Summary:
Linking with the /OPT:REF linker flag when building COFF files causes
the linker to strip SanitizerCoverage's constructors. Prevent this by
giving the constructors WeakODR linkage and by passing the linker a
directive to include sancov.module_ctor.
Include a test in compiler-rt to verify libFuzzer can be linked using
/OPT:REF
Reviewers: morehouse, rnk
Reviewed By: morehouse, rnk
Subscribers: rnk, morehouse, hiraditya
Differential Revision: https://reviews.llvm.org/D52119
llvm-svn: 344391
Some seem fragile, some fail, and some just take a really long time to run.
It does not seem to make sense to support some subset of value profiling tests.
Differential Revision: https://reviews.llvm.org/D53047
llvm-svn: 344105
Without this change, tests in coverage.test and dump_coverage.test are
failing on non-x86_64 platforms.
The diff is copied from sanitizer_common library, an alternative would
be to link it together with libFuzzer.
Differential Revision: https://reviews.llvm.org/D53040
llvm-svn: 344104
r342698 removed the 'UNSUPPORTED: windows' flags from a number of fuzzer
tests, however too many lines were removed from 'fuzzer-flags.test',
including the run-line that generates the test binary, which breaks that
test for me (and the clang-x64-ninja-win7 buildbot). This patch just
re-adds that line.
llvm-svn: 342720
Summary:
Replace FuzzerExtFunctionsDlsymWin.cpp with FuzzerExtFunctionsWeakAlias.cpp
to get externally defined functions (eg: LLVMFuzzerInitialize,
LLVMFuzzerCustomMutator, etc) working again.
Also enable tests that depended on these functions (on windows)
Reviewers: rnk, morehouse
Reviewed By: rnk, morehouse
Subscribers: rnk, morehouse, mgorny
Differential Revision: https://reviews.llvm.org/D51700
llvm-svn: 342698
Summary:
Enables trace-malloc-unbalanced.test on Windows, fixing two problems it had with Windows before.
The first fix is specifying python instead of relying on a script's shebang since they can't be used on Windows.
The second fix is making the regex tolerate windows' implementation of the "%p" format string.
Reviewers: Dor1s
Reviewed By: Dor1s
Subscribers: morehouse
Differential Revision: https://reviews.llvm.org/D51760
llvm-svn: 341632
Summary:
When targeting MSVC: compile using clang's cl driver mode (this is needed for
libfuzzer's exit_on_src_pos feature). Don't use -lstdc++ when linking,
it isn't needed and causes a warning.
On Windows: Fix exit_on_src_pos.test by making sure debug info isn't
overwritten during compilation of second binary by using .exe extension.
Reviewers: morehouse
Reviewed By: morehouse
Subscribers: aprantl, JDevlieghere
Differential Revision: https://reviews.llvm.org/D51757
llvm-svn: 341622
Summary:
Remove unneeded #include of <unistd.h> from SymbolizeDeadlock.cpp to
enable compilation on Windows and enable symbolize-deadlock.test on
Windows.
Patch by Jonathan Metzman (@metzman).
Reviewers: Dor1s
Reviewed By: Dor1s
Subscribers: kcc, delcypher, llvm-commits, #sanitizers
Differential Revision: https://reviews.llvm.org/D51685
llvm-svn: 341473
Summary:
Enable tests on Windows and make check-fuzzer pass on it. Make
check-fuzzer pass on Windows by fixing libFuzzer, fixing tests, and by
disabling tests on Windows. Most of these are disabled temporarily as
support for the tests and the features they test will be added
incrementally. Other tests will not be enabled since they require
things that are not on Windows (eg: afl_driver tests). Every test
that was explicitly disabled on Windows has a comment explaining why
(unless obvious like merge-posix.test).
The lit.cfg file was modified to support running tests on windows.
fuzzer-dirs.test was fixed by making the Windows implementation print
the same error message as the posix version.
merge-control-file.test was fixed by making the test binary end with
the ".exe" extension (on all platforms).
