Bug 174272 - LayoutTest inspector/canvas/css-canvas-clients.html is a flaky timeout
Summary: LayoutTest inspector/canvas/css-canvas-clients.html is a flaky timeout
Status: NEW
Alias: None
Product: WebKit
Classification: Unclassified
Component: Web Inspector (show other bugs)
Version: WebKit Nightly Build
Hardware: All All
: P2 Normal
Assignee: Devin Rousso
URL:
Keywords: InRadar
Depends on: 173965
Blocks:
  Show dependency treegraph
 
Reported: 2017-07-07 13:50 PDT by Devin Rousso
Modified: 2021-07-16 10:59 PDT (History)
15 users (show)

See Also:


Attachments
[Patch] Remove TestExpectation (991 bytes, patch)
2019-05-11 14:02 PDT, Devin Rousso
ews-watchlist: commit-queue-
Details | Formatted Diff | Diff
Archive of layout-test-results from ews100 for mac-highsierra (3.19 MB, application/zip)
2019-05-11 14:54 PDT, EWS Watchlist
no flags Details
Archive of layout-test-results from ews114 for mac-highsierra (2.88 MB, application/zip)
2019-05-11 15:59 PDT, EWS Watchlist
no flags Details
Archive of layout-test-results from ews215 for win-future (13.47 MB, application/zip)
2019-05-11 19:01 PDT, EWS Watchlist
no flags Details
[Patch] Extra Logging (6.49 KB, patch)
2019-05-12 13:10 PDT, Devin Rousso
hi: commit-queue-
Details | Formatted Diff | Diff
[Patch] Extra Logging (6.49 KB, patch)
2019-05-12 20:01 PDT, Devin Rousso
hi: commit-queue-
Details | Formatted Diff | Diff
Archive of layout-test-results from ews102 for mac-highsierra (3.21 MB, application/zip)
2019-05-12 21:06 PDT, EWS Watchlist
no flags Details
Archive of layout-test-results from ews114 for mac-highsierra (3.04 MB, application/zip)
2019-05-12 21:58 PDT, EWS Watchlist
no flags Details
[Patch] Extra Logging (8.86 KB, patch)
2019-05-12 23:03 PDT, Devin Rousso
hi: commit-queue-
Details | Formatted Diff | Diff
[Patch] Extra Logging (9.58 KB, patch)
2019-05-12 23:05 PDT, Devin Rousso
hi: commit-queue-
Details | Formatted Diff | Diff
Archive of layout-test-results from ews100 for mac-highsierra (3.23 MB, application/zip)
2019-05-13 00:17 PDT, EWS Watchlist
no flags Details
[Patch] Extra Logging (10.87 KB, patch)
2019-05-13 00:47 PDT, Devin Rousso
hi: commit-queue-
Details | Formatted Diff | Diff
[Patch] Extra Logging (11.23 KB, patch)
2019-05-13 00:52 PDT, Devin Rousso
hi: commit-queue-
Details | Formatted Diff | Diff
[Patch] Extra Logging (11.83 KB, patch)
2019-05-13 00:57 PDT, Devin Rousso
hi: commit-queue-
Details | Formatted Diff | Diff
[Patch] Extra Logging (11.24 KB, patch)
2019-05-13 00:57 PDT, Devin Rousso
hi: commit-queue-
Details | Formatted Diff | Diff
Archive of layout-test-results from ews105 for mac-highsierra-wk2 (3.09 MB, application/zip)
2019-05-13 01:54 PDT, EWS Watchlist
no flags Details
Archive of layout-test-results from ews101 for mac-highsierra (3.11 MB, application/zip)
2019-05-13 02:04 PDT, EWS Watchlist
no flags Details
[Patch] Extra Logging (11.90 KB, patch)
2019-05-13 02:36 PDT, Devin Rousso
hi: commit-queue-
Details | Formatted Diff | Diff
[Patch] Extra Logging (10.84 KB, patch)
2019-05-13 11:57 PDT, Devin Rousso
no flags Details | Formatted Diff | Diff
Patch (7.06 KB, patch)
2019-05-13 13:29 PDT, Devin Rousso
no flags Details | Formatted Diff | Diff
Archive of layout-test-results from ews213 for win-future (13.41 MB, application/zip)
2019-05-14 09:24 PDT, EWS Watchlist
no flags Details
Update Test Expectations (1.73 KB, patch)
2021-07-16 10:51 PDT, Eric Hutchison
no flags Details | Formatted Diff | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Devin Rousso 2017-07-07 13:50:41 PDT
LayoutTest inspector/canvas/css-canvas-clients.html is a flaky timeout

