[Bug 103819] New: [SKL] Black screen after suspend/resume with eDP

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=103819

            Bug ID: 103819
           Summary: [SKL] Black screen after suspend/resume with eDP
           Product: DRI
           Version: unspecified
          Hardware: x86-64 (AMD64)
                OS: Linux (All)
            Status: NEW
          Severity: normal
          Priority: medium
         Component: DRM/Intel
          Assignee: intel-gfx-bugs at lists.freedesktop.org
          Reporter: peter at lekensteyn.nl
        QA Contact: intel-gfx-bugs at lists.freedesktop.org
                CC: intel-gfx-bugs at lists.freedesktop.org

Created attachment 135592
  --> https://bugs.freedesktop.org/attachment.cgi?id=135592&action=edit
journal for Linux v4.14-2225-gb4da24717364

Bug description:
After a system sleep/resume (and sometimes after leaving the laptop idle,
allowing DPMS to kick in?), the screen remains black.

System environment:
-- chipset: SKL (HD Graphics 530)
-- system architecture: 64-bit
-- xf86-video-intel: 1:2.99.917+800+g37a682aa-1
-- xserver: 1.19.5-1
-- mesa: 17.2.5-1
-- libdrm: 2.4.88-1
-- kernel: 4.14-2 (and also 4.13.10 and drm-tip v4.14-2225-gb4da24717364)
-- Linux distribution: Arch Linux
-- Machine or mobo model: Clevo P651RA laptop
-- Display connector: eDP

Reproducing steps:
1. Suspend system (via sleep button or "systemctl suspend")
2. Resume system by pressing the power button

The issue occurs both from X and from the console (with no X running) and
occurs
both with and without nouveau.

Additional info:
The laptop has hybrid graphics ("Nvidia Optimus"), with Intel i7-6700HQ (using
i915) + Nvidia GTX 965M (using nouveau). It has a mux which can be configured
in
BIOS to "Hybrid" or "Discrete", external connectors are wired to the Nvidia
GPU.

I had a workaround for this black screen since at least 4.11.6-1-ARCH, invoking
this from "/usr/lib/systemd/system-sleep/fix-edp1-black-screen post suspend":

    echo detect > /sys/class/drm/card0-eDP-1/status

Before I automated that, I used to trigger this from a KDE Plasma shortcut:

    xset dpms force off; xset dpms force on

This worked until kernel 4.12.10, after an upgrade to 4.13.10 (and also 4.14),
the above methods fail. With 4.14, I observe the following:

 1. The initial boot works.
 2. The first/second s/r cycle may survive. While the backlight is turned on, a
    flicker (with no picture) is visible and it takes 1-3 (?) seconds before
the
    console (or SDDM lock screen with X) becomes visible.
 3. Further sessions that fail result in a black screen (no flickering is
    visible at all even if the backlight is on). "xrandr" shows no modes
    associated with eDP1 (but reports it as "connected").

It seems that the black screen occurs in the next s/r cycle when
/sys/class/drm/card0-eDP-1/modes is empty (with 4.14).

Attached is a full journal with current drm-tip with the following events:

 1. Boot to console (no Xorg), everything is OK
 2. Suspend+resume, still OK
 3. Suspend+resume, black screen. Attempts to "fix" this with
    "echo detect > /sys/class/drm/card0-eDP-1/status" have no effect.
 4. Suspend+resume, still black screen. Writing "detect" still does not work.
 5. Shutdown.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
You are the assignee for the bug.
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20171119/a94ffdf7/attachment-0001.html>