added gcc_plugin

This commit is contained in:
van Hauser 2019-07-16 08:34:17 +02:00
parent 3252523823
commit 8a4cdd56d4
10 changed files with 1137 additions and 51 deletions

View File

@ -181,7 +181,7 @@ all_done: test_build
.NOTPARALLEL: clean
clean:
rm -f $(PROGS) afl-as as afl-g++ afl-clang afl-clang++ *.o *~ a.out core core.[1-9][0-9]* *.stackdump test .test .test1 .test2 test-instr .test-instr0 .test-instr1 qemu_mode/qemu-3.1.0.tar.xz afl-qemu-trace
rm -f $(PROGS) afl-as as afl-g++ afl-clang afl-clang++ *.o *~ a.out core core.[1-9][0-9]* *.stackdump test .test .test1 .test2 test-instr .test-instr0 .test-instr1 qemu_mode/qemu-3.1.0.tar.xz afl-qemu-trace afl-gcc-fast afl-gcc-pass.so afl-gcc-rt.o afl-g++-fast
rm -rf out_dir qemu_mode/qemu-3.1.0
$(MAKE) -C llvm_mode clean
$(MAKE) -C libdislocator clean
@ -193,6 +193,7 @@ install: all
install -m 755 $(PROGS) $(SH_PROGS) $${DESTDIR}$(BIN_PATH)
rm -f $${DESTDIR}$(BIN_PATH)/afl-as
if [ -f afl-qemu-trace ]; then install -m 755 afl-qemu-trace $${DESTDIR}$(BIN_PATH); fi
if [ -f afl-gcc-fast ]; then set e; install -m 755 afl-gcc-fast $${DESTDIR}$(BIN_PATH); ln -sf afl-gcc-fast $${DESTDIR}$(BIN_PATH)/afl-g++-fast; install -m 755 afl-gcc-pass.so afl-gcc-rt.o $${DESTDIR}$(HELPER_PATH); fi
ifndef AFL_TRACE_PC
if [ -f afl-clang-fast -a -f libLLVMInsTrim.so -a -f afl-llvm-rt.o ]; then set -e; install -m 755 afl-clang-fast $${DESTDIR}$(BIN_PATH); ln -sf afl-clang-fast $${DESTDIR}$(BIN_PATH)/afl-clang-fast++; install -m 755 libLLVMInsTrim.so afl-llvm-pass.so afl-llvm-rt.o $${DESTDIR}$(HELPER_PATH); fi
else
@ -204,7 +205,9 @@ endif
if [ -f split-compares-pass.so ]; then set -e; install -m 755 split-compares-pass.so $${DESTDIR}$(HELPER_PATH); fi
if [ -f split-switches-pass.so ]; then set -e; install -m 755 split-switches-pass.so $${DESTDIR}$(HELPER_PATH); fi
set -e; for i in afl-g++ afl-clang afl-clang++; do ln -sf afl-gcc $${DESTDIR}$(BIN_PATH)/$$i; done
set -e; ln -sf afl-gcc $${DESTDIR}$(BIN_PATH)/afl-g++
set -e; if [ -f afl-clang-fast ] ; then ln -sf afl-clang-fast $${DESTDIR}$(BIN_PATH)/afl-clang ; ln -sf afl-clang-fast $${DESTDIR}$(BIN_PATH)/afl-clang++ ; else ln -sf afl-gcc $${DESTDIR}$(BIN_PATH)/afl-clang ; ln -sf afl-gcc $${DESTDIR}$(BIN_PATH)/afl-clang++; fi
install -m 755 afl-as $${DESTDIR}$(HELPER_PATH)
ln -sf afl-as $${DESTDIR}$(HELPER_PATH)/as
install -m 644 docs/README docs/ChangeLog docs/*.txt $${DESTDIR}$(DOC_PATH)

View File

@ -5234,6 +5234,7 @@ static u32 calculate_score(struct queue_entry* q) {
// MOpt mode
if (limit_time_sig != 0 && max_depth - q->depth < 3) perf_score *= 2;
else if (perf_score < 1) perf_score = 1; // Add a lower bound to AFLFast's energy assignment strategies
/* Make sure that we don't go over limit. */
@ -10224,8 +10225,7 @@ static u8 core_fuzzing(char** argv) {
stage_max = (doing_det ? HAVOC_CYCLES_INIT : HAVOC_CYCLES) *
perf_score / havoc_div / 100;
}
else {
} else {
static u8 tmp[32];
@ -10243,20 +10243,13 @@ static u8 core_fuzzing(char** argv) {
//for (; swarm_now < swarm_num; swarm_now++)
{
if (key_puppet == 1)
{
if (unlikely(orig_hit_cnt_puppet == 0))
{
if (key_puppet == 1) {
if (unlikely(orig_hit_cnt_puppet == 0)) {
orig_hit_cnt_puppet = queued_paths + unique_crashes;
last_limit_time_start = get_cur_time();
SPLICE_CYCLES_puppet = (UR(SPLICE_CYCLES_puppet_up - SPLICE_CYCLES_puppet_low + 1) + SPLICE_CYCLES_puppet_low);
}
}
{
havoc_stage_puppet:
@ -10266,14 +10259,11 @@ static u8 core_fuzzing(char** argv) {
splice_cycle variable is set, generate different descriptions and such. */
if (!splice_cycle) {
stage_name = "MOpt core avoc";
stage_short = "MOpt core havoc";
stage_max = (doing_det ? HAVOC_CYCLES_INIT : HAVOC_CYCLES) *
perf_score / havoc_div / 100;
}
else {
} else {
static u8 tmp[32];
perf_score = orig_perf;
sprintf(tmp, "MOpt core splice %u", splice_cycle);
@ -10282,31 +10272,20 @@ static u8 core_fuzzing(char** argv) {
stage_max = SPLICE_HAVOC * perf_score / havoc_div / 100;
}
if (stage_max < HAVOC_MIN) stage_max = HAVOC_MIN;
temp_len = len;
orig_hit_cnt = queued_paths + unique_crashes;
havoc_queued = queued_paths;
for (stage_cur = 0; stage_cur < stage_max; stage_cur++) {
u32 use_stacking = 1 << (1 + UR(HAVOC_STACK_POW2));
stage_cur_val = use_stacking;
for (i = 0; i < operator_num; i++)
{
for (i = 0; i < operator_num; i++) {
core_operator_cycles_puppet_v3[i] = core_operator_cycles_puppet_v2[i];
}
for (i = 0; i < use_stacking; i++) {
switch (select_algorithm()) {
@ -10366,8 +10345,7 @@ static u8 core_fuzzing(char** argv) {
if (UR(2)) {
u32 pos = UR(temp_len - 1);
*(u16*)(out_buf + pos) -= 1 + UR(ARITH_MAX);
}
else {
} else {
u32 pos = UR(temp_len - 1);
u16 num = 1 + UR(ARITH_MAX);
*(u16*)(out_buf + pos) =
@ -10377,8 +10355,7 @@ static u8 core_fuzzing(char** argv) {
if (UR(2)) {
u32 pos = UR(temp_len - 1);
*(u16*)(out_buf + pos) += 1 + UR(ARITH_MAX);
}
else {
} else {
u32 pos = UR(temp_len - 1);
u16 num = 1 + UR(ARITH_MAX);
*(u16*)(out_buf + pos) =
@ -10394,8 +10371,7 @@ static u8 core_fuzzing(char** argv) {
if (UR(2)) {
u32 pos = UR(temp_len - 3);
*(u32*)(out_buf + pos) -= 1 + UR(ARITH_MAX);
}
else {
} else {
u32 pos = UR(temp_len - 3);
u32 num = 1 + UR(ARITH_MAX);
*(u32*)(out_buf + pos) =
@ -10405,8 +10381,7 @@ static u8 core_fuzzing(char** argv) {
if (UR(2)) {
u32 pos = UR(temp_len - 3);
*(u32*)(out_buf + pos) += 1 + UR(ARITH_MAX);
}
else {
} else {
u32 pos = UR(temp_len - 3);
u32 num = 1 + UR(ARITH_MAX);
*(u32*)(out_buf + pos) =
@ -10429,8 +10404,7 @@ static u8 core_fuzzing(char** argv) {
if (UR(2)) {
*(u16*)(out_buf + UR(temp_len - 1)) =
interesting_16[UR(sizeof(interesting_16) >> 1)];
}
else {
} else {
*(u16*)(out_buf + UR(temp_len - 1)) = SWAP16(
interesting_16[UR(sizeof(interesting_16) >> 1)]);
}
@ -10446,8 +10420,7 @@ static u8 core_fuzzing(char** argv) {
if (UR(2)) {
*(u32*)(out_buf + UR(temp_len - 3)) =
interesting_32[UR(sizeof(interesting_32) >> 2)];
}
else {
} else {
*(u32*)(out_buf + UR(temp_len - 3)) = SWAP32(
interesting_32[UR(sizeof(interesting_32) >> 2)]);
}
@ -10466,7 +10439,6 @@ static u8 core_fuzzing(char** argv) {
break;
case 13: {
/* Delete bytes. We're making this a bit more likely
@ -10507,8 +10479,7 @@ static u8 core_fuzzing(char** argv) {
clone_len = choose_block_len(temp_len);
clone_from = UR(temp_len - clone_len + 1);
}
else {
} else {
clone_len = choose_block_len(HAVOC_BLK_XL);
clone_from = 0;
@ -10575,17 +10546,10 @@ static u8 core_fuzzing(char** argv) {
}
tmp_core_time += 1;
u64 temp_total_found = queued_paths + unique_crashes;
if (common_fuzz_stuff(argv, out_buf, temp_len))
goto abandon_entry_puppet;

View File

@ -17,6 +17,10 @@ sending a mail to <afl-users+subscribe@googlegroups.com>.
Version ++2.52d (tbd):
-----------------------------
- if llvm_mode was compiled, afl-clang/afl-clang++ will point to these
instead of afl-gcc
- added gcc_plugin which is like llvm_mode but for gcc. This version
supports gcc version 5 to 8. See gcc_plugin/README (https://github.com/T12z/afl)
- added instrim, a much faster llvm_mode instrumentation at the cost of
path discovery. See llvm_mode/README.instrim (https://github.com/csienslab/instrim)
- added MOpt (github.com/puppet-meteor/MOpt-AFL) mode, see docs/README.MOpt

View File

@ -17,6 +17,7 @@ afl-qemu-optimize-entrypoint.diff by mh(at)mh-sec(dot)de
afl-qemu-speed.diff by abiondo on github
afl-qemu-optimize-map.diff by mh(at)mh-sec(dot)de
+ gcc_plugin (https://github.com/T12z/afl)
+ instrim (https://github.com/csienslab/instrim) was integrated
+ MOpt (github.com/puppet-meteor/MOpt-AFL) was imported
+ AFLfast additions (github.com/mboehme/aflfast) were incorporated.

View File

@ -25,6 +25,9 @@ american fuzzy lop plus plus
Also newly integrated is instrim, a very effective CFG llvm_mode
instrumentation implementation from https://github.com/csienslab/instrim
Similar to llvm_mode a gcc_plugin that supports versions 5 to 8 is
now available (from https://github.com/T12z/afl)
A more thorough list is available in the PATCHES file.
So all in all this is the best-of AFL that is currently out there :-)

