SRU for Q - https://lists.ubuntu.com/archives/kernel-
team/2026-March/166266.html

** Description changed:

  [Impact]
  HP and Dell systems with Intel Panther Lake-H (PTL-H) CPU and eDP panel 
configured for Panel Replay suffer a permanent black screen after logout from 
the desktop. Only a hard reboot restores the display.
  
  [Fix]
  Remove the two erroneous register writes from intel_alpm_disable() so that 
only the actual
  enable bits (ALPM_CTL_ALPM_ENABLE and ALPM_CTL_LOBF_ENABLE) are cleared on 
disable.
  PORT_ALPM_CTL is no longer touched during the disable path, and 
ALPM_CTL_ALPM_AUX_LESS_ENABLE
  is left intact for the subsequent intel_alpm_configure() call during link 
training.
  The fix is in linux-next as of 2026-02-27:
    008304c9ae75c772d3460040de56e12112cdf5e6  drm/i915/alpm: ALPM disable fixes
  In mainline b7.0-rc2
-   eb4a7139e9737 drm/i915/alpm: ALPM disable fixes
+   eb4a7139e9737 drm/i915/alpm: ALPM disable fixes
  
  Patchwork: https://patchwork.freedesktop.org/patch/704253/
  
  [Test Plan]
  Requires Intel PTL-H system with a Panel Replay + DSC eDP panel (DISPLAY_VER 
= 20).
  Click logout button from the top-right menu and then the GDM shows up.
  
  [Where problems could occur]
- It may break Intel i915/xe display driver Panel Replay and PSR functionality 
on PTL-class hardware (DISPLAY_VER >= 20).
- If the fix incorrectly omits a necessary cleanup step in 
intel_alpm_disable(), the ALPM hardware state might not be properly quiesced 
before subsequent link training, potentially causing Panel Replay or PSR2 to 
fail to enable at all (black screen at boot or first modeset rather than only 
on resume).  It could also manifest as display corruption, link training 
failures, or AUX communication errors logged in dmesg on affected hardware.
- The change is confined to the DISPLAY_VER >= 20 code path; older hardware is 
not affected.
+ The change modifies the ALPM disable path in the Intel i915 display driver, 
which
+ is exercised whenever PSR2 or Panel Replay is active on eDP panels.
+ If the removal of the PORT_ALPM_CTL write is incorrect (i.e., if some hardware
+ requires it to be cleared at disable time), ALPM may fail to disable properly,
+ leaving the link in a low-power state that prevents display from recovering. 
This
+ would manifest as a black screen or display not coming back after 
suspend/resume.
+ Similarly, if removing the ALPM_CTL_ALPM_AUX_LESS_ENABLE clear has unintended
+ side effects on some panel configurations, the ALPM mode (AUX-Less vs 
AUX-Wake)
+ could be stuck in the wrong state, causing link training failures or display
+ flicker on affected eDP panels.
+ The fix is reviewed by Intel display maintainers (Reviewed-by: Michał 
Grzelak),
+ reducing regression risk, but panels with non-standard ALPM implementations 
may
+ behave differently.
  
  [Other Info]
  Remaining open issue: a ~20-second GDM login delay after logout (unrelated to 
ALPM) is
  tracked upstream at 
https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/7497 and is
  outside the scope of this SRU.

** Description changed:

  [Impact]
  HP and Dell systems with Intel Panther Lake-H (PTL-H) CPU and eDP panel 
configured for Panel Replay suffer a permanent black screen after logout from 
the desktop. Only a hard reboot restores the display.
  
  [Fix]
- Remove the two erroneous register writes from intel_alpm_disable() so that 
only the actual
- enable bits (ALPM_CTL_ALPM_ENABLE and ALPM_CTL_LOBF_ENABLE) are cleared on 
disable.
- PORT_ALPM_CTL is no longer touched during the disable path, and 
ALPM_CTL_ALPM_AUX_LESS_ENABLE
- is left intact for the subsequent intel_alpm_configure() call during link 
training.
+ Fix the intel_alpm_disable() function in 
drivers/gpu/drm/i915/display/intel_alpm.c:
+ 1. Remove the PORT_ALPM_CTL write from intel_alpm_disable() — this register
+    must only be written before link training, not during disable.
+ 2. Stop clearing ALPM_CTL_ALPM_AUX_LESS_ENABLE in intel_alpm_disable() — this
+    bit is about mode switching (AUX-Less vs AUX-Wake), not ALPM 
enable/disable.
+ 
  The fix is in linux-next as of 2026-02-27:
    008304c9ae75c772d3460040de56e12112cdf5e6  drm/i915/alpm: ALPM disable fixes
  In mainline b7.0-rc2
    eb4a7139e9737 drm/i915/alpm: ALPM disable fixes
  
  Patchwork: https://patchwork.freedesktop.org/patch/704253/
  
  [Test Plan]
- Requires Intel PTL-H system with a Panel Replay + DSC eDP panel (DISPLAY_VER 
= 20).
+ Requires a system with Intel integrated graphics and an eDP panel that 
supports
+ Panel Replay (typically recent Intel platforms with Panel Replay-capable eDP 
displays).
+ 
  Click logout button from the top-right menu and then the GDM shows up.
  
  [Where problems could occur]
  The change modifies the ALPM disable path in the Intel i915 display driver, 
which
  is exercised whenever PSR2 or Panel Replay is active on eDP panels.
  If the removal of the PORT_ALPM_CTL write is incorrect (i.e., if some hardware
  requires it to be cleared at disable time), ALPM may fail to disable properly,
  leaving the link in a low-power state that prevents display from recovering. 
This
  would manifest as a black screen or display not coming back after 
suspend/resume.
  Similarly, if removing the ALPM_CTL_ALPM_AUX_LESS_ENABLE clear has unintended
  side effects on some panel configurations, the ALPM mode (AUX-Less vs 
AUX-Wake)
  could be stuck in the wrong state, causing link training failures or display
  flicker on affected eDP panels.
  The fix is reviewed by Intel display maintainers (Reviewed-by: Michał 
Grzelak),
  reducing regression risk, but panels with non-standard ALPM implementations 
may
  behave differently.
  
  [Other Info]
  Remaining open issue: a ~20-second GDM login delay after logout (unrelated to 
ALPM) is
  tracked upstream at 
https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/7497 and is
  outside the scope of this SRU.

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2143100

Title:
  Got black screen after clicked logout button

To manage notifications about this bug go to:
https://bugs.launchpad.net/hwe-next/+bug/2143100/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to