Bug 115601 – WebProcess consuming very high CPU on linkedin.com

• Component: WebKit2
• Status: RESOLVED FIXED

[Luc Jallois]

Whenever I'm opening www.linkedin.com the "SafariForWebKitDevelopment Web Content" process is consuming very high CPU (between 50 and 100%).

[Alexey Proskuryakov]

I think that this is the same as <rdar://problem/13794359> (Apple internal link). Zalan?

[alan]

(In reply to comment #1)
> I think that this is the same as <rdar://problem/13794359> (Apple internal link). Zalan?

Yes it is the same (just checked, that soon after finishing the page load, setTimeout(f(), 300) hits and creates clones. Within seconds, we've got over 100K objects with the same setTimeout(300)-creating-clones pattern -> high CPU usage.)

[alan]

Created attachment 201328 [details]
Patch

[Build Bot]

Comment on attachment 201328 [details]
Patch

Attachment 201328 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-queues.appspot.com/results/317088

[alan]

The build system does not see that JSDOMWindow.h/.cpp needs updating. 
DOMWindow.idl has the corresponding functions behind flags, but not the generated code in JSDOMWindow.h/.cpp
Any thoughts on how to fix this other than deleting JSDOMWindow* in DervidedSoruces?

[Build Bot]

Comment on attachment 201328 [details]
Patch

Attachment 201328 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.appspot.com/results/458065

New failing tests:
fast/js/global-constructors-attributes.html

[Build Bot]

Created attachment 201337 [details]
Archive of layout-test-results from webkit-ews-06 for mac-mountainlion

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: webkit-ews-06  Port: mac-mountainlion  Platform: Mac OS X 10.8.2

[Ryosuke Niwa]

Comment on attachment 201328 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=201328&action=review

> Source/JavaScriptCore/ChangeLog:10
> +        Turn off window.performance and performance.now(). Some JS frameworks expect
> +        additional Web Timing APIs, when performance.now() is available.

That's unfortunate. Can we add some internals method to give access to performance.now()?
performance tests ran by run-perf-tests uses performance.now() and going back to Date.now() is going to make the tests less reliable.

[alan]

(In reply to comment #8)
> (From update of attachment 201328 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=201328&action=review
> 
> > Source/JavaScriptCore/ChangeLog:10
> > +        Turn off window.performance and performance.now(). Some JS frameworks expect
> > +        additional Web Timing APIs, when performance.now() is available.
> 
> That's unfortunate. Can we add some internals method to give access to performance.now()?
> performance tests ran by run-perf-tests uses performance.now() and going back to Date.now() is going to make the tests less reliable.
we could easily mimic performance.now() in internals as it simply returns a monotonic time (since navigation start)

[alan]

Created attachment 201436 [details]
Patch

[Build Bot]

Comment on attachment 201436 [details]
Patch

Attachment 201436 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.appspot.com/results/405169

New failing tests:
fast/js/global-constructors-attributes.html

[Build Bot]

Created attachment 201445 [details]
Archive of layout-test-results from webkit-ews-04 for mac-mountainlion

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: webkit-ews-04  Port: mac-mountainlion  Platform: Mac OS X 10.8.2

[Build Bot]

Comment on attachment 201436 [details]
Patch

Attachment 201436 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.appspot.com/results/277251

New failing tests:
fast/js/global-constructors-attributes.html

[Build Bot]

Created attachment 201446 [details]
Archive of layout-test-results from webkit-ews-08 for mac-mountainlion

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: webkit-ews-08  Port: mac-mountainlion  Platform: Mac OS X 10.8.2

[Benjamin Poulain]

Comment on attachment 201436 [details]
Patch

I think we should just remove ENABLE_WEB_TIMING_MINIMAL. It was a nice workaround but it will always break some feature detection code.

[alan]

(In reply to comment #15)
> (From update of attachment 201436 [details])
> I think we should just remove ENABLE_WEB_TIMING_MINIMAL. It was a nice workaround but it will always break some feature detection code.

With the exception of the Memory API. It is behind WEB_TIMING, but not necessarily something we want to expose for obvious reasons. I agree, though, that missing parts of PerformanceNavigation and PerformanceTiming have the potential of breaking detection code at any point of time. (even though performance.now() ought to just work as a standalone api, really as a monotonic timer. )
I am inclined to remove this completely and opt out of Memory when we enable WEB_TIMING.

[alan]

Created attachment 201455 [details]
Patch

[Build Bot]

Comment on attachment 201455 [details]
Patch

Attachment 201455 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.appspot.com/results/346221

New failing tests:
fast/js/global-constructors-attributes.html

[Build Bot]

Created attachment 201472 [details]
Archive of layout-test-results from webkit-ews-06 for mac-mountainlion

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: webkit-ews-06  Port: mac-mountainlion  Platform: Mac OS X 10.8.2

[Andreas Kling]

Comment on attachment 201455 [details]
Patch

r=me with failures not failing.

[Build Bot]

Comment on attachment 201455 [details]
Patch

Attachment 201455 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.appspot.com/results/461207

New failing tests:
fast/js/global-constructors-attributes.html

[Build Bot]

Created attachment 201473 [details]
Archive of layout-test-results from webkit-ews-01 for mac-mountainlion

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: webkit-ews-01  Port: mac-mountainlion  Platform: Mac OS X 10.8.2

[alan]

Created attachment 201488 [details]
Patch

[alan]

Comment on attachment 201488 [details]
Patch

passed mac-ews

[WebKit Commit Bot]

Comment on attachment 201488 [details]
Patch

Clearing flags on attachment: 201488

Committed r149999: <http://trac.webkit.org/changeset/149999>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.