Re: [PATCH][v3.19.y-ckt] IB/ipoib: change init sequence ordering

[Doug Ledford <dledford@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 1714 bytes --]

On 07/01/2015 01:27 PM, Joseph Salisbury wrote:
> Hello,
> 
> Please consider including upstream commits
> be7aa663fc1d9156798f5af3c60e6df45e1fe5de and
> e135106fac9525352feb8e49077c8f46c3eaf288 in the next v3.19.y-ckt
> release.  They were included mainline as of v4.1-rc1.  These commits
> have been tested and confirmed to resolve
> http://bugs.launchpad.net/bugs/1467912 .
> 
> 
> commit be7aa663fc1d9156798f5af3c60e6df45e1fe5de
> Author: Doug Ledford <dledford@redhat.com>
> Date:   Sat Feb 21 19:27:00 2015 -0500
> 
>     IB/ipoib: change init sequence ordering
> 
> commit e135106fac9525352feb8e49077c8f46c3eaf288
> Author: Doug Ledford <dledford@redhat.com>
> Date:   Sat Feb 21 19:26:59 2015 -0500
> 
>     IB/ipoib: factor out ah flushing
> 
> The bug reporter states this bug is only happening with the 3.19 kernel,
> these commits may not be needed in other stable kernels.

I have no objection to these two patches going into a 3.19 stable
kernel.  They are both stand alone, safe patches.

However, there is no difference in the IPoIB code between 3.18 and 3.19,
and I didn't think this bug existed in 3.19 (my commit logs called the
work preparatory for upcoming patches).  If your testing does in fact
show that this is only a 3.19 bug, and 3.18 tests clear, then that means
the bug was introduced by some other change in the core kernel (probably
in how work queues are scheduled and we likely had a situation where our
code worked based upon how they used to be scheduled, but with the
change we now need explicit flushes).

In any case, ack from me for these two patches.


-- 
Doug Ledford <dledford@redhat.com>
              GPG KeyID: 0E572FDD



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 884 bytes --]