Subject: | |
From: | |
Reply To: | |
Date: | Wed, 1 Sep 2010 09:26:29 +0100 |
Content-Type: | TEXT/PLAIN |
Parts/Attachments: |
|
|
Good day,
Someone in our local LUG recently posed about this, he was seeing
> Jul 21 07:39:04 linux kernel: INFO: task mount:5258 blocked for more than 120 seconds.
> Jul 21 07:39:04 linux kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> Jul 21 07:39:04 linux kernel: mount D 0000000000000002 0 / 5258 4842 0x00000080
> Jul 21 07:39:04 linux kernel: ffff880037b4fb28 0000000000000082 / ffff880037b4fa78 ffffffff812d6ea0
> Jul 21 07:39:04 linux kernel: ffff880037b4fad8 ffffffff812ef2f9 / ffff880037b4ffd8 ffff880037b4ffd8
He found a solution:
> Googling around suggests that it might be a recurrent kernel bug (as
> it appears to come and go between versions) and that it might be related to
> the method the kernel uses to idle the processor. The most power-efficient
> technique is to use the MWAIT instruction, which is the default on systems
> that support it. The least power-efficient (and most performant) is use a
> polling idle loop. In between is to use the HALT instruction.
>
> Since starting to boot my kernel with
>
> idle=halt
>
> (and pcie_aspm=off, FWIW, but I don't think my system was using it anyway),
> I haven't had any recurrance.
Does that help?
|
|
|