https://webkit-queues.webkit.org/results/4046639
Comment 2 Radar WebKit Bug Importer 2019-04-08 13:35:28 PDT
<rdar://problem/49709077>
Comment 3 Shawn Roberts 2019-04-08 13:49:27 PDT
Skipped test in https://trac.webkit.org/changeset/244041/webkit while waiting for a fix.
Comment 4 Shawn Roberts 2019-04-09 11:08:14 PDT
Unskipped test and just marked as a flaky failure since timeout no longer occurs, it just fails due to code changes.

https://trac.webkit.org/changeset/244084/webkit
Comment 5 Devin Rousso 2019-05-11 14:02:37 PDT
Created attachment 369660 [details]
[Patch] Remove TestExpectation

This is passing for me, so let's see what the bots think
Comment 6 EWS Watchlist 2019-05-11 14:54:40 PDT Comment hidden (obsolete)
Comment 7 EWS Watchlist 2019-05-11 14:54:41 PDT Comment hidden (obsolete)
Comment 8 EWS Watchlist 2019-05-11 15:59:19 PDT Comment hidden (obsolete)
Comment 9 EWS Watchlist 2019-05-11 15:59:21 PDT Comment hidden (obsolete)
Comment 10 EWS Watchlist 2019-05-11 19:01:10 PDT Comment hidden (obsolete)
Comment 11 EWS Watchlist 2019-05-11 19:01:12 PDT Comment hidden (obsolete)
Comment 12 Devin Rousso 2019-05-12 13:10:04 PDT
Created attachment 369682 [details]
[Patch] Extra Logging
Comment 13 Devin Rousso 2019-05-12 20:01:08 PDT
Created attachment 369689 [details]
[Patch] Extra Logging
Comment 14 EWS Watchlist 2019-05-12 21:06:47 PDT Comment hidden (obsolete)
Comment 15 EWS Watchlist 2019-05-12 21:06:49 PDT Comment hidden (obsolete)
Comment 16 EWS Watchlist 2019-05-12 21:58:24 PDT Comment hidden (obsolete)
Comment 17 EWS Watchlist 2019-05-12 21:58:26 PDT Comment hidden (obsolete)
Comment 18 Devin Rousso 2019-05-12 23:03:29 PDT
Created attachment 369696 [details]
[Patch] Extra Logging
Comment 19 Devin Rousso 2019-05-12 23:05:36 PDT
Created attachment 369697 [details]
[Patch] Extra Logging
Comment 20 EWS Watchlist 2019-05-13 00:17:57 PDT Comment hidden (obsolete)
Comment 21 EWS Watchlist 2019-05-13 00:17:59 PDT Comment hidden (obsolete)
Comment 22 Devin Rousso 2019-05-13 00:32:46 PDT
Comment on attachment 369697 [details]
[Patch] Extra Logging

View in context: https://bugs.webkit.org/attachment.cgi?id=369697&action=review

> Source/WebCore/css/CSSImageGeneratorValue.cpp:81
>          if (HTMLCanvasElement* canvasElement = downcast<CSSCanvasValue>(this)->element())

It looks like this line is the primary issue.  `WebCore::CSSCanvasValue::element()` just fetches the stored `m_element` value, which (in this case) hasn't been initialized yet.

