Skip to contentSkip to conversation

2.6.23.8, ondemand scaling governor: "BUG: soft lockup detected on CPU#0!"

DISCUSSION Harald Dunkel 2 participants 5 messages Nov 20, 2007

lore
Share & tools
lore
#0 lore raw

Hi folks,

using the ondemand scaling governour I see some error messages in kern.log, e.g.:

Nov 20 01:00:46 bugs kernel: BUG: soft lockup detected on CPU#0!
Nov 20 01:00:46 bugs kernel:  [<c013cf8d>] softlockup_tick+0x91/0xa6
Nov 20 01:00:46 bugs kernel:  [<c012269c>] update_process_times+0x3a/0x5d
Nov 20 01:00:46 bugs kernel:  [<c0131219>] tick_sched_timer+0x115/0x164
Nov 20 01:00:46 bugs kernel:  [<c012d311>] hrtimer_interrupt+0x102/0x191
Nov 20 01:00:46 bugs kernel:  [<c0106cd6>] timer_interrupt+0x2e/0x34
Nov 20 01:00:46 bugs kernel:  [<c013d1f6>] handle_IRQ_event+0x1a/0x3f
Nov 20 01:00:46 bugs kernel:  [<c013e4e1>] handle_level_irq+0xa8/0xb7
Nov 20 01:00:46 bugs kernel:  [<c0106367>] do_IRQ+0x53/0x6c
Nov 20 01:00:46 bugs kernel:  [<c0104853>] common_interrupt+0x23/0x28
Nov 20 01:00:46 bugs kernel:  [<c011007b>] smp_apic_timer_interrupt+0x1a/0x70
Nov 20 01:00:46 bugs kernel:  [<c0102a36>] default_idle+0x27/0x39
Nov 20 01:00:46 bugs kernel:  [<c010234c>] cpu_idle+0x46/0x68
Nov 20 01:00:46 bugs kernel:  [<c032e9e8>] start_kernel+0x24d/0x252
Nov 20 01:00:46 bugs kernel:  [<c032e317>] unknown_bootoption+0x0/0x196
Nov 20 01:00:46 bugs kernel:  =======================

This seems to happen when the load drops below the threshold and the ondemand governor changes the CPU from 2GHz to 400MHz. If I use the "performance" governor instead, then there is no such message. If I set it back to "ondemand", then the message is printed immediately.

# uname -a
Linux bugs 2.6.23.8 #1 PREEMPT Sun Nov 18 09:14:13 CET 2007 i686 GNU/Linux
# cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 13
model name      : Intel(R) Pentium(R) M processor 2.00GHz
stepping        : 8
cpu MHz         : 400.000
cache size      : 2048 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 2
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss tm pbe nx bts est tm2
bogomips        : 798.34
clflush size    : 64

Please mail if I can help to track this down.

Regards

Harri

#1 lore raw
Quoted 68 lines

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Harald Dunkel
Sent: Monday, November 19, 2007 4:19 PM
To: Linux Kernel list
Subject: 2.6.23.8, ondemand scaling governor: "BUG: soft lockup detected on CPU#0!"

Hi folks,

using the ondemand scaling governour I see some error messages in kern.log, e.g.:

Nov 20 01:00:46 bugs kernel: BUG: soft lockup detected on CPU#0!
Nov 20 01:00:46 bugs kernel:  [<c013cf8d>] softlockup_tick+0x91/0xa6
Nov 20 01:00:46 bugs kernel:  [<c012269c>] 
update_process_times+0x3a/0x5d
Nov 20 01:00:46 bugs kernel:  [<c0131219>] tick_sched_timer+0x115/0x164
Nov 20 01:00:46 bugs kernel:  [<c012d311>] 
hrtimer_interrupt+0x102/0x191
Nov 20 01:00:46 bugs kernel:  [<c0106cd6>] timer_interrupt+0x2e/0x34
Nov 20 01:00:46 bugs kernel:  [<c013d1f6>] handle_IRQ_event+0x1a/0x3f
Nov 20 01:00:46 bugs kernel:  [<c013e4e1>] handle_level_irq+0xa8/0xb7
Nov 20 01:00:46 bugs kernel:  [<c0106367>] do_IRQ+0x53/0x6c
Nov 20 01:00:46 bugs kernel:  [<c0104853>] common_interrupt+0x23/0x28
Nov 20 01:00:46 bugs kernel:  [<c011007b>] 
smp_apic_timer_interrupt+0x1a/0x70
Nov 20 01:00:46 bugs kernel:  [<c0102a36>] default_idle+0x27/0x39
Nov 20 01:00:46 bugs kernel:  [<c010234c>] cpu_idle+0x46/0x68
Nov 20 01:00:46 bugs kernel:  [<c032e9e8>] start_kernel+0x24d/0x252
Nov 20 01:00:46 bugs kernel:  [<c032e317>] unknown_bootoption+0x0/0x196
Nov 20 01:00:46 bugs kernel:  =======================

This seems to happen when the load drops below the threshold and the ondemand governor changes the CPU from 2GHz to 400MHz. If I use the "performance" governor instead, then there is no such message. If I set it back to "ondemand", then the message is printed immediately.

