-
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
[WASI][AOT] LLVM ERROR: out of memory #101533
Comments
If this is with the 'vanilla' clang from WASI SDK (it looks to be like that from the logs), this may be WebAssembly/wasi-sdk#326. |
It is. The first recorded failure was in #100993 and the second was in a rolling build with that commit in it so I wonder if we're hitting a new path (for mono) that is triggering that failure. The other possibility is some trimming changes that may have exposed it. |
|
see also #99150 |
Tagging subscribers to this area: @lambdageek, @steveisok |
@fanyang-mono this is likely instrinics related |
Ah, that's interesting. CI for that PR was run several times, with the very last one just being a rerun for extra precaution before P4 and all of the CI failures had existing issues attributed to them (in this case, it looks like it was attributed to #99150). Given the test the OP is referencing, it doesn't look like the issue had shown up in the prior CI runs either. At a glance: https://github.com/dotnet/runtime/runs/24118131821 (which is on top of 11b53b2) doesn't have S.C.SpecializedTests failing vs https://github.com/dotnet/runtime/runs/24176375150 (which is on top of 40bc2d8) which did. There's an 80 commit difference made to main between those two: 11b53b2...40bc2d8 (but notably no change to the PR code itself, the same was true for the commit from several days before that which likewise just merged Is this one not consistently failing perhaps? |
I remember Zoltan turned on dead code elimination in mono when AOTing llvm bitcode. Was that only for browser-wasm, or all llvm? is it possible we didn't turn it on for wasi? |
Instead of worrying about what caused it to start failing recently the bundled clang is definitely still 32bit (as @SingleAccretion |
The clang in the wasi-sdk is a 32bit exe and it is failing to do aot compilation correctly so disable the lane until that is fixed dotnet#101533
opened #101827 to disable the failing bits in the meantime |
The clang in the wasi-sdk is a 32bit exe and it is failing to do aot compilation correctly so disable the lane until that is fixed dotnet#101533
reopening to reenable the disabled tests and verify the fix
|
The clang in the wasi-sdk is a 32bit exe and it is failing to do aot compilation correctly so disable the lane until that is fixed dotnet#101533
The clang in the wasi-sdk is a 32bit exe and it is failing to do aot compilation correctly so disable the lane until that is fixed dotnet#101533
we should try enabling this with the wasi sdk update |
when we get to LLVM 19, right ? |
|
ah, ok. |
Not hit for a week since enabling while previously it had a frequency of hundreds of hits per week. |
Note both LLVM 17 and 18 have the same problem with AOT of
Invariant.Tests
Log
Log
Build Information
Build: https://dev.azure.com/dnceng-public/public/_build/results?buildId=655571
Build error leg or test failing:
Error Message
Fill the error message using step by step known issues guidance.
Known issue validation
Build: 🔎 https://dev.azure.com/dnceng-public/public/_build/results?buildId=655571
Error message validated:
[LLVM ERROR: out of memory
]Result validation: ✅ Known issue matched with the provided build.
Validation performed at: 4/25/2024 9:35:23 AM UTC
Report
Summary
The text was updated successfully, but these errors were encountered: