2015-04-24 05:29:37 +08:00
|
|
|
=================
|
2015-04-24 04:40:04 +08:00
|
|
|
SanitizerCoverage
|
2015-04-24 05:29:37 +08:00
|
|
|
=================
|
2015-04-24 04:40:04 +08:00
|
|
|
|
|
|
|
.. contents::
|
|
|
|
:local:
|
|
|
|
|
|
|
|
Introduction
|
|
|
|
============
|
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
LLVM has a simple code coverage instrumentation built in (SanitizerCoverage).
|
|
|
|
It inserts calls to user-defined functions on function-, basic-block-, and edge- levels.
|
|
|
|
Default implementations of those callbacks are provided and implement
|
|
|
|
simple coverage reporting and visualization,
|
|
|
|
however if you need *just* coverage visualization you may want to use
|
|
|
|
:doc:`SourceBasedCodeCoverage <SourceBasedCodeCoverage>` instead.
|
2016-02-18 05:34:43 +08:00
|
|
|
|
2016-09-14 09:39:49 +08:00
|
|
|
Tracing PCs with guards
|
|
|
|
=======================
|
|
|
|
|
|
|
|
With ``-fsanitize-coverage=trace-pc-guard`` the compiler will insert the following code
|
|
|
|
on every edge:
|
|
|
|
|
|
|
|
.. code-block:: none
|
|
|
|
|
2017-01-24 08:57:31 +08:00
|
|
|
__sanitizer_cov_trace_pc_guard(&guard_variable)
|
2016-09-14 09:39:49 +08:00
|
|
|
|
2016-09-30 01:43:24 +08:00
|
|
|
Every edge will have its own `guard_variable` (uint32_t).
|
2016-09-16 06:11:08 +08:00
|
|
|
|
2020-01-19 22:51:25 +08:00
|
|
|
The compiler will also insert calls to a module constructor:
|
2016-09-14 09:39:49 +08:00
|
|
|
|
|
|
|
.. code-block:: c++
|
|
|
|
|
2016-09-17 13:03:05 +08:00
|
|
|
// The guards are [start, stop).
|
2017-02-09 05:28:39 +08:00
|
|
|
// This function will be called at least once per DSO and may be called
|
|
|
|
// more than once with the same values of start/stop.
|
2016-09-30 02:34:40 +08:00
|
|
|
__sanitizer_cov_trace_pc_guard_init(uint32_t *start, uint32_t *stop);
|
2016-09-17 13:03:05 +08:00
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
With an additional ``...=trace-pc,indirect-calls`` flag
|
2016-09-17 13:03:05 +08:00
|
|
|
``__sanitizer_cov_trace_pc_indirect(void *callee)`` will be inserted on every indirect call.
|
2016-09-14 09:39:49 +08:00
|
|
|
|
2016-09-17 13:03:05 +08:00
|
|
|
The functions `__sanitizer_cov_trace_pc_*` should be defined by the user.
|
2016-09-14 09:39:49 +08:00
|
|
|
|
2016-09-30 02:58:17 +08:00
|
|
|
Example:
|
|
|
|
|
|
|
|
.. code-block:: c++
|
|
|
|
|
|
|
|
// trace-pc-guard-cb.cc
|
|
|
|
#include <stdint.h>
|
|
|
|
#include <stdio.h>
|
|
|
|
#include <sanitizer/coverage_interface.h>
|
|
|
|
|
|
|
|
// This callback is inserted by the compiler as a module constructor
|
2017-02-09 05:28:39 +08:00
|
|
|
// into every DSO. 'start' and 'stop' correspond to the
|
2016-09-30 02:58:17 +08:00
|
|
|
// beginning and end of the section with the guards for the entire
|
2017-02-09 05:28:39 +08:00
|
|
|
// binary (executable or DSO). The callback will be called at least
|
|
|
|
// once per DSO and may be called multiple times with the same parameters.
|
2016-09-30 02:58:17 +08:00
|
|
|
extern "C" void __sanitizer_cov_trace_pc_guard_init(uint32_t *start,
|
|
|
|
uint32_t *stop) {
|
|
|
|
static uint64_t N; // Counter for the guards.
|
|
|
|
if (start == stop || *start) return; // Initialize only once.
|
|
|
|
printf("INIT: %p %p\n", start, stop);
|
|
|
|
for (uint32_t *x = start; x < stop; x++)
|
|
|
|
*x = ++N; // Guards should start from 1.
|
|
|
|
}
|
|
|
|
|
|
|
|
// This callback is inserted by the compiler on every edge in the
|
|
|
|
// control flow (some optimizations apply).
|
|
|
|
// Typically, the compiler will emit the code like this:
|
|
|
|
// if(*guard)
|
|
|
|
// __sanitizer_cov_trace_pc_guard(guard);
|
|
|
|
// But for large functions it will emit a simple call:
|
|
|
|
// __sanitizer_cov_trace_pc_guard(guard);
|
|
|
|
extern "C" void __sanitizer_cov_trace_pc_guard(uint32_t *guard) {
|
|
|
|
if (!*guard) return; // Duplicate the guard check.
|
|
|
|
// If you set *guard to 0 this code will not be called again for this edge.
|
|
|
|
// Now you can get the PC and do whatever you want:
|
|
|
|
// store it somewhere or symbolize it and print right away.
|
|
|
|
// The values of `*guard` are as you set them in
|
2016-09-30 03:06:09 +08:00
|
|
|
// __sanitizer_cov_trace_pc_guard_init and so you can make them consecutive
|
2016-09-30 02:58:17 +08:00
|
|
|
// and use them to dereference an array or a bit vector.
|
|
|
|
void *PC = __builtin_return_address(0);
|
|
|
|
char PcDescr[1024];
|
|
|
|
// This function is a part of the sanitizer run-time.
|
|
|
|
// To use it, link with AddressSanitizer or other sanitizer.
|
|
|
|
__sanitizer_symbolize_pc(PC, "%p %F %L", PcDescr, sizeof(PcDescr));
|
|
|
|
printf("guard: %p %x PC %s\n", guard, *guard, PcDescr);
|
|
|
|
}
|
|
|
|
|
|
|
|
.. code-block:: c++
|
|
|
|
|
|
|
|
// trace-pc-guard-example.cc
|
|
|
|
void foo() { }
|
|
|
|
int main(int argc, char **argv) {
|
|
|
|
if (argc > 1) foo();
|
|
|
|
}
|
|
|
|
|
|
|
|
.. code-block:: console
|
|
|
|
|
|
|
|
clang++ -g -fsanitize-coverage=trace-pc-guard trace-pc-guard-example.cc -c
|
|
|
|
clang++ trace-pc-guard-cb.cc trace-pc-guard-example.o -fsanitize=address
|
|
|
|
ASAN_OPTIONS=strip_path_prefix=`pwd`/ ./a.out
|
|
|
|
|
|
|
|
.. code-block:: console
|
|
|
|
|
|
|
|
INIT: 0x71bcd0 0x71bce0
|
|
|
|
guard: 0x71bcd4 2 PC 0x4ecd5b in main trace-pc-guard-example.cc:2
|
|
|
|
guard: 0x71bcd8 3 PC 0x4ecd9e in main trace-pc-guard-example.cc:3:7
|
|
|
|
|
2016-09-30 03:06:09 +08:00
|
|
|
.. code-block:: console
|
|
|
|
|
|
|
|
ASAN_OPTIONS=strip_path_prefix=`pwd`/ ./a.out with-foo
|
|
|
|
|
|
|
|
|
|
|
|
.. code-block:: console
|
|
|
|
|
|
|
|
INIT: 0x71bcd0 0x71bce0
|
|
|
|
guard: 0x71bcd4 2 PC 0x4ecd5b in main trace-pc-guard-example.cc:3
|
|
|
|
guard: 0x71bcdc 4 PC 0x4ecdc7 in main trace-pc-guard-example.cc:4:17
|
|
|
|
guard: 0x71bcd0 1 PC 0x4ecd20 in foo() trace-pc-guard-example.cc:2:14
|
|
|
|
|
2017-08-25 06:40:03 +08:00
|
|
|
Inline 8bit-counters
|
|
|
|
====================
|
|
|
|
|
|
|
|
**Experimental, may change or disappear in future**
|
|
|
|
|
|
|
|
With ``-fsanitize-coverage=inline-8bit-counters`` the compiler will insert
|
|
|
|
inline counter increments on every edge.
|
|
|
|
This is similar to ``-fsanitize-coverage=trace-pc-guard`` but instead of a
|
|
|
|
callback the instrumentation simply increments a counter.
|
|
|
|
|
|
|
|
Users need to implement a single function to capture the counters at startup.
|
|
|
|
|
|
|
|
.. code-block:: c++
|
|
|
|
|
|
|
|
extern "C"
|
|
|
|
void __sanitizer_cov_8bit_counters_init(char *start, char *end) {
|
|
|
|
// [start,end) is the array of 8-bit counters created for the current DSO.
|
|
|
|
// Capture this array in order to read/modify the counters.
|
|
|
|
}
|
|
|
|
|
2020-04-09 17:33:36 +08:00
|
|
|
|
|
|
|
Inline bool-flag
|
|
|
|
================
|
|
|
|
|
|
|
|
**Experimental, may change or disappear in future**
|
|
|
|
|
|
|
|
With ``-fsanitize-coverage=inline-bool-flag`` the compiler will insert
|
|
|
|
setting an inline boolean to true on every edge.
|
|
|
|
This is similar to ``-fsanitize-coverage=inline-8bit-counter`` but instead of
|
|
|
|
an increment of a counter, it just sets a boolean to true.
|
|
|
|
|
|
|
|
Users need to implement a single function to capture the flags at startup.
|
|
|
|
|
|
|
|
.. code-block:: c++
|
|
|
|
|
|
|
|
extern "C"
|
|
|
|
void __sanitizer_cov_bool_flag_init(bool *start, bool *end) {
|
|
|
|
// [start,end) is the array of boolean flags created for the current DSO.
|
|
|
|
// Capture this array in order to read/modify the flags.
