Public bug reported:
SUMMARY
-------
On a Lenovo ThinkPad T450s (Intel Broadwell HD Graphics 5500, AU Optronics eDP
panel), two apparently related bugs occur on Ubuntu 26.04 LTS with kernel
7.0.0-28-generic:
1. After system suspend, the machine wakes but the screen stays black, even
though the kernel itself resumes successfully (confirmed via journalctl: "PM:
suspend exit" and "System returned from sleep operation 'suspend'" are both
logged normally).
2. Screen brightness cannot be changed via hotkeys or the GNOME brightness
slider. Writes to /sys/class/backlight/intel_backlight/brightness are accepted
(the sysfs value updates) but have no visible effect on the panel.
Both issues disappeared after adding "i915.enable_dpcd_backlight=0" to
the kernel command line, which strongly suggests a shared root cause in
the i915 eDP backlight handling.
RELEASE / VERSION
------------------
Distro: Ubuntu 26.04 LTS (Resolute Raccoon)
Kernel: 7.0.0-28-generic
Desktop: GNOME Shell 50.1, Wayland session (GNOME 50 removed the X11 session
entirely, so testing under X11 for comparison was not possible)
Mesa: 26.0.3-1ubuntu1
HARDWARE
--------
Machine: Lenovo ThinkPad T450s (20BWS0090B)
GPU: Intel Broadwell-U GT2 [HD Graphics 5500], PCI ID 8086:1616, driver i915
Panel: eDP-1, AU Optronics 0x303e, 1600x900
EXPECTED BEHAVIOR
------------------
- Display is active immediately after resuming from suspend.
- Brightness hotkeys/slider visibly change screen brightness.
ACTUAL BEHAVIOR
----------------
- Display stays black after every resume from suspend, despite the kernel
completing the resume normally per logs.
- Brightness slider and hotkeys accept input (the sysfs value changes) but
panel brightness never visibly changes.
STEPS TO REPRODUCE
-------------------
1. Boot Ubuntu 26.04 LTS on the hardware above with default kernel parameters.
2. Suspend the system (systemctl suspend, lid close, or Suspend from the GNOME
menu).
3. Wake the system. Screen remains black despite the kernel completing the
resume.
4. Separately, try changing brightness via Fn keys or Settings > Display. No
visible change occurs, although
/sys/class/backlight/intel_backlight/actual_brightness reflects the requested
value.
RELEVANT LOG EXCERPT (journalctl -b -1, around suspend entry)
---------------------------------------------------------------
gnome-shell[3472]: Cursor update failed: drmModeAtomicCommit: Invalid argument
kernel: PM: suspend entry (deep)
kernel: i915 0000:00:02.0: [drm] CPU PWM1 enabled
kernel: WARNING: drivers/gpu/drm/i915/display/intel_display_power.c:1225 at
assert_can_disable_lcpll+0x741/0x760 [i915], CPU#0: kworker/u16:26/5166
kernel: RIP: 0010:assert_can_disable_lcpll+0x747/0x760 [i915]
kernel: hsw_disable_lcpll.constprop.0+0x18/0x430 [i915]
kernel: hsw_enable_pc8+0x46/0xc0 [i915]
kernel: intel_display_power_suspend_late+0x69/0xf0 [i915]
kernel: i915_drm_suspend_late+0x7e/0x100 [i915]
kernel: i915 0000:00:02.0: [drm] PCH PWM1 enabled
kernel: WARNING: drivers/gpu/drm/i915/display/intel_display_power.c:1232 at
assert_can_disable_lcpll+0x5e4/0x760 [i915], CPU#0: kworker/u16:26/5166
kernel: ACPI: PM: Waking up from system sleep state S3
kernel: PM: suspend exit
(Full journal excerpt attached separately.)
WORKAROUND
----------
Adding the following kernel parameter (in /etc/default/grub,
GRUB_CMDLINE_LINUX_DEFAULT) resolves both symptoms:
i915.enable_dpcd_backlight=0
After rebooting with this parameter:
- The assert_can_disable_lcpll warning no longer appears on suspend.
- The system reliably wakes with the display active over multiple
suspend/resume cycles.
- Brightness hotkeys and slider work correctly.
SUSPECTED ROOT CAUSE / POSSIBLE REGRESSION
--------------------------------------------
This warning signature (assert_can_disable_lcpll, with "CPU PWM1 enabled"
immediately before it, in the hsw_enable_pc8 / hsw_disable_lcpll path) matches
a bug reported and fixed upstream in 2021 on Broadwell hardware (fi-bdw-samus
test machine), where DPCD-based eDP backlight control could leave the panel PWM
enabled while the driver tried to power down the LCPLL. That was addressed by
commit fe7d52bccab6 ("drm/i915/dp: Don't use DPCD backlights that need PWM
enable/disable") and a follow-up fix around April 2021 (drm-tip commit
0d733c1e9408). Reference:
https://lore.kernel.org/lkml/20210204090229.GH17757@xsang-OptiPlex-9020/T/
Seeing the identical warning on kernel 7.0.0-28-generic in 2026 suggests either:
(a) a regression reintroducing this issue, or
(b) this specific AU Optronics panel (0x303e) is not covered by whatever
heuristic currently excludes affected panels from the DPCD backlight path.
Forcing i915.enable_dpcd_backlight=0 avoids the DPCD path entirely and
resolves both the suspend/black-screen issue and the non-functional
brightness control, supporting a shared root cause.
ADDITIONAL NOTES
-----------------
- Full kernel cmdline at time of testing:
BOOT_IMAGE=/boot/vmlinuz-7.0.0-28-generic
root=UUID=0a4365c1-be8e-4995-855b-909bf99c3e4c ro quiet splash
acpi_backlight=native i915.enable_psr=0 i915.enable_dc=0
intel_idle.max_cstate=1 i915.enable_dpcd_backlight=0
crashkernel=2G-4G:320M,4G-32G:512M,32G-64G:1024M,64G-128G:2048M,128G-:4096M
- Only one backlight interface is present: /sys/class/backlight/intel_backlight
(no competing ACPI/vendor interface).
- GNOME 50 removed the X11 session entirely, so it was not possible to confirm
whether the black-screen symptom is Wayland/Mutter-specific.
** Affects: linux (Ubuntu)
Importance: Undecided
Status: New
** Tags: backlight broadwell i915 resume suspend wayland
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2161826
Title:
i915 DPCD backlight bug on Broadwell (ThinkPad T450s): black screen
after suspend + broken brightness, fixed by
i915.enable_dpcd_backlight=0
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2161826/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
