Bump LLVM to 80603c6672226cb48798bd60120c8f859dbeca19.#10047
Bump LLVM to 80603c6672226cb48798bd60120c8f859dbeca19.#10047mikeurbach wants to merge 1 commit intomainfrom
Conversation
This is a clean LLVM bump. While testing, I did notice one circt-synth test that was not gated behind a libz3 requirement, so I added that.
|
I think we've got competing bumps, now! Does this subsume: #10010 ? |
|
This subsumes so let's use this one. |
| @@ -1,3 +1,5 @@ | |||
| // REQUIRES: libz3 | |||
|
Results of circt-tests run for 11ffa1f compared to results for 93c30ea: sv-testsChanges in emitted diagnostics:
Introduced 304 segfaults:
|
|
Bah sorry @uenoku I didn't realize your PR was open. |
|
I guess the Windows CI check was shown to be flaky already so I'm retrying 😕 But the circt-tests report an short integration tests seem to point to a real issue in circt-verilog... @fabianschuiki does anything ring a bell here? |
|
Huh this is very weird. I wouldn't expect there to be that many changes in circt-verilog due to the bump. Let me dig… |
|
The fact that this now triggers an exception is very weird 🤔 And also, why does this not happen on the actual build-and-test runs, only in short integration? |
|
Yeah I am surprised. Maybe it is a different flavor of the wonkiness we saw in the Windows build, now cropping up in Linux? |
|
Hmmm yeah that could be it. The exception smells a bit like it's some file access that might fail for whatever reson 😢 |
|
Let me rerun the short integration test to see if it's reproducibly failing at the exact same point, or if it fails in a different place. |
This is a clean LLVM bump.
While testing, I did notice one circt-synth test that was not gated behind a libz3 requirement, so I added that.