Hojun-Cho
4fe2e25da4
Every interned symbol -- including branch labels defined via p.label -- was emitted STB_GLOBAL, so labels were collide-able across objects and the header's "GLOBAL symbols only" contract was enforced nowhere (asym.isglobal was set but never read). Labels stay interned for fixup resolution; the symtab now carries only exported definitions, undefined externs, and reloc-referenced syms (a reloc pre-pass marks those). rt/ensure.o drops from 15 emitted syms to 2. Bootstrap 991-995 hold byte-identical on the new format.
318 lines
8.4 KiB
C
318 lines
8.4 KiB
C
/*
|
|
* Layout (in file order):
|
|
* [0] ELF header
|
|
* [1] Section .text (program bytes)
|
|
* [2] Section .rela.text (relocations)
|
|
* [3] Section .data (writable; only present if datalen > 0)
|
|
* [4] Section .symtab
|
|
* [5] Section .strtab
|
|
* [6] Section .shstrtab
|
|
* [7] Section header table
|
|
*
|
|
* When no DATAW directive appears in the input, the .data section is
|
|
* omitted entirely so output stays byte-identical to the pre-DATAW
|
|
* format. Test 991 (selfhost .o byte-diff) depends on this.
|
|
*
|
|
* Symtab indices: 0 = STN_UNDEF, 1.. = our syms (no file entry).
|
|
* Every emitted symbol is GLOBAL (no LOCAL ordering rules); defined
|
|
* non-exported label syms are pruned unless a reloc references them.
|
|
*/
|
|
#include "a.h"
|
|
#include <stdlib.h>
|
|
#include <string.h>
|
|
#include <stdio.h>
|
|
|
|
#define ELFMAG "\x7f""ELF"
|
|
#define ELFCLASS64 2
|
|
#define ELFDATA2LSB 1
|
|
#define EV_CURRENT 1
|
|
#define ET_REL 1
|
|
#define EM_X86_64 62
|
|
|
|
#define SHT_NULL 0
|
|
#define SHT_PROGBITS 1
|
|
#define SHT_SYMTAB 2
|
|
#define SHT_STRTAB 3
|
|
#define SHT_RELA 4
|
|
|
|
#define SHF_WRITE 0x1
|
|
#define SHF_ALLOC 0x2
|
|
#define SHF_EXECINSTR 0x4
|
|
#define SHF_INFO_LINK 0x40
|
|
|
|
#define STB_LOCAL 0
|
|
#define STB_GLOBAL 1
|
|
#define STT_NOTYPE 0
|
|
#define STT_OBJECT 1
|
|
#define STT_FUNC 2
|
|
#define ELF64_ST_INFO(b,t) (((b) << 4) + ((t) & 0xf))
|
|
|
|
#define R_X86_64_PC32 2
|
|
#define R_X86_64_PLT32 4
|
|
#define ELF64_R_INFO(s,t) (((u64)(s) << 32) | ((u64)(t) & 0xffffffff))
|
|
|
|
typedef struct Buf Buf;
|
|
struct Buf { u8 *p; size_t n, cap; };
|
|
|
|
static void
|
|
bput(Buf *b, const void *src, size_t n)
|
|
{
|
|
if (b->n + n > b->cap) {
|
|
size_t nc = b->cap ? b->cap * 2 : 256;
|
|
while (nc < b->n + n) nc *= 2;
|
|
b->p = realloc(b->p, nc);
|
|
b->cap = nc;
|
|
}
|
|
memcpy(b->p + b->n, src, n);
|
|
b->n += n;
|
|
}
|
|
|
|
