On Lenovo T440s with kernel 3.13.0-031300rc1-generic are only 5 modes of brightness. First mode shutdown back light completely off, so there are only 4 modes of brightness. 4 modes are too bright or too low, there is nothing between. It hurts my eyes. as a parameter using this: GRUB_CMDLINE_LINUX_DEFAULT="quiet splash pcie_aspm=force acpi_backlight=vendor" root@dell-xps:/sys/class/backlight/intel_backlight/device/intel_backlight/subsystem/intel_backlight# ll total 0 drwxr-xr-x 3 root root 0 Nov 24 12:05 ./ drwxr-xr-x 5 root root 0 Nov 24 12:05 ../ -r--r--r-- 1 root root 4096 Nov 24 12:05 actual_brightness -rw-r--r-- 1 root root 4096 Nov 24 12:14 bl_power -rw-r--r-- 1 root root 4096 Nov 24 12:13 brightness lrwxrwxrwx 1 root root 0 Nov 24 12:14 device -> ../../card0-eDP-1/ -r--r--r-- 1 root root 4096 Nov 24 12:05 max_brightness drwxr-xr-x 2 root root 0 Nov 24 12:14 power/ lrwxrwxrwx 1 root root 0 Nov 24 12:05 subsystem -> ../../../../../../../class/backlight/ -r--r--r-- 1 root root 4096 Nov 24 12:14 type -rw-r--r-- 1 root root 4096 Nov 24 12:05 uevent root@dell-xps:/sys/class/backlight/intel_backlight/device/intel_backlight/subsystem/intel_backlight#
What's the value of max_brightness? BTW, is the acpi_video interface broken? Please also attach your acpidump, thanks.
Brightness: bio@dell-xps:/sys/class/backlight/intel_backlight/device/intel_backlight/subsystem/intel_backlight$ cat actual_brightness step1: 0 step2: 170 step3: 340 step4: 510 step5: 680 step6: 851 - ok. 6 steps...(sorry i wrote previously only 5 steps) bio@dell-xps:/sys/class/backlight/intel_backlight/device/intel_backlight/subsystem/intel_backlight$ cat max_brightness 851 I removed option pcie_aspm=force and it's running fine. for the acpidump i tried, but got error: bio@dell-xps:~/Desktop$ sudo acpidump > acpi.dat Wrong checksum for FADT! bio@dell-xps:~/Desktop$ sudo acpidump -t SSDT > SSDT.dat Wrong checksum for FADT!
The step is decided by GUI, not kernel. Kernel just exposes backlight control interface for GUI to use. Which GUI are you using?
Xfce 4.10, but I haven't change anything else except kernel upgrade.
What's the value of max_brightness on the good_old kernel?
Hi, in good old kernel "Linux t440s 3.12.0-031200-generic #201311031935 SMP Mon Nov 4 00:36:54 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux" is max brigtness 765. steps are following: bio@t440s:/sys/class/backlight/intel_backlight/device/intel_backlight/subsystem/intel_backlight$ cat brightness 9 15 24 30 39 45 54 75 102 138 186 252 330 441 579 765 765
(In reply to tomasi from comment #2) > I removed option pcie_aspm=force and it's running fine. What does this mean, the step is OK without pcie_aspm=force?
i meant system is running fine, it boots, there is no black screen after boot...
If intel_backlight's max_brightness is 765 you have all the steps in range 0..765. You can try them by running sudo sh -c "echo $val > /sys/class/backlight/intel_backlight/brightness" with $val in range 0..765. How does that work? Is there a specific reason you use acpi_backlight=vendor? Why does comment #0 say "Lenovo T440s" and subject says "dell-xps"? Which one is it?
I will try change $val later. Now I have older kernel. Is there a specific reason you use acpi_backlight=vendor? - yes, otherwise buttons (brightness up and down) on laptop doesn't work. Why does comment #0 say "Lenovo T440s" and subject says "dell-xps"? Which one is it? - I upgraded laptop and didn't change hostname. Defect is on T440s.
sudo sh -c "echo 110 > /sys/class/backlight/intel_backlight/brightness" is changing brightness. Still in version 3.13.0-031300rc2 there are only 6 steps of brightness. 3.13.0-031300rc2-generic #201311291635 SMP Fri Nov 29 21:36:49 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux
(In reply to tomasi from comment #11) > sudo sh -c "echo 110 > /sys/class/backlight/intel_backlight/brightness" > is changing brightness. > Still in version 3.13.0-031300rc2 there are only 6 steps of brightness. Please be more specific. If you echo 0, 50, 100, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, and 750 into intel_backlight/brightness, are there 6 or 16 steps of brightness?!
Also please mount debugfs and attach /sys/kernel/debug/dri/0/i915_opregion.
I have only 6 steps when I'm pressing FN+brightess buttons on the keyboard.
Created attachment 117931 [details] i915
There is no issue with settings brightness via commandline. Issue is, when using FN+brighness up/down butttons. From kernel 3.13 (kernel 3.12 was ok) there are only 6 steps of brightness. i915_opregion is attached, I'm not very familiar with debugfs.
(In reply to tomasi from comment #16) > There is no issue with settings brightness via commandline. > Issue is, when using FN+brighness up/down butttons. From kernel 3.13 (kernel > 3.12 was ok) there are only 6 steps of brightness. > i915_opregion is attached, I'm not very familiar with debugfs. That sounds more like a regression somewhere else than in the i915.ko driver. Have you attempted to bisect where this regression was introduced?