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2017-08-25 06:40:03 +08:00
|
|
|
PC-Table
|
|
|
|
========
|
|
|
|
|
|
|
|
**Experimental, may change or disappear in future**
|
|
|
|
|
2019-02-02 01:12:35 +08:00
|
|
|
**Note:** this instrumentation might be incompatible with dead code stripping
|
|
|
|
(``-Wl,-gc-sections``) for linkers other than LLD, thus resulting in a
|
|
|
|
significant binary size overhead. For more information, see
|
|
|
|
`Bug 34636 <https://bugs.llvm.org/show_bug.cgi?id=34636>`_.
|
|
|
|
|
2017-08-25 06:40:03 +08:00
|
|
|
With ``-fsanitize-coverage=pc-table`` the compiler will create a table of
|
2020-04-09 17:33:36 +08:00
|
|
|
instrumented PCs. Requires either ``-fsanitize-coverage=inline-8bit-counters``,
|
|
|
|
or ``-fsanitize-coverage=inline-bool-flag``, or ``-fsanitize-coverage=trace-pc-guard``.
|
2017-08-25 06:40:03 +08:00
|
|
|
|
2017-08-26 03:29:47 +08:00
|
|
|
Users need to implement a single function to capture the PC table at startup:
|
2017-08-25 06:40:03 +08:00
|
|
|
|
|
|
|
.. code-block:: c++
|
|
|
|
|
|
|
|
extern "C"
|
2017-08-26 03:29:47 +08:00
|
|
|
void __sanitizer_cov_pcs_init(const uintptr_t *pcs_beg,
|
|
|
|
const uintptr_t *pcs_end) {
|
2017-08-25 06:40:03 +08:00
|
|
|
// [pcs_beg,pcs_end) is the array of ptr-sized integers representing
|
2017-08-26 03:29:47 +08:00
|
|
|
// pairs [PC,PCFlags] for every instrumented block in the current DSO.
|
|
|
|
// Capture this array in order to read the PCs and their Flags.
|
|
|
|
// The number of PCs and PCFlags for a given DSO is the same as the number
|
2020-04-09 17:33:36 +08:00
|
|
|
// of 8-bit counters (-fsanitize-coverage=inline-8bit-counters), or
|
|
|
|
// boolean flags (-fsanitize-coverage=inline=bool-flags), or trace_pc_guard
|
|
|
|
// callbacks (-fsanitize-coverage=trace-pc-guard).
|
2017-08-26 03:29:47 +08:00
|
|
|
// A PCFlags describes the basic block:
|
|
|
|
// * bit0: 1 if the block is the function entry block, 0 otherwise.
|
2017-08-25 06:40:03 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
Tracing PCs
|
|
|
|
===========
|
|
|
|
|
|
|
|
With ``-fsanitize-coverage=trace-pc`` the compiler will insert
|
|
|
|
``__sanitizer_cov_trace_pc()`` on every edge.
|
|
|
|
With an additional ``...=trace-pc,indirect-calls`` flag
|
|
|
|
``__sanitizer_cov_trace_pc_indirect(void *callee)`` will be inserted on every indirect call.
|
|
|
|
These callbacks are not implemented in the Sanitizer run-time and should be defined
|
|
|
|
by the user.
|
|
|
|
This mechanism is used for fuzzing the Linux kernel
|
|
|
|
(https://github.com/google/syzkaller).
|
|
|
|
|
|
|
|
Instrumentation points
|
|
|
|
======================
|
|
|
|
Sanitizer Coverage offers different levels of instrumentation.
|
|
|
|
|
|
|
|
* ``edge`` (default): edges are instrumented (see below).
|
|
|
|
* ``bb``: basic blocks are instrumented.
|
|
|
|
* ``func``: only the entry block of every function will be instrumented.
|
|
|
|
|
|
|
|
Use these flags together with ``trace-pc-guard`` or ``trace-pc``,
|
|
|
|
like this: ``-fsanitize-coverage=func,trace-pc-guard``.
|
|
|
|
|
|
|
|
When ``edge`` or ``bb`` is used, some of the edges/blocks may still be left
|
2017-05-06 07:28:18 +08:00
|
|
|
uninstrumented (pruned) if such instrumentation is considered redundant.
|
|
|
|
Use ``no-prune`` (e.g. ``-fsanitize-coverage=bb,no-prune,trace-pc-guard``)
|
|
|
|
to disable pruning. This could be useful for better coverage visualization.
|
2017-05-02 08:32:57 +08:00
|
|
|
|
|
|
|
|
|
|
|
Edge coverage
|
|
|
|
-------------
|
|
|
|
|
|
|
|
Consider this code:
|
|
|
|
|
|
|
|
.. code-block:: c++
|
|
|
|
|
|
|
|
void foo(int *a) {
|
|
|
|
if (a)
|
|
|
|
*a = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
It contains 3 basic blocks, let's name them A, B, C:
|
|
|
|
|
|
|
|
.. code-block:: none
|
|
|
|
|
|
|
|
A
|
|
|
|
|\
|
|
|
|
| \
|
|
|
|
| B
|
|
|
|
| /
|
|
|
|
|/
|
|
|
|
C
|
|
|
|
|
|
|
|
If blocks A, B, and C are all covered we know for certain that the edges A=>B
|
|
|
|
and B=>C were executed, but we still don't know if the edge A=>C was executed.
|
|
|
|
Such edges of control flow graph are called
|
2019-01-24 04:39:07 +08:00
|
|
|
`critical <https://en.wikipedia.org/wiki/Control_flow_graph#Special_edges>`_.
|
|
|
|
The edge-level coverage simply splits all critical edges by introducing new
|
|
|
|
dummy blocks and then instruments those blocks:
|
2017-05-02 08:32:57 +08:00
|
|
|
|
|
|
|
.. code-block:: none
|
|
|
|
|
|
|
|
A
|
|
|
|
|\
|
|
|
|
| \
|
|
|
|
D B
|
|
|
|
| /
|
|
|
|
|/
|
|
|
|
C
|
2016-09-30 02:58:17 +08:00
|
|
|
|
2015-08-01 05:48:10 +08:00
|
|
|
Tracing data flow
|
|
|
|
=================
|
|
|
|
|
2016-08-30 09:27:03 +08:00
|
|
|
Support for data-flow-guided fuzzing.
|
2015-08-01 05:48:10 +08:00
|
|
|
With ``-fsanitize-coverage=trace-cmp`` the compiler will insert extra instrumentation
|
|
|
|
around comparison instructions and switch statements.
|
2016-08-30 09:27:03 +08:00
|
|
|
Similarly, with ``-fsanitize-coverage=trace-div`` the compiler will instrument
|
|
|
|
integer division instructions (to capture the right argument of division)
|
|
|
|
and with ``-fsanitize-coverage=trace-gep`` --
|
2018-11-05 01:02:00 +08:00
|
|
|
the `LLVM GEP instructions <https://llvm.org/docs/GetElementPtr.html>`_
|
2016-08-30 09:27:03 +08:00
|
|
|
(to capture array indices).
|
2015-08-01 05:48:10 +08:00
|
|
|
|
2019-02-01 07:43:00 +08:00
|
|
|
Unless ``no-prune`` option is provided, some of the comparison instructions
|
|
|
|
will not be instrumented.
|
|
|
|
|
2015-08-01 05:48:10 +08:00
|
|
|
.. code-block:: c++
|
|
|
|
|
|
|
|
// Called before a comparison instruction.
|
|
|
|
// Arg1 and Arg2 are arguments of the comparison.
|
2016-08-18 09:26:36 +08:00
|
|
|
void __sanitizer_cov_trace_cmp1(uint8_t Arg1, uint8_t Arg2);
|
|
|
|
void __sanitizer_cov_trace_cmp2(uint16_t Arg1, uint16_t Arg2);
|
|
|
|
void __sanitizer_cov_trace_cmp4(uint32_t Arg1, uint32_t Arg2);
|
|
|
|
void __sanitizer_cov_trace_cmp8(uint64_t Arg1, uint64_t Arg2);
|
2015-08-01 05:48:10 +08:00
|
|
|
|
2017-08-10 23:00:13 +08:00
|
|
|
// Called before a comparison instruction if exactly one of the arguments is constant.