75
gcc_plugin/Makefile Normal file
View File

@ -0,0 +1,75 @@
#
# american fuzzy lop - GCC plugin instrumentation
# -----------------------------------------------
#
# Written by Austin Seipp <aseipp@pobox.com> and
# Laszlo Szekeres <lszekeres@google.com> and
# Michal Zalewski <lcamtuf@google.com>
#
# GCC integration design is based on the LLVM design, which comes
# from Laszlo Szekeres.
#
# Copyright 2015 Google Inc. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at:
#
# http://www.apache.org/licenses/LICENSE-2.0
#
PREFIX ?= /usr/local
HELPER_PATH = $(PREFIX)/lib/afl
BIN_PATH = $(PREFIX)/bin
CFLAGS ?= -O3 -g -funroll-loops
CFLAGS += -Wall -D_FORTIFY_SOURCE=2 -Wno-pointer-sign \
-DAFL_PATH=\"$(HELPER_PATH)\" -DBIN_PATH=\"$(BIN_PATH)\" \
CXXFLAGS ?= -O3 -g -funroll-loops
CXXEFLAGS := $(CXXFLAGS) -Wall -D_FORTIFY_SOURCE=2
CC ?= gcc
CXX ?= g++
PLUGIN_FLAGS = -fPIC -fno-rtti -I"$(shell $(CC) -print-file-name=plugin)/include"
PROGS = ../afl-gcc-fast ../afl-gcc-pass.so ../afl-gcc-rt.o
all: test_deps $(PROGS) test_build all_done
test_deps:
@echo "[*] Checking for working '$(CC)'..."
@which $(CC) >/dev/null 2>&1 || ( echo "[-] Oops, can't find '$(CC)'. Make sure that it's in your \$$PATH (or set \$$CC and \$$CXX)."; exit 1 )
@echo "[*] Checking for '../afl-showmap'..."
@test -f ../afl-showmap || ( echo "[-] Oops, can't find '../afl-showmap'. Be sure to compile AFL first."; exit 1 )
@echo "[+] All set and ready to build."
../afl-gcc-fast: afl-gcc-fast.c | test_deps
$(CC) $(CFLAGS) $< -o $@ $(LDFLAGS)
ln -sf afl-gcc-fast ../afl-g++-fast
../afl-gcc-pass.so: afl-gcc-pass.so.cc | test_deps
$(CXX) $(CXXEFLAGS) $(PLUGIN_FLAGS) -shared $< -o $@
../afl-gcc-rt.o: afl-gcc-rt.o.c | test_deps
$(CC) $(CFLAGS) -fPIC -c $< -o $@
test_build: $(PROGS)
@echo "[*] Testing the CC wrapper and instrumentation output..."
# unset AFL_USE_ASAN AFL_USE_MSAN; AFL_QUIET=1 AFL_INST_RATIO=100 AFL_PATH=. AFL_CC=$(CC) ../afl-gcc-fast $(CFLAGS) ../test-instr.c -o test-instr $(LDFLAGS)
unset AFL_USE_ASAN AFL_USE_MSAN; AFL_INST_RATIO=100 AFL_PATH=. AFL_CC=$(CC) ../afl-gcc-fast $(CFLAGS) ../test-instr.c -o test-instr $(LDFLAGS)
echo 0 | ../afl-showmap -m none -q -o .test-instr0 ./test-instr
echo 1 | ../afl-showmap -m none -q -o .test-instr1 ./test-instr
@rm -f test-instr
@cmp -s .test-instr0 .test-instr1; DR="$$?"; rm -f .test-instr0 .test-instr1; if [ "$$DR" = "0" ]; then echo; echo "Oops, the instrumentation does not seem to be behaving correctly!"; echo; echo "Please ping <lcamtuf@google.com> to troubleshoot the issue."; echo; exit 1; fi
@echo "[+] All right, the instrumentation seems to be working!"
all_done: test_build
@echo "[+] All done! You can now use '../afl-gcc-fast' to compile programs."
.NOTPARALLEL: clean
clean:
rm -f *.o *.so *~ a.out core core.[1-9][0-9]* test-instr .test-instr0 .test-instr1
rm -f $(PROGS) ../afl-g++-fast