static u32 stput(Buf *st, const char *s) {
|
|
u32 off = (u32)st->n;
|
|
bput(st, s, strlen(s) + 1);
|
|
return off;
|
|
}
|
|
|
|
#pragma pack(push, 1)
|
|
typedef struct {
|
|
u8 e_ident[16];
|
|
u16 e_type, e_machine;
|
|
u32 e_version;
|
|
u64 e_entry, e_phoff, e_shoff;
|
|
u32 e_flags;
|
|
u16 e_ehsize, e_phentsize, e_phnum, e_shentsize, e_shnum, e_shstrndx;
|
|
} Ehdr;
|
|
|
|
typedef struct {
|
|
u32 sh_name, sh_type;
|
|
u64 sh_flags, sh_addr, sh_offset, sh_size;
|
|
u32 sh_link, sh_info;
|
|
u64 sh_addralign, sh_entsize;
|
|
} Shdr;
|
|
|
|
typedef struct {
|
|
u32 st_name;
|
|
u8 st_info, st_other;
|
|
u16 st_shndx;
|
|
u64 st_value, st_size;
|
|
} Sym64;
|
|
|
|
typedef struct {
|
|
u64 r_offset;
|
|
u64 r_info;
|
|
i64 r_addend;
|
|
} Rela64;
|
|
#pragma pack(pop)
|
|
|
|
int
|
|
a_emit_elf(Asm *a, FILE *f)
|
|
{
|
|
Buf shstr = {0}, str = {0}, sym = {0}, rela = {0}, relad = {0};
|
|
stput(&shstr, ""); /* idx 0 = empty */
|
|
stput(&str, "");
|
|
|
|
const int has_data = (a->datalen > 0);
|
|
int has_data_relocs = 0;
|
|
for (Areloc *r = a->relocs; r; r = r->next) {
|
|
if (r->section == 1) { has_data_relocs = 1; break; }
|
|
}
|
|
|
|
/* Section indices.
|
|
* Without data, without data-relocs:
|
|
* 1=.text, 2=.rela.text, 3=.symtab, 4=.strtab, 5=.shstrtab
|
|
* With data, no data-relocs:
|
|
* 1=.text, 2=.rela.text, 3=.data, 4=.symtab, 5=.strtab, 6=.shstrtab
|
|
* With data + data-relocs:
|
|
* 1=.text, 2=.rela.text, 3=.data, 4=.rela.data, 5=.symtab,
|
|
* 6=.strtab, 7=.shstrtab
|
|
*/
|
|
const u16 SH_TEXT = 1;
|
|
const u16 SH_DATA = has_data ? 3 : 0;
|
|
const u16 SH_RELAD = (has_data && has_data_relocs) ? 4 : 0;
|
|
const u16 SH_SYMTAB = has_data
|
|
? (has_data_relocs ? 5 : 4)
|
|
: 3;
|
|
const u16 SH_STRTAB = (u16)(SH_SYMTAB + 1);
|
|
const u16 SH_SHSTR = (u16)(SH_STRTAB + 1);
|
|
|
|
/* Section name offsets. Append .data's name only when used so the
|
|
* .shstrtab buffer stays byte-identical for the no-DATAW case.
|
|
* Same for .rela.data — only present when data relocs exist. */
|
|
u32 shn_text = stput(&shstr, ".text");
|
|
u32 shn_rela = stput(&shstr, ".rela.text");
|
|
u32 shn_data = has_data ? stput(&shstr, ".data") : 0;
|
|
u32 shn_relad = (has_data && has_data_relocs)
|
|
? stput(&shstr, ".rela.data") : 0;
|
|
u32 shn_symtab = stput(&shstr, ".symtab");
|
|
u32 shn_strtab = stput(&shstr, ".strtab");
|
|
u32 shn_shstrtab = stput(&shstr, ".shstrtab");
|
|
|
|
/* Symbol 0 — STN_UNDEF */
|
|
{
|
|
Sym64 z = {0};
|
|
bput(&sym, &z, sizeof z);
|
|
