Re: [PATCH] blk: optimization for classic polling

2018-02-21 Thread Peter Zijlstra
On Tue, Feb 20, 2018 at 12:37:07PM -1000, Jens Axboe wrote: > On 2/20/18 3:21 AM, Peter Zijlstra wrote: > > On Sun, May 30, 2083 at 09:51:06AM +0530, Nitesh Shetty wrote: > >> This removes the dependency on interrupts to wake up task. Set task > >> state as TASK_RUNNING, if need_resched() returns

Re: [PATCH] blk: optimization for classic polling

2018-02-20 Thread Jens Axboe
On 2/20/18 3:21 AM, Peter Zijlstra wrote: > On Sun, May 30, 2083 at 09:51:06AM +0530, Nitesh Shetty wrote: >> This removes the dependency on interrupts to wake up task. Set task >> state as TASK_RUNNING, if need_resched() returns true, >> while polling for IO completion. >> Earlier, polling task

Re: [PATCH] blk: optimization for classic polling

2018-02-20 Thread Keith Busch
On Tue, Feb 20, 2018 at 02:21:37PM +0100, Peter Zijlstra wrote: > Also, set_current_state(TASK_RUNNING) is dodgy (similarly in > __blk_mq_poll), why do you need that memory barrier? You're right. The subsequent revision that was committed removed the barrier. The commit is here:

Re: [PATCH] blk: optimization for classic polling

2018-02-20 Thread Peter Zijlstra
On Sun, May 30, 2083 at 09:51:06AM +0530, Nitesh Shetty wrote: > This removes the dependency on interrupts to wake up task. Set task > state as TASK_RUNNING, if need_resched() returns true, > while polling for IO completion. > Earlier, polling task used to sleep, relying on interrupt to wake it

Re: [PATCH] blk: optimization for classic polling

2018-02-12 Thread Bart Van Assche
On 05/29/83 20:21, Nitesh Shetty wrote: [ ... ] Hello Nitesh, Can you check the clock of the system you used to send this e-mail? In the header of your e-mail I found the following: Date: Sun, 30 May 2083 09:51:06 +0530 Thanks, Bart.

Re: [PATCH] blk: optimization for classic polling

2018-02-08 Thread Sagi Grimberg
I think it'd be simpler to have blk_poll set it back to running if need_resched is true rather than repeat this patter across all the callers: --- diff --git a/block/blk-mq.c b/block/blk-mq.c index df93102e2149..40285fe1c8ad 100644 --- a/block/blk-mq.c +++ b/block/blk-mq.c @@ -3164,6 +3164,7

Re: [PATCH] blk: optimization for classic polling

2018-02-08 Thread Keith Busch
On Sun, May 30, 2083 at 09:51:06AM +0530, Nitesh Shetty wrote: > This removes the dependency on interrupts to wake up task. Set task > state as TASK_RUNNING, if need_resched() returns true, > while polling for IO completion. > Earlier, polling task used to sleep, relying on interrupt to wake it

[PATCH] blk: optimization for classic polling

2018-02-08 Thread Nitesh Shetty
This removes the dependency on interrupts to wake up task. Set task state as TASK_RUNNING, if need_resched() returns true, while polling for IO completion. Earlier, polling task used to sleep, relying on interrupt to wake it up. This made some IO take very long when interrupt-coalescing is enabled