6502/wdc-utils
1
0
Fork 0
mirror of https://github.com/ksherlock/wdc-utils.git synced 2024-12-12 19:30:38 +00:00
Code Issues Projects Releases Wiki Activity

move most disassembler logic to separate class.

Browse Source
This commit is contained in:
Kelvin Sherlock 2017-01-04 23:50:35 -05:00
parent fefdbb5ad0
commit 2cd1ad1ce5
7 changed files with 523 additions and 392 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
Makefile
View File

@ -2,7 +2,7 @@ LINK.o = $(LINK.cc)
CXXFLAGS = -std=c++11 -g CXXFLAGS = -std=c++11 -g
CCFLAGS = -g CCFLAGS = -g
OBJS = dumpobj.o disassembler.o OBJS = dumpobj.o disassembler.o zrdz_disassembler.o
#UNAME_S := $(shell uname -s) #UNAME_S := $(shell uname -s)
#ifeq ($(UNAME_S),MINGW64_NT-10.0) #ifeq ($(UNAME_S),MINGW64_NT-10.0)
@ -16,7 +16,8 @@ dumpobj : $(OBJS)
disassembler.o : disassembler.cpp disassembler.h disassembler.o : disassembler.cpp disassembler.h
dumpobj.o : dumpobj.cpp disassembler.h zrdz_disassembler.o : zrdz_disassembler.cpp zrdz_disassembler.h disassembler.h
dumpobj.o : dumpobj.cpp zrdz_disassembler.h disassembler.h
mingw/err.o : mingw/err.c mingw/err.h mingw/err.o : mingw/err.c mingw/err.h
.PHONY: clean .PHONY: clean

78
disassembler.cpp
View File

@ -333,8 +333,21 @@ static constexpr const int modes[] =
}; };
constexpr const int kOpcodeTab = 20;
constexpr const int kOperandTab = 30;
constexpr const int kCommentTab = 80;
static std::string to_x(uint32_t x, unsigned bytes, char prefix = 0) {
void indent_to(std::string &line, unsigned position) {
if (line.length() < position)
line.resize(position, ' ');
}
disassembler::~disassembler() {
}
std::string disassembler::to_x(uint32_t x, unsigned bytes, char prefix) {
std::string s; std::string s;
char buffer[16]; char buffer[16];
if (prefix) s.push_back(prefix); if (prefix) s.push_back(prefix);
@ -356,20 +369,68 @@ static std::string to_x(uint32_t x, unsigned bytes, char prefix = 0) {
} }
void disassembler::emit(const std::string &label) {
fputs(label.c_str(), stdout);
fputc('\n', stdout);
}
void disassembler::emit(const std::string &label, const std::string &opcode) {
std::string tmp;
tmp = label;
int column = tmp.length();
if (!opcode.empty()) {
do {
tmp.push_back(' ');
column++;
} while (column < kOpcodeTab);
tmp += opcode;
}
tmp.push_back('\n');
fputs(tmp.c_str(), stdout);
}
void disassembler::emit(const std::string &label, const std::string &opcode, const std::string &operand) {
std::string tmp;
tmp = label;
int column = tmp.length();
if (!opcode.empty()) {
do {
tmp.push_back(' ');
column++;
} while (column < kOpcodeTab);
tmp += opcode;
}
column = tmp.length();
if (!operand.empty()) {
do {
tmp.push_back(' ');
column++;
} while (column < kOperandTab);
tmp += operand;
}
tmp.push_back('\n');
fputs(tmp.c_str(), stdout);
}
void disassembler::reset() { void disassembler::reset() {
_arg = 0; _arg = 0;
_st = 0; _st = 0;
} }
constexpr const int kOpcodeTab = 20;
constexpr const int kOperandTab = 30;
constexpr const int kCommentTab = 80;
void indent_to(std::string &line, unsigned position) {
if (line.length() < position)
line.resize(position, ' ');
}
void disassembler::dump() { void disassembler::dump() {
@ -430,6 +491,7 @@ void disassembler::dump(const std::string &expr, unsigned size) {
void disassembler::flush() { void disassembler::flush() {
if (_st) dump(); if (_st) dump();
check_labels();
} }

