[Devel] Re: seems to be a flaw in cfq

[Vasily Tarasov]

Hello,

On 2.6.19 "buffered randread" bizarre behavior persists. So it was fixed
somewhere in 2.6.20... Can you give me a guess, what changes in 2.6.20
cause this fix? The point is that we want to port it back to 2.6.18
(rhel-5), if possible.

Thanks,
Vasily.

Jens Axboe wrote:
> On Thu, Dec 21 2006, Jens Axboe wrote:
>   
>> On Thu, Dec 21 2006, Vasily Tarasov wrote:
>>     
>>> Hello, Jens,
>>>
>>> Sorry for late answer.
>>> The situation is the following:
>>>
>>> direct seqread
>>> (2.6.18) - reproducable
>>> (2.6.20-rc1) - reproducable
>>> (2.6.20-rc1 + your fix) - not reproducable
>>>
>>> buffered randread
>>> (2.6.18) - reproducable
>>> (2.6.20-rc1) - not reproducable
>>> (2.6.20-rc1 + your fix) - not reproducable
>>>
>>> So the conclusion: problem with "direct seqread" is fixed by your patch
>>> (thanks!) and
>>> the problem with "buffered randread" was fixed in 2.6.19/2.6.20.
>>>       
>> Perfect, explains why I didn't see much badness with buffered io on
>> 2.6.20-rc1'ish. Thanks for retesting and the initial report, the fix I
>> sent you is going upstream (sitting in the 'for-linus' branch awaiting a
>> pull) for 2.6.20-rc2.
>>     
>
> Oh, Linus just pulled a few minutes ago, the fix is already upstream
> now.