATA;
std::vector<section> local_sections = read_sections(section_data);
std::vector<symbol> local_symbols = read_symbols(symbol_data);
2017-01-16 21:16:33 +00:00
2017-01-10 19:16:02 +00:00
// convert local sections to global
for (auto &s : local_sections) {
2017-01-16 21:16:33 +00:00
//printf("section %20s %d\n", s.name.c_str(), s.number);
2017-01-10 19:16:02 +00:00
if (s.number <= SECT_UDATA) {
sections[s.number].size += s.size; // for page0 / udata sections.
continue;
}
2017-01-16 21:16:33 +00:00
2017-01-10 19:16:02 +00:00
// todo -- should install section name as global symbol?
auto iter = section_map.find(s.name);
if (iter == section_map.end()) {
2017-01-16 21:16:33 +00:00
2017-01-10 19:16:02 +00:00
int virtual_section = sections.size();
remap_section[s.number] = virtual_section;
s.number = virtual_section;
if (!(s.flags & SEC_NONAME)) {
/* generate an _BEG_name_ and _END_name */
symbol sym;
sym.name = "_BEG_" + s.name;
upper_case(sym.name);
sym.section = virtual_section;
sym.type = S_REL; // check if section has offset?
sym.flags = SF_DEF | SF_GBL;
auto iter = symbol_map.find(sym.name);
if (iter == symbol_map.end()) {
symbol_map.emplace(sym.name, symbols.size());
symbols.emplace_back(std::move(sym));
} else {
// duplicate label error...
}
/*
// add entry for name? or handle via undefined symbol lookup below?
handled at end if symbol undeinfed.
sym.name = s.name;
sym.section = virtual_section;
sym.type = S_REL;
sym.flags = SF_DEF | SF_GBL;
iter = symbol_map.find(sym.name);
if (iter == symbol_map.end()) {
symbol_map.emplace(sym.name, symbols.size());
symbols.emplace_back(std::move(sym));
} else {
// duplicate label error...
}
*/
sym.name = "_END_" + s.name;
upper_case(sym.name);
sym.section = symbols.size();
sym.type = S_UND;
sym.flags = 0;
iter = symbol_map.find(sym.name);
if (iter == symbol_map.end()) {
s.end_symbol = sym.section;
symbol_map.emplace(sym.name, sym.section);
symbols.emplace_back(std::move(sym));
} else {
// duplicate label...
}
}
2017-01-10 19:16:02 +00:00
sections.emplace_back(s);
2017-01-16 21:16:33 +00:00
section_map.emplace(s.name, virtual_section);
2017-01-10 19:16:02 +00:00
} else {
auto &ss = sections[iter->second];
assert(ss.flags == s.flags); // check org????
remap_section[s.number] = iter->second;
s.number = iter->second;
// update size (for ref-only sections)
ss.size += s.size;
}
}
2017-01-16 21:16:33 +00:00
2017-01-10 19:16:02 +00:00
// convert local symbols to global.
for (auto &s : local_symbols) {
2017-01-16 21:16:33 +00:00
if (flags._v) {
const char *status = "";
if (s.type == S_UND) status = "extern";
else if (s.flags & SF_GBL) status = "public";
else status = "private";
fprintf(stderr, " %-20s [%s]\n", s.name.c_str(), status);
}
2017-01-10 19:16:02 +00:00
if (s.type == S_UND) {
2017-01-16 02:21:33 +00:00
2017-01-10 19:16:02 +00:00
auto iter = symbol_map.find(s.name);
if (iter == symbol_map.end()) {
s.section = symbols.size();
symbol_map.emplace(s.name, s.section);
2017-01-16 21:16:33 +00:00
symbols.emplace_back(s);
2017-01-10 19:16:02 +00:00
undefined_symbols.emplace(s.name);
2017-01-16 21:16:33 +00:00
fprintf(stderr, "Adding %s to undefined symbols\n", s.name.c_str());
2017-01-10 19:16:02 +00:00
}
else {
// already exists...
const auto &ss = symbols[iter->second];
2017-01-16 21:16:33 +00:00
/* if (ss.type != S_UND) */
// always copy over since s.section is a big deal.
s = ss;
2017-01-10 19:16:02 +00:00
}
2017-01-16 02:21:33 +00:00
if (local_undefined) {
if (s.type == S_UND) local_undefined->emplace(s.name);
}
2017-01-10 19:16:02 +00:00
continue;
}
// remap and fudge the offset.
if ((s.type & 0x0f) == S_REL) {
int virtual_section = remap_section[s.section];
assert(virtual_section != -1);
s.section = virtual_section;
s.offset += sections[virtual_section].data.size();
} else {
s.section = -1;
}
2017-01-16 21:16:33 +00:00
2017-01-10 19:16:02 +00:00
constexpr const unsigned mask = SF_GBL | SF_DEF;
if ((s.flags & mask) == mask) {
auto iter = symbol_map.find(s.name);
if (iter == symbol_map.end()) {
unsigned tmp = symbols.size();
symbol_map.emplace(s.name, tmp);
symbols.emplace_back(s);
} else {
auto &ss = symbols[iter->second];
// if it was undefined, define it!