12
disassembler.h
View File

@ -9,6 +9,7 @@ class disassembler {
public: public:
disassembler() = default; disassembler() = default;
virtual ~disassembler();
void operator()(uint8_t byte); void operator()(uint8_t byte);
void operator()(const std::string &expr, unsigned size); void operator()(const std::string &expr, unsigned size);
@ -53,6 +54,17 @@ class disassembler {
else _next_label = -1; else _next_label = -1;
} }
static std::string to_x(uint32_t value, unsigned bytes, char prefix = 0);
static void emit(const std::string &label);
static void emit(const std::string &label, const std::string &opcode);
static void emit(const std::string &label, const std::string &opcode, const std::string &operand);
protected:
virtual int32_t next_label(int32_t pc) {
return -1;
}
private: private:

419
dumpobj.cpp
View File

@ -14,7 +14,7 @@
#include <iterator> #include <iterator>
#include "obj816.h" #include "obj816.h"
#include "disassembler.h" #include "zrdz_disassembler.h"
#ifndef O_BINARY #ifndef O_BINARY
#define O_BINARY 0 #define O_BINARY 0
@ -40,13 +40,6 @@ enum class endian {
}; };
//extern unsigned init_flags(bool longM, bool longX);
//void dump(const std::vector<uint8_t> &data, unsigned &pc);
//extern void disasm(const std::vector<uint8_t> &data, unsigned &flags, unsigned &pc);
template<class T> template<class T>
void swap_if(T &t, std::false_type) {} void swap_if(T &t, std::false_type) {}
@ -77,9 +70,9 @@ void usage() {
#pragma pack(push, 1) #pragma pack(push, 1)
struct Header { struct Header {
uint32_t h_magic; /* magic number for detection */ uint32_t magic; /* magic number for detection */
uint16_t h_version; /* version number of object format */ uint16_t version; /* version number of object format */
uint8_t h_filtyp; /* file type, object or library */ uint8_t filetype; /* file type, object or library */
}; };
#pragma pack(pop) #pragma pack(pop)
@ -150,26 +143,6 @@ std::string read_pstring(T &iter) {
} }
struct symbol {
std::string name;
uint8_t type = S_UND;
uint8_t flags = 0;
uint8_t section = 0xff;
uint32_t offset = 0;
};
struct section {
std::string name;
uint8_t number = 0;
uint8_t flags = 0;
uint32_t size = 0;
uint32_t org = 0;
// for disassembly tracking...
uint32_t pc = 0;
std::vector<symbol> symbols;
};
std::vector<section> read_sections(const std::vector<uint8_t> &section_data) { std::vector<section> read_sections(const std::vector<uint8_t> &section_data) {
std::vector<section> sections; std::vector<section> sections;
@ -185,9 +158,7 @@ std::vector<section> read_sections(const std::vector<uint8_t> &section_data) {
if (!(s.flags & SEC_NONAME)) s.name = read_cstring(iter); if (!(s.flags & SEC_NONAME)) s.name = read_cstring(iter);
if (sections.size() < s.number) sections.resize(s.number); sections.emplace_back(std::move(s));
if (sections.size() == s.number) sections.emplace_back(std::move(s));
else sections[s.number] = std::move(s);
} }
return sections; return sections;
} }
@ -213,111 +184,6 @@ std::vector<symbol> read_symbols(const std::vector<uint8_t> &symbol_data) {
return symbols; return symbols;
} }
std::vector<symbol> labels_for_section(const std::vector<symbol> &symbols, unsigned section) {
std::vector<symbol> out;
std::copy_if(symbols.begin(), symbols.end(), std::back_inserter(out),
[section](const symbol &s) { return s.section == section && s.type != S_UND; }
);
std::sort(out.begin(), out.end(), [](const symbol &a, const symbol &b){
return a.offset > b.offset;
});
return out;
}
symbol find_symbol(const std::vector<symbol> &symbols, unsigned section, unsigned offset) {
auto iter = std::find_if(symbols.begin(), symbols.end(), [section, offset](const symbol &s){
return s.section == section && s.offset == offset && s.type != S_UND;
});
if (iter != symbols.end()) return *iter;
return symbol{};
}
void place_labels(std::vector<symbol> &labels, uint32_t pc) {
while (!labels.empty()) {
auto &label = labels.back();
if (label.offset > pc) return;
if (label.offset == pc) {
printf("%s:\n", label.name.c_str());
} else {
warnx("Unable to place label %s (offset $%04x)", label.name.c_str(), label.offset);
}
labels.pop_back();
}
}
void emit(const std::string &label) {
fputs(label.c_str(), stdout);
//fputc(':', stdout); // special case.
fputc('\n', stdout);
}
void emit(const std::string &label, const std::string &opcode) {
fputs(label.c_str(), stdout);
int column = label.length();
if (!opcode.empty()) {
do {
