This decreases eviction by up to 20%, by improving the fragmentation
quality. No harm in normal cases that fit VRAM fully (PTS gaming suite).

In some cases, even the VRAM-fitting cases improved slightly (openarena, urban terror).

512kb was measured as the most optimal threshold for 3d workloads common to radeon.
Other drivers may need different thresholds according to their workloads.

v2: Nicer formatting
Signed-off-by: Lauri Kasanen <cand@gmx.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Clients like i915 need to segregate cache domains within the GTT which
can lead to small amounts of fragmentation. By allocating the uncached
buffers from the bottom and the cacheable buffers from the top, we can
reduce the amount of wasted space and also optimize allocation of the
mappable portion of the GTT to only those buffers that require CPU
access through the GTT.

For other drivers, allocating small bos from one end and large ones
from the other helps improve the quality of fragmentation.

Based on drm_mm work by Chris Wilson.

v3: Changed to use a TTM placement flag
v2: Updated kerneldoc

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <ben@bwidawsk.net>
Cc: Christian König <deathsimple@vodafone.de>
Signed-off-by: Lauri Kasanen <cand@gmx.com>
Signed-off-by: David Airlie <airlied@redhat.com>

* 'msm-next' of git://people.freedesktop.org/~robclark/linux:
  drm/omap: Don't dereference list head when the connectors list is empty
  drm/msm/mdp: add timeout for irq wait
  drm/msm: validate flags, etc
  drm/msm: use componentised device support
  drm/msm: add chip-id param
  drm/msm: crank down gpu when inactive
  drm/msm: spin helper
  drm/msm: add hang_debug module param
  drm/msm: hdmi audio support

The rcar_du_encoder_init() function can fail and return an error code.
Don't ignore it.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

- Rework of finding the right PLL numbers for display
- Couple of different bugfixes

* 'drm-next-3.15' of git://people.freedesktop.org/~deathsimple/linux:
  drm/radeon: fix typo in spectre_golden_registers
  drm/radeon: fix endian swap on hawaii clear state buffer setup
  drm/radeon: call drm_edid_to_eld when we update the edid
  drm/radeon: rework finding display PLL numbers v2
  drm/radeon: fix resuming mode in pm runtime resume path
  drm/radeon: fix runtime suspend breaking secondary GPUs
  drm/radeon: clear needs_reset flag if IB test fails

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org

Need to swap on BE.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org

This needs to be done to update some of the fields in
the connector structure used by the audio code.

Noticed by several users on irc.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org

This completely reworks how the PLL parameters are generated and
should result in better matching dot clock frequencies.

Probably needs quite a bit of testing.

bugs: https://bugs.freedesktop.org/show_bug.cgi?id=76564

v2: more cleanup and comments.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>

For runtime pm we'd never suspend with the modesetting hw turned on,
so don't try and resume the modesetting hw, as that path will take
locks that the interface that is causing us to wake up might also
take.
Signed-off-by: Dave Airlie <airlied@redhat.com>

Same fix as for nouveau, when we fail with EINVAL, subsequent
gets fail hard, causing the device not to open.
Signed-off-by: Dave Airlie <airlied@redhat.com>

If the IB test fails we don't want to reset the card over
and over again, just accept that it isn't working.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=76501Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

- Inherit/reuse firmwar framebuffers (for real this time) from Jesse, less
  flicker for fastbooting.
- More flexible cloning for hdmi (Ville).
- Some PPGTT fixes from Ben.
- Ring init fixes from Naresh Kumar.
- set_cache_level regression fixes for the vma conversion from Ville&Chris.
- Conversion to the new dp aux helpers (Jani).
- Unification of runtime pm with pc8 support from Paulo, prep work for runtime
  pm on other platforms than HSW.
- Larger cursor sizes (Sagar Kamble).
- Piles of improvements and fixes all over, as usual.

* tag 'drm-intel-next-2014-03-21' of git://anongit.freedesktop.org/drm-intel: (75 commits)
  drm/i915: Include a note about the dangers of I915_READ64/I915_WRITE64
  drm/i915/sdvo: fix questionable return value check
  drm/i915: Fix unsafe loop iteration over vma whilst unbinding them
  drm/i915: Enabling 128x128 and 256x256 ARGB Cursor Support
  drm/i915: Print how many objects are shared in per-process stats
  drm/i915: Per-process stats work better when evaluated per-process
  drm/i915: remove rps local variables
  drm/i915: Remove extraneous MMIO for RPS
  drm/i915: Rename and comment all the RPS *stuff*
  drm/i915: Store the HW min frequency as min_freq
  drm/i915: Fix coding style for RPS
  drm/i915: Reorganize the overclock code
  drm/i915: init pm.suspended earlier
  drm/i915: update the PC8 and runtime PM documentation
  drm/i915: rename __hsw_do_{en, dis}able_pc8
  drm/i915: kill struct i915_package_c8
  drm/i915: move pc8.irqs_disabled to pm.irqs_disabled
  drm/i915: remove dev_priv->pc8.enabled
  drm/i915: don't get/put PC8 when getting/putting power wells
  drm/i915: make intel_aux_display_runtime_get get runtime PM, not PC8
  ...