|
|
|
|
// Arg1 and Arg2 are arguments of the comparison, Arg1 is a compile-time constant.
|
|
|
|
// These callbacks are emitted by -fsanitize-coverage=trace-cmp since 2017-08-11
|
|
|
|
void __sanitizer_cov_trace_const_cmp1(uint8_t Arg1, uint8_t Arg2);
|
|
|
|
void __sanitizer_cov_trace_const_cmp2(uint16_t Arg1, uint16_t Arg2);
|
|
|
|
void __sanitizer_cov_trace_const_cmp4(uint32_t Arg1, uint32_t Arg2);
|
|
|
|
void __sanitizer_cov_trace_const_cmp8(uint64_t Arg1, uint64_t Arg2);
|
|
|
|
|
2015-08-01 05:48:10 +08:00
|
|
|
// Called before a switch statement.
|
|
|
|
// Val is the switch operand.
|
|
|
|
// Cases[0] is the number of case constants.
|
|
|
|
// Cases[1] is the size of Val in bits.
|
|
|
|
// Cases[2:] are the case constants.
|
|
|
|
void __sanitizer_cov_trace_switch(uint64_t Val, uint64_t *Cases);
|
|
|
|
|
2016-08-30 09:27:03 +08:00
|
|
|
// Called before a division statement.
|
|
|
|
// Val is the second argument of division.
|
|
|
|
void __sanitizer_cov_trace_div4(uint32_t Val);
|
|
|
|
void __sanitizer_cov_trace_div8(uint64_t Val);
|
|
|
|
|
|
|
|
// Called before a GetElemementPtr (GEP) instruction
|
|
|
|
// for every non-constant array index.
|
|
|
|
void __sanitizer_cov_trace_gep(uintptr_t Idx);
|
|
|
|
|
Implement `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist` for clang
Summary:
This commit adds two command-line options to clang.
These options let the user decide which functions will receive SanitizerCoverage instrumentation.
This is most useful in the libFuzzer use case, where it enables targeted coverage-guided fuzzing.
Patch by Yannis Juglaret of DGA-MI, Rennes, France
libFuzzer tests its target against an evolving corpus, and relies on SanitizerCoverage instrumentation to collect the code coverage information that drives corpus evolution. Currently, libFuzzer collects such information for all functions of the target under test, and adds to the corpus every mutated sample that finds a new code coverage path in any function of the target. We propose instead to let the user specify which functions' code coverage information is relevant for building the upcoming fuzzing campaign's corpus. To this end, we add two new command line options for clang, enabling targeted coverage-guided fuzzing with libFuzzer. We see targeted coverage guided fuzzing as a simple way to leverage libFuzzer for big targets with thousands of functions or multiple dependencies. We publish this patch as work from DGA-MI of Rennes, France, with proper authorization from the hierarchy.
Targeted coverage-guided fuzzing can accelerate bug finding for two reasons. First, the compiler will avoid costly instrumentation for non-relevant functions, accelerating fuzzer execution for each call to any of these functions. Second, the built fuzzer will produce and use a more accurate corpus, because it will not keep the samples that find new coverage paths in non-relevant functions.
The two new command line options are `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist`. They accept files in the same format as the existing `-fsanitize-blacklist` option <https://clang.llvm.org/docs/SanitizerSpecialCaseList.html#format>. The new options influence SanitizerCoverage so that it will only instrument a subset of the functions in the target. We explain these options in detail in `clang/docs/SanitizerCoverage.rst`.
Consider now the woff2 fuzzing example from the libFuzzer tutorial <https://github.com/google/fuzzer-test-suite/blob/master/tutorial/libFuzzerTutorial.md>. We are aware that we cannot conclude much from this example because mutating compressed data is generally a bad idea, but let us use it anyway as an illustration for its simplicity. Let us use an empty blacklist together with one of the three following whitelists:
```
# (a)
src:*
fun:*
# (b)
src:SRC/*
fun:*
# (c)
src:SRC/src/woff2_dec.cc
fun:*
```
Running the built fuzzers shows how many instrumentation points the compiler adds, the fuzzer will output //XXX PCs//. Whitelist (a) is the instrument-everything whitelist, it produces 11912 instrumentation points. Whitelist (b) focuses coverage to instrument woff2 source code only, ignoring the dependency code for brotli (de)compression; it produces 3984 instrumented instrumentation points. Whitelist (c) focuses coverage to only instrument functions in the main file that deals with WOFF2 to TTF conversion, resulting in 1056 instrumentation points.
For experimentation purposes, we ran each fuzzer approximately 100 times, single process, with the initial corpus provided in the tutorial. We let the fuzzer run until it either found the heap buffer overflow or went out of memory. On this simple example, whitelists (b) and (c) found the heap buffer overflow more reliably and 5x faster than whitelist (a). The average execution times when finding the heap buffer overflow were as follows: (a) 904 s, (b) 156 s, and (c) 176 s.
We explain these results by the fact that WOFF2 to TTF conversion calls the brotli decompression algorithm's functions, which are mostly irrelevant for finding bugs in WOFF2 font reconstruction but nevertheless instrumented and used by whitelist (a) to guide fuzzing. This results in longer execution time for these functions and a partially irrelevant corpus. Contrary to whitelist (a), whitelists (b) and (c) will execute brotli-related functions without instrumentation overhead, and ignore new code paths found in them. This results in faster bug finding for WOFF2 font reconstruction.
The results for whitelist (b) are similar to the ones for whitelist (c). Indeed, WOFF2 to TTF conversion calls functions that are mostly located in SRC/src/woff2_dec.cc. The 2892 extra instrumentation points allowed by whitelist (b) do not tamper with bug finding, even though they are mostly irrelevant, simply because most of these functions do not get called. We get a slightly faster average time for bug finding with whitelist (b), which might indicate that some of the extra instrumentation points are actually relevant, or might just be random noise.
Reviewers: kcc, morehouse, vitalybuka
Reviewed By: morehouse, vitalybuka
Subscribers: pratyai, vitalybuka, eternalsakura, xwlin222, dende, srhines, kubamracek, #sanitizers, lebedev.ri, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D63616
2020-04-11 01:42:41 +08:00
|
|
|
Partially disabling instrumentation
|
|
|
|
===================================
|
|
|
|
|
|
|
|
It is sometimes useful to tell SanitizerCoverage to instrument only a subset of the
|
|
|
|
functions in your target.
|
2020-06-20 13:22:47 +08:00
|
|
|
With ``-fsanitize-coverage-allowlist=allowlist.txt``
|
|
|
|
and ``-fsanitize-coverage-blocklist=blocklist.txt``,
|
|
|
|
you can specify such a subset through the combination of a allowlist and a blocklist.
|
Implement `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist` for clang
Summary:
This commit adds two command-line options to clang.
These options let the user decide which functions will receive SanitizerCoverage instrumentation.
This is most useful in the libFuzzer use case, where it enables targeted coverage-guided fuzzing.
Patch by Yannis Juglaret of DGA-MI, Rennes, France
libFuzzer tests its target against an evolving corpus, and relies on SanitizerCoverage instrumentation to collect the code coverage information that drives corpus evolution. Currently, libFuzzer collects such information for all functions of the target under test, and adds to the corpus every mutated sample that finds a new code coverage path in any function of the target. We propose instead to let the user specify which functions' code coverage information is relevant for building the upcoming fuzzing campaign's corpus. To this end, we add two new command line options for clang, enabling targeted coverage-guided fuzzing with libFuzzer. We see targeted coverage guided fuzzing as a simple way to leverage libFuzzer for big targets with thousands of functions or multiple dependencies. We publish this patch as work from DGA-MI of Rennes, France, with proper authorization from the hierarchy.
Targeted coverage-guided fuzzing can accelerate bug finding for two reasons. First, the compiler will avoid costly instrumentation for non-relevant functions, accelerating fuzzer execution for each call to any of these functions. Second, the built fuzzer will produce and use a more accurate corpus, because it will not keep the samples that find new coverage paths in non-relevant functions.
The two new command line options are `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist`. They accept files in the same format as the existing `-fsanitize-blacklist` option <https://clang.llvm.org/docs/SanitizerSpecialCaseList.html#format>. The new options influence SanitizerCoverage so that it will only instrument a subset of the functions in the target. We explain these options in detail in `clang/docs/SanitizerCoverage.rst`.
Consider now the woff2 fuzzing example from the libFuzzer tutorial <https://github.com/google/fuzzer-test-suite/blob/master/tutorial/libFuzzerTutorial.md>. We are aware that we cannot conclude much from this example because mutating compressed data is generally a bad idea, but let us use it anyway as an illustration for its simplicity. Let us use an empty blacklist together with one of the three following whitelists:
```
# (a)
src:*
fun:*
# (b)
src:SRC/*
fun:*
# (c)
src:SRC/src/woff2_dec.cc
fun:*
```
Running the built fuzzers shows how many instrumentation points the compiler adds, the fuzzer will output //XXX PCs//. Whitelist (a) is the instrument-everything whitelist, it produces 11912 instrumentation points. Whitelist (b) focuses coverage to instrument woff2 source code only, ignoring the dependency code for brotli (de)compression; it produces 3984 instrumented instrumentation points. Whitelist (c) focuses coverage to only instrument functions in the main file that deals with WOFF2 to TTF conversion, resulting in 1056 instrumentation points.
