Mercurial > hg
view tests/test-fuzz-targets.t @ 45825:8f07f5a9c3de
worker: raise exception instead of calling sys.exit() with child's code
When a worker process returns an error code, we would call
`sys.exit()` with that exit code on the main process. The `SystemExit`
exception would then get caught in `scmutil.callcatch()`, which would
return that error code. The comment there says "Commands shouldn't
sys.exit directly", which I agree with. This patch changes it so we
raise a specific exception when a worker fails so we can catch
instead. I think that means that `SystemExit` is now always an
internal error.
(I had earlier thought that this call to `sys.exit()` was from within
the child process until Matt Harbison made me look again, so thanks
for that!)
Differential Revision: https://phab.mercurial-scm.org/D9287
author | Martin von Zweigbergk <martinvonz@google.com> |
---|---|
date | Sat, 07 Nov 2020 21:50:28 -0800 |
parents | 19da643dc10c |
children | b918494198f7 |
line wrap: on
line source
#require test-repo $ cd $TESTDIR/../contrib/fuzz $ OUT=$TESTTMP ; export OUT which(1) could exit nonzero, but that's fine because we'll still end up without a valid executable, so we don't need to check $? here. $ if which gmake >/dev/null 2>&1; then > MAKE=gmake > else > MAKE=make > fi $ havefuzz() { > cat > $TESTTMP/dummy.cc <<EOF > #include <stdlib.h> > #include <stdint.h> > int LLVMFuzzerTestOneInput(const uint8_t *Data, size_t Size) { return 0; } > int main(int argc, char **argv) { > const char data[] = "asdf"; > return LLVMFuzzerTestOneInput((const uint8_t *)data, 4); > } > EOF > $CXX $TESTTMP/dummy.cc -o $TESTTMP/dummy \ > -fsanitize=fuzzer-no-link,address || return 1 > } #if clang-libfuzzer $ CXX=clang++ havefuzz || exit 80 $ $MAKE -s clean all PYTHON_CONFIG=`which python-config` #endif #if no-clang-libfuzzer clang-6.0 $ CXX=clang++-6.0 havefuzz || exit 80 $ $MAKE -s clean all CC=clang-6.0 CXX=clang++-6.0 PYTHON_CONFIG=`which python-config` #endif #if no-clang-libfuzzer no-clang-6.0 $ exit 80 #endif $ cd $TESTTMP Run each fuzzer using dummy.cc as a fake input, to make sure it runs at all. In the future we should instead unpack the corpus for each fuzzer and use that instead. $ for fuzzer in `ls *_fuzzer | sort` ; do > echo run $fuzzer... > ./$fuzzer dummy.cc > /dev/null 2>&1 > done run bdiff_fuzzer... run dirs_fuzzer... run dirstate_fuzzer... run fm1readmarkers_fuzzer... run fncache_fuzzer... run jsonescapeu8fast_fuzzer... run manifest_fuzzer... run mpatch_fuzzer... run revlog_fuzzer... run xdiff_fuzzer... Clean up. $ cd $TESTDIR/../contrib/fuzz $ $MAKE -s clean