(In reply to Daniel Vetter from comment #17) > (In reply to tomasi from comment #16) > > There is no issue with settings brightness via commandline. > > Issue is, when using FN+brighness up/down butttons. From kernel 3.13 > (kernel > > 3.12 was ok) there are only 6 steps of brightness. > > i915_opregion is attached, I'm not very familiar with debugfs. > > That sounds more like a regression somewhere else than in the i915.ko > driver. Have you attempted to bisect where this regression was introduced? No, i haven't tried. What should I attach - logs/debugs?
I mean doing a kernel bisect on the git trees: http://www.reactivated.net/weblog/archives/2006/01/using-git-bisect-to-find-buggy-kernel-patches/ There's also nice howtos for the distro-specific stuff around.
Sorry for long answer, but I have no results. Bisection involves building kernel.. Is there any other way?
If the /sys/class/backlight/intel_backlight/brightness interface works fine and supports as many brightness levels as you like, this is not a bug in i915. If you're using acpi_backlight=vendor kernel parameter, I believe ACPI video in 3.13 will only report the backlight keys to userspace, and won't change the brightness. Aaron, is this correct? In this case, it's also not a bug in ACPI video. I think what's left is userspace. Aaron, any other things to check?
(In reply to Jani Nikula from comment #21) > If the /sys/class/backlight/intel_backlight/brightness interface works fine > and supports as many brightness levels as you like, this is not a bug in > i915. Agree. > > If you're using acpi_backlight=vendor kernel parameter, I believe ACPI video > in 3.13 will only report the backlight keys to userspace, and won't change > the brightness. Aaron, is this correct? In this case, it's also not a bug in > ACPI video. Yes, except that I'm not sure if the vendor driver(thinkpad-acpi in this case) would do something weird. Hi tomasi, Please boot v3.13 kernel with video.use_native_backlight=1 and remove acpi_backlight=vendor, see if that makes a difference. That should make thinkpad-acpi do nothing for video related key.
Hi, I guess, this solved my issue. video.use_native_backlight=1. Bellow are two examples. bio@t440s:~$ uname -a Linux t440s 3.13.0-031300-generic #201401192235 SMP Mon Jan 20 03:36:48 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux a/- parameter was wrong (so basically without parameters) GRUB_CMDLINE_LINUX_DEFAULT="quiet splash use_native_backlight=1" bio@t440s:/sys/class/backlight/intel_backlight/device/intel_backlight/subsystem/intel_backlight$ cat brightness 75, 138, 765 Only three modes are avalable b/ - good parameters - video.use_native_backlight=1 GRUB_CMDLINE_LINUX_DEFAULT="quiet splash video.use_native_backlight=1" modes are: brigtness from down to max 0, 85, 170, 255, 340, 425, 510, 595, 680, 765, 851 from max to low 766, 681, 596, 511, 426, 341, 256, 171, 86, 1 remark: 0 .. means black screen Strange is 1 difference when going up and then down. So... parameters helped, but I prefer more modes around 100 value like i wrote in comment #6. But, ok. What can I do.
side effect :-( after removing power cord brightness is on max level, after attaching is on max level again.
Please attach your acpidump: # acpidump > acpidump.txt And dmi info: # dmidecode > dmi.info
bio@t440s:/tmp/acpi$ sudo acpidump > acpidump.txt Wrong checksum for FADT! dmidecode attached
Created attachment 123481 [details] dmidecode
I suppose the acpidump.txt is still generated?
Created attachment 123531 [details] acpidump
(In reply to tomasi from comment #24) > side effect :-( > after removing power cord brightness is on max level, after attaching is on > max level again. This may be related to GUI. If you boot into console, i.e. do not start any GUI, and when you remove the power cord, does the brightness change?
It's not related to the GUI. I stopped lightdm service. results: - fn + brightness down/up keys were not working - brightness was on max when attaching or removing power cord from laptop.. (I had to use echo $val > /sys/class/backlight/intel_backlight/brightness for changing brightness down)
Sorry for replying back. Please boot into console mode and then enable backlight debug message by: # cd /sys/kernel/debug/dynamic_debug # echo 'module backlight +pfmt' > control After this, remove/attach power cord and see if you get any messages by checking dmesg.
latest kernel: Linux t440s 3.14.0-031400rc2-generic #201402092235 SMP Mon Feb 10 03:37:33 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux dmesg shows following: --cord removed -- brightness max [ 265.939458] init: anacron main process (4640) killed by TERM signal [ 265.978072] thinkpad_acpi: EC reports that Thermal Table has changed -- cord attached -- brightness max [ 275.652095] thinkpad_acpi: EC reports that Thermal Table has changed [ 275.655289] e1000e 0000:00:19.0: irq 58 for MSI/MSI-X -- and so on... --cord removed -- brightness max [ 305.191235] init: anacron main process (4906) killed by TERM signal [ 305.232855] thinkpad_acpi: EC reports that Thermal Table has changed -- cord attached -- brightness max [ 313.184167] thinkpad_acpi: EC reports that Thermal Table has changed [ 313.187649] e1000e 0000:00:19.0: irq 58 for MSI/MSI-X
This suggests the firmware did something on power cord attach/detach instead of kernel.
Not fixable for the backlight max on power cord attach/detach.