[Lavapipe] Mark tests affected by gallivm APInt assertion by joaosaffran · Pull Request #1580 · llvm/offload-test-suite

joaosaffran · 2026-10-14T00:14:25Z

Marking test affected by issue https://gitlab.freedesktop.org/mesa/mesa/-/work_items/16596 as XFAIL

farzonl · 2026-10-14T17:37:16Z

+# Driver bug: https://github.com/llvm/offload-test-suite/issues/1553
 # UNSUPPORTED: Lavapipe


this is where things become unclear. because you have been using #1553 to indicate # XFAIL: Lavapipe but now you are doing # UNSUPPORTED: Lavapipe. That is a bit of inconsistency that is hard for me to wrap my head around.

Issue #1552 is flaky, issue #1553 is not flaky. This, and some other tests, are being affected by both issues. My logic is: If a test is flaky, use UNSUPPORTED. Otherwise, if #1553 is fixed before #1552, we will see this test XPASS half the time it runs and generate noise to whoever is on triage. Let me know if there is a better approach for this.

Can we do something like:

# Driver bug: https://github.com/llvm/offload-test-suite/issues/1552 (flakey) # UNSUPPORTED: Lavapipe # https://github.com/llvm/offload-test-suite/issues/1553 # XFAIL: Lavapipe

farzonl · 2026-10-19T17:26:02Z

 # UNSUPPORTED: WARP && DXC

+# Driver Bug: https://gitlab.freedesktop.org/api/v4/projects/mesa%2Fmesa/issues/16596
+# XFAIL: Lavapipe && AMD && DXC


WAIT I don't think AMD means what you think it means. AMD is a GPU. Lavapipe && AMD makes no sense it is one or the other. I think you are trying to distinguish ARM from x86 which the right way would be to look at how we do platform.machine() in lit. I see things like XFAIL: Lavapipe && host-arm64 We may need to add one for x86/x86_64.

Yeah sorry, you are right, I made some confusion indeed. Is !host-arm64 the same as saying x86/x86_64 in this scenario?

No it would not be. LLVM runs on RISC-V among other CPU hardwares as does lavapipe. A linux sku could technically run our offload test suite.

I would prefer if we just future proof things and explicilty check for x86.

I've pushed this new PR to the stack: #1613, which adds the requested x86 targets

farzonl

All of these are some derivative of Lavapipe && AMD. That needs to be fix before you take this out of draft.

damyanp · 2026-10-26T21:57:27Z

Looks like the PR description needs to be updated since it currently refers to the wrong issue.

damyanp · 2026-10-26T21:58:42Z

 # XFAIL: Clang && Vulkan

+# Driver Bug: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/16596
+# XFAIL: Lavapipe && (host-x86 || host-x86_64) && DXC


Why && DXC? Presumably once #1435 is completed this would start failing on clang as well?

Yeah, most likely. I will remove the && DXC here

Stacked on the #1552 branch. Tests affected only by #1553 are XFAIL: Lavapipe. Tests also affected by #1552 stay UNSUPPORTED: Lavapipe and link both issues. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 643c3c11-3596-47b0-acc4-1fa90c8e4d85

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 643c3c11-3596-47b0-acc4-1fa90c8e4d85

Annotate 107 tests for Mesa issue 16596: 53 fail with both compilers, 52 only with DXC, and 2 only with Clang. Scope the XFAILs to the requested Lavapipe and AMD features. Remove obsolete issue annotations and inherited no-output access-violation exclusions that are absent from both logs. The AMD feature currently identifies AMD GPU descriptions, not the llvmpipe host runner; these requested conditions require AMD to be supplied separately for that runner. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: ea4455aa-9cd1-48c2-8568-2a6d5786a339

joaosaffran · 2026-10-27T00:38:22Z

 # XFAIL: Metal

+# Driver Bug: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/16596
+# XFAIL: Lavapipe && (host-x86 || host-x86_64) && DXC


This test is weird; it says clang doesn't implement TextureCubeArray, but it doesn't xfail either. So I think the right approach here would be to XFAIL on clang as well, since it is not implemented. But that change is not in scope with this patch. I am happy to address it in a follow-up PR if reviewers agree with this fix.

joaosaffran mentioned this pull request Oct 14, 2026

[Lavapipe] Mark tests affected by intermittent access violation (#1552) #1575

Closed

joaosaffran added this pull request to stack #1582 October 14, 2026 00:14

joaosaffran marked this pull request as ready for review October 14, 2026 00:18

joaosaffran added the test-lavapipe Run lavapipe tests, only label Oct 14, 2026

joaosaffran marked this pull request as draft October 14, 2026 00:19

joaosaffran marked this pull request as ready for review October 14, 2026 00:19

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch from da0a186 to bd3b087 Compare October 14, 2026 00:21

