drm/vc4: Don't wait for vblank when updating the cursor

Commonly used desktop environments such as xfce4 and gnome
on debian sid can flood the graphics drivers with cursor
updates.  Because the current implementation is waiting
for a vblank between cursor updates, this will cause the
display to hang for a long time since a typical refresh
rate is only 60Hz.

This is unnecessary and unexpected by user mode software,
so simply swap out the cursor frame buffer without waiting.

Signed-off-by: Michael Zoran <mzoran@crowfest.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Link: http://patchwork.freedesktop.org/patch/msgid/20170224015431.24583-1-mzoran@crowfest.net

drm: qxl: Enable atomic modesetting ioctl

Now that atomic support is implemented, enable the atomic flag.

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-15-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Atomic phase 3: Wire up atomic page_flip helper

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-14-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Atomic phase 3: Wire up atomic set_config helper

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-13-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Atomic phase 3: Use atomic handlers for planes

Now that the state objects are wired up, we can move to the final atomic
handlers.

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-12-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Atomic phase 2: Use drm_atomic_set_fb_for_plane helper

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-11-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Atomic phase 2: Wire up state object handlers

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-10-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Atomic phase 1: Don't unpin primary when disabling crtc

In the qxl atomic model, the primary doesn't stay pinned all the time,
instead it is only pinned/unpinned between prepare_fb and cleanup_fb.
So, we no longer need a final unpin of the primary framebuffer when
disabling the crtc.

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-9-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Atomic phase 1: Implement mode_set_nofb

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-8-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Atomic phase 1: Use drm_plane_helpers for primary plane

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-7-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Atomic phase 1: convert cursor to universal plane

In preparation for atomic conversion, let's use the transitional atomic
helpers drm_plane_helper_update/disable.

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-6-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Expose creation of universal primary plane

Let's expose the primary plane initialization inside the qxl driver in
preparation for universal planes and atomic.

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-5-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Don't initialize vblank support

qxl don't have support for hardware vblanks so we can't initialize it
here, otherwise we risk getting stuck in drm_wait_one_vblank.

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-4-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Consolidate bo reservation when pinning

Every attempt to pin/unpin objects in memory requires
qxl_bo_reserve/unreserve calls around the pinning operation to protect
the object from concurrent access, which causes that call sequence to be
reproduced every place where pinning is needed.  In some cases, that
sequence was not executed correctly, resulting in potential unprotected
pinning operations.

This commit encapsulates the reservation inside a new wrapper to make
sure it is always handled properly.  In cases where reservation must be
done beforehand, for some reason, one can use the unprotected version
__qxl_bo_pin/unpin.

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Reviewed-by: Gustavo Padovan <gustavo.padovan@collabora.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-3-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: qxl: Drop device flags attribute

There are no device specific flags that we need to keep track of here.
Let it vanish.

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Reviewed-by: Gustavo Padovan <gustavo.padovan@collabora.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227204328.18761-2-krisman@collabora.co.uk
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>

drm: Rename connector list iterator API

Currently the functions that initialize and tear down a connector
iterator use the _get() and _put() suffixes. However, these suffixes
are typically used by reference counting functions.

Make these function names a little more consistent by changing the
suffixes to _begin() and _end(), which is a fairly common pattern in
the rest of the Linux kernel.

Suggested-by: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170228144643.5668-8-thierry.reding@gmail.com

drm: Introduce drm_property_blob_{get,put}()

For consistency with other reference counting APIs in the kernel, add
drm_property_blob_get() and drm_property_blob_put() to reference count
DRM blob properties.

Compatibility aliases are added to keep existing code working. To help
speed up the transition, all the instances of the old functions in the
DRM core are already replaced in this commit.

A semantic patch is provided that can be used to convert all drivers to
the new helpers.

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170228144643.5668-7-thierry.reding@gmail.com

drm: Introduce drm_gem_object_{get,put}()

For consistency with other reference counting APIs in the kernel, add
drm_gem_object_get() and drm_gem_object_put(), as well as an unlocked
variant of the latter, to reference count GEM buffer objects.

