-
Notifications
You must be signed in to change notification settings - Fork 4.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Linux][arm] System.Memory.Tests failed with SIGSEGV on rolling build #60825
Comments
Tagging subscribers to this area: @GrabYourPitchforks, @dotnet/area-system-memory Issue DetailsRolling build: https://dev.azure.com/dnceng/public/_build/results?buildId=1437467&view=results
in case it's needed you can get full log
|
can we get a dump? |
There's a dump file in this folder:
But since it's a Linux arm64 dump file, you need to follow the instructions in the
The I opened the dump file with WinDBG, I made sure to specify the right symbols folder, and this is what I got: Error message:
Callstack:
coreclr/gc/gc.cpp: inline unsigned int gc_heap::mark_array_marked(uint8_t* add)
{
return mark_array [mark_word_of (add)] & (1 << mark_bit_bit_of (add));
} |
Tagging subscribers to this area: @dotnet/gc Issue DetailsRolling build: https://dev.azure.com/dnceng/public/_build/results?buildId=1437467&view=results
in case it's needed you can get full log
|
Thanks @carlossanlop ! GC folks, do you believe this is GC, or more likely some memory corruption caused by a libraries bug? |
Is this a consistent repro? This looks to be some kind of heap/stack corruption. |
@krwq how often is it happening? |
Definitely not consistent repro, For rolling builds that's a single instance I've seen in the last month. I'm not sure about PRs or other pipelines. I'll see if I can find the data (runfo is crashing when I try the search) |
Jobs
| where Queued > ago (180d)
| where Source == "ci/public/dotnet/runtime/refs/heads/main"
| distinct Name
| join WorkItems on $left.Name == $right.JobName
| where FriendlyName == "System.Memory.Tests"
| where ExitCode == 139
| project WorkItemId, Finished, MachineName, QueueName//, ConsoleUri
just looking at the last 30 days, it does seem Arm32 is particularly segfaulty: Jobs
| where Queued > ago (30d)
| where Source == "ci/public/dotnet/runtime/refs/heads/main"
| distinct Name
| join WorkItems on $left.Name == $right.JobName
| where ExitCode == 139
| summarize count() by FriendlyName, QueueName
| sort by count_ desc
That might be consistent with a runtime/GC bug. |
This was hit on #64028 (unrelated PR) after running |
This was being hit early during the 7 timeframe, but the same assert hasnt reproed since Jan, so closing for now. |
Rolling build: https://dev.azure.com/dnceng/public/_build/results?buildId=1437467&view=results
in case it's needed you can get full log
net7.0-Linux-Release-arm64-CoreCLR_release-(Ubuntu.1804.ArmArch.Open)Ubuntu.1804.ArmArch.Open@mcr.microsoft.com/dotnet-buildtools/prereqs:ubuntu-16.04-helix-arm64v8-20210106155927-56c6673
The text was updated successfully, but these errors were encountered: