linux.git/drivers/gpu, branch v3.4-rc3

Merge branch 'exynos-drm-fixes' of git://git.infradead.org/users/kmpark/linux-samsung into drm-intel-fixes

2012-04-12T16:42:01+00:00

* 'exynos-drm-fixes' of git://git.infradead.org/users/kmpark/linux-samsung:
  drm/exynos: fixed exynos broken ioctl
  drm/exynos: fix to pointer manager member of struct exynos_drm_subdrv
  drm/exynos: fix struct for operation callback functions to driver name
  drm/exynos: use define instead of default_win member in struct mixer_context
  drm/exynos: rename s/HDMI_OVERLAY_NUMBER/MIXER_WIN_NR
  drm/exynos: remove unused codes in hdmi and mixer
  drm/exynos: remove unnecessary type conversion of hdmi and mixer
  drm/exynos: add format list of plane
  drm/exynos: fixed duplicated page allocation bug.
  drm/exynos: fixed page align and code clean.

Merge branch 'drm-intel-fixes' of git://people.freedesktop.org/~danvet/drm-intel into drm-intel-fixes

2012-04-12T16:41:10+00:00

Daniel Vetter writes:
3 regression fixes:
- disable gmbus again, too broken for 3.4, we'll try again for 3.5
- dp bandwidth computation fix, we've lost the 6bpc dithering flag
 sometimes, this is a 3.3 regression (maybe even earlier for some
 configurations).
- fix resume regression caused by the gen2/3 fencing fix merged into -rc2.

And a few other fixes:
- gpu hang fix for i845 (Chris)
- sprite fix (Armin Reese)
- crtc disable vs. scanlinewait race fix (Chris)
- rc6 module option read-only, it confused testers (Jesse)
- fbc related blitter death hw workaround, note that we disable fbc on snb
 by default anyway.

With these fixes we have one 3.4 regression outstanding: One of the
cleanup patches for the interlaced support managed to confuse the lvds
panel fitter when upscaling. The root-cause is still unclear, but test
patches are awaiting feedback from the reporter.

* 'drm-intel-fixes' of git://people.freedesktop.org/~danvet/drm-intel:
  drm/i915: clear fencing tracking state when retiring requests
  drm/i915: make rc6 module parameter read-only
  drm/i915: implement ColorBlt w/a
  drm/i915/ringbuffer: Exclude last 2 cachlines of ring on 845g
  Revert "drm/i915: reenable gmbus on gen3+ again"
  drm/i915: properly compute dp dithering for user-created modes
  drm/i915: Finish any pending operations on the framebuffer before disabling
  drm/i915: Removed IVB forced enable of sprite dest key.

drm/i915: clear fencing tracking state when retiring requests

2012-04-12T07:02:37+00:00

This fixes a resume regression introduced in

commit 7dd4906586274f3945f2aeaaa5a33b451c3b4bba
Author: Chris Wilson 
Date:   Wed Mar 21 10:48:18 2012 +0000

    drm/i915: Mark untiled BLT commands as fenced on gen2/3

which fixed fencing tracking for untiled blt commands.

A side effect of that patch was that now also untiled objects have a
non-zero obj->last_fenced_seqno to track when a fence can be set up
after a pipelined tiling change. Unfortunately this was only cleared
by the fence setup and teardown code, resulting in tons of untiled but
inactive objects with non-zero last_fenced_seqno.

Now after resume we completely reset the seqno tracking, both on the
driver side (by setting dev_priv->next_seqno = 1) and on the hw side
(by allocating a new hws page, which contains the seqnos). Hilarity
and indefinite waits ensued from the stale seqnos in
obj->last_fenced_seqno from before the suspend.

The fix is to properly clear the fencing tracking state like we
already do for the normal gpu rendering while moving objects off the
active list.

Reported-and-tested-by: "Rafael J. Wysocki" 
Cc: Jiri Slaby 
Reviewed-by: Chris Wilson 
Signed-Off-by: Daniel Vetter

drm/exynos: fix to pointer manager member of struct exynos_drm_subdrv

2012-04-12T05:51:27+00:00

The struct exynos_drm_manager has to exist for exynos drm sub driver
using encoder and connector. If it isn't NULL to member of struct
exynos_drm_subdrv, will create encoder and connector else will not. And
the is_local member also doesn't need.

Signed-off-by: Joonyoung Shim 
Signed-off-by: Kyungmin Park 
Signed-off-by: Inki Dae

drm/exynos: fix struct for operation callback functions to driver name

2012-04-12T05:51:26+00:00

The mixer driver and hdmi driver have each operation callback functions
and they is registered to hdmi common driver. Their struct names in hdmi
common driver include display, manager and overlay. It confuses to
appear whose operation and two driver cannot register same operation
callback functions at the same time. Use their struct names to driver
name.

Signed-off-by: Joonyoung Shim 
Signed-off-by: Kyungmin Park 
Signed-off-by: Inki Dae

drm/exynos: use define instead of default_win member in struct mixer_context

2012-04-12T05:51:24+00:00

The default_win member in struct mixer_context isn't change its value
after initialized to 0, so it's better using to define.

Signed-off-by: Joonyoung Shim 
Signed-off-by: Kyungmin Park 
Signed-off-by: Inki Dae

drm/exynos: rename s/HDMI_OVERLAY_NUMBER/MIXER_WIN_NR

2012-04-12T05:51:21+00:00

HDMI_OVERLAY_NUMBER is specific of mixer driver and be used "windows
layer" term in exynos user manaual, so rename it.

Signed-off-by: Joonyoung Shim 
Signed-off-by: Kyungmin Park 
Signed-off-by: Inki Dae

drm/exynos: remove unused codes in hdmi and mixer

2012-04-12T05:51:18+00:00

Some members in struct mixer_context aren't used and the define
HDMI_OVERLAY_NUMBER is unused in hdmi driver, remove them.

Signed-off-by: Joonyoung Shim 
Signed-off-by: Kyungmin Park 
Signed-off-by: Inki Dae

drm/exynos: remove unnecessary type conversion of hdmi and mixer

2012-04-12T05:51:16+00:00

When the void pointer type variable is assigned to the specific pointer
type variable, don't need to do type conversion.

Signed-off-by: Joonyoung Shim 
Signed-off-by: Kyungmin Park 
Signed-off-by: Inki Dae

drm/i915: make rc6 module parameter read-only

2012-04-11T19:36:38+00:00

People have been getting confused and thinking this is a runtime control.

Cc: stable@vger.kernel.org
Signed-off-by: Jesse Barnes 
Signed-off-by: Daniel Vetter