Created attachment 119648 [details] System Information (Manjaro on LENOVO L540) Please see: https://bugs.archlinux.org/task/45501?project=1&cat[0]=2&string=xf86-video-intel Description: After upgrading this package (Manajro package management) and after resuming from sleep and logging back in (from lightdm) the backlight turns off. Switching to a tty doesn't turn it back on. Running commands to increase the brightness like using /sys/class/backlight/intel_backlight/brightness do not help. I know it's just the backlight because if I shine a light on the screen I can make out features and run commands in a terminal. Additional info: * Upgrade from 2.99.917-5 to 1:2.99.917+364+gb24e758-1 Steps to reproduce: * Upgrade xf86-video-intel Downgrading to 2.99.917-5 fixed it again
Confirmed on a Lenovo T440. The bug reports on Arch seem to suggest that the bug is mainly present on Lenovo hardware ...
What does "grep -r . /sys/class/backlight/intel_backlight/" report?
Before suspend: /sys/class/backlight/intel_backlight/type:raw /sys/class/backlight/intel_backlight/brightness:426 /sys/class/backlight/intel_backlight/power/control:auto /sys/class/backlight/intel_backlight/power/async:disabled /sys/class/backlight/intel_backlight/power/runtime_enabled:disabled /sys/class/backlight/intel_backlight/power/runtime_active_kids:0 /sys/class/backlight/intel_backlight/power/runtime_active_time:0 /sys/class/backlight/intel_backlight/power/runtime_status:unsupported /sys/class/backlight/intel_backlight/power/runtime_usage:0 /sys/class/backlight/intel_backlight/power/runtime_suspended_time:0 /sys/class/backlight/intel_backlight/bl_power:0 /sys/class/backlight/intel_backlight/max_brightness:851 /sys/class/backlight/intel_backlight/actual_brightness:426 After resume: sys/class/backlight/intel_backlight/type:raw /sys/class/backlight/intel_backlight/brightness:426 /sys/class/backlight/intel_backlight/power/control:auto /sys/class/backlight/intel_backlight/power/async:disabled /sys/class/backlight/intel_backlight/power/runtime_enabled:disabled /sys/class/backlight/intel_backlight/power/runtime_active_kids:0 /sys/class/backlight/intel_backlight/power/runtime_active_time:0 /sys/class/backlight/intel_backlight/power/runtime_status:unsupported /sys/class/backlight/intel_backlight/power/runtime_usage:0 /sys/class/backlight/intel_backlight/power/runtime_suspended_time:0 /sys/class/backlight/intel_backlight/bl_power:4 /sys/class/backlight/intel_backlight/max_brightness:851 /sys/class/backlight/intel_backlight/actual_brightness:426 It's purely Xorg by the way; LightDM revives the screen correctly after resume, but after logging in again, the screen is dimmed completely.
Here are my results from "grep -r . /sys/class/backlight/intel_backlight/": /sys/class/backlight/intel_backlight/type:raw /sys/class/backlight/intel_backlight/brightness:958 /sys/class/backlight/intel_backlight/power/control:auto /sys/class/backlight/intel_backlight/power/async:disabled /sys/class/backlight/intel_backlight/power/runtime_enabled:disabled /sys/class/backlight/intel_backlight/power/runtime_active_kids:0 /sys/class/backlight/intel_backlight/power/runtime_active_time:0 grep: /sys/class/backlight/intel_backlight/power/autosuspend_delay_ms: Eingabe-/Ausgabefehler /sys/class/backlight/intel_backlight/power/runtime_status:unsupported /sys/class/backlight/intel_backlight/power/runtime_usage:0 /sys/class/backlight/intel_backlight/power/runtime_suspended_time:0 /sys/class/backlight/intel_backlight/bl_power:0 /sys/class/backlight/intel_backlight/max_brightness:4794 /sys/class/backlight/intel_backlight/actual_brightness:958
And after resume: /sys/class/backlight/intel_backlight/type:raw /sys/class/backlight/intel_backlight/brightness:958 /sys/class/backlight/intel_backlight/power/control:auto /sys/class/backlight/intel_backlight/power/async:disabled /sys/class/backlight/intel_backlight/power/runtime_enabled:disabled /sys/class/backlight/intel_backlight/power/runtime_active_kids:0 /sys/class/backlight/intel_backlight/power/runtime_active_time:0 grep: /sys/class/backlight/intel_backlight/power/autosuspend_delay_ms: Eingabe-/Ausgabefehler /sys/class/backlight/intel_backlight/power/runtime_status:unsupported /sys/class/backlight/intel_backlight/power/runtime_usage:0 /sys/class/backlight/intel_backlight/power/runtime_suspended_time:0 /sys/class/backlight/intel_backlight/bl_power:4 /sys/class/backlight/intel_backlight/max_brightness:4794 /sys/class/backlight/intel_backlight/actual_brightness:958
In this case what you probably want to do is strace Xorg across the suspend and resume (and attach the log). Is your system using the xf86-video-intel-backlight-helper?
O.K., I brought my machine with "systemctl set-default multi-user.targe" to boot to console only. As a user I started "strace -o logfile.log startx" and the result is attached. Oddly, after resume the display stays very bright for a couple of seconds and goes to normal. But the backlight stays on! It seems to me (just guessing) that the display manager maybe is causing the issues because in this test >lightdm< isn't started. But this is much about my knowledge. THX.
Created attachment 119663 [details] Xorg strace - L540
Is "bl_power:4" of any concern?
Ok, if you strace startx, please pass -f to strace (i.e. strace -f -o x.strace startx)
Created attachment 119665 [details] strace -f -o x.strace startx I hope the new x.strace is useable but I needed to zip it, it's huge (37MB).
You have a non-root ddx and so the ddx is using the backlight helper. The backlight helper doesn't set the power sysfs value, and that is being left off across suspend.
I've looked up DDX and sounds like rocket science to me. Something else I can do to support troubleshooting?
Please add drm.debug=14 module parameter and attach dmesg from early boot to the suspend resume cycle reproducing the problem.
Created attachment 119700 [details] drm.debug=14, dmesg before resume, L540 drm.debug=14, dmesg before resume, L540
Created attachment 119701 [details] drm.debug=14, dmesg after resume, L540 drm.debug=14, dmesg after resume, L540
After resume: [ 63.147738] [drm:intel_edp_backlight_on] [ 63.147739] [drm:intel_panel_enable_backlight] pipe A [ 63.147742] [drm:intel_panel_actually_set_backlight] set backlight PWM = 1480 bl_power gets set to 0 at this point (that means "on"). And a bit later: [ 63.803210] [drm:intel_backlight_device_update_status] updating intel_backlight, brightness=1441/4794 [ 63.803213] [drm:intel_panel_actually_set_backlight] set backlight PWM = 1480 [ 63.803219] [drm:intel_edp_backlight_power] panel power control backlight disable This is a request from userspace to set bl_power to 4 (that means "off"). It's not a kernel issue.
bug report for lightdm: https://bugs.launchpad.net/lightdm/+bug/1516622
Ok, so there's a bug filed against LightDM; but why is it only present with current revisions of the Intel display driver and not with the old one?
Issue was fixed on Arch with version xf86-video-intel1:2.99.917+519+g8229390-0.1 of the video driver in combination with latest kernel 4.3.3
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.