putc(' ', stdout);
column++;
} while (column < 20);
fputs(opcode.c_str(), stdout);
}
fputc('\n', stdout);
}
void emit(const std::string &label, const std::string &opcode, const std::string &operand) {
fputs(label.c_str(), stdout);
int column = label.length();
if (!opcode.empty()) {
do {
putc(' ', stdout);
column++;
} while (column < 20);
fputs(opcode.c_str(), stdout);
column += opcode.length();
}
if (!operand.empty()) {
do {
putc(' ', stdout);
column++;
} while (column < 30);
fputs(operand.c_str(), stdout);
column += operand.length();
}
fputc('\n', stdout);
return;
}
static std::string to_x(uint32_t x, unsigned bytes, char prefix = 0) {
std::string s;
char buffer[16];
if (prefix) s.push_back(prefix);
if (x > 0xff && bytes < 4) bytes = 4;
if (x > 0xffff && bytes < 6) bytes = 6;
if (x > 0xffffff && bytes < 8) bytes = 8;
memset(buffer, '0', sizeof(buffer));
int i = 16;
while (x) {
buffer[--i] = "0123456789abcdef"[x & 0x0f];
x >>= 4;
}
s.append(buffer + 16 - bytes, buffer + 16);
return s;
}
bool dump_obj(const char *name, int fd) bool dump_obj(const char *name, int fd)
{ {
static const char *kSections[] = { "PAGE0", "CODE", "KDATA", "DATA", "UDATA" }; static const char *kSections[] = { "PAGE0", "CODE", "KDATA", "DATA", "UDATA" };
@ -356,16 +222,6 @@ bool dump_obj(const char *name, int fd)
if (ok != h.h_namlen) errx(EX_DATAERR, "%s", name); if (ok != h.h_namlen) errx(EX_DATAERR, "%s", name);
/*
printf("name: %s\n", oname.data());
printf("record size : $%04x\n", h.h_recsize);
printf("section size : $%04x\n", h.h_secsize);
printf("symbol size : $%04x\n", h.h_symsize);
printf("option size : $%04x\n", h.h_optsize);
printf("number sections: $%04x\n", h.h_num_secs);
printf("number symbols : $%04x\n", h.h_num_syms);
*/
// records [until record_eof] // records [until record_eof]
std::vector<uint8_t> data; std::vector<uint8_t> data;
@ -385,168 +241,19 @@ bool dump_obj(const char *name, int fd)
ok = read(fd, symbol_data.data(), h.h_symsize); ok = read(fd, symbol_data.data(), h.h_symsize);
if (ok != h.h_symsize) errx(EX_DATAERR, "%s symbols truncated", name); if (ok != h.h_symsize) errx(EX_DATAERR, "%s symbols truncated", name);
std::vector<symbol> symbols = read_symbols(symbol_data); zrdz_disassembler d(read_sections(section_data), read_symbols(symbol_data));
std::vector<section> sections = read_sections(section_data);
disassembler d;
uint8_t op = REC_END; uint8_t op = REC_END;
unsigned section = SECT_CODE; // default section = CODE unsigned section = SECT_CODE; // default section = CODE
unsigned line = 0; unsigned line = 0;
std::string file; std::string file;
if (sections.size() < 5) sections.resize(5);
for (int i = 0; i < 5; ++i)
if (sections[i].name.empty()) sections[i].name = kSections[i];
d.front_matter(std::string(oname.data()));
/* custom sections... */
if (sections.size() > 5) {
printf("\n");
for (auto iter = sections.begin() + 5; iter != sections.end(); ++iter) {
std::string attr;
bool comma = false;
int flags = iter->flags;
if (flags & SEC_OFFSET) {
if (comma) attr += ", ";
attr += "SEC_OFFSET $";
attr += to_x(iter->org, 4);
comma = true;
}
if (flags & SEC_INDIRECT) {
if (comma) attr += ", ";
attr += "SEC_INDIRECT $";
attr += to_x(iter->org, 4);
comma = true;
}
#define _(x) if (flags & x) { if (comma) attr += ", "; attr += #x; comma = true; }
_(SEC_STACKED)
_(SEC_REF_ONLY)
_(SEC_CONST)
_(SEC_DIRECT)
_(SEC_NONAME)
_(SEC_DATA)
#undef _
emit(iter->name, "section", attr);
}
printf("\n");
}
emit("", "MODULE", std::string(oname.data()));
printf("\n");
d.set_pc(0);
d.set_code(true);
bool newline = false;
for (const auto &s : symbols) {
if (s.type != S_UND) continue;
emit("", "extern", s.name);
newline = true;
}
for (const auto &s : symbols) {
if (s.type == S_UND) continue;
if (s.type == S_ABS) continue; // ? equ/gequ
// only if s.type == S_REL?
sections[s.section].symbols.push_back(s);
if (s.flags & SF_GBL) {
emit("", "public", s.name);
newline = true;
}
}
if (newline) printf("\n");
for (auto &section : sections) {
std::sort(section.symbols.begin(), section.symbols.end(), [](const symbol &a, const symbol &b){
return a.offset > b.offset;
});
}
// print out any reference-only sections and symbols.
for (auto &section : sections) {
if ((section.flags & SEC_REF_ONLY) == 0) continue;
if (section.size == 0) continue;
emit("", section.name);
if (section.org) emit("",".org", to_x(section.org,4,'$'));
uint32_t pc = section.org;
