Bug 71531

Summary:	eDP disabled forever if lid closed at boot time
Product:	Drivers	Reporter:	Matthieu Imbert (matthieu.imbert)
Component:	Video(DRI - Intel)	Assignee:	Daniel Vetter (daniel)
Status:	RESOLVED CODE_FIX
Severity:	normal	CC:	intel-gfx-bugs, marek.szuba
Priority:	P3
Hardware:	x86-64
OS:	Linux
Kernel Version:	3.12.1	Subsystem:
Regression:	No	Bisected commit-id:
Attachments:	kernel log with drm.debug=0xe, when booting with laptop screen closed kernel log with drm.debug=0xe, when booting with laptop screen open kernel log with drm.debug=0xe, when booting with laptop screen closed kernel log with drm.debug=0xe, when booting with laptop screen open output of intel_reg_dumper after boot with laptop screen closed output of intel_reg_dumper after boot with laptop screen open kernel log with drm.debug=0xe, when booting with laptop screen closed kernel log with drm.debug=0xe, when booting with laptop screen open stdout of intel_reg_dumper after boot with laptop screen closed stdout of intel_reg_dumper after boot with laptop screen open

Description Matthieu Imbert 2014-03-05 10:33:26 UTC

On a dell e4310 laptop with arrandale dual-core i5-540m integrated graphics, driver i915: If the laptop is booted with its lid closed, the eDP is not initialized at all and is lost forever.

dmesg says: [drm] failed to retrieve link info, disabling eDP

Booting with lid closed is a common situation for anyone using a docking station: A user puts the laptop closed on the docking station, boots (lid closed), using external screen. Later, the user needs to move and then undock the laptop, opens the lid and expects to have the eDP working (X + some desktop env correctly handle these situations).

This bug occurs with i915 started with option modeset=1 (KMS enabled).

I use the current default debian testing kernel. If needed I can recompile a specific kernel (possibly patched) for testing.

Feel free to ask for any needed details (logs, anything..)

Matthieu

Comment 1 Jani Nikula 2014-03-05 11:16:59 UTC

Unfortunately there still are cases where we fail to enable eDP if the BIOS didn't do that already. Perhaps we'll find a clue this time, fingers crossed.

Please attach dmesg with drm.debug=0xe module parameter set *and* the output of intel_reg_dumper (from intel-gpu-tools package) for both the two cases (booted with lid closed, booted with lid open).

Comment 2 Matthieu Imbert 2014-03-05 15:20:15 UTC

Created attachment 128171 [details]
kernel log with drm.debug=0xe, when booting with laptop screen closed

Comment 3 Matthieu Imbert 2014-03-05 15:20:47 UTC

Created attachment 128181 [details]
kernel log with drm.debug=0xe, when booting with laptop screen open

Comment 4 Matthieu Imbert 2014-03-05 15:21:43 UTC

In the two previous attachements I put the content of kernel.log rather than the output of dmesg because in the case when the laptop screen is open, drm.debug output is so verbose that dmesg buffer wraps and some part of the log is lost. dmesg and kernel.log are the same, right?

for intel_reg_dumper, I installed debian package intel-gpu-tools, but I get this error message: "Couldn't map MMIO region: Resource temporarily unavailable" (debian bug 715278 https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=715278). Should I try to build the latest from http://anonscm.debian.org/gitweb/?p=pkg-xorg/app/intel-gpu-tools.git ?

Comment 5 Matthieu Imbert 2014-03-05 20:37:35 UTC

Created attachment 128211 [details]
kernel log with drm.debug=0xe, when booting with laptop screen closed

Comment 6 Matthieu Imbert 2014-03-05 20:38:16 UTC

Created attachment 128221 [details]
kernel log with drm.debug=0xe, when booting with laptop screen open

Comment 7 Matthieu Imbert 2014-03-05 20:41:16 UTC

Created attachment 128231 [details]
output of intel_reg_dumper after boot with laptop screen closed

while running intel_reg_dumper, I got this on stderr (I don't know if it's important or not):

Couldn't find path to dri/debugfs entry
i915 loaded; not proceeding.

Comment 8 Matthieu Imbert 2014-03-05 20:42:57 UTC

Created attachment 128241 [details]
output of intel_reg_dumper after boot with laptop screen open

while running intel_reg_dumper, I got this on stderr (I don't know if it's important or not):

Couldn't find path to dri/debugfs entry
i915 loaded; not proceeding.

Comment 9 Matthieu Imbert 2014-03-05 20:45:47 UTC

FYI: I compiled intel_reg_dumper commit cf16da03b87267cfbb516f1745de218207093e1c from http://anonscm.debian.org/gitweb/?p=pkg-xorg/app/intel-gpu-tools.git to avoid the "Couldn't map MMIO region: Resource temporarily unavailable"

Comment 10 Daniel Vetter 2014-03-10 09:45:20 UTC

The debian package is ridiculously outdated. Please use the upstream git repo from

http://cgit.freedesktop.org/xorg/app/intel-gpu-tools/

Comment 11 Matthieu Imbert 2014-03-10 10:29:55 UTC

Ok. Do you need that I attach new dumps obtained with a version compiled from the upstream git repo or are the current ones enough for your needs?

Comment 12 Jani Nikula 2014-03-11 13:16:31 UTC

(In reply to Matthieu Imbert from comment #11)
> Ok. Do you need that I attach new dumps obtained with a version compiled
> from the upstream git repo or are the current ones enough for your needs?

Please attach new dumps.

Comment 13 Matthieu Imbert 2014-03-12 07:44:19 UTC

Created attachment 129111 [details]
kernel log with drm.debug=0xe, when booting with laptop screen closed

Comment 14 Matthieu Imbert 2014-03-12 07:45:31 UTC

Created attachment 129121 [details]
kernel log with drm.debug=0xe, when booting with laptop screen open

Comment 15 Matthieu Imbert 2014-03-12 07:48:06 UTC

Created attachment 129131 [details]
stdout of intel_reg_dumper after boot with laptop screen closed

stderr was:

Couldn't find path to dri/debugfs entry
warning: forcewake will not be handled

Comment 16 Matthieu Imbert 2014-03-12 07:49:20 UTC

Created attachment 129141 [details]
stdout of intel_reg_dumper after boot with laptop screen open

stderr was:

Couldn't find path to dri/debugfs entry
warning: forcewake will not be handled

Comment 17 Matthieu Imbert 2014-03-12 07:51:32 UTC

intel_reg_dumper version used: version compiled from HEAD c769d0988b7fae0e758046ba165c9c52c4f2816a from http://cgit.freedesktop.org/xorg/app/intel-gpu-tools/

Comment 18 Jani Nikula 2014-08-14 12:41:38 UTC

*** Bug 79991 has been marked as a duplicate of this bug. ***

Comment 19 Marek Szuba 2014-08-14 13:34:40 UTC

For the record, the problem no longer seems to present in kernel version 3.15.5. There is now a problem with incorrect detection of maximum resolution of the screen connected to the docking station (output DP2; xrandr doesn't show anything above 1280x1024) which I did not encounter with 3.14, that however is a subject for a separate bug report.

Comment 20 Jani Nikula 2014-09-12 09:56:49 UTC

(In reply to Marek Szuba from comment #19)
> For the record, the problem no longer seems to present in kernel version

Great, though would be nice to know what fixed it. Closing, please reopen or file a new bug at https://bugs.freedesktop.org/enter_bug.cgi?product=DRI on DRM/Intel if the problem persists. Thanks for the report.