}
|
|
|
|
/* Data symbols carry STT_OBJECT and st_shndx=SH_DATA; everything
|
|
* else keeps the legacy STT_FUNC/SH_TEXT shape so non-DATAW
|
|
* outputs stay byte-identical. */
|
|
/* Branch labels are interned for fixup resolution but are NOT
|
|
* link-visible: emitting them (all bindings here are GLOBAL)
|
|
* made every label collide-able across objects. Only exported
|
|
* definitions (is_global), undefined externs, and reloc-
|
|
* referenced syms reach the symtab. */
|
|
for (Areloc *r = a->relocs; r; r = r->next)
|
|
r->sym->has_reloc = 1;
|
|
int idx = 1;
|
|
for (Asym *s = a->syms; s; s = s->next) {
|
|
if (s->defined && !s->is_global && !s->has_reloc)
|
|
continue;
|
|
Sym64 e = {0};
|
|
e.st_name = stput(&str, s->name);
|
|
if (s->defined) {
|
|
if (s->is_data) {
|
|
e.st_info = ELF64_ST_INFO(STB_GLOBAL, STT_OBJECT);
|
|
e.st_shndx = SH_DATA;
|
|
} else {
|
|
e.st_info = ELF64_ST_INFO(STB_GLOBAL, STT_FUNC);
|
|
e.st_shndx = SH_TEXT;
|
|
}
|
|
e.st_value = s->addr;
|
|
e.st_size = 0;
|
|
} else {
|
|
e.st_info = ELF64_ST_INFO(STB_GLOBAL, STT_NOTYPE);
|
|
e.st_shndx = 0;
|
|
}
|
|
bput(&sym, &e, sizeof e);
|
|
s->idx = idx++;
|
|
}
|
|
|
|
/* Build relocations — split into two buffers so we can emit
|
|
* .rela.text and (conditionally) .rela.data separately. */
|
|
for (Areloc *r = a->relocs; r; r = r->next) {
|
|
Rela64 re;
|
|
re.r_offset = r->off;
|
|
re.r_info = ELF64_R_INFO((u64)r->sym->idx, (u64)r->kind);
|
|
re.r_addend = r->addend;
|
|
bput(r->section == 1 ? &relad : &rela, &re, sizeof re);
|
|
}
|
|
|
|
u64 off = sizeof(Ehdr);
|
|
u64 off_text = off; off += a->textlen;
|
|
u64 off_rela = off; off += rela.n;
|
|
u64 off_data = off; if (has_data) off += a->datalen;
|
|
u64 off_relad = off; if (has_data && has_data_relocs) off += relad.n;
|
|
u64 off_sym = off; off += sym.n;
|
|
u64 off_str = off; off += str.n;
|
|
u64 off_shstr= off; off += shstr.n;
|
|
while (off % 8) off++;
|
|
u64 off_shdr = off;
|
|
const int NSECT = has_data
|
|
? (has_data_relocs ? 8 : 7)
|
|
: 6; /* null + reals */
|
|
|
|
Ehdr eh = {0};
|
|
memcpy(eh.e_ident, "\x7f""ELF", 4);
|
|
eh.e_ident[4] = ELFCLASS64;
|
|
eh.e_ident[5] = ELFDATA2LSB;
|
|
eh.e_ident[6] = EV_CURRENT;
|
|
eh.e_type = ET_REL;
|
|
eh.e_machine = EM_X86_64;
|
|
eh.e_version = EV_CURRENT;
|
|
eh.e_shoff = off_shdr;
|
|
eh.e_ehsize = sizeof(Ehdr);
|
|
eh.e_shentsize = sizeof(Shdr);
|
|
eh.e_shnum = NSECT;
|
|
eh.e_shstrndx = SH_SHSTR;
|
|
|
|
fwrite(&eh, 1, sizeof eh, f);
|
|
if (a->textlen) fwrite(a->text, 1, a->textlen, f);
