Re: [ck] Re: SD still better than CFS for 3d ?

[Martin Schwidefsky <schwidefsky@de.ibm.com>]

On Tue, 2007-07-31 at 13:07 +1000, Matthew Hawkins wrote:
> On 7/31/07, Roland Dreier <rdreier@cisco.com> wrote:
> >  >      Fuck you Martin!
> >
> > I think you meant to yell at Matthew, not Martin ;)
> 
> What's amusing about this is he's yelling at me for something I didn't
> do, can't even get my name right, and has the audacity to claim that
> *I* am the one looking like a fool!  While we're descending into
> primary school theatrics, may I just say "takes one to know one" ;-)

Pouring oil into the fire ?

> I took the time to track down what caused a breakage - in an "illegal
> binary driver" (not against the law here, though defamation certainly
> is...) no less.  And contacted the vendor (separately).  Other people
> on desktop machines with an ATI card using the fglrx driver may have
> been interested to know that they can't do the benchmarking some
> people here on lkml and -mm are asking for with a current 2.6.23 git
> kernel, hence my post.

To inform the vendor and to post a warning about the issue on lkml was
the right thing to do. It is the wording of your post that obviously
irked some people.

> Martin's cleanup patch is good and I never claimed otherwise, I just
> said the comment on the commit was a bad call (as there are users of
> that interface).  Certainly ATI should fix their dodgy drivers.
> That's been the cry of the community for a long time...

The commit message could have been better. The correct thing to say
would have been "Nobody in the official kernel is using
ptep_test_and_clear_dirty and ptep_clear_flush_dirty."
nvidia will have to adapt their binary driver. This is not the first
time it breaks and it won't be the last time. We do not really have a
problem and we should all calm down and put that issue to rest.

-- 
blue skies,
  Martin.

"Reality continues to ruin my life." - Calvin.