[ANNOUNCE] BFQ-v2 for 2.6.38 (and BFQ-v2-r1 for 2.6.37 and 2.6.37.3)
Paolo Valente
at last I have found the time to port BFQ to 2.6.38. I took the
opportunity to also introduce further improvements.
I introduced these improvements also into BFQ for 2.6.37 and 2.6.37.3
(for these kernels, I named the new version BFQ-v2-r1).
BFQ-v2-r1 improves BFQ-v2 as follows (see the patch to switch from v2 to
v2-r1 for details).
* An issue that may cause little throughput loss on fast disks has been
solved. BFQ-v1 and CFQ may suffer from this problem.
* The disk-idling timeout has been better tuned to further file latency
(especially for the idle- or light-loaded-disk scenarios).
* One of the parameters of the low-latency heuristics has been tuned a
little bit more, so as to reduce the probability that a disk-bound
process may hamper the reduction of the latency of interactive and soft
real-time applications.
---------------------------------------------------------------
You can find the new patchsets here:
BFQ-v2-r1 for 2.6.37:
http://algo.ing.unimo.it/people/paolo/disk_sched/patches/2.6.37/
BFQ-v2-r1 for 2.6.37.3:
http://algo.ing.unimo.it/people/paolo/disk_sched/patches/2.6.37.3-v2-r1/
BFQ-v2 for 2.6.38:
http://algo.ing.unimo.it/people/paolo/disk_sched/patches/2.6.38/
Switch from BFQ-v2 to BFQ-v2-r1:
http://algo.ing.unimo.it/people/paolo/disk_sched/patches/previous_versions_revisions/incremental_patches/from_v2-to_v2-r1/
-----------------------------------------------------------------
I have done everything in a rush, I hope I did not make big mistakes,
Paolo
PS.
If someone is interested, I have created a google group bfq-iosched, for
discussion and announcements about BFQ. You can find it here:
http://groups.google.com/group/bfq-iosched
Steven Barrett
I'm really stoked to give your new patches a spin. Although I won't
be building a new kernel soon, when I do, I'll let you know if I
notice anything significant.
By the way, the new patches are already pushed to Zen if anyone
prefers to pull patches in that manner.
On Mar 25, 11:36 am, Paolo Valente <posta_pa...@yahoo.it> wrote:
> Hi,
> at last I have found the time to port BFQ to 2.6.38. I took the
> opportunity to also introduce further improvements.
> I introduced these improvements also into BFQ for 2.6.37 and 2.6.37.3
> (for these kernels, I named the new version BFQ-v2-r1).
> BFQ-v2-r1 improves BFQ-v2 as follows (see the patch to switch from v2 to
> v2-r1 for details).
>
> * An issue that may cause little throughput loss on fast disks has been
> solved. BFQ-v1 and CFQ may suffer from this problem.
> * The disk-idling timeout has been better tuned to further file latency
> (especially for the idle- or light-loaded-disk scenarios).
> * One of the parameters of the low-latency heuristics has been tuned a
> little bit more, so as to reduce the probability that a disk-bound
> process may hamper the reduction of the latency of interactive and soft
> real-time applications.
>
> ---------------------------------------------------------------
> You can find the new patchsets here:
> BFQ-v2-r1 for 2.6.37:http://algo.ing.unimo.it/people/paolo/disk_sched/patches/2.6.37/
>
> BFQ-v2-r1 for 2.6.37.3:http://algo.ing.unimo.it/people/paolo/disk_sched/patches/2.6.37.3-v2-r1/
>
> BFQ-v2 for 2.6.38:http://algo.ing.unimo.it/people/paolo/disk_sched/patches/2.6.38/
>