if (ss.type == S_UND) {
ss = s;
undefined_symbols.erase(s.name);
}
else {
// ok if symbols are identical..
if (ss.type != s.type || ss.flags != s.flags || ss.section != s.section || ss.offset != s.offset) {
warnx("Duplicate label %s", s.name.c_str());
flags._errors++;
}
}
}
}
}
2017-01-16 21:16:33 +00:00
// set it here. sections may be resized above.
int current_section = SECT_CODE;
std::vector<uint8_t> *data_ptr = &sections[current_section].data;
2017-01-10 19:16:02 +00:00
auto iter = data.begin();
for(;;) {
uint8_t op = read_8(iter);
if (op == REC_END) return;
if (op < 0xf0) {
data_ptr->insert(data_ptr->end(), iter, iter + op);
iter += op;
continue;
}
switch(op) {
case REC_SPACE: {
uint16_t count = read_16(iter);
data_ptr->insert(data_ptr->end(), count, 0);
break;
}
case REC_SECT: {
/* switch sections */
uint8_t s = read_8(iter);
current_section = remap_section[s];
assert(current_section > 0 && current_section < sections.size());
data_ptr = &sections[current_section].data;
break;
}
case REC_ORG: {
assert(!"ORG not supported.");
break;
}
case REC_RELEXP:
case REC_EXPR: {
expression e;
e.relative = op == REC_RELEXP;
2017-01-13 01:55:47 +00:00
e.section = current_section;
2017-01-10 19:16:02 +00:00
e.offset = data_ptr->size();
e.size = read_8(iter);
2017-01-16 02:21:33 +00:00
2017-01-10 19:16:02 +00:00
data_ptr->insert(data_ptr->end(), e.size, 0);
int reduced_size = 0;
2017-01-10 19:16:02 +00:00
/**/
for(;;) {
op = read_8(iter);
if (op == OP_END) break;
switch(op) {
case OP_VAL: {
reduced_size++;
2017-01-10 19:16:02 +00:00
uint32_t offset = read_32(iter);
e.stack.emplace_back(op, offset);
break;
}
case OP_SYM: {
reduced_size++;
2017-01-10 19:16:02 +00:00
uint16_t symbol = read_16(iter);
assert(symbol < local_symbols.size());
auto &s = local_symbols[symbol];
switch (s.type & 0x0f) {
case S_UND:
// S_UND indicates it's still undefined globally.
e.stack.emplace_back(OP_SYM, 0, s.section); /* section is actually a symbol number */
e.undefined = true;
break;
2017-01-16 21:16:33 +00:00
2017-01-10 19:16:02 +00:00
case S_REL:
e.stack.emplace_back(OP_LOC, s.offset, s.section);
break;
case S_ABS:
e.stack.emplace_back(OP_VAL, s.offset);
break;
default:
assert(!"unsupported symbol flags.");
}
break;
}
case OP_LOC: {
reduced_size++;
2017-01-10 19:16:02 +00:00
uint8_t section = read_8(iter);
uint32_t offset = read_32(iter);
int real_section = remap_section[section];
assert(real_section >= 0);
e.stack.emplace_back(op, offset, real_section);
break;
}
// operations..
//unary
case OP_NOT:
case OP_NEG:
case OP_FLP:
e.stack.emplace_back(op);
break;
2017-01-10 19:16:02 +00:00
// binary
case OP_EXP:
case OP_MUL:
case OP_DIV:
case OP_MOD:
case OP_SHR:
case OP_SHL:
case OP_ADD:
case OP_SUB:
case OP_AND:
case OP_OR:
case OP_XOR:
case OP_EQ:
case OP_GT:
case OP_LT:
case OP_UGT:
case OP_ULT:
reduced_size--;
2017-01-10 19:16:02 +00:00
e.stack.emplace_back(op);
break;
default:
assert(!"unsupported expression opcode.");
}
}
if (reduced_size != 1) {
expr_error(true, e, "Malformed expression");
}
2017-01-10 19:16:02 +00:00
sections[current_section].expressions.emplace_back(std::move(e));
break;
}
case REC_LINE: break;
case REC_DEBUG: {
uint16_t size = read_16(iter);
iter += size;
break;
}
}
}
}
/*
* n.b -- UDATA and PAGE0 are ref only, therefore no data is generated.
* as a special case for UDATA (but not PAGE0) have a flag so it will be 0-filled and generate data?
