Hi, I upgraded to 2.6.26.1 and my X61 backlight (which has never really worked right) gave me more troubles than usual, so I started to try to track things down a little more, and found a mess that I'd like some help sorting out. I think most of these are userspace issues but I think the kernel is having some troubles as well. For the userspace stuff, I'd like to get some input from people more knowledgeable before I go filing bugs. It's insanely complicated how all of this software works together so I'll just explain everything I've found: - I need to load thinkpad-acpi with experimental=1 so I can control WAN power with /proc/acpi/ibm/wan. Other than that thinkpad-acpi seems to be doing its job correctly for brightness control: thinkpad_acpi: ThinkPad ACPI Extras v0.20 thinkpad_acpi: http://ibm-acpi.sf.net/ thinkpad_acpi: ThinkPad BIOS 7NETA6WW (2.06 ), EC 7MHT25WW-1.03 thinkpad_acpi: Lenovo ThinkPad X61, model 7675CTO thinkpad_acpi: radio switch found; radios are enabled thinkpad_acpi: This ThinkPad has standard ACPI backlight brightness control, supported by the ACPI video driver thinkpad_acpi: Disabling thinkpad-acpi brightness events by default... thinkpad_acpi: Lenovo BIOS switched to ACPI backlight control mode thinkpad_acpi: standard ACPI backlight interface available, not loading native one... - I'm loading ACPI video.ko too. It generates hotkey events correctly: video LCD0 00000086 00000000 video LCD0 00000087 00000000 - Bug: two backlight devices show up: /sys/class/backlight/acpi_video0 /sys/class/backlight/acpi_video1 - Pressing the brightness hotkeys (single-user mode, no hal, no gnome-power-manager, no acpid) causes both actual_brightness values to change, but the backlight does not physically change brightness. - Similarly, manually writing to these controls acts strangely: # cat acpi_video0/actual_brightness 15 # cat acpi_video1/actual_brightness 15 # echo 8 > acpi_video0/brightness # the backlight does NOT change # cat acpi_video0/actual_brightness 8 # cat acpi_video1/actual_brightness 8 # echo 10 > acpi_video1/brightness # the backlight DOES change # cat acpi_video0/actual_brightness 8 # cat acpi_video1/actual_brightness 10 - It seems that: - Writing to acpi_video0/brightness changes actual_brightness for both, but has no actual effect. - Pressing hotkey buttons is the same as writing to acpi_video0/brightness. - Writing to acpi_video1/brightness changes actual_brightness for only acpi_video1, and actually works. - Gnome-power-manager saw the hotkeys but didn't change the backlight. When hal & gnome-power-manager & acpid are running, the brightness key ACPI events are delivered as fake keystrokes which g-p-m receives. When it wants to change brightness, it invokes the HAL GetBrightness function to check, then uses SetBrightness to set, but ONLY if they differ. Roughly, GetBrightness = cat acpi_videoX/actual_brightness SetBrightness = echo > acpi_videoX/brightness - Since pressing hotkey buttons DOES change acpi_video1/actual_brightness, g-p-m thinks someone already set the proper brightness, and so it never calls SetBrightness. Result: backlight does not change. - But, it still half-worked, sometimes. Why? Because of the terrible Debian package "hotkey-setup". This package first writes 0xffffffff to /proc/acpi/ibm/hotkey at startup, which is not good. Then, it runs a program "thinkpad_keys" which directly polls /dev/nvram for hotkeys, and sends brightness keys as fake keyboard inputs using /dev/input/uinput. g-p-m was therefore receiving brightness keys TWICE (one from acpid, one from thinkpad_keys). This means the bug with actual_brightness didn't matter because g-p-m wanted the value to change twice, so it would still do a SetBrightness. (It was still buggy, though; I think thinkpad_keys missed some of the keypresses). I've since purged this hotkey-setup package from my system because I definitely don't want it to change /proc/acpi/ibm/hotkey or do its own nvram polling. - With hotkey-setup gone, I can make g-p-m work correctly by: echo N > /sys/module/video/parameters/brightness_switch_enabled This causes acpi_video1/actual_brightness to NOT change when I hit hotkeys, so now g-p-m does the right thing (sees that it is unchanged, and changes it. Good!) It seems like this option should be unnecessary. It's as if the BIOS thinks it's changing the brightness but failing to do so. Is this a DSDT bug maybe? Or a bug in video.ko? - After all this, brightness mostly works. There is still an issue with plugging/unplugging the AC adapter. In single-user-mode: On AC: # echo 0 > acpi_video1/brightness # dark # echo 15 > acpi_video1/brightness # fully bright Unplugged AC here # no change # echo 15 > acpi_video1/brightness # no change # echo 0 > acpi_video1/brightness # dark # echo 15 > acpi_video1/brightness # --HALF-- bright Plugged AC here # no change # echo 15 > acpi_video1/brightness # no change # echo 0 > acpi_video1/brightness # dark # echo 15 > acpi_video1/brightness # fully bright Basically removing the AC adapter reduces the available brightness range, BUT doesn't take effect until I actually try to change the brightness. I don't know what should change the brightness in this case. I could add an ACPI hook, but I'd need something like: b=$(((`cat acpi_video1/brightness` + 8) % 16)) echo $(($b + 1) % 16)) > acpi_video1/brightness echo $b > acpi_video1/brightness to ensure that the value actually gets changed, which just seems like a hack. By the way, the Debian package "acpi-support" also sets up a modprobe.d file containing: options thinkpad_acpi hotkey=enable,0xffffbf experimental=1 which I have removed. Another broken userspace package? Apparently this file was placed here to fix problems on someone's x60; this whole setup is so complicated it seems like it would be an endless problem of fixing one laptop by breaking another. -jim ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ ibm-acpi-devel mailing list ibm-acpi-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.sourceforge.net/lists/listinfo/ibm-acpi-devel
