Thread (41 messages) 41 messages, 5 authors, 2012-01-06

Re: [PATCH block:for-3.3/core] cfq: merged request shouldn't jump to a different cfqq

From: Shaohua Li <hidden>
Date: 2012-01-06 03:20:00
Also in: linux-next, linux-scsi, lkml

On Thu, 2012-01-05 at 19:04 -0800, Tejun Heo wrote:
Hello,

On Fri, Jan 06, 2012 at 11:14:15AM +0800, Shaohua Li wrote:
quoted
quoted
So, yeah, the right fix would be using elv_former/latter_request()
instead.  Maybe we should strip out rqhash altogether and change
elevator handle everything?  I don't know.  I'll prepare a different
fix patch soon.
So not allow merge from two cfq queues strictly? This will impact
performance. I don't know how important the strict isolation is. we even
allow two cfq queues merge to improve performance.
That's how cfq has behaved before this recent plug merge breakage and
IIRC why the cooperating queue thing is there.  If you want to change
the behavior, that should be an explicit separate patch.
My point is both cooperating merge and the plug merge of different cfq
are merge, no reason we allow one but disallow the other. plug merge
isn't a breakage to me.

Thanks,
Shaohua
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help