[linux-2.6-block.git] / Documentation / vDSO / parse_vdso.c

/*
 * parse_vdso.c: Linux reference vDSO parser
 * Written by Andrew Lutomirski, 2011-2014.
 *
 * This code is meant to be linked in to various programs that run on Linux.
 * As such, it is available with as few restrictions as possible.  This file
 * is licensed under the Creative Commons Zero License, version 1.0,
 * available at http://creativecommons.org/publicdomain/zero/1.0/legalcode
 *
 * The vDSO is a regular ELF DSO that the kernel maps into user space when
 * it starts a program.  It works equally well in statically and dynamically
 * linked binaries.
 *
 * This code is tested on x86.  In principle it should work on any
 * architecture that has a vDSO.
 */

#include <stdbool.h>
#include <stdint.h>
#include <string.h>
#include <limits.h>
#include <elf.h>

/*
 * To use this vDSO parser, first call one of the vdso_init_* functions.
 * If you've already parsed auxv, then pass the value of AT_SYSINFO_EHDR
 * to vdso_init_from_sysinfo_ehdr.  Otherwise pass auxv to vdso_init_from_auxv.
 * Then call vdso_sym for each symbol you want.  For example, to look up
 * gettimeofday on x86_64, use:
 *
 *     <some pointer> = vdso_sym("LINUX_2.6", "gettimeofday");
 * or
 *     <some pointer> = vdso_sym("LINUX_2.6", "__vdso_gettimeofday");
 *
 * vdso_sym will return 0 if the symbol doesn't exist or if the init function
 * failed or was not called.  vdso_sym is a little slow, so its return value
 * should be cached.
 *
 * vdso_sym is threadsafe; the init functions are not.
 *
 * These are the prototypes:
 */
extern void vdso_init_from_auxv(void *auxv);
extern void vdso_init_from_sysinfo_ehdr(uintptr_t base);
extern void *vdso_sym(const char *version, const char *name);


/* And here's the code. */
#ifndef ELF_BITS
# if ULONG_MAX > 0xffffffffUL
#  define ELF_BITS 64
# else
#  define ELF_BITS 32
# endif
#endif

#define ELF_BITS_XFORM2(bits, x) Elf##bits##_##x
#define ELF_BITS_XFORM(bits, x) ELF_BITS_XFORM2(bits, x)
#define ELF(x) ELF_BITS_XFORM(ELF_BITS, x)

static struct vdso_info
{
	bool valid;

	/* Load information */
	uintptr_t load_addr;
	uintptr_t load_offset;  /* load_addr - recorded vaddr */

	/* Symbol table */
	ELF(Sym) *symtab;
	const char *symstrings;
	ELF(Word) *bucket, *chain;
	ELF(Word) nbucket, nchain;

	/* Version table */
	ELF(Versym) *versym;
	ELF(Verdef) *verdef;
} vdso_info;

/* Straight from the ELF specification. */
static unsigned long elf_hash(const unsigned char *name)
{
	unsigned long h = 0, g;
	while (*name)
	{
		h = (h << 4) + *name++;
		if (g = h & 0xf0000000)
			h ^= g >> 24;
		h &= ~g;
	}
	return h;
}

void vdso_init_from_sysinfo_ehdr(uintptr_t base)
{
	size_t i;
	bool found_vaddr = false;

	vdso_info.valid = false;

	vdso_info.load_addr = base;

	ELF(Ehdr) *hdr = (ELF(Ehdr)*)base;
	if (hdr->e_ident[EI_CLASS] !=
	    (ELF_BITS == 32 ? ELFCLASS32 : ELFCLASS64)) {
		return;  /* Wrong ELF class -- check ELF_BITS */
	}

	ELF(Phdr) *pt = (ELF(Phdr)*)(vdso_info.load_addr + hdr->e_phoff);
	ELF(Dyn) *dyn = 0;

