Re: RCU hang on cpu re-hotplug with 2.6.27rc8

From: Andi Kleen
Date: Tue Oct 07 2008 - 03:09:50 EST


> Could you please try the patch at the following URL (from Thomas
> Gleixner)?
>
> http://www.rdrop.com/users/paulmck/patches/2.6.27-rc7-tglx-timer-1.patch
>
> This fixed some CPU hotplug hangs that I was seeing in 2.6.27-rc7 and
> -rc8. Alternatively, try 2.6.27-rc9, which seems to include Thomas's
> patch.

Still happens with rc9. This time with frame pointers:

bash D 00000000ffff975e 0 4770 4757
ffff8804bcda3bf8 0000000000000082 ffff8804bcda3c18 0000000000000296
ffff8804bde4e6b0 ffff8804be6fc630 ffff8804bde4e8f8 0000000e00000296
ffff8804bcda3c18 ffffffff808cfd98 ffff880288064780 ffff88027d18ec98
Call Trace:
[<ffffffff805c4915>] schedule_timeout+0x22/0xb4
[<ffffffff8020a009>] ? __switch_to+0x320/0x330
[<ffffffff802602ed>] ? cpupri_set+0xc5/0xd8
[<ffffffff805c476f>] wait_for_common+0xcd/0x131
[<ffffffff8022db87>] ? default_wake_function+0x0/0xf
[<ffffffff805c485d>] wait_for_completion+0x18/0x1a
[<ffffffff80243fbf>] synchronize_rcu+0x35/0x3c
[<ffffffff8024403e>] ? wakeme_after_rcu+0x0/0x12
[<ffffffff8022ed25>] partition_sched_domains+0x9b/0x1dd
[<ffffffff8022dbb3>] ? wake_up_process+0x10/0x12
[<ffffffff8022ee95>] update_sched_domains+0x2e/0x35
[<ffffffff805c8342>] notifier_call_chain+0x33/0x5b
[<ffffffff80249291>] __raw_notifier_call_chain+0x9/0xb
[<ffffffff802492a2>] raw_notifier_call_chain+0xf/0x11
[<ffffffff805c1c3a>] _cpu_up+0xd3/0x10c
[<ffffffff805c1cca>] cpu_up+0x57/0x67
[<ffffffff805a5787>] store_online+0x4d/0x75
[<ffffffff803e67bb>] sysdev_store+0x1b/0x1d
[<ffffffff802cce10>] sysfs_write_file+0xe0/0x11c
[<ffffffff8028b661>] vfs_write+0xae/0x137
[<ffffffff8028bb0a>] sys_write+0x47/0x6f
[<ffffffff8020b34b>] system_call_fastpath+0x16/0x1b

BTW it also happens reliably every time. Reproduced it three times
now. I just do

laut:~ # echo 0 > /sys/devices/system/cpu/cpu15/online
laut:~ # echo 0 > /sys/devices/system/cpu/cpu14/online
laut:~ # echo 1 > /sys/devices/system/cpu/cpu14/online
laut:~ # echo 1 > /sys/devices/system/cpu/cpu15/online
<hang>

-Andi

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/