Conflicts:
	drivers/gpu/drm/i915/intel_display.c
	drivers/gpu/drm/i915/intel_dp.c

Here's the latest iteration of the universal planes work, which I believe is
finally ready for merging.  Aside from the minor driver patches to use the
new drm_for_each_legacy_plane() macro for plane loops, these should all have
an r-b from Rob Clark now.

Actual userspace-visibility is currently hidden behind a
drm.universal_planes module parameter so that we can do some experimental
testing of this before flipping it on universally.

* 'primary-plane' of git://people.freedesktop.org/~robclark/linux:
  drm/doc: Update plane documentation and add plane helper library
  drm: Allow userspace to ask for universal plane list (v2)
  drm: Remove unused drm_crtc->fb
  drm: Replace crtc fb with primary plane fb (v3)
  drm/msm: Switch to universal plane API's
  drm: Add drm_crtc_init_with_planes() (v2)
  drm: Add plane type property (v2)
  drm: Add drm_universal_plane_init()
  drm: Add primary plane helpers (v3)
  drm: Make drm_crtc_check_viewport non-static
  drm/shmobile: Restrict plane loops to only operate on legacy planes
  drm/i915: Restrict plane loops to only operate on overlay planes (v2)
  drm/exynos: Restrict plane loops to only operate on overlay planes (v2)
  drm: Add support for multiple plane types (v2)

This is the equivalent change in the crtc helpers as done to the i915
modeset infrastructure in

commit b0a2658a
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Tue Dec 18 09:37:54 2012 +0100

    drm/i915: don't disable disconnected outputs

This was originally introduced to make encoder sharing on radone
easier for userspace, but:

- It is policy and as such belongs into userspace. E.g. personally I'm
  fairly annoyed that a flaky cable results in permanent changes of
  the desktop layout, so I'll kick out DEs which do this. Worse if the
  kernel also tries to be clever.

- It's inconsistent: We only kill disconnected outputs on setCrtc
  (which userspace might also call when just changing the
  framebuffer), but not when e.g. we receive a hpd event or in the
  output poll worker.

- It's unexpected behaviour for the userspace driver, at least in the
  intel ddx we've had tons of bugs where the driver fell over and
  killed the X session becuase pageflips/vblanks suddenly stopped
  working. We've had to fix this by wrapping every single setCrtc int
  a big "recover kms state from the kernel again" operation.

- It's suprising for the kernel, too: It took a few mails between Rob,
  Matt and me for them to notice that little dragon wreaking havoc
  with the universal plane framebuffer refcounting.

- Userspace can cope with it and e.g. Gnome already kills disconnected
  outputs and reconfigures the desktop automatically. And since there
  have been no regression reports for the i915 change from over 1 year
  ago I think all other DEs are also ready.

Note that the lines removed in this patch go back to

commit a3a0544b
Author: Dave Airlie <airlied@redhat.com>
Date:   Mon Aug 31 15:16:30 2009 +1000

    drm/kms: add explicit encoder disable function and detach harder.

Unfortunately the patch itself doesn't explain a hole lot about why it
was added ...

Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>

This was introduced in

commit 25f397a4
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Fri Jul 19 18:57:11 2013 +0200

    drm/crtc-helper: explicit DPMS on after modeset

but due to a bit of rebase fail on my side the patch actually merged
put one hunk on the wrong side of a break statement. Fix this up.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>

This fixes a BUG_ON(bo->sync_obj != NULL); in ttm_bo_release_list.

Cc: stable@vger.kernel.org #v3.10+
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Earlier this week, there was a bit of confusion about those new
capabilities, to the point I think it's better to document the intention
and API contract.

The comment documents the current situation:
 - the radeon driver returns the only valid size for the hw
 - i915 returns the maximun cursor size
 - other drivers fall back to returning 64x64

The common contract is to return a valid cursor size.

Cc: Sagar Kamble <sagar.a.kamble@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Imre Deak <imre.deak@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

This should ensure we don't hit a locking problem when someone
wakes us up via a connector, we should never go into suspend
while the display is on anyways.
Signed-off-by: Dave Airlie <airlied@redhat.com>

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

Userspace clients which wish to receive all DRM planes (primary and
cursor planes in addition to the traditional overlay planes) may set the
DRM_CLIENT_CAP_UNIVERSAL_PLANES capability.

v2: Hide behind drm.universal_planes module option [suggested by
    Daniel Vetter]
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

Now that CRTC's have a primary plane, there's no need to track the
framebuffer in the CRTC.  Replace all references to the CRTC fb with the
primary plane's fb.

This patch was generated by the Coccinelle semantic patching tool using
the following rules:

        @@ struct drm_crtc C; @@
        -   (C).fb
        +   C.primary->fb

        @@ struct drm_crtc *C; @@
        -   (C)->fb
        +   C->primary->fb

v3: Generate patch via coccinelle.  Actual removal of crtc->fb has been
    moved to a subsequent patch.

v2: Fixup several lingering crtc->fb instances that were missed in the
    first patch iteration.  [Rob Clark]
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

Use drm_universal_plane_init() and drm_crtc_init_with_planes() rather
than the legacy drm_plane_init() / drm_crtc_init().  This will ensure
that the proper primary plane is registered with the DRM (and eventually
exposed to userspace in future patches).

Cc: Rob Clark <robdclark@gmail.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Rob Clark <robdclark@gmail.com>

Add a new drm_crtc_init_with_planes() to allow drivers to provide
specific primary and cursor planes at CRTC initialization.  The existing
drm_crtc_init() interface remains to avoid driver churn in existing
drivers; it will initialize the CRTC with a plane helper-created primary
plane and no cursor plane.

v2:
  - Move drm_crtc_init() to plane helper file so that nothing in the DRM
    core depends on helpers.  [suggested by Daniel Vetter]
  - Keep cursor parameter to drm_crtc_init_with_planes() a void* until
    we actually add cursor support.  [suggested by Daniel Vetter]
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

Add a plane type property to allow userspace to distinguish plane types.

v2: Driver-specific churn eliminated now that drm_plane_init() and
    drm_universal_plane_init() were separated out in a previous patch.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

Add a new plane initialization interface for universal plane support
that allows a specific plane type (primary, cursor, or overlay) to
be specified.

drm_plane_init() remains as a compatibility API to reduce churn in
existing drivers.  The 'bool priv' parameter has been changed to
'bool is_primary' under the assumption that all existing uses of
private planes were representing primary planes.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

When we expose non-overlay planes to userspace, they will become
accessible via standard userspace plane API's.  We should be able to
handle the standard plane operations against primary planes in a generic
way via the modeset handler.

Drivers that can program primary planes more efficiently, that want to
use their own primary plane structure to track additional information,
or that don't have the limitations assumed by the helpers are free to
provide their own implementation of some or all of these handlers.

v3: Tweak kerneldoc formatting slightly to avoid ugliness
v2:
 - Move plane helpers to a new file (drm_plane_helper.c)
 - Tighten checks on update handler (check for scaling, CRTC coverage,
   subpixel positioning)
 - Pass proper panning parameters to modeset interface
 - Disallow disabling primary plane (and thus CRTC) if other planes are
   still active on the CRTC.
 - Use a minimal format list that should work on all hardware/drivers.
   Drivers may call this function with a more accurate plane list to
   enable additional formats they can support.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

This function will be used by the universal plane helpers and may also
be useful for individual drivers.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

Ensure that existing driver loops over all planes do not change behavior
when we begin adding new types of planes (primary and cursor) to the DRM
plane list in future patches.
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>

Ensure that existing driver loops over all planes do not change behavior
when we begin adding new types of planes (primary and cursor) to the DRM
plane list in future patches.

v2: Switch to using drm_for_each_legacy_plane()

Cc: Intel Graphics Development <intel-gfx@lists.freedesktop.org>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>

Ensure that existing driver loops over all planes do not change behavior
when we begin adding new types of planes (primary and cursor) to the DRM
plane list in future patches.

v2: Switch to using drm_for_each_legacy_plane()

Cc: Inki Dae <inki.dae@samsung.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>

The DRM core currently only tracks "overlay"-style planes.  Start
refactoring the plane handling to allow other plane types (primary and
cursor) to also be placed on the DRM plane list.

v2: Add drm_for_each_legacy_plane() iterator to smooth transition
    of drivers with plane loops.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

The connectors list iterator returns the list head when the list is
empty. Fix it by returning NULL in that case.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Rob Clark <robdclark@gmail.com>

Make things recover a bit more gracefully if we get stuck with no vblank
irq ever coming.
Signed-off-by: Rob Clark <robdclark@gmail.com>

After reading a nice article on LWN[1], I went back and double checked
my handling of invalid-input checking.  Turns out there were a couple
places I had missed.

Since the driver is fairly young, and the devices it supports are really
only just barely usable for basic stuff (serial console) with an
upstream kernel, I think we should fix this now and revert specific
parts of this patch later in the unlikely event that a regression is
reported.

[1] https://lwn.net/Articles/588444/Signed-off-by: Rob Clark <robdclark@gmail.com>

Signed-off-by: Rob Clark <robdclark@gmail.com>

Some of the w/a or different behavior of userspace blob driver seem to
be keyed to gpu patch revision, rather than gpu-id.  So expose the full
chip-id to userspace so it can DTRT.
Signed-off-by: Rob Clark <robdclark@gmail.com>

Shut down the clks when the gpu has nothing to do.  A short inactivity
timer is used to provide a low pass filter for power transitions.
Signed-off-by: Rob Clark <robdclark@gmail.com>

Helper macro to simplify places where we need to poll with timeout
waiting for gpu.
Signed-off-by: Rob Clark <robdclark@gmail.com>