Compatibility aliases are added to keep existing code working. To help
speed up the transition, all the instances of the old functions in the
DRM core are already replaced in this commit.

The existing semantic patch for the DRM subsystem-wide conversion is
extended to account for these new helpers.

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170228144643.5668-6-thierry.reding@gmail.com

drm: Introduce drm_framebuffer_{get,put}()

For consistency with other reference counting APIs in the kernel, add
drm_framebuffer_get() and drm_framebuffer_put() to reference count DRM
framebuffers.

Compatibility aliases are added to keep existing code working. To help
speed up the transition, all the instances of the old functions in the
DRM core are already replaced in this commit.

The existing semantic patch for the DRM subsystem-wide conversion is
extended to account for these new helpers.

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170228144643.5668-5-thierry.reding@gmail.com

drm: Introduce drm_connector_{get,put}()

For consistency with other reference counting APIs in the kernel, add
drm_connector_get() and drm_connector_put() functions to reference count
connectors.

Compatibility aliases are added to keep existing code working. To help
speed up the transition, all the instances of the old functions in the
DRM core are already replaced in this commit.

The existing semantic patch for mode object reference count conversion
is extended for these new helpers.

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170228144643.5668-4-thierry.reding@gmail.com

drm: Introduce drm_mode_object_{get,put}()

For consistency with other reference counting APIs in the kernel, add
drm_mode_object_get() and drm_mode_object_put() to reference count DRM
mode objects.

Compatibility aliases are added to keep existing code working. To help
speed up the transition, all the instances of the old functions in the
DRM core are already replaced in this commit.

A semantic patch is provided that can be used to convert all drivers to
the new helpers.

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170228144643.5668-3-thierry.reding@gmail.com

drm: Rename drm_mode_object_get()

Subsequent patches will introduce reference counting APIs that are more
consistent with similar APIs throughout the Linux kernel. These APIs use
the _get() and _put() suffixes and will collide with this existing
function.

Rename the function to drm_mode_object_add() which is a slightly more
accurate description of what it does. Also the kerneldoc for this
function gives an indication that it's badly named because it doesn't
actually acquire a reference to anything.

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170228144643.5668-2-thierry.reding@gmail.com

gpu: drm: core: Convert printk(KERN_<LEVEL> to pr_<level>

Use a more common logging style.

Miscellanea:

o Coalesce formats and realign arguments

Signed-off-by: Joe Perches <joe@perches.com>
[danvet: Resolve minor conflict in drm_edid.c]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drm/doc: Capture TODO for deferred fbdev setup

Also became a bit a FAQ recently.

Cc: John Stultz <john.stultz@linaro.org
Cc: Thierry Reding <thierry.reding@gmail.com>
Acked-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170226193851.3245-2-daniel.vetter@ffwll.ch

drm/docs: Record TODO about plane clipping

It's such a mess that it's become a FAQ :(

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Russell King - ARM Linux <linux@armlinux.org.uk>
Acked-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20170226193851.3245-1-daniel.vetter@ffwll.ch

drm: Convert drm_framebuffer_remove to atomic, v4.

Instead of trying to do everything in 1 go, just do a basic safe
conversion first. We've been bitten by too many regressions in the
past.

This patch only converts drm_framebuffer_remove to atomic. The
regression sensitive part is split out to a separate patch.

v2:
- Remove plane->fb assignment, done by drm_atomic_clean_old_fb.
- Add WARN_ON when atomic_remove_fb fails.
- Always call drm_atomic_state_put.
v3:
- Use drm_drv_uses_atomic_modeset
- Handle the case where the first plane-disable-only commit fails
  with -EINVAL. Some drivers do not support this, fall back to
  disabling all crtc's in this case.
v4:
- Solve vmwgfx compatibility issue in their driver, was fixed in this
  patch by v3.
- Move only disabling primary to a separate patch.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1487685102-31991-3-git-send-email-maarten.lankhorst@linux.intel.com

drm/atomic: Make disable_all helper fully disable the crtc.

