SPMI: More information on CI failures#77361
Conversation
We've seen some SPMI diffs fail due to the JitStdOutFile not existing after running the subprocess. This throws an error with some more information in those case.
|
Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch Issue DetailsWe've seen some SPMI diffs fail due to the JitStdOutFile not existing after running the subprocess. This throws an error with some more information in those case.
|
BruceForstall
left a comment
There was a problem hiding this comment.
I'll go ahead and approve in case I'm wrong about needing to remove the existing proc.communicate line, or in case you can fix it and merge with the fix long before I wake up so we can make progress investigating this issue.
|
@BruceForstall I've updated this PR with some more enhancements, can you please take another look?
|
BruceForstall
left a comment
There was a problem hiding this comment.
LGTM.
When you said "fail on superpmi sub process failure" I thought maybe you'd fixed #56506. Oh well :-)
Afraid not. I think the ideal way to fix that is if we can start using threads instead of processes. Would simplify a lot of other things, too, like if we want to add compression support. |
This enhances parts of the SPMI process with more information on failures.
Also fixes a bug I introduced when I added tpdiffs: when
asmdiffsfails, it was not marking the entire pipeline as failing.