On Thu, 2008-09-11 at 11:06 -0400, Jeremy Katz wrote:
... we really shouldn't shove in major changes like this for a
significant chunk of hardware post-beta. Putting it in the beta means
that we get a relatively large chunk of testing done due to the
visibility surrounding the beta announce. As well as having time to fix
large problems that are uncovered. Later milestones both a) lack the
time for changes after them and b) the visibility.
Yeah, I'm not so keen on doing all of this post-beta. We've got enough
stability problems with F10 as it is. This honestly feels like one of
those features that just isn't 'testable' by Beta, and thus needs to
enact the contingency plan.
How do things look for that contingency plan. Is it even possible at
this point?
If we're not going to enable it by default, I would rather the modesetting code
be included, but disabled by default, and let users enable it on the kernel
command line.
-- Chris
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
09-11-2008, 05:08 PM
Adam Jackson
modesetting feature status
On Thu, 2008-09-11 at 10:18 -0600, Stephen Warren wrote:
> On Thu, September 11, 2008 1:18 am, Dave Airlie wrote:
> > Hi,
> >
> > So one of the features we've been trying to land for F10 that is a major
> > change in how we deal with graphics hardware in the Linux world.
> >
> > https://fedoraproject.org/wiki/Features/KernelModesetting
>
> What will happen on unsupported hardware (NVIDIA, unsupported ATI/Intel
> chipsets)? Will the system simply fall back to the existing F9-style
> startup sequence.
They'll get text boot, yes.
> I assume the X.org drivers won't be coded to assume that
> kernel modesetting has already run? How will X.org drivers detect this?
They have been changed to notice if kms has already happened.
> How will this interact with proprietary binary drivers from NVIDIA and
> ATI, which don't know anything about this new feature?
Well, it's irrelevant for nvidia since we don't even pretend to do kms
there. For fglrx, you'll probably have to force it to text boot, since
otherwise the radeon driver will load, locking out fglrx.
- ajax
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
09-11-2008, 05:09 PM
Adam Jackson
modesetting feature status
On Thu, 2008-09-11 at 11:06 -0400, Jeremy Katz wrote:
> On Thu, 2008-09-11 at 17:18 +1000, Dave Airlie wrote:
> > For the beta we are only going to enable Radeon support for the r300 to
> > r500 class of hardware, while we await upstream changes to the Intel
> > driver.
>
> ... we really shouldn't shove in major changes like this for a
> significant chunk of hardware post-beta. Putting it in the beta means
> that we get a relatively large chunk of testing done due to the
> visibility surrounding the beta announce. As well as having time to fix
> large problems that are uncovered. Later milestones both a) lack the
> time for changes after them and b) the visibility.
So are you saying:
a) enable it for both, see what breaks
b) remove both, because, screw it
?
- ajax
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
09-11-2008, 05:46 PM
Jeremy Katz
modesetting feature status
On Thu, 2008-09-11 at 13:09 -0400, Adam Jackson wrote:
> On Thu, 2008-09-11 at 11:06 -0400, Jeremy Katz wrote:
> > On Thu, 2008-09-11 at 17:18 +1000, Dave Airlie wrote:
> > > For the beta we are only going to enable Radeon support for the r300 to
> > > r500 class of hardware, while we await upstream changes to the Intel
> > > driver.
> >
> > ... we really shouldn't shove in major changes like this for a
> > significant chunk of hardware post-beta. Putting it in the beta means
> > that we get a relatively large chunk of testing done due to the
> > visibility surrounding the beta announce. As well as having time to fix
> > large problems that are uncovered. Later milestones both a) lack the
> > time for changes after them and b) the visibility.
>
> So are you saying:
>
> a) enable it for both, see what breaks
I've been advocating us getting the pieces merged for a long time so
that we could see what was broken rather than trying to wait for a
mythical "perfection" that's been just another week or two out for on
the order of months now.
> b) remove both, because, screw it
This is a part of the contingency plan. But removing for both seems a
little extreme.
There's also c) Just enable it for radeon. Intel users get pretty
modesetting for F11. It's only six months away. Adding the Intel
support post-beta just leaves us with way too little runway of testing
on what's a significant amount of the hardware running Fedora. And one
of the big reasons behind the six month release cycle is that if
something is late to the party for release n, n+1 is not that far away.
