fix(profiling): re-initialise string tables post-fork#18063
Conversation
Codeowners resolved as |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: a9deb21451
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
b37f328 to
17cbcbc
Compare
emmettbutler
left a comment
There was a problem hiding this comment.
Oh cool. I wonder if the same logic is applied to the string table that underlies the TraceWriter.
|
@emmettbutler From what I know, I don't think so. We only need to do that in the Profiling Sampling Thread because it's a native Thread (not a PeriodicThread nor Python-owned Thread) that we don't stop at fork, which means post-fork everything that Thread touches without synchronisation can be in any kind of weird state. If the Trace Writer Thread is a Periodic Thread (like for example the Profiling Scheduler / Uploader Thread is), then I believe now we definitely stop/join those Threads before fork which means they can't corrupt anything from suddenly stopping to exist at fork. If it's not a PeriodicThread, though, things may be different (but I'd be surprised, as I haven't seen any crashes that would indicate so). |
17cbcbc to
512674c
Compare
|
4450da7 to
74cd85b
Compare
74cd85b to
d9381a6
Compare
|
/merge -f --reason "microbenchmarks broken for unrelated reason" |
|
View all feedbacks in Devflow UI.
The expected merge time in
Warning This change was merged without running any pre merge CI checks Reason: microbenchmarks broken for unrelated reason |
f1f6901
into
main
…s to CI **Deterministic unit test** (`test_ddup_atfork_handler_registered_before_stack_sampler`): - Mocks `ddup.start` and `StackCollector._init` to record their call order during profiler startup, then asserts ddup was first. - POSIX guarantees FIFO ordering for post-fork child handlers, so whichever component calls pthread_atfork first gets its child handler run first. ddup must run first to rebuild profile state before the stack sampler clears and re-registers threads. - Catches the regression deterministically without needing timing luck or fork() calls — a swap of the init order in profiler.py fails immediately. - Regression guard for PRs #17183/#17042/#18063. **CI: stress-ng CPU saturation + 20 iterations**: - Runs `stress-ng --cpu $(nproc)` in the background during the fork loop to saturate all CPUs and widen the pthread_atfork timing race window — the same load profile that makes the bug manifest in production Gunicorn/gthread. - Increased iterations from 10 to 20 to give more chances to surface the race. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
This change is marked for backport to 4.8 and it does not conflict with that branch. |
## Description This PR ensures we clear the string tables post-fork. This is meant to address a crash happening on (at least) 4.8.0rc5. ``` Error UnixSignal: Process terminated with SEGV_MAPERR (SIGSEGV) #0 0x0000000000000000 std::__detail::_Hashtable_base<unsigned long, std::pair<unsigned long const, std::string>, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Hashtable_traits<false, false, true> >::_M_equals(unsigned long const&, unsigned long, std::__detail::_Hash_node<std::pair<unsigned long const, std::string>, false>*) const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable_policy.h:1804) #1 0x000073afd44ec124 std::__detail::_Hashtable_base<unsigned long, std::pair<unsigned long const, std::string>, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Hashtable_traits<false, false, true> >::_M_equals(unsigned long const&, unsigned long, std::__detail::_Hash_node<std::pair<unsigned long const, std::string>, false>*) const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable_policy.h:1804) #2 0x000073afd44ec124 std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::_M_find_before_node(unsigned long, unsigned long const&, unsigned long) const [clone .constprop.0] (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:1580) #3 0x000073afd44ece4b std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::_M_find_node(unsigned long, unsigned long const&, unsigned long) const [clone .isra.0] (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:694) #4 0x0000000000000000 std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:1455) #5 0x0000000000000000 std::unordered_map<unsigned long, std::string, std::hash<unsigned long>, std::equal_to<unsigned long>, std::allocator<std::pair<unsigned long const, std::string> > >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/unordered_map.h:920) #6 0x000073afd44e4e61 std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:1455) #7 0x000073afd44e4e61 std::unordered_map<unsigned long, std::string, std::hash<unsigned long>, std::equal_to<unsigned long>, std::allocator<std::pair<unsigned long const, std::string> > >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/unordered_map.h:920) #8 0x000073afd44e4e61 StringTable::key(_object*, StringTag) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/strings.cc:58) #9 0x000073afd44e517f Frame::create(EchionSampler&, PyCodeObject*, int) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/frame.cc:24) #10 0x000073afd44e5436 Frame::get(EchionSampler&, PyCodeObject*, int) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/frame.cc:214) #11 0x000073afd44e55b1 Frame::read(EchionSampler&, _PyInterpreterFrame*, _PyInterpreterFrame**) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/frame.cc:153) #12 0x000073afd44e56c0 unwind_frame(EchionSampler&, _object*, FrameStack&, unsigned long) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/stacks.cc:56) #13 0x000073afd44e5859 unwind_python_stack(EchionSampler&, _ts*, FrameStack&) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/stacks.cc:103) #14 0x000073afd44e7bbf ThreadInfo::unwind(EchionSampler&, _ts*) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/threads.cc:15) #15 0x0000000000000000 __gnu_cxx::__normal_iterator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const*, std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > > >::__normal_iterator(std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const* const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_iterator.h:979) #16 0x0000000000000000 