Another (possibly bigger) issue is that `WebCore::CSSCanvasValue::element(Document&)` never gets called, which means we never add an observer.  Not sure why that is.
Comment 23 Devin Rousso 2019-05-13 00:47:37 PDT
Created attachment 369703 [details]
[Patch] Extra Logging
Comment 24 Devin Rousso 2019-05-13 00:48:47 PDT
Comment on attachment 369703 [details]
[Patch] Extra Logging

View in context: https://bugs.webkit.org/attachment.cgi?id=369703&action=review

> Source/WebCore/css/CSSCanvasValue.cpp:46
> +        if (auto* canvasElement = element(renderer.document()))

This is probably "wrong", in that we shouldn't be eagerly creating the HTMLCanvasElement _only_ when Web Inspector is open.  I'm curious to see if this will fix the test, however.
Comment 25 Devin Rousso 2019-05-13 00:52:58 PDT
Created attachment 369704 [details]
[Patch] Extra Logging
Comment 26 Devin Rousso 2019-05-13 00:57:03 PDT
Created attachment 369705 [details]
[Patch] Extra Logging
Comment 27 Devin Rousso 2019-05-13 00:57:52 PDT
Created attachment 369706 [details]
[Patch] Extra Logging
Comment 28 EWS Watchlist 2019-05-13 01:54:54 PDT Comment hidden (obsolete)
Comment 29 EWS Watchlist 2019-05-13 01:54:56 PDT Comment hidden (obsolete)
Comment 30 EWS Watchlist 2019-05-13 02:04:15 PDT Comment hidden (obsolete)
Comment 31 EWS Watchlist 2019-05-13 02:04:16 PDT Comment hidden (obsolete)
Comment 32 Devin Rousso 2019-05-13 02:36:31 PDT
Created attachment 369720 [details]
[Patch] Extra Logging
Comment 33 Devin Rousso 2019-05-13 11:57:33 PDT
Created attachment 369758 [details]
[Patch] Extra Logging
Comment 34 Devin Rousso 2019-05-13 13:29:14 PDT
Created attachment 369775 [details]
Patch

This feels "wrong" in that Web Inspector shouldn't be forcing/eagerly-creating anything.  It's even weirder that it's only an issue in debug...
Comment 35 EWS Watchlist 2019-05-14 09:24:32 PDT Comment hidden (obsolete)
Comment 36 EWS Watchlist 2019-05-14 09:24:35 PDT Comment hidden (obsolete)
Comment 37 Devin Rousso 2019-06-06 17:24:30 PDT
Comment on attachment 369775 [details]
Patch

I don't think this is the right way forward.
Comment 38 Eric Hutchison 2021-07-16 10:36:13 PDT
Timeouts recurring on all Mac platforms as of r279104:

Diff:
--- /Volumes/Data/worker/bigsur-debug-applesilicon-tests-wk2/build/layout-test-results/inspector/canvas/requestClientNodes-css-expected.txt
+++ /Volumes/Data/worker/bigsur-debug-applesilicon-tests-wk2/build/layout-test-results/inspector/canvas/requestClientNodes-css-actual.txt
@@ -1,13 +1,5 @@
-Test that CanvasAgent tracks changes in the client nodes of a CSS canvas.
+#PID UNRESPONSIVE - WebKitTestRunner (pid 59241)
+FAIL: Timed out waiting for notifyDone to be called
 
-
-== Running test suite: Canvas.requestClientNodes.CSS
--- Running test case: Canvas.requestClientNodes.CSS.Create
-PASS: Canvas with created client should have CSS name "css-canvas".
-PASS: There should be one client node.
-PASS: Client node "div" is valid.
-
--- Running test case: Canvas.requestClientNodes.CSS.Destroy
-PASS: Canvas with destroyed client should have CSS name "css-canvas".
-PASS: There should be no client nodes.
-
+#EOF
+#EOF
Comment 39 Eric Hutchison 2021-07-16 10:51:45 PDT
Created attachment 433685 [details]
Update Test Expectations
Comment 40 Robert Jenner 2021-07-16 10:59:19 PDT
Comment on attachment 433685 [details]
Update Test Expectations

Clearing flags on attachment: 433685

Committed r279989 (239732@main): <https://commits.webkit.org/239732@main>