157
gcc_plugin/README.gcc Normal file
View File

@ -0,0 +1,157 @@
===========================================
Fast GCC-based instrumentation for afl-fuzz
===========================================
(See ../docs/README for the general instruction manual.)
(See ../llvm_mode/README.llvm for the LLVM-based instrumentation.)
1) Introduction
---------------
The code in this directory allows you to instrument programs for AFL using
true compiler-level instrumentation, instead of the more crude
assembly-level rewriting approach taken by afl-gcc and afl-clang. This has
several interesting properties:
- The compiler can make many optimizations that are hard to pull off when
manually inserting assembly. As a result, some slow, CPU-bound programs will
run up to around 2x faster.
The gains are less pronounced for fast binaries, where the speed is limited
chiefly by the cost of creating new processes. In such cases, the gain will
probably stay within 10%.
- The instrumentation is CPU-independent. At least in principle, you should
be able to rely on it to fuzz programs on non-x86 architectures (after
building afl-fuzz with AFL_NOX86=1).
- Because the feature relies on the internals of GCC, it is gcc-specific
and will *not* work with LLVM (see ../llvm_plugin for an alternative).
Once this implementation is shown to be sufficiently robust and portable, it
will probably replace afl-gcc. For now, it can be built separately and
co-exists with the original code.
The idea and much of the implementation comes from Laszlo Szekeres.
2) How to use
-------------
In order to leverage this mechanism, you need to have GCC and the plugin headers
installed on your system. That should be all you need. On Debian machines, these
headers can be acquired by installing the `gcc-<VERSION>-plugin-dev` packages.
To build the instrumentation itself, type 'make'. This will generate binaries
called afl-gcc-fast and afl-g++-fast in the parent directory. Once this
is done, you can instrument third-party code in a way similar to the standard
operating mode of AFL, e.g.:
CC=/path/to/afl/afl-gcc-fast ./configure [...options...]
make
Be sure to also include CXX set to afl-g++-fast for C++ code.
The tool honors roughly the same environmental variables as afl-gcc (see
../docs/env_variables.txt). This includes AFL_INST_RATIO, AFL_USE_ASAN,
AFL_HARDEN, and AFL_DONT_OPTIMIZE.
Note: if you want the GCC plugin to be installed on your system for all
users, you need to build it before issuing 'make install' in the parent
directory.
3) Gotchas, feedback, bugs
--------------------------
This is an early-stage mechanism, so field reports are welcome. You can send bug
reports to <aseipp@pobox.com>.
4) Bonus feature #1: deferred instrumentation
---------------------------------------------
AFL tries to optimize performance by executing the targeted binary just once,
stopping it just before main(), and then cloning this "master" process to get
a steady supply of targets to fuzz.
Although this approach eliminates much of the OS-, linker- and libc-level
costs of executing the program, it does not always help with binaries that
perform other time-consuming initialization steps - say, parsing a large config
file before getting to the fuzzed data.
In such cases, it's beneficial to initialize the forkserver a bit later, once
most of the initialization work is already done, but before the binary attempts
to read the fuzzed input and parse it; in some cases, this can offer a 10x+
performance gain. You can implement delayed initialization in LLVM mode in a
fairly simple way.
First, locate a suitable location in the code where the delayed cloning can
take place. This needs to be done with *extreme* care to avoid breaking the
binary. In particular, the program will probably malfunction if you select
a location after:
- The creation of any vital threads or child processes - since the forkserver
can't clone them easily.
- The initialization of timers via setitimer() or equivalent calls.
- The creation of temporary files, network sockets, offset-sensitive file
descriptors, and similar shared-state resources - but only provided that
their state meaningfully influences the behavior of the program later on.
- Any access to the fuzzed input, including reading the metadata about its
size.
With the location selected, add this code in the appropriate spot:
#ifdef __AFL_HAVE_MANUAL_CONTROL
__AFL_INIT();
#endif
You don't need the #ifdef guards, but they will make the program still work as
usual when compiled with a tool other than afl-gcc-fast/afl-clang-fast.
Finally, recompile the program with afl-gcc-fast (afl-gcc or afl-clang will
*not* generate a deferred-initialization binary) - and you should be all set!
5) Bonus feature #2: persistent mode
------------------------------------
Some libraries provide APIs that are stateless, or whose state can be reset in
between processing different input files. When such a reset is performed, a
single long-lived process can be reused to try out multiple test cases,
eliminating the need for repeated fork() calls and the associated OS overhead.
The basic structure of the program that does this would be:
while (__AFL_LOOP(1000)) {
/* Read input data. */
/* Call library code to be fuzzed. */
/* Reset state. */
}
/* Exit normally */
The numerical value specified within the loop controls the maximum number
of iterations before AFL will restart the process from scratch. This minimizes
the impact of memory leaks and similar glitches; 1000 is a good starting point.
A more detailed template is shown in ../experimental/persistent_demo/.
Similarly to the previous mode, the feature works only with afl-gcc-fast or
afl-clang-fast; #ifdef guards can be used to suppress it when using other
compilers.
Note that as with the previous mode, the feature is easy to misuse; if you
do not reset the critical state fully, you may end up with false positives or
waste a whole lot of CPU power doing nothing useful at all. Be particularly
wary of memory leaks and the state of file descriptors.
When running in this mode, the execution paths will inherently vary a bit
depending on whether the input loop is being entered for the first time or
executed again. To avoid spurious warnings, the feature implies
AFL_NO_VAR_CHECK and hides the "variable path" warnings in the UI.
PS. Because there are task switches still involved, the mode isn't as fast as
"pure" in-process fuzzing offered, say, by LLVM's LibFuzzer; but it is a lot
faster than the normal fork() model, and compared to in-process fuzzing,
should be a lot more robust.

