Bug 104451

Summary: [SKL] Black screen on 4.15-rc6 (DP link training error) [regression]
Product: DRI Reporter: Direx <direx>
Component: DRM/IntelAssignee: Intel GFX Bugs mailing list <intel-gfx-bugs>
Status: CLOSED FIXED QA Contact: Intel GFX Bugs mailing list <intel-gfx-bugs>
Severity: major    
Priority: high CC: intel-gfx-bugs
Version: DRI gitKeywords: bisect_pending
Hardware: x86-64 (AMD64)   
OS: Linux (All)   
Whiteboard:
i915 platform: SKL i915 features: display/DP, display/DP MST
Attachments:
Description Flags
DP link training errors in dmesg with drm.debug=0x1e
none
Full kernel debug log (with DMC 1.27) none

Description Direx 2018-01-02 08:48:49 UTC
Created attachment 136486 [details]
DP link training errors in dmesg with drm.debug=0x1e

I've just booted 4.15-rc6 and I am getting a black external screen (Thinkpad L460 with Ultra Dock and external DP monitor). Stable Kernels (4.14 and older) work just fine.

The dmesg is flooded with link training errors, so I was unable to capture a full system log through SSH. The attached dmesg lines are the last link training messages with drm.debug=0x1e. Thousands of these messages appear every second.

The messages appear right when modesetting kicks in, before Wayland is starting.

This is a major issue, as this is a regression which cause DP displays to fail.


My Configuration:

- Thinkpad L460 with Ultra Dock (contains a DP MST hub)
- External DP monitor connected to Ultra Dock
- Arch Linux
- Kernel 4.15-rc6
- Intel Core i5-6200U
Comment 1 Direx 2018-01-02 08:55:03 UTC
FYI:

The issue is about DP-3 (DP2-1 in X11). eDP1 is working.
Comment 2 Direx 2018-01-03 09:08:22 UTC
Created attachment 136515 [details]
Full kernel debug log (with DMC 1.27)

I managed to get a full debug log, which should be more helpful.

Meanwhile I have also applied https://patchwork.freedesktop.org/patch/187559/ (the log is with new DMC 1.27), but it does not change anything. But in case a log without that patch is required (DMC 1.26) let me know.
Comment 3 Elizabeth 2018-01-03 21:06:20 UTC
Hello Direx, is it possible for you to bisect to find the culprit commit?? Thank you.
Comment 4 Direx 2018-01-04 12:22:57 UTC
Sorry, I am neither a kernel developer nor a git expert, just an enthusiast end user. I've tried fiddling with git, but after all the rebasing in drm-intel I have difficulties bisecting.

All I can say is that 069d40f5834ad26a58f269225a7e13af17019062 is a good commit and 5ea2355a100a3c6304901d058aee06d3a6be69bc is a bad commit.
Comment 5 Elizabeth 2018-01-04 22:49:06 UTC
As reference:
good:
commit 069d40f5834ad26a58f269225a7e13af17019062
Author: Imre Deak <imre.deak@intel.com>
Date:   Tue Oct 3 12:51:59 2017 +0300

    drm/i915/glk: Fix DMC/DC state idleness calculation
bad:
commit 5ea2355a100a3c6304901d058aee06d3a6be69bc
Author: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Date:   Tue Oct 3 17:22:11 2017 +0300

    drm/i915/mst: Use MST sideband message transactions for dpms control
Comment 6 Direx 2018-01-05 10:32:07 UTC
This has been fixed today by:

commit 30414f3010aff95ffdb6bed7b9dce62cde94fdc7
Author: Lucas De Marchi <lucas.demarchi@intel.com>
Date:   Tue Jan 2 12:18:37 2018 -0800

    drm/i915: Apply Display WA #1183 on skl, kbl, and cfl


Maybe you want to add a test for this, so I am leaving the bug open.
Comment 7 Jani Nikula 2018-01-19 14:51:57 UTC
Thanks for the follow-up, closing.

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.