For experimentation purposes, we ran each fuzzer approximately 100 times, single process, with the initial corpus provided in the tutorial. We let the fuzzer run until it either found the heap buffer overflow or went out of memory. On this simple example, whitelists (b) and (c) found the heap buffer overflow more reliably and 5x faster than whitelist (a). The average execution times when finding the heap buffer overflow were as follows: (a) 904 s, (b) 156 s, and (c) 176 s.
We explain these results by the fact that WOFF2 to TTF conversion calls the brotli decompression algorithm's functions, which are mostly irrelevant for finding bugs in WOFF2 font reconstruction but nevertheless instrumented and used by whitelist (a) to guide fuzzing. This results in longer execution time for these functions and a partially irrelevant corpus. Contrary to whitelist (a), whitelists (b) and (c) will execute brotli-related functions without instrumentation overhead, and ignore new code paths found in them. This results in faster bug finding for WOFF2 font reconstruction.
The results for whitelist (b) are similar to the ones for whitelist (c). Indeed, WOFF2 to TTF conversion calls functions that are mostly located in SRC/src/woff2_dec.cc. The 2892 extra instrumentation points allowed by whitelist (b) do not tamper with bug finding, even though they are mostly irrelevant, simply because most of these functions do not get called. We get a slightly faster average time for bug finding with whitelist (b), which might indicate that some of the extra instrumentation points are actually relevant, or might just be random noise.
Reviewers: kcc, morehouse, vitalybuka
Reviewed By: morehouse, vitalybuka
Subscribers: pratyai, vitalybuka, eternalsakura, xwlin222, dende, srhines, kubamracek, #sanitizers, lebedev.ri, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D63616
2020-04-11 01:42:41 +08:00
|
|
|
|
|
|
|
SanitizerCoverage will only instrument functions that satisfy two conditions.
|
2020-06-20 13:22:47 +08:00
|
|
|
First, the function should belong to a source file with a path that is both allowlisted
|
|
|
|
and not blocklisted.
|
|
|
|
Second, the function should have a mangled name that is both allowlisted and not blocklisted.
|
Implement `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist` for clang
Summary:
This commit adds two command-line options to clang.
These options let the user decide which functions will receive SanitizerCoverage instrumentation.
This is most useful in the libFuzzer use case, where it enables targeted coverage-guided fuzzing.
Patch by Yannis Juglaret of DGA-MI, Rennes, France
libFuzzer tests its target against an evolving corpus, and relies on SanitizerCoverage instrumentation to collect the code coverage information that drives corpus evolution. Currently, libFuzzer collects such information for all functions of the target under test, and adds to the corpus every mutated sample that finds a new code coverage path in any function of the target. We propose instead to let the user specify which functions' code coverage information is relevant for building the upcoming fuzzing campaign's corpus. To this end, we add two new command line options for clang, enabling targeted coverage-guided fuzzing with libFuzzer. We see targeted coverage guided fuzzing as a simple way to leverage libFuzzer for big targets with thousands of functions or multiple dependencies. We publish this patch as work from DGA-MI of Rennes, France, with proper authorization from the hierarchy.
Targeted coverage-guided fuzzing can accelerate bug finding for two reasons. First, the compiler will avoid costly instrumentation for non-relevant functions, accelerating fuzzer execution for each call to any of these functions. Second, the built fuzzer will produce and use a more accurate corpus, because it will not keep the samples that find new coverage paths in non-relevant functions.
The two new command line options are `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist`. They accept files in the same format as the existing `-fsanitize-blacklist` option <https://clang.llvm.org/docs/SanitizerSpecialCaseList.html#format>. The new options influence SanitizerCoverage so that it will only instrument a subset of the functions in the target. We explain these options in detail in `clang/docs/SanitizerCoverage.rst`.
Consider now the woff2 fuzzing example from the libFuzzer tutorial <https://github.com/google/fuzzer-test-suite/blob/master/tutorial/libFuzzerTutorial.md>. We are aware that we cannot conclude much from this example because mutating compressed data is generally a bad idea, but let us use it anyway as an illustration for its simplicity. Let us use an empty blacklist together with one of the three following whitelists:
```
# (a)
src:*
fun:*
# (b)
src:SRC/*
fun:*
# (c)
src:SRC/src/woff2_dec.cc
fun:*
```
Running the built fuzzers shows how many instrumentation points the compiler adds, the fuzzer will output //XXX PCs//. Whitelist (a) is the instrument-everything whitelist, it produces 11912 instrumentation points. Whitelist (b) focuses coverage to instrument woff2 source code only, ignoring the dependency code for brotli (de)compression; it produces 3984 instrumented instrumentation points. Whitelist (c) focuses coverage to only instrument functions in the main file that deals with WOFF2 to TTF conversion, resulting in 1056 instrumentation points.
For experimentation purposes, we ran each fuzzer approximately 100 times, single process, with the initial corpus provided in the tutorial. We let the fuzzer run until it either found the heap buffer overflow or went out of memory. On this simple example, whitelists (b) and (c) found the heap buffer overflow more reliably and 5x faster than whitelist (a). The average execution times when finding the heap buffer overflow were as follows: (a) 904 s, (b) 156 s, and (c) 176 s.
We explain these results by the fact that WOFF2 to TTF conversion calls the brotli decompression algorithm's functions, which are mostly irrelevant for finding bugs in WOFF2 font reconstruction but nevertheless instrumented and used by whitelist (a) to guide fuzzing. This results in longer execution time for these functions and a partially irrelevant corpus. Contrary to whitelist (a), whitelists (b) and (c) will execute brotli-related functions without instrumentation overhead, and ignore new code paths found in them. This results in faster bug finding for WOFF2 font reconstruction.
The results for whitelist (b) are similar to the ones for whitelist (c). Indeed, WOFF2 to TTF conversion calls functions that are mostly located in SRC/src/woff2_dec.cc. The 2892 extra instrumentation points allowed by whitelist (b) do not tamper with bug finding, even though they are mostly irrelevant, simply because most of these functions do not get called. We get a slightly faster average time for bug finding with whitelist (b), which might indicate that some of the extra instrumentation points are actually relevant, or might just be random noise.
Reviewers: kcc, morehouse, vitalybuka
Reviewed By: morehouse, vitalybuka
Subscribers: pratyai, vitalybuka, eternalsakura, xwlin222, dende, srhines, kubamracek, #sanitizers, lebedev.ri, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D63616
2020-04-11 01:42:41 +08:00
|
|
|
|
2020-06-20 13:22:47 +08:00
|
|
|
The allowlist and blocklist format is similar to that of the sanitizer blocklist format.
|
|
|
|
The default allowlist will match every source file and every function.
|
|
|
|
The default blocklist will match no source file and no function.
|
Implement `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist` for clang
Summary:
This commit adds two command-line options to clang.
These options let the user decide which functions will receive SanitizerCoverage instrumentation.
This is most useful in the libFuzzer use case, where it enables targeted coverage-guided fuzzing.
Patch by Yannis Juglaret of DGA-MI, Rennes, France
libFuzzer tests its target against an evolving corpus, and relies on SanitizerCoverage instrumentation to collect the code coverage information that drives corpus evolution. Currently, libFuzzer collects such information for all functions of the target under test, and adds to the corpus every mutated sample that finds a new code coverage path in any function of the target. We propose instead to let the user specify which functions' code coverage information is relevant for building the upcoming fuzzing campaign's corpus. To this end, we add two new command line options for clang, enabling targeted coverage-guided fuzzing with libFuzzer. We see targeted coverage guided fuzzing as a simple way to leverage libFuzzer for big targets with thousands of functions or multiple dependencies. We publish this patch as work from DGA-MI of Rennes, France, with proper authorization from the hierarchy.
Targeted coverage-guided fuzzing can accelerate bug finding for two reasons. First, the compiler will avoid costly instrumentation for non-relevant functions, accelerating fuzzer execution for each call to any of these functions. Second, the built fuzzer will produce and use a more accurate corpus, because it will not keep the samples that find new coverage paths in non-relevant functions.
The two new command line options are `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist`. They accept files in the same format as the existing `-fsanitize-blacklist` option <https://clang.llvm.org/docs/SanitizerSpecialCaseList.html#format>. The new options influence SanitizerCoverage so that it will only instrument a subset of the functions in the target. We explain these options in detail in `clang/docs/SanitizerCoverage.rst`.