It seems that nouveau requires this, so best to do this in the helper.
This allows nouveau to use the atomic suspend helper.

Cc: nouveau@lists.freedesktop.org
Acked-by: Ben Skeggs <bskeggs@redhat.com> #irc
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1487685102-31991-2-git-send-email-maarten.lankhorst@linux.intel.com

drm/atmel-hlcdc: Simplify the HLCDC layer logic

An HLCDC layers in Atmel's nomenclature is either a DRM plane or a 'Post
Processing Layer' which can be used to output the results of the HLCDC
composition in a memory buffer.

atmel_hlcdc_layer.c was designed to be generic enough to be re-usable in
both cases, but we're not exposing the post-processing layer yet, and
even if we were, I'm not sure the code would provide the necessary tools
to manipulate this kind of layer.

Moreover, the code in atmel_hlcdc_{plane,layer}.c was designed before the
atomic modesetting API, and was trying solve the
check-setting/commit-if-ok/rollback-otherwise problem, which is now
entirely solved by the existing core infrastructure.

And finally, the code in atmel_hlcdc_layer.c is over-complicated compared
to what we really need. This rework is a good excuse to simplify it. Note
that this rework solves an existing resource leak (leading to a -EBUSY
error) which I failed to clearly identify.

Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Nicolas Ferre <nicolas.ferre@microchip.com>

drm: add drm_get_connector_force_name

Follow the naming in debugfs also for logging, add "unknown" for values
beyond the enumerated ones.

v2: add \n in connector_show, make internal to drm (Chris)

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1487580708-29340-1-git-send-email-jani.nikula@intel.com

drm/via: use get_user_pages_unlocked()

Moving from get_user_pages() to get_user_pages_unlocked() simplifies the code
and takes advantage of VM_FAULT_RETRY functionality when faulting in pages.

Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20170227215008.21457-1-lstoakes@gmail.com

drm: sti: make driver use devm_of_platform_populate()

This make sure that of_platform_depopulate() is called if an error
occur in probe after populating the date from the device tree.

Signed-off-by: Benjamin Gaignard <benjamin.gaignard@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1487952874-23635-3-git-send-email-benjamin.gaignard@linaro.org

of: add devm_ functions for populate and depopulate

Lots of calls to of_platform_populate() are not unbalanced by a call
to of_platform_depopulate(). This create issues while drivers are
bind/unbind.

In way to solve those issues is to add devm_of_platform_populate()
which will call of_platform_depopulate() when the device is unbound
from the bus.

Signed-off-by: Benjamin Gaignard <benjamin.gaignard@linaro.org>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1487952874-23635-2-git-send-email-benjamin.gaignard@linaro.org

drm: Add a new connector atomic property for link status

At the time userspace does setcrtc, we've already promised the mode
would work. The promise is based on the theoretical capabilities of
the link, but it's possible we can't reach this in practice. The DP
spec describes how the link should be reduced, but we can't reduce
the link below the requirements of the mode. Black screen follows.

One idea would be to have setcrtc return a failure. However, it
already should not fail as the atomic checks have passed. It would
also conflict with the idea of making setcrtc asynchronous in the
future, returning before the actual mode setting and link training.

Another idea is to train the link "upfront" at hotplug time, before
pruning the mode list, so that we can do the pruning based on
practical not theoretical capabilities. However, the changes for link
training are pretty drastic, all for the sake of error handling and
DP compliance, when the most common happy day scenario is the current
approach of link training at mode setting time, using the optimal
parameters for the mode. It is also not certain all hardware could do
this without the pipe on; not even all our hardware can do this. Some
of this can be solved, but not trivially.

Both of the above ideas also fail to address link degradation *during*
operation.