*
*/
void init() {
sections.resize(5);
sections[SECT_PAGE0].number = SECT_PAGE0;
sections[SECT_PAGE0].flags = SEC_DATA | SEC_NONAME | SEC_DIRECT | SEC_REF_ONLY;
sections[SECT_PAGE0].name = "page0";
sections[SECT_CODE].number = SECT_CODE;
sections[SECT_CODE].flags = SEC_NONAME;
sections[SECT_CODE].name = "code";
sections[SECT_KDATA].number = SECT_KDATA;
sections[SECT_KDATA].flags = SEC_DATA | SEC_NONAME;
sections[SECT_KDATA].name = "kdata";
sections[SECT_DATA].number = SECT_DATA;
sections[SECT_DATA].flags = SEC_DATA | SEC_NONAME;
sections[SECT_DATA].name = "data";
sections[SECT_UDATA].number = SECT_UDATA;
sections[SECT_UDATA].flags = SEC_DATA | SEC_NONAME | SEC_REF_ONLY;
sections[SECT_UDATA].name = "udata";
/*
* For each section, [the linker] creates three symbols,
* _ROM_BEG_secname, _BEG_secname and _END_secname, which
* correspond to the rom location and the execution beginning
* and end of the section. These will be used more in the next
* two sections of code.
*/
// n.b - only for pre-defined sections [?], skip the _ROM_BEG_* symbols...
static std::string names[] = {
"_BEG_PAGE0", "_END_PAGE0",
"_BEG_CODE", "_END_CODE",
"_BEG_KDATA", "_END_KDATA",
"_BEG_DATA", "_END_DATA",
"_BEG_UDATA", "_END_UDATA",
};
for (int i = 0; i < 5; ++i) {
// begin is 0.
symbol s;
s.name = names[i * 2];
s.section = i;
s.type = S_REL;
s.flags = SF_DEF | SF_GBL;
symbol_map.emplace(s.name, i * 2);
symbols.emplace_back(s);
// end is undefined...
s.name = names[i * 2 + 1];
s.section = i * 2 + 1; // symbol number.
s.type = S_UND;
s.flags = 0;
sections[i].end_symbol = s.section;
2017-01-10 19:16:02 +00:00
symbol_map.emplace(s.name, s.section);
symbols.emplace_back(s);
2017-01-16 21:16:33 +00:00
// even though it's undefined, don't add it to the undefined list ...
// don't need to search libraries for it!
2017-01-10 19:16:02 +00:00
}
}
2017-01-21 17:43:18 +00:00
/*
* add and return an undefined symbol. if known, adds it to the missing set.
* if symbol is already defined, returns it.
*/
symbol &reserve_symbol(const std::string &name, bool open = true) {
auto iter = symbol_map.find(name);
if (iter != symbol_map.end()) return symbols[iter->second];
if (open) undefined_symbols.emplace(name);
else undefined_symbols.erase(name);
symbols.emplace_back();
auto &sym = symbols.back();
sym.name = name;
sym.type = S_UND;
sym.section = symbols.size() - 1;
symbol_map.emplace(name, sym.section);
return sym;
}
2017-01-10 19:16:02 +00:00
void generate_end() {
/*
const std::string names[] = {
"_END_PAGE0",
"_END_CODE",
"_END_KDATA"
"_END_DATA"
"_END_UDATA"
};
*/
/*
2017-01-10 19:16:02 +00:00
for (int i = 0; i < 5; ++i) {
symbol s;
s.section = i;
s.type = S_REL;
s.flags = SF_DEF | SF_GBL;
s.offset = sections[i].size; // data.size() doesn't word w/ ref_only
symbols[i * 2 + 1] = s;
}
*/
for (const auto &s : sections) {
// if there is an undefined symbol matching the section name, add it.
if ((s.flags & SEC_NONAME) == 0) {
auto iter = symbol_map.find(s.name);
if (iter != symbol_map.end()) {
symbol &sym = symbols[iter->second];
if (sym.type == S_UND) {
undefined_symbols.erase(s.name);
sym.section = s.number;
sym.type = S_REL;
sym.flags = SF_DEF | SF_GBL;
sym.offset = 0;
}
}
}
if (s.end_symbol) {
symbol &sym = symbols[s.end_symbol];
sym.section = s.number;
sym.type = S_REL;
sym.flags = SF_DEF | SF_GBL;
sym.offset = s.size;
}
}
2017-01-10 19:16:02 +00:00
}
std::vector<omf::segment> omf_segments;
template<class T>
void append(std::vector<T> &to, std::vector<T> &from) {
2017-03-06 05:40:00 +00:00
// std::move(from.begin(), from.end(), std::back_inserter(to));
2017-01-10 19:16:02 +00:00
to.insert(to.end(),
std::make_move_iterator(from.begin()),
std::make_move_iterator(from.end())
);
}
template<class T>
void append(std::vector<T> &to, const std::vector<T> &from) {
to.insert(to.end(),
from.begin(),
from.end()
);
}
template<class T>
void append(std::vector<T> &to, unsigned count, const T& value) {
to.insert(to.end(),
count,
value
);
}
/*
* convert a wdc expression to an omf reloc/interseg record.