Jeremy
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
09-11-2008, 07:09 PM
"Joshua C."
modesetting feature status
2008/9/11 Jeremy Katz <katzj@redhat.com>:
> On Thu, 2008-09-11 at 13:09 -0400, Adam Jackson wrote:
>> On Thu, 2008-09-11 at 11:06 -0400, Jeremy Katz wrote:
>> > On Thu, 2008-09-11 at 17:18 +1000, Dave Airlie wrote:
>> > > For the beta we are only going to enable Radeon support for the r300 to
>> > > r500 class of hardware, while we await upstream changes to the Intel
>> > > driver.
>> >
>> > ... we really shouldn't shove in major changes like this for a
>> > significant chunk of hardware post-beta. Putting it in the beta means
>> > that we get a relatively large chunk of testing done due to the
>> > visibility surrounding the beta announce. As well as having time to fix
>> > large problems that are uncovered. Later milestones both a) lack the
>> > time for changes after them and b) the visibility.
>>
>> So are you saying:
>>
>> a) enable it for both, see what breaks
>
> I've been advocating us getting the pieces merged for a long time so
> that we could see what was broken rather than trying to wait for a
> mythical "perfection" that's been just another week or two out for on
> the order of months now.
>
>> b) remove both, because, screw it
>
> This is a part of the contingency plan. But removing for both seems a
> little extreme.
>
> There's also c) Just enable it for radeon. Intel users get pretty
> modesetting for F11. It's only six months away. Adding the Intel
> support post-beta just leaves us with way too little runway of testing
> on what's a significant amount of the hardware running Fedora. And one
> of the big reasons behind the six month release cycle is that if
> something is late to the party for release n, n+1 is not that far away.
>
> Jeremy
>
> --
> fedora-devel-list mailing list
> fedora-devel-list@redhat.com
> https://www.redhat.com/mailman/listinfo/fedora-devel-list
>
Is there a list of cards that work with this (already tested)? f10 is
just a month away and my rs485 still cannot function with this code.
It's an important feature and if it cannot work, then maybe the
contingency plan should be applied.
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
09-11-2008, 07:55 PM
Josh Boyer
modesetting feature status
On Thu, Sep 11, 2008 at 01:46:04PM -0400, Jeremy Katz wrote:
>On Thu, 2008-09-11 at 13:09 -0400, Adam Jackson wrote:
>> On Thu, 2008-09-11 at 11:06 -0400, Jeremy Katz wrote:
>> > On Thu, 2008-09-11 at 17:18 +1000, Dave Airlie wrote:
>> > > For the beta we are only going to enable Radeon support for the r300 to
>> > > r500 class of hardware, while we await upstream changes to the Intel
>> > > driver.
>> >
>> > ... we really shouldn't shove in major changes like this for a
>> > significant chunk of hardware post-beta. Putting it in the beta means
>> > that we get a relatively large chunk of testing done due to the
>> > visibility surrounding the beta announce. As well as having time to fix
>> > large problems that are uncovered. Later milestones both a) lack the
>> > time for changes after them and b) the visibility.
>>
>> So are you saying:
>>
>> a) enable it for both, see what breaks
>
>I've been advocating us getting the pieces merged for a long time so
>that we could see what was broken rather than trying to wait for a
>mythical "perfection" that's been just another week or two out for on
>the order of months now.
>
>> b) remove both, because, screw it
>
>This is a part of the contingency plan. But removing for both seems a
>little extreme.
>
>There's also c) Just enable it for radeon. Intel users get pretty
It's not working very well on radeon either. My laptop never gets a
display (i686), and it's hosed on ppc in general.
josh
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
09-11-2008, 11:04 PM
"David Nielsen"
modesetting feature status
2008/9/11 Josh Boyer <jwboyer@gmail.com>
On Thu, Sep 11, 2008 at 01:46:04PM -0400, Jeremy Katz wrote:
>On Thu, 2008-09-11 at 13:09 -0400, Adam Jackson wrote:
>> On Thu, 2008-09-11 at 11:06 -0400, Jeremy Katz wrote:
>> > On Thu, 2008-09-11 at 17:18 +1000, Dave Airlie wrote:
>> > > For the beta we are only going to enable Radeon support for the r300 to
>> > > r500 class of hardware, while we await upstream changes to the Intel
>> > > driver.
