Bug 39322

Please run acpi_listen while clicking the Fn + brightness up/down keys to check whether the even has been triggered for backlight changing.

After echo <number> > /sys/class/backlight/acpi_video0/brightness, cat the brightness and actual_brightness to identify whether the number has been changed.

>Has anyone got an idea? I have tried with kernel 2.6.38.5, and with 
>acpi_backlight=video, acpi_backlight=legacy and acpi_backlight=vendor but it 
>still doesn't work.
>When acpi_backlight=vendor, the Fn + backlight up and down button just don't
>>work, and with the two others options, they work, but the backlight doesn't
>>change.

What meaning of "with the two others options, they work" with "but the backlight doesn't change"?

Hello,

acpi_listen doesn't show anything when Fn + brightness up/down buttons are pressed.

The number in brightness and actual_brightness has been changed, but my brightness still doesn't work.

"with the two others options, they work, but the backlight doesn't change" means that when pressing Fn + brightness up/down button, a progress bar appears and is changed, but my brightness doesn't change.
Actually, this also happens without any boot options, I just need to press Fn + brightness up/down buttons longer.

"my brightness still doesn't work" -> "my brightness still doesn't change"

Please try acpi_listen in the kernel 2.6.37.6 which works correctly.

Attach the output of the acpidump.

Actually, I have just realized that I had made an error : it did not work in 2.6.37.6 but in 2.6.37.5. The 2.6.37.6 version has never been packaged for Archlinux. They have jumped from 2.6.37.5 to 2.6.38.2.

With 2.6.37.5 kernel, acpi_listen doesn't show anything when Fn + brightness up/down buttons are pressed, but brightness changes.

I have attached the output of acpidump.

Created attachment 65592 [details]
Acpidump output

Please try to bisect between 2.6.37.5. and 2.6.38.2. to identify which version  became bad. 

Attach the output of "grep . /sys/firmware/acpi/interrupts/*" before and after  Fn + brightness up/down buttons are pressed on the bad kernel version. 

Is there anything different if you boot with kernel parameter acpi_osi="!Windows 2009" on the bad kernel version?

Is there anything different if you boot with kernel parameter acpi_osi="!Windows 2009" AND acpi_osi="!Windows 2006" on the bad kernel version?

Does "echo <number> > /sys/class/backlight/acpi_video0/brightness" work on the 2.6.37.5?

Hello,

I have tried to compile 2.6.37.6 : the brightness worked.
Unfortunately, I could compile 2.6.38.1, because it gave me this error : 
/tmp/ccCUigGd.s: Assembler messages:
/tmp/ccCUigGd.s: Error: .size expression for apf_page_fault does not evaluate to a constant
make[2]: *** [arch/x86/kernel/entry_32.o] Erreur 1
make[1]: *** [arch/x86/kernel] Erreur 2
make: *** [arch/x86] Erreur 2
(I compiled it with the ArchLinux configuration for 2.6.38.2)

So the bug was either introduced in 2.6.38.1 or 2.6.38.2.

"echo <number> > /sys/class/backlight/acpi_video0/brightness" works with 2.6.37.5.

For the bad kernel version, I have used the latest : 2.6.39.3.

I will attach the output of "grep . /sys/firmware/acpi/interrupts/*".

With acpi_osi="!Windows 2009" and with acpi_osi="!Windows 2009" AND acpi_osi="!Windows 2006", the brightness still doesn't change, but acpi -i says : 

Battery 0: design capacity 786 mAh, last full capacity 786 mAh = 100%

instead of : 

Battery 0: design capacity 4623 mAh, last full capacity 786 mAh = 17%

Created attachment 65722 [details]
Output of "grep . /sys/firmware/acpi/interrupts/*" before pressing Fn + brightness up/down

Created attachment 65732 [details]
Output of "grep . /sys/firmware/acpi/interrupts/*" after pressing Fn + brightness up/down

Please attach the output of "grep . /sys/firmware/acpi/interrupts/*" before and after Fn + brightness up/down buttons are pressed on the 2.6.37.5.