The solution is to add a new "link-status" connector property in order
to address link training failure in a way that:
a) changes the current happy day scenario as little as possible, to
avoid regressions, b) can be implemented the same way by all drm
drivers, c) is still opt-in for the drivers and userspace, and opting
out doesn't regress the user experience, d) doesn't prevent drivers
from implementing better or alternate approaches, possibly without
userspace involvement. And, of course, handles all the issues presented.
In the usual happy day scenario, this is always "good". If something
fails during or after a mode set, the kernel driver can set the link
status to "bad" and issue a hotplug uevent for userspace to have it
re-check the valid modes through GET_CONNECTOR IOCTL, and try modeset
again. If the theoretical capabilities of the link can't be reached,
the mode list is trimmed based on that.

v7 by Jani:
* Rebase, simplify set property while at it, checkpatch fix
v6:
* Fix a typo in kernel doc (Sean Paul)
v5:
* Clarify doc for silent rejection of atomic properties by driver (Daniel Vetter)
v4:
* Add comments in kernel-doc format (Daniel Vetter)
* Update the kernel-doc for link-status (Sean Paul)
v3:
* Fixed a build error (Jani Saarinen)
v2:
* Removed connector->link_status (Daniel Vetter)
* Set connector->state->link_status in drm_mode_connector_set_link_status_property
(Daniel Vetter)
* Set the connector_changed flag to true if connector->state->link_status changed.
* Reset link_status to GOOD in update_output_state (Daniel Vetter)
* Never allow userspace to set link status from Good To Bad (Daniel Vetter)

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Tony Cheng <tony.cheng@amd.com>
Acked-by: Harry Wentland <harry.wentland@amd.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Ville Syrjala <ville.syrjala@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Manasi Navare <manasi.d.navare@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Acked-by: Eric Anholt <eric@anholt.net> (for the -modesetting patch)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/0182487051aa9f1594820e35a4853de2f8747b4e.1481883920.git.jani.nikula@intel.com

drm/doc: Fix up some kms function names

A couple of the kms functions didn't have the correct/newest names.
This prevented them to be identified as refs in the html doc.

Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
Tested-by: Eric Engestrom <eric.engestrom@imgtec.com>
Signed-off-by: Archit Taneja <architt@codeaurora.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20170222084741.8485-1-architt@codeaurora.org

drm/tinydrm: fix semicolon.cocci warnings

drivers/gpu/drm/tinydrm/core/tinydrm-helpers.c:454:2-3: Unneeded semicolon

 Remove unneeded semicolon.

Generated by: scripts/coccinelle/misc/semicolon.cocci

Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20170223164647.GA2519@lkp-ws02

drm: kselftest: fix spelling mistake: "misalinged" -> "misaligned"

trivial fix to spelling mistake in pr_err message

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20170223000717.8898-1-colin.king@canonical.com

dma-buf/reservation: Wrap ww_mutex_trylock

In a similar fashion to reservation_object_lock() and
reservation_object_unlock(), ww_mutex_trylock is also useful and so is
worth wrapping for consistency.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
[danvet: Add __must_check Joonas wants.]
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20170221093000.22802-1-chris@chris-wilson.co.uk

drm/fb-helper: Add multi buffer support for cma fbdev

This patch add a config to support to create multi buffer for cma fbdev.
Such as double buffer and triple buffer.

Cma fbdev is convient to add a legency fbdev. And still many Android
devices use fbdev now and at least double buffer is needed for these
Android devices, so that a buffer flip can be operated. It will need
some time for Android device vendors to abondon legency fbdev. So multi
buffer for fbdev is needed.

Signed-off-by: Xinliang Liu <xinliang.liu@linaro.org>
[s.christ@phytec.de: Picking patch from
                     https://lkml.org/lkml/2015/9/14/188]
Signed-off-by: Stefan Christ <s.christ@phytec.de>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Acked-by: Dave Airlie <airlied@gmail.com>
Tested-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/075ffb50cc16ab055b5d47b30163401bb356ab51.1487175046.git-series.maxime.ripard@free-electrons.com

drm: Always prepare null framebuffer in transitional helper

Despite the documentation claim that cleanup_fb will match prior calls
to prepare_fb, in case of NULL framebuffers in the transitional helpers,
the code will skip the call to prepare_fb but not the corresponding
cleanup_fb call.  This asymmetry in semantics is unnecessarily surprising
for developers transitioning drivers to atomic model, specially because
the final atomic handlers don't have the issue - the prepare_fb is
always called, despite the new state framebuffer being null.

