Baytrail problems (was Re: Linux 3.13-rc8)

Baytrail problems (was Re: Linux 3.13-rc8)

[Linus Torvalds]

Adding in the appropriate people...

On Tue, Jan 14, 2014 at 7:01 PM, One Thousand Gnomes
<gnomes@lxorguk.ukuu.org.uk> wrote:
>
> It seems to hate Baytrail/T
>
> My ASUS T100TA has gone from 3.11 'needs video=VGA-1:blah' to get the mode
> right but otherwise running nicely and playing 3D games to 3.13-rc8
>
> - crashes and burns during boot if the Baytrail pinctrl driver is compiled
>   in

Andy, Paul? The changes to the pinctrl driver since 3.11 look trivial,
but I guess the GPIO and ACPI ID addition ends up also enabling all
the old code that Alan probably never ran in 3.11 because the driver
didn't trigger on his machine.

Alan, do you have a crash picture/log?

> - i915 console doesn't work

Daniel/Dave?

> - X can be kind of hacked to work by adding mode lines wit xrandr and
>   turning the vga back on
>
> Bugs filed on both issues. The pinctrl one is nasty as it makes the i915
> driver crash during initialisation, so the only way you can debug it is
> pretty much a USB debug cable.
>
> Without pinctrl the crash goes away but the display still doesn't work.

Thanks,

              Linus

Re: Baytrail problems (was Re: Linux 3.13-rc8)

[Linus Walleij]

On Thu, Jan 16, 2014 at 2:49 AM, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
> Adding in the appropriate people...
>
> On Tue, Jan 14, 2014 at 7:01 PM, One Thousand Gnomes
> <gnomes@lxorguk.ukuu.org.uk> wrote:
>>
>> It seems to hate Baytrail/T
>>
>> My ASUS T100TA has gone from 3.11 'needs video=VGA-1:blah' to get the mode
>> right but otherwise running nicely and playing 3D games to 3.13-rc8
>>
>> - crashes and burns during boot if the Baytrail pinctrl driver is compiled
>>   in
>
> Andy, Paul? The changes to the pinctrl driver since 3.11 look trivial,
> but I guess the GPIO and ACPI ID addition ends up also enabling all
> the old code that Alan probably never ran in 3.11 because the driver
> didn't trigger on his machine.

So that is commit f6308b36c411dc5
"ACPI: Add BayTrail SoC GPIO and LPSS ACPI IDs"?

This seems to have been selected for -stable as well so the
problem will be spreading as distros start pushing stable kernel
updates :-(

That came in through the ACPI tree, Rafael do you want to
revert it if there is no better quickfix?

This adds an interesting twist to the general statement that
new PCI/ACPI IDs should go into stable BTW, maybe they
should not always do that.

Yours,
Linus Walleij

Re: Baytrail problems (was Re: Linux 3.13-rc8)

[Andy Shevchenko]

On Thu, 2014-01-16 at 10:29 +0100, Linus Walleij wrote:
> On Thu, Jan 16, 2014 at 2:49 AM, Linus Torvalds
> <torvalds@linux-foundation.org> wrote:
> > Adding in the appropriate people...
> >
> > On Tue, Jan 14, 2014 at 7:01 PM, One Thousand Gnomes
> > <gnomes@lxorguk.ukuu.org.uk> wrote:
> >>
> >> It seems to hate Baytrail/T
> >>
> >> My ASUS T100TA has gone from 3.11 'needs video=VGA-1:blah' to get the mode
> >> right but otherwise running nicely and playing 3D games to 3.13-rc8
> >>
> >> - crashes and burns during boot if the Baytrail pinctrl driver is compiled
> >>   in
> >
> > Andy, Paul? The changes to the pinctrl driver since 3.11 look trivial,
> > but I guess the GPIO and ACPI ID addition ends up also enabling all
> > the old code that Alan probably never ran in 3.11 because the driver
> > didn't trigger on his machine.
> 
> So that is commit f6308b36c411dc5
> "ACPI: Add BayTrail SoC GPIO and LPSS ACPI IDs"?
> 
> This seems to have been selected for -stable as well so the
> problem will be spreading as distros start pushing stable kernel
> updates :-(
> 
> That came in through the ACPI tree, Rafael do you want to
> revert it if there is no better quickfix?

It seems Mathias and Mika (Cc:ed) already heard about the issue and have
been working on the solution.

> 
> This adds an interesting twist to the general statement that
> new PCI/ACPI IDs should go into stable BTW, maybe they
> should not always do that.


-- 
Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Intel Finland Oy

Re: Baytrail problems (was Re: Linux 3.13-rc8)

[Daniel Vetter]

On Thu, Jan 16, 2014 at 2:49 AM, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>> - i915 console doesn't work
>
> Daniel/Dave?

