Brice Goglin -> Kirill A. Korinskiy @ Sat, 22 Dec 2007 17:59:55 +0100:
BG> Login by ssh, and run
BG> gdb -p $(pidof X)
BG> then you can interrupt X within gdb with ctrl-c and get the current
BG> stack with
BG> bt full
BG> X can run again if you enter "c" (for "continue"). Then you can ctrl-c
BG> and get another stack. If you get always the same backtrace, then we
BG> will know where the hang would be.
BG> This bug could be related to a well-known hang that we just fixed (bug
BG> #451989).
--8<---------------cut here---------------start------------->8---
(gdb) bt full
#0 0x0809a1c0 in EnqueueEvent ()
No symbol table info available.
#1 0x08099688 in ComputeFreezes ()
No symbol table info available.
#2 0x08099e8b in CheckGrabForSyncs ()
No symbol table info available.
#3 0x08099fdf in ActivateKeyboardGrab ()
No symbol table info available.
#4 0x0809471c in GrabDevice ()
No symbol table info available.
#5 0x08094925 in ProcGrabKeyboard ()