421813 - LTR version of "ASSERTION: Couldn't find glyph for trailing marker" with LSEP, ZWJ, LRO, Arabic

Status: RESOLVED WORKSFORME

(Core :: Graphics, defect)

Description

[Jesse Ruderman]

Attachment: testcase (seems to cause random memory corruption)

Loading the testcase triggers:

###!!! ASSERTION: Couldn't find glyph for trailing marker: 'glyphRecords[numGlyphs - 1].originalOffset == aSegmentLength*2', file /Users/jruderman/trunk/mozilla/gfx/thebes/src/gfxAtsuiFonts.cpp, line 1050

###!!! ASSERTION: Font change inside character group!: 'Error', file /Users/jruderman/trunk/mozilla/gfx/thebes/src/gfxAtsuiFonts.cpp, line 971


Sometimes it also triggers:

###!!! ASSERTION: Invalid char passed in: '!gfxFontGroup::IsInvalidChar(aString[firstOffset/2])', file /Users/jruderman/trunk/mozilla/gfx/thebes/src/gfxAtsuiFonts.cpp, line 900


Reload a few times and you'll probably see a random-looking crash.


There are several existing bug reports where the RTL version of the "Couldn't find glyph for trailing marker" assertion fires (bug 420945 and bug 421393), but this is the only one I know about where the LTR version fires.

Comment 1

[Jesse Ruderman]

WFM using Mac trunk (no assertions, no crash) (tested with and without Malloc*Scribble).

Comment 2

[Jesse Ruderman]

Crashtest checked in.