283
gcc_plugin/afl-gcc-fast.c Normal file
View File

@ -0,0 +1,283 @@
/*
american fuzzy lop - GCC wrapper for GCC plugin
------------------------------------------------
Written by Austin Seipp <aseipp@pobox.com> and
Laszlo Szekeres <lszekeres@google.com> and
Michal Zalewski <lcamtuf@google.com>
GCC integration design is based on the LLVM design, which comes
from Laszlo Szekeres.
Copyright 2015 Google Inc. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at:
http://www.apache.org/licenses/LICENSE-2.0
This program is a drop-in replacement for gcc, similar in most
respects to ../afl-gcc, but with compiler instrumentation through a
plugin. It tries to figure out compilation mode, adds a bunch of
flags, and then calls the real compiler.
*/
#define AFL_MAIN
#include "../config.h"
#include "../types.h"
#include "../debug.h"
#include "../alloc-inl.h"
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <string.h>
static u8* obj_path; /* Path to runtime libraries */
static u8** cc_params; /* Parameters passed to the real CC */
static u32 cc_par_cnt = 1; /* Param count, including argv0 */
/* Try to find the runtime libraries. If that fails, abort. */
static void find_obj(u8* argv0) {
u8 *afl_path = getenv("AFL_PATH");
u8 *slash, *tmp;
if (afl_path) {
tmp = alloc_printf("%s/afl-gcc-rt.o", afl_path);
if (!access(tmp, R_OK)) {
obj_path = afl_path;
ck_free(tmp);
return;
}
ck_free(tmp);
}
slash = strrchr(argv0, '/');
if (slash) {
u8 *dir;
*slash = 0;
dir = ck_strdup(argv0);
*slash = '/';
tmp = alloc_printf("%s/afl-gcc-rt.o", dir);
if (!access(tmp, R_OK)) {
obj_path = dir;
ck_free(tmp);
return;
}
ck_free(tmp);
ck_free(dir);
}
if (!access(AFL_PATH "/afl-gcc-rt.o", R_OK)) {
obj_path = AFL_PATH;
return;
}
FATAL("Unable to find 'afl-gcc-rt.o' or 'afl-gcc-pass.so'. Please set AFL_PATH");
}
/* Copy argv to cc_params, making the necessary edits. */
static void edit_params(u32 argc, char** argv) {
u8 fortify_set = 0, asan_set = 0, x_set = 0, maybe_linking = 1;
u8 *name;
cc_params = ck_alloc((argc + 64) * sizeof(u8*));
name = strrchr(argv[0], '/');
if (!name) name = argv[0]; else name++;
if (!strcmp(name, "afl-g++-fast")) {
u8* alt_cxx = getenv("AFL_CXX");
cc_params[0] = alt_cxx ? alt_cxx : (u8*)"g++";
} else {
u8* alt_cc = getenv("AFL_CC");
cc_params[0] = alt_cc ? alt_cc : (u8*)"gcc";
}
char* fplugin_arg = alloc_printf("-fplugin=%s/afl-gcc-pass.so", obj_path);
cc_params[cc_par_cnt++] = fplugin_arg;
while (--argc) {
u8* cur = *(++argv);
#if defined(__x86_64__)
if (!strcmp(cur, "-m32")) FATAL("-m32 is not supported");
#endif
if (!strcmp(cur, "-x")) x_set = 1;
if (!strcmp(cur, "-c") || !strcmp(cur, "-S") || !strcmp(cur, "-E") ||
!strcmp(cur, "-v")) maybe_linking = 0;
if (!strcmp(cur, "-fsanitize=address") ||
!strcmp(cur, "-fsanitize=memory")) asan_set = 1;
if (strstr(cur, "FORTIFY_SOURCE")) fortify_set = 1;
cc_params[cc_par_cnt++] = cur;
}
if (getenv("AFL_HARDEN")) {
cc_params[cc_par_cnt++] = "-fstack-protector-all";
if (!fortify_set)
cc_params[cc_par_cnt++] = "-D_FORTIFY_SOURCE=2";
}
if (!asan_set) {
if (getenv("AFL_USE_ASAN")) {
cc_params[cc_par_cnt++] = "-fsanitize=address";
if (getenv("AFL_USE_MSAN"))
FATAL("ASAN and MSAN are mutually exclusive");
} else if (getenv("AFL_USE_MSAN")) {
cc_params[cc_par_cnt++] = "-fsanitize=memory";
if (getenv("AFL_USE_ASAN"))
FATAL("ASAN and MSAN are mutually exclusive");
}
}
if (!getenv("AFL_DONT_OPTIMIZE")) {
cc_params[cc_par_cnt++] = "-g";
cc_params[cc_par_cnt++] = "-O3";
cc_params[cc_par_cnt++] = "-funroll-loops";
}
cc_params[cc_par_cnt++] = "-D__AFL_HAVE_MANUAL_CONTROL=1";
/* When the user tries to use persistent or deferred forkserver modes by
appending a single line to the program, we want to reliably inject a
signature into the binary (to be picked up by afl-fuzz) and we want
to call a function from the runtime .o file. This is unnecessarily
painful for three reasons:
1) We need to convince the compiler not to optimize out the signature.
This is done with __attribute__((used)).
2) We need to convince the linker, when called with -Wl,--gc-sections,
not to do the same. This is done by forcing an assignment to a
'volatile' pointer.
3) We need to declare __afl_persistent_loop() in the global namespace,
but doing this within a method in a class is hard - :: and extern "C"
are forbidden and __attribute__((alias(...))) doesn't work. Hence the
__asm__ aliasing trick.
*/
cc_params[cc_par_cnt++] = "-D__AFL_LOOP(_A)="
"({ static volatile char *_B __attribute__((used)); "
" _B = (char*)\"" PERSIST_SIG "\"; "
#ifdef __APPLE__
"int _L(unsigned int) __asm__(\"___afl_persistent_loop\"); "
#else
"int _L(unsigned int) __asm__(\"__afl_persistent_loop\"); "
#endif /* ^__APPLE__ */
"_L(_A); })";
cc_params[cc_par_cnt++] = "-D__AFL_INIT()="
"do { static volatile char *_A __attribute__((used)); "
" _A = (char*)\"" DEFER_SIG "\"; "
#ifdef __APPLE__
"void _I(void) __asm__(\"___afl_manual_init\"); "
#else
"void _I(void) __asm__(\"__afl_manual_init\"); "
#endif /* ^__APPLE__ */
"_I(); } while (0)";
if (maybe_linking) {
if (x_set) {
cc_params[cc_par_cnt++] = "-x";
cc_params[cc_par_cnt++] = "none";
}
cc_params[cc_par_cnt++] = alloc_printf("%s/afl-gcc-rt.o", obj_path);
}
cc_params[cc_par_cnt] = NULL;
}
/* Main entry point */
int main(int argc, char** argv) {
if (isatty(2) && !getenv("AFL_QUIET")) {
SAYF(cCYA "afl-gcc-fast" cRST " initial version 1.94 by <aseipp@pobox.com>, updated to " cBRI VERSION cRST " by <thorsten.schulz@uni-rostock.de>\n");
}
if (argc < 2) {
SAYF("\n"
"This is a helper application for afl-fuzz. It serves as a drop-in replacement\n"
"for gcc, letting you recompile third-party code with the required runtime\n"
"instrumentation. A common use pattern would be one of the following:\n\n"
" CC=%s/afl-gcc-fast ./configure\n"
" CXX=%s/afl-g++-fast ./configure\n\n"
"In contrast to the traditional afl-gcc tool, this version is implemented as\n"
"a GCC plugin and tends to offer improved performance with slow programs\n"
"(similarly to the LLVM plugin used by afl-clang-fast).\n\n"
"You can specify custom next-stage toolchain via AFL_CC and AFL_CXX. Setting\n"
"AFL_HARDEN enables hardening optimizations in the compiled code.\n\n",
BIN_PATH, BIN_PATH);
exit(1);
}
find_obj(argv[0]);
edit_params(argc, argv);
/*if (isatty(2) && !getenv("AFL_QUIET")) {
printf("Calling \"%s\" with:\n", cc_params[0]);
for(int i=1; i<cc_par_cnt; i++) printf("%s\n", cc_params[i]);
}
*/
execvp(cc_params[0], (char**)cc_params);
FATAL("Oops, failed to execute '%s' - check your PATH", cc_params[0]);
return 0;
}