	/*
	 * We need two things from the segment table: the load offset
	 * and the dynamic table.
	 */
	for (i = 0; i < hdr->e_phnum; i++)
	{
		if (pt[i].p_type == PT_LOAD && !found_vaddr) {
			found_vaddr = true;
			vdso_info.load_offset =	base
				+ (uintptr_t)pt[i].p_offset
				- (uintptr_t)pt[i].p_vaddr;
		} else if (pt[i].p_type == PT_DYNAMIC) {
			dyn = (ELF(Dyn)*)(base + pt[i].p_offset);
		}
	}

	if (!found_vaddr || !dyn)
		return;  /* Failed */

	/*
	 * Fish out the useful bits of the dynamic table.
	 */
	ELF(Word) *hash = 0;
	vdso_info.symstrings = 0;
	vdso_info.symtab = 0;
	vdso_info.versym = 0;
	vdso_info.verdef = 0;
	for (i = 0; dyn[i].d_tag != DT_NULL; i++) {
		switch (dyn[i].d_tag) {
		case DT_STRTAB:
			vdso_info.symstrings = (const char *)
				((uintptr_t)dyn[i].d_un.d_ptr
				 + vdso_info.load_offset);
			break;
		case DT_SYMTAB:
			vdso_info.symtab = (ELF(Sym) *)
				((uintptr_t)dyn[i].d_un.d_ptr
				 + vdso_info.load_offset);
			break;
		case DT_HASH:
			hash = (ELF(Word) *)
				((uintptr_t)dyn[i].d_un.d_ptr
				 + vdso_info.load_offset);
			break;
		case DT_VERSYM:
			vdso_info.versym = (ELF(Versym) *)
				((uintptr_t)dyn[i].d_un.d_ptr
				 + vdso_info.load_offset);
			break;
		case DT_VERDEF:
			vdso_info.verdef = (ELF(Verdef) *)
				((uintptr_t)dyn[i].d_un.d_ptr
				 + vdso_info.load_offset);
			break;
		}
	}
	if (!vdso_info.symstrings || !vdso_info.symtab || !hash)
		return;  /* Failed */

	if (!vdso_info.verdef)
		vdso_info.versym = 0;

	/* Parse the hash table header. */
	vdso_info.nbucket = hash[0];
	vdso_info.nchain = hash[1];
	vdso_info.bucket = &hash[2];
	vdso_info.chain = &hash[vdso_info.nbucket + 2];

	/* That's all we need. */
	vdso_info.valid = true;
}

static bool vdso_match_version(ELF(Versym) ver,
			       const char *name, ELF(Word) hash)
{
	/*
	 * This is a helper function to check if the version indexed by
	 * ver matches name (which hashes to hash).
	 *
	 * The version definition table is a mess, and I don't know how
	 * to do this in better than linear time without allocating memory
	 * to build an index.  I also don't know why the table has
	 * variable size entries in the first place.
	 *
	 * For added fun, I can't find a comprehensible specification of how
	 * to parse all the weird flags in the table.
	 *
	 * So I just parse the whole table every time.
	 */

	/* First step: find the version definition */
	ver &= 0x7fff;  /* Apparently bit 15 means "hidden" */
	ELF(Verdef) *def = vdso_info.verdef;
	while(true) {
		if ((def->vd_flags & VER_FLG_BASE) == 0
		    && (def->vd_ndx & 0x7fff) == ver)
			break;

		if (def->vd_next == 0)
			return false;  /* No definition. */

		def = (ELF(Verdef) *)((char *)def + def->vd_next);
	}

	/* Now figure out whether it matches. */
	ELF(Verdaux) *aux = (ELF(Verdaux)*)((char *)def + def->vd_aux);
	return def->vd_hash == hash
		&& !strcmp(name, vdso_info.symstrings + aux->vda_name);
}

void *vdso_sym(const char *version, const char *name)
{
	unsigned long ver_hash;
	if (!vdso_info.valid)
		return 0;

	ver_hash = elf_hash(version);
	ELF(Word) chain = vdso_info.bucket[elf_hash(name) % vdso_info.nbucket];

	for (; chain != STN_UNDEF; chain = vdso_info.chain[chain]) {
		ELF(Sym) *sym = &vdso_info.symtab[chain];

		/* Check for a defined global or weak function w/ right name. */
		if (ELF64_ST_TYPE(sym->st_info) != STT_FUNC)
			continue;
		if (ELF64_ST_BIND(sym->st_info) != STB_GLOBAL &&
		    ELF64_ST_BIND(sym->st_info) != STB_WEAK)
			continue;
		if (sym->st_shndx == SHN_UNDEF)
			continue;
		if (strcmp(name, vdso_info.symstrings + sym->st_name))
			continue;

