path: root/elf.c

/*
 * revdep - reverse dependency checker for rawnix
 * elf.c - ELF parsing, hash-table cache, library resolution
 *
 * Copyright (c) 2025 zorz@gmx.com
 * Licensed under the BSD 2-Clause License. See revdep.h.
 *
 * x86_64-only.  musl-only.  No ld.so.conf (musl uses DT_RUNPATH).
 */

#define _POSIX_C_SOURCE 200809L
#define _DEFAULT_SOURCE

#include "revdep.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <fcntl.h>
#include <unistd.h>
#include <libgen.h>
#include <limits.h>
#include <gelf.h>

/* -------------------------------------------------------------------- */
/*  ELF inspection                                                      */
/* -------------------------------------------------------------------- */

static bool
elf_is_valid(Elf *e)
{
	GElf_Ehdr ehdr;

	if (!e || elf_kind(e) != ELF_K_ELF)
		return false;

	if (!gelf_getehdr(e, &ehdr))
		return false;

	/* x86_64 only */
	if (ehdr.e_machine != EM_X86_64)
		return false;

	/* executables and shared objects */
	if (ehdr.e_type != ET_EXEC && ehdr.e_type != ET_DYN)
		return false;

	/* Linux or unspecified */
	if (ehdr.e_ident[EI_OSABI] != ELFOSABI_NONE &&
	    ehdr.e_ident[EI_OSABI] != ELFOSABI_LINUX)
		return false;

	return true;
}

static bool
elf_read_dynamic(Elf *e, elfinfo_t *info)
{
	size_t phnum, i, j, count;
	GElf_Phdr phdr;
	Elf_Scn *scn;
	GElf_Shdr shdr;
	Elf_Data *data;
	GElf_Dyn dyn;
	const char *str;

	if (elf_getphdrnum(e, &phnum) == -1)
		return false;

	for (i = 0; i < phnum; i++) {
		if (!gelf_getphdr(e, (int)i, &phdr))
			continue;
		if (phdr.p_type != PT_DYNAMIC)
			continue;

		scn = gelf_offscn(e, phdr.p_offset);
		if (!scn)
			continue;
		if (!gelf_getshdr(scn, &shdr))
			continue;
		if (shdr.sh_type != SHT_DYNAMIC)
			continue;

		data = elf_getdata(scn, NULL);
		if (!data)
			return false;

		count = shdr.sh_size /
		        gelf_fsize(e, ELF_T_DYN, 1, EV_CURRENT);

		for (j = 0; j < count; j++) {
			if (!gelf_getdyn(data, (int)j, &dyn))
				break;

			str = elf_strptr(e, shdr.sh_link, dyn.d_un.d_val);
			if (!str)
				continue;

			switch (dyn.d_tag) {
			case DT_NEEDED:
				strvec_push(&info->needed, str);
				break;
			case DT_RPATH:
				strsplit(str, ':', &info->rpath);
				break;
			case DT_RUNPATH:
				strsplit(str, ':', &info->runpath);
				break;
			default:
				break;
			}
		}
		return true;
	}

	return false;
}

elfinfo_t *
elf_inspect(const char *path)
{
	elfinfo_t *info;
	int fd;
	Elf *e;

	if (!path)
		return NULL;

	info = calloc(1, sizeof(*info));
	if (!info)
		return NULL;

	info->path = xstrdup(path);
	strvec_init(&info->needed);
	strvec_init(&info->rpath);
	strvec_init(&info->runpath);
	info->valid = false;

	fd = open(path, O_RDONLY);
	if (fd == -1) {
		elf_free(info);
		return NULL;
	}

	e = elf_begin(fd, ELF_C_READ_MMAP, NULL);
	if (!e) {
		close(fd);
		elf_free(info);
		return NULL;
	}

	if (!elf_is_valid(e) || !elf_read_dynamic(e, info)) {
		elf_end(e);
		close(fd);
		elf_free(info);
		return NULL;
	}

	info->valid = true;
	elf_end(e);
	close(fd);
	return info;
}

void
elf_free(elfinfo_t *e)
{
	if (!e)
		return;
	free(e->path);
	strvec_free(&e->needed);
	strvec_free(&e->rpath);
	strvec_free(&e->runpath);
	free(e);
}

/* -------------------------------------------------------------------- */
/*  Hash-table ELF cache                                                */
/*                                                                      */
/*  Many packages share the same libraries (libz, libc++, etc).         */
/*  Without caching, we'd re-open and re-parse the same .so hundreds    */
/*  of times.  The hash table gives O(1) lookup by path.                */
/* -------------------------------------------------------------------- */

#define CACHE_BUCKETS 2048

typedef struct cache_node {
	char              *key;
	elfinfo_t         *val;
	struct cache_node *next;
} cache_node_t;

static cache_node_t *cache_table[CACHE_BUCKETS];

static unsigned long
hash_djb2(const char *s)
{
	unsigned long h = 5381;
	int c;

	while ((c = *s++))
		h = ((h << 5) + h) + (unsigned long)c;
	return h % CACHE_BUCKETS;
}

elfinfo_t *
cache_lookup(const char *path)
{
	unsigned long h;
	cache_node_t *n;
	elfinfo_t *info;

	if (!path)
		return NULL;

	h = hash_djb2(path);

	for (n = cache_table[h]; n; n = n->next) {
		if (strcmp(n->key, path) == 0)
			return n->val;
	}