View File

@ -0,0 +1,372 @@
//
// There are two TODOs in this file:
// - dont instrument blocks that are uninterested
// - implement whitelist feature
//
/*
american fuzzy lop - GCC instrumentation pass
---------------------------------------------
Written by Austin Seipp <aseipp@pobox.com> with bits from
Emese Revfy <re.emese@gmail.com>
Updated by Thorsten Schulz <thorsten.schulz@uni-rostock.de>
GCC integration design is based on the LLVM design, which comes
from Laszlo Szekeres. Some of the boilerplate code below for
afl_pass to adapt to different GCC versions was taken from Emese
Revfy's Size Overflow plugin for GCC, licensed under the GPLv2/v3.
(NOTE: this plugin code is under GPLv3, in order to comply with the
GCC runtime library exception, which states that you may distribute
"Target Code" from the compiler under a license of your choice, as
long as the "Compilation Process" is "Eligible", and contains no
GPL-incompatible software in GCC "during the process of
transforming high level code to target code". In this case, the
plugin will be used to generate "Target Code" during the
"Compilation Process", and thus it must be GPLv3 to be "eligible".)
Copyright (C) 2015 Austin Seipp
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
//#define BUILD_INLINE_INST
#include "../config.h"
#include "../debug.h"
/* clear helper AFL types pulls in, which intervene with gcc-plugin geaders from GCC-8 */
#ifdef likely
#undef likely
#endif
#ifdef unlikely
#undef unlikely
#endif
#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <gcc-plugin.h>
#include <plugin-version.h>
#include <diagnostic.h>
#include <tree.h>
#include <tree-ssa.h>
#include <tree-pass.h>
#include <tree-ssa-alias.h>
#include <basic-block.h>
#include <gimple-expr.h>
#include <gimple.h>
#include <gimple-iterator.h>
#include <version.h>
#include <toplev.h>
#include <intl.h>
#include <context.h>
#include <stringpool.h>
#include <cgraph.h>
#include <cfgloop.h>
/* -------------------------------------------------------------------------- */
/* -- AFL instrumentation pass ---------------------------------------------- */
static int be_quiet = 0;
static unsigned int inst_ratio = 100;
static bool inst_ext = true; /* I reckon inline is broken / unfunctional */
static unsigned int ext_call_instrument(function *fun) {
/* Instrument all the things! */
basic_block bb;
unsigned finst_blocks = 0;
unsigned fcnt_blocks = 0;
FOR_ALL_BB_FN(bb, fun) {
gimple_seq fcall;
gimple_seq seq = NULL;
gimple_stmt_iterator bentry;
if (!fcnt_blocks++) continue; /* skip block 0 */
// TODO: if the predecessor does not have to destinations
// then skip this block :TODO
/* Bail on this block if we trip the specified ratio */
if (R(100) >= inst_ratio) continue;
/* Make up cur_loc */
unsigned int rand_loc = R(MAP_SIZE);
tree cur_loc = build_int_cst(uint64_type_node, rand_loc);
/* Update bitmap via external call */
/* to quote:
* /+ Trace a basic block with some ID +/
* void __afl_trace(u16 x);
*/
tree fntype = build_function_type_list(
void_type_node, /* return */
uint16_type_node, /* args */
NULL_TREE); /* done */
tree fndecl = build_fn_decl("__afl_trace", fntype);
TREE_STATIC(fndecl) = 1; /* Defined elsewhere */
TREE_PUBLIC(fndecl) = 1; /* Public */
DECL_EXTERNAL(fndecl) = 1; /* External linkage */
DECL_ARTIFICIAL(fndecl) = 1; /* Injected by compiler */
fcall = gimple_build_call(fndecl, 1, cur_loc); /* generate the function _call_ to above built reference, with *1* parameter -> the random const for the location */
gimple_seq_add_stmt(&seq, fcall); /* and insert into a sequence */
/* Done - grab the entry to the block and insert sequence */
bentry = gsi_start_bb(bb);
gsi_insert_seq_before(&bentry, seq, GSI_SAME_STMT);
finst_blocks++;
}
fcnt_blocks--; /* discard the first in the count */
/* Say something nice. */
if (!be_quiet) {
if (!finst_blocks)
WARNF(G_("No instrumentation targets found in " cBRI "%s" cRST ),
function_name(fun));
else if (finst_blocks < fcnt_blocks)
OKF(G_("Instrumented %2u /%2u locations in " cBRI "%s" cRST ),
finst_blocks, fcnt_blocks,
function_name(fun));
else
OKF(G_("Instrumented %2u locations in " cBRI "%s" cRST ),
finst_blocks,
function_name(fun));
}
return 0;
}
static unsigned int inline_instrument(function *fun) {
#ifdef BUILD_INLINE_INST /* ifdef inline away, so I don't have to refactor it */
/* Instrument all the things! */
basic_block bb;
unsigned finst_blocks = 0;
unsigned fcnt_blocks = 0;
/* Set up global type declarations */
tree map_type = build_pointer_type(unsigned_char_type_node);
tree map_ptr_g = build_decl(UNKNOWN_LOCATION, VAR_DECL, get_identifier_with_length("__afl_area_ptr",14), map_type);
TREE_USED(map_ptr_g) = 1;
TREE_STATIC(map_ptr_g) = 1; /* Defined elsewhere */
DECL_EXTERNAL(map_ptr_g) = 1; /* External linkage */
DECL_PRESERVE_P(map_ptr_g) = 1;
DECL_ARTIFICIAL(map_ptr_g) = 1;
rest_of_decl_compilation(map_ptr_g, 1, 0);
tree prev_loc_g = build_decl(UNKNOWN_LOCATION, VAR_DECL, get_identifier_with_length("__afl_prev_loc",14), uint16_type_node);
TREE_USED(prev_loc_g) = 1;
TREE_STATIC(prev_loc_g) = 1; /* Defined elsewhere */
DECL_EXTERNAL(prev_loc_g) = 1; /* External linkage */
DECL_PRESERVE_P(prev_loc_g) = 1;
DECL_ARTIFICIAL(prev_loc_g) = 1;
rest_of_decl_compilation(prev_loc_g, 1, 0);
FOR_ALL_BB_FN(bb, fun) {
gimple *g;
gimple_seq seq = NULL;
gimple_stmt_iterator bentry;
if (!fcnt_blocks++) continue; /* skip block 0 */
/* Bail on this block if we trip the specified ratio */
if (R(100) >= inst_ratio) continue;
