Bug 81371

Summary:	Dell Latitude Docker's display port gets switched on slowly
Product:	Drivers	Reporter:	LÉVAI Dániel (kernel.refutable485)
Component:	Video(DRI - Intel)	Assignee:	intel-gfx-bugs (intel-gfx-bugs)
Status:	CLOSED OBSOLETE
Severity:	normal	CC:	alan, intel-gfx-bugs, tianyu.lan
Priority:	P3
Hardware:	i386
OS:	Linux
Kernel Version:	3.15.7	Subsystem:
Regression:	No	Bisected commit-id:

Description LÉVAI Dániel 2014-07-30 08:23:08 UTC

Hi!

So here's another weird one...
With 3.14.x, I was struggling with my Latitude E5440's docker's display port output: it wouldn't switch on the connected monitor for at least a minute (after turning it on, or waking the laptop up from sleep).
Switching to 3.15 solved this, I was very happy: the monitor switching was snappy - almost instant - after waking the computer up from sleep.
The last good version for me was 3.15.6, and now with 3.15.7, the display port got slow again. Admittingly, not *as* slow as with 3.14, but much slower than 3.15.0 <= 3.15.6.
I don't even know if this is a power management issue, to be honest. Is this something you guys would be concerned with?

-- 
Daniel

Comment 1 LÉVAI Dániel 2014-08-01 11:53:34 UTC

It's the same with 3.15.8.

Comment 2 LÉVAI Dániel 2014-08-01 17:07:48 UTC

It's much more worse with 3.16-rc7. It goes back to 3.14-era, and never seems to turn on the displayport monitor.

Comment 3 Lan Tianyu 2014-08-04 02:36:52 UTC

Could you do a bisect between v3.15 and v3.16-rc6 to find which commit makes it worse? This seems not related with PM and so reassign to video component.

Comment 4 Lan Tianyu 2014-08-04 02:37:06 UTC

Could you do a bisect between v3.15 and v3.16-rc6 to find which commit makes it worse? This seems not related with PM and so reassign to video component.

Comment 5 LÉVAI Dániel 2014-08-04 09:12:22 UTC

Sure thing.
It seems that the offending commit was:

commit 1c46b5d7cd5d9912687a9e65626cdc24b1074ae1
Author: Dave Airlie <airlied@redhat.com>
Date:   Mon Jul 14 11:04:39 2014 +1000
    Revert "drm/i915: reverse dp link param selection, prefer fast over wide again"

If I checkout 3.15.8 and revert this one commit, then the external monitor detection/switching on the display port is fast again.


The original commit between 3.15-rc5 and rc6, which was revert by the above one, was:

commit f4cdbc21444a45d207a8dc175f44d2facfbd0845
Author: Jani Nikula <jani.nikula@intel.com>
Date:   Wed May 14 13:02:19 2014 +0300
    drm/i915/dp: force eDP lane count to max available lanes on BDW


-- 
Daniel

Comment 6 LÉVAI Dániel 2014-08-04 09:16:59 UTC

Oh, but I just realized that I read what I wanted to read :)
So, I did the bisect between 3.15.6 and 3.15.7, I never went to 3.16.
Is this okay anyways?


-- 
Daniel

Comment 7 LÉVAI Dániel 2014-08-05 15:47:55 UTC

Sorry, the original commit was not that, it was just referenced by that one... Anyway, I know it's not my job to tell you guys these, so I'll just shut up and wait for the fix (he-he) ;->


-- 
Daniel

Comment 8 LÉVAI Dániel 2014-08-10 08:08:48 UTC

Anything else I can help you guys with to fix this? Don't drop the ball...

Comment 9 Jani Nikula 2014-09-23 14:24:10 UTC

Please try drm-intel-nightly branch from http://cgit.freedesktop.org/drm-intel, and attach full dmesg from boot to the problem with drm.debug=0xe module parameter set.

Comment 10 LÉVAI Dániel 2014-10-10 11:31:41 UTC

Nevermind; I had stopped using this hardware as its support gets worse with every kernel release.