	/* not cached — inspect and store */
	info = elf_inspect(path);
	if (!info || !info->valid) {
		elf_free(info);
		return NULL;
	}

	n = malloc(sizeof(*n));
	if (!n) {
		elf_free(info);
		return NULL;
	}

	n->key  = xstrdup(path);
	n->val  = info;
	n->next = cache_table[h];
	cache_table[h] = n;

	return info;
}

void
cache_clear(void)
{
	size_t i;
	cache_node_t *n, *next;

	for (i = 0; i < CACHE_BUCKETS; i++) {
		for (n = cache_table[i]; n; n = next) {
			next = n->next;
			free(n->key);
			elf_free(n->val);
			free(n);
		}
		cache_table[i] = NULL;
	}
}

/* -------------------------------------------------------------------- */
/*  $ORIGIN expansion                                                   */
/*                                                                      */
/*  musl only needs $ORIGIN (and $LIB = "lib").                          */
/*  No $PLATFORM, no multi-lib.                                         */
/* -------------------------------------------------------------------- */

static char *
expand_origin(const elfinfo_t *elf, const char *dir)
{
	char origin[PATH_MAX];
	char result[PATH_MAX];
	const char *src;
	char *dst, *end, *d;
	size_t len, dlen;

	if (!elf || !dir)
		return NULL;

	/* get directory of the ELF file */
	len = strlen(elf->path);
	if (len >= PATH_MAX)
		return NULL;
	memcpy(origin, elf->path, len + 1);
	d = dirname(origin);
	dlen = strlen(d);

	src = dir;
	dst = result;
	end = result + PATH_MAX - 1;

	while (*src && dst < end) {
		if (*src == '$') {
			if (strncmp(src, "$ORIGIN", 7) == 0 ||
			    strncmp(src, "${ORIGIN}", 9) == 0) {
				size_t skip = (src[1] == '{') ? 9 : 7;
				if (dst + dlen >= end)
					return NULL;
				memcpy(dst, d, dlen);
				dst += dlen;
				src += skip;
				continue;
			}
			if (strncmp(src, "$LIB", 4) == 0 ||
			    strncmp(src, "${LIB}", 6) == 0) {
				size_t skip = (src[1] == '{') ? 6 : 4;
				if (dst + 3 >= end)
					return NULL;
				memcpy(dst, "lib", 3);
				dst += 3;
				src += skip;
				continue;
			}
		}
		*dst++ = *src++;
	}
	*dst = '\0';

	return xstrdup(result);
}

/* -------------------------------------------------------------------- */
/*  Library resolution — follows ld.so search order                     */
/* -------------------------------------------------------------------- */

static bool
try_lib_in_dirs(const elfinfo_t *elf, const char *lib,
                const strvec_t *dirs, bool do_expand)
{
	size_t i;
	char *dir, *full, *real;
	elfinfo_t *libelf;

	if (!elf || !lib || !dirs)
		return false;

	for (i = 0; i < dirs->count; i++) {
		if (do_expand) {
			dir = expand_origin(elf, dirs->items[i]);
			if (!dir)
				continue;
		} else {
			dir = dirs->items[i];
		}

		full = xpathjoin(dir, lib);
		if (do_expand)
			free(dir);
		if (!full)
			continue;

		real = realpath_safe(full);
		free(full);
		if (!real)
			continue;

		libelf = cache_lookup(real);
		free(real);

		if (libelf)
			return true;
	}

	return false;
}

/*
 * Search order (simplified for musl — no ld.so.conf, no ldconfig cache):
 *
 *  1. DT_RPATH   (only if DT_RUNPATH absent)
 *  2. DT_RUNPATH
 *  3. System dirs (/lib, /usr/lib)
 *  4. Package-specific dirs (/etc/revdep.d/<pkg>)
 */
bool
lib_find(const elfinfo_t *elf, const pkg_t *pkg,
         const char *lib, const strvec_t *sys_dirs)
{
	if (!elf || !lib || !sys_dirs)
		return false;

	/* absolute or relative path in DT_NEEDED — resolve directly */
	if (strchr(lib, '/')) {
		char *real = realpath_safe(lib);
		if (!real)
			return false;
		elfinfo_t *e = cache_lookup(real);
		free(real);
		return e != NULL;
	}

	/* 1. DT_RPATH (deprecated, but still used by some software) */
	if (elf->rpath.count > 0 && elf->runpath.count == 0) {
		if (try_lib_in_dirs(elf, lib, &elf->rpath, true))
			return true;
	}

	/* 2. DT_RUNPATH */
	if (elf->runpath.count > 0) {
		if (try_lib_in_dirs(elf, lib, &elf->runpath, true))
			return true;
	}

	/* 3. system directories */
	if (try_lib_in_dirs(elf, lib, sys_dirs, false))
		return true;

	/* 4. per-package directories from /etc/revdep.d */
	if (pkg && pkg->dirs.count > 0) {
		if (try_lib_in_dirs(elf, lib, &pkg->dirs, false))
			return true;
	}

	return false;
}