/* Make up cur_loc */
unsigned int rand_loc = R(MAP_SIZE);
tree cur_loc = build_int_cst(uint64_type_node, rand_loc);
/* Load prev_loc, xor with cur_loc */
tree area_off = create_tmp_var(uint64_type_node, "area_off");
g = gimple_build_assign(area_off, BIT_XOR_EXPR, prev_loc_g, cur_loc);
gimple_seq_add_stmt(&seq, g); // area_off = prev_loc ^ cur_loc
/* Update bitmap */
// tree zero = build_int_cst(unsigned_char_type_node, 0);
tree one = build_int_cst(unsigned_char_type_node, 1);
tree tmp1 = create_tmp_var(map_type, "tmp1");
g = gimple_build_assign(tmp1, PLUS_EXPR, map_ptr_g, area_off);
gimple_seq_add_stmt(&seq, g); // tmp1 = __afl_area_ptr + area_off
SAYF(G_("%d,"), fcnt_blocks);
tree tmp2 = create_tmp_var(unsigned_char_type_node, "tmp2");
//tree tmp1_ptr = build_simple_mem_ref_loc(UNKNOWN_LOCATION, tmp1);
g = gimple_build_assign(tmp2, INDIRECT_REF, tmp1);
gimple_seq_add_stmt(&seq, g); // tmp2 = *tmp1
tree tmp3 = create_tmp_var(unsigned_char_type_node, "tmp3");
g = gimple_build_assign(tmp3, PLUS_EXPR, tmp2, one);
gimple_seq_add_stmt(&seq, g); // tmp3 = tmp2 + 1
// tree tmp4 = create_tmp_var(map_type, "tmp4");
// g = gimple_build_assign(tmp4, PLUS_EXPR, map_ptr_g, area_off);
// gimple_seq_add_stmt(&seq, g); // tmp4 = __afl_area_ptr + area_off
// tree deref2 = build2(MEM_REF, map_type, tmp4, zero);
tree deref2 = build4(ARRAY_REF, map_type, map_ptr_g, area_off, NULL, NULL);
g = gimple_build_assign(deref2, MODIFY_EXPR, tmp3);
gimple_seq_add_stmt(&seq, g); // *tmp4 = tmp3
SAYF(G_("+%d,"), fcnt_blocks);
/* Set prev_loc to cur_loc >> 1 */
tree shifted_loc = build_int_cst(TREE_TYPE(prev_loc_g), rand_loc >> 1);
g = gimple_build_assign(prev_loc_g, MODIFY_EXPR, shifted_loc);
gimple_seq_add_stmt(&seq, g); // __afl_pred_loc = cur_loc >> 1
/* Done - grab the entry to the block and insert sequence */
bentry = gsi_start_bb(bb);
gsi_insert_seq_before(&bentry, seq, GSI_SAME_STMT);
inst_blocks++;
finst_blocks++;
}
/* Say something nice. */
if (!be_quiet) {
if (!finst_blocks)
WARNF(G_("No instrumentation targets found in " cBRI "%s" cRST ),
function_name(fun));
else if (finst_blocks < fcnt_blocks)
OKF(G_("Instrumented %2u /%2u locations in " cBRI "%s" cRST ),
finst_blocks, fcnt_blocks,
function_name(fun));
else
OKF(G_("Instrumented %2u locations in " cBRI "%s" cRST ),
finst_blocks,
function_name(fun));
}
#endif
return 0;
}
/* -------------------------------------------------------------------------- */
/* -- Boilerplate and initialization ---------------------------------------- */
static const struct pass_data afl_pass_data = {
.type = GIMPLE_PASS,
.name = "afl-inst",
.optinfo_flags = OPTGROUP_NONE,
.tv_id = TV_NONE,
.properties_required = 0,
.properties_provided = 0,
.properties_destroyed = 0,
.todo_flags_start = 0,
// NOTE(aseipp): it's very, very important to include
// at least 'TODO_update_ssa' here so that GCC will
// properly update the resulting SSA form, e.g., to
// include new PHI nodes for newly added symbols or
// names. Do not remove this. Do not taunt Happy Fun
// Ball.
.todo_flags_finish = TODO_update_ssa | TODO_verify_il | TODO_cleanup_cfg,
};
namespace {
class afl_pass : public gimple_opt_pass {
private:
bool do_ext_call;
public:
afl_pass(bool ext_call, gcc::context *g) : gimple_opt_pass(afl_pass_data, g), do_ext_call(ext_call) {}
virtual unsigned int execute(function *fun) {
// TODO: implement whitelist feature here :TODO
return do_ext_call ? ext_call_instrument(fun) : inline_instrument(fun);
}
}; /* class afl_pass */
} /* anon namespace */
static struct opt_pass *make_afl_pass(bool ext_call, gcc::context *ctxt) {
return new afl_pass(ext_call, ctxt);
}
/* -------------------------------------------------------------------------- */
/* -- Initialization -------------------------------------------------------- */
int plugin_is_GPL_compatible = 1;
static struct plugin_info afl_plugin_info = {
.version = "20181200",
.help = "AFL gcc plugin\n",
};
int plugin_init(struct plugin_name_args *plugin_info,
struct plugin_gcc_version *version) {
struct register_pass_info afl_pass_info;
struct timeval tv;
struct timezone tz;
u32 rand_seed;
/* Setup random() so we get Actually Random(TM) outputs from R() */
gettimeofday(&tv, &tz);
rand_seed = tv.tv_sec ^ tv.tv_usec ^ getpid();
srandom(rand_seed);
/* Pass information */
afl_pass_info.pass = make_afl_pass(inst_ext, g);
afl_pass_info.reference_pass_name = "ssa";
afl_pass_info.ref_pass_instance_number = 1;
afl_pass_info.pos_op = PASS_POS_INSERT_AFTER;
if (!plugin_default_version_check(version, &gcc_version)) {
FATAL(G_("Incompatible gcc/plugin versions!"));
}
/* Show a banner */
if (isatty(2) && !getenv("AFL_QUIET")) {
SAYF(G_(cCYA "afl-gcc-pass" cRST " initial version 1.94 by <aseipp@pobox.com>, updated to " cBRI VERSION cRST " by <thorsten.schulz@uni-rostock.de>\n"));
} else
be_quiet = 1;
/* Decide instrumentation ratio */
char* inst_ratio_str = getenv("AFL_INST_RATIO");
if (inst_ratio_str) {
if (sscanf(inst_ratio_str, "%u", &inst_ratio) != 1 || !inst_ratio || inst_ratio > 100)
FATAL(G_("Bad value of AFL_INST_RATIO (must be between 1 and 100)"));
else {
if (!be_quiet)
ACTF(G_("%s instrumentation at ratio of %u%% in %s mode."),
inst_ext ? G_("Call-based") : G_("Inline"),
inst_ratio,
getenv("AFL_HARDEN") ? G_("hardened") : G_("non-hardened"));
}
}
/* Go go gadget */
register_callback(plugin_info->base_name, PLUGIN_INFO, NULL, &afl_plugin_info);
register_callback(plugin_info->base_name, PLUGIN_PASS_MANAGER_SETUP, NULL, &afl_pass_info);
return 0;
}