Consider now the woff2 fuzzing example from the libFuzzer tutorial <https://github.com/google/fuzzer-test-suite/blob/master/tutorial/libFuzzerTutorial.md>. We are aware that we cannot conclude much from this example because mutating compressed data is generally a bad idea, but let us use it anyway as an illustration for its simplicity. Let us use an empty blacklist together with one of the three following whitelists:
```
# (a)
src:*
fun:*
# (b)
src:SRC/*
fun:*
# (c)
src:SRC/src/woff2_dec.cc
fun:*
```
Running the built fuzzers shows how many instrumentation points the compiler adds, the fuzzer will output //XXX PCs//. Whitelist (a) is the instrument-everything whitelist, it produces 11912 instrumentation points. Whitelist (b) focuses coverage to instrument woff2 source code only, ignoring the dependency code for brotli (de)compression; it produces 3984 instrumented instrumentation points. Whitelist (c) focuses coverage to only instrument functions in the main file that deals with WOFF2 to TTF conversion, resulting in 1056 instrumentation points.
For experimentation purposes, we ran each fuzzer approximately 100 times, single process, with the initial corpus provided in the tutorial. We let the fuzzer run until it either found the heap buffer overflow or went out of memory. On this simple example, whitelists (b) and (c) found the heap buffer overflow more reliably and 5x faster than whitelist (a). The average execution times when finding the heap buffer overflow were as follows: (a) 904 s, (b) 156 s, and (c) 176 s.
We explain these results by the fact that WOFF2 to TTF conversion calls the brotli decompression algorithm's functions, which are mostly irrelevant for finding bugs in WOFF2 font reconstruction but nevertheless instrumented and used by whitelist (a) to guide fuzzing. This results in longer execution time for these functions and a partially irrelevant corpus. Contrary to whitelist (a), whitelists (b) and (c) will execute brotli-related functions without instrumentation overhead, and ignore new code paths found in them. This results in faster bug finding for WOFF2 font reconstruction.
The results for whitelist (b) are similar to the ones for whitelist (c). Indeed, WOFF2 to TTF conversion calls functions that are mostly located in SRC/src/woff2_dec.cc. The 2892 extra instrumentation points allowed by whitelist (b) do not tamper with bug finding, even though they are mostly irrelevant, simply because most of these functions do not get called. We get a slightly faster average time for bug finding with whitelist (b), which might indicate that some of the extra instrumentation points are actually relevant, or might just be random noise.
Reviewers: kcc, morehouse, vitalybuka
Reviewed By: morehouse, vitalybuka
Subscribers: pratyai, vitalybuka, eternalsakura, xwlin222, dende, srhines, kubamracek, #sanitizers, lebedev.ri, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D63616
2020-04-11 01:42:41 +08:00
|
|
|
|
2020-06-20 13:22:47 +08:00
|
|
|
A common use case is to have the allowlist list folders or source files for which you want
|
|
|
|
instrumentation and allow all function names, while the blocklist will opt out some specific
|
|
|
|
files or functions that the allowlist loosely allowed.
|
Implement `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist` for clang
Summary:
This commit adds two command-line options to clang.
These options let the user decide which functions will receive SanitizerCoverage instrumentation.
This is most useful in the libFuzzer use case, where it enables targeted coverage-guided fuzzing.
Patch by Yannis Juglaret of DGA-MI, Rennes, France
libFuzzer tests its target against an evolving corpus, and relies on SanitizerCoverage instrumentation to collect the code coverage information that drives corpus evolution. Currently, libFuzzer collects such information for all functions of the target under test, and adds to the corpus every mutated sample that finds a new code coverage path in any function of the target. We propose instead to let the user specify which functions' code coverage information is relevant for building the upcoming fuzzing campaign's corpus. To this end, we add two new command line options for clang, enabling targeted coverage-guided fuzzing with libFuzzer. We see targeted coverage guided fuzzing as a simple way to leverage libFuzzer for big targets with thousands of functions or multiple dependencies. We publish this patch as work from DGA-MI of Rennes, France, with proper authorization from the hierarchy.
Targeted coverage-guided fuzzing can accelerate bug finding for two reasons. First, the compiler will avoid costly instrumentation for non-relevant functions, accelerating fuzzer execution for each call to any of these functions. Second, the built fuzzer will produce and use a more accurate corpus, because it will not keep the samples that find new coverage paths in non-relevant functions.
The two new command line options are `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist`. They accept files in the same format as the existing `-fsanitize-blacklist` option <https://clang.llvm.org/docs/SanitizerSpecialCaseList.html#format>. The new options influence SanitizerCoverage so that it will only instrument a subset of the functions in the target. We explain these options in detail in `clang/docs/SanitizerCoverage.rst`.
Consider now the woff2 fuzzing example from the libFuzzer tutorial <https://github.com/google/fuzzer-test-suite/blob/master/tutorial/libFuzzerTutorial.md>. We are aware that we cannot conclude much from this example because mutating compressed data is generally a bad idea, but let us use it anyway as an illustration for its simplicity. Let us use an empty blacklist together with one of the three following whitelists:
```
# (a)
src:*
fun:*
# (b)
src:SRC/*
fun:*
# (c)
src:SRC/src/woff2_dec.cc
fun:*
```
Running the built fuzzers shows how many instrumentation points the compiler adds, the fuzzer will output //XXX PCs//. Whitelist (a) is the instrument-everything whitelist, it produces 11912 instrumentation points. Whitelist (b) focuses coverage to instrument woff2 source code only, ignoring the dependency code for brotli (de)compression; it produces 3984 instrumented instrumentation points. Whitelist (c) focuses coverage to only instrument functions in the main file that deals with WOFF2 to TTF conversion, resulting in 1056 instrumentation points.
For experimentation purposes, we ran each fuzzer approximately 100 times, single process, with the initial corpus provided in the tutorial. We let the fuzzer run until it either found the heap buffer overflow or went out of memory. On this simple example, whitelists (b) and (c) found the heap buffer overflow more reliably and 5x faster than whitelist (a). The average execution times when finding the heap buffer overflow were as follows: (a) 904 s, (b) 156 s, and (c) 176 s.
We explain these results by the fact that WOFF2 to TTF conversion calls the brotli decompression algorithm's functions, which are mostly irrelevant for finding bugs in WOFF2 font reconstruction but nevertheless instrumented and used by whitelist (a) to guide fuzzing. This results in longer execution time for these functions and a partially irrelevant corpus. Contrary to whitelist (a), whitelists (b) and (c) will execute brotli-related functions without instrumentation overhead, and ignore new code paths found in them. This results in faster bug finding for WOFF2 font reconstruction.
The results for whitelist (b) are similar to the ones for whitelist (c). Indeed, WOFF2 to TTF conversion calls functions that are mostly located in SRC/src/woff2_dec.cc. The 2892 extra instrumentation points allowed by whitelist (b) do not tamper with bug finding, even though they are mostly irrelevant, simply because most of these functions do not get called. We get a slightly faster average time for bug finding with whitelist (b), which might indicate that some of the extra instrumentation points are actually relevant, or might just be random noise.
Reviewers: kcc, morehouse, vitalybuka
Reviewed By: morehouse, vitalybuka
Subscribers: pratyai, vitalybuka, eternalsakura, xwlin222, dende, srhines, kubamracek, #sanitizers, lebedev.ri, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D63616
2020-04-11 01:42:41 +08:00
|
|
|
|
2020-06-20 13:22:47 +08:00
|
|
|
Here is an example allowlist:
|
Implement `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist` for clang
Summary:
This commit adds two command-line options to clang.
These options let the user decide which functions will receive SanitizerCoverage instrumentation.
This is most useful in the libFuzzer use case, where it enables targeted coverage-guided fuzzing.
Patch by Yannis Juglaret of DGA-MI, Rennes, France
libFuzzer tests its target against an evolving corpus, and relies on SanitizerCoverage instrumentation to collect the code coverage information that drives corpus evolution. Currently, libFuzzer collects such information for all functions of the target under test, and adds to the corpus every mutated sample that finds a new code coverage path in any function of the target. We propose instead to let the user specify which functions' code coverage information is relevant for building the upcoming fuzzing campaign's corpus. To this end, we add two new command line options for clang, enabling targeted coverage-guided fuzzing with libFuzzer. We see targeted coverage guided fuzzing as a simple way to leverage libFuzzer for big targets with thousands of functions or multiple dependencies. We publish this patch as work from DGA-MI of Rennes, France, with proper authorization from the hierarchy.
Targeted coverage-guided fuzzing can accelerate bug finding for two reasons. First, the compiler will avoid costly instrumentation for non-relevant functions, accelerating fuzzer execution for each call to any of these functions. Second, the built fuzzer will produce and use a more accurate corpus, because it will not keep the samples that find new coverage paths in non-relevant functions.
The two new command line options are `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist`. They accept files in the same format as the existing `-fsanitize-blacklist` option <https://clang.llvm.org/docs/SanitizerSpecialCaseList.html#format>. The new options influence SanitizerCoverage so that it will only instrument a subset of the functions in the target. We explain these options in detail in `clang/docs/SanitizerCoverage.rst`.
