[GIT] lock scaling and bug fixes for 2.6.38

[GIT] lock scaling and bug fixes for 2.6.38

[David Dillow]

Roland,

Please pull from

    git://git.kernel.org/pub/scm/linux/kernel/git/dad/srp-initiator.git srp-lock-scaling

to get the following commits. This is the series posted and reviewed on
the RDMA mailing list. I expect to have another pull request once you
sort out how you want to handle work queues in the IB stack so that we
can remove flush_scheduled_work() from the initiator.

Thanks!

Bart Van Assche (6):
  IB/srp: consolidate state change code
  IB/srp: allow lockless work posting
  IB/srp: don't move active requests to their own list
  IB/srp: reduce local coverage for command submission and EH
  IB/srp: reduce lock coverage of command completion
  IB/srp: stop sharing the host lock with SCSI

David Dillow (8):
  IB/srp: allow task management without a previous request
  IB/srp: consolidate hot-path variables into cache lines

 drivers/infiniband/ulp/srp/ib_srp.c |  392 ++++++++++++++++-------------------
 drivers/infiniband/ulp/srp/ib_srp.h |   46 +++--
 2 files changed, 206 insertions(+), 232 deletions(-)

-- 
Dave Dillow
National Center for Computational Science
Oak Ridge National Laboratory
(865) 241-6602 office

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [GIT] lock scaling and bug fixes for 2.6.38

[Roland Dreier]

 >    git://git.kernel.org/pub/scm/linux/kernel/git/dad/srp-initiator.git srp-lock-scaling

Thanks, pulled.

Will try to merge Tejun's work queue patch shortly.

 - R.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [GIT] lock scaling and bug fixes for 2.6.38

[Bart Van Assche]

On Tue, Jan 11, 2011 at 2:45 AM, Roland Dreier <rdreier-FYB4Gu1CFyUAvxtiuMwx3w@public.gmane.org> wrote:
>
>  >    git://git.kernel.org/pub/scm/linux/kernel/git/dad/srp-initiator.git srp-lock-scaling
>
> Thanks, pulled.
>
> Will try to merge Tejun's work queue patch shortly.

Maybe it's a good idea to add a comment above the declaration of the
global variable "extern struct workqueue_struct *ib_wq;" that explains
that that variable has been added during the work queue conversion and
that the preferred approach for new code is to define a new work queue
instead of using ib_wq ?

Bart.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [GIT] lock scaling and bug fixes for 2.6.38

[Tejun Heo]

Hello,

On Tue, Jan 11, 2011 at 06:29:52PM +0100, Bart Van Assche wrote:
> On Tue, Jan 11, 2011 at 2:45 AM, Roland Dreier <rdreier-FYB4Gu1CFyUAvxtiuMwx3w@public.gmane.org> wrote:
> >
> >  >    git://git.kernel.org/pub/scm/linux/kernel/git/dad/srp-initiator.git srp-lock-scaling
> >
> > Thanks, pulled.
> >
> > Will try to merge Tejun's work queue patch shortly.
> 
> Maybe it's a good idea to add a comment above the declaration of the
> global variable "extern struct workqueue_struct *ib_wq;" that explains
> that that variable has been added during the work queue conversion and
> that the preferred approach for new code is to define a new work queue
> instead of using ib_wq ?

But why is defining more workqueues preferred?  Unless there's a
specific need, using what's already there (system or ib_wq) should be
preferable.

Thank you.

-- 
tejun
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [GIT] lock scaling and bug fixes for 2.6.38

[Bart Van Assche]

On Tue, Jan 11, 2011 at 6:43 PM, Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org> wrote:
> Hello,
>
> On Tue, Jan 11, 2011 at 06:29:52PM +0100, Bart Van Assche wrote:
>> On Tue, Jan 11, 2011 at 2:45 AM, Roland Dreier <rdreier-FYB4Gu1CFyUAvxtiuMwx3w@public.gmane.org> wrote:
>> >
>> >  >    git://git.kernel.org/pub/scm/linux/kernel/git/dad/srp-initiator.git srp-lock-scaling
>> >
>> > Thanks, pulled.
>> >
>> > Will try to merge Tejun's work queue patch shortly.
>>
>> Maybe it's a good idea to add a comment above the declaration of the
>> global variable "extern struct workqueue_struct *ib_wq;" that explains
>> that that variable has been added during the work queue conversion and
>> that the preferred approach for new code is to define a new work queue
>> instead of using ib_wq ?
>
> But why is defining more workqueues preferred?  Unless there's a
> specific need, using what's already there (system or ib_wq) should be
> preferable.

Sorry, but I just noticed that I had missed the part of the discussion
that happened on the LKML and where linux-rdma was not CC-ed:
https://lkml.org/lkml/2010/12/16/244. I think that thread contains
valuable information.

Bart.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html