Re: bcache: Fix a writeback performance regression

[Stefan Priebe <s.priebe-2Lf/h1ldwEHR5kwTpVNS9A@public.gmane.org>]

Am 22.08.2013 01:47, schrieb Kent Overstreet:
> On Tue, Aug 20, 2013 at 10:07:45AM +0200, Stefan Priebe - Profihost AG wrote:
>> Am 20.08.2013 10:01, schrieb Stefan Priebe - Profihost AG:
>>> Am 20.08.2013 00:27, schrieb Kent Overstreet:
>>>> On Mon, Aug 19, 2013 at 12:09:24AM +0200, Stefan Priebe wrote:
>>>>>
>>>>> Vanilla 3.10.7 + bcache: Fix a writeback performance regression
>>>>>
>>>>> http://pastebin.com/raw.php?i=LXZk4cMH
>>>>
>>>> Whoops, at first I thought this was the same bug as one I'd already been
>>>> chasing down that had been a harmless bug - turns out I didn't look
>>>> closely enough at the backtrace.
>>>>
>>>> What happened is background writeback is deadlocking, because for some
>>>> reason the workqueue it's running out of is a singlethreaded workqueue,
>>>> so as soon as it decides to queue enough writeback bios that it has to
>>>> sleep on that semaphore (which often won't happen due to the PD
>>>> controller based ratelimiting) - boom, deadlock.
>>>>
>>>> Here's the fixup patch I just tested and am applying:
>>
>> Oh i'm now seeing very high CPU spikes of kworker... i don't see if i
>> remove bcache: Fix a writeback performance regression.
>
> Are you able to reproduce it? I'm not having any luck reproducing it...

Hi,

yes but only on production system it seems it happens when there is a 
big bunch in dirty and having heavy 4K random I/O.

Stefan