[Nouveau] Blank Screen on Mac Book Retina

[Harald Jung]

Hi,

we have a blank screen on a mac book retina since kernel 3.16.
The behavour is a bit weird:
In kernel 3.16 and higher it is like this, when you reset all display 
and xrandr settings, sometimes the screen comes up in the highest 
possible resolution.
But when you try to choose another resolution the screen stays turns black.
But in 80% of the tries it stays black, when the saved screen 
resolutions is another than the highest resolution it always stays black.

As imirkin mentioned on irc, the problem is caused by the "nouveau E[   
PDISP][0000:01:00.0] 00:0006:0f44: link training failed ".
The last working kernel was 3.15.10.

I've tried to bisect it, but i wasn't able to have the notebook for the 
whole time.
And so we tried to do the rest of the bisects remote, but i think we've 
messed something up in the coordination
I have no chance to try a bisect again (not when i start with 3.15.10 
good and 3.16 bad), it takes to much time and i don't have access to the 
notebook for so long.
If we really need a bisect, maybe you could help me with a better 
starting and ending version to try

Logfiles and stuff can be found at the end of the mail.

best regards
Harald

Hardware Info can be found at: http://www.hjit.de/macretina.txt

Here is the bisect log:
Bisecting: a merge base must be tested
[1860e379875dfe7271c649058aeddffe5afd9d0d] Linux 3.15
Bisecting: 6940 revisions left to test after this (roughly 13 steps)
[7b215de3d0abbc4f6daf2efd19e8809af0564490] Merge branch 'i2c/for-next' 
of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux into next
Bisecting: 3892 revisions left to test after this (roughly 12 steps)
[0430e49b6e7c6b5e076be8fefdee089958c9adad] ima: introduce ima_kernel_read()
Bisecting: 1567 revisions left to test after this (roughly 11 steps)
[2937f5efa5754754daf46de745f67350f7f06ec2] Merge branch 'for_linus' of 
git://cavan.codon.org.uk/platform-drivers-x86
Bisecting: 785 revisions left to test after this (roughly 10 steps)
[bc1dfff04a5d4064ba0db1fab13f84ab4f333d2b] Merge branch 
'drm-nouveau-next' of 
git://anongit.freedesktop.org/git/nouveau/linux-2.6 into drm-next
Bisecting: 402 revisions left to test after this (roughly 9 steps)
[4fa62c890cea83f28c30e1d5dc8fc86f61210280] drm/i915: Move buffer pinning 
and ring selection to intel_crtc_page_flip()
Bisecting: 189 revisions left to test after this (roughly 8 steps)
[aeab0b5af7df88284d101abf8d121f0e913b81ff] drm/i915: disable runtime PM 
if RC6 is disabled
Bisecting: 94 revisions left to test after this (roughly 7 steps)
[9d0a6fa6c5e618bd978d625a215dc4a240ba3b3c] drm/i915: add render state 
initialization
Bisecting: 47 revisions left to test after this (roughly 6 steps)
[b3a3f03d7b1cfecf055e35289371f42a401fdd94] drm/i915: Fix ILK GPU reset 
domain bits
Bisecting: 23 revisions left to test after this (roughly 5 steps)
[c0c353299c48b19c95de1e7bda494c0c71248c0c] drm/i915/chv: Fix gmbus for 
port D
Bisecting: 11 revisions left to test after this (roughly 4 steps)
[1966e59ec1075597eff4d7feb3d11536a242d0eb] drm/i915/chv: Use RMW to 
toggle swing calc init
Bisecting: 5 revisions left to test after this (roughly 3 steps)
[525997e00bd32c3e25d5f77c26703f76c498f424] drm/i915: shuffle panel code
Bisecting: 2 revisions left to test after this (roughly 2 steps)
[75f7f3ec600524c9544cc31695155f1a9ddbe1d9] drm/i915: Fix mmio vs. CS 
flip race on ILK+
Bisecting: 0 revisions left to test after this (roughly 1 step)
[6304cd91e7f05f8802ea6f91287cac09741d9c46] drm/i915: Drop the excessive 
vblank waits from modeset codepaths
4fa62c890cea83f28c30e1d5dc8fc86f61210280 is the first bad commit
commit 4fa62c890cea83f28c30e1d5dc8fc86f61210280
Author: Ville Syrjälä <ville.syrjala at linux.intel.com>
Date:   Tue Apr 15 21:41:38 2014 +0300

     drm/i915: Move buffer pinning and ring selection to 
intel_crtc_page_flip()

     All of the .queue_flip() callbacks duplicate the same code to pin the
     buffers and calculate the gtt_offset. Move that code to
     intel_crtc_page_flip(). In order to do that we must also move the ring
     selection logic there.

     Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
     Reviewed-by: Chris Wilson <chris at chris-wilson.co.uk>
     Signed-off-by: Daniel Vetter <daniel.vetter at ffwll.ch>

:040000 040000 829ebf0fa1e7b31c1623aa498efc92bca429a483 
511140874cd4759c4eadef87ee9968808efa5f2a M      drivers