Hi,
We run into a process hang, ps -elf shows the status as 'D'. The stack
trace shows it might be xfs_read. We are running xfs 1.0.1 with a few
more patches from the TAKES, see list below on powerpc linux kernel
2.4.2 by MontaVista. This is reproducable though the duration to make it
happen varies, took around an hour last time and the process can vary
and has included insmod and rpm.
Any comments or directions are welcome. I can email the System.map or
other files as required.
Thanks
Amit
------------------------------------------------------
ps -elf
cp1:root> ps -elf
F S UID PID PPID C PRI NI ADDR SZ WCHAN STIME TTY
TIME CMD
100 S root 1 0 1 68 0 - 312 4a630 14:41 ?
00:00:04 init
040 S root 2 1 0 69 0 - 0 22140 14:41 ?
00:00:00 [keventd]
040 S root 3 1 0 69 0 - 0 2fe10 14:41 ?
00:00:00 [kswapd]
040 S root 4 1 0 69 0 - 0 2ff04 14:41 ?
00:00:00 [kreclaimd]
040 S root 5 1 0 69 0 - 0 3c358 14:41 ?
00:00:00 [bdflush]
040 S root 6 1 0 69 0 - 0 3c450 14:41 ?
00:00:00 [kupdate]
040 S root 7 1 0 69 0 - 0 83664 14:41 ?
00:00:00 [pagebuf_daemon]
000 S root 93 1 0 69 0 - 700 16bf0 14:41 ?
00:00:00 /bin/sh /etc/rc.d/rc 3
140 S bin 126 1 0 69 0 - 310 4a630 14:41 ?
00:00:00 /sbin/portmap
140 S root 136 1 0 69 0 - 342 4a630 14:41 ?
00:00:00 /usr/sbin/inetd
040 S root 166 1 0 69 0 - 341 4a630 14:41 ?
00:00:00 /usr/sbin/syslogd -m 0
140 S root 167 1 0 69 0 - 307 1286c 14:41 ?
00:00:00 /usr/sbin/klogd -x
000 S root 175 93 0 69 0 - 569 16bf0 14:41 ?
00:00:00 /bin/sh /etc/rc.d/rc3.d/S80fstest start
000 D root 309 175 0 69 0 - 598 26be4 14:42 ?
00:00:01 rpm -V apache-1.3.14-2
000 S root 310 175 0 69 0 - 340 42f14 14:42 ?
00:00:00 grep /lib
000 S root 311 136 1 73 0 - 520 4a630 14:45 ?
00:00:00 in.telnetd: 192.168.198.232
100 S root 312 311 1 72 0 - 743
16bf0 14:45 pts/0 00:00:00 -sh
000 R root 329 312 0 74 0 - 651 - 14:46 pts/0
00:00:00 ps -elf
>>
>> ps
ADDR UID PID PPID STATE FLAGS NAME
===============================================================================
c01bd020 0 0 0 0 0 swapper
c0454000 0 1 0 1 100 init
c7e02000 0 2 1 1 40 keventd
c04fe000 0 3 1 1 840 kswapd
c04fc000 0 4 1 1 840 kreclaimd
c04fa000 0 5 1 1 40 bdflush
c04f8000 0 6 1 1 40 kupdate
c7e46000 0 7 1 1 840 pagebuf_daemon
c778a000 0 93 1 1 0 rc
c781e000 1 126 1 1 140 portmap
c7fb0000 0 136 1 1 140 inetd
c7f6e000 0 166 1 1 40 syslogd
c7f00000 0 167 1 1 140 klogd
c7358000 0 175 93 1 0 S80fstest
c67b0000 0 309 175 2 0 rpm
c6f7e000 0 310 175 1 0 grep
c6164000 0 311 136 1 0 in.telnetd
c6148000 0 312 311 1 100 sh
c5efa000 0 330 312 0 100 gdbserver
===============================================================================
19 active task structs found
>> bt 0xc67b0000
=====================================================================
STACK TRACE FOR TASK: 0xc67b0000 (rpm)
0 _switch_to+104 [0xc00052e0]
1 schedule+792 [0xc000f86c]
2 __lock_page+144 [0xc0026be4]
3 lock_page+32 [0xc0026c5c]
4 do_generic_file_read+628 [0xc0027524]
5 generic_file_read+64 [0xc0027920]
6 xfs_read+380 [0xc00e48dc]
7 linvfs_read+100 [0xc00e02a0]
8 sys_read+132 [0xc0036f70]
9 ret_from_syscall_1 [0xc00026b4]
10 <stack chain invalid (reverse stack direction)>+<ERROR> [0x0]
=====================================================================
>>
TAKE - work around gcc bug during xfs_growfs
http://marc.theaimsgroup.com/?l=linux-xfs&m=100274698924296&w=2
URL:
Re: Busy inodes after umount
http://marc.theaimsgroup.com/?l=linux-xfs&m=99558966006128&w=2
http://marc.theaimsgroup.com/?l=linux-xfs&m=99670088622262&w=2
TAKE - fix hang with mmap writes to a full filesystem
http://marc.theaimsgroup.com/?l=linux-xfs&m=100206169000709&w=2
TAKE - fix a hang in parallel copies & dbench
http://marc.theaimsgroup.com/?l=linux-xfs&m=100196296016505&w=2
|