auto &symbols = section.symbols;
while (!symbols.empty()) {
auto &s = symbols.back();
if (s.offset > pc) {
emit("","ds", std::to_string(s.offset - pc));
pc = s.offset;
}
emit(s.name);
symbols.pop_back();
}
if (pc < section.size)
emit("","ds",std::to_string(section.size - pc));
emit("","ends");
printf("\n");
}
//
// print equates.
// doesn the section matter?
newline = false;
for (const auto &s : symbols) {
if (s.type == (ST_EQU << 4) + S_ABS) {
emit(s.name,"gequ", to_x(s.offset, 4, '$'));
newline = true;
}
}
if (newline) printf("\n");
d.set_label_callback([&section, &sections](int32_t offset) -> int32_t {
auto &symbols = sections[section].symbols;
if (offset >= 0) {
while(!symbols.empty()) {
auto &s = symbols.back();
if (s.offset > offset) return s.offset;
if (s.offset == offset) {
emit(s.name);
} else {
std::string tmp = "; ";
tmp += s.name;
tmp += " = $";
tmp += to_x(s.offset, 4);
emit("", tmp);
}
symbols.pop_back();
continue;
}
}
if (symbols.empty()) return -1;
return symbols.back().offset;
});
//std::vector<symbol> labels = labels_for_section(symbols, section);
emit("", "CODE", "; section 1");
auto iter = data.begin(); auto iter = data.begin();
while (iter != data.end()) { while (iter != data.end()) {
//place_labels(labels, d.pc());
op = read_8(iter); op = read_8(iter);
if (op == 0) break; if (op == 0) break;
@ -554,7 +261,6 @@ bool dump_obj(const char *name, int fd)
auto end = iter + op; auto end = iter + op;
while (iter != end) { while (iter != end) {
d(*iter++); d(*iter++);
//place_labels(labels, d.pc());
} }
continue; continue;
} }
@ -587,41 +293,18 @@ bool dump_obj(const char *name, int fd)
uint8_t section = read_8(iter); uint8_t section = read_8(iter);
uint32_t offset = read_32(iter); uint32_t offset = read_32(iter);
std::string name; std::string name = d.location_name(section, offset);
stack.emplace_back(std::move(name));
symbol s = find_symbol(symbols, section, offset);
if (s.type) {
name = s.name;
} else {
if (section < sizeof(kSections) / sizeof(kSections[0]))
name = kSections[section];
else {
snprintf(buffer, sizeof(buffer), "section%d", section);
name = buffer;
}
}
if (offset) {
snprintf(buffer, sizeof(buffer), "+$%04x", offset);
name += buffer;
}
stack.push_back(name);
break; break;
} }
case OP_VAL: case OP_VAL:
snprintf(buffer, sizeof(buffer), "$%04x", read_32(iter)); stack.push_back(d.to_x(read_32(iter), 4, '$'));
stack.push_back(buffer);
break; break;
case OP_SYM: { case OP_SYM: {
uint16_t symbol = read_16(iter); uint16_t symbol = read_16(iter);
if (symbol < symbols.size()) stack.push_back(symbols[symbol].name); stack.emplace_back(d.symbol_name(symbol));
else {
snprintf(buffer, sizeof(buffer), "symbol $%02x", symbol);
stack.push_back(buffer);
}
break; break;
} }
@ -695,45 +378,45 @@ bool dump_obj(const char *name, int fd)
switch(op) { switch(op) {
case D_LONGA_ON: case D_LONGA_ON:
d.set_m(true); d.set_m(true);
emit("", "longa", "on"); d.emit("", "longa", "on");
break; break;
case D_LONGA_OFF: case D_LONGA_OFF:
d.set_m(false); d.set_m(false);
emit("", "longa", "off"); d.emit("", "longa", "off");
break; break;
case D_LONGI_ON: case D_LONGI_ON:
d.set_x(true); d.set_x(true);
emit("", "longi", "on"); d.emit("", "longi", "on");
break; break;
case D_LONGI_OFF: case D_LONGI_OFF:
d.set_x(false); d.set_x(false);
emit("", "longi", "off"); d.emit("", "longi", "off");
break; break;
case D_C_FILE: { case D_C_FILE: {
file = read_cstring(iter); file = read_cstring(iter);
line = read_16(iter); line = read_16(iter);
std::string tmp = file + ", " + std::to_string(line); std::string tmp = file + ", " + std::to_string(line);
emit("", ".file", tmp); d.emit("", ".file", tmp);
break; break;
} }
case D_C_LINE: { case D_C_LINE: {
line = read_16(iter); line = read_16(iter);
emit("",".line", std::to_string(line)); d.emit("",".line", std::to_string(line));
break; break;
} }
case D_C_BLOCK: { case D_C_BLOCK: {
uint16_t block = read_16(iter); uint16_t block = read_16(iter);
emit("",".block", std::to_string(line)); d.emit("",".block", std::to_string(line));
break; break;
} }
case D_C_ENDBLOCK: { case D_C_ENDBLOCK: {
uint16_t line = read_16(iter); uint16_t line = read_16(iter);
emit("",".endblock", std::to_string(line)); d.emit("",".endblock", std::to_string(line));
break; break;
} }
case D_C_FUNC: { case D_C_FUNC: {
uint16_t arg = read_16(iter); uint16_t arg = read_16(iter);