Consider now the woff2 fuzzing example from the libFuzzer tutorial <https://github.com/google/fuzzer-test-suite/blob/master/tutorial/libFuzzerTutorial.md>. We are aware that we cannot conclude much from this example because mutating compressed data is generally a bad idea, but let us use it anyway as an illustration for its simplicity. Let us use an empty blacklist together with one of the three following whitelists:
```
# (a)
src:*
fun:*
# (b)
src:SRC/*
fun:*
# (c)
src:SRC/src/woff2_dec.cc
fun:*
```
Running the built fuzzers shows how many instrumentation points the compiler adds, the fuzzer will output //XXX PCs//. Whitelist (a) is the instrument-everything whitelist, it produces 11912 instrumentation points. Whitelist (b) focuses coverage to instrument woff2 source code only, ignoring the dependency code for brotli (de)compression; it produces 3984 instrumented instrumentation points. Whitelist (c) focuses coverage to only instrument functions in the main file that deals with WOFF2 to TTF conversion, resulting in 1056 instrumentation points.
For experimentation purposes, we ran each fuzzer approximately 100 times, single process, with the initial corpus provided in the tutorial. We let the fuzzer run until it either found the heap buffer overflow or went out of memory. On this simple example, whitelists (b) and (c) found the heap buffer overflow more reliably and 5x faster than whitelist (a). The average execution times when finding the heap buffer overflow were as follows: (a) 904 s, (b) 156 s, and (c) 176 s.
We explain these results by the fact that WOFF2 to TTF conversion calls the brotli decompression algorithm's functions, which are mostly irrelevant for finding bugs in WOFF2 font reconstruction but nevertheless instrumented and used by whitelist (a) to guide fuzzing. This results in longer execution time for these functions and a partially irrelevant corpus. Contrary to whitelist (a), whitelists (b) and (c) will execute brotli-related functions without instrumentation overhead, and ignore new code paths found in them. This results in faster bug finding for WOFF2 font reconstruction.
The results for whitelist (b) are similar to the ones for whitelist (c). Indeed, WOFF2 to TTF conversion calls functions that are mostly located in SRC/src/woff2_dec.cc. The 2892 extra instrumentation points allowed by whitelist (b) do not tamper with bug finding, even though they are mostly irrelevant, simply because most of these functions do not get called. We get a slightly faster average time for bug finding with whitelist (b), which might indicate that some of the extra instrumentation points are actually relevant, or might just be random noise.
Reviewers: kcc, morehouse, vitalybuka
Reviewed By: morehouse, vitalybuka
Subscribers: pratyai, vitalybuka, eternalsakura, xwlin222, dende, srhines, kubamracek, #sanitizers, lebedev.ri, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D63616
2020-04-11 01:42:41 +08:00
|
|
|
|
|
|
|
.. code-block:: none
|
|
|
|
|
|
|
|
# Enable instrumentation for a whole folder
|
|
|
|
src:bar/*
|
|
|
|
# Enable instrumentation for a specific source file
|
|
|
|
src:foo/a.cpp
|
|
|
|
# Enable instrumentation for all functions in those files
|
|
|
|
fun:*
|
|
|
|
|
2020-06-20 13:22:47 +08:00
|
|
|
And an example blocklist:
|
Implement `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist` for clang
Summary:
This commit adds two command-line options to clang.
These options let the user decide which functions will receive SanitizerCoverage instrumentation.
This is most useful in the libFuzzer use case, where it enables targeted coverage-guided fuzzing.
Patch by Yannis Juglaret of DGA-MI, Rennes, France
libFuzzer tests its target against an evolving corpus, and relies on SanitizerCoverage instrumentation to collect the code coverage information that drives corpus evolution. Currently, libFuzzer collects such information for all functions of the target under test, and adds to the corpus every mutated sample that finds a new code coverage path in any function of the target. We propose instead to let the user specify which functions' code coverage information is relevant for building the upcoming fuzzing campaign's corpus. To this end, we add two new command line options for clang, enabling targeted coverage-guided fuzzing with libFuzzer. We see targeted coverage guided fuzzing as a simple way to leverage libFuzzer for big targets with thousands of functions or multiple dependencies. We publish this patch as work from DGA-MI of Rennes, France, with proper authorization from the hierarchy.
Targeted coverage-guided fuzzing can accelerate bug finding for two reasons. First, the compiler will avoid costly instrumentation for non-relevant functions, accelerating fuzzer execution for each call to any of these functions. Second, the built fuzzer will produce and use a more accurate corpus, because it will not keep the samples that find new coverage paths in non-relevant functions.
The two new command line options are `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist`. They accept files in the same format as the existing `-fsanitize-blacklist` option <https://clang.llvm.org/docs/SanitizerSpecialCaseList.html#format>. The new options influence SanitizerCoverage so that it will only instrument a subset of the functions in the target. We explain these options in detail in `clang/docs/SanitizerCoverage.rst`.
Consider now the woff2 fuzzing example from the libFuzzer tutorial <https://github.com/google/fuzzer-test-suite/blob/master/tutorial/libFuzzerTutorial.md>. We are aware that we cannot conclude much from this example because mutating compressed data is generally a bad idea, but let us use it anyway as an illustration for its simplicity. Let us use an empty blacklist together with one of the three following whitelists:
```
# (a)
src:*
fun:*
# (b)
src:SRC/*
fun:*
# (c)
src:SRC/src/woff2_dec.cc
fun:*
```
Running the built fuzzers shows how many instrumentation points the compiler adds, the fuzzer will output //XXX PCs//. Whitelist (a) is the instrument-everything whitelist, it produces 11912 instrumentation points. Whitelist (b) focuses coverage to instrument woff2 source code only, ignoring the dependency code for brotli (de)compression; it produces 3984 instrumented instrumentation points. Whitelist (c) focuses coverage to only instrument functions in the main file that deals with WOFF2 to TTF conversion, resulting in 1056 instrumentation points.
For experimentation purposes, we ran each fuzzer approximately 100 times, single process, with the initial corpus provided in the tutorial. We let the fuzzer run until it either found the heap buffer overflow or went out of memory. On this simple example, whitelists (b) and (c) found the heap buffer overflow more reliably and 5x faster than whitelist (a). The average execution times when finding the heap buffer overflow were as follows: (a) 904 s, (b) 156 s, and (c) 176 s.
We explain these results by the fact that WOFF2 to TTF conversion calls the brotli decompression algorithm's functions, which are mostly irrelevant for finding bugs in WOFF2 font reconstruction but nevertheless instrumented and used by whitelist (a) to guide fuzzing. This results in longer execution time for these functions and a partially irrelevant corpus. Contrary to whitelist (a), whitelists (b) and (c) will execute brotli-related functions without instrumentation overhead, and ignore new code paths found in them. This results in faster bug finding for WOFF2 font reconstruction.
The results for whitelist (b) are similar to the ones for whitelist (c). Indeed, WOFF2 to TTF conversion calls functions that are mostly located in SRC/src/woff2_dec.cc. The 2892 extra instrumentation points allowed by whitelist (b) do not tamper with bug finding, even though they are mostly irrelevant, simply because most of these functions do not get called. We get a slightly faster average time for bug finding with whitelist (b), which might indicate that some of the extra instrumentation points are actually relevant, or might just be random noise.
Reviewers: kcc, morehouse, vitalybuka
Reviewed By: morehouse, vitalybuka
Subscribers: pratyai, vitalybuka, eternalsakura, xwlin222, dende, srhines, kubamracek, #sanitizers, lebedev.ri, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D63616
2020-04-11 01:42:41 +08:00
|
|
|
|
|
|
|
.. code-block:: none
|
|
|
|
|
2020-06-20 13:22:47 +08:00
|
|
|
# Disable instrumentation for a specific source file that the allowlist allowed
|
Implement `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist` for clang
Summary:
This commit adds two command-line options to clang.
These options let the user decide which functions will receive SanitizerCoverage instrumentation.
This is most useful in the libFuzzer use case, where it enables targeted coverage-guided fuzzing.
Patch by Yannis Juglaret of DGA-MI, Rennes, France
libFuzzer tests its target against an evolving corpus, and relies on SanitizerCoverage instrumentation to collect the code coverage information that drives corpus evolution. Currently, libFuzzer collects such information for all functions of the target under test, and adds to the corpus every mutated sample that finds a new code coverage path in any function of the target. We propose instead to let the user specify which functions' code coverage information is relevant for building the upcoming fuzzing campaign's corpus. To this end, we add two new command line options for clang, enabling targeted coverage-guided fuzzing with libFuzzer. We see targeted coverage guided fuzzing as a simple way to leverage libFuzzer for big targets with thousands of functions or multiple dependencies. We publish this patch as work from DGA-MI of Rennes, France, with proper authorization from the hierarchy.
Targeted coverage-guided fuzzing can accelerate bug finding for two reasons. First, the compiler will avoid costly instrumentation for non-relevant functions, accelerating fuzzer execution for each call to any of these functions. Second, the built fuzzer will produce and use a more accurate corpus, because it will not keep the samples that find new coverage paths in non-relevant functions.
The two new command line options are `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist`. They accept files in the same format as the existing `-fsanitize-blacklist` option <https://clang.llvm.org/docs/SanitizerSpecialCaseList.html#format>. The new options influence SanitizerCoverage so that it will only instrument a subset of the functions in the target. We explain these options in detail in `clang/docs/SanitizerCoverage.rst`.