# uname -a
Linux bugs 2.6.23.8 #1 PREEMPT Sun Nov 18 09:14:13 CET 2007 
i686 GNU/Linux
# cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 13
model name      : Intel(R) Pentium(R) M processor 2.00GHz
stepping        : 8
cpu MHz         : 400.000
cache size      : 2048 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 2
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 sep mtrr 
pge mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss tm pbe 
nx bts est tm2
bogomips        : 798.34
clflush size    : 64
Quoted 3 lines

Please mail if I can help to track this down.

Can you try switching to powersave governor (which should always run CPU at 400MHz) and see whether you see similar error?

Thanks,
Venki

#2 lore raw
Quoted 4 lines · Venkatesh

Can you try switching to powersave governor (which should always run CPU at 400MHz) and see whether you see similar error?

Yes, if I move from performance to powersave, then I see a similar error:

Nov 20 09:06:48 bugs kernel: BUG: soft lockup detected on CPU#0!
Nov 20 09:06:48 bugs kernel:  [<c013cf8d>] softlockup_tick+0x91/0xa6
Nov 20 09:06:48 bugs kernel:  [<c012269c>] update_process_times+0x3a/0x5d
Nov 20 09:06:48 bugs kernel:  [<c0131219>] tick_sched_timer+0x115/0x164
Nov 20 09:06:48 bugs kernel:  [<c012d311>] hrtimer_interrupt+0x102/0x191
Nov 20 09:06:48 bugs kernel:  [<c0106cd6>] timer_interrupt+0x2e/0x34
Nov 20 09:06:48 bugs kernel:  [<c013d1f6>] handle_IRQ_event+0x1a/0x3f
Nov 20 09:06:48 bugs kernel:  [<c013e4e1>] handle_level_irq+0xa8/0xb7
Nov 20 09:06:48 bugs kernel:  [<c0106367>] do_IRQ+0x53/0x6c
Nov 20 09:06:48 bugs kernel:  [<c0104853>] common_interrupt+0x23/0x28
Nov 20 09:06:48 bugs kernel:  =======================

Please mail. Regards

Harri

#3 lore raw
Quoted 20 lines

-----Original Message-----
From: Harald Dunkel [mailto:[email protected]]
Sent: Tuesday, November 20, 2007 12:17 AM
To: Pallipadi, Venkatesh
Cc: Linux Kernel list
Subject: Re: 2.6.23.8, ondemand scaling governor: "BUG: soft lockup detected on CPU#0!"

Quoted 2 lines · Venkatesh

Can you try switching to powersave governor (which should

always run CPU

Quoted 2 lines

at 400MHz) and see whether you see similar error?

Yes, if I move from performance to powersave, then I see a similar error:

Quoted 15 lines
Nov 20 09:06:48 bugs kernel: BUG: soft lockup detected on CPU#0!
Nov 20 09:06:48 bugs kernel:  [<c013cf8d>] softlockup_tick+0x91/0xa6
Nov 20 09:06:48 bugs kernel:  [<c012269c>] 
update_process_times+0x3a/0x5d
Nov 20 09:06:48 bugs kernel:  [<c0131219>] tick_sched_timer+0x115/0x164
Nov 20 09:06:48 bugs kernel:  [<c012d311>] 
hrtimer_interrupt+0x102/0x191
Nov 20 09:06:48 bugs kernel:  [<c0106cd6>] timer_interrupt+0x2e/0x34
Nov 20 09:06:48 bugs kernel:  [<c013d1f6>] handle_IRQ_event+0x1a/0x3f
Nov 20 09:06:48 bugs kernel:  [<c013e4e1>] handle_level_irq+0xa8/0xb7
Nov 20 09:06:48 bugs kernel:  [<c0106367>] do_IRQ+0x53/0x6c
Nov 20 09:06:48 bugs kernel:  [<c0104853>] common_interrupt+0x23/0x28
Nov 20 09:06:48 bugs kernel:  =======================

This looks like TSC related issue. Ingo's patch commit id a3b13c23f186ecb57204580cc1f2dbe9c284953a
http://git.kernel.org/gitweb.cgi?p=linux/kernel/git/torvalds/linux-2.6.g
it;a=commit;h=a3b13c23f186ecb57204580cc1f2dbe9c284953a should help.

Thanks,
Venki