		/* Check symbol version. */
		if (vdso_info.versym
		    && !vdso_match_version(vdso_info.versym[chain],
					   version, ver_hash))
			continue;

		return (void *)(vdso_info.load_offset + sym->st_value);
	}

	return 0;
}

void vdso_init_from_auxv(void *auxv)
{
	ELF(auxv_t) *elf_auxv = auxv;
	for (int i = 0; elf_auxv[i].a_type != AT_NULL; i++)
	{
		if (elf_auxv[i].a_type == AT_SYSINFO_EHDR) {
			vdso_init_from_sysinfo_ehdr(elf_auxv[i].a_un.a_val);
			return;
		}
	}

	vdso_info.valid = false;
}
Commit	Line	Data
98eedc3a AL	1	/*
98eedc3a AL	2	* parse_vdso.c: Linux reference vDSO parser
4ebbefd6	3	* Written by Andrew Lutomirski, 2011-2014.
98eedc3a AL	4	*
	5	* This code is meant to be linked in to various programs that run on Linux.
	6	* As such, it is available with as few restrictions as possible. This file
	7	* is licensed under the Creative Commons Zero License, version 1.0,
	8	* available at http://creativecommons.org/publicdomain/zero/1.0/legalcode
	9	*
	10	* The vDSO is a regular ELF DSO that the kernel maps into user space when
	11	* it starts a program. It works equally well in statically and dynamically
	12	* linked binaries.
	13	*
4ebbefd6	14	* This code is tested on x86. In principle it should work on any
98eedc3a AL	15	* architecture that has a vDSO.
	16	*/
	17
	18	#include <stdbool.h>
	19	#include <stdint.h>
	20	#include <string.h>
4ebbefd6	21	#include <limits.h>
98eedc3a AL	22	#include <elf.h>
	23
	24	/*
	25	* To use this vDSO parser, first call one of the vdso_init_* functions.
	26	* If you've already parsed auxv, then pass the value of AT_SYSINFO_EHDR
	27	* to vdso_init_from_sysinfo_ehdr. Otherwise pass auxv to vdso_init_from_auxv.
	28	* Then call vdso_sym for each symbol you want. For example, to look up
	29	* gettimeofday on x86_64, use:
	30	*
	31	* <some pointer> = vdso_sym("LINUX_2.6", "gettimeofday");
	32	* or
	33	* <some pointer> = vdso_sym("LINUX_2.6", "__vdso_gettimeofday");
	34	*
	35	* vdso_sym will return 0 if the symbol doesn't exist or if the init function
	36	* failed or was not called. vdso_sym is a little slow, so its return value
	37	* should be cached.
	38	*
	39	* vdso_sym is threadsafe; the init functions are not.
	40	*
	41	* These are the prototypes:
	42	*/
	43	extern void vdso_init_from_auxv(void *auxv);
	44	extern void vdso_init_from_sysinfo_ehdr(uintptr_t base);
	45	extern void vdso_sym(const char version, const char *name);
	46
	47
	48	/* And here's the code. */
4ebbefd6 AL	49	#ifndef ELF_BITS
	50	# if ULONG_MAX > 0xffffffffUL
	51	# define ELF_BITS 64
	52	# else
	53	# define ELF_BITS 32
	54	# endif
98eedc3a AL	55	#endif
98eedc3a AL	56
4ebbefd6 AL	57	#define ELF_BITS_XFORM2(bits, x) Elf##bits##_##x
	58	#define ELF_BITS_XFORM(bits, x) ELF_BITS_XFORM2(bits, x)
	59	#define ELF(x) ELF_BITS_XFORM(ELF_BITS, x)
	60
98eedc3a AL	61	static struct vdso_info
	62	{
	63	bool valid;
	64
	65	/* Load information */
	66	uintptr_t load_addr;
	67	uintptr_t load_offset; /* load_addr - recorded vaddr */