emit("",".function", std::to_string(line)); d.emit("",".function", std::to_string(line));
break; break;
} }
case D_C_ENDFUNC: { case D_C_ENDFUNC: {
@ -744,7 +427,7 @@ bool dump_obj(const char *name, int fd)
tmp = std::to_string(line) + ", " tmp = std::to_string(line) + ", "
+ std::to_string(local_offset) + ", " + std::to_string(local_offset) + ", "
+ std::to_string(arg_offset); + std::to_string(arg_offset);
emit("",".endfunc", tmp); d.emit("",".endfunc", tmp);
break; break;
} }
@ -761,11 +444,11 @@ bool dump_obj(const char *name, int fd)
std::string tmp; std::string tmp;
tmp = name + ", " + std::to_string(size) + ", " + std::to_string(tag); tmp = name + ", " + std::to_string(size) + ", " + std::to_string(tag);
emit("", opname, tmp); d.emit("", opname, tmp);
break; break;
} }
case D_C_EOS: { case D_C_EOS: {
emit("", ".eos"); d.emit("", ".eos");
break; break;
} }
@ -791,7 +474,7 @@ bool dump_obj(const char *name, int fd)
if (version == 0) { if (version == 0) {
std::string svalue; std::string svalue;
svalue = symbols[value].name; svalue = d.symbol_name(value);
attr = name + ", " + svalue; attr = name + ", " + svalue;
} }
@ -827,7 +510,7 @@ bool dump_obj(const char *name, int fd)
} }
emit("", opname, attr); d.emit("", opname, attr);
break; break;
} }
@ -842,33 +525,15 @@ bool dump_obj(const char *name, int fd)
break; break;
case REC_SECT: { case REC_SECT: {
d.flush();
uint8_t sec = read_8(iter); uint8_t sec = read_8(iter);
//printf("\t.sect\t%d\n", sec); d.set_section(sec);
if (sec != section) {
if (sec >= sections.size()) {
warnx("Undefined section %d", sec);
}
const auto &s = sections[sec];
emit("", "ends", std::string("; end section ") + std::to_string(section));
printf("\n");
emit("", s.name, std::string("; section ") + std::to_string(sec));
sections[section].pc = d.pc();
section = sec;
d.set_pc(s.pc);
d.recalc_next_label();
d.set_code((s.flags & SEC_DATA) == 0);
}
break; break;
} }
case REC_ORG: { case REC_ORG: {
d.flush(); d.flush();
uint32_t org = read_32(iter); uint32_t org = read_32(iter);
emit("", ".org", to_x(org, 4, '$')); d.emit("", ".org", d.to_x(org, 4, '$'));
d.set_pc(org); d.set_pc(org);
break; break;
} }
@ -877,7 +542,7 @@ bool dump_obj(const char *name, int fd)
d.flush(); d.flush();
uint16_t count = read_16(iter); uint16_t count = read_16(iter);
// todo -- need to coordinate with label printer/disassembler. // todo -- need to coordinate with label printer/disassembler.
emit("", "ds", to_x(count, 4, '$')); d.emit("", "ds", d.to_x(count, 4, '$'));
d.set_pc(d.pc() + count); d.set_pc(d.pc() + count);
break; break;
} }
@ -893,23 +558,13 @@ bool dump_obj(const char *name, int fd)
errx(EX_DATAERR, "%s: unknown opcode %02x", name, op); errx(EX_DATAERR, "%s: unknown opcode %02x", name, op);
} }
} }
// dump any unfinished business.
d.flush();
//place_labels(labels, d.pc());
/* d.back_matter();
for(auto &label : labels) {
warnx("Unable to place label %s (offset $%04x)", label.name.c_str(), label.offset);
}
*/
emit("", "ends");
printf("\n");
emit("", "endmod");
printf("\n");
if (iter != data.end() || op != REC_END) errx(EX_DATAERR, "%s records ended early", name); if (iter != data.end() || op != REC_END) errx(EX_DATAERR, "%s records ended early", name);
#if 0
if (flags._S) { if (flags._S) {
printf("; sections\n"); printf("; sections\n");
for (auto &s : sections) { for (auto &s : sections) {
@ -925,7 +580,7 @@ bool dump_obj(const char *name, int fd)
s.name.c_str(), s.type, s.flags, s.section, s.offset); s.name.c_str(), s.type, s.flags, s.section, s.offset);
} }
} }
#endif
return true; return true;
} }
@ -1006,15 +661,15 @@ void dump(const char *name) {
if (ok != sizeof(h)) if (ok != sizeof(h))
errx(EX_DATAERR, "%s is not an object file", name); errx(EX_DATAERR, "%s is not an object file", name);
le_to_host(h.h_magic); le_to_host(h.magic);
le_to_host(h.h_version); le_to_host(h.version);
le_to_host(h.h_filtyp); le_to_host(h.filetype);
if (h.h_magic != MOD_MAGIC || h.h_version != 1 || h.h_filtyp > 2) if (h.magic != MOD_MAGIC || h.version != 1 || h.filetype > 2)
errx(EX_DATAERR, "%s is not an object file", name); errx(EX_DATAERR, "%s is not an object file", name);
lseek(fd, 0, SEEK_SET); lseek(fd, 0, SEEK_SET);
if (h.h_filtyp == 2) dump_lib(name, fd); if (h.filetype == 2) dump_lib(name, fd);
// files may contain multiple modules. // files may contain multiple modules.
while (dump_obj(name, fd)) /* ... */; while (dump_obj(name, fd)) /* ... */;

