Re: khugepaged blocks suspend2ram (3.2.0-rc1-00252-g8f042aa)

[Hillf Danton]

On Sat, Nov 12, 2011 at 3:48 PM, Sergei Trofimovich <slyich@xxxxxxxxx> wrote:
> $ cat /sys/kernel/mm/transparent_hugepage/enabled
> [always] madvise never
>
> $ echo mem > /sys/power/state
>
> [ 4036.239222] PM: Syncing filesystems ... done.
> [ 4040.198755] PM: Preparing system for mem sleep
> [ 4040.625898] Freezing user space processes ... (elapsed 0.01 seconds) done.
> [ 4040.637103] Freezing remaining freezable tasks ...
> [ 4060.646142] Freezing of tasks failed after 20.00 seconds (1 tasks refusing to freeze, wq_busy=0):
> [ 4060.646180] khugepaged   ÂS 0000000000000000   0  514   Â2 0x00800000
> [ 4060.646194] Âffff88007b231c70 0000000000000046 ffff88007b231c00 ffffffff00000000
> [ 4060.646209] Âffff88007bb40000 0000000000010ac0 ffff88007b231fd8 ffff88007b230000
> [ 4060.646222] Â0000000000010ac0 0000000000004000 ffff88007b231fd8 0000000000010ac0
> [ 4060.646235] Call Trace:
> [ 4060.646254] Â[<ffffffff810af8b1>] ? try_to_free_pages+0x71/0x80
> [ 4060.646265] Â[<ffffffff8103eaf1>] ? get_parent_ip+0x11/0x50
> [ 4060.646275] Â[<ffffffff8103ec9d>] ? add_preempt_count+0x9d/0xd0
> [ 4060.646286] Â[<ffffffff81432f3a>] schedule+0x3a/0x50
> [ 4060.646294] Â[<ffffffff81433408>] schedule_timeout+0x158/0x230
> [ 4060.646304] Â[<ffffffff81052db0>] ? init_timer_deferrable_key+0x20/0x20
> [ 4060.646314] Â[<ffffffff81433539>] schedule_timeout_interruptible+0x19/0x20
> [ 4060.646324] Â[<ffffffff810de107>] khugepaged_alloc_hugepage+0xc7/0xf0
> [ 4060.646334] Â[<ffffffff81063700>] ? wake_up_bit+0x40/0x40
> [ 4060.646344] Â[<ffffffff814351f2>] ? _raw_spin_unlock_irqrestore+0x12/0x40
> [ 4060.646353] Â[<ffffffff810de5ad>] khugepaged+0x9d/0x1240
> [ 4060.646364] Â[<ffffffff81063700>] ? wake_up_bit+0x40/0x40
> [ 4060.646372] Â[<ffffffff810de510>] ? prepare_pmd_huge_pte.clone.29+0x60/0x60
> [ 4060.646382] Â[<ffffffff81063266>] kthread+0x96/0xa0
> [ 4060.646392] Â[<ffffffff81436f54>] kernel_thread_helper+0x4/0x10
> [ 4060.646403] Â[<ffffffff810631d0>] ? kthread_worker_fn+0x190/0x190
> [ 4060.646412] Â[<ffffffff81436f50>] ? gs_change+0xb/0xb
> [ 4060.646521]
> [ 4060.646525] Restarting tasks ... done.
> [ 4060.660646] video LNXVIDEO:00: Restoring backlight state
>
> $ echo never > /sys/kernel/mm/transparent_hugepage/enabled
> $ cat /sys/kernel/mm/transparent_hugepage/enabled
> always madvise [never]
>
> $ echo mem > /sys/power/state
> suspended/resumed fine
>
> $ uname -a
> Linux sf 3.2.0-rc1-00252-g8f042aa #159 SMP PREEMPT Sat Nov 12 09:28:53 FET 2011 x86_64 Intel(R) Core(TM)2 Duo CPU U7600 @ 1.20GHz GenuineIntel GNU/Linux
>
> Tried to reproduce 2 times. Both are successful
>

Hi ÂSergei

Would you please try the recent approach, at

            https://lkml.org/lkml/2011/11/9/312

Best regards
Hillf
8<--------------------------

From: Andrea Arcangeli <aarcange@xxxxxxxxxx>
Subject: thp: reduce khugepaged freezing latency

Use wait_event_freezable_timeout() instead of
schedule_timeout_interruptible() to avoid missing freezer wakeups. A
try_to_freeze() would have been needed in the
khugepaged_alloc_hugepage tight loop too in case of the allocation
failing repeatedly, and wait_event_freezable_timeout will provide it
too.

khugepaged would still freeze just fine by trying again the next
minute but it's better if it freezes immediately.

Reported-by: Jiri Slaby <jslaby@xxxxxxx>
Signed-off-by: Andrea Arcangeli <aarcange@xxxxxxxxxx>
--
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/