The only current user of the transitional helper that doesn't take care
of null framebuffers explicitly inside the prepare_fb hook is
atmel_hlcdc, so we take special care to make sure we don't break
anything there.

Signed-off-by: Gabriel Krisman Bertazi <krisman@collabora.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20170216164442.28704-1-krisman@collabora.co.uk

drm: drm_printer: add __printf validation

drm_printf does not currently use the compiler to verify
format and arguments.  Make it do so.

Miscellanea:

o Add appropriate #include files for __printf and struct va_format
o Convert dev_printk to dev_info

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/133858f214e9b90f92bb8eb44c6b1dc04429933d.1487201526.git.joe@perches.com

Merge airlied/drm-next into drm-misc-next

Backmerge the main pull request to sync up with all the newly landed
drivers. Otherwise we'll have chaos even before 4.12 started in
earnest.

Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>

lib/Kconfig: make PRIME_NUMBERS not user selectable.

Linus doesn't like it user selectable, so kill it until
someone needs it for something else.

Signed-off-by: Dave Airlie <airlied@redhat.com>

drm/tinydrm: helpers: Properly fix backlight dependency

BACKLIGHT_CLASS_DEVICE was selected in the last version of the
tinydrm patchset to fix the backlight dependency, but the
ifdef CONFIG_BACKLIGHT_CLASS_DEVICE was forgotten. Fix that.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>

drm/tinydrm: mipi-dbi: Fix field width specifier warning

This warning is seen on 64-bit builds in functions:
   'mipi_dbi_typec1_command':
   'mipi_dbi_typec3_command_read':
   'mipi_dbi_typec3_command':

>> drivers/gpu/drm/tinydrm/mipi-dbi.c:65:20: warning: field width specifier '*' expects argument of type 'int', but argument 5 has type 'size_t {aka long unsigned int}' [-Wformat=]
      DRM_DEBUG_DRIVER("cmd=%02x, par=%*ph\n", cmd, len, data); \
                       ^
   include/drm/drmP.h:228:40: note: in definition of macro 'DRM_DEBUG_DRIVER'
     drm_printk(KERN_DEBUG, DRM_UT_DRIVER, fmt, ##__VA_ARGS__)
                                           ^~~
>> drivers/gpu/drm/tinydrm/mipi-dbi.c:671:2: note: in expansion of macro 'MIPI_DBI_DEBUG_COMMAND'
     MIPI_DBI_DEBUG_COMMAND(cmd, parameters, num);
     ^~~~~~~~~~~~~~~~~~~~~~

Fix by casting 'len' to int in the macro MIPI_DBI_DEBUG_COMMAND().
There is no chance of overflow.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>

drm/tinydrm: mipi-dbi: Silence: ‘cmd’ may be used uninitialized

Fix this warning:
drivers/gpu/drm/tinydrm/mipi-dbi.c: In function ‘mipi_dbi_debugfs_command_write’:
drivers/gpu/drm/tinydrm/mipi-dbi.c:905:8: warning: ‘cmd’ may be used uninitialized in this function [-Wmaybe-uninitialized]
  ret = mipi_dbi_command_buf(mipi, cmd, parameters, i);
        ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

cmd can't be used uninitialized, but to satisfy the compiler,
initialize it to zero.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Merge branch 'drm-next-4.11' of git://people.freedesktop.org/~agd5f/linux into drm-next

Some ttm/amd fixes.

* 'drm-next-4.11' of git://people.freedesktop.org/~agd5f/linux:
  drm/amd/powerplay: fix PSI feature on Polars12.
  drm/amdgpu: refuse to reserve io mem for split VRAM buffers
  drm/ttm: fix use-after-free races in vm fault handling
  drm/amd/amdgpu: post card if there is real hw resetting performed

Merge tag 'drm/panel/for-4.11-rc1' of git://anongit.freedesktop.org/tegra/linux into drm-next

drm/panel: Changes for v4.11-rc1

This set contains a couple of cleanups as well as support for a few more
simple panels.