*
*/
2017-01-13 01:55:47 +00:00
inline bool in_range(int value, int low, int high) {
return value >= low && value <= high;
}
2017-01-10 19:16:02 +00:00
2017-01-13 01:55:47 +00:00
void to_omf(const expression &e, omf::segment &seg) {
if (e.stack.empty() || e.size == 0) {
2017-01-13 02:32:11 +00:00
expr_error(false, e, "Expression empty");
2017-01-10 19:16:02 +00:00
return;
}
2017-01-16 02:21:33 +00:00
if (e.size < 1 || e.size > 4) {
expr_error(true, e, "Expression size must be 1-4 bytes");
}
2017-01-10 19:16:02 +00:00
2017-01-13 01:55:47 +00:00
if (e.stack.size() == 1) {
auto &a = e.stack[0];
uint32_t value = a.value;
if (e.relative) {
if (a.tag == OP_VAL || a.tag == OP_LOC) {
int tmp = (int)value - (int)e.offset - (int)e.size;
bool ok = false;
if (e.size >= 2 && in_range(tmp, -32768, 32767)) ok = true;
if (e.size == 1 && in_range(tmp, -128, 127)) ok = true;
if (!ok) {
2017-01-13 02:32:11 +00:00
expr_error(true, e, "Relative branch out of range");
2017-01-13 01:55:47 +00:00
return;
}
for (int i = 0; i < e.size; ++i, tmp >>= 8)
seg.data[e.offset + i] = tmp & 0xff;
return;
}
2017-01-10 19:16:02 +00:00
2017-01-13 02:32:11 +00:00
expr_error(true, e, "Relative expression too complex");
2017-01-10 19:16:02 +00:00
return;
}
2017-01-13 01:55:47 +00:00
if (a.tag == OP_VAL) {
2017-01-10 19:16:02 +00:00
for (int i = 0; i < e.size; ++i, value >>= 8)
seg.data[e.offset + i] = value & 0xff;
2017-01-13 01:55:47 +00:00
return;
}
2017-01-10 19:16:02 +00:00
2017-01-13 01:55:47 +00:00
if (a.tag == OP_LOC) {
auto &loc = a;
2017-01-13 02:32:11 +00:00
2017-01-13 01:55:47 +00:00
if (loc.section == 0) {
2017-01-13 02:32:11 +00:00
expr_error(true, e, "Invalid segment");
2017-01-13 01:55:47 +00:00
return;
}
if (loc.section == seg.segnum) {
omf::reloc r;
r.size = e.size;
r.offset = e.offset;
r.value = value;
seg.relocs.emplace_back(r);
} else {
omf::interseg r;
r.size = e.size;
r.offset = e.offset;
r.segment = loc.section;
r.segment_offset = loc.value;
seg.intersegs.emplace_back(r);
2017-01-10 19:16:02 +00:00
2017-01-13 01:55:47 +00:00
// if generating super, store
}
return;
2017-01-10 19:16:02 +00:00
}
2017-01-13 01:55:47 +00:00
// error handled below.
2017-01-10 19:16:02 +00:00
}
2017-01-13 01:55:47 +00:00
if (e.stack.size() == 3) {
2017-01-10 19:16:02 +00:00
auto &loc = e.stack[0];
auto &shift = e.stack[1];
auto &op = e.stack[2];
2017-01-13 01:55:47 +00:00
if (loc.tag == OP_LOC && shift.tag == OP_VAL && (op.tag == OP_SHL || op.tag == OP_SHR)) {
2017-01-10 19:16:02 +00:00
2017-01-13 01:55:47 +00:00
if (shift.value > 24) {
2017-01-13 02:32:11 +00:00
expr_error(false, e, "Shift too large");
2017-01-13 01:55:47 +00:00
// data is already pre-zeroed.
return;
}
2017-01-10 19:16:02 +00:00
2017-01-13 01:55:47 +00:00
if (loc.section == 0) {
2017-01-13 02:32:11 +00:00
expr_error(true, e, "Invalid segment");
2017-01-13 01:55:47 +00:00
return;
}
2017-01-10 19:16:02 +00:00
2017-01-13 01:55:47 +00:00
uint32_t value = loc.value;
uint8_t shift_value = shift.value;
if (op.tag == OP_SHR) {
value >>= shift_value;
shift_value = -shift_value;
} else {
value <<= shift_value;
}
2017-01-10 19:16:02 +00:00
2017-01-13 01:55:47 +00:00
if (loc.section == seg.segnum) {
omf::reloc r;
r.size = e.size;
r.offset = e.offset;
r.value = loc.value;
r.shift = shift_value;
2017-01-10 19:16:02 +00:00
2017-01-13 01:55:47 +00:00
#if 0
// also store value in data
for (int i = 0; i < e.size; ++i, value >>= 8)
seg.data[e.offset + i] = value & 0xff;
#endif
2017-01-10 19:16:02 +00:00
2017-01-13 01:55:47 +00:00
seg.relocs.emplace_back(r);
} else {
omf::interseg r;
r.size = e.size;
r.offset = e.offset;
r.segment = loc.section;
r.segment_offset = loc.value;
r.shift = shift_value;
seg.intersegs.emplace_back(r);
}
return;
2017-01-10 19:16:02 +00:00
}
}
2017-01-13 02:32:11 +00:00
expr_error(true, e, "Expression too complex");
2017-01-13 01:55:47 +00:00
// should also pretty-print the expression.
