Source code

Revision control

Copy as Markdown

Other Tools

/* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#include <stdint.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/mman.h>
#include <elf.h>
/* The Android NDK headers define those */
#undef Elf_Ehdr
#undef Elf_Addr
#if defined(__LP64__)
# define Elf_Ehdr Elf64_Ehdr
# define Elf_Phdr Elf64_Phdr
# define Elf_Addr Elf64_Addr
# define Elf_Word Elf64_Word
# define Elf_Dyn Elf64_Dyn
#else
# define Elf_Phdr Elf32_Phdr
# define Elf_Ehdr Elf32_Ehdr
# define Elf_Addr Elf32_Addr
# define Elf_Word Elf32_Word
# define Elf_Dyn Elf32_Dyn
#endif
#ifdef RELRHACK
# include "relrhack.h"
# define mprotect_cb mprotect
# define sysconf_cb sysconf
#else
// On ARM, PC-relative function calls have a limit in how far they can jump,
// which might not be enough for e.g. libxul.so. The easy way out would be
// to use the long_call attribute, which forces the compiler to generate code
// that can call anywhere, but clang doesn't support the attribute yet
// (https://bugs.llvm.org/show_bug.cgi?id=40623), and while the command-line
// equivalent does exist, it's currently broken
// (https://bugs.llvm.org/show_bug.cgi?id=40624). So we create a manual
// trampoline, corresponding to the code GCC generates with long_call.
# ifdef __arm__
__attribute__((section(".text._init_trampoline"), naked)) int init_trampoline(
int argc, char** argv, char** env) {
__asm__ __volatile__(
// thumb doesn't allow to use r12/ip with ldr, and thus would require an
// additional push/pop to save/restore the modified register, which would
// also change the call into a blx. It's simpler to switch to arm.
".arm\n"
" ldr ip, .LADDR\n"
".LAFTER:\n"
" add ip, pc, ip\n"
" bx ip\n"
".LADDR:\n"
" .word real_original_init-(.LAFTER+8)\n");
}
# endif
// On aarch64, a similar problem exists, but long_call is not an option at all
// (even GCC doesn't support them on aarch64).
# ifdef __aarch64__
__attribute__((section(".text._init_trampoline"), naked)) int init_trampoline(
int argc, char** argv, char** env) {
__asm__ __volatile__(
" adrp x8, .LADDR\n"
" add x8, x8, :lo12:.LADDR\n" // adrp + add gives us the full address
// for .LADDR
" ldr x0, [x8]\n" // Load the address of real_original_init relative to
// .LADDR
" add x0, x8, x0\n" // Add the address of .LADDR
" br x0\n" // Branch to real_original_init
".LADDR:\n"
" .xword real_original_init-.LADDR\n");
}
# endif
extern __attribute__((visibility("hidden"))) void original_init(int argc,
char** argv,
char** env);
extern __attribute__((visibility("hidden"))) Elf_Addr relhack[];
extern __attribute__((visibility("hidden"))) Elf_Addr relhack_end[];
extern __attribute__((visibility("hidden"))) int (*mprotect_cb)(void* addr,
size_t len,
int prot);
extern __attribute__((visibility("hidden"))) long (*sysconf_cb)(int name);
extern __attribute__((visibility("hidden"))) char relro_start[];
extern __attribute__((visibility("hidden"))) char relro_end[];
#endif
extern __attribute__((visibility("hidden"))) Elf_Ehdr __ehdr_start;
static inline __attribute__((always_inline)) void do_relocations(
Elf_Addr* relhack, Elf_Addr* relhack_end) {
Elf_Addr* ptr;
for (Elf_Addr* entry = relhack; entry < relhack_end; entry++) {
if ((*entry & 1) == 0) {
ptr = (Elf_Addr*)((intptr_t)&__ehdr_start + *entry);
*ptr += (intptr_t)&__ehdr_start;
} else {
Elf_Addr bits = *entry;
Elf_Addr* end = ptr + 8 * sizeof(Elf_Addr) - 1;
do {
ptr++;
bits >>= 1;
if (bits & 1) {
*ptr += (intptr_t)&__ehdr_start;
}
} while (ptr < end);
}
}
}
#ifndef RELRHACK
__attribute__((section(".text._init_noinit"))) int init_noinit(int argc,
char** argv,
char** env) {
do_relocations(relhack, relhack_end);
return 0;
}
__attribute__((section(".text._init"))) int init(int argc, char** argv,
char** env) {
