I finally did sort out the problem below, just for the archives.

fc-cache wants to write a fonts.cache-1 file in each directory 
referenced in the font directory list in /etc/fonts/fonts.conf.  Since 
the Live DVD has a read-only /usr, this doesn't work, and it ends up 
creating a .fonts.cache-1 in root's home directory.  Applications 
attempt to use that and crash as shown below; I could reproduce it just 
trying to run gedit under CDE

The workaround/fix is to run a chroot'ed fc-cache in the image 
construction process when the filesystem is writable.

Anyone tested whether this stuff works right with a diskless client? 
They also mount /usr read-only and would seem to have the same problem.

Dave

Dave Miner wrote:
> Laszlo (Laca) Peter wrote:
>> Run fc-cache --force as root.
>>
> 
> Unfortunately, neither your suggestion nor James's made any difference. 
>   I did end up with a .fonts.cache-1 in root's home directory, but still 
> end up with the same core when attempting to login.  Any other suggestions?
> 
> Dave
> 
>> Laca
>>
>> On Sat, 2007-01-20 at 11:45 -0800, Dave Miner wrote:
>>> Using snv_55, I'm getting crashes from gnome-session, the sessionlog has:
>> (...)
>>> -bash-3.00# pstack core
>>> core 'core' of 24142:   /usr/lib/gnome_segv gnome-session 11 2.16.1
>>>  cb9b74fb FcPatternPosition (8287a08, 8127d40) + 6b
>>>  cb9b758e FcPatternFindElt (8287a08, cb9c7c00) + 2e
>>>  cb9b81fe FcPatternGet (8287a08, cb9c7c00, 0, 8044424) + 2e
>>>  cb9b8596 FcPatternGetCharSet (8287a08, cb9c7c00, 0, 8044468) + 36
>>>  cb9b3dfb FcSortWalk (81d3514, 619, 82e49e0, 80444e4, 1) + 6b
>>>  cb9b45e8 FcFontSetSort (811e300, 8044538, 1, 811ed40, 1, 0) + 668
>>>  cb9b471a FcFontSort (811e300, 811ed40, 1, 0, 8044608) + aa
>>>  cb84a926 pango_fc_font_map_get_patterns (80821e8, 80e5a40, 80e5900, 
>>> 8114de0) + 1ba
>>>  cb84ad06 pango_fc_font_map_load_fontset (80821e8, 80e5a40, 80e5900, 
>>> 8114de0) + 36
>>>  cbad671b pango_font_map_load_fontset (80821e8, 80e5a40, 80e5900, 8114de0) 
>>> + 3b
>>>  cbad4dad itemize_state_update_for_new_run (8044728) + 135
>> (...)
>>
>>
> 
> 


Reply via email to