Bug 181836 – REGRESSION (r226981): ASSERTION FAILED: startY >= 0 && endY <= height && startY < endY in WebCore::FEMorphology::platformApplyGeneric

• Component: New Bugs
• Status: RESOLVED FIXED

[Ryan Haddad]

This assertion failure is seen on iOS Debug WK2 with LayoutTest svg/filters/feMorphology-invalid-radius.svg:

ASSERTION FAILED: startY >= 0 && endY <= height && startY < endY

Thread 16 Crashed:: Dispatch queue: com.apple.root.default-qos
0   JavaScriptCore                	0x000000011be46ec4 WTFCrash + 36
1   com.apple.WebCore             	0x000000011ea6e45d WebCore::FEMorphology::platformApplyGeneric(WebCore::FEMorphology::PaintingData const&, int, int) + 317
2   com.apple.WebCore             	0x000000011ea6ebbe WebCore::FEMorphology::platformApplyWorker(WebCore::FEMorphology::PlatformApplyParameters*) + 46
3   com.apple.WebCore             	0x000000011ea720b0 ___ZN3WTF19ParallelEnvironment7executeEPh_block_invoke + 48
4   libdispatch.dylib             	0x00000001151dd2e6 _dispatch_client_callout2 + 8
5   libdispatch.dylib             	0x00000001151e9ebc _dispatch_apply_invoke + 164
6   libdispatch.dylib             	0x00000001151dd2b5 _dispatch_client_callout + 8
7   libdispatch.dylib             	0x00000001151e7c1b _dispatch_root_queue_drain + 993
8   libdispatch.dylib             	0x00000001151e77dc _dispatch_worker_thread3 + 119
9   libsystem_pthread.dylib       	0x00000001156985a2 _pthread_wqthread + 1299
10  libsystem_pthread.dylib       	0x000000011569807d start_wqthread + 13

https://build.webkit.org/results/Apple%20iOS%2011%20Simulator%20Debug%20WK2%20(Tests)/r227176%20(2176)/results.html

https://webkit-test-results.webkit.org/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=svg%2Ffilters%2FfeMorphology-invalid-radius.svg

[Simon Fraser (smfr)]

Weird that this happens on iOS but not macOS. Will look.

[Simon Fraser (smfr)]

Happens on macOS too.

FEMorphology::platformApplySoftware() was relying in implicit conversion between large floats and ints, which used to result in negative radius values, but now causes large positive values.

[Simon Fraser (smfr)]

This logic is bad:

            const int jobSize = paintingData.height / numOfThreads;
            const int jobsWithExtra = paintingData.height % numOfThreads;

when numOfThreads > paintingData.height, because you end up with jobSize=0 and jobsWithExtra = paintingData.height, but we loop on numOfThreads.

[Simon Fraser (smfr)]

Created attachment 331835 [details]
Patch

[Simon Fraser (smfr)]

Created attachment 331844 [details]
Patch

[Simon Fraser (smfr)]

Created attachment 331845 [details]
Patch

[EWS Watchlist]

Comment on attachment 331845 [details]
Patch

Attachment 331845 [details] did not pass win-ews (win):
Output: http://webkit-queues.webkit.org/results/6151177

New failing tests:
svg/filters/feTurbulence-parallel-jobs-wide.svg
svg/filters/feLighting-parallel-jobs.svg

[EWS Watchlist]

Created attachment 331847 [details]
Archive of layout-test-results from ews206 for win-future

The attached test failures were seen while running run-webkit-tests on the win-ews.
Bot: ews206  Port: win-future  Platform: CYGWIN_NT-6.1-2.9.0-0.318-5-3-x86_64-64bit

[Simon Fraser (smfr)]

Created attachment 331950 [details]
Patch

[EWS Watchlist]

Comment on attachment 331950 [details]
Patch

Attachment 331950 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-queues.webkit.org/results/6173128

New failing tests:
media/video-main-content-autoplay.html

[EWS Watchlist]

Created attachment 331958 [details]
Archive of layout-test-results from ews107 for mac-sierra-wk2

The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews107  Port: mac-sierra-wk2  Platform: Mac OS X 10.12.6

[Tim Horton]

Comment on attachment 331950 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=331950&action=review

> Source/WebCore/platform/graphics/filters/FEMorphology.cpp:197
> +        WTF::ParallelJobs<PlatformApplyParameters> parallelJobs(&WebCore::FEMorphology::platformApplyWorker, optimalThreadNumber);

Why the WTF::

[Ryan Haddad]

Comment on attachment 331950 [details]
Patch

Marking CQ+ again because the failure is unrelated.

[WebKit Commit Bot]

Comment on attachment 331950 [details]
Patch

Clearing flags on attachment: 331950

Committed r227374: <https://trac.webkit.org/changeset/227374>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/36757037>