Arch Linux

Aakko

Member

Registered: 2007-10-23

Posts: 65

Kernel 2.6.23.9-1 BUG: soft lockup detected on CPU#0!

With Kernel 2.6.23.9-1 I get dmesg flooded with cpu soft lockup bugs like this:

BUG: soft lockup detected on CPU#0!
 [<c0157d9a>] softlockup_tick+0xea/0x120
 [<c0135853>] update_process_times+0x33/0x80
 [<c0149077>] tick_sched_timer+0x77/0xf0
 [<c0143fc3>] hrtimer_interrupt+0x163/0x1f0
 [<c0149000>] tick_sched_timer+0x0/0xf0
 [<c0107d61>] timer_interrupt+0x31/0x40
 [<c0158140>] handle_IRQ_event+0x30/0x60
 [<c0159a9d>] handle_level_irq+0x7d/0xf0
 [<c010717b>] do_IRQ+0x3b/0x70
 [<c0104ee7>] common_interrupt+0x23/0x28
 [<c02c2d50>] mousedev_poll+0x0/0x50
 [<c02c2d53>] mousedev_poll+0x3/0x50
 [<c018bdb0>] do_select+0x260/0x4a0
 [<c018c560>] __pollwait+0x0/0xe0
 [<c01260e0>] default_wake_function+0x0/0x10
 [<c01260e0>] default_wake_function+0x0/0x10
 [<c01260e0>] default_wake_function+0x0/0x10
 [<c01260e0>] default_wake_function+0x0/0x10
 [<c01260e0>] default_wake_function+0x0/0x10
 [<c01260e0>] default_wake_function+0x0/0x10
 [<c01260e0>] default_wake_function+0x0/0x10
 [<c01260e0>] default_wake_function+0x0/0x10
 [<c01260e0>] default_wake_function+0x0/0x10
 [<c0125d91>] try_to_wake_up+0x41/0x390
 [<c012415d>] update_curr+0x12d/0x140
 [<c0123a8c>] update_stats_wait_end+0x9c/0xd0
 [<c01030d3>] __switch_to+0xc3/0x150
 [<c035fb40>] schedule+0x1e0/0x600
 [<c035fb68>] schedule+0x208/0x600
 [<c0123709>] __wake_up_common+0x39/0x60
 [<c0124ac8>] __wake_up+0x38/0x50
 [<c03600c3>] preempt_schedule+0x53/0x80
 [<c033e208>] unix_stream_sendmsg+0x1a8/0x350
 [<c033e1b6>] unix_stream_sendmsg+0x156/0x350
 [<c02cd4f8>] sock_aio_write+0xe8/0x100
 [<c018c1c8>] core_sys_select+0x1d8/0x300
 [<c01035cb>] setup_sigcontext+0x10b/0x190
 [<c01360cb>] recalc_sigpending+0xb/0x20
 [<c0103ddc>] do_notify_resume+0x3ac/0x720
 [<c01095a5>] convert_fxsr_from_user+0x15/0xe0
 [<c018c72a>] sys_select+0xea/0x1b0
 [<c0104482>] sysenter_past_esp+0x6b/0xa1
 =======================

I have laptop with Athlon XP M -processor. I think the bug has something to do with the new kernel's scheduler  The bugs started to appear when i upgraded to Kernel 2.6.23.9-1 from the repo.

Fackamato

Member

Registered: 2006-03-31

Posts: 579

Re: Kernel 2.6.23.9-1 BUG: soft lockup detected on CPU#0!

From what kernel did you upgrade? (I.e. which kernel did *not* produce these errors?)

Maybe your CPU is overheating with the new kernel, and therefore produces these errors? (just a quick thought)

Aakko

Member

Registered: 2007-10-23

Posts: 65

Re: Kernel 2.6.23.9-1 BUG: soft lockup detected on CPU#0!

I upgraded from kernel26-2.6.23.8-1 and after that the bugs started to appear.

I checked cpu-temperature values and it seems that the problem isn't overheating.

elide

Member

From: Russia

Registered: 2007-12-02

Posts: 40

Re: Kernel 2.6.23.9-1 BUG: soft lockup detected on CPU#0!

That's very strange.
I've read at LKML, that this bug is specific for 2.6.23.8 and it is fixed in 2.6.23.9.
I have this bug in .8
Try to unload acpi-cpufreq and all cpufreq governors. since I have this bug only with cpufreq-ondemand, but not with cpufreq-performance.

Doehni

Member

From: Germany

Registered: 2007-05-19

Posts: 175

Re: Kernel 2.6.23.9-1 BUG: soft lockup detected on CPU#0!

Hi
I'm also having that issue.
My kernel: 2.6.23.8-1
My Processor: Pentium 4 mobile (using acpi-cpufreq...)

My dmesg code:

BUG: soft lockup detected on CPU#0!
 [<c0157d9a>] softlockup_tick+0xea/0x120
 [<c0135853>] update_process_times+0x33/0x80
 [<c0149077>] tick_sched_timer+0x77/0xf0
 [<c0143fc3>] hrtimer_interrupt+0x163/0x1f0
 [<c0149000>] tick_sched_timer+0x0/0xf0
 [<c0107d61>] timer_interrupt+0x31/0x40
 [<c0158140>] handle_IRQ_event+0x30/0x60
 [<c0159a9d>] handle_level_irq+0x7d/0xf0
 [<c010717b>] do_IRQ+0x3b/0x70
 [<c010ac33>] sched_clock+0x13/0x30
 [<c0104ee7>] common_interrupt+0x23/0x28
 [<dcf460b0>] acpi_processor_idle+0x29c/0x434 [processor]
 [<c0102453>] cpu_idle+0x73/0xe0
 [<c0435a6a>] start_kernel+0x30a/0x3a0
 [<c0435140>] unknown_bootoption+0x0/0x1f0
 =======================

dale77

Member

From: Down under

Registered: 2007-02-10

Posts: 102

Website

Re: Kernel 2.6.23.9-1 BUG: soft lockup detected on CPU#0!

Hmm. I'm also seeing this in 2.6.23.8-1

Athlon64 3000+

Trying out .9-1 now.

dale77

Member

From: Down under

Registered: 2007-02-10

Posts: 102

Website

Re: Kernel 2.6.23.9-1 BUG: soft lockup detected on CPU#0!

Seems OK now with 2.6.23.9-1

Doehni

Member

From: Germany

Registered: 2007-05-19

Posts: 175

Re: Kernel 2.6.23.9-1 BUG: soft lockup detected on CPU#0!

I'm also not having any lockup with the 2.6.23.9-1 kernel.

Aakko

Member

Registered: 2007-10-23

Posts: 65

Re: Kernel 2.6.23.9-1 BUG: soft lockup detected on CPU#0!

Try to unload acpi-cpufreq and all cpufreq governors. since I have this bug only with cpufreq-ondemand, but not with cpufreq-performance.

I changed cpufreq governor from ondemand to performance and the bugs are gone.