[Fontconfig] fontconfig 2.8.0

[Derek Leach]

Hi Raimund,

So the tracking information on the customer's target machine is useless, looks like it was a desktop build cooked by someone, so not true idea if it is 2.8.0 or not.
Below is the segfault that occurred, where the first seven frames are similar (frames 2,3,4 are not defined in the bug report).
Any ideas?  Not really a lot of information to go on ...

Thanks,
Derek

#0  0xb8a8e502 in IA__FcStrListCreate () from ...........
No symbol table info available.
#1  0xb8a87c09 in IA__FcLangSetHasLang () from ............
No symbol table info available.
#2  0xb8a89dfc in FcCompareLang () from ...........
No symbol table info available.
#3  0xb8a8969c in FcCompareValueList () from ............
No symbol table info available.
#4  0xb8a898fc in FcCompare () from ..............
No symbol table info available.
#5  0xb8a89fa9 in IA__FcFontSetSort () from ...........
No symbol table info available.
#6  0xb8a8a489 in IA__FcFontSort () from ........
No symbol table info available.
#7  0xb87e3613 in FontConfig::Find (
    family=0x87d4220 <Address 0x87d4220 out of bounds>, isBold=false, 
    isItalic=false)
    at .../platform/unix/unixfontconfig.cpp:124
	result = 3098046891
	match = <value optimized out>
	fs = <value optimized out>
	weight = <value optimized out>
	filename = <value optimized out>
	slant = <value optimized out>
	pattern = (FcPattern *) 0x87da3e8
	tempString = 0x87d4560 <Address 0x87d4560 out of bounds>
#8  0xb87e3710 in FontConfig::FindFont (
    family=0x87d4220 <Address 0x87d4220 out of bounds>, isBold=88, 
    isItalic=<value optimized out>)
    at .../platform/unix/unixfontconfig.cpp:74
	font = <value optimized out>
	fontString = <value optimized out>
#9  0xb85528e2 in PlatformEDevice::FindFont (this=0x97048b0, 
    family=0x87d4500 <Address 0x87d4500 out of bounds>, alias=0x0)
    at .../platform/unix/unixfonts.cpp:170
	tmpFamily = {static NPOS = <optimized out>, 
  m_buf = 0x87d4220 <Address 0x87d4220 out of bounds>, m_len = 22, 
  m_size = 23}
	isItalic = <value optimized out>
	cached = 88
	font = (UnixFont *) 0x0
	addr = (void *) 0x0
	isBold = <value optimized out>
	cachedName = {static NPOS = <optimized out>, 
  m_buf = 0x8e7b160 <Address 0x8e7b160 out of bounds>, m_len = 31, 
  m_size = 286}
#10 0xb8552aec in PlatformEDevice::ResolveFontName (this=0x97048b0, 
    result=0x8044fac "mb222demiitalic-global", version=10)
    at .../platform/unix/unixfonts.cpp:223
	str = <value optimized out>
	cached = <value optimized out>
	resolved = <value optimized out>
	font = (UnixFont *) 0x0
	addr = (void *) 0x0
	familyName = <value optimized out>
	cachedName = {static NPOS = <optimized out>, 
  m_buf = 0x8d2e110 <Address 0x8d2e110 out of bounds>, m_len = 31, 
  m_size = 286}
[snip]

-----Original Message-----
From: Raimund Steger [mailto:rs at mytum.de] 
Sent: April-02-14 6:18 PM
To: Derek Leach; 'fontconfig at lists.freedesktop.org'
Subject: Re: [Fontconfig] fontconfig 2.8.0

Derek Leach wrote:
> Hello All,
>
> I have encountered a bug on a internal distribution that uses 
> fontconfig 2.8.0, the backtrace is identical to this bug:
>
> https://bugzilla.redhat.com/show_bug.cgi?id=532718
>
> at least frames #0 - #7.  I have been trying to find the code change 
> that fixed this bug, but have not been successful.
>
> Can someone direct me to the code change that was used to fix this segfault?
>

Judging by the date of the comments in this bug I would guess 1f4e6fecde22fd4ce8336b01a5c32c533fcb8bac, i. e.
http://cgit.freedesktop.org/fontconfig/commit/?id=1f4e6fecde22fd4ce8336b01a5c32c533fcb8bac

However, if I understand the bug correctly this was supposed to be fixed with 2.8.0, and even before only happened if the same cache was used by different fontconfig versions. Are you sure it's what you're seeing?

Raimund

--
Worringer Str 31 Duesseldorf 40211 DE  home: <rs at mytum.de>
+49-179-2981632 icq 16845346           work: <rs at interface-ag.de>