Xglamo loops and hangs (and crashed before)
Harald Koenig
koenig at tat.physik.uni-tuebingen.de
Tue Aug 26 17:20:52 CEST 2008
On Aug 26, Holger Freyther wrote:
> > Xglamo sucks cpu time and doesn't accept new xclients, they just hang.
> > this is the tail of "ps uaxOT" output:
>
> (and top would be the better utility to see what sucks your cpu...)
of course I used top to monitor it -- but the nice dynamic output of top
doesn't really fit into a static email (and I don't want to include a mpeg movie ;-))
voila:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3768 root 19 -1 14280 4448 448 R 95.7 3.6 259:56.42 Xglamo
8169 root 20 0 2252 1108 884 R 3.7 0.9 0:00.56 top
5598 root 20 0 2400 620 316 S 0.3 0.5 1:43.87 dropbear
...
> https://docs.openmoko.org/trac/ticket/1315
>
> if you find a way to reproduce this with up to five minutes of usage that
> would be very welcome.
seems to be difficult, since it happend to be for the first time now (fortuneately;).
is there a reasonable way to debug the Xglamo ? I still keep it running/looping
to get more information what's going on...
[where] can I get debug info for that Xglamo binary ?
I tried to attach gdb to Xglamo and single step it a bit
-------------------------------------------------------------------------------
0x00014ee8 in ?? ()
(gdb) where
#0 0x00014ee8 in ?? ()
(gdb) disp/i $pc
1: x/i $pc 0x14ee8 <shmget+10460>: and r1, r12, r7
(gdb) si
0x00014eec in ?? ()
1: x/i $pc 0x14eec <shmget+10464>: cmp r1, r6
(gdb)
0x00014ef0 in ?? ()
1: x/i $pc 0x14ef0 <shmget+10468>: beq 0x14f40 <shmget+10548>
(gdb)
0x00014ef4 in ?? ()
1: x/i $pc 0x14ef4 <shmget+10472>: mov r1, #0 ; 0x0
(gdb)
0x00014ef8 in ?? ()
1: x/i $pc 0x14ef8 <shmget+10476>: mov r0, sp
(gdb)
0x00014efc in ?? ()
1: x/i $pc 0x14efc <shmget+10480>: bl 0x12048 <gettimeofday>
(gdb)
0x00012048 in ?? ()
1: x/i $pc 0x12048 <gettimeofday>: add r12, pc, #1048576 ; 0x100000
(gdb)
0x0001204c in ?? ()
1: x/i $pc 0x1204c <gettimeofday+4>: add r12, r12, #0 ; 0x0
(gdb)
0x00012050 in ?? ()
1: x/i $pc 0x12050 <gettimeofday+8>: ldr pc, [r12, #1144]!
(gdb)
0x401f88a0 in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88a0 <gettimeofday>: mov r12, r7
(gdb)
0x401f88a4 in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88a4 <gettimeofday+4>: mov r7, #78 ; 0x4e
(gdb)
0x401f88a8 in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88a8 <gettimeofday+8>: svc 0x00000000
(gdb)
0x401f88ac in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88ac <gettimeofday+12>: mov r7, r12
(gdb)
0x401f88b0 in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88b0 <gettimeofday+16>: cmn r0, #4096 ; 0x1000
(gdb)
0x401f88b4 in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88b4 <gettimeofday+20>: bxcc lr
(gdb)
[ gdb hangs and "wait" for CTRL-C :-[
Program received signal SIGINT, Interrupt.
0x401f88ac in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88ac <gettimeofday+12>: mov r7, r12
(gdb) si
0x401f88b0 in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88b0 <gettimeofday+16>: cmn r0, #4096 ; 0x1000
(gdb)
0x401f88b4 in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88b4 <gettimeofday+20>: bxcc lr
(gdb)
Program received signal SIGINT, Interrupt.
0x401f88ac in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88ac <gettimeofday+12>: mov r7, r12
(gdb) si
0x401f88b0 in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88b0 <gettimeofday+16>: cmn r0, #4096 ; 0x1000
(gdb)
0x401f88b4 in gettimeofday () from /lib/libc.so.6
1: x/i $pc 0x401f88b4 <gettimeofday+20>: bxcc lr
(gdb)
Program received signal SIGINT, Interrupt.
0x401f88ac in gettimeofday () from /lib/libc.so.6
-------------------------------------------------------------------------------
this hang in gettimeofday matches the $PC address which "strace -i" showed.
how can I do instruction stepping (si) in gdb on OM ?
any other suggestion how to analyze Xglamo ?
thanks for any hint!
Harald
--
"I hope to die ___ _____
before I *have* to use Microsoft Word.", 0--,| /OOOOOOO\
Donald E. Knuth, 02-Oct-2001 in Tuebingen. <_/ / /OOOOOOOOOOO\
\ \/OOOOOOOOOOOOOOO\
\ OOOOOOOOOOOOOOOOO|//
\/\/\/\/\/\/\/\/\/
Harald Koenig // / \\ \
koenig at tat.physik.uni-tuebingen.de ^^^^^ ^^^^^
More information about the devel
mailing list