2.6.25-rc1 regression - IBM ACPI backlight

2.6.25-rc1 regression - IBM ACPI backlight

[Lukas Hejtmanek]

Hello,

as of 2.6.25-rc1, there is no more /sys/class/backlight/acpi_video1 which
controlled LVDS backlight on Lenovo ThinkPad T61. There is still acpi_video0
which seems to have sane values but echo N > brightness has no effect at all.

thinkpad_acpi module reports that ACPI controlled backlight is present.

-- 
Lukáš Hejtmánek

Re: 2.6.25-rc1 regression - IBM ACPI backlight

[Henrique de Moraes Holschuh]

On Thu, 14 Feb 2008, Lukas Hejtmanek wrote:
> as of 2.6.25-rc1, there is no more /sys/class/backlight/acpi_video1 which
> controlled LVDS backlight on Lenovo ThinkPad T61. There is still acpi_video0
> which seems to have sane values but echo N > brightness has no effect at all.
> 
> thinkpad_acpi module reports that ACPI controlled backlight is present.

Please report this to linux-acpi @ vger . kernel . org...

It is likely caused by the stuff to locate and disable the unused video
device in the DSDT (apparently it got the wrong device?!).

However, BEFORE reporting to linux-acpi, please test it in single user mode
without X *ever* being loaded on that boot.  New X disables the BIOS
brightness changes that ACPI uses.

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Re: 2.6.25-rc1 regression - IBM ACPI backlight

[Lukas Hejtmanek]

On Thu, Feb 14, 2008 at 10:24:52AM -0200, Henrique de Moraes Holschuh wrote:
> > as of 2.6.25-rc1, there is no more /sys/class/backlight/acpi_video1 which
> > controlled LVDS backlight on Lenovo ThinkPad T61. There is still
> > acpi_video0 which seems to have sane values but echo N > brightness has no
> > effect at all.
> > 
> > thinkpad_acpi module reports that ACPI controlled backlight is present.
> 
> Please report this to linux-acpi @ vger . kernel . org...
> 
> It is likely caused by the stuff to locate and disable the unused video
> device in the DSDT (apparently it got the wrong device?!).
> 
> However, BEFORE reporting to linux-acpi, please test it in single user mode
> without X *ever* being loaded on that boot.  New X disables the BIOS
> brightness changes that ACPI uses.

2.6.24 kernel has two acpi_video[01]. 2.6.25-rc1 does not, it contains only
acpi_video0 (even in text mode in single user without X being loaded) which
does not work at all.

-- 
Lukáš Hejtmánek

Re: 2.6.25-rc1 regression - IBM ACPI backlight

[Henrique de Moraes Holschuh]

On Fri, 15 Feb 2008, Lukas Hejtmanek wrote:
> On Thu, Feb 14, 2008 at 10:24:52AM -0200, Henrique de Moraes Holschuh wrote:
> > > as of 2.6.25-rc1, there is no more /sys/class/backlight/acpi_video1 which
> > > controlled LVDS backlight on Lenovo ThinkPad T61. There is still
> > > acpi_video0 which seems to have sane values but echo N > brightness has no
> > > effect at all.
> > > 
> > > thinkpad_acpi module reports that ACPI controlled backlight is present.
> > 
> > Please report this to linux-acpi @ vger . kernel . org...
> > 
> > It is likely caused by the stuff to locate and disable the unused video
> > device in the DSDT (apparently it got the wrong device?!).
> > 
> > However, BEFORE reporting to linux-acpi, please test it in single user mode
> > without X *ever* being loaded on that boot.  New X disables the BIOS
> > brightness changes that ACPI uses.
> 
> 2.6.24 kernel has two acpi_video[01]. 2.6.25-rc1 does not, it contains only
> acpi_video0 (even in text mode in single user without X being loaded) which
> does not work at all.

The thinkpad DSDT exports two acpi video nodes, but only one is active.
Apparently, ACPI video chose the wrong one.  You did the test by booting
directly to single user mode (without X loading at all since the last
reboot), correct?  That means there is a bug to report, indeed.

Please report it through http://buzilla.kernel.org, as that's the preferred
method of bug reporting by the ACPI people.

-- 
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Re: 2.6.25-rc1 regression - IBM ACPI backlight

[Matthew Garrett]

On Fri, Feb 15, 2008 at 12:02:13PM +0100, Lukas Hejtmanek wrote:

> 2.6.24 kernel has two acpi_video[01]. 2.6.25-rc1 does not, it contains only
> acpi_video0 (even in text mode in single user without X being loaded) which
> does not work at all.

Sigh. Ok. Could you attach the output of acpidump and lspci, please? 
I'll try to figure out why this is wrong.

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: 2.6.25-rc1 regression - IBM ACPI backlight

[Lukas Hejtmanek]

On Fri, Feb 15, 2008 at 12:07:20PM +0000, Matthew Garrett wrote:
> On Fri, Feb 15, 2008 at 12:02:13PM +0100, Lukas Hejtmanek wrote:
> > 2.6.24 kernel has two acpi_video[01]. 2.6.25-rc1 does not, it contains only
> > acpi_video0 (even in text mode in single user without X being loaded) which
> > does not work at all.
> 
> Sigh. Ok. Could you attach the output of acpidump and lspci, please? 
> I'll try to figure out why this is wrong.

I created a bug report here including acpidump and lspci:
http://bugzilla.kernel.org/show_bug.cgi?id=9995

-- 
Lukáš Hejtmánek