>> >
>> > ... we really shouldn't shove in major changes like this for a
>> > significant chunk of hardware post-beta. *Putting it in the beta means
>> > that we get a relatively large chunk of testing done due to the
>> > visibility surrounding the beta announce. *As well as having time to fix
>> > large problems that are uncovered. *Later milestones both a) lack the
>> > time for changes after them and b) the visibility.
>>
>> So are you saying:
>>
>> a) enable it for both, see what breaks
>
>I've been advocating us getting the pieces merged for a long time so
>that we could see what was broken rather than trying to wait for a
>mythical "perfection" that's been just another week or two out for on
>the order of months now.
>
>> b) remove both, because, screw it
>
>This is a part of the contingency plan. *But removing for both seems a
>little extreme.
>
>There's also c) Just enable it for radeon. *Intel users get pretty
It's not working very well on radeon either. *My laptop never gets a
display (i686), and it's hosed on ppc in general.
Neither my trusty old r200 nor my shiny new r600 experience any issues booting in the default configuration, however there is no sign of acceleration so the X experience is so slow as to be practically useless. I trust the regression will go away by stable, I will help any way I am able to make it so (which frankly does not go beyond testing new code and reporting back).
- David
- David
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
09-11-2008, 11:10 PM
"Paul W. Frields"
modesetting feature status
On Thu, 2008-09-11 at 21:09 +0200, Joshua C. wrote:
> 2008/9/11 Jeremy Katz <katzj@redhat.com>:
> > On Thu, 2008-09-11 at 13:09 -0400, Adam Jackson wrote:
> >> On Thu, 2008-09-11 at 11:06 -0400, Jeremy Katz wrote:
> >> > On Thu, 2008-09-11 at 17:18 +1000, Dave Airlie wrote:
> >> > > For the beta we are only going to enable Radeon support for the r300 to
> >> > > r500 class of hardware, while we await upstream changes to the Intel
> >> > > driver.
> >> >
> >> > ... we really shouldn't shove in major changes like this for a
> >> > significant chunk of hardware post-beta. Putting it in the beta means
> >> > that we get a relatively large chunk of testing done due to the
> >> > visibility surrounding the beta announce. As well as having time to fix
> >> > large problems that are uncovered. Later milestones both a) lack the
> >> > time for changes after them and b) the visibility.
> >>
> >> So are you saying:
> >>
> >> a) enable it for both, see what breaks
> >
> > I've been advocating us getting the pieces merged for a long time so
> > that we could see what was broken rather than trying to wait for a
> > mythical "perfection" that's been just another week or two out for on
> > the order of months now.
> >
> >> b) remove both, because, screw it
> >
> > This is a part of the contingency plan. But removing for both seems a
> > little extreme.
> >
> > There's also c) Just enable it for radeon. Intel users get pretty
> > modesetting for F11. It's only six months away. Adding the Intel
> > support post-beta just leaves us with way too little runway of testing
> > on what's a significant amount of the hardware running Fedora. And one
> > of the big reasons behind the six month release cycle is that if
> > something is late to the party for release n, n+1 is not that far away.
> >
> > Jeremy
> >
> > --
> > fedora-devel-list mailing list
> > fedora-devel-list@redhat.com
> > https://www.redhat.com/mailman/listinfo/fedora-devel-list
> >
>
> Is there a list of cards that work with this (already tested)? f10 is
> just a month away and my rs485 still cannot function with this code.
> It's an important feature and if it cannot work, then maybe the
> contingency plan should be applied.
By my calendar, F10 is a little over two months away...
On Fri, 2008-09-12 at 01:04 +0200, David Nielsen wrote:
>
>
> 2008/9/11 Josh Boyer <jwboyer@gmail.com>
> On Thu, Sep 11, 2008 at 01:46:04PM -0400, Jeremy Katz wrote:
> >On Thu, 2008-09-11 at 13:09 -0400, Adam Jackson wrote:
> >> On Thu, 2008-09-11 at 11:06 -0400, Jeremy Katz wrote:
> >> > On Thu, 2008-09-11 at 17:18 +1000, Dave Airlie wrote:
> >> > > For the beta we are only going to enable Radeon support
> for the r300 to
> >> > > r500 class of hardware, while we await upstream changes
> to the Intel
> >> > > driver.