Patch By: metzman
Reviewers: morehouse
Reviewed By: morehouse
Subscribers: srhines, mgorny
Differential Revision: https://reviews.llvm.org/D51549
llvm-svn: 341385
Summary:
This was an experimental feature. After evaluating it with:
1) https://github.com/google/fuzzer-test-suite/tree/master/engine-comparison
2) enabling on real world fuzz targets running at ClusterFuzz and OSS-Fuzz
The following conclusions were made:
1) With fuzz targets that have reached a code coverage plateau, the feature does
not improve libFuzzer's ability to discover new coverage and may actually
negatively impact it.
2) With fuzz targets that have not yet reached a code coverage plateau, the
feature might speed up new units discovery in some cases, but it is quite
rare and hard to confirm with a high level on confidence.
Revert of https://reviews.llvm.org/D48054 and https://reviews.llvm.org/D49621.
Reviewers: metzman, morehouse
Reviewed By: metzman, morehouse
Subscribers: delcypher, #sanitizers, llvm-commits, kcc
Differential Revision: https://reviews.llvm.org/D51455
llvm-svn: 340976
Summary:
Without this metadata LLD strips unused PC table entries
but won't strip unused guards. This metadata also seems
to influence the linker to change the ordering in the PC
guard section to match that of the PC table section.
The libFuzzer runtime library depends on the ordering
of the PC table and PC guard sections being the same. This
is not generally guaranteed, so we may need to redesign
PC tables/guards/counters in the future.
Reviewers: eugenis
Reviewed By: eugenis
Subscribers: kcc, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D50483
llvm-svn: 339733
After https://reviews.llvm.org/D48800, shrink.test started failing on
x86_64h architecture.
Looking into this, the optimization pass is too eager to unroll the loop
on x86_64h, possibly leading to worse coverage data.
Alternative solutions include not unrolling the loop when fuzzing, or
disabling this test on that architecture.
Differential Revision: https://reviews.llvm.org/D50484
llvm-svn: 339303
Summary:
There may be cases in which a user wants to know which part of their code is unstable.
We use ObservedFuncs and UnstableCounters to print at exit which of the ObservedFunctions
are unstable under the -print_unstable_stats flag.
Patch by Kyungtak Woo (@kevinwkt).
Reviewers: Dor1s, metzman, morehouse
Reviewed By: Dor1s, metzman, morehouse
Subscribers: delcypher, #sanitizers, llvm-commits, kcc
Differential Revision: https://reviews.llvm.org/D50264
llvm-svn: 339081
Summary:
Added functions that calculate stats while fuzz targets are running and give
mutations weight based on how much new coverage they provide, and choose better
performing mutations more often.
Patch by Kodé Williams (@kodewilliams).
Reviewers: Dor1s, metzman, morehouse
Reviewed By: Dor1s, morehouse
Subscribers: delcypher, kcc, llvm-commits, #sanitizers
Differential Revision: https://reviews.llvm.org/D49621
llvm-svn: 338776
Summary:
Added a new mode within flag -handle_unstable for new unstable handling algorithm that does the following:
When an edge is shown as unstable, copy to UnstableCounters the value 0.
During ApplyUnstableCounters we copy back the value 0 to ModuleInline8bitCounters if the edge was unstable.
This way we would be ignoring completely features that were collected through non-determinism.
Unstable hits would be counted as if it never hit.
Reviewers: metzman, Dor1s, kcc, morehouse
Reviewed By: metzman, morehouse
Subscribers: delcypher, llvm-commits, #sanitizers
Differential Revision: https://reviews.llvm.org/D49684
llvm-svn: 337853
Summary:
Created unstable_handle flag that takes 1 or 2, depending on the handling type.
Modified RunOne to accommodate the following heuristic:
Use the first CollectFeatures to count how many features there are.
If no new features, CollectFeatures like before.
If there is new feature, we run CB 2 more times,
Check which edges are unstable per input and we store the least amount of hit counts for each edge.
Apply these hit counts back to inline8bitcounters so that CollectFeatures can work as intended.