5
obj816.h
View File

@ -1,3 +1,6 @@
#ifndef __obj816_h__
#define __obj816_h__
#include <stdint.h> #include <stdint.h>
#pragma pack(push, 1) #pragma pack(push, 1)
@ -174,3 +177,5 @@ enum {SECT_PAGE0, SECT_CODE, SECT_KDATA, SECT_DATA, SECT_UDATA };
/**************************************************/ /**************************************************/
#pragma pack(pop) #pragma pack(pop)
#endif

322
zrdz_disassembler.cpp Normal file
View File

@ -0,0 +1,322 @@
#include "zrdz_disassembler.h"
#include <err.h>
#include <stdio.h>
static constexpr const int equ_type = (ST_EQU << 4) | S_ABS;
zrdz_disassembler::zrdz_disassembler(std::vector<section> &&sections, std::vector<symbol> &&symbols) :
_symbols(std::move(symbols))
{
// do not sort _symbols ... order matters for lookup by entry number.
#if 0
// sort by section / address.
std::sort(_symbols.begin(), _symbols.end(), [](const symbol &a, const symbol &b){
if (a.section < b.section) return true;
if (a.section == b.section && a.offset < b.offset) return true;
return false;
});
#endif
for (auto &s : sections) {
if (s.number >= _sections.size()) _sections.resize(s.number + 1);
_sections[s.number].valid = true;
_sections[s.number].pc = s.org;
_sections[s.number] = std::move(s);
}
for (auto &s : _symbols) {
if (s.type == S_UND) continue;
if (s.section >= _sections.size()) {
warnx("invalid section %d for symbol %s", s.section, s.name.c_str());
continue;
}
auto &e = _sections[s.section];
if (!e.valid) {
warnx("invalid section %d for symbol %s", s.section, s.name.c_str());
continue;
}
if (s.type == equ_type) continue;
e.symbols.emplace_back(s);
}
if (_sections.size() < 5) _sections.resize(5);
_sections[SECT_PAGE0].name = "page0";
_sections[SECT_CODE].name = "code";
_sections[SECT_KDATA].name = "kdata";
_sections[SECT_DATA].name = "data";
_sections[SECT_UDATA].name = "udata";
#if 0
// now place equates at the end...
for (auto &s : _symbols) {
if (s.type != equ_type) continue;
auto &sec = _sections[s.section];
sec.symbols.emplace_back(s);
}
#endif
}
zrdz_disassembler::~zrdz_disassembler() {
}
void zrdz_disassembler::front_matter(const std::string &module) {
emit("", "module", module);
putchar('\n');
print_externs();
// reference-only sections.
for (auto &e : _sections) {
if (!e.valid) continue;
if ((e.flags & SEC_REF_ONLY) == 0) continue;
if (e.size == 0 && e.symbols.empty()) continue;
e.processed = true;
print_section(e);
print_globals(e.number);
print_equs(e.number);
if (e.org) emit("", ".org", to_x(e.org, 4, '$'));
uint32_t pc = e.org;
for (; !e.symbols.empty(); e.symbols.pop_back()) {
auto &s = e.symbols.back();
if (s.offset > pc) {
emit("","ds", std::to_string(s.offset - pc));
pc = s.offset;
}
emit(s.name);
}
if (pc < e.size)
emit("","ds",std::to_string(e.size - pc));
emit("", "ends");
putchar('\n');
}
set_section(1);
/*
emit("","code");
print_globals(1);
print_equs(1);
_sections[1].processed = true;
*/
}
void zrdz_disassembler::back_matter() {
flush();
_section = -1;
// todo -- print remaining symbols.
// todo -- print any empty sections?
emit("", "ends");
putchar('\n');
for (auto &e : _sections) {
if (e.processed) continue;
if (!e.valid) continue;
print_section(e);
print_globals(e.number);
print_equs(e.number);
emit("","ends");
e.processed = true;
// ...check for symbols?
}
emit("", "endmod");
putchar('\n');
}
void zrdz_disassembler::print_externs() {
std::vector<std::string> tmp;
for (auto &s : _symbols) {
if (s.type == S_UND) tmp.push_back(s.name);
