* 2.6.5-rc2-mm3 blizzard of "bad: scheduling while atomic" with PREEMPT
@ 2004-03-25 18:02 Steven Cole
2004-03-25 18:35 ` Andrew Morton
0 siblings, 1 reply; 6+ messages in thread
From: Steven Cole @ 2004-03-25 18:02 UTC (permalink / raw)
To: Andrew Morton; +Cc: linux-mm
Apologies in advance if this is a known problem. I looked through
some recent archives and didn't find this exact situation, so here it
is. I'm just starting to test the -mm kernels again after a pause,
so I'm not up on the current problem set.
Kernel is 2.6.5-rc2-mm3 with SMP and PREEMPT. Box is dual PIII.
Base distro is Mandrake 10.
These messages were the start of about 8,000 lines of similar.
The "while atomic" message came out about 374 times before it
seemed to stop.
Recompiling without PREEMPT made this go away.
Once, with 2.6.5-rc2-mm3 and PREEMPT, the box hung (during password
entry) while attempting to login via the KDE graphical login. I have
not been able to reproduce this yet.
I usually reboot this box at least once a day with the current 2.6-bk
kernel, and haven't noticed any problems of that sort.
Steven
Mar 25 08:21:04 spc0 syslogd 1.4.1: restart.
Mar 25 08:21:04 spc0 kernel: klogd 1.4.1, log source = /proc/kmsg started.
Mar 25 08:21:04 spc0 kernel: Inspecting /boot/System.map-2.6.5-rc2-mm3
Mar 25 08:21:05 spc0 kernel: Loaded 20371 symbols from /boot/System.map-2.6.5-rc2-mm3.
Mar 25 08:21:05 spc0 kernel: Symbols match kernel version 2.6.5.
Mar 25 08:21:05 spc0 kernel: No module symbols loaded - kernel modules not enabled.
Mar 25 08:21:05 spc0 kernel: 0/0x260
Mar 25 08:21:05 spc0 partmon: Checking if partitions have enough free diskspace:
Mar 25 08:21:05 spc0 kernel: [do_page_fault+294/1274] do_page_fault+0x126/0x4fa
Mar 25 08:21:05 spc0 kernel: [<c0117676>] do_page_fault+0x126/0x4fa
Mar 25 08:21:05 spc0 kernel: [do_sigaction+443/656] do_sigaction+0x1bb/0x290
Mar 25 08:21:05 spc0 kernel: [<c012c92b>] do_sigaction+0x1bb/0x290
Mar 25 08:21:05 spc0 kernel: [sys_rt_sigaction+192/256] sys_rt_sigaction+0xc0/0x100
Mar 25 08:21:05 spc0 kernel: [<c012cdd0>] sys_rt_sigaction+0xc0/0x100
Mar 25 08:21:05 spc0 kernel: [getname+130/176] getname+0x82/0xb0
Mar 25 08:21:05 spc0 kernel: [<c0164012>] getname+0x82/0xb0
Mar 25 08:21:05 spc0 kernel: [sys_execve+53/112] sys_execve+0x35/0x70
Mar 25 08:21:05 spc0 kernel: [<c01073e5>] sys_execve+0x35/0x70
Mar 25 08:21:05 spc0 kernel: [sysenter_past_esp+67/101] sysenter_past_esp+0x43/0x65
Mar 25 08:21:05 spc0 kernel: [<c031b882>] sysenter_past_esp+0x43/0x65
Mar 25 08:21:05 spc0 kernel:
Mar 25 08:21:05 spc0 kernel: bad: scheduling while atomic!
Mar 25 08:21:05 spc0 kernel: Call Trace:
Mar 25 08:21:05 spc0 kernel: [schedule+1391/1472] schedule+0x56f/0x5c0
Mar 25 08:21:05 spc0 kernel: [<c011b03f>] schedule+0x56f/0x5c0
Mar 25 08:21:05 spc0 kernel: [recalc_task_prio+139/416] recalc_task_prio+0x8b/0x1a0
Mar 25 08:21:05 spc0 kernel: [<c011886b>] recalc_task_prio+0x8b/0x1a0
Mar 25 08:21:05 spc0 kernel: [wait_for_completion+156/304] wait_for_completion+0x9c/0x130
Mar 25 08:21:05 spc0 kernel: [<c011b3ec>] wait_for_completion+0x9c/0x130
Mar 25 08:21:05 spc0 kernel: [default_wake_function+0/16] default_wake_function+0x0/0x10
Mar 25 08:21:05 spc0 kernel: [<c011b0e0>] default_wake_function+0x0/0x10
Mar 25 08:21:05 spc0 kernel: [default_wake_function+0/16] default_wake_function+0x0/0x10
