Random reaction to brightness hotkeys

I don't know if it's related but if the laptop stays idle it will change brightness for a second and then go back to original brightness. Same when I touch the touchpad after idle time
Here is lspci

On my laptop I traced it back to kernel and here is the fix: http://marc.info/?l=linux-acpi&m=124456088526538&w=2

Thank you for your report and for finding and testing that patch! I'm marking confirmed in both gnome-power-manager and listing as an ACPI bug.

There are several other bug reprots with similar symptoms that are now trying this patch.

Dmitriy Geels just added two dupes to this report. He has down some debugging at Bug #350912, I just wanted to make sure his info is carried over (there are more logs and files there, but the following info confirms the behavior the patch in this report fixes):

This laptop has 3 backlight control buttons: dim/bright and 'eco' (which should put laptop into economy mode).
Now, how buttons work (I press buttons not 1 time per second or less):
dmig@dmig-laptop:~$ watch -n 1 cat /proc/acpi/video/VGA1/LCDD/brightness
Every 1,0s: cat /proc/acpi/video/VGA1/LCDD/brightness Sun Mar 29 16:08:54 2009

levels: 0 1 2 3 4 5 6
current: 5

dim button: no changes, gnome brightness popup shows zero level brightness
bright button:
1st press
levels: 0 1 2 3 4 5 6
current: 1
2nd press
levels: 0 1 2 3 4 5 6
current: 4
3rd presslevels: 0 1 2 3 4 5 6
current: 1
4th press
levels: 0 1 2 3 4 5 6
current: 4

eco button:
eco off:
levels: 0 1 2 3 4 5 6
current: 4

eco on: no change
eco off:
backlight blinks twice, then:
levels: 0 1 2 3 4 5 6
current: 2

eco on: no change
eco off: screen blinks 3 times, then level rests on 2

Applied this patch to .30 kernel -- it helped! Brightness control works as expected now. But gnome-brightness-applet is often loosing dbus connection to g-p-m with .30 kernel.

Thank you for your bug report. This bug has been reported to the developers of the software. You can track it and make comments at: http://bugzilla.kernel.org/show_bug.cgi?id=13511

@phcoder: The linux-acpi developers have asked that you resend your patch to the linux-acpi mailing list again [1]. When you do, reference linux kernel bug #13511. Thanks for your help and fixing this bug for many HP laptop owners.

[1] http://bugzilla.kernel.org/show_bug.cgi?id=13511#c2

Relaying a message from the upstream developers:

http://bugzilla.kernel.org/show_bug.cgi?id=13511#c4

patch: create symbol link from backlight sys device to ACPI video device

you have only one backlight sysfs I/F, i.e. /sys/class/backlight/acpi_video0,
right?
please apply this patch [1], and get the output of "cat
/sys/class/backlight/acpi_video0/device/path".
this will help us to know which ACPI video device is used.

[1] http://bugzilla.kernel.org/attachment.cgi?id=21952

sorry for the extra email, upstream updated the patch:
http://bugzilla.kernel.org/attachment.cgi?id=21978

use this instead of the one from my comment on 6/16

Just disassembled laptop's DSDT, there a 3 device LCDD descriptors. All have method _BCL.
First 2 instances:
Device (LCDD)
{
...
Method (_BCL, 0, NotSerialized)
{
    Return (Package (0x10)
    {
        0x0F,
        0x0E,
        0x0D,
        0x0C,
        0x0B,
        0x0A,
        0x09,
        0x08,
        0x07,
        0x06,
        0x05,
        0x04,
        0x03,
        0x02,
        One,
        Zero
    })
}
...
}
And here is third one:
Device (LCDD)
{
...
Method (_BCL, 0, NotSerialized)
{
    Return (Package (0x09)
    {
        0x08,
        0x07,
        0x06,
        0x05,
        0x04,
        0x03,
        0x02,
        One,
        Zero
    })
}
...
}

As far as I understand, this causes backlight control problems and modifying first 2 instances to be like 3d one will fix problem without kernel modification (by overriding DSDT in initrd)?

Hmm... This should be the reason, but patched DSDT didn't fix the problem. Or kernel refused to load my DSDT.
Dammit!
# CONFIG_ACPI_CUSTOM_DSDT is not set
CONFIG_ACPI_CUSTOM_DSDT_FILE=""

Bug #246222 reported for Intrepid alpha 3, marked as fixed, is still there!
Config from 2.6.28 contains one more line:
CONFIG_ACPI_CUSTOM_DSDT_INITRD=y

Kernel refused to load my DSDT:
# CONFIG_ACPI_CUSTOM_DSDT is not set
CONFIG_ACPI_CUSTOM_DSDT_FILE=""

Bug #246222 reported for Intrepid alpha 3 and marked as fixed is still there!
Default kernel, 2.6.28 has one more line:
CONFIG_ACPI_CUSTOM_DSDT_INITRD=y

Thanks Dmitriy, can you post your comments at http://bugzilla.kernel.org/show_bug.cgi?id=13511? The patch and fix work is going on there, and it would probably be best if we directly work with them on that site.

After applying the patch, what was the output of:
cat /sys/class/backlight/acpi_video0/device/path

Could you post the result of that to the kernel bugzilla address?

Oh, I haven't applied later patch. I will do that and post the result