}
if (tmp.empty()) return;
std::sort(tmp.begin(), tmp.end());
for (const auto &s : tmp) emit("", "extern", s);
putchar('\n');
}
void zrdz_disassembler::print_globals(int section) {
std::vector<std::string> tmp;
for (auto &s : _symbols) {
if (s.section == section && s.flags & SF_GBL) tmp.push_back(s.name);
}
if (tmp.empty()) return;
std::sort(tmp.begin(), tmp.end());
for (const auto &s : tmp) emit("", "public", s);
putchar('\n');
}
void zrdz_disassembler::print_equs(int section) {
std::vector<symbol> tmp;
std::copy_if(_symbols.begin(), _symbols.end(), std::back_inserter(tmp),
[=](const symbol &s) { return s.section == section && s.type == equ_type; }
);
if (tmp.empty()) return;
std::sort(tmp.begin(), tmp.end(),
[](const symbol &a, const symbol &b) { return a.name < b.name; }
);
for (const auto &s : tmp) emit(s.name, "gequ", to_x(s.offset, 4, '$'));
putchar('\n');
}
void zrdz_disassembler::print_section(const entry &e) {
std::string attr;
bool comma = false;
if (e.processed || e.number < 5) {
emit("", e.name);
return;
}
if (e.flags & SEC_OFFSET) {
if (comma) attr += ", ";
attr += "SEC_OFFSET $";
attr += to_x(e.org, 4);
comma = true;
}
if (e.flags & SEC_INDIRECT) {
if (comma) attr += ", ";
attr += "SEC_INDIRECT $";
attr += to_x(e.org, 4);
comma = true;
}
#define _(x) if (e.flags & x) { if (comma) attr += ", "; attr += #x; comma = true; }
_(SEC_STACKED)
_(SEC_REF_ONLY)
_(SEC_CONST)
_(SEC_DIRECT)
_(SEC_NONAME)
_(SEC_DATA)
#undef _
emit(e.name, "section", attr);
putchar('\n');
}
void zrdz_disassembler::set_section(int section) {
if (section == _section) return;
if (section >= _sections.size()) {
warnx("Invalid section %d", section);
return;
}
auto &e = _sections[section];
if (!e.valid) {
warnx("Invalid section %d", section);
return;
}
if (_section >= 0) {
flush();
_sections[_section].pc = pc();
emit("", "ends");
putchar('\n');
}
print_section(e);
if (!e.processed) {
print_globals(section);
print_equs(section);
e.processed = true;
}
_section = section;
set_pc(e.pc);
set_code((e.flags & SEC_DATA) == 0);
recalc_next_label();
}
int32_t zrdz_disassembler::next_label(int32_t pc) {
if (_section < 0) return -1;
auto &symbols = _sections[_section].symbols;
if (pc >= 0) {
for(; !symbols.empty(); symbols.pop_back()) {
auto &s = symbols.back();
if (s.offset > pc) return s.offset;
if (s.offset == pc) emit(s.name);
else {
warnx("Unable to place symbol %s at offset $%04x in section %d",
s.name.c_str(), s.offset, _section);
}
}
}
if (symbols.empty()) return -1;
return symbols.back().offset;
}
std::string zrdz_disassembler::location_name(unsigned section, uint32_t offset) const {
if (section >= _sections.size()) {
warnx("Invalid section %d", section);
return "$";
}
auto &e = _sections[section];
if (!e.valid) {
warnx("Invalid section %d", section);
return "$";
}
// todo -- need to verify relative/absolute value
auto iter = std::find_if(_symbols.begin(), _symbols.end(), [=](const symbol &s){
if (s.type == equ_type) return false;
if (s.type == S_UND) return false;
return s.section == section && s.offset == offset;
});
if (iter != _symbols.end()) return iter->name;
//fallback to section name + offset
std::string tmp = e.name;
if (tmp.empty()) tmp = "section" + std::to_string(section);
if (offset) tmp += "+$" + to_x(offset, 4);
return tmp;
}
std::string zrdz_disassembler::symbol_name(unsigned entry) const {
if (entry >= _symbols.size()) {
warnx("Invalid symbol %d", entry);
return "$";
}
return _symbols[entry].name;
}

