[ewg] Re: [ofa-general] Re: OFED 1.2 Mar-12 meeting summary on beta readiness

[Michael S. Tsirkin]

> Quoting Bryan O'Sullivan <bos at pathscale.com>:
> Subject: Re: [ofa-general] Re: OFED 1.2 Mar-12 meeting summary on beta readiness
> 
> Michael S. Tsirkin wrote:
> >>  Kernel code re-base to 2.6.21: Decided not to change the kernel base 
> >>  code
> >>  Reason was the risk that is can insert and delaying the release.
> >
> >I agree it makes sense for beta.
> 
> Are you saying that we *should* rebase to 2.6.21-rc3?

Yes.

> Not without some 
> notion of the concrete, valuable benefits it will bring, and how thosse 
> weigh against the pain, annoyance, and churn, we shouldn't.

I expect very minor pain.
I'll do most of the work. It's probably 1 day or so.

> Unless 
> there's a solidly compelling reason to rebase, I am quite against the idea.

The point of this is to be good citizens of upstream kernel.

We currently have 3 code-bases: 2.6.20, OFED and 2.6.21. This increases
support load (e.g. preparing patches for stable releases
requires doing the same work 3 times, and OFED and upstream testing
needs to be done 3 times).

I know this does not help ipath since apparently you didn't submit
your patches for upstream inclusion, but for core, chelsio, ehca and mthca this
makes support much simpler as we have a single code-stream
between upstreamand OFED.

I will post more detail later.

-- 
MST