Modified UnstableCounters to 8int_t and created a bitset UnstableSet to tell which edges are unstable.
Patch by Kyungtak Woo (@kevinwkt).
Reviewers: Dor1s, metzman, morehouse
Reviewed By: Dor1s, morehouse
Subscribers: delcypher, #sanitizers, llvm-commits, kcc
Differential Revision: https://reviews.llvm.org/D49525
llvm-svn: 337696
Previously, check-all failed many tests for me. It was running the
X86_64DefaultLinuxConfig, X86_64LibcxxLinuxConfig, and
X86_64StaticLibcxxLinuxConfig configs out of
llvm-build/projects/compiler-rt/test/fuzzer. Now, it runs them out of
separate subdirectories there, and most tests pass.
Reviewed By: morehouse, george.karpenkov
Differential Revision: https://reviews.llvm.org/D49249
llvm-svn: 337521
Summary:
Code now exists to track number of mutations that are used in fuzzing in total
and ones that produce new coverage. The stats are currently being dumped to the
command line.
Patch by Kodé Williams (@kodewilliams).
Reviewers: metzman, Dor1s, morehouse, kcc
Reviewed By: Dor1s, morehouse, kcc
Subscribers: delcypher, kubamracek, kcc, morehouse, llvm-commits, #sanitizers, mgorny
Differential Revision: https://reviews.llvm.org/D48054
llvm-svn: 337324
This is a fix for bug 37047.
https://bugs.llvm.org/show_bug.cgi?id=37047
Implemented by basically reversing the logic. Previously all strings
were considered, with some operations excluded. Now strings are excluded
by default, and only strings during the CB considered.
Patch By: pdknsk
Differential Revision: https://reviews.llvm.org/D48800
llvm-svn: 337296
Summary:
Code now exists to track number of mutations that are used in fuzzing in total
and ones that produce new coverage. The stats are currently being dumped to the
command line.
Patch by Kodé Williams (@kodewilliams).
Reviewers: metzman, Dor1s, morehouse, kcc
Reviewed By: Dor1s, morehouse, kcc
Subscribers: delcypher, kubamracek, kcc, morehouse, llvm-commits, #sanitizers, mgorny
Differential Revision: https://reviews.llvm.org/D48054
llvm-svn: 337194
Summary:
Created a -print_unstable_stats flag.
When -print_unstable_stats=1, we run it 2 more times on interesting inputs poisoning unstable edges in an array.
On program termination, we run PrintUnstableStats() which will print a line with a stability percentage like AFL does.
Patch by Kyungtak Woo (@kevinwkt).
Reviewers: metzman, Dor1s, kcc, morehouse
Reviewed By: metzman, Dor1s, morehouse
Subscribers: delcypher, llvm-commits, #sanitizers, kcc, morehouse, Dor1s
Differential Revision: https://reviews.llvm.org/D49212
llvm-svn: 337187
Summary:
Created a -print_unstable_stats flag.
When -print_unstable_stats=1, we run it 2 more times on interesting inputs poisoning unstable edges in an array.
On program termination, we run PrintUnstableStats() which will print a line with a stability percentage like AFL does.
Patch by Kyungtak Woo (@kevinwkt).
Reviewers: metzman, Dor1s, kcc, morehouse
Reviewed By: metzman, Dor1s, morehouse
Subscribers: delcypher, llvm-commits, #sanitizers, kcc, morehouse, Dor1s
Differential Revision: https://reviews.llvm.org/D49212
llvm-svn: 337175
Summary:
This allows counters associated with unused functions to be
dead-stripped along with their functions. This approach is the same one
we used for PC tables.
Fixes an issue where LLD removes an unused PC table but leaves the 8-bit
counter.
Reviewers: eugenis
Reviewed By: eugenis
Subscribers: llvm-commits, hiraditya, kcc
Differential Revision: https://reviews.llvm.org/D49264
llvm-svn: 336941
Summary:
After my recent change to allow MSan + libFuzzer, the
ExplodeDFSanLabelsTest.cpp test started to overflow the stack with
recursive function SetBytesForLabel() on an AArch64 bot. Perhaps that
bot has a smaller stack size, or maybe AArch64 has larger stack frames
for this particular function.