Consider now the woff2 fuzzing example from the libFuzzer tutorial <https://github.com/google/fuzzer-test-suite/blob/master/tutorial/libFuzzerTutorial.md>. We are aware that we cannot conclude much from this example because mutating compressed data is generally a bad idea, but let us use it anyway as an illustration for its simplicity. Let us use an empty blacklist together with one of the three following whitelists:
```
# (a)
src:*
fun:*
# (b)
src:SRC/*
fun:*
# (c)
src:SRC/src/woff2_dec.cc
fun:*
```
Running the built fuzzers shows how many instrumentation points the compiler adds, the fuzzer will output //XXX PCs//. Whitelist (a) is the instrument-everything whitelist, it produces 11912 instrumentation points. Whitelist (b) focuses coverage to instrument woff2 source code only, ignoring the dependency code for brotli (de)compression; it produces 3984 instrumented instrumentation points. Whitelist (c) focuses coverage to only instrument functions in the main file that deals with WOFF2 to TTF conversion, resulting in 1056 instrumentation points.
For experimentation purposes, we ran each fuzzer approximately 100 times, single process, with the initial corpus provided in the tutorial. We let the fuzzer run until it either found the heap buffer overflow or went out of memory. On this simple example, whitelists (b) and (c) found the heap buffer overflow more reliably and 5x faster than whitelist (a). The average execution times when finding the heap buffer overflow were as follows: (a) 904 s, (b) 156 s, and (c) 176 s.
We explain these results by the fact that WOFF2 to TTF conversion calls the brotli decompression algorithm's functions, which are mostly irrelevant for finding bugs in WOFF2 font reconstruction but nevertheless instrumented and used by whitelist (a) to guide fuzzing. This results in longer execution time for these functions and a partially irrelevant corpus. Contrary to whitelist (a), whitelists (b) and (c) will execute brotli-related functions without instrumentation overhead, and ignore new code paths found in them. This results in faster bug finding for WOFF2 font reconstruction.
The results for whitelist (b) are similar to the ones for whitelist (c). Indeed, WOFF2 to TTF conversion calls functions that are mostly located in SRC/src/woff2_dec.cc. The 2892 extra instrumentation points allowed by whitelist (b) do not tamper with bug finding, even though they are mostly irrelevant, simply because most of these functions do not get called. We get a slightly faster average time for bug finding with whitelist (b), which might indicate that some of the extra instrumentation points are actually relevant, or might just be random noise.
Reviewers: kcc, morehouse, vitalybuka
Reviewed By: morehouse, vitalybuka
Subscribers: pratyai, vitalybuka, eternalsakura, xwlin222, dende, srhines, kubamracek, #sanitizers, lebedev.ri, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D63616
2020-04-11 01:42:41 +08:00
|
|
|
src:bar/b.cpp
|
2020-06-20 13:22:47 +08:00
|
|
|
# Disable instrumentation for a specific function that the allowlist allowed
|
Implement `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist` for clang
Summary:
This commit adds two command-line options to clang.
These options let the user decide which functions will receive SanitizerCoverage instrumentation.
This is most useful in the libFuzzer use case, where it enables targeted coverage-guided fuzzing.
Patch by Yannis Juglaret of DGA-MI, Rennes, France
libFuzzer tests its target against an evolving corpus, and relies on SanitizerCoverage instrumentation to collect the code coverage information that drives corpus evolution. Currently, libFuzzer collects such information for all functions of the target under test, and adds to the corpus every mutated sample that finds a new code coverage path in any function of the target. We propose instead to let the user specify which functions' code coverage information is relevant for building the upcoming fuzzing campaign's corpus. To this end, we add two new command line options for clang, enabling targeted coverage-guided fuzzing with libFuzzer. We see targeted coverage guided fuzzing as a simple way to leverage libFuzzer for big targets with thousands of functions or multiple dependencies. We publish this patch as work from DGA-MI of Rennes, France, with proper authorization from the hierarchy.
Targeted coverage-guided fuzzing can accelerate bug finding for two reasons. First, the compiler will avoid costly instrumentation for non-relevant functions, accelerating fuzzer execution for each call to any of these functions. Second, the built fuzzer will produce and use a more accurate corpus, because it will not keep the samples that find new coverage paths in non-relevant functions.
The two new command line options are `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist`. They accept files in the same format as the existing `-fsanitize-blacklist` option <https://clang.llvm.org/docs/SanitizerSpecialCaseList.html#format>. The new options influence SanitizerCoverage so that it will only instrument a subset of the functions in the target. We explain these options in detail in `clang/docs/SanitizerCoverage.rst`.
Consider now the woff2 fuzzing example from the libFuzzer tutorial <https://github.com/google/fuzzer-test-suite/blob/master/tutorial/libFuzzerTutorial.md>. We are aware that we cannot conclude much from this example because mutating compressed data is generally a bad idea, but let us use it anyway as an illustration for its simplicity. Let us use an empty blacklist together with one of the three following whitelists:
```
# (a)
src:*
fun:*
# (b)
src:SRC/*
fun:*
# (c)
src:SRC/src/woff2_dec.cc
fun:*
```
Running the built fuzzers shows how many instrumentation points the compiler adds, the fuzzer will output //XXX PCs//. Whitelist (a) is the instrument-everything whitelist, it produces 11912 instrumentation points. Whitelist (b) focuses coverage to instrument woff2 source code only, ignoring the dependency code for brotli (de)compression; it produces 3984 instrumented instrumentation points. Whitelist (c) focuses coverage to only instrument functions in the main file that deals with WOFF2 to TTF conversion, resulting in 1056 instrumentation points.
For experimentation purposes, we ran each fuzzer approximately 100 times, single process, with the initial corpus provided in the tutorial. We let the fuzzer run until it either found the heap buffer overflow or went out of memory. On this simple example, whitelists (b) and (c) found the heap buffer overflow more reliably and 5x faster than whitelist (a). The average execution times when finding the heap buffer overflow were as follows: (a) 904 s, (b) 156 s, and (c) 176 s.
We explain these results by the fact that WOFF2 to TTF conversion calls the brotli decompression algorithm's functions, which are mostly irrelevant for finding bugs in WOFF2 font reconstruction but nevertheless instrumented and used by whitelist (a) to guide fuzzing. This results in longer execution time for these functions and a partially irrelevant corpus. Contrary to whitelist (a), whitelists (b) and (c) will execute brotli-related functions without instrumentation overhead, and ignore new code paths found in them. This results in faster bug finding for WOFF2 font reconstruction.
The results for whitelist (b) are similar to the ones for whitelist (c). Indeed, WOFF2 to TTF conversion calls functions that are mostly located in SRC/src/woff2_dec.cc. The 2892 extra instrumentation points allowed by whitelist (b) do not tamper with bug finding, even though they are mostly irrelevant, simply because most of these functions do not get called. We get a slightly faster average time for bug finding with whitelist (b), which might indicate that some of the extra instrumentation points are actually relevant, or might just be random noise.
Reviewers: kcc, morehouse, vitalybuka
Reviewed By: morehouse, vitalybuka
Subscribers: pratyai, vitalybuka, eternalsakura, xwlin222, dende, srhines, kubamracek, #sanitizers, lebedev.ri, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D63616
2020-04-11 01:42:41 +08:00
|
|
|
fun:*myFunc*
|
|
|
|
|
|
|
|
The use of ``*`` wildcards above is required because function names are matched after mangling.
|
|
|
|
Without the wildcards, one would have to write the whole mangled name.
|
|
|
|
|
|
|
|
Be careful that the paths of source files are matched exactly as they are provided on the clang
|
|
|
|
command line.
|
2020-06-20 13:22:47 +08:00
|
|
|
For example, the allowlist above would include file ``bar/b.cpp`` if the path was provided
|
Implement `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist` for clang
Summary:
This commit adds two command-line options to clang.
These options let the user decide which functions will receive SanitizerCoverage instrumentation.
This is most useful in the libFuzzer use case, where it enables targeted coverage-guided fuzzing.
Patch by Yannis Juglaret of DGA-MI, Rennes, France
libFuzzer tests its target against an evolving corpus, and relies on SanitizerCoverage instrumentation to collect the code coverage information that drives corpus evolution. Currently, libFuzzer collects such information for all functions of the target under test, and adds to the corpus every mutated sample that finds a new code coverage path in any function of the target. We propose instead to let the user specify which functions' code coverage information is relevant for building the upcoming fuzzing campaign's corpus. To this end, we add two new command line options for clang, enabling targeted coverage-guided fuzzing with libFuzzer. We see targeted coverage guided fuzzing as a simple way to leverage libFuzzer for big targets with thousands of functions or multiple dependencies. We publish this patch as work from DGA-MI of Rennes, France, with proper authorization from the hierarchy.
Targeted coverage-guided fuzzing can accelerate bug finding for two reasons. First, the compiler will avoid costly instrumentation for non-relevant functions, accelerating fuzzer execution for each call to any of these functions. Second, the built fuzzer will produce and use a more accurate corpus, because it will not keep the samples that find new coverage paths in non-relevant functions.
