Re: please revert 869184a675662bddcdf76c5b95665272facff2b8 ASAP

Re: please revert 869184a675662bddcdf76c5b95665272facff2b8 ASAP

[Dave Airlie]

On Mon, Nov 29, 2010 at 7:44 PM, Chris Wilson <chris@chris-wilson.co.uk> wrote:
> On Mon, 29 Nov 2010 16:18:44 +1000, Dave Airlie <airlied@gmail.com> wrote:
>> This commit breaks the eDP panel on the my HP Elitebook 2540p,
>> reverting it brings the panel back to life.
>
> Happy to test the revert just as soon as I work out what the equivalent
> commit is in my tree... ;-)
> -Chris

oops

869184a675662bddcdf76c5b95665272facff2b8

must have sent the revert commit id locally or something ;-)

drm/i915/dp: use VBT provided eDP params if available

Looks like trusting the BIOS is a bad plan, also that commit seems to
do a lot more than just use some VBT values, probably needs more
splitting up from jbarnes.

Dave.

Re: please revert 869184a675662bddcdf76c5b95665272facff2b8 ASAP

[Chris Wilson]

On Mon, 29 Nov 2010 20:06:23 +1000, Dave Airlie <airlied@gmail.com> wrote:
> drm/i915/dp: use VBT provided eDP params if available
> 
> Looks like trusting the BIOS is a bad plan, also that commit seems to
> do a lot more than just use some VBT values, probably needs more
> splitting up from jbarnes.

I've had one lucky Vaio owner report that this revert breaks his machine.
Looks like a partial revert is in order. eDP, what fun.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre

Re: please revert 869184a675662bddcdf76c5b95665272facff2b8 ASAP

[Dave Airlie]

On Tue, Nov 30, 2010 at 5:14 AM, Chris Wilson <chris@chris-wilson.co.uk> wrote:
> On Mon, 29 Nov 2010 20:06:23 +1000, Dave Airlie <airlied@gmail.com> wrote:
>> drm/i915/dp: use VBT provided eDP params if available
>>
>> Looks like trusting the BIOS is a bad plan, also that commit seems to
>> do a lot more than just use some VBT values, probably needs more
>> splitting up from jbarnes.
>
> I've had one lucky Vaio owner report that this revert breaks his machine.
> Looks like a partial revert is in order. eDP, what fun.
> -Chris

his vaio was broken in 2.6.36, he is no worse off in 2.6.37, my HP
worked in 2.6.36, I'm worse off.

regression rules are revert first, ask questions later.

I can't even send my current fixes tree to Linus because it won't boot
on my hardware, granted its new
hw but still its hard to send a known broken tree.

Dave.

Re: please revert 869184a675662bddcdf76c5b95665272facff2b8 ASAP

[Chris Wilson]

On Tue, 30 Nov 2010 06:19:09 +1000, Dave Airlie <airlied@gmail.com> wrote:
> his vaio was broken in 2.6.36, he is no worse off in 2.6.37, my HP
> worked in 2.6.36, I'm worse off.
> 
> regression rules are revert first, ask questions later.
> 
> I can't even send my current fixes tree to Linus because it won't boot
> on my hardware, granted its new
> hw but still its hard to send a known broken tree.

Done. Hopefully your test machine boots again and we can continue to make
Linus unhappy.

Jesse, can you look at which bits of that patch are required for bringing
up your Vaio Z - hopefully without now breaking QA's or Dave's machine ;-)
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre

Re: please revert 869184a675662bddcdf76c5b95665272facff2b8 ASAP

[Joep Daandels]

Chris Wilson <chris <at> chris-wilson.co.uk> writes:

> 
> On Mon, 29 Nov 2010 20:06:23 +1000, Dave Airlie <airlied <at> gmail.com> wrote:
> > drm/i915/dp: use VBT provided eDP params if available
> > 
> > Looks like trusting the BIOS is a bad plan, also that commit seems to
> > do a lot more than just use some VBT values, probably needs more
> > splitting up from jbarnes.
> 
> I've had one lucky Vaio owner report that this revert breaks his machine.
> Looks like a partial revert is in order. eDP, what fun.
> -Chris
> 


I'm afraid he's not the only vaio owner with problems. I was quite happy with
2.6.37 rc 2 to 4 because finally the display port was detected again after
skipping 2.6.36 because of the issue. 

Only just found this thread with an explaination why I was unable to work 2.6.36.

Is there any chance eDP will work again in the final release without breaking
other laptops:

Re: please revert 869184a675662bddcdf76c5b95665272facff2b8 ASAP

[Jesse Barnes]

On Thu, 9 Dec 2010 20:05:44 +0000 (UTC)
Joep Daandels <jdaandels@home.nl> wrote:

> Chris Wilson <chris <at> chris-wilson.co.uk> writes:
> 
> > 
> > On Mon, 29 Nov 2010 20:06:23 +1000, Dave Airlie <airlied <at> gmail.com> wrote:
> > > drm/i915/dp: use VBT provided eDP params if available
> > > 
> > > Looks like trusting the BIOS is a bad plan, also that commit seems to
> > > do a lot more than just use some VBT values, probably needs more
> > > splitting up from jbarnes.
> > 
> > I've had one lucky Vaio owner report that this revert breaks his machine.
> > Looks like a partial revert is in order. eDP, what fun.
> > -Chris
> > 
> 
> 
> I'm afraid he's not the only vaio owner with problems. I was quite happy with
> 2.6.37 rc 2 to 4 because finally the display port was detected again after
> skipping 2.6.36 because of the issue. 
> 
> Only just found this thread with an explaination why I was unable to work 2.6.36.
> 
> Is there any chance eDP will work again in the final release without breaking
> other laptops:

Possible but unlikely.  I'm still waiting to receive a couple more
laptops for testing so I can fix all of them at once rather than fixing
one and breaking another...

