ww/cmd/w6l/dyn.c

/*
 * dyn.c — load a shared object (ET_DYN) so the linker knows which
 * symbols it exports and which DT_NEEDED entry to record. We do not
 * pull bytes from the .so; the dynamic loader maps it at runtime.
 *
 * Each call appends one Lso to lnk->sos. `l_so_provides` answers
 * "does this .so export the named symbol?" — l_resolve uses that to
 * promote unresolved references to dynamic.
 */
#include "l.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#define ET_DYN			3

#define SHT_DYNAMIC		6
#define SHT_DYNSYM		11
/* GNU extensions, sh_type values. */
#define SHT_GNU_VERDEF		0x6ffffffd
#define SHT_GNU_VERNEED		0x6ffffffe
#define SHT_GNU_VERSYM		0x6fffffff

#define DT_NULL			0
#define DT_SONAME		14
#define DT_STRTAB		5

/* Versym special values: 0 = local, 1 = base/global. */
#define VER_NDX_LOCAL		0
#define VER_NDX_GLOBAL		1
#define VER_FLG_BASE		1
#define VERSYM_HIDDEN		0x8000
#define VERSYM_VERSION		0x7fff

#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 {
	i64 d_tag;
	u64 d_val;
} Dyn64;

typedef struct {
	u16 vd_version;
	u16 vd_flags;
	u16 vd_ndx;
	u16 vd_cnt;
	u32 vd_hash;
	u32 vd_aux;
	u32 vd_next;
} Verdef;

typedef struct {
	u32 vda_name;
	u32 vda_next;
} Verdaux;
#pragma pack(pop)

/* ELF binding values. STB_GLOBAL = 1, STB_WEAK = 2. */
#define ST_BIND(i)	((i) >> 4)

int
l_load_so(Lnk *l, const char *path)
{
	u8 *buf;
	u64 len;
	if (l_read_all(path, &buf, &len) < 0) {
		fprintf(stderr, "w6l: %s: cannot read\n", path);
		return -1;
	}
	if (len < sizeof(Ehdr)) {
		fprintf(stderr, "w6l: %s: short ELF\n", path);
		free(buf);
		return -1;
	}
	Ehdr *eh = (Ehdr *)buf;
	if (memcmp(eh->e_ident, "\x7f""ELF", 4) != 0
	    || eh->e_ident[4] != 2
	    || eh->e_machine != 62 /* EM_X86_64 */
	    || eh->e_type != ET_DYN) {
		fprintf(stderr, "w6l: %s: not an amd64 ET_DYN\n", path);
		free(buf);
		return -1;
	}
	if (eh->e_shoff == 0 || eh->e_shnum == 0) {
		fprintf(stderr, "w6l: %s: stripped .so unsupported\n", path);
		free(buf);
		return -1;
	}

	Shdr *sh = (Shdr *)(buf + eh->e_shoff);
	int idx_dynsym = -1, idx_dynamic = -1;
	int idx_versym = -1, idx_verdef = -1;
	for (u16 i = 0; i < eh->e_shnum; i++) {
		if (sh[i].sh_type == SHT_DYNSYM)      idx_dynsym  = i;
		if (sh[i].sh_type == SHT_DYNAMIC)     idx_dynamic = i;
		if (sh[i].sh_type == SHT_GNU_VERSYM)  idx_versym  = i;
		if (sh[i].sh_type == SHT_GNU_VERDEF)  idx_verdef  = i;
	}
	if (idx_dynsym < 0) {
		fprintf(stderr, "w6l: %s: no .dynsym\n", path);
		free(buf);
		return -1;
	}
	int idx_dynstr = sh[idx_dynsym].sh_link;
	const char *str = (const char *)(buf + sh[idx_dynstr].sh_offset);
	Sym64 *syms = (Sym64 *)(buf + sh[idx_dynsym].sh_offset);
	u64 nsyms = sh[idx_dynsym].sh_size / sizeof(Sym64);

	/* SONAME: the .dynamic section's strings live in the section pointed
	 * at by its sh_link (not necessarily .dynstr — though usually). */
	const char *soname = NULL;
	if (idx_dynamic >= 0) {
		int idx_dstr = sh[idx_dynamic].sh_link;
		const char *dstr = (const char *)(buf + sh[idx_dstr].sh_offset);
		Dyn64 *d = (Dyn64 *)(buf + sh[idx_dynamic].sh_offset);
		u64 nd = sh[idx_dynamic].sh_size / sizeof(Dyn64);
		for (u64 i = 0; i < nd; i++) {
			if (d[i].d_tag == DT_NULL) break;
			if (d[i].d_tag == DT_SONAME) {
				soname = dstr + d[i].d_val;
				break;
			}
		}
	}
	if (soname == NULL) {
		const char *bn = strrchr(path, '/');
		soname = bn ? bn + 1 : path;
	}

