|Summary:||[chv][regression][bisected] Brightness adjustment on Lenovo Miix 320-10ICR and Lenovo Miix2 8 is broken since Linux 5.3rc1|
|Component:||DRM/Intel||Assignee:||Intel GFX Bugs mailing list <intel-gfx-bugs>|
|Status:||RESOLVED FIXED||QA Contact:||Intel GFX Bugs mailing list <intel-gfx-bugs>|
|Priority:||medium||CC:||chris, intel-gfx-bugs, thierry.reding|
|i915 platform:||BSW/CHT||i915 features:||display/backlight|
Description russianneuromancer 2019-08-03 11:02:30 UTC
Created attachment 144941 [details] drm-tip dmesg with drm.debug=0x1e from Lenovo Miix 320-10ICR Hello! I find that brightness adjustment is broken on Lenovo Miix 320-10ICR since Linux 5.3rc1, git bisect lead to this commit: https://github.com/torvalds/linux/commit/4a6ef8e37c4d9a40f09438068da1734fd965bd75 drm-tip 2019-07-30 4040b4c4ab647422d82100c8b091d34b6a82f572 log with drm.debug=0x1e log_buf_len=8M where I trying to adjust brightness is attached. Brightness adjustment works fine on Linux 5.0-5.2 (earlies releases wasn't tested).
Comment 1 Chris Wilson 2019-08-03 11:40:11 UTC
You should definitely mention what interface you are using to change the backlight.
Comment 2 russianneuromancer 2019-08-03 13:13:45 UTC
On Linux 5.0-5.2 I adjusted backlight via Gnome Shell slider in upper right corner. On Linux 5.3rc1+ this is no longer works, according to git bisect this is due to mentioned commit. That all I know right now. If I can provide more data from working/mom-working kernel - please let me know how to do so.
Comment 3 Lakshmi 2019-08-06 08:11:39 UTC
[7.984343] [drm:intel_panel_vbt_fixed_mode [i915]] [CONNECTOR:105:DSI-1] using mode from VBT: [ 7.984387] [drm:drm_mode_debug_printmodeline [drm]] Modeline "1200x1920": 0 162980 1200 1280 1284 1344 1920 1980 1981 2021 0x8 0xa [ 7.984494] [drm:pwm_setup_backlight [i915]] *ERROR* Failed to own the pwm chip [ 7.984600] [drm:intel_panel_setup_backlight [i915]] failed to setup backlight for connector DSI-1 [ 7.984698] [drm:intel_dsi_get_hw_state [i915]] [ 7.984922] [drm:intel_modeset_readout_hw_state [i915]] [CRTC:47:pipe A] hw state readout: enabled @Chris, any suggestion here? Does this error give any clue?
Comment 4 Chris Wilson 2019-08-06 08:14:50 UTC
Two drivers for one bit of backlight HW, doomed to fail. If the acpi driver exposed a /sys/class/backlight, xf86-video-intel should have still found it.
Comment 5 russianneuromancer 2019-08-06 16:52:11 UTC
Created attachment 144959 [details] drm-tip dmesg with drm.debug=0x1e from Lenovo Miix2 8 As I just found same issue is reproducible on Lenovo Miix2 8 too. Attached log is from drm-tip 2019-08-05 (2bb7df95d4e058e61d9de39f1b97404592708d24) > If the acpi driver exposed a /sys/class/backlight, xf86-video-intel should have still found it. Lenovo Miix2 8, Lenovo Miix 320-10IRC On Linux 5.2 there is just one intel_backlight, adjusting brightness via intel_backlight works. Lenovo Miix2 8: On Linux 5.3 there is acpi_video0 and acpi_video1, echo values to brightness file in each one doesn't change backlight level. Lenovo Miix 320-10IRC: On Linux 5.3 there is... ~# ls /sys/class/backlight acpi_video0 acpi_video1 acpi_video2 acpi_video3 acpi_video4 acpi_video5 acpi_video6 acpi_video7 As with Lenovo Miix2 8, echo values to brightness file in each acpi_video doesn't change backlight level.
Comment 6 russianneuromancer 2019-08-07 14:48:46 UTC
For comparison below is logs from Linux 5.2.7 with drm.debug=0x1e and working backlight adjustment:
Comment 7 russianneuromancer 2019-08-07 14:49:13 UTC
Created attachment 144972 [details] Linux 5.2.7 dmesg with drm.debug=0x1e from Lenovo Miix 320-10ICR
Comment 8 russianneuromancer 2019-08-07 14:49:24 UTC
Created attachment 144973 [details] Linux 5.2.7 dmesg with drm.debug=0x1e from Lenovo Miix2 8
Comment 9 Jani Nikula 2019-08-08 07:30:42 UTC
Cc Thierry, committer of the regressing commit commit 4a6ef8e37c4d9a40f09438068da1734fd965bd75 Author: Nikolaus Voss <firstname.lastname@example.org> Date: Wed Jun 12 10:36:07 2019 +0200 pwm: Add support referencing PWMs from ACPI Any thoughts?
Comment 10 Jani Nikula 2019-08-09 08:07:15 UTC
Per Thierry, the fix is , on its way to v5.3-rc4. Thanks for the report, closing. Please reopen if the problem persists with that commit.  https://kernel.googlesource.com/pub/scm/linux/kernel/git/thierry.reding/linux-pwm/+/6cf9481b440da6d6d86bd8e4c99a8b553b9d1271
Comment 11 russianneuromancer 2019-08-19 15:00:01 UTC
Issue is no longer reproducible with Linux 5.3rc4.