	68
	69	/* Symbol table */
4ebbefd6	70	ELF(Sym) *symtab;
98eedc3a	71	const char *symstrings;
4ebbefd6 AL	72	ELF(Word) bucket, chain;
4ebbefd6 AL	73	ELF(Word) nbucket, nchain;
98eedc3a AL	74
98eedc3a AL	75	/* Version table */
4ebbefd6 AL	76	ELF(Versym) *versym;
4ebbefd6 AL	77	ELF(Verdef) *verdef;
98eedc3a AL	78	} vdso_info;
	79
	80	/* Straight from the ELF specification. */
	81	static unsigned long elf_hash(const unsigned char *name)
	82	{
	83	unsigned long h = 0, g;
	84	while (*name)
	85	{
	86	h = (h << 4) + *name++;
	87	if (g = h & 0xf0000000)
	88	h ^= g >> 24;
	89	h &= ~g;
	90	}
	91	return h;
	92	}
	93
	94	void vdso_init_from_sysinfo_ehdr(uintptr_t base)
	95	{
	96	size_t i;
	97	bool found_vaddr = false;
	98
	99	vdso_info.valid = false;
	100
	101	vdso_info.load_addr = base;
	102
4ebbefd6 AL	103	ELF(Ehdr) hdr = (ELF(Ehdr))base;
	104	if (hdr->e_ident[EI_CLASS] !=
	105	(ELF_BITS == 32 ? ELFCLASS32 : ELFCLASS64)) {
	106	return; /* Wrong ELF class -- check ELF_BITS */
	107	}
	108
	109	ELF(Phdr) pt = (ELF(Phdr))(vdso_info.load_addr + hdr->e_phoff);
	110	ELF(Dyn) *dyn = 0;
98eedc3a AL	111
	112	/*
	113	* We need two things from the segment table: the load offset
	114	* and the dynamic table.
	115	*/
	116	for (i = 0; i < hdr->e_phnum; i++)
	117	{
	118	if (pt[i].p_type == PT_LOAD && !found_vaddr) {
	119	found_vaddr = true;
	120	vdso_info.load_offset = base
	121	+ (uintptr_t)pt[i].p_offset
	122	- (uintptr_t)pt[i].p_vaddr;
	123	} else if (pt[i].p_type == PT_DYNAMIC) {
4ebbefd6	124	dyn = (ELF(Dyn)*)(base + pt[i].p_offset);
98eedc3a AL	125	}
	126	}
	127
	128	if (!found_vaddr \|\| !dyn)
	129	return; /* Failed */
	130
	131	/*
	132	* Fish out the useful bits of the dynamic table.
	133	*/
4ebbefd6	134	ELF(Word) *hash = 0;
98eedc3a AL	135	vdso_info.symstrings = 0;
	136	vdso_info.symtab = 0;
	137	vdso_info.versym = 0;
	138	vdso_info.verdef = 0;
	139	for (i = 0; dyn[i].d_tag != DT_NULL; i++) {
	140	switch (dyn[i].d_tag) {
	141	case DT_STRTAB:
	142	vdso_info.symstrings = (const char *)
	143	((uintptr_t)dyn[i].d_un.d_ptr
	144	+ vdso_info.load_offset);
	145	break;
	146	case DT_SYMTAB:
4ebbefd6	147	vdso_info.symtab = (ELF(Sym) *)
98eedc3a AL	148	((uintptr_t)dyn[i].d_un.d_ptr
	149	+ vdso_info.load_offset);
	150	break;
	151	case DT_HASH:
4ebbefd6	152	hash = (ELF(Word) *)
98eedc3a AL	153	((uintptr_t)dyn[i].d_un.d_ptr
	154	+ vdso_info.load_offset);
	155	break;
	156	case DT_VERSYM:
4ebbefd6	157	vdso_info.versym = (ELF(Versym) *)
98eedc3a AL	158	((uintptr_t)dyn[i].d_un.d_ptr
	159	+ vdso_info.load_offset);
	160	break;
	161	case DT_VERDEF:
4ebbefd6	162	vdso_info.verdef = (ELF(Verdef) *)
98eedc3a AL	163	((uintptr_t)dyn[i].d_un.d_ptr
	164	+ vdso_info.load_offset);
	165	break;
	166	}
	167	}
	168	if (!vdso_info.symstrings \|\| !vdso_info.symtab \|\| !hash)
	169	return; /* Failed */
	170
	171	if (!vdso_info.verdef)
	172	vdso_info.versym = 0;
	173
	174	/* Parse the hash table header. */
	175	vdso_info.nbucket = hash[0];
	176	vdso_info.nchain = hash[1];
