194644 - ISO10646-1 fonts not recognised + Talkback crashes in Xprint module [@ nsRenderingContextXlib::GetTextDimensions]

Status: RESOLVED FIXED

(Core Graveyard :: Printing: Xprint, defect, P1)

Description

[Simon Montagu :smontagu]

Some recent talkback reports with crashes in Xprint

Comment 1

[Simon Montagu :smontagu]

Attachment: Talkback report

The TB IDs are:
17296826
17285461
17285457
17285455
17218471
17204322
17204321

Comment 2

[Roland Mainz]

Swapping QA<---->Owner...

Comment 3

[Roland Mainz]

Looks something in nsRenderingContextXlib::GetTextDimensions() is going wrong...
two problems:
- The crash seems to be related to the existance of certain fonts. For example I
cannot reproduce the crash while printing http://www.aljazeera.net when I do not
have any arabic fonts installed.
- The "fast test measurement" code path is only enabled on i386 platforms,
marking it hard for me to hunt this issue... ;-(

Comment 4

[Roland Mainz]

I filed bug 194651 ("Need workaround for crashers in
nsRenderingContextXlib::GetTextDimensions()") to get a quick workaround for
1.3final and use this bug to hunt the root of the problem...

Comment 5

[Roland Mainz]

Per
http://groups.google.com/groups?q=GetTextDimensions&hl=en&lr=&ie=UTF-8&safe=off&scoring=d&selm=200302200940.BAA03481%40beckett.mcom.com&rnum=1
|GetTextDimensions| seems to be causing some trouble on other platforms, too...
;-/

Comment 6

[Roland Mainz]

rbs/smontagu:
Is it (theoretically) possible that someone may (accidently) pass |nsnull| for
the |GetTextDimensions|'s |aString| argument ?

Comment 7

[Simon Montagu :smontagu]

The problem in nsRenderingContextXlib::GetTextDimensions when printing Arabic is
conceivably related to bug 188759, which I hope to check in the fix for today.

Comment 8

[Simon Montagu :smontagu]

Attachment: Stack from TB 17218471

The stacks in the talkback reports are all much the same. Here is one example.

Comment 9

[Roland Mainz]

Looks that |mCCMap| is |nsnull|:
-- snip --
WARNING: SetUpFontCharSetInfo: charset = 'jis_0201', file nsFontMetricsXlib.cpp,
line 2581

Program received signal SIGSEGV, Segmentation fault.
0x42a604b9 in nsRenderingContextXlib::GetTextDimensions (this=0x8a09d40,
aString=0xbfff5ac4, aLength=7, aDimensions=@0xbfff5820, 
    aFontID=0x0) at nsRenderingContextXlib.cpp:2027
2027            if (CCMAP_HAS_CHAR((*font)->mCCMap, c)) {
(gdb) print font
$1 = (nsFontXlib **) 0x8a68cd4
(gdb) print *font
$2 = (nsFontXlib *) 0x8a6b5b0
(gdb) print (*font)->mCCMap
$3 = (PRUint16 *) 0x0
-- snip --

Comment 10

[Roland Mainz]

Attachment: Patch for 2003-02-07-08-trunk

The issue was more simple than I thought... :)

Actually the issue is identical to the bug that all iso106464-1 fonts cause
crashes or are inaccessible by the Xlib port.
After some reaching I figured-out that the original patch from bug 187125
("GTK+ and Xlib gfx should support multiple device instances") forgot to handle
the special case of the ISO-10646-1 fonts - this patch now fixes that.

Comment 11

[Roland Mainz]

Comment on attachment 115444 [details] [diff] [review]
Patch for 2003-02-07-08-trunk

Requesting r=/sr= ...

Comment 12

[Simon Montagu :smontagu]

Comment on attachment 115444 [details] [diff] [review]
Patch for 2003-02-07-08-trunk

r=smontagu

Comment 13

[Roland Mainz]

Comment on attachment 115444 [details] [diff] [review]
Patch for 2003-02-07-08-trunk

Requesting a= for 1.3final branch...

Comment 14

[Roland Mainz]

Patch checked-in into "trunk"
(http://bonsai.mozilla.org/cvsquery.cgi?module=MozillaTinderboxAll&branch=HEAD&cvsroot=/cvsroot&date=explicit&mindate=1046142360&maxdate=1046142720&who=timeless%25mozdev.org)...

... waiting for a{1.3final}= ...

Comment 15

[Asa Dotzler [:asa]]

Comment on attachment 115444 [details] [diff] [review]
Patch for 2003-02-07-08-trunk

a=asa (on behalf of drivers) for checkin to the 1.3 branch.

Comment 16

[Jan Carpenter]

adding topcrash to keywords (for posterity)

Comment 17

[Roland Mainz]

Patch checked-in into 1.3-branch
(http://bonsai.mozilla.org/cvsquery.cgi?module=MozillaTinderboxAll&branch=MOZILLA_1_3_BRANCH&cvsroot=/cvsroot&date=explicit&mindate=1046216220&maxdate=1046217780&who=smontagu%25netscape.com),
marking bug as FIXED.