On Thu, Oct 8, 2009 at 20:49:25 +0200, Stefan Klinger wrote: > On 08 October 2009, Brice Goglin wrote with possible deletions: > > You should start X normally,and then connect to the machine through ssh > > and type gdp -p $(pidof X) > > Then enter 'c' for continue in gdb, reproduce the bug, and get the > > backtrace with 'bt full' in gdb. > > Ok. That's what I did. the results are attached. > > There are 3*3=9 files attached. Their names indicate the conditions under > which the error showed up: > > - The files triggered.* com from a crash triggered by calling 'dmesg 9' as > non-privoleged user. Yes, it is 'dmesg 9' not 'dmesg -n 9'. > > - The random.* files come from a crash that I did not consiously trigger. I > bet I did, but I dont know why or how. > > - The normal.* files come from "normal" operation. I include them, because X > seems to terminate with a SEGFAULT when the windowmanager disconnects and > my > session terminates. > Any chance you could try again to get gdb traces, but this time with 'handle SIGPIPE nostop' at the gdb prompt before reproducing? SIGPIPE in the X server happens in normal operation, when a client disconnects.
The 'normal.*' segfault (on server restart) is a known bug, fixed in intel 2.9.0, so don't worry about that one. Thanks, Julien -- To UNSUBSCRIBE, email to debian-x-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