std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::end() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:839) #17 0x0000000000000000 std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::empty() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:1008) #18 0x000073afd44eb2eb __gnu_cxx::__normal_iterator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const*, std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > > >::__normal_iterator(std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const* const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_iterator.h:979) #19 0x000073afd44eb2eb std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::end() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:839) #20 0x000073afd44eb2eb std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::empty() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:1008) #21 0x000073afd44eb2eb ThreadInfo::sample(EchionSampler&, _ts*, long) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/threads.cc:698) #22 0x000073afd44eb789 Datadog::Sampler::sampling_thread(unsigned long) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/sampler.cpp:287) #23 0x000073afd44ebad0 call_sampling_thread(void*) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/sampler.cpp:45) #24 0x000073afdf06bac3 start_thread (nptl/nptl/pthread_create.c:442) #25 0x000073afdf0fd8d0 __clone3 ``` Co-authored-by: thomas.kowalski <thomas.kowalski@datadoghq.com> (cherry picked from commit f1f6901) Co-authored-by: Thomas Kowalski <thomas.kowalski@datadoghq.com>
|
This change is marked for backport to 4.9 and it does not conflict with that branch. |
|
This change is marked for backport to 4.10 and it does not conflict with that branch. |
## Description This PR ensures we clear the string tables post-fork. This is meant to address a crash happening on (at least) 4.8.0rc5. ``` Error UnixSignal: Process terminated with SEGV_MAPERR (SIGSEGV) #0 0x0000000000000000 std::__detail::_Hashtable_base<unsigned long, std::pair<unsigned long const, std::string>, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Hashtable_traits<false, false, true> >::_M_equals(unsigned long const&, unsigned long, std::__detail::_Hash_node<std::pair<unsigned long const, std::string>, false>*) const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable_policy.h:1804) #1 0x000073afd44ec124 std::__detail::_Hashtable_base<unsigned long, std::pair<unsigned long const, std::string>, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Hashtable_traits<false, false, true> >::_M_equals(unsigned long const&, unsigned long, std::__detail::_Hash_node<std::pair<unsigned long const, std::string>, false>*) const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable_policy.h:1804) #2 0x000073afd44ec124 std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::_M_find_before_node(unsigned long, unsigned long const&, unsigned long) const [clone .constprop.0] (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:1580) #3 0x000073afd44ece4b std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::_M_find_node(unsigned long, unsigned long const&, unsigned long) const [clone .isra.0] (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:694) #4 0x0000000000000000 std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:1455) #5 0x0000000000000000 std::unordered_map<unsigned long, std::string, std::hash<unsigned long>, std::equal_to<unsigned long>, std::allocator<std::pair<unsigned long const, std::string> > >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/unordered_map.h:920) #6 0x000073afd44e4e61 std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:1455) #7 0x000073afd44e4e61 std::unordered_map<unsigned long, std::string, std::hash<unsigned long>, std::equal_to<unsigned long>, std::allocator<std::pair<unsigned long const, std::string> > >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/unordered_map.h:920) #8 0x000073afd44e4e61 StringTable::key(_object*, StringTag) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/strings.cc:58) #9 0x000073afd44e517f Frame::create(EchionSampler&, PyCodeObject*, int) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/frame.cc:24) #10 0x000073afd44e5436 Frame::get(EchionSampler&, PyCodeObject*, int) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/frame.cc:214) #11 0x000073afd44e55b1 Frame::read(EchionSampler&, _PyInterpreterFrame*, _PyInterpreterFrame**) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/frame.cc:153) #12 0x000073afd44e56c0 unwind_frame(EchionSampler&, _object*, FrameStack&, unsigned long) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/stacks.cc:56) #13 0x000073afd44e5859 unwind_python_stack(EchionSampler&, _ts*, FrameStack&) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/stacks.cc:103) #14 0x000073afd44e7bbf ThreadInfo::unwind(EchionSampler&, _ts*) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/threads.cc:15) #15 0x0000000000000000 __gnu_cxx::__normal_iterator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const*, std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > > >::__normal_iterator(std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const* const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_iterator.h:979) #16 0x0000000000000000 std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::end() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:839) #17 0x0000000000000000 std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::empty() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:1008) #18 0x000073afd44eb2eb __gnu_cxx::__normal_iterator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const*, std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > > >::__normal_iterator(std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const* const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_iterator.h:979) #19 0x000073afd44eb2eb std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::end() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:839) #20 0x000073afd44eb2eb std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::empty() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:1008) #21 0x000073afd44eb2eb ThreadInfo::sample(EchionSampler&, _ts*, long) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/threads.cc:698) #22 0x000073afd44eb789 