2017-01-10 19:16:02 +00:00
return;
}
void build_omf_segments() {
std::vector< std::pair<unsigned, uint32_t> > remap;
remap.resize(sections.size());
// if data + code can fit in one bank, merge them
// otherwise, merge all data sections and 1 omf segment
// per code section.
// code is next segment...
unsigned code_segment = 0;
unsigned data_segment = 0;
{
omf_segments.emplace_back();
auto &seg = omf_segments.back();
code_segment = data_segment = seg.segnum = omf_segments.size();
seg.kind = 0x0000; // static code segment.
auto &s = sections[SECT_CODE];
remap[s.number] = std::make_pair(code_segment, 0);
append(seg.data, s.data);
s.data.clear();
}
uint32_t total_code_size = 0;
uint32_t total_data_size = 0;
for (const auto &s : sections) {
if (s.flags & SEC_REF_ONLY) continue;
if (s.flags & SEC_DATA) {
total_data_size += s.size;
} else {
total_code_size += s.size;
}
}
// add in UDATA
total_data_size += sections[SECT_UDATA].size;
if (total_data_size + sections[SECT_CODE].size > 0xffff) {
omf_segments.emplace_back();
auto &seg = omf_segments.back();
data_segment = seg.segnum = omf_segments.size();
seg.kind = 0x0001; // static data segment.
}
{
omf::segment &data_seg = omf_segments[data_segment-1];
2017-01-10 19:16:02 +00:00
// KDATA, DATA, UDATA, other segment order.
for (auto &s : sections) {
if (s.flags & SEC_REF_ONLY) continue;
if ((s.flags & SEC_DATA) == 0) continue;
2017-01-10 19:16:02 +00:00
remap[s.number] = std::make_pair(data_segment, data_seg.data.size());
2017-01-10 19:16:02 +00:00
append(data_seg.data, s.data);
s.data.clear();
}
// add in UDATA
{
auto &s = sections[SECT_UDATA];
remap[s.number] = std::make_pair(data_segment, data_seg.data.size());
append(data_seg.data, s.size, (uint8_t)0);
}
// data_seg no longer valid since emplace_back() may invalidate.
2017-01-10 19:16:02 +00:00
}
// for all other sections, create a new segment.
for (auto &s : sections) {
if (s.flags & SEC_REF_ONLY) continue;
if (s.flags & SEC_DATA) continue;
if (s.number == SECT_CODE) continue;
omf_segments.emplace_back();
auto &seg = omf_segments.back();
seg.segnum = omf_segments.size();
2017-01-10 19:16:02 +00:00
seg.kind = 0x0000; // static code.
seg.data = std::move(s.data);
seg.segname = s.name;
s.data.clear();
remap[s.number] = std::make_pair(seg.segnum, 0);
}
// add a stack segment at the end
if (flags._S) {
auto &s = sections[SECT_PAGE0];
// create stack/dp segment.
uint32_t size = s.size;
if (size) {
// ????
size = (size + 255) & ~255;
omf_segments.emplace_back();
auto &seg = omf_segments.back();
seg.segnum = omf_segments.size();
2017-01-10 19:16:02 +00:00
seg.kind = 0x12; // static dp/stack segment.
seg.data.resize(size, 0);
seg.loadname = "~Stack";
omf_segments.emplace_back(std::move(seg));
// remap SECT_PAGE0...
remap[s.number] = std::make_pair(seg.segnum, 0);
} else {
warnx("page0 is 0 sized. Stack/dp segment not created.");
}
}
// now adjust all the expressions, simplify, and convert to reloc records.
for (auto &s :sections) {
auto &x = remap[s.number];
2017-01-10 19:16:02 +00:00
for (auto &e : s.expressions) {
e.offset += x.second;
2017-01-10 19:16:02 +00:00
for (auto &t : e.stack) {
if (t.tag == OP_LOC) {
const auto &x = remap[t.section];
t.section = x.first;
t.value += x.second;
}
}
simplify_expression(e);
unsigned segnum = remap[s.number].first;
to_omf(e, omf_segments.at(segnum-1));
2017-01-10 19:16:02 +00:00
}
}
// and we're done...