Mar 25 08:21:05 spc0 kernel: [<c011b0e0>] default_wake_function+0x0/0x10
Mar 25 08:21:05 spc0 kernel: [sched_migrate_task+141/176] sched_migrate_task+0x8d/0xb0
Mar 25 08:21:05 spc0 kernel: [<c011977d>] sched_migrate_task+0x8d/0xb0
Mar 25 08:21:05 spc0 kernel: [sched_balance_exec+92/128] sched_balance_exec+0x5c/0x80
Mar 25 08:21:05 spc0 kernel: [<c01198fc>] sched_balance_exec+0x5c/0x80
Mar 25 08:21:05 spc0 kernel: [do_execve+46/608] do_execve+0x2e/0x260
Mar 25 08:21:05 spc0 kernel: [<c016269e>] do_execve+0x2e/0x260
Mar 25 08:21:05 spc0 kernel: [buffered_rmqueue+269/576] buffered_rmqueue+0x10d/0x240
Mar 25 08:21:05 spc0 kernel: [<c013aafd>] buffered_rmqueue+0x10d/0x240
Mar 25 08:21:05 spc0 kernel: [__alloc_pages+176/752] __alloc_pages+0xb0/0x2f0
Mar 25 08:21:05 spc0 kernel: [<c013ace0>] __alloc_pages+0xb0/0x2f0
Mar 25 08:21:05 spc0 kernel: [page_remove_rmap+36/448] page_remove_rmap+0x24/0x1c0
Mar 25 08:21:05 spc0 kernel: [<c014be44>] page_remove_rmap+0x24/0x1c0
Mar 25 08:21:05 spc0 kernel: [do_wp_page+613/1008] do_wp_page+0x265/0x3f0
Mar 25 08:21:05 spc0 kernel: [<c01465f5>] do_wp_page+0x265/0x3f0
Mar 25 08:21:05 spc0 kernel: [handle_mm_fault+480/608] handle_mm_fault+0x1e0/0x260
Mar 25 08:21:05 spc0 kernel: [<c0147860>] handle_mm_fault+0x1e0/0x260
Mar 25 08:21:05 spc0 kernel: [do_page_fault+294/1274] do_page_fault+0x126/0x4fa
Mar 25 08:21:05 spc0 kernel: [<c0117676>] do_page_fault+0x126/0x4fa
Mar 25 08:21:05 spc0 kernel: [do_sigaction+443/656] do_sigaction+0x1bb/0x290
Mar 25 08:21:05 spc0 kernel: [<c012c92b>] do_sigaction+0x1bb/0x290
Mar 25 08:21:05 spc0 kernel: [sys_rt_sigaction+192/256] sys_rt_sigaction+0xc0/0x100
Mar 25 08:21:05 spc0 kernel: [<c012cdd0>] sys_rt_sigaction+0xc0/0x100
Mar 25 08:21:05 spc0 kernel: [getname+130/176] getname+0x82/0xb0
Mar 25 08:21:05 spc0 kernel: [<c0164012>] getname+0x82/0xb0
Mar 25 08:21:05 spc0 kernel: [sys_execve+53/112] sys_execve+0x35/0x70
Mar 25 08:21:05 spc0 kernel: [<c01073e5>] sys_execve+0x35/0x70
Mar 25 08:21:05 spc0 kernel: [sysenter_past_esp+67/101] sysenter_past_esp+0x43/0x65
Mar 25 08:21:05 spc0 kernel: [<c031b882>] sysenter_past_esp+0x43/0x65
Mar 25 08:21:05 spc0 kernel:
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: 2.6.5-rc2-mm3 blizzard of "bad: scheduling while atomic" with PREEMPT
2004-03-25 18:02 2.6.5-rc2-mm3 blizzard of "bad: scheduling while atomic" with PREEMPT Steven Cole
@ 2004-03-25 18:35 ` Andrew Morton
2004-03-25 18:38 ` Ingo Molnar
` (2 more replies)
0 siblings, 3 replies; 6+ messages in thread
From: Andrew Morton @ 2004-03-25 18:35 UTC (permalink / raw)
To: Steven Cole; +Cc: linux-mm, Ingo Molnar, Nick Piggin
Steven Cole <elenstev@mesatop.com> wrote:
>
> Apologies in advance if this is a known problem. I looked through
> some recent archives and didn't find this exact situation, so here it
> is. I'm just starting to test the -mm kernels again after a pause,
> so I'm not up on the current problem set.
>
> Kernel is 2.6.5-rc2-mm3 with SMP and PREEMPT. Box is dual PIII.
> Base distro is Mandrake 10.
>
> These messages were the start of about 8,000 lines of similar.
> The "while atomic" message came out about 374 times before it
> seemed to stop.
>
> Recompiling without PREEMPT made this go away.
err, yes. Ingo broke it ;)
bad: scheduling while atomic!