Datadog::Sampler::sampling_thread(unsigned long) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/sampler.cpp:287) #23 0x000073afd44ebad0 call_sampling_thread(void*) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/sampler.cpp:45) #24 0x000073afdf06bac3 start_thread (nptl/nptl/pthread_create.c:442) #25 0x000073afdf0fd8d0 __clone3 ``` Co-authored-by: thomas.kowalski <thomas.kowalski@datadoghq.com> (cherry picked from commit f1f6901) Co-authored-by: Thomas Kowalski <thomas.kowalski@datadoghq.com>
## Description This PR ensures we clear the string tables post-fork. This is meant to address a crash happening on (at least) 4.8.0rc5. ``` Error UnixSignal: Process terminated with SEGV_MAPERR (SIGSEGV) #0 0x0000000000000000 std::__detail::_Hashtable_base<unsigned long, std::pair<unsigned long const, std::string>, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Hashtable_traits<false, false, true> >::_M_equals(unsigned long const&, unsigned long, std::__detail::_Hash_node<std::pair<unsigned long const, std::string>, false>*) const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable_policy.h:1804) #1 0x000073afd44ec124 std::__detail::_Hashtable_base<unsigned long, std::pair<unsigned long const, std::string>, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Hashtable_traits<false, false, true> >::_M_equals(unsigned long const&, unsigned long, std::__detail::_Hash_node<std::pair<unsigned long const, std::string>, false>*) const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable_policy.h:1804) #2 0x000073afd44ec124 std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::_M_find_before_node(unsigned long, unsigned long const&, unsigned long) const [clone .constprop.0] (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:1580) #3 0x000073afd44ece4b std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::_M_find_node(unsigned long, unsigned long const&, unsigned long) const [clone .isra.0] (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:694) #4 0x0000000000000000 std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:1455) #5 0x0000000000000000 std::unordered_map<unsigned long, std::string, std::hash<unsigned long>, std::equal_to<unsigned long>, std::allocator<std::pair<unsigned long const, std::string> > >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/unordered_map.h:920) #6 0x000073afd44e4e61 std::_Hashtable<unsigned long, std::pair<unsigned long const, std::string>, std::allocator<std::pair<unsigned long const, std::string> >, std::__detail::_Select1st, std::equal_to<unsigned long>, std::hash<unsigned long>, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, false, true> >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/hashtable.h:1455) #7 0x000073afd44e4e61 std::unordered_map<unsigned long, std::string, std::hash<unsigned long>, std::equal_to<unsigned long>, std::allocator<std::pair<unsigned long const, std::string> > >::find(unsigned long const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/unordered_map.h:920) #8 0x000073afd44e4e61 StringTable::key(_object*, StringTag) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/strings.cc:58) #9 0x000073afd44e517f Frame::create(EchionSampler&, PyCodeObject*, int) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/frame.cc:24) #10 0x000073afd44e5436 Frame::get(EchionSampler&, PyCodeObject*, int) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/frame.cc:214) #11 0x000073afd44e55b1 Frame::read(EchionSampler&, _PyInterpreterFrame*, _PyInterpreterFrame**) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/frame.cc:153) #12 0x000073afd44e56c0 unwind_frame(EchionSampler&, _object*, FrameStack&, unsigned long) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/stacks.cc:56) #13 0x000073afd44e5859 unwind_python_stack(EchionSampler&, _ts*, FrameStack&) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/stacks.cc:103) #14 0x000073afd44e7bbf ThreadInfo::unwind(EchionSampler&, _ts*) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/threads.cc:15) #15 0x0000000000000000 __gnu_cxx::__normal_iterator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const*, std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > > >::__normal_iterator(std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const* const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_iterator.h:979) #16 0x0000000000000000 std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::end() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:839) #17 0x0000000000000000 std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::empty() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:1008) #18 0x000073afd44eb2eb __gnu_cxx::__normal_iterator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const*, std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > > >::__normal_iterator(std::unique_ptr<StackInfo, std::default_delete<StackInfo> > const* const&) (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_iterator.h:979) #19 0x000073afd44eb2eb std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::end() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:839) #20 0x000073afd44eb2eb std::vector<std::unique_ptr<StackInfo, std::default_delete<StackInfo> >, std::allocator<std::unique_ptr<StackInfo, std::default_delete<StackInfo> > > >::empty() const (/opt/rh/devtoolset-10/root/usr/include/c++/10/bits/stl_vector.h:1008) #21 0x000073afd44eb2eb ThreadInfo::sample(EchionSampler&, _ts*, long) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/echion/threads.cc:698) #22 0x000073afd44eb789 Datadog::Sampler::sampling_thread(unsigned long) [clone .localalias] (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/sampler.cpp:287) #23 0x000073afd44ebad0 call_sampling_thread(void*) (/go/src/github.com/DataDog/apm-reliability/dd-trace-py/ddtrace/internal/datadog/profiling/stack/src/sampler.cpp:45) #24 0x000073afdf06bac3 start_thread (nptl/nptl/pthread_create.c:442) #25 0x000073afdf0fd8d0 __clone3 ``` Co-authored-by: thomas.kowalski <thomas.kowalski@datadoghq.com> (cherry picked from commit f1f6901) Co-authored-by: Thomas Kowalski <thomas.kowalski@datadoghq.com>
Description
This PR ensures we clear the string tables post-fork. This is meant to address a crash happening on (at least) 4.8.0rc5.