[Webkit-unassigned] [Bug 52422] [chromium] More crash in FontFallbackList::determinePitch(const Font* font)

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=52422





--- Comment #3 from Hironori Bono <hbono at chromium.org>  2011-01-18 22:58:43 PST ---
Greetings,

Thank you for your comments. I would like to describe some more details about this issue before updating my change.

(In reply to comment #2)
> (From update of attachment 78889 [details])
> oops. :)
> This line should be removed.  Ideally there would be a test.

Unfortunately, I do not have any idea why this crash still happens, i.e. I do not have any ideas how to reproduce this crash with Chromium with my r72116 <http://trac.webkit.org/changeset/73116>. This change just tries emulating Safari/Win and sees it makes any differences. If it does not make any differences, I need another change.

> What about doing what you mentioned in the bug or something like that?

When I tried enumerating all the fonts with a EnumFontFamiliesEx() call, it took long time (>1s) on my virtual machine and blocks WebKit during the period. Even though WebKit needs to enumerate all the installed fonts only once (and it is pretty rare to fall back to my code), I'm wondering if it is OK to add such huge blocking code.

> Would that cause a crash that is fixed by this change?

In brief, I do not have any clear ideas about this crash except that my r72116 did not fix the crash. As shown in the crash dump, this crash happens in FontFallbackList::determinePitch(), some time after FontCache::getLastResortFallbackFont() returns 0. So, this crash dump does not include any information about why FontCache::getLastResortFallbackFont() returns 0. (I noticed ASSERT_NOT_REACHED() did not cause crashes on Release builds.) If we add code that intentionally crashes as ASSERT_NOT_REACHED() does, we can investigate why this function returns 0. Nevertheless, I'm not sure it is acceptable. (I assume this is a reason why this crash has not been fixed for a long time.)

It is definitely welcome if there are any suggestions about how to identify why this crash still happens and fix it.

Regards,

Hironori Bono

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.