reassign 397012 libimlib2 found 397012 1.3.0.0debian1-2 tags 397012 patch thanks
On Tue, Nov 07, 2006 at 04:55:10PM +0100, Jonas Meurer wrote: > On 06/11/2006 Steve Langasek wrote: > > > i neither have any directories inside ~/.idesktop/ nor have a background > > > image stored there. > > > See the ~/.ideskrc file attached. > > I can't reproduce the crash on amd64 using the provided .ideskrc alone. Can > > you provide a minimal test case for the crash, including images and any > > parts of your .idesktop that are needed in order to trigger the crash? > ok, even with only one file (gimp.lnk) in ~/.idesktop/, idesk crashes. > I used the .ideskrc file that you already know, and the following inside > ~/.idesktop/gimp.lnk: > table Icon > Caption: Gimp > Icon: /usr/share/icons/crystalsvg/48x48/apps/gimp.png > X: 30 > Y: 730 > Command[0]: /usr/bin/gimp > CaptionTip: Gimp > end > > Also, do you have any tiffs anywhere in your config? I notice that bug > > #381177 has just been closed in libimlib2... > no, only png images. > i could rebuild libimlib2 with DEB_BUILD_OPTIONS=nostrip and send again > the gdb backtrace and strace output. Turns out I am able to reproduce this, I just misunderstood the symptoms because something is trapping the segfault when not running under gdb and idesk was therefore exiting silently instead of throwing an error message. And I've tracked the problem down to some amd64 assembly code in libImlib2.so.1: #0 0x00002b9b2be201ff in __imlib_amd64_blend_rgba_to_rgb_cmod () from /usr/lib/libImlib2.so.1 #1 0x00002b9b2bde8d14 in imlib_blend_image_onto_image (source_image=0x553800, merge_alpha=1 '\001', source_x=0, source_y=0, source_width=5593504, source_height=48, destination_x=0, destination_y=0, destination_width=48, destination_height=48) at api.c:2015 [...] The instruction at 0x00002b9b2be201ff is: 0x00002b08934011ff <__imlib_amd64_blend_rgba_to_rgb_cmod+47>: movdqa 1155461(%rip),%xmm5 # 0x2b089351b38c <c1> I have no idea what this does, other than that it segfaults. The easiest fix I see is to simply disable the amd64 assembly in libimlib2 until someone can take a look at and fix it; this can be done by adding a --disable-amd64 option to the configure command in debian/rules. I've confirmed that doing so fixes the segfault on my test system. Cheers, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. [EMAIL PROTECTED] http://www.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]