Reviewers: kcc, javed.absar
Reviewed By: kcc
Subscribers: kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D49150
llvm-svn: 336725
This reverts rL334510 due to breakage of afl_driver's command line
interface.
Patch By: Jonathan Metzman
Differential Revision: https://reviews.llvm.org/D49141
llvm-svn: 336719
Code now exists to track number of mutations that are used in fuzzing in
total and ones that produce new coverage. The stats are currently being
dumped to the command line.
Patch By: Kode Williams
Differntial Revision: https://reviews.llvm.org/D48054
llvm-svn: 336597
This change adds a support for multiarch style runtimes layout, so in
addition to the existing layout where runtimes get installed to:
lib/clang/$version/lib/$os
Clang now allows runtimes to be installed to:
lib/clang/$version/$target/lib
This also includes libc++, libc++abi and libunwind; today those are
assumed to be in Clang library directory built for host, with the
new layout it is possible to install libc++, libc++abi and libunwind
into the runtime directory built for different targets.
The use of new layout is enabled by setting the
LLVM_ENABLE_RUNTIME_TARGET_DIR CMake variable and is supported by both
projects and runtimes layouts. The runtimes CMake build has been further
modified to use the new layout when building runtimes for multiple
targets.
Differential Revision: https://reviews.llvm.org/D45604
llvm-svn: 335809
Under load, these tests tend to fail sporadically on our bots.
In my understanding, the signal handling is not guaranteed to happen
within 2 seconds, and the test is inherently flaky.
llvm-svn: 335792
There's more platforms than just "ios" and "iossim" that we should support, and adding more lit config variables for each platform isn't great. Let's generalize and have a single value that determines what the platform under test is.
Differential Revision: https://reviews.llvm.org/D48309
llvm-svn: 335123
Summary:
Using associated metadata rather than llvm.used allows linkers to
perform dead stripping with -fsanitize-coverage=pc-table. Unfortunately
in my local tests, LLD was the only linker that made use of this metadata.
Partially addresses https://bugs.llvm.org/show_bug.cgi?id=34636 and fixes
https://github.com/google/sanitizers/issues/971.
Reviewers: eugenis
Reviewed By: eugenis
Subscribers: Dor1s, hiraditya, llvm-commits, kcc
Differential Revision: https://reviews.llvm.org/D48203
llvm-svn: 334858
For paths without "%t" inside uniqueness is not guaranteed, and
potential collisions might be present.
Differential Revision: https://reviews.llvm.org/D47288
llvm-svn: 334546
Handling differently the iterations with the type limit and eventually an error message.
Reviewers: morehouse, kcc
Reviewed By: morehouse
Differential Revision: https://reviews.llvm.org/D47880
llvm-svn: 334510
The ordering of stderr/stdout writes might not be guaranteed,
and is irrelevant for this test.
Differential Revision: https://reviews.llvm.org/D47294
llvm-svn: 333245
With %run commands being present now, we want to make sure that they
stay present.
Differential Revision: https://reviews.llvm.org/D47293
llvm-svn: 333244
The bug was not previously exposed as only one architecture is currently
supported for libFuzzer.
Differential Revision: https://reviews.llvm.org/D47287
llvm-svn: 333242
Summary:
This can be solved just in seconds with KLEE. Current libFuzzer
is able to satistfy 101 constraints out of 410 constraints presented during
the first hour of running with -use_value_profile=1 and -max_len=20.
During the next 3 hours, libFuzzer is able to generate ~50 NEW inputs,
bot none of those solve any new constraint.
During the next 20 hours, it didn't find any NEW inputs.
This test might be interesting for experimenting with the data flow tracing
approach started in https://reviews.llvm.org/D46666.