> >> >
> >> > ... we really shouldn't shove in major changes like this
> for a
> >> > significant chunk of hardware post-beta. Putting it in
> the beta means
> >> > that we get a relatively large chunk of testing done due
> to the
> >> > visibility surrounding the beta announce. As well as
> having time to fix
> >> > large problems that are uncovered. Later milestones both
> a) lack the
> >> > time for changes after them and b) the visibility.
> >>
> >> So are you saying:
> >>
> >> a) enable it for both, see what breaks
> >
> >I've been advocating us getting the pieces merged for a long
> time so
> >that we could see what was broken rather than trying to wait
> for a
> >mythical "perfection" that's been just another week or two
> out for on
> >the order of months now.
> >
> >> b) remove both, because, screw it
> >
> >This is a part of the contingency plan. But removing for
> both seems a
> >little extreme.
> >
> >There's also c) Just enable it for radeon. Intel users get
> pretty
>
>
> It's not working very well on radeon either. My laptop never
> gets a
> display (i686), and it's hosed on ppc in general.
>
> Neither my trusty old r200 nor my shiny new r600 experience any issues
> booting in the default configuration, however there is no sign of
> acceleration so the X experience is so slow as to be practically
> useless. I trust the regression will go away by stable, I will help
> any way I am able to make it so (which frankly does not go beyond
> testing new code and reporting back).
>
Neither of these should have changed much if at all so they should just
be as they were in F9. Actually maybe r200 is now using EXA by default..
but r600 has no acceleration support to speak off yet.
Dave.
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
09-11-2008, 11:16 PM
Dave Airlie
modesetting feature status
On Thu, 2008-09-11 at 21:09 +0200, Joshua C. wrote:
> 2008/9/11 Jeremy Katz <katzj@redhat.com>:
> > On Thu, 2008-09-11 at 13:09 -0400, Adam Jackson wrote:
> >> On Thu, 2008-09-11 at 11:06 -0400, Jeremy Katz wrote:
> >> > On Thu, 2008-09-11 at 17:18 +1000, Dave Airlie wrote:
> >> > > For the beta we are only going to enable Radeon support for the r300 to
> >> > > r500 class of hardware, while we await upstream changes to the Intel
> >> > > driver.
> >> >
> >> > ... we really shouldn't shove in major changes like this for a
> >> > significant chunk of hardware post-beta. Putting it in the beta means
> >> > that we get a relatively large chunk of testing done due to the
> >> > visibility surrounding the beta announce. As well as having time to fix
> >> > large problems that are uncovered. Later milestones both a) lack the
> >> > time for changes after them and b) the visibility.
> >>
> >> So are you saying:
> >>
> >> a) enable it for both, see what breaks
> >
> > I've been advocating us getting the pieces merged for a long time so
> > that we could see what was broken rather than trying to wait for a
> > mythical "perfection" that's been just another week or two out for on
> > the order of months now.
> >
> >> b) remove both, because, screw it
> >
> > This is a part of the contingency plan. But removing for both seems a
> > little extreme.
> >
> > There's also c) Just enable it for radeon. Intel users get pretty
> > modesetting for F11. It's only six months away. Adding the Intel
> > support post-beta just leaves us with way too little runway of testing
> > on what's a significant amount of the hardware running Fedora. And one
> > of the big reasons behind the six month release cycle is that if
> > something is late to the party for release n, n+1 is not that far away.
> >
> > Jeremy
> >
> > --
> > fedora-devel-list mailing list
> > fedora-devel-list@redhat.com
> > https://www.redhat.com/mailman/listinfo/fedora-devel-list
> >
>
> Is there a list of cards that work with this (already tested)? f10 is
> just a month away and my rs485 still cannot function with this code.
> It's an important feature and if it cannot work, then maybe the
> contingency plan should be applied.
>
bug number?
what kernel did you last test with?
stop scare mongering its a beta release, if it still doesn't work at
devel freeze I'll blacklist all the broken machines.
Dave.
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
modesetting feature status