The two new command line options are `-fsanitize-coverage-whitelist` and `-fsanitize-coverage-blacklist`. They accept files in the same format as the existing `-fsanitize-blacklist` option <https://clang.llvm.org/docs/SanitizerSpecialCaseList.html#format>. The new options influence SanitizerCoverage so that it will only instrument a subset of the functions in the target. We explain these options in detail in `clang/docs/SanitizerCoverage.rst`.
Consider now the woff2 fuzzing example from the libFuzzer tutorial <https://github.com/google/fuzzer-test-suite/blob/master/tutorial/libFuzzerTutorial.md>. We are aware that we cannot conclude much from this example because mutating compressed data is generally a bad idea, but let us use it anyway as an illustration for its simplicity. Let us use an empty blacklist together with one of the three following whitelists:
```
# (a)
src:*
fun:*
# (b)
src:SRC/*
fun:*
# (c)
src:SRC/src/woff2_dec.cc
fun:*
```
Running the built fuzzers shows how many instrumentation points the compiler adds, the fuzzer will output //XXX PCs//. Whitelist (a) is the instrument-everything whitelist, it produces 11912 instrumentation points. Whitelist (b) focuses coverage to instrument woff2 source code only, ignoring the dependency code for brotli (de)compression; it produces 3984 instrumented instrumentation points. Whitelist (c) focuses coverage to only instrument functions in the main file that deals with WOFF2 to TTF conversion, resulting in 1056 instrumentation points.
For experimentation purposes, we ran each fuzzer approximately 100 times, single process, with the initial corpus provided in the tutorial. We let the fuzzer run until it either found the heap buffer overflow or went out of memory. On this simple example, whitelists (b) and (c) found the heap buffer overflow more reliably and 5x faster than whitelist (a). The average execution times when finding the heap buffer overflow were as follows: (a) 904 s, (b) 156 s, and (c) 176 s.
We explain these results by the fact that WOFF2 to TTF conversion calls the brotli decompression algorithm's functions, which are mostly irrelevant for finding bugs in WOFF2 font reconstruction but nevertheless instrumented and used by whitelist (a) to guide fuzzing. This results in longer execution time for these functions and a partially irrelevant corpus. Contrary to whitelist (a), whitelists (b) and (c) will execute brotli-related functions without instrumentation overhead, and ignore new code paths found in them. This results in faster bug finding for WOFF2 font reconstruction.
The results for whitelist (b) are similar to the ones for whitelist (c). Indeed, WOFF2 to TTF conversion calls functions that are mostly located in SRC/src/woff2_dec.cc. The 2892 extra instrumentation points allowed by whitelist (b) do not tamper with bug finding, even though they are mostly irrelevant, simply because most of these functions do not get called. We get a slightly faster average time for bug finding with whitelist (b), which might indicate that some of the extra instrumentation points are actually relevant, or might just be random noise.
Reviewers: kcc, morehouse, vitalybuka
Reviewed By: morehouse, vitalybuka
Subscribers: pratyai, vitalybuka, eternalsakura, xwlin222, dende, srhines, kubamracek, #sanitizers, lebedev.ri, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D63616
2020-04-11 01:42:41 +08:00
|
|
|
exactly like this, but would it would fail to include it with other ways to refer to the same
|
|
|
|
file such as ``./bar/b.cpp``, or ``bar\b.cpp`` on Windows.
|
|
|
|
So, please make sure to always double check that your lists are correctly applied.
|
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
Default implementation
|
|
|
|
======================
|
2015-04-24 04:40:04 +08:00
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
The sanitizer run-time (AddressSanitizer, MemorySanitizer, etc) provide a
|
|
|
|
default implementations of some of the coverage callbacks.
|
|
|
|
You may use this implementation to dump the coverage on disk at the process
|
|
|
|
exit.
|
|
|
|
|
|
|
|
Example:
|
2015-04-24 04:40:04 +08:00
|
|
|
|
|
|
|
.. code-block:: console
|
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
% cat -n cov.cc
|
|
|
|
1 #include <stdio.h>
|
|
|
|
2 __attribute__((noinline))
|
|
|
|
3 void foo() { printf("foo\n"); }
|
|
|
|
4
|
|
|
|
5 int main(int argc, char **argv) {
|
|
|
|
6 if (argc == 2)
|
|
|
|
7 foo();
|
|
|
|
8 printf("main\n");
|
|
|
|
9 }
|
|
|
|
% clang++ -g cov.cc -fsanitize=address -fsanitize-coverage=trace-pc-guard
|
|
|
|
% ASAN_OPTIONS=coverage=1 ./a.out; wc -c *.sancov
|
|
|
|
main
|
|
|
|
SanitizerCoverage: ./a.out.7312.sancov 2 PCs written
|
|
|
|
24 a.out.7312.sancov
|
|
|
|
% ASAN_OPTIONS=coverage=1 ./a.out foo ; wc -c *.sancov
|
|
|
|
foo
|
|
|
|
main
|
|
|
|
SanitizerCoverage: ./a.out.7316.sancov 3 PCs written
|
|
|
|
24 a.out.7312.sancov
|
|
|
|
32 a.out.7316.sancov
|
2015-04-24 04:40:04 +08:00
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
Every time you run an executable instrumented with SanitizerCoverage
|
|
|
|
one ``*.sancov`` file is created during the process shutdown.
|
|
|
|
If the executable is dynamically linked against instrumented DSOs,
|
|
|
|
one ``*.sancov`` file will be also created for every DSO.
|
2015-04-24 04:40:04 +08:00
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
Sancov data format
|
|
|
|
------------------
|
2017-05-02 03:52:01 +08:00
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
The format of ``*.sancov`` files is very simple: the first 8 bytes is the magic,
|
|
|
|
one of ``0xC0BFFFFFFFFFFF64`` and ``0xC0BFFFFFFFFFFF32``. The last byte of the
|
|
|
|
magic defines the size of the following offsets. The rest of the data is the
|
|
|
|
offsets in the corresponding binary/DSO that were executed during the run.
|
2015-04-24 04:40:04 +08:00
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
Sancov Tool
|
|
|
|
-----------
|
2015-04-24 04:40:04 +08:00
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
An simple ``sancov`` tool is provided to process coverage files.
|
|
|
|
The tool is part of LLVM project and is currently supported only on Linux.
|
|
|
|
It can handle symbolization tasks autonomously without any extra support
|
|
|
|
from the environment. You need to pass .sancov files (named
|
|
|
|
``<module_name>.<pid>.sancov`` and paths to all corresponding binary elf files.
|
|
|
|
Sancov matches these files using module names and binaries file names.
|
2015-04-24 04:40:04 +08:00
|
|
|
|
|
|
|
.. code-block:: console
|
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
USAGE: sancov [options] <action> (<binary file>|<.sancov file>)...
|
|
|
|
|
|
|
|
Action (required)
|
|
|
|
-print - Print coverage addresses
|
|
|
|
-covered-functions - Print all covered functions.
|
|
|
|
-not-covered-functions - Print all not covered functions.
|
|
|
|
-symbolize - Symbolizes the report.
|
|
|
|
|
|
|
|
Options
|
2020-06-20 13:22:47 +08:00
|
|
|
-blocklist=<string> - Blocklist file (sanitizer blocklist format).
|
2017-05-02 08:32:57 +08:00
|
|
|
-demangle - Print demangled function name.
|
|
|
|
-strip_path_prefix=<string> - Strip this prefix from file paths in reports
|
|
|
|
|
|
|
|
|
|
|
|
Coverage Reports
|
|
|
|
----------------
|
2015-04-24 04:40:04 +08:00
|
|
|
|
2017-05-02 08:32:57 +08:00
|
|
|
**Experimental**
|
|
|
|
|
|
|
|
``.sancov`` files do not contain enough information to generate a source-level
|
|
|
|
coverage report. The missing information is contained
|
|
|
|
in debug info of the binary. Thus the ``.sancov`` has to be symbolized
|
|
|
|
to produce a ``.symcov`` file first:
|
|
|
|
|
|
|
|
.. code-block:: console
|
|
|
|
|
|
|
|
sancov -symbolize my_program.123.sancov my_program > my_program.123.symcov
|
|
|
|
|
2020-01-19 22:51:25 +08:00
|
|
|
The ``.symcov`` file can be browsed overlaid over the source code by
|
2017-05-02 08:32:57 +08:00
|
|
|
running ``tools/sancov/coverage-report-server.py`` script that will start
|
|
|
|
an HTTP server.
|
|
|
|
|
|
|
|
Output directory
|
|
|
|
----------------
|
|
|
|
|
|
|
|
By default, .sancov files are created in the current working directory.
|
|
|
|
This can be changed with ``ASAN_OPTIONS=coverage_dir=/path``:
|
|
|
|
|
|
|
|
.. code-block:: console
|
|
|
|
|
|
|
|
% ASAN_OPTIONS="coverage=1:coverage_dir=/tmp/cov" ./a.out foo
|
|
|
|
% ls -l /tmp/cov/*sancov
|
|
|
|
-rw-r----- 1 kcc eng 4 Nov 27 12:21 a.out.22673.sancov
|
|
|
|
-rw-r----- 1 kcc eng 8 Nov 27 12:21 a.out.22679.sancov
|