* tag 'drm/panel/for-4.11-rc1' of git://anongit.freedesktop.org/tegra/linux:
  drm/panel: simple: Specify bus width and flags for EDT displays
  drm/panel: simple: Add Netron DY E231732
  of: Add vendor prefix for Netron DY
  drm/panel: simple: Add support for Tianma TM070JDHG30
  of: Add vendor prefix for Tianma Micro-electronics
  drm/panel: simple: Add support BOE NV101WXMN51
  dt-bindings: display: Add BOE NV101WXMN51 panel binding
  drm/panel: Constify device node argument to of_drm_find_panel()

Merge tag 'drm/tegra/for-4.11-rc1' of git://anongit.freedesktop.org/tegra/linux into drm-next

drm/tegra: Changes for v4.11-rc1

Just a single change that hooks up the Tegra DRM parent device to the
correct device tree node.

* tag 'drm/tegra/for-4.11-rc1' of git://anongit.freedesktop.org/tegra/linux:
  gpu: host1x: Set OF node for new host1x devices

drm: Add name for DRM_DP_DUAL_MODE_LSPCON

Handle DRM_DP_DUAL_MODE_LSPCON in drm_dp_get_dual_mode_type_name(),
otherwise a call to that function can theoretically trigger a WARN.

Fixes: 056996b95686 ("drm: Helper for lspcon in drm_dp_dual_mode")
Cc: Shashank Sharma <shashank.sharma@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Dave Airlie <airlied@gmail.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Cc: Sean Paul <seanpaul@chromium.org>
Cc: David Airlie <airlied@linux.ie>
Cc: dri-devel@lists.freedesktop.org
Cc: intel-gfx@lists.freedesktop.org
Reviewed-by: Shashank Sharma <shashank.sharma@intel.com>
Signed-off-by: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@intel.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20170222125928.30223-1-ander.conselvan.de.oliveira@intel.com

drm/bridge/sii8620: add missing error handling in probe

devm_request_threaded_irq result should be checked for errors.

Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1487764057-23362-1-git-send-email-a.hajda@samsung.com

drm/sti: fix build warnings in sti_drv.c and sti_vtg.c files

Fix compilation warning introduced by:
commit 0c7ff84f7f9d ("drm/sti: remove deprecated legacy vtg slave")
commit 5e60f595d6ca ("drm/sti: use atomic_helper for commit")

Signed-off-by: Vincent Abriou <vincent.abriou@st.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Merge tag 'imx-drm-fixes-2017-02-17' of https://git.pengutronix.de/git/pza/linux into drm-next

imx-drm: TVE regulator, fb size limit, and ipu-v3 module fixes

- Fix i.MX5 TV encoder probing in case no dac-supply regulator
  is set in the device tree.
- Remove 64 pixel min_width/height limit, which unnecessarily
  prohibits creation of small frame buffers.
- Add missing ipu_csi_set_downsize export, for media drivers
  built as modules.
- Stop modifying pdev->dev.of_node for IPU client devices that
  do not have an OF modalias to fix module autoloading.

* tag 'imx-drm-fixes-2017-02-17' of https://git.pengutronix.de/git/pza/linux:
  gpu: ipu-v3: Stop overwriting pdev->dev.of_node of child devices
  gpu: ipu-v3: export ipu_csi_set_downsize
  drm/imx: lift 64x64 pixel minimum framebuffer size requirement
  drm/imx: imx-tve: Do not set the regulator voltage

Merge tag 'v4.10-rc8' into drm-next

Linux 4.10-rc8

Backmerge Linus rc8 to fix some conflicts, but also
to avoid pulling it in via a fixes pull from someone.