}
2017-01-16 02:21:33 +00:00
bool one_module(const std::string &name, int fd, std::set<std::string> *local_undefined = nullptr) {
Mod_head h;
ssize_t ok;
ok = read(fd, &h, sizeof(h));
if (ok == 0) return false;
if (ok < sizeof(h)) {
warnx("Invalid object file: %s", name.c_str());
return false;;
}
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);
2017-01-10 19:16:02 +00:00
2017-01-16 02:21:33 +00:00
assert(h.h_magic == MOD_MAGIC);
assert(h.h_version == 1);
assert(h.h_filtyp == 1);
std::string module_name;
{
// now read the name (h_namlen includes 0 terminator.)
std::vector<char> tmp;
tmp.resize(h.h_namlen);
ok = read(fd, tmp.data(), h.h_namlen);
if (ok != h.h_namlen) {
warnx("Invalid object file: %s", name.c_str());
return false;
}
module_name.assign(tmp.data());
}
std::vector<uint8_t> record_data;
std::vector<uint8_t> symbol_data;
std::vector<uint8_t> section_data;
record_data.resize(h.h_recsize);
ok = read(fd, record_data.data(), h.h_recsize);
if (ok != h.h_recsize) {
warnx("Truncated object file: %s", name.c_str());
return false;
}
section_data.resize(h.h_secsize);
ok = read(fd, section_data.data(), h.h_secsize);
if (ok != h.h_secsize) {
warnx("Truncated object file: %s", name.c_str());
return false;
}
symbol_data.resize(h.h_symsize);
ok = read(fd, symbol_data.data(), h.h_symsize);
if (ok != h.h_symsize) {
warnx("Truncated object file: %s", name.c_str());
return false;
}
if (flags._v) {
printf("Processing %s:%s\n", name.c_str(), module_name.c_str());
}
// should probably pass in name and module....
one_module(record_data, section_data, symbol_data, local_undefined);
if (h.h_optsize) lseek(fd, h.h_optsize, SEEK_CUR);
return true;
}
2017-01-10 19:16:02 +00:00
bool one_file(const std::string &name) {
2017-01-16 02:21:33 +00:00
if (flags._v) printf("Processing %s\n", name.c_str());
2017-01-10 19:16:02 +00:00
int fd = open(name.c_str(), O_RDONLY | O_BINARY);
if (fd < 0) {
warn("Unable to open %s", name.c_str());
return false;
}
bool rv = false;
Header h;
ssize_t ok;
ok = read(fd, &h, sizeof(h));
if (ok != sizeof(h)) {
warnx("Invalid object file: %s", name.c_str());
close(fd);
return false;
}
le_to_host(h.magic);
le_to_host(h.version);
le_to_host(h.filetype);
if (h.magic != MOD_MAGIC || h.version != MOD_VERSION || h.filetype < MOD_OBJECT || h.filetype > MOD_LIBRARY) {
warnx("Invalid object file: %s", name.c_str());
close(fd);
return false;
}
if (h.filetype == MOD_LIBRARY) {
warnx("%s is a library", name.c_str());
close(fd);
// todo -- add to library list...
return true;
}
lseek(fd, 0, SEEK_SET);
2017-01-16 02:21:33 +00:00
while(one_module(name, fd)) ;
2017-01-10 19:16:02 +00:00
2017-01-16 02:21:33 +00:00
close(fd);
return true;
}
2017-01-10 19:16:02 +00:00
2017-01-16 02:21:33 +00:00
enum {
kPending = 1,
kProcessed = 2,
};
2017-01-10 19:16:02 +00:00
/*
* observation: the library symbol size will far exceed the missing symbols size.
* Therefore, library symbols should be an unordered_map and explicitely look up
* each undefined symbol.
*
* output c is a map (and thus sorted) to guarantee reproducable builds.
* (could use a vector then sort/unique it...)
*/
bool intersection(const std::unordered_map<std::string, uint32_t> &a,
const std::set<std::string> &b,
std::map<uint32_t, int> &c)
{
bool rv = false;
for (const auto &name : b) {
auto iter = a.find(name);
if (iter == a.end()) continue;
rv = true;
c.emplace(iter->second, kPending);
}
return rv;
}
#if 0
2017-01-16 02:21:33 +00:00
bool intersection(const std::map<std::string, uint32_t> &a,
const std::set<std::string> &b,
std::map<uint32_t, int> &c)
{
auto a_iter = a.begin();
auto b_iter = b.begin();
2017-01-10 19:16:02 +00:00
2017-01-16 02:21:33 +00:00
bool rv = false;
2017-01-10 19:16:02 +00:00
2017-01-16 02:21:33 +00:00
while (a_iter != a.end() && b_iter != b.end()) {
2017-01-16 21:16:33 +00:00
//fprintf(stderr, "comparing %s - %s\n", a_iter->first.c_str(), b_iter->c_str());
2017-01-16 02:21:33 +00:00
int cmp = strcmp(a_iter->first.c_str(), b_iter->c_str());
if (cmp < 0) a_iter++;
else if (cmp > 0) b_iter++;
else {
// insert/emplace does not overwrite a previous value.
c.emplace(a_iter->second, kPending);
a_iter++;
b_iter++;
rv = true;
2017-01-10 19:16:02 +00:00
}
2017-01-16 02:21:33 +00:00
}
2017-01-10 19:16:02 +00:00
2017-01-16 02:21:33 +00:00
return rv;
}
#endif
2017-01-16 02:21:33 +00:00
2017-01-10 19:16:02 +00:00
2017-01-16 02:21:33 +00:00
bool one_lib(const std::string &path) {
Lib_head h;
int fd = open(path.c_str(), O_RDONLY | O_BINARY);
if (fd < 0) {
if (errno == ENOENT) return false;
2017-01-10 19:16:02 +00:00
}
2017-01-16 02:21:33 +00:00
if (flags._v) printf("Processing library %s\n", path.c_str());
if (fd < 0) {
warn("Unable to open %s", path.c_str());
return false;
}
ssize_t ok;
ok = read(fd, &h, sizeof(h));
if (ok != sizeof(h)) {
warnx("Invalid library file: %s", path.c_str());
close(fd);
return false;
}
le_to_host(h.l_magic);
le_to_host(h.l_version);
le_to_host(h.l_filtyp);
le_to_host(h.l_unused1);
le_to_host(h.l_modstart);
le_to_host(h.l_numsyms);
le_to_host(h.l_symsize);
le_to_host(h.l_numfiles);
if (h.l_magic != MOD_MAGIC || h.l_version != MOD_VERSION || h.l_filtyp != MOD_LIBRARY) {
warnx("Invalid library file: %s", path.c_str());
close(fd);
return false;
}
// read the symbol dictionary.
std::vector<uint8_t> data;
2017-01-16 21:16:33 +00:00
data.resize(h.l_modstart - sizeof(h));
2017-01-16 02:21:33 +00:00
ok = read(fd, data.data(), data.size());
if (ok != data.size()) {
warnx("Invalid library file: %s", path.c_str());
return false;
}
auto iter = data.begin();
auto end = data.end();
// files -- only reading since it's variable length.
for (unsigned i = 0; i < h.l_numfiles; ++i) {
// fileno, pstring file name
uint16_t fileno = read_16(iter);
2017-01-16 21:16:33 +00:00
std::string s = read_pstring(iter);
//uint8_t size = read_8(iter);
//iter += size; // don't care about the name.
2017-01-16 02:21:33 +00:00
}
std::unordered_map<std::string, uint32_t> lib_symbol_map;
2017-01-16 02:21:33 +00:00
// map of which modules have been loaded or are pending processing.
std::map<uint32_t, int> modules;
auto name_iter = iter + h.l_numsyms * 8;
for (unsigned 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) + h.l_modstart;
auto tmp = name_iter + name_offset;
std::string name = read_pstring(tmp);
lib_symbol_map.emplace(std::move(name), offset);
//modules[offset] = 0;
}
// find an intersection of undefined symbols and symbols defined in lib_symbol_map
if (!intersection(lib_symbol_map, undefined_symbols, modules)) {
close(fd);
return true;
}
for(;;) {
bool delta = false;
std::set<std::string> local_undefined_symbols;
for (auto &x : modules) {
uint32_t offset = x.first;
int status = x.second;
if (status == kPending) {
x.second = kProcessed;
lseek(fd, offset, SEEK_SET);
one_module(path, fd, &local_undefined_symbols);
delta = true;
}
}
if (!delta) break;
delta = intersection(lib_symbol_map, local_undefined_symbols, modules);
if (!delta) break;
}
2017-01-10 19:16:02 +00:00
close(fd);
2017-01-16 02:21:33 +00:00
return true;
}
void libraries() {
if (undefined_symbols.empty()) return;
for (auto &l : flags._l) {
for (auto &L : flags._L) {
2017-01-16 21:16:33 +00:00
//std::string path = L + "lib" + l;
std::string path = L + l + ".lib";
2017-01-16 02:21:33 +00:00
if (one_lib(path)) break;
}
if (undefined_symbols.empty()) break;
}
2017-01-10 19:16:02 +00:00
}
2017-01-12 01:17:06 +00:00
#if 0
bool parse_ft(const std::string &s) {
// xx
// xx:xxxx or xx,xxxx
auto lambda = [](const optional<int> &lhs, const uint8_t rhs) {
if (!lhs) return lhs;
if (rhs >= '0' && rhs <= '9')
return optional<int>((*lhs << 4) + rhs);
if (rhs >= 'a' && rhs <= 'f')
return optional<int>((*lhs << 4) + (rhs - 'a' + 10));
if (rhs >= 'A' && rhs <= 'F')
return optional<int>((*lhs << 4) + (rhs - 'A') + 10);
return optional<int>();
};
optional<int> ft;
optional<int> at;
if (s.length() == 2 || s.length() == 7) {
ft = std::accumulate(s.begin(), s.begin() + 2, optional<int>(0), lambda);
if (s.length() == 7) {
ft = optional<int>();
if (s[2] == ':' || s[2] == ',')
at = std::accumulate(s.begin() + 3, s.end(), optional<int>(0), lambda);