	/* Build verdef-index → version-name table. The version name is in
	 * the first Verdaux (the rest are predecessor names — version
	 * inheritance for stable ABI within a release line). For our
	 * purposes only the leading name matters. */
	const char **verdef_names = NULL;
	int verdef_max = 0;
	if (idx_verdef >= 0) {
		const u8 *vbase = buf + sh[idx_verdef].sh_offset;
		const char *vstr = (const char *)
		    (buf + sh[sh[idx_verdef].sh_link].sh_offset);
		/* First pass: discover max ndx so we can size the table. */
		u64 vd_off = 0;
		while (vd_off < sh[idx_verdef].sh_size) {
			Verdef *vd = (Verdef *)(vbase + vd_off);
			if ((int)vd->vd_ndx > verdef_max)
				verdef_max = vd->vd_ndx;
			if (vd->vd_next == 0) break;
			vd_off += vd->vd_next;
		}
		verdef_names = calloc((size_t)verdef_max + 1,
		    sizeof *verdef_names);
		vd_off = 0;
		while (vd_off < sh[idx_verdef].sh_size) {
			Verdef *vd = (Verdef *)(vbase + vd_off);
			Verdaux *va = (Verdaux *)((u8 *)vd + vd->vd_aux);
			verdef_names[vd->vd_ndx] = vstr + va->vda_name;
			if (vd->vd_next == 0) break;
			vd_off += vd->vd_next;
		}
	}

	/* Versym is one u16 per .dynsym entry. */
	const u16 *versym = NULL;
	if (idx_versym >= 0)
		versym = (const u16 *)(buf + sh[idx_versym].sh_offset);

	Lso *so = calloc(1, sizeof *so);
	so->path     = strdup(path);
	so->soname   = strdup(soname);
	so->exports  = calloc((size_t)nsyms + 1, sizeof *so->exports);
	so->versions = calloc((size_t)nsyms + 1, sizeof *so->versions);

	int n = 0;
	for (u64 i = 1; i < nsyms; i++) {
		if (syms[i].st_shndx == 0) continue;	/* SHN_UNDEF */
		u8 b = ST_BIND(syms[i].st_info);
		if (b != 1 && b != 2) continue;		/* GLOBAL or WEAK */
		const char *nm = str + syms[i].st_name;
		if (nm[0] == '\0') continue;

		/* Skip non-default versions: when a name has multiple version
		 * definitions, the loader binds an unversioned reference to
		 * the one whose Versym entry has the hidden bit clear. */
		const char *vername = NULL;
		if (versym != NULL) {
			u16 v = versym[i];
			if (v & VERSYM_HIDDEN) continue;	/* non-default */
			u16 vidx = v & VERSYM_VERSION;
			if (vidx == VER_NDX_LOCAL) continue;	/* not exported */
			if (vidx == VER_NDX_GLOBAL) {
				vername = NULL;			/* unversioned */
			} else if (verdef_names != NULL
			    && (int)vidx <= verdef_max
			    && verdef_names[vidx] != NULL) {
				/* index 1 in glibc's Verdef is the SONAME with
				 * VER_FLG_BASE — we skip its export entries
				 * as a side effect of vidx==1 mapping to the
				 * BASE name (e.g. "libc.so.6"), which never
				 * appears as a reference target. Treat any
				 * lookup that lands on the BASE entry as
				 * unversioned. */
				if (vidx == 1)
					vername = NULL;
				else
					vername = verdef_names[vidx];
			}
		}

		so->exports[n]  = strdup(nm);
		so->versions[n] = vername ? strdup(vername) : NULL;
		n++;
	}
	so->exports[n]  = NULL;
	so->versions[n] = NULL;

	if (verdef_names != NULL) free(verdef_names);

	so->next = l->sos;
	l->sos = so;
	free(buf);
	return 0;
}

int
l_so_provides(Lso *so, const char *name)
{
	const char *v;
	return l_so_provides_v(so, name, &v);
}

int
l_so_provides_v(Lso *so, const char *name, const char **out_version)
{
	if (so == NULL || so->exports == NULL) return 0;
	for (int i = 0; so->exports[i]; i++) {
		if (strcmp(so->exports[i], name) != 0) continue;
		*out_version = so->versions ? so->versions[i] : NULL;
		return 1;
	}
	return 0;
}