Skip to content

Conversation

@AllanZyne
Copy link
Contributor

@AllanZyne AllanZyne commented Oct 25, 2024

fix #15461

  • Update lit config to make sure tests will run on cpu or gpu device
  • disable GEN12 test

@AllanZyne AllanZyne force-pushed the review/yang/fix_asan_tests branch from be4cf3d to 6543647 Compare October 25, 2024 03:26
@AllanZyne AllanZyne removed request for a team October 25, 2024 03:27
@AllanZyne AllanZyne changed the title [DeviceAsan][NFC] Fix ASan e2e tests, enable CPU or DG2 device only [DeviceAsan][NFC] Fix ASan e2e tests, enable CPU or GPU tests Oct 25, 2024
@@ -1,4 +1,4 @@
// REQUIRES: linux
// REQUIRES: linux, cpu || (gpu && level_zero)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

level_zero implies gpu, right?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No. Intel NPU device is exposed as Level Zero device as well.

@AllanZyne
Copy link
Contributor Author

@AllanZyne - Runtime review isn't needed, so you have all the approvals you need.

That said, before we can merge this, you need to remove sycl/test-e2e/AddressSanitizer/nullpointer/global_nullptr.cpp from sycl/test/e2e_test_requirements/no-unsupported-without-info.cpp and decrement the expected number of unsupported tests in that file. That may require runtime approval, so I'll give you that preemptively.

Thanks, updated.

@AllanZyne
Copy link
Contributor Author

@intel/llvm-gatekeepers, please merge, thanks!

@AllanZyne
Copy link
Contributor Author

If I remember correctly, I pushed the last commit yesterday.
I think the CI test is hang.

@steffenlarsen
Copy link
Contributor

If I remember correctly, I pushed the last commit yesterday. I think the CI test is hang.

CI failed, so I restarted it recently.

@steffenlarsen
Copy link
Contributor

Tests do indeed seem to be running for a long time. @intel/llvm-gatekeepers - Is this something anyone else have observed? If not, we will have to assume that it is caused by the tests being enabled here, until we observe otherwise.

@sarnex
Copy link
Contributor

sarnex commented Oct 31, 2024

Probably there is some hanging test here, we're not seeing this on other PRs. I cancelled the job here to not hog the CI.

@AllanZyne @intel/dpcpp-sanitizers-review Can someone try to reproduce a Gen12 Linux hang introduced by this PR? Thanks

@yingcong-wu
Copy link
Contributor

Hi @sarnex @AllanZyne , We noticed that Device Address Sanitizer run into some weird problem with gen 12 and we suspect the problem is with IGC codegen. Therefore I suggest that we skip running on gen 12 for the moment.

@AllanZyne
Copy link
Contributor Author

Hi @sarnex @AllanZyne , We noticed that Device Address Sanitizer run into some weird problem with gen 12 and we suspect the problem is with IGC codegen. Therefore I suggest that we skip running on gen 12 for the moment.

Sure, thank you for your investigation!

@sarnex
Copy link
Contributor

sarnex commented Nov 4, 2024

Still seeing timeouts on Linux gen12 it seems.

@AllanZyne
Copy link
Contributor Author

AllanZyne commented Nov 22, 2024

Still seeing timeouts on Linux gen12 it seems.

I have disable GEN12 test now.
I tested them on GEN12, and it seems there're some issues in igc/level0.

@AllanZyne
Copy link
Contributor Author

Hi @intel/llvm-gatekeepers, please merge. Thanks.

@martygrant martygrant merged commit 48730da into sycl Nov 25, 2024
13 checks passed
@AllanZyne AllanZyne deleted the review/yang/fix_asan_tests branch November 25, 2024 09:29
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Many AddressSanitizer fails on OCL CPU in Nightly

9 participants