	177	vdso_info.bucket = &hash[2];
	178	vdso_info.chain = &hash[vdso_info.nbucket + 2];
	179
	180	/* That's all we need. */
	181	vdso_info.valid = true;
	182	}
	183
4ebbefd6 AL	184	static bool vdso_match_version(ELF(Versym) ver,
4ebbefd6 AL	185	const char *name, ELF(Word) hash)
98eedc3a AL	186	{
	187	/*
	188	* This is a helper function to check if the version indexed by
	189	* ver matches name (which hashes to hash).
	190	*
	191	* The version definition table is a mess, and I don't know how
	192	* to do this in better than linear time without allocating memory
	193	* to build an index. I also don't know why the table has
	194	* variable size entries in the first place.
	195	*
	196	* For added fun, I can't find a comprehensible specification of how
	197	* to parse all the weird flags in the table.
	198	*
	199	* So I just parse the whole table every time.
	200	*/
	201
	202	/* First step: find the version definition */
	203	ver &= 0x7fff; /* Apparently bit 15 means "hidden" */
4ebbefd6	204	ELF(Verdef) *def = vdso_info.verdef;
98eedc3a AL	205	while(true) {
	206	if ((def->vd_flags & VER_FLG_BASE) == 0
	207	&& (def->vd_ndx & 0x7fff) == ver)
	208	break;
	209
	210	if (def->vd_next == 0)
	211	return false; /* No definition. */
	212
4ebbefd6	213	def = (ELF(Verdef) )((char )def + def->vd_next);
98eedc3a AL	214	}
	215
	216	/* Now figure out whether it matches. */
4ebbefd6	217	ELF(Verdaux) aux = (ELF(Verdaux))((char *)def + def->vd_aux);
98eedc3a AL	218	return def->vd_hash == hash
	219	&& !strcmp(name, vdso_info.symstrings + aux->vda_name);
	220	}
	221
	222	void vdso_sym(const char version, const char *name)
	223	{
	224	unsigned long ver_hash;
	225	if (!vdso_info.valid)
	226	return 0;
	227
	228	ver_hash = elf_hash(version);
4ebbefd6	229	ELF(Word) chain = vdso_info.bucket[elf_hash(name) % vdso_info.nbucket];
98eedc3a AL	230
98eedc3a AL	231	for (; chain != STN_UNDEF; chain = vdso_info.chain[chain]) {
4ebbefd6	232	ELF(Sym) *sym = &vdso_info.symtab[chain];
98eedc3a AL	233
	234	/* Check for a defined global or weak function w/ right name. */
	235	if (ELF64_ST_TYPE(sym->st_info) != STT_FUNC)
	236	continue;
	237	if (ELF64_ST_BIND(sym->st_info) != STB_GLOBAL &&
	238	ELF64_ST_BIND(sym->st_info) != STB_WEAK)
	239	continue;
	240	if (sym->st_shndx == SHN_UNDEF)
	241	continue;
	242	if (strcmp(name, vdso_info.symstrings + sym->st_name))
	243	continue;
	244
	245	/* Check symbol version. */
	246	if (vdso_info.versym
	247	&& !vdso_match_version(vdso_info.versym[chain],
	248	version, ver_hash))
	249	continue;
	250
	251	return (void *)(vdso_info.load_offset + sym->st_value);
	252	}
	253
	254	return 0;
	255	}
	256
	257	void vdso_init_from_auxv(void *auxv)
	258	{
4ebbefd6	259	ELF(auxv_t) *elf_auxv = auxv;
98eedc3a AL	260	for (int i = 0; elf_auxv[i].a_type != AT_NULL; i++)
	261	{
	262	if (elf_auxv[i].a_type == AT_SYSINFO_EHDR) {
	263	vdso_init_from_sysinfo_ehdr(elf_auxv[i].a_un.a_val);
	264	return;
	265	}
	266	}
	267
	268	vdso_info.valid = false;
	269	}