I noticed this when I tried to get Blastwave's KDE build running from within a
zone (to not pollute the global zone). First thought it'd be KDE, but then I
saw it's also going it with gnome-session.
It appears that closing any window created inside Xephyr will make it crash.
Considering it generates different traces whether Compiz is running or not, I'm
at a loss where to look for a cause for this. Though truss output seems to
create link between both traces.
To crash it, any window created in it just needs to be closed. Whether the
application uses Unix domain sockets or TCP/IP doesn't seem to matter, nor if
it's in the local or global zone. Starting gedit in Xephyr and quitting it
makes it crash. Opening gedit, then opening the file dialog and exiting that
one, it crashes. So, anything that involves window closure makes it quit. The
stack traces are different ones depending whether Compiz ran or not. But when
Compiz is running, it's interesting to see that it's apparently crashing in
compDestroyWindow in Xephyr (???).
This happens in OpenSolaris image-updated to snv_94 and as well snv_95. Haven't
tried using Xephyr before snv_94. Any way to pinpoint this down?
Regards,
-mg
Usual stacktrace with Metacity:
=======================================================
hyperborea:~ > pstack core
core 'core' of 556: iiim-xbe
----------------- lwp# 1 / thread# 1 --------------------
feefd2a5 __nanosleep (8047b90, 8047b98) + 15
feeec227 sleep (989680, 8047c00, 8047bc8, 805dac0) + 35
0805dacd main (1, 8047c00, 8047c08, fef97080) + 29
0805d39d _start (1, 8047ce4, 0, 8047ced, 8047d20, 8047d36) + 7d
----------------- lwp# 2 / thread# 2 --------------------
fe7bbe26 property_notify_switcher_window (809a688, c00003, fe2aef64, 80b11b8)
+ 32
feca7c4f XFilterEvent (fe2aef64, 0, a00003, 805f761) + d7
0805f7a6 __1cLXIMProtocolLxevent_loop6Fpv_1_ (808bb78) + 52
feef9ce0 _thrp_setup (fe1a0200) + 70
feef9f40 _lwp_start (fe1a0200, 0, 0, 0, 0, 0)
----------------- lwp# 3 / thread# 3 --------------------
feefdfb5 __pollsys (80a5048, 1, 0, 0) + 15
feeaf3a2 poll (80a5048, 1, ffffffff) + 52
fec92612 _XWaitForReadable (80a4a88) + de
fec9243e _XRead (80a4a88, 809abd8, 20) + 1b6
fecc74db _XReadEvents (80a4a88) + 29b
fec9e783 XNextEvent (80a4a88, fe19ef64) + 37
08068544 __1cMR6IMProtocolLxevent_loop6Fpv_1_ (808c138) + 28
feef9ce0 _thrp_setup (fe1a0a00) + 70
feef9f40 _lwp_start (fe1a0a00, 0, 0, 0, 0, 0)
Usual stacktrace with Compiz
=======================================================
hyperborea:~ > pstack core
core 'core' of 3026: Xephyr :20 -ac -extension GLX -dumb
00000000008928f6 compDestroyWindow () + 96
0000000000546be6 FreeWindowResources () + f6
000000000054b0ca DeleteWindow () + 1fa
000000000053fc1f FreeResource () + 1af
0000000000522d11 ProcDestroyWindow () + 61
0000000000529b82 Dispatch () + 4c2
000000000053cf05 main () + 495
0000000000518f5c _start () + 6c
Tail of truss when it crashes
under Compiz. There's IIIM, too.
=======================================================
read(5, "100105\0\v\0\0\0 I I I M".., 4096) = 20
read(5, 0x00D00590, 4096) Err#11 EAGAIN
writev(5, 0xFFFFFD7FFFDFF270, 1) = 32
clock_gettime(4, 0xFFFFFD7FFFDF7280) = 0
ioctl(0, FIONREAD, 0xFFFFFD7FFFDF6DC8) = 0
pollsys(0xFFFFFD7FFFDF69D0, 4, 0xFFFFFD7FFFDF7290, 0x00000000) = 1
clock_gettime(4, 0xFFFFFD7FFFDF7280) = 0
read(5, "100107\011\0\0\0 I I I M".., 4096) = 28
read(5, 0x00D00590, 4096) Err#11 EAGAIN
writev(5, 0xFFFFFD7FFFDFF270, 1) = 32
clock_gettime(4, 0xFFFFFD7FFFDF7280) = 0
ioctl(0, FIONREAD, 0xFFFFFD7FFFDF6DC8) = 0
pollsys(0xFFFFFD7FFFDF69D0, 4, 0xFFFFFD7FFFDF7290, 0x00000000) = 0
Received signal #14, SIGALRM [caught]
lwp_sigmask(SIG_SETMASK, 0x00002000, 0x00000000) = 0xFFBFFEFF [0x0000FFFF]
setitimer(ITIMER_REAL, 0xFFFFFD7FFFDF6480, 0x00000000) = 0
setcontext(0xFFFFFD7FFFDF6160)
setitimer(ITIMER_REAL, 0xFFFFFD7FFFDF7280, 0x00000000) = 0
clock_gettime(4, 0xFFFFFD7FFFDF7280) = 0
clock_gettime(4, 0xFFFFFD7FFFDF7280) = 0
ioctl(0, FIONREAD, 0xFFFFFD7FFFDF6DC8) = 0
pollsys(0xFFFFFD7FFFDF69D0, 4, 0xFFFFFD7FFFDF7290, 0x00000000) = 1
clock_gettime(4, 0xFFFFFD7FFFDF7280) = 0
read(5, "040102\0 A\0 \004 I02\0".., 4096) = 568
clock_gettime(4, 0xFFFFFD7FFFDFF0C0) = 0
Incurred fault #6, FLTBOUNDS %pc = 0x008928F6
siginfo: SIGSEGV SEGV_MAPERR addr=0x00000008
Received signal #11, SIGSEGV [default]
siginfo: SIGSEGV SEGV_MAPERR addr=0x00000008
Usual stacktrace with Compiz
32bit Xephyr for a test
=======================================================
hyperborea:~ > pstack core
core 'core' of 3228: /usr/X11/bin/i386/Xephyr :20 -ac -dumb -extension GLX
08487589 compDestroyWindow (88a7700, 1, 1, 8503000, 88a7658, 0) + a9
08103eb1 FreeWindowResources (88a7700, 8047330, 1, 0, 88a7700, 1) + e1
081081a8 DeleteWindow (88a7700, 200041, 88a7700, 84d0272, 8815894, 74) + 188
080fca94 FreeResource (200041, 0, 4, 8503000, 4, 88afee0) + 154
080de0c9 ProcDestroyWindow (88afee0, 88afee0, 8519060, 8503000, 88afee0,
881af20) + 79
0819ea1d XaceCatchDispatchProc (88afee0, 88afee0, 2, 1, 88c2038, 8a0c2bc) + 8d
080e5a49 Dispatch (2, 0, 8047a00, 80f9821, fee644c0, 80d10ac) + 4f9
080f9ca1 main (6, 8047a00, 8047a1c) + 491
080d36f0 _start (6, 8047b20, 8047b39, 8047b3d, 8047b41, 8047b47) + 80
This message posted from opensolaris.org