|
|
fwrite(rela.p, 1, rela.n, f);
|
|
if (has_data) fwrite(a->data, 1, a->datalen, f);
|
|
if (has_data && has_data_relocs) fwrite(relad.p, 1, relad.n, f);
|
|
fwrite(sym.p, 1, sym.n, f);
|
|
fwrite(str.p, 1, str.n, f);
|
|
fwrite(shstr.p, 1, shstr.n, f);
|
|
while ((u64)ftell(f) % 8) fputc(0, f);
|
|
|
|
/* section header table */
|
|
Shdr sh = {0};
|
|
fwrite(&sh, 1, sizeof sh, f); /* SHT_NULL */
|
|
|
|
memset(&sh, 0, sizeof sh);
|
|
sh.sh_name = shn_text;
|
|
sh.sh_type = SHT_PROGBITS;
|
|
sh.sh_flags = SHF_ALLOC | SHF_EXECINSTR;
|
|
sh.sh_offset = off_text;
|
|
sh.sh_size = a->textlen;
|
|
sh.sh_addralign = 1;
|
|
fwrite(&sh, 1, sizeof sh, f);
|
|
|
|
memset(&sh, 0, sizeof sh);
|
|
sh.sh_name = shn_rela;
|
|
sh.sh_type = SHT_RELA;
|
|
sh.sh_flags = SHF_INFO_LINK;
|
|
sh.sh_offset = off_rela;
|
|
sh.sh_size = rela.n;
|
|
sh.sh_link = SH_SYMTAB;
|
|
sh.sh_info = 1; /* applies to .text */
|
|
sh.sh_addralign = 8;
|
|
sh.sh_entsize = sizeof(Rela64);
|
|
fwrite(&sh, 1, sizeof sh, f);
|
|
|
|
if (has_data) {
|
|
memset(&sh, 0, sizeof sh);
|
|
sh.sh_name = shn_data;
|
|
sh.sh_type = SHT_PROGBITS;
|
|
sh.sh_flags = SHF_ALLOC | SHF_WRITE;
|
|
sh.sh_offset = off_data;
|
|
sh.sh_size = a->datalen;
|
|
sh.sh_addralign = 8;
|
|
fwrite(&sh, 1, sizeof sh, f);
|
|
}
|
|
|
|
if (has_data && has_data_relocs) {
|
|
memset(&sh, 0, sizeof sh);
|
|
sh.sh_name = shn_relad;
|
|
sh.sh_type = SHT_RELA;
|
|
sh.sh_flags = SHF_INFO_LINK;
|
|
sh.sh_offset = off_relad;
|
|
sh.sh_size = relad.n;
|
|
sh.sh_link = SH_SYMTAB;
|
|
sh.sh_info = SH_DATA; /* applies to .data */
|
|
sh.sh_addralign = 8;
|
|
sh.sh_entsize = sizeof(Rela64);
|
|
fwrite(&sh, 1, sizeof sh, f);
|
|
}
|
|
|
|
memset(&sh, 0, sizeof sh);
|
|
sh.sh_name = shn_symtab;
|
|
sh.sh_type = SHT_SYMTAB;
|
|
sh.sh_offset = off_sym;
|
|
sh.sh_size = sym.n;
|
|
sh.sh_link = SH_STRTAB;
|
|
sh.sh_info = 1; /* one local: STN_UNDEF */
|
|
sh.sh_addralign = 8;
|
|
sh.sh_entsize = sizeof(Sym64);
|
|
fwrite(&sh, 1, sizeof sh, f);
|
|
|
|
memset(&sh, 0, sizeof sh);
|
|
sh.sh_name = shn_strtab;
|
|
sh.sh_type = SHT_STRTAB;
|
|
sh.sh_offset = off_str;
|
|
sh.sh_size = str.n;
|
|
sh.sh_addralign = 1;
|
|
fwrite(&sh, 1, sizeof sh, f);
|
|
|
|
memset(&sh, 0, sizeof sh);
|
|
sh.sh_name = shn_shstrtab;
|
|
sh.sh_type = SHT_STRTAB;
|
|
sh.sh_offset = off_shstr;
|
|
sh.sh_size = shstr.n;
|
|
sh.sh_addralign = 1;
|
|
fwrite(&sh, 1, sizeof sh, f);
|
|
|
|
free(shstr.p); free(str.p); free(sym.p); free(rela.p); free(relad.p);
|
|
return 0;
|
|
}
|