GNOME Bugzilla – Bug 364570
gnome-pilot crashes backing up T|E with pilot-link 0.11.8
Last modified: 2006-10-24 10:58:51 UTC
Steps to reproduce: I installed recently pilot-link 0.11.8, gnome-pilot(and -conduits) 2.0.14 and Evolution (that's why I didn't install pilot-link 0.12.1). This was the first atempt to backup files from the PDA, a Tungsten E, with the configuration. The backup was going well untill it got to the photos. It hanged for a while then gnome-pilot crashed. Stack trace: Memory status: size: 58347520 vsize: 0 resident: 58347520 share: 0 rss: 14315520 rss_rlim: 0 CPU usage: start_time: 1161597852 rtime: 0 utime: 909 stime: 0 cutime:468 cstime: 0 timeout: 441 it_real_value: 0 frequency: 0 Backtrace was generated from '/usr/libexec/gnome-pilot' Using host libthread_db library "/lib/libthread_db.so.1". [Thread debugging using libthread_db enabled] [New Thread -1225820496 (LWP 9379)] [New Thread -1245156448 (LWP 10976)] [New Thread -1236763744 (LWP 10973)] 0xffffe410 in ?? ()
+ Trace 77941
Other information: Pilot-link < 0.12 is known to segfault when it tries to backup ImgFile-Foto.prc and Jpeg-Foto.prc from Tungsten E [*]; maybe gnome-pilot (when linked to pilot-link 0.11.x) should just exclude the files from the backup list, and warn the user about that. I'm running Gentoo Linux with conservative settings: -02, -pipe, -ggdb, and -march=athlon-xp. *. http://lists.pilot-link.org/pipermail/pilot-link-devel/2003-November/000580.html
It is possible to exclude particular files from being backed up by gnome-pilot. See comment nine in bug 136010: http://bugzilla.gnome.org/show_bug.cgi?id=136010 Alternatively, there is a patch available to port the evolution conduits to pilot-link 0.12.x. See the link on the gnome-pilot wiki page: http://live.gnome.org/GnomePilot Also on the wiki page you will see that there's a patch to fix a memory leak found since the release of 2.0.14. If you're not using HAL with gpilotd, you'll definitely want to apply that patch. Matt p.s. It's odd that you don't seem to have symbols from gpilotd in your backtrace, despite using '-ggdb'... *** This bug has been marked as a duplicate of 136010 ***
(In reply to comment #1) > It is possible to exclude particular files from being backed up by gnome-pilot. > See comment nine in bug 136010: > http://bugzilla.gnome.org/show_bug.cgi?id=136010 Thanks! > Alternatively, there is a patch available to port the evolution conduits to > pilot-link 0.12.x. See the link on the gnome-pilot wiki page: > http://live.gnome.org/GnomePilot Is it know to be stable? In other words, could a distro include this patch? > Also on the wiki page you will see that there's a patch to fix a memory leak > found since the release of 2.0.14. If you're not using HAL with gpilotd, > you'll definitely want to apply that patch. Sorry asking here, but is there anything in my trace suggesting myt gnome-pilot installation isn't taking benefit from HAL? > p.s. It's odd that you don't seem to have symbols from gpilotd in your > backtrace, despite using '-ggdb'... I thought so, too.
The evolution conduits patch is currently being shipped by Ubuntu (edgy), AFAIK. It has not caused problems, again AFAIK. gpilotd 2.0.14 uses HAL by default. It needs to be compiled in (check the status messages at the end of the ./configure output - it will report whether suitable HAL headers/libraries are available). At runtime, if HAL can't be initialised gpilotd falls back to polling the usb devices via sysfs. If you run gpilotd from a terminal window you will see a message "seems that HAL is not running" or something like that. In any case, the patch for the leak is linked to from the wiki page: http://live.gnome.org/GnomePilot and should be added. It is a one-liner.