224
gcc_plugin/afl-gcc-rt.o.c Normal file
View File

@ -0,0 +1,224 @@
/*
american fuzzy lop - GCC plugin instrumentation bootstrap
---------------------------------------------------------
Written by Austin Seipp <aseipp@pobox.com> and
Laszlo Szekeres <lszekeres@google.com> and
Michal Zalewski <lcamtuf@google.com>
GCC integration design is based on the LLVM design, which comes
from Laszlo Szekeres.
Copyright 2015 Google Inc. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at:
http://www.apache.org/licenses/LICENSE-2.0
This code is the rewrite of afl-as.h's main_payload.
*/
#include "../config.h"
#include "../types.h"
#include <stdlib.h>
#include <signal.h>
#include <unistd.h>
#include <assert.h>
#include <sys/mman.h>
#include <sys/shm.h>
#include <sys/wait.h>
#include <sys/types.h>
/* Globals needed by the injected instrumentation. The __afl_area_initial region
is used for instrumentation output before __afl_map_shm() has a chance to run.
It will end up as .comm, so it shouldn't be too wasteful. */
u8 __afl_area_initial[MAP_SIZE];
u8* __afl_area_ptr = __afl_area_initial;
u16 __afl_prev_loc;
/* Running in persistent mode? */
static u8 is_persistent;
/* Trace a basic block with some ID */
void __afl_trace(u16 x) {
u16 l = __afl_prev_loc;
u16 n = l ^ x;
*(__afl_area_ptr+n) += 1;
__afl_prev_loc = (x >> 1);
return;
}
/* SHM setup. */
static void __afl_map_shm(void) {
u8 *id_str = getenv(SHM_ENV_VAR);
/* If we're running under AFL, attach to the appropriate region, replacing the
early-stage __afl_area_initial region that is needed to allow some really
hacky .init code to work correctly in projects such as OpenSSL. */
if (id_str) {
u32 shm_id = atoi(id_str);
__afl_area_ptr = shmat(shm_id, NULL, 0);
/* Whooooops. */
if (__afl_area_ptr == (void *)-1) exit(1);
/* Write something into the bitmap so that even with low AFL_INST_RATIO,
our parent doesn't give up on us. */
__afl_area_ptr[0] = 1;
}
}
/* Fork server logic. */
static void __afl_start_forkserver(void) {
static u8 tmp[4];
s32 child_pid;
u8 child_stopped = 0;
/* Phone home and tell the parent that we're OK. If parent isn't there,
assume we're not running in forkserver mode and just execute program. */
if (write(FORKSRV_FD + 1, tmp, 4) != 4) return;
while (1) {
u32 was_killed;
int status;
/* Wait for parent by reading from the pipe. Abort if read fails. */
if (read(FORKSRV_FD, &was_killed, 4) != 4) exit(1);
/* If we stopped the child in persistent mode, but there was a race
condition and afl-fuzz already issued SIGKILL, write off the old
process. */
if (child_stopped && was_killed) {
child_stopped = 0;
if (waitpid(child_pid, &status, 0) < 0) exit(1);
}
if (!child_stopped) {
/* Once woken up, create a clone of our process. */
child_pid = fork();
if (child_pid < 0) exit(1);
/* In child process: close fds, resume execution. */
if (!child_pid) {
close(FORKSRV_FD);
close(FORKSRV_FD + 1);
return;
}
} else {
/* Special handling for persistent mode: if the child is alive but
currently stopped, simply restart it with SIGCONT. */
kill(child_pid, SIGCONT);
child_stopped = 0;
}
/* In parent process: write PID to pipe, then wait for child. */
if (write(FORKSRV_FD + 1, &child_pid, 4) != 4) exit(1);
if (waitpid(child_pid, &status, is_persistent ? WUNTRACED : 0) < 0)
exit(1);
/* In persistent mode, the child stops itself with SIGSTOP to indicate
a successful run. In this case, we want to wake it up without forking
again. */
if (WIFSTOPPED(status)) child_stopped = 1;
/* Relay wait status to pipe, then loop back. */
if (write(FORKSRV_FD + 1, &status, 4) != 4) exit(1);
}
}
/* A simplified persistent mode handler, used as explained in README.llvm. */
int __afl_persistent_loop(unsigned int max_cnt) {
static u8 first_pass = 1;
static u32 cycle_cnt;
if (first_pass) {
cycle_cnt = max_cnt;
first_pass = 0;
return 1;
}
if (is_persistent && --cycle_cnt) {
raise(SIGSTOP);
return 1;
} else return 0;
}
/* This one can be called from user code when deferred forkserver mode
is enabled. */
void __afl_manual_init(void) {
static u8 init_done;
if (!init_done) {
__afl_map_shm();
__afl_start_forkserver();
init_done = 1;
}
}
/* Proper initialization routine. */
__attribute__((constructor(101))) void __afl_auto_init(void) {
is_persistent = !!getenv(PERSIST_ENV_VAR);
if (getenv(DEFER_ENV_VAR)) return;
__afl_manual_init();
}