For the solution with KLEE and other information, see
https://github.com/Dor1s/codegate2017-quals-angrybird
Reviewers: kcc
Reviewed By: kcc
Subscribers: delcypher, llvm-commits, #sanitizers
Differential Revision: https://reviews.llvm.org/D46924
llvm-svn: 332507
Keeping fuzzer.test together as a gigantic test has no advantages and multiple disadvantages:
- Worse test parallelization (fuzzer.test is always blocking the test
run on a machine with many cores)
- Debugging test failures is made more difficult (not clear what fails
from fuzzer.test failing)
- Makes porting tests to new platforms more difficult (whenever
fuzzer.test fails have to inspect the output to figure out what is going
on, and then restart all of it)
- Hides dead code (in our case, "Done1000000" FileCheck variable was
never used, DISABLED: not %t-UninstrumentedTest-Uninstrumented was never
compiled, and there was small amount of duplication vs. simple-cmp.test)
- Makes correspondence between LIT .test files and .cpp files less
straightforward
Differential Revision: https://reviews.llvm.org/D46557
llvm-svn: 332145
This patch does not do anything, but paves the way for future changes,
where %run command will be expanded into a script performing the testing
on device.
Differential Revision: https://reviews.llvm.org/D46553
llvm-svn: 332144
Summary:
Experimental data flow tracer for fuzz targets.
Allows to tell which bytes of the input affect which functions of the fuzz target.
We previously attempted to use DFSan directly in the libFuzzer process,
and that didn't work nicely.
Now we will try to collect the data flow information for the seed corpus
in a separate process (using this tracer), and then use it in the regular libFuzzer runs.
Reviewers: morehouse, pcc, Dor1s
Reviewed By: morehouse, Dor1s
Subscribers: delcypher, #sanitizers, llvm-commits
Differential Revision: https://reviews.llvm.org/D46666
llvm-svn: 332029
Summary:
Fixed two non-standard usages of sem_open in the libFuzzer library and
one NetBSD-related modification with test script.
- The return value to indicate error should be SEM_FAILED instead of
(void *)-1 (please refer to "RETURN VALUE" section in this [[
http://pubs.opengroup.org/onlinepubs/9699919799/functions/sem_open.html
| page ]]). Actually, SEM_FAILED != (void *)-1 holds in NetBSD.
- The SharedMemoryRegion::SemName function should return name
starting with slash. Because the behaviour of name which does not
start with slash is unspecified as the [[
http://pubs.opengroup.org/onlinepubs/9699919799/functions/sem_open.html
| "DESCRIPTION" section ]] specified:
> If name does not begin with the <slash> character, the effect is implementation-defined.
- The length of name is limited to 14 in NetBSD, it is suggested to
reduce the length of equivalence server name in the test script.
Patch by: Yang Zheng
Reviewers: vitalybuka, krytarowski, kcc
Reviewed By: kcc
Subscribers: kcc, #sanitizers, llvm-commits, joerg
Differential Revision: https://reviews.llvm.org/D46622
llvm-svn: 332003
The try-lock guard change seems to be making this test fail on Mac, but
I haven't been able to reproduce the failure. Disabling the test on Mac
to fix build bot.
llvm-svn: 331894
Summary:
When out-of-memory or timeout occurs, threads can be stopped during
symbolization, thereby causing a deadlock when the OOM/TO handlers
attempt symbolization. We avoid this deadlock by skipping symbolization
if another thread is symbolizing.
Reviewers: kcc
Reviewed By: kcc
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D46605
llvm-svn: 331825
Summary:
Fixes https://github.com/google/sanitizers/issues/788/, a deadlock
caused by multiple crashes happening at the same time. Before printing
a crash report, we now test and set an atomic flag. If the flag was
already set, the crash handler returns immediately.
Reviewers: kcc
Reviewed By: kcc
Subscribers: llvm-commits, kubamracek
Differential Revision: https://reviews.llvm.org/D46277
llvm-svn: 331310
opened.
Summary:
Currently if the directory cannot be opened for a reason other than
non-existence (e.g. too many open file descriptors) the error message
printed is incredibly confusing.