Created attachment 65912 [details]
Output of "grep . /sys/firmware/acpi/interrupts/*" before pressing Fn + brightness up/down with 2.6.37.5.

Created attachment 65922 [details]
Output of "grep . /sys/firmware/acpi/interrupts/*" after pressing Fn + brightness up/down with 2.6.37.5.

Hello,
I have attached the files.

The two attachment are the same file "before_26375". Please have a check. Thanks.

Created attachment 65972 [details]
Output of "grep . /sys/firmware/acpi/interrupts/*" before pressing Fn + brightness up/down with 2.6.37.5.

Created attachment 65982 [details]
Output of "grep . /sys/firmware/acpi/interrupts/*" after pressing Fn + brightness up/down with 2.6.37.5.

Hello,

Sorry for that.
I have reuploaded the good files.

Please override the DSDT with my attched DSDT.hex respectively in the 2.6.37.5 and 2.6.38.2 or up.

http://www.lesswatts.org/projects/acpi/overridingDSDT.php

Do following operations on the two kernel:
  
(1)Run "echo 1 > /sys/module/acpi/parameters/aml_debug_output".
   Pressing Fn + brightness up/down
   Attach the dmesg.

(2)Run "echo 1 > /sys/module/acpi/parameters/aml_debug_output".
   Run "echo <number> > /sys/class/backlight/acpi_video0/brightness"
   Attach the dmesg.

Check whether "CONFIG_ACPI_PROC_EVENT=y" in the kernel configure or not.

Created attachment 66172 [details]
DSDT.hex

Hello,

I have tried to recompile the kernel 2.6.37.5 to override the DSDT but unfortunatly there was an error : 
  CC [M]  drivers/net/igbvf/ethtool.o
  In file included from drivers/net/igbvf/ethtool.c:36:0:
  drivers/net/igbvf/igbvf.h:129:15: erreur: duplicate member ‘page’
  make[3]: *** [drivers/net/igbvf/ethtool.o] Erreur 1
  make[2]: *** [drivers/net/igbvf] Erreur 2
  make[1]: *** [drivers/net] Erreur 2
  make: *** [drivers] Erreur 2

So I have compiled the kernel 2.6.37.6 and overrided the DSDT (the brightness works in this version), and did what you told me.

I have compiled the kernel 2.6.39.3 and override the DSDT too. I have also done what you told me.

I will upload the files.

CONFIG_ACPI_PROC_EVENT=y is in the kernel config.

Created attachment 66242 [details]
dmesg output after doing Fn + brightness up/down (with debug on and with kernel 2.6.37.6)

Created attachment 66252 [details]
dmesg output after doing echo <number> /sys/class/backlight/acpi_video0/brightness (with debug on and with kernel 2.6.37.6)

Created attachment 66262 [details]
dmesg output after doing echo <number> /sys/class/backlight/acpi_video0/brightness (with debug on and with kernel 2.6.37.6)

Created attachment 66272 [details]
dmesg output after doing Fn + brightness up/down (with debug on and with kernel 2.6.39.3)

Created attachment 66282 [details]
dmesg output after doing echo <number> > /sys/class/backlight/acpi_video0/brightness (with debug on and with kernel 2.6.39.3)

(I have done first Fn + brightness up/down and then echo <number> > /sys/class/backlight/acpi_video0/brightness for each kernel)

Please make clean 2.6.38.1 kernel and make to compile successfully. 
Identify which version the issue is introduced. Thanks.

Hello,

I really can't compile 2.6.38.1, even after doing a make clean (and even a make mrproper).

The error is the same :
  LD      arch/x86/crypto/built-in.o
  CC      arch/x86/kernel/process_32.o
  CC      arch/x86/kernel/signal.o
  AS      arch/x86/kernel/entry_32.o
/tmp/cc5ufEEA.s: Assembler messages:
/tmp/cc5ufEEA.s: Error: .size expression for apf_page_fault does not evaluate to a constant
make[2]: *** [arch/x86/kernel/entry_32.o] Erreur 1
make[1]: *** [arch/x86/kernel] Erreur 2
make: *** [arch/x86] Erreur 2

What can I do?

git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git linux-2.6
git checkout v2.6.38-rc1.
I have check v2.6.38-rc1 on my machine. It's ok.

I have tried with git and it still doesn't work (same error). 
What should I do?

Please update gcc and try again.

Hello,

I am using the last version of gcc : 4.6.1.

Hi,

I'm also affected by the same issue. I own a HP Probook 6560b that has an integrated ATI Radeon 6470M card and I run the 3.0 kernel, however the brightness does not change when I press the corresponding keyboard keys. acpi_listen recognizes the events:

video/brightnessdown BRTDN 00000087 00000000
video/brightnessup BRTUP 00000086 00000000
video/brightnessdown BRTDN 00000087 00000000
video/brightnessdown BRTDN 00000087 00000000
video/brightnessdown BRTDN 00000087 00000000
video/brightnessup BRTUP 00000086 00000000

but kernel message log reports:

Aug  5 09:11:32 localhost logger: ACPI event unhandled: video/brightnessup BRTUP 00000086 00000000
Aug  5 09:11:32 localhost logger: ACPI event unhandled: video/brightnessdown BRTDN 00000087 00000000
Aug  5 09:11:33 localhost logger: ACPI event unhandled: video/brightnessdown BRTDN 00000087 00000000
Aug  5 09:11:33 localhost logger: ACPI event unhandled: video/brightnessdown BRTDN 00000087 00000000
Aug  5 09:11:33 localhost logger: ACPI event unhandled: video/brightnessup BRTUP 00000086 00000000
Aug  5 09:11:33 localhost logger: ACPI event unhandled: video/brightnessup BRTUP 00000086 00000000

If you need more information, please say so.

Hello,

I have tried with Kernel 3.0, and it still doesn't work.

HI Vasileios:
   Is this a regression from 2.6.38? If no, please apply a new bug. This maybe is another bug.

HI bugzilla@ealexp.eu:
   I have no idea about the compile error. There are other releases between 2.6.37.6 and 2.6.38.1. (From git:2.6.37-rc7 2.6.37-rc8 and 2.6.38).I think you can bisect these to identify which version become bad. 

Did you test on the ARCHlinux kernel or upstream kernel?

(In reply to comment #37)
> HI Vasileios:
>    Is this a regression from 2.6.38? If no, please apply a new bug. This
>    maybe
> is another bug.

Since, this is a new setup, I wouldn't know if the issue persists from 2.6.38. I used 3.0.0 from the beginning. I will open a new bug, thanks.

Hello,

What does "bisect" mean?

First I used the ArchLinux kernel, then I downloaded the kernel from kernel.org and compiled it. But I compiled it with the archlinux configuration. Is that bad? Should I configure it myself?

> What does "bisect" mean?
You can regard it as test those kernels to find a good one.

> First I used the ArchLinux kernel, then I downloaded the kernel from
> kernel.org
> and compiled it. But I compiled it with the archlinux configuration. Is that
> bad? Should I configure it myself?
That's ok with the archlinux configuration to compile the upstream kernel.

(In reply to comment #40)
> > What does "bisect" mean?
> You can regard it as test those kernels to find a good one.
Sorry. You can regard it as test those kernels to find the one which begin to become bad.

Hello,

I will try that, but this means downloading and compiling a lot of kernels. Are you sure there isn't a quicker way to do it?

Hello,

Do you know where I could download "git:2.6.37-rc7 2.6.37-rc8 and 2.6.38" kernel versions from, please?

Under the git kernel root directory.
run "git checkout v2.6.37-rc7"
others follow this rule.

bug closed as there is no response from the bug reporter.
please feel free to reopen it if the problem still exists in the latest upstream kernel.