do_relocations(relhack, relhack_end);
original_init(argc, argv, env);
// Ensure there is no tail-call optimization, avoiding the use of the
// B.W instruction in Thumb for the call above.
return 0;
}
#endif
static inline __attribute__((always_inline)) void do_relocations_with_relro(
Elf_Addr* relhack, Elf_Addr* relhack_end, char* relro_start,
char* relro_end) {
long page_size = sysconf_cb(_SC_PAGESIZE);
uintptr_t aligned_relro_start = ((uintptr_t)relro_start) & ~(page_size - 1);
// The relro segment may not end at a page boundary. If that's the case, the
// remainder of the page needs to stay read-write, so the last page is never
// set read-only. Thus the aligned relro end is page-rounded down.
uintptr_t aligned_relro_end = ((uintptr_t)relro_end) & ~(page_size - 1);
// By the time the injected code runs, the relro segment is read-only. But
// we want to apply relocations in it, so we set it r/w first. We'll restore
// it to read-only in relro_post.
mprotect_cb((void*)aligned_relro_start,
aligned_relro_end - aligned_relro_start, PROT_READ | PROT_WRITE);
do_relocations(relhack, relhack_end);
mprotect_cb((void*)aligned_relro_start,
aligned_relro_end - aligned_relro_start, PROT_READ);
#ifndef RELRHACK
// mprotect_cb and sysconf_cb are allocated in .bss, so we need to restore
// them to a NULL value.
mprotect_cb = NULL;
sysconf_cb = NULL;
#endif
}
#ifndef RELRHACK
__attribute__((section(".text._init_noinit_relro"))) int init_noinit_relro(
int argc, char** argv, char** env) {
do_relocations_with_relro(relhack, relhack_end, relro_start, relro_end);
return 0;
}
__attribute__((section(".text._init_relro"))) int init_relro(int argc,
char** argv,
char** env) {
do_relocations_with_relro(relhack, relhack_end, relro_start, relro_end);
original_init(argc, argv, env);
return 0;
}
#else
extern __attribute__((visibility("hidden"))) Elf_Dyn _DYNAMIC[];
static void _relrhack_init(void) {
// Get the location of the SHT_RELR data from the PT_DYNAMIC segment.
uintptr_t elf_header = (uintptr_t)&__ehdr_start;
Elf_Addr* relhack = NULL;
Elf_Word size = 0;
for (Elf_Dyn* dyn = _DYNAMIC; dyn->d_tag != DT_NULL; dyn++) {
if ((dyn->d_tag & ~DT_RELRHACK_BIT) == DT_RELR) {
relhack = (Elf_Addr*)(elf_header + dyn->d_un.d_ptr);
} else if ((dyn->d_tag & ~DT_RELRHACK_BIT) == DT_RELRSZ) {
size = dyn->d_un.d_val;
}
}
Elf_Addr* relhack_end = (Elf_Addr*)((uintptr_t)relhack + size);
// Find the location of the PT_GNU_RELRO segment in the program headers.
Elf_Phdr* phdr = (Elf_Phdr*)(elf_header + __ehdr_start.e_phoff);
char* relro_start = NULL;
char* relro_end = NULL;
for (int i = 0; i < __ehdr_start.e_phnum; i++) {
if (phdr[i].p_type == PT_GNU_RELRO) {
relro_start = (char*)(elf_header + phdr[i].p_vaddr);
relro_end = (char*)(relro_start + phdr[i].p_memsz);
break;
}
}
if (relro_start != relro_end) {
do_relocations_with_relro(relhack, relhack_end, relro_start, relro_end);
} else {
do_relocations(relhack, relhack_end);
}
}
# ifdef ANDROID
// This creates a value that uses a relative relocation.
// In the case we've not set DT_RELRHACK_BIT on the RELR tags in the
// dynamic section, if the dynamic loader applied the relocation,
// the value will correctly point to the function address in memory.
__attribute__((visibility("hidden"))) void (*__relrhack_init)(void) =
_relrhack_init;
# endif
// The Android CRT doesn't contain an init function.
# ifndef ANDROID
extern __attribute__((visibility("hidden"))) void _init(int argc, char** argv,
char** env);
# endif
void _relrhack_wrap_init(int argc, char** argv, char** env) {
# ifdef ANDROID
// When the dynamic loader has applied the relocations itself, do nothing.
// (see comment above __relrhack_init)
if (__relrhack_init == _relrhack_init) {
return;
}
# endif
_relrhack_init();
# ifndef ANDROID
_init(argc, argv, env);
# endif
}
#endif