74
zrdz_disassembler.h Normal file
View File

@ -0,0 +1,74 @@
#ifndef __zrdz_disassembler_h__
#define __zrdz_disassembler_h__
#include <string>
#include <vector>
#include <stdint.h>
#include "disassembler.h"
#include "obj816.h"
struct symbol {
std::string name;
uint8_t type = S_UND;
uint8_t flags = 0;
uint8_t section = 0xff;
uint32_t offset = 0;
};
struct section {
std::string name;
uint8_t number = 0;
uint8_t flags = 0;
uint32_t size = 0;
uint32_t org = 0;
};
class zrdz_disassembler final : public disassembler {
public:
zrdz_disassembler(std::vector<section> &&, std::vector<symbol> &&);
virtual ~zrdz_disassembler();
//int section() const { return _section; }
void set_section(int);
void front_matter(const std::string &module_name);
void back_matter();
std::string location_name(unsigned section, uint32_t offset) const;
std::string symbol_name(unsigned entry) const;
protected:
virtual int32_t next_label(int32_t pc) override;
private:
struct entry : public section {
using section::section;
using section::operator=;
bool processed = false;
bool valid = false;
std::vector<symbol> symbols;
uint32_t pc = 0;
};
std::vector<symbol> _symbols;
std::vector<entry> _sections;
int _section = -1;
void print_section(const entry &e);
void print_externs();
void print_globals(int section);
void print_equs(int section);
};
#endif