-- 
Jesse Barnes, Intel Open Source Technology Center

Re: please revert 869184a675662bddcdf76c5b95665272facff2b8 ASAP

[Jesse Barnes]

On Mon, 29 Nov 2010 20:06:23 +1000
Dave Airlie <airlied@gmail.com> wrote:

> On Mon, Nov 29, 2010 at 7:44 PM, Chris Wilson <chris@chris-wilson.co.uk> wrote:
> > On Mon, 29 Nov 2010 16:18:44 +1000, Dave Airlie <airlied@gmail.com> wrote:
> >> This commit breaks the eDP panel on the my HP Elitebook 2540p,
> >> reverting it brings the panel back to life.
> >
> > Happy to test the revert just as soon as I work out what the equivalent
> > commit is in my tree... ;-)
> > -Chris
> 
> oops
> 
> 869184a675662bddcdf76c5b95665272facff2b8
> 
> must have sent the revert commit id locally or something ;-)
> 
> drm/i915/dp: use VBT provided eDP params if available
> 
> Looks like trusting the BIOS is a bad plan, also that commit seems to
> do a lot more than just use some VBT values, probably needs more
> splitting up from jbarnes.

!#$!@!!

One day we'll have reliable eDP mode setting, I promise.  Time to
harass some more people internally about how to get sane values and
whether skipping training is generally safe (assuming we use the right
values that is!).

-- 
Jesse Barnes, Intel Open Source Technology Center

Re: please revert 869184a675662bddcdf76c5b95665272facff2b8 ASAP

[Dave Airlie]

On Tue, Dec 7, 2010 at 4:04 AM, Jesse Barnes <jbarnes@virtuousgeek.org> wrote:
> On Mon, 29 Nov 2010 20:06:23 +1000
> Dave Airlie <airlied@gmail.com> wrote:
>
>> On Mon, Nov 29, 2010 at 7:44 PM, Chris Wilson <chris@chris-wilson.co.uk> wrote:
>> > On Mon, 29 Nov 2010 16:18:44 +1000, Dave Airlie <airlied@gmail.com> wrote:
>> >> This commit breaks the eDP panel on the my HP Elitebook 2540p,
>> >> reverting it brings the panel back to life.
>> >
>> > Happy to test the revert just as soon as I work out what the equivalent
>> > commit is in my tree... ;-)
>> > -Chris
>>
>> oops
>>
>> 869184a675662bddcdf76c5b95665272facff2b8
>>
>> must have sent the revert commit id locally or something ;-)
>>
>> drm/i915/dp: use VBT provided eDP params if available
>>
>> Looks like trusting the BIOS is a bad plan, also that commit seems to
>> do a lot more than just use some VBT values, probably needs more
>> splitting up from jbarnes.
>
> !#$!@!!
>
> One day we'll have reliable eDP mode setting, I promise.  Time to
> harass some more people internally about how to get sane values and
> whether skipping training is generally safe (assuming we use the right
> values that is!).

I suspect you should split that patch also, it did two things, and I
wasn't sure which one way the one I wanted.

Dave.

Re: please revert 869184a675662bddcdf76c5b95665272facff2b8 ASAP

[Jesse Barnes]

On Tue, 7 Dec 2010 08:32:49 +1000
Dave Airlie <airlied@gmail.com> wrote:

> On Tue, Dec 7, 2010 at 4:04 AM, Jesse Barnes <jbarnes@virtuousgeek.org> wrote:
> > On Mon, 29 Nov 2010 20:06:23 +1000
> > Dave Airlie <airlied@gmail.com> wrote:
> >
> >> On Mon, Nov 29, 2010 at 7:44 PM, Chris Wilson <chris@chris-wilson.co.uk> wrote:
> >> > On Mon, 29 Nov 2010 16:18:44 +1000, Dave Airlie <airlied@gmail.com> wrote:
> >> >> This commit breaks the eDP panel on the my HP Elitebook 2540p,
> >> >> reverting it brings the panel back to life.
> >> >
> >> > Happy to test the revert just as soon as I work out what the equivalent
> >> > commit is in my tree... ;-)
> >> > -Chris
> >>
> >> oops
> >>
> >> 869184a675662bddcdf76c5b95665272facff2b8
> >>
> >> must have sent the revert commit id locally or something ;-)
> >>
> >> drm/i915/dp: use VBT provided eDP params if available
> >>
> >> Looks like trusting the BIOS is a bad plan, also that commit seems to
> >> do a lot more than just use some VBT values, probably needs more
> >> splitting up from jbarnes.
> >
> > !#$!@!!
> >
> > One day we'll have reliable eDP mode setting, I promise.  Time to
> > harass some more people internally about how to get sane values and
> > whether skipping training is generally safe (assuming we use the right
> > values that is!).
> 
> I suspect you should split that patch also, it did two things, and I
> wasn't sure which one way the one I wanted.

You mean using the VBT values and avoiding the training?  I suppose
those could be split, but for eDP they should amount to the same thing
(but otoh "should" is a big word when talking about eDP).

-- 
Jesse Barnes, Intel Open Source Technology Center
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx