Repository navigation
Test failure GC\\Features\\HeapExpansion\\plug\\plug.cmd #63854
Description
Run: runtime-coreclr outerloop 20220116.2
Failed test:
CoreCLR windows arm64 Checked no_tiered_compilation @ Windows.10.Arm64v8.Open
- GC\\Features\\HeapExpansion\\plug\\plug.cmd
CoreCLR Linux arm64 Checked @ (Ubuntu.1804.Arm64.Open)Ubuntu.1804.Armarch.Open@mcr.microsoft.com/dotnet-buildtools/prereqs:ubuntu-18.04-helix-arm64v8-20210531091519-97d8652
- GC/Scenarios/ServerModel/servermodel/servermodel.sh
CoreCLR Linux arm64 Checked no_tiered_compilation @ (Alpine.314.Arm64.Open)Ubuntu.1804.ArmArch.Open@mcr.microsoft.com/dotnet-buildtools/prereqs:alpine-3.14-helix-arm64v8-20210910135810-8a6f4f3
- GC/Scenarios/ServerModel/servermodel/servermodel.sh
Error messgae:
Return code: 1
Raw output file: D:\h\w\BC070A55\w\AED00960\uploads\Reports\GC.Features\HeapExpansion\plug\plug.output.txt
Raw output:
BEGIN EXECUTION
"D:\h\w\BC070A55\p\corerun.exe" -p "System.Reflection.Metadata.MetadataUpdater.IsSupported=false" plug.dll
Running 100 iterations
Expected: 100
Actual: -2147483645
END EXECUTION - FAILED
FAILED
Test Harness Exitcode is : 1
To run the test:
set CORE_ROOT=D:\h\w\BC070A55\p
D:\h\w\BC070A55\w\AED00960\e\GC\Features\HeapExpansion\plug\plug.cmd
Expected: True
Actual: False
Stack trace
at GC_Features._HeapExpansion_plug_plug_._HeapExpansion_plug_plug_cmd()
or
createdump: /__w/1/s/src/coreclr/pal/src/libunwind/src/dwarf/Gfind_proc_info-lsb.c:929: int _Uaarch64_dwarf_search_unwind_table(unw_addr_space_t, unw_word_t, unw_dyn_info_t *, unw_proc_info_t *, int, void *): Assertion `ip >= di->start_ip && ip < di->end_ip' failed.
/root/helix/work/workitem/e/GC/Scenarios/ServerModel/servermodel/servermodel.sh: line 418: 4134 Aborted (core dumped) $LAUNCHER $ExePath "${CLRTestExecutionArguments[@]}"
Return code: 1
Raw output file: /root/helix/work/workitem/uploads/Reports/GC.Scenarios/ServerModel/servermodel/servermodel.output.txt
Raw output:
BEGIN EXECUTION
/root/helix/work/correlation/corerun -p System.Reflection.Metadata.MetadataUpdater.IsSupported=false servermodel.dll '/numrequests:100'
Using 361542648 as random seed
took 512 iteration to reach steady state
258 reqs/sec
306 reqs/sec
Expected: 100
Actual: 134
END EXECUTION - FAILED
Test Harness Exitcode is : 1
To run the test:
set CORE_ROOT=/root/helix/work/correlation
/root/helix/work/workitem/e/GC/Scenarios/ServerModel/servermodel/servermodel.sh
Expected: True
Actual: False
Stack trace
at GC_Scenarios._ServerModel_servermodel_servermodel_._ServerModel_servermodel_servermodel_sh()
Activity
janvorli commentedon Jan 17, 2022
I was looking at the libunwind assert on Friday (using the dump from CI), there is a problem where the remote unwinding code doesn't recognize a managed frame as managed and tries to unwind one more time using the libunwind. That results in 0 instruction pointer and the next call to unwinder fails with that assert due to that. Might be a DACization issue, it is not clear yet.
As for the GC issue, here is the call stack at the point of failure, unfortunately all the locals from the interesting frames were optimized out by the compiler.
janvorli commentedon Jan 17, 2022
@dotnet/gc I have found I can repro the issue reliably locally in a couple of runs of the
System.Text.Json.Testslibraries tests on macOS arm64 (I've tried that in checked build so far).janvorli commentedon Jan 17, 2022
It is reproducible in debug build as well, so all locals are preserved. I can see that at the crash time, the
mark_array_addrstarting at indexmarkwcontains multiple non-zero entries. I've dumped a bit of that memory below:janvorli commentedon Jan 17, 2022
@dotnet/gc I can share the dump from the debug build run.
mangod9 commentedon Jan 18, 2022
Hi @janvorli, could you please share the debug dump for this? Were you able to repro the failure consistently with regions enabled?
VincentBu commentedon Mar 14, 2022
Failed again in: runtime-coreclr outerloop 20220313.2
Failed test:
Error message:
mangod9 commentedon Mar 16, 2022
Should be fixed with this: #66696, will leave it open but hoping we dont see these failures in the next few days.
mangod9 commentedon Mar 21, 2022
Closing since this hasnt recurred since the fix.
JulieLeeMSFT commentedon Mar 21, 2022
Seen today Link.
mangod9 commentedon Mar 21, 2022
@Maoni0 fyi.
Maoni0 commentedon Mar 21, 2022
this was not run with my change. it's still this commit 17e435d6 which was from 7 days ago.
if you want to trigger a run with updated src you need to do /azp.
mangod9 commentedon Mar 22, 2022
Closing again.