Call Trace:
[<c011d0f0>] schedule+0x3c/0x58c
[<c0109f69>] dump_stack+0x19/0x20
[<c011f146>] __might_sleep+0xaa/0xb4
[<c011d91a>] wait_for_completion+0xae/0x110
[<c011d688>] default_wake_function+0x0/0x1c
[<c011d688>] default_wake_function+0x0/0x1c
[<c011be8b>] sched_migrate_task+0x6b/0x9c
[<c011c023>] sched_balance_exec+0x63/0x8c
[<c0167108>] do_execve+0x14/0x200
[<c0141ef5>] buffered_rmqueue+0x1c5/0x1d4
[<c0141fae>] __alloc_pages+0xaa/0x2fc
[<c014c5c2>] do_wp_page+0x3c2/0x478
[<c014d6c3>] handle_mm_fault+0x107/0x174
[<c01196b5>] do_page_fault+0x15d/0x4ba
[<c0119558>] do_page_fault+0x0/0x4ba
[<c02865b6>] tiocspgrp+0x72/0x9c
[<c012aca9>] recalc_sigpending+0x11/0x18
[<c0168a6f>] getname+0x5f/0xa0
[<c0107a6f>] sys_execve+0x2f/0x68
[<c0109185>] sysenter_past_esp+0x52/0x71
Fortunately vger seems to have gobbled the mm3 announcement.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: 2.6.5-rc2-mm3 blizzard of "bad: scheduling while atomic" with PREEMPT
2004-03-25 18:35 ` Andrew Morton
@ 2004-03-25 18:38 ` Ingo Molnar
2004-03-25 19:06 ` Ingo Molnar
2004-03-25 19:16 ` Ingo Molnar
2 siblings, 0 replies; 6+ messages in thread
From: Ingo Molnar @ 2004-03-25 18:38 UTC (permalink / raw)
To: Andrew Morton; +Cc: Steven Cole, linux-mm, Nick Piggin
* Andrew Morton <akpm@osdl.org> wrote:
> > Recompiling without PREEMPT made this go away.
>
> err, yes. Ingo broke it ;)
ugh. Checking it.
Ingo
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: 2.6.5-rc2-mm3 blizzard of "bad: scheduling while atomic" with PREEMPT
2004-03-25 18:35 ` Andrew Morton
2004-03-25 18:38 ` Ingo Molnar
@ 2004-03-25 19:06 ` Ingo Molnar
2004-03-25 19:17 ` Andrew Morton
2004-03-25 19:16 ` Ingo Molnar
2 siblings, 1 reply; 6+ messages in thread
From: Ingo Molnar @ 2004-03-25 19:06 UTC (permalink / raw)
To: Andrew Morton; +Cc: Steven Cole, linux-mm, Nick Piggin
* Andrew Morton <akpm@osdl.org> wrote:
> > Recompiling without PREEMPT made this go away.
>
> err, yes. Ingo broke it ;)
ok, this replacement patch should fix it:
http://redhat.com/~mingo/scheduler-patches/sched-2.6.5-rc2-mm2-A5
(i'll soon send a fix against -mm3 too.)
the patch also includes prototype fixes for the NUMA fork() balancing
problem.
Ingo
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: 2.6.5-rc2-mm3 blizzard of "bad: scheduling while atomic" with PREEMPT
2004-03-25 18:35 ` Andrew Morton
2004-03-25 18:38 ` Ingo Molnar
2004-03-25 19:06 ` Ingo Molnar
@ 2004-03-25 19:16 ` Ingo Molnar
2 siblings, 0 replies; 6+ messages in thread
From: Ingo Molnar @ 2004-03-25 19:16 UTC (permalink / raw)
To: Andrew Morton; +Cc: Steven Cole, linux-mm, Nick Piggin
> bad: scheduling while atomic!
> Call Trace:
> [<c011d0f0>] schedule+0x3c/0x58c
> [<c0109f69>] dump_stack+0x19/0x20
> [<c011f146>] __might_sleep+0xaa/0xb4
> [<c011d91a>] wait_for_completion+0xae/0x110
> [<c011d688>] default_wake_function+0x0/0x1c
> [<c011d688>] default_wake_function+0x0/0x1c
> [<c011be8b>] sched_migrate_task+0x6b/0x9c
> [<c011c023>] sched_balance_exec+0x63/0x8c
the patch against -mm3 is at:
redhat.com/~mingo/scheduler-patches/sched-2.6.5-rc2-mm3-A0
this also includes the fork-time balancing changes.
Ingo
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: 2.6.5-rc2-mm3 blizzard of "bad: scheduling while atomic" with PREEMPT
2004-03-25 19:06 ` Ingo Molnar
@ 2004-03-25 19:17 ` Andrew Morton
0 siblings, 0 replies; 6+ messages in thread
From: Andrew Morton @ 2004-03-25 19:17 UTC (permalink / raw)
To: Ingo Molnar; +Cc: elenstev, linux-mm, nickpiggin
Ingo Molnar <mingo@elte.hu> wrote:
>
> ok, this replacement patch should fix it:
>
> http://redhat.com/~mingo/scheduler-patches/sched-2.6.5-rc2-mm2-A5
Thanks, I swapped out the old one for this.
There are rejects against ppc64 files. But I dropped the ppc64
sched-domain patches from rc2-mm2, so I'm not sure what's going on there.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2004-03-25 19:17 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2004-03-25 18:02 2.6.5-rc2-mm3 blizzard of "bad: scheduling while atomic" with PREEMPT Steven Cole
2004-03-25 18:35 ` Andrew Morton
2004-03-25 18:38 ` Ingo Molnar
2004-03-25 19:06 ` Ingo Molnar
2004-03-25 19:17 ` Andrew Morton
2004-03-25 19:16 ` Ingo Molnar
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox