Bug 214763 – REGRESSION (r264845): webgl/2.0.0/conformance2/textures/misc/tex-unpack-params.html is failing

• Status: ASSIGNED
• Component: WebGL

[Ryan Haddad]

webgl/2.0.0/conformance2/textures/misc/tex-unpack-params.html is failing after the rebaseline in https://trac.webkit.org/changeset/264845

https://results.webkit.org/?suite=layout-tests&test=webgl%2F2.0.0%2Fconformance2%2Ftextures%2Fmisc%2Ftex-unpack-params.html

--- /Volumes/Data/slave/catalina-release-tests-wk2/build/layout-test-results/webgl/2.0.0/conformance2/textures/misc/tex-unpack-params-expected.txt
+++ /Volumes/Data/slave/catalina-release-tests-wk2/build/layout-test-results/webgl/2.0.0/conformance2/textures/misc/tex-unpack-params-actual.txt
@@ -1,5 +1,990 @@
 This test runs the WebGL Test listed below in an iframe and reports PASS or FAIL.
 
 Test: ../../../resources/webgl_test_files/conformance2/textures/misc/tex-unpack-params.html
-[ PASS ] All tests passed
+[ 266: FAIL ] compare pixels: 32 pixels differ
+[ 340: FAIL ] compare pixels: 70 pixels differ
+[ 345: FAIL ] compare pixels: 12 pixels differ
+[ 363: FAIL ] compare pixels: 84 pixels differ
+[ 368: FAIL ] compare pixels: 20 pixels differ
+[ 386: FAIL ] compare pixels: 98 pixels differ
+[ 391: FAIL ] compare pixels: 15 pixels differ
+[ 409: FAIL ] compare pixels: 112 pixels differ
+[ 414: FAIL ] compare pixels: 24 pixels differ
+[ 765: FAIL ] compare pixels: 16 pixels differ
+[ 835: FAIL ] compare pixels: 70 pixels differ
+[ 840: FAIL ] compare pixels: 12 pixels differ
+[ 858: FAIL ] compare pixels: 84 pixels differ
+[ 863: FAIL ] compare pixels: 20 pixels differ
+[ 881: FAIL ] compare pixels: 49 pixels differ
+[ 904: FAIL ] compare pixels: 112 pixels differ

[Radar WebKit Bug Importer]

<rdar://problem/66075979>

[Ryan Haddad]

I truncated the diff since it is quite long.. full one here: https://build.webkit.org/results/Apple-Catalina-Release-WK2-Tests/r264848%20(7369)/webgl/2.0.0/conformance2/textures/misc/tex-unpack-params-diff.txt

[Ryan Haddad]

I suppose this is still a progression since there were 206 subtests failing before, now they are only 16.

[Kenneth Russell]

Sorry about the regression. I'll suppress the failure while investigating.

[Kenneth Russell]

BTW, this test passed on my MacBook Pro - that's where the rebaseline was done. Do you know how the failure made it through the commit queue?

[Kenneth Russell]

Created attachment 405193 [details]
Patch

[Kenneth Russell]

Created attachment 405194 [details]
Patch

[Ryan Haddad]

(In reply to Kenneth Russell from comment #5)
> BTW, this test passed on my MacBook Pro - that's where the rebaseline was
> done. Do you know how the failure made it through the commit queue?
Maybe it is hardware dependent? The CQ runs on MacPro6,1 hardware and the post-commit testing is with MacMini8,1 (only integrated graphics).

[Ryan Haddad]

Comment on attachment 405194 [details]
Patch

Clearing flags on attachment: 405194

Committed r264865: <https://trac.webkit.org/changeset/264865>

[Kenneth Russell]

Yes, possible this test passes on discrete GPUs on macOS and not Intel integrated ones. It's also possible the test was revised after WebKit's snapshot was taken. I'll investigate.