Bug 100444

Summary:	VT switch broken after DP-MST docking
Product:	DRI	Reporter:	Takashi Iwai <tiwai>
Component:	DRM/Intel	Assignee:	Intel GFX Bugs mailing list <intel-gfx-bugs>
Status:	CLOSED FIXED	QA Contact:	Intel GFX Bugs mailing list <intel-gfx-bugs>
Severity:	normal
Priority:	medium	CC:	intel-gfx-bugs
Version:	XOrg git
Hardware:	Other
OS:	All
Whiteboard:
i915 platform:		i915 features:	display/DP MST

Description Takashi Iwai 2017-03-29 09:50:41 UTC

Originally reported to openSUSE bugzilla:
  https://bugzilla.opensuse.org/show_bug.cgi?id=1031462

The i915 in 4.11-rc kernel seems to have a regression regarding VT switch.
When the machine is booted with an external monitor (HDMI or DP), the mode is set on both eDP and HDMI/DP, then switching to VT1 from X leads to the frozen screen.

It's not really a crash but likely the atomic modeset failure like bsc#1018358, as you can go back to X via alt-F7 again.

The 4.10 kernel works fine, so it's a regression.  The problem is reproduced on Dell E7270 laptop with Skylake chip.

Comment 1 Takashi Iwai 2017-03-29 09:51:06 UTC

The bisect leads to the commit:

commit 3c6b29b2df12fe5783b17fbf73bc2d53e385cdbd
Author: Paulo Zanoni <paulo.r.zanoni@intel.com>
Date:   Thu Dec 15 11:23:55 2016 -0200

    drm/i915: fully apply WaSkipStolenMemoryFirstPage

Unfortunately, it's not trivial to revert this on top of 4.11-rc.

Comment 2 Chris Wilson 2017-03-30 12:21:30 UTC

Sounds like commit 12ffed96d4369f086261ba2ee734fa8c932d7f55
Author: Michel Dänzer <michel.daenzer@amd.com>
Date:   Thu Mar 23 17:53:26 2017 +0900

    drm/fb-helper: Allow var->x/yres(_virtual) < fb->width/height again

should fix this?

Comment 3 Takashi Iwai 2017-03-30 13:08:22 UTC

Indeed, this VT switch bug seems solved in 4.11-rc4.  Thanks for checking!

While we're at it: there is still another VT switch bug showing the similar symptom.

Boot with a docking station with DP-MST monitor.  Then undock, and go to VT1.  Now  no VT switch occurs.  The machine is the same one, Dell Latitude E7270.

IIRC, this problem is present for long time, and no regression in 4.11-rc.  Rather it's been since the atomic mode setting.

Comment 4 Jari Tahvanainen 2017-04-03 10:00:21 UTC

Hello Takashi, please open another bug related to the symptoms described on comment 3. And if original failure is fixed, like you said, then please mark this bug resolved+fixed.

Comment 5 yann 2017-04-11 14:04:48 UTC

(In reply to Takashi Iwai from comment #3)
> Indeed, this VT switch bug seems solved in 4.11-rc4.  Thanks for checking!
> 
> While we're at it: there is still another VT switch bug showing the similar
> symptom.
> 
> Boot with a docking station with DP-MST monitor.  Then undock, and go to
> VT1.  Now  no VT switch occurs.  The machine is the same one, Dell Latitude
> E7270.
> 
> IIRC, this problem is present for long time, and no regression in 4.11-rc. 
> Rather it's been since the atomic mode setting.

Takashi Iwai, likes Jari said, please open a new bug for this defect. I am closing now this one as fixed to avoid confusion with the new one.

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.