We have a bugzilla from Alan Cox but haven't gained too much clue yet
what's been going wrong. Especially that old kernels managed to get it
to work by forcing the vga output is rather puzzling:

https://bugzilla.kernel.org/show_bug.cgi?id=68451

Cheers, Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch

Re: Baytrail problems (was Re: Linux 3.13-rc8)

[Rafael J. Wysocki]

On Thursday, January 16, 2014 10:29:45 AM Linus Walleij wrote:
> On Thu, Jan 16, 2014 at 2:49 AM, Linus Torvalds
> <torvalds@linux-foundation.org> wrote:
> > Adding in the appropriate people...
> >
> > On Tue, Jan 14, 2014 at 7:01 PM, One Thousand Gnomes
> > <gnomes@lxorguk.ukuu.org.uk> wrote:
> >>
> >> It seems to hate Baytrail/T
> >>
> >> My ASUS T100TA has gone from 3.11 'needs video=VGA-1:blah' to get the mode
> >> right but otherwise running nicely and playing 3D games to 3.13-rc8
> >>
> >> - crashes and burns during boot if the Baytrail pinctrl driver is compiled
> >>   in
> >
> > Andy, Paul? The changes to the pinctrl driver since 3.11 look trivial,
> > but I guess the GPIO and ACPI ID addition ends up also enabling all
> > the old code that Alan probably never ran in 3.11 because the driver
> > didn't trigger on his machine.
> 
> So that is commit f6308b36c411dc5
> "ACPI: Add BayTrail SoC GPIO and LPSS ACPI IDs"?
> 
> This seems to have been selected for -stable as well so the
> problem will be spreading as distros start pushing stable kernel
> updates :-(
> 
> That came in through the ACPI tree, Rafael do you want to
> revert it if there is no better quickfix?

Yes, I'll push either a revert or a fix before the weekend.

> This adds an interesting twist to the general statement that
> new PCI/ACPI IDs should go into stable BTW, maybe they
> should not always do that.

Well, this is the first time we've had such a problem with ACPI IDs, but
since we do have this problem, I'm not going to tag any commits like that for
stable any more (and I'm not going to fast-track them any more for that
matter).

Thanks!

-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.

Re: Baytrail problems (was Re: Linux 3.13-rc8)

[Rafael J. Wysocki]

On Thursday, January 16, 2014 05:01:19 PM Rafael J. Wysocki wrote:
> On Thursday, January 16, 2014 10:29:45 AM Linus Walleij wrote:
> > On Thu, Jan 16, 2014 at 2:49 AM, Linus Torvalds
> > <torvalds@linux-foundation.org> wrote:
> > > Adding in the appropriate people...
> > >
> > > On Tue, Jan 14, 2014 at 7:01 PM, One Thousand Gnomes
> > > <gnomes@lxorguk.ukuu.org.uk> wrote:
> > >>
> > >> It seems to hate Baytrail/T
> > >>
> > >> My ASUS T100TA has gone from 3.11 'needs video=VGA-1:blah' to get the mode
> > >> right but otherwise running nicely and playing 3D games to 3.13-rc8
> > >>
> > >> - crashes and burns during boot if the Baytrail pinctrl driver is compiled
> > >>   in
> > >
> > > Andy, Paul? The changes to the pinctrl driver since 3.11 look trivial,
> > > but I guess the GPIO and ACPI ID addition ends up also enabling all
> > > the old code that Alan probably never ran in 3.11 because the driver
> > > didn't trigger on his machine.
> > 
> > So that is commit f6308b36c411dc5
> > "ACPI: Add BayTrail SoC GPIO and LPSS ACPI IDs"?
> > 
> > This seems to have been selected for -stable as well so the
> > problem will be spreading as distros start pushing stable kernel
> > updates :-(
> > 
> > That came in through the ACPI tree, Rafael do you want to
> > revert it if there is no better quickfix?
> 
> Yes, I'll push either a revert or a fix before the weekend.

OK, I haven't seen any fixes for this so far, so I'm queing up a revert of
commit f6308b36c411dc5 to be pushed later today.

Thanks!

-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.

Re: Baytrail problems (was Re: Linux 3.13-rc8)

[Rafael J. Wysocki]

On Friday, January 17, 2014 02:38:32 PM Rafael J. Wysocki wrote:
> On Thursday, January 16, 2014 05:01:19 PM Rafael J. Wysocki wrote:
> > On Thursday, January 16, 2014 10:29:45 AM Linus Walleij wrote:
> > > On Thu, Jan 16, 2014 at 2:49 AM, Linus Torvalds
> > > <torvalds@linux-foundation.org> wrote:
> > > > Adding in the appropriate people...
> > > >
> > > > On Tue, Jan 14, 2014 at 7:01 PM, One Thousand Gnomes
> > > > <gnomes@lxorguk.ukuu.org.uk> wrote:
> > > >>
> > > >> It seems to hate Baytrail/T
> > > >>
> > > >> My ASUS T100TA has gone from 3.11 'needs video=VGA-1:blah' to get the mode
> > > >> right but otherwise running nicely and playing 3D games to 3.13-rc8
> > > >>
> > > >> - crashes and burns during boot if the Baytrail pinctrl driver is compiled
> > > >>   in
> > > >
> > > > Andy, Paul? The changes to the pinctrl driver since 3.11 look trivial,
> > > > but I guess the GPIO and ACPI ID addition ends up also enabling all
> > > > the old code that Alan probably never ran in 3.11 because the driver
> > > > didn't trigger on his machine.
> > > 
> > > So that is commit f6308b36c411dc5
> > > "ACPI: Add BayTrail SoC GPIO and LPSS ACPI IDs"?
> > > 
> > > This seems to have been selected for -stable as well so the
> > > problem will be spreading as distros start pushing stable kernel
> > > updates :-(
> > > 
> > > That came in through the ACPI tree, Rafael do you want to
> > > revert it if there is no better quickfix?
> > 
> > Yes, I'll push either a revert or a fix before the weekend.
> 
> OK, I haven't seen any fixes for this so far, so I'm queing up a revert of
> commit f6308b36c411dc5 to be pushed later today.

BTW, I haven't tagged it for stable, so I'm not sure who has requested
stable to include it and why.

Thanks!

-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.