Bug 153051 – Mixing Content Blocking of fonts and display:none rules causes battery drain

• Component: New Bugs
• Status: RESOLVED FIXED

[Myles C. Maxfield]

Mixing Content Blocking of fonts and display:none rules causes battery drain

[Myles C. Maxfield]

Created attachment 268835 [details]
Patch

[Build Bot]

Comment on attachment 268835 [details]
Patch

Attachment 268835 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.webkit.org/results/684709

New failing tests:
http/tests/security/contentSecurityPolicy/user-style-sheet-font-crasher.html

[Build Bot]

Created attachment 268843 [details]
Archive of layout-test-results from ews100 for mac-yosemite

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: ews100  Port: mac-yosemite  Platform: Mac OS X 10.10.5

[Build Bot]

Comment on attachment 268835 [details]
Patch

Attachment 268835 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-queues.webkit.org/results/684713

New failing tests:
http/tests/security/contentSecurityPolicy/user-style-sheet-font-crasher.html

[Build Bot]

Created attachment 268845 [details]
Archive of layout-test-results from ews107 for mac-yosemite-wk2

The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews107  Port: mac-yosemite-wk2  Platform: Mac OS X 10.10.5

[Build Bot]

Comment on attachment 268835 [details]
Patch

Attachment 268835 [details] did not pass mac-debug-ews (mac):
Output: http://webkit-queues.webkit.org/results/684705

New failing tests:
http/tests/security/contentSecurityPolicy/user-style-sheet-font-crasher.html

[Build Bot]

Created attachment 268847 [details]
Archive of layout-test-results from ews114 for mac-yosemite

The attached test failures were seen while running run-webkit-tests on the mac-debug-ews.
Bot: ews114  Port: mac-yosemite  Platform: Mac OS X 10.10.5

[Myles C. Maxfield]

Created attachment 268848 [details]
Patch

[Myles C. Maxfield]

<rdar://problem/23187709>

[Myles C. Maxfield]

Yeah; it's as I suspected; this patch doesn't actually fix the bug. Any time you clear the style resolver, a layout will occur, which means the font will be used again, which means we will go through content blockers, which means we may clear the style resolver again.

[Myles C. Maxfield]

(In reply to comment #10)
> Yeah; it's as I suspected; this patch doesn't actually fix the bug. Any time
> you clear the style resolver, a layout will occur, which means the font will
> be used again, which means we will go through content blockers, which means
> we may clear the style resolver again.

It looks like I'll need some bit of state saying "this font was blocked; don't try again." However, this means that I'll need some way of knowing when content blocker rules change (to clear the state).

[Myles C. Maxfield]

(In reply to comment #11)
> (In reply to comment #10)
> > Yeah; it's as I suspected; this patch doesn't actually fix the bug. Any time
> > you clear the style resolver, a layout will occur, which means the font will
> > be used again, which means we will go through content blockers, which means
> > we may clear the style resolver again.
> 
> It looks like I'll need some bit of state saying "this font was blocked;
> don't try again." However, this means that I'll need some way of knowing
> when content blocker rules change (to clear the state).

Maybe the key here is that layout triggers the download, rather than setting a src attribute or something like that (for an image).

[Myles C. Maxfield]

Created attachment 269021 [details]
Patch

[Alex Christensen]

Comment on attachment 269021 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=269021&action=review

> Source/WebCore/ChangeLog:8
> +        If we have applied a rule before, don't apply it again (thereby avoiding a relayout).

If we have applied a rule before and we are not applying it again, don't resolve the style again.

The problem wasn't that we were applying it again, the problem was that we were doing another style resolution when we weren't changing anything, which often led to infinite loops.

[Myles C. Maxfield]

Committed r195088: <http://trac.webkit.org/changeset/195088>

[Myles C. Maxfield]

Follow-up patch at http://trac.webkit.org/changeset/195089

[Myles C. Maxfield]

Committed r195161: <http://trac.webkit.org/changeset/195161>