GNOME Bugzilla – Bug 84083
nautilus crashes when it enters a directory with a bitmap file
Last modified: 2010-07-10 04:06:14 UTC
Gnome 2.0, nightly 30th May 2002. Solaris sparc 9 Nautilus will crash if user enters a directory containing a bitmap file. Also if user attempts to use a bitmap (*.bmp) as a default background, gnome-backgroud-properties will crash.
Created attachment 8973 [details] sample bitmap graphic file
Can you also please provide a backtrace of the crash...
Created attachment 8993 [details] pstack of nautilus core after entering directory with bitmap image
This works fine for me; what version of gtk+ are you using ? Can you run a gtk+ test program for me on this file - it's possibly a Solaris issue: gtk+/demos/test-pixbuf <my-crasher-pixbuf> If that crashes it's almost certainly a gtk+ / gdk-pixbuf problem. Otherwise it might be that some code in there is not threadsafe for some reason.
gtk+ version is 2.1.0 here is the output from testpixbuf, it does not open the sample file. Hope this is of some help?. ./testpixbuf /export/home/gn_tst1/test_files/sample.bmp Visual type = pseudo color, depth = 8, 0:0:0 (system); score=4111 Visual type = direct color, depth = 8, 7:38:c0; score=4100 Visual type = true color, depth = 8, 7:38:c0; score=4101 Visual type = static color, depth = 8, 0:0:0; score=4100 Visual type = grayscale, depth = 8, 0:0:0; score=4100 Visual type = static gray, depth = 8, 0:0:0; score=4100 color cube: 6 x 6 x 6 Chose visual type=pseudo color depth=8, image bpp=8, msb first Bus Error
Could we please get a stack trace with line number information? Either nautilus or test-pixbuf are fine --- test-pixbuf should be easier to track down. Cute image, BTW :)
It may also be useful to get a disassembly of DecodeHeader().
Nautilus also cores or freezes when creating a thumbnail for a .ras file. EOG also cores when given the .bmp file I can't get a good line number traceback on this stripped build. Here are the library calls in the top thread: (if it is at all useful I can attach the other threads, otherwise I'll try to find a debug build with this problem) lwp# 8 / thread# 8 -------------------- fcce0fac DecodeHeader (3de888, 3de896, 77cac0, 0, 0, 3de888) + c fcce2950 gdk_pixbuf__bmp_image_load_increment (77cac0, 76973a, 66, 0, fea27a64, 1023c) + 110 fea09090 gdk_pixbuf_loader_load_module (3efb20, 0, 0, 4, 7e, 0) + 1b0 fea0934c gdk_pixbuf_loader_write (3efb20, fcdebed4, 10000, 0, 0, 769718) + 24c fef39ee0 eel_gdk_pixbuf_load (3f1d98, 60, 3de4c8, 1cedc, ff291d00, fde7a0d4) + e0 ff291d2c thumbnail_thread_make_thumbnail (28fdf0, 1c00, 1e38, 7697f8, 433d80, 16) + 2cc ff291fa4 thumbnail_thread_start (0, 0, 0, 0, 0, 0) + c4 fdec58f0 _lwp_start (0, 0, 0, 0, 0, 0)
We do need the exact line number. Also, please provide a disassembly of DecodeHeader() to see just what's going on. I thought it could be an alignment issue but after looking at the code it doesn't look like it at all.
OK, I have a patch that fixes it. Awaiting approval from the release team.
Created attachment 9070 [details] [review] Patch with proposed fix.
*** Bug 84515 has been marked as a duplicate of this bug. ***
Federico: please make sure you put this in sun_patches ASAP, BTW. Thanks.
I have verified that this fixes the core dump. Now Nautilus correctly displays the BMP file as a thumbnail. I just added this patch to sun-patches as: sun-patches/gtk+/208-84083-s.diff I notice this patch is not checked into the gtk+ module though.
Fixed on CVS. Should I leave the patch in sun-patches?
*** Bug 84292 has been marked as a duplicate of this bug. ***
Federico: don't know, to be honest. Not sure of the current build procedure.
fix verified in the latest sun pre-beta build (respin of build3) marking as fix -> verified
closing