} else at = optional<int>(0);
}
if (at && ft) {
flags._file_type = *ft;
flags._aux_type = *at;
return true;
}
return false;
}
#endif
bool parse_ft(const std::string &s) {
2017-01-12 01:37:03 +00:00
// gcc doesn't like std::xdigit w/ std::all_of
2017-01-12 01:17:06 +00:00
if (s.length() != 2 && s.length() != 7) return false;
2017-01-12 01:37:03 +00:00
if (!std::all_of(s.begin(), s.begin() + 2, isxdigit)) return false;
2017-01-12 01:17:06 +00:00
if (s.length() == 7) {
if (s[2] != ',' && s[2] != ':') return false;
2017-01-12 01:37:03 +00:00
if (!std::all_of(s.begin() + 3, s.end(), isxdigit)) return false;
2017-01-12 01:17:06 +00:00
}
auto lambda = [](int lhs, uint8_t rhs){
lhs <<= 4;
if (rhs <= '9') return lhs + rhs - '0';
return lhs + (rhs | 0x20) - 'a' + 10;
};
flags._file_type = std::accumulate(s.begin(), s.begin() + 2, 0, lambda);
flags._aux_type = 0;
if (s.length() == 7)
flags._aux_type = std::accumulate(s.begin() + 3, s.end(), 0, lambda);
return true;
}
2017-01-10 19:16:02 +00:00
void help() {
exit(0);
}
void usage() {
exit(EX_USAGE);
}
int main(int argc, char **argv) {
int c;
2017-01-11 19:05:12 +00:00
while ((c = getopt(argc, argv, "vCXL:l:o:t:")) != -1) {
2017-01-10 19:16:02 +00:00
switch(c) {
case 'v': flags._v = true; break;
case 'X': flags._X = true; break;
case 'C': flags._C = true; break;
case 'o': flags._o = optarg; break;
2017-01-16 02:21:33 +00:00
case 'l': {
if (*optarg) flags._l.emplace_back(optarg);
break;
}
case 'L': {
std::string tmp(optarg);
if (tmp.empty()) tmp = ".";
if (tmp.back() != '/') tmp.push_back('/');
flags._L.emplace_back(std::move(tmp));
break;
}
2017-01-10 19:16:02 +00:00
case 'h': help(); break;
2017-01-11 19:05:12 +00:00
case 't': {
// -t xx[:xxxx] -- set file/auxtype.
2017-01-12 01:17:06 +00:00
if (!parse_ft(optarg)) {
errx(EX_USAGE, "Invalid -t argument: %s", optarg);
}
2017-01-11 19:05:12 +00:00
break;
}
2017-01-10 19:16:02 +00:00
case ':':
case '?':
default:
usage();
}
}
argc -= optind;
argv += optind;
if (argc == 0) usage();
init();
for (int i = 0 ; i < argc; ++i) {
if (!one_file(argv[i])) flags._errors++;
}
2017-01-16 02:21:33 +00:00
if (flags._v && !undefined_symbols.empty()) {
printf("Undefined Symbols:\n");
for (const auto & s : undefined_symbols) {
printf("%s\n", s.c_str());
}
printf("\n");
2017-01-10 19:16:02 +00:00
}
2017-01-16 02:21:33 +00:00
if (!undefined_symbols.empty()) libraries();
//
2017-01-16 21:16:33 +00:00
if (!undefined_symbols.empty()) {
fprintf(stderr, "Unable to resolve the following symbols:\n");
for (auto &s : undefined_symbols) fprintf(stderr, "%s\n", s.c_str());
exit(EX_DATAERR);
}
2017-01-10 19:16:02 +00:00
generate_end();
2017-01-16 02:21:33 +00:00
simplify();
2017-01-10 19:16:02 +00:00
if (flags._v) {
for (const auto &s : sections) {
//if (s.flags & SEC_REF_ONLY) continue;
printf("section %3d %-20s $%04x $%04x\n",
s.number, s.name.c_str(), (uint32_t)s.data.size(), s.size);
}
fputs("\n", stdout);
}
build_omf_segments();
if (flags._v) {
for (const auto &s : omf_segments) {
printf("segment %3d %-20s $%04x\n",
s.segnum, s.segname.c_str(), (uint32_t)s.data.size());
for (auto &r : s.relocs) {
printf(" %02x %02x %06x %06x\n",
r.size, r.shift, r.offset, r.value);
}
for (auto &r : s.intersegs) {
printf(" %02x %02x %06x %02x %04x %06x\n",
r.size, r.shift, r.offset, r.file, r.segment, r.segment_offset);
}
}
}
2017-01-11 05:36:03 +00:00
if (flags._o.empty()) flags._o = "out.omf";
2017-01-11 19:05:12 +00:00
if (!flags._file_type) {
flags._file_type = 0xb3;
}
2017-01-13 15:39:02 +00:00
void save_omf(const std::string &path, std::vector<omf::segment> &segments, bool compress, bool expressload);
2017-01-11 19:05:12 +00:00
int set_file_type(const std::string &path, uint16_t file_type, uint32_t aux_type);
2017-01-10 19:16:02 +00:00
2017-01-13 15:39:02 +00:00
save_omf(flags._o, omf_segments, !flags._C, !flags._X);
2017-01-11 19:05:12 +00:00
set_file_type(flags._o, flags._file_type, flags._aux_type);
2017-01-10 19:16:02 +00:00
}