Bug 23985 - Crash on WebCore::InlineFlowBox::addToLine
Summary: Crash on WebCore::InlineFlowBox::addToLine
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: WebCore Misc. (show other bugs)
Version: 528+ (Nightly build)
Hardware: All All
: P2 Major
Assignee: Dave Hyatt
URL: http://www.familyguyx.net/episode/bab...
Keywords:
: 23998 (view as bug list)
Depends on:
Blocks: 24281
  Show dependency treegraph
 
Reported: 2009-02-16 19:27 PST by Dimitri Glazkov (Google)
Modified: 2009-03-01 17:07 PST (History)
1 user (show)

See Also:


Attachments
Crash log (34.03 KB, text/plain)
2009-02-16 20:23 PST, Dimitri Glazkov (Google)
no flags Details
Rendering slowness reduction. (287 bytes, text/html)
2009-02-17 10:04 PST, Dimitri Glazkov (Google)
no flags Details
Force legends to be display:block (1.65 KB, patch)
2009-02-17 10:17 PST, Dave Hyatt
no flags Details | Formatted Diff | Diff
Force legends to be display:block (931 bytes, patch)
2009-02-17 10:18 PST, Dave Hyatt
no flags Details | Formatted Diff | Diff
Force legends to be display:block and eliminate the RenderLegend that caused the trouble in the first place (12.77 KB, patch)
2009-02-17 10:30 PST, Dave Hyatt
hyatt: review-
Details | Formatted Diff | Diff
Force legends to be display:block (60.61 KB, patch)
2009-02-17 10:51 PST, Dave Hyatt
eric: review+
Details | Formatted Diff | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Dimitri Glazkov (Google) 2009-02-16 19:27:30 PST
This bug was initially filed here http://crbug.com/7775. Stack trace coming up (see Chromium's for information for now).
Comment 1 Dimitri Glazkov (Google) 2009-02-16 20:23:30 PST
Created attachment 27717 [details]
Crash log

Here's the crash log. Working on reduction ...
Comment 2 Dimitri Glazkov (Google) 2009-02-17 10:04:20 PST
Created attachment 27732 [details]
Rendering slowness reduction.

Ok, this is interesting. Here's a reduction that doesn't cause a crash, but it definitely hangs the renderer for a while (up to 30 secs on my machine) and produces clearly invalid output.
Comment 3 Dave Hyatt 2009-02-17 10:17:52 PST
Created attachment 27733 [details]
Force legends to be display:block
Comment 4 Dave Hyatt 2009-02-17 10:18:28 PST
Created attachment 27734 [details]
Force legends to be display:block
Comment 5 Dave Hyatt 2009-02-17 10:19:19 PST
This patch fixes the rendering issues with this test case (and another one Dan pointed out to me a week or so ago).  It might still be worth investigating how we hit the assert in createLineBoxes when we let the legend be inline.
Comment 6 Dave Hyatt 2009-02-17 10:30:06 PST
Created attachment 27735 [details]
Force legends to be display:block and eliminate the RenderLegend that caused the trouble in the first place
Comment 7 Dave Hyatt 2009-02-17 10:32:24 PST
Comment on attachment 27735 [details]
Force legends to be display:block and eliminate the RenderLegend that caused the trouble in the first place

Need new changelog and a test case. Minusing myself.
Comment 8 Dave Hyatt 2009-02-17 10:51:55 PST
Created attachment 27737 [details]
Force legends to be display:block
Comment 9 Dimitri Glazkov (Google) 2009-02-17 10:53:57 PST
It's gorgeous. Somebody r+ this!
Comment 10 Eric Seidel (no email) 2009-02-17 11:27:36 PST
Comment on attachment 27737 [details]
Force legends to be display:block

Looks fine.  I can't claim to be an expert here, but Hyatt's change looks sane to me.
Comment 11 Dave Hyatt 2009-02-17 11:34:02 PST
Fixed in r41041.

Comment 12 Jon@Chromium 2009-02-17 15:43:20 PST
*** Bug 23998 has been marked as a duplicate of this bug. ***