Patch By: Alex Gaynor
Reviewers: kcc, morehouse
Reviewed By: morehouse
Subscribers: delcypher, llvm-commits, Sanitizers
Differential Revision: https://reviews.llvm.org/D45322
llvm-svn: 329438
For a few tests, volatile was placed before the '*' in pointer
declarations, resulting in it applying to the underlying data rather
than the pointer itself. Placing volatile after the '*' allows us to
switch those tests to -O2.
llvm-svn: 328633
Summary:
Disables certain CMP optimizations to improve fuzzing signal under -O1
and -O2.
Switches all fuzzer tests to -O2 except for a few leak tests where the
leak is optimized out under -O2.
Reviewers: kcc, vitalybuka
Reviewed By: vitalybuka
Subscribers: cfe-commits, llvm-commits
Differential Revision: https://reviews.llvm.org/D44798
llvm-svn: 328384
This changes the add_custom_libcxx macro to resemble the
llvm_ExternalProject_Add. The primary motivation is to avoid
unnecessary libFuzzer rebuilds that are being done on every
Ninja/Make invocation. The libc++ should be only rebuilt whenever
the libc++ source itself changes.
Differential Revision: https://reviews.llvm.org/D43213
llvm-svn: 326921
This is needed in case the users of libFuzzer use libc++ in their
code, which the fuzz target (libFuzzer) will be linked against.
When libc++ source is available, we build a private version of it
and link it against libFuzzer which allows using the same static
library against codebases which use both libc++ and libstdc++.
Differential Revision: https://reviews.llvm.org/D37631
llvm-svn: 322755
This is needed in case the users of libFuzzer use libc++ in their
code, which the fuzz target (libFuzzer) will be linked against.
When libc++ source is available, we build a private version of it
and link it against libFuzzer which allows using the same static
library against codebases which use both libc++ and libstdc++.
Differential Revision: https://reviews.llvm.org/D37631
llvm-svn: 322604
Fails on darwin
Revert "[fuzzer] Script to detect unbalanced allocation in -trace_malloc output"
Needs previous one.
This reverts commit r317034, r317036.
llvm-svn: 317061
Summary:
It can be enabled via "-use_clang_coverage=1" flag. Reason for disabling:
libFuzzer resets Clang Counters and makes it impossible to generate coverage
report for a regular fuzz target (i.e. not standalone build).
Reviewers: kcc
Reviewed By: kcc
Subscribers: kcc
Differential Revision: https://reviews.llvm.org/D38604
llvm-svn: 315029
Summary:
Current implementation does not work if CMAKE_OSX_SYSROOT is not specified.
It silently generates invalid command with the following flags:
`-std=c++11 -lc++ -gline-tables-only -isysroot -fsanitize=address,fuzzer`
and then fails with the following error:
```
warning: no such sysroot directory: '-fsanitize=address,fuzzer' [-Wmissing-sysroot]"
<...>/RepeatedBytesTest.cpp:5:10: fatal error: 'assert.h' file not found
#include <assert.h>
^~~~~~~~~~
1 error generated.
```
However, if you have Command Line Tools installed, you have '/usr/include' dir.
In that case, it is not necessary to specify isysroot path.
Also, with the patch, in case of '/usr/include' does not exist, the '-sysroot'
path would be resolved automatically in compiler-rt/cmake/base-config-ix.cmake.
For more context, see the comment at `compiler-rt/cmake/base-config-ix.cmake#L76`
Reviewers: kcc, george.karpenkov
Reviewed By: kcc, george.karpenkov
Differential Revision: https://reviews.llvm.org/D37721
llvm-svn: 313033
Summary:
Use the initialexec TLS type and eliminate calls to the TLS
wrapper. Fixes the sanitizer-x86_64-linux-fuzzer bot failure.
Reviewers: vitalybuka, kcc
Reviewed By: kcc
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D37026
llvm-svn: 311490
Resulting library binaries will be named libclang_rt.fuzzer*, and will
be placed in Clang toolchain, allowing redistribution.
Differential Revision: https://reviews.llvm.org/D36908
llvm-svn: 311407
Roman Lebedev