joaosaffran mentioned this pull request Oct 14, 2026

[Lavapipe] Mark ray tracing tests affected by SBT alignment bug #1583

Closed

joaosaffran changed the title ~~[Lavapipe] Mark tests affected by gallivm APInt assertion (#1553)~~ [Lavapipe] Mark tests affected by gallivm APInt assertion Oct 14, 2026

joaosaffran marked this pull request as draft October 14, 2026 00:34

joaosaffran removed this pull request from stack #1582 October 14, 2026 00:42

joaosaffran mentioned this pull request Oct 14, 2026

Require goldenimage for mesh shader tests #1584

Open

joaosaffran added this pull request to stack #1592 October 14, 2026 00:43

joaosaffran marked this pull request as ready for review October 14, 2026 00:43

farzonl reviewed Oct 14, 2026

View reviewed changes

Comment thread test/Feature/RT/closest-hit-barycentrics.test Outdated

farzonl reviewed Oct 14, 2026

View reviewed changes

joaosaffran requested a review from farzonl October 14, 2026 20:41

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch from 69a7238 to bd3b087 Compare October 15, 2026 22:20

joaosaffran marked this pull request as draft October 15, 2026 22:20

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch from bd3b087 to 8a88c41 Compare October 15, 2026 22:23

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch 3 times, most recently from 8ca7acf to 8c49f95 Compare October 17, 2026 01:48

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch from 8c49f95 to db1828b Compare October 17, 2026 01:54

joaosaffran requested a review from damyanp October 17, 2026 01:55

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch from db1828b to eb111e5 Compare October 19, 2026 17:16

farzonl reviewed Oct 19, 2026

View reviewed changes

farzonl requested changes Oct 19, 2026

View reviewed changes

joaosaffran changed the base branch from users/joaosaffran/lavapipe-access-violation-1552 to users/joaosaffran/create-x86-host October 20, 2026 00:11

joaosaffran added this pull request to stack #1614 October 20, 2026 00:11

joaosaffran requested a review from farzonl October 20, 2026 00:12

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch from 3edcd26 to 853c2ad Compare October 20, 2026 00:48

joaosaffran marked this pull request as ready for review October 20, 2026 22:31

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch from 625425e to e5eecef Compare October 21, 2026 23:52

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch from e5eecef to 625425e Compare October 22, 2026 20:39

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch 2 times, most recently from 625425e to 3a1c567 Compare October 22, 2026 20:42

damyanp reviewed Oct 26, 2026

View reviewed changes

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch from 16f6b1f to 7f04d62 Compare October 27, 2026 00:06

joaosaffran-zz and others added 11 commits October 26, 2026 17:33

Retrigger CI

0cc1962

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Copilot-Session: 643c3c11-3596-47b0-acc4-1fa90c8e4d85

fixup! Match Lavapipe APInt XFAILs to AMD runner logs

c5beac8

fixup! Match Lavapipe APInt XFAILs to AMD runner logs

6fe8a48

change host to x86

699745c

add DXC xfails

aa0c1a4

remove DXC xpass

8b8e518

fix xfails

ae94bf9

add one more test failure

124faf4

remove DXC specific failures

556ac4d

joaosaffran force-pushed the users/joaosaffran/lavapipe-gallivm-assert-1553 branch from 7f04d62 to 556ac4d Compare October 27, 2026 00:35

joaosaffran requested a review from damyanp October 27, 2026 00:38

joaosaffran commented Oct 27, 2026

View reviewed changes

Navigation Menu

Repository navigation

[Lavapipe] Mark tests affected by gallivm APInt assertion - #1580

[Lavapipe] Mark tests affected by gallivm APInt assertion#1580
joaosaffran wants to merge 11 commits into
users/joaosaffran/create-x86-hostfrom
users/joaosaffran/lavapipe-gallivm-assert-1553

joaosaffran commented Oct 14, 2026 •

edited

Loading

Uh oh!

Uh oh!

farzonl Oct 14, 2026

Uh oh!

joaosaffran Oct 14, 2026

Uh oh!

damyanp Oct 15, 2026

Uh oh!

farzonl Oct 19, 2026

Uh oh!

joaosaffran Oct 19, 2026 •

edited

Loading

Uh oh!

farzonl Oct 19, 2026

Uh oh!

joaosaffran Oct 20, 2026

Uh oh!

farzonl left a comment

Uh oh!

damyanp commented Oct 26, 2026

Uh oh!

damyanp Oct 26, 2026

Uh oh!

joaosaffran Oct 27, 2026

Uh oh!

joaosaffran Oct 27, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

		# Driver bug: https://github.com/llvm/offload-test-suite/issues/1553
		# UNSUPPORTED: Lavapipe

Conversation

joaosaffran commented Oct 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

joaosaffran Oct 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

farzonl left a comment

Choose a reason for hiding this comment

Uh oh!

damyanp commented Oct 26, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

joaosaffran commented Oct 14, 2026 •

edited

Loading

joaosaffran Oct 19, 2026 •

edited

Loading