Merge tag 'drm-tinydrm-2017-02-18' of https://github.com/notro/linux into drm-next

Add tinydrm

* tag 'drm-tinydrm-2017-02-18' of https://github.com/notro/linux:
  drm/tinydrm: Add support for Multi-Inno MI0283QT display
  dt-bindings: Add Multi-Inno MI0283QT binding
  dt-bindings: display/panel: Add common rotation property
  of: Add vendor prefix for Multi-Inno
  drm/tinydrm: Add MIPI DBI support
  drm/tinydrm: Add helper functions
  drm: Add DRM support for tiny LCD displays

drm/amd/powerplay: fix PSI feature on Polars12.

Signed-off-by: Rex Zhu <Rex.Zhu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/amdgpu: refuse to reserve io mem for split VRAM buffers

When the fast blit path fails while attempting to move a buffer from RAM
to VRAM, we fall back to a CPU-based memcpy that cannot handle split VRAM
buffers. Instead of crashing, simply fail the buffer move.

Ideally, we would teach TTM about split buffers so that the fallback still
works in this case, but that is quite involved. So for now, apply the
simplest possible fix.

Fixes: 40361bb1704b ("drm/amdgpu: add VRAM manager v2")
Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm/ttm: fix use-after-free races in vm fault handling

The vm fault handler relies on the fact that the VMA owns a reference
to the BO. However, once mmap_sem is released, other tasks are free to
destroy the VMA, which can lead to the BO being freed. Fix two code
paths where that can happen, both related to vm fault retries.

Found via a lock debugging warning which flagged &bo->wu_mutex as
locked while being destroyed.

Fixes: cbe12e74ee4e ("drm/ttm: Allow vm fault retries")
Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drm: qxl: use vblank hooks in struct drm_crtc_funcs

The vblank hooks in struct drm_driver are deprecated and only meant for
legacy drivers.  For modern drivers with DRIVER_MODESET flag, the hooks
in struct drm_crtc_funcs should be used instead.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Cc: Dave Airlie <airlied@redhat.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1486458995-31018-16-git-send-email-shawnguo@kernel.org

drm: mediatek: use vblank hooks in struct drm_crtc_funcs

The vblank hooks in struct drm_driver are deprecated and only meant for
legacy drivers.  For modern drivers with DRIVER_MODESET flag, the hooks
in struct drm_crtc_funcs should be used instead.

Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Cc: CK Hu <ck.hu@mediatek.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1486458995-31018-14-git-send-email-shawnguo@kernel.org

drm: kirin: use vblank hooks in struct drm_crtc_funcs

The vblank hooks in struct drm_driver are deprecated and only meant for
legacy drivers.  For modern drivers with DRIVER_MODESET flag, the hooks
in struct drm_crtc_funcs should be used instead.

Reviewed-by: Xinliang Liu <xinliang.liu@linaro.org>
Signed-off-by: Shawn Guo <shawn.guo@linaro.org>
Cc: Xinliang Liu <z.liuxinliang@hisilicon.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1486458995-31018-12-git-send-email-shawnguo@kernel.org

drm/edid: respect connector force for drm_get_edid ddc probe

Skip DDC probe for forced connector status. Don't try to read the EDID
if the connector is forced off. Skipping probe for forced on connectors
will make more sense when drm_do_get_edid() will handle override and
firmware EDIDs.

Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1487344854-18777-4-git-send-email-jani.nikula@intel.com

drm: do not debug log about missing CEA extensions on NULL edid

Make the drm_edid_to_eld() function useful for resetting, not just
setting, the ELD and HDMI VSDB data, without debug warnings about
missing CEA extensions.

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1487344854-18777-3-git-send-email-jani.nikula@intel.com

drm: move edid property update and add modes out of edid firmware loader

Make the firmware loader more generic and generally useful.

Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1487344854-18777-2-git-send-email-jani.nikula@intel.com

drm/tinydrm: Add support for Multi-Inno MI0283QT display

Add driver to support the Multi-Inno MI0283QT display panel.
It has an ILI9341 MIPI DBI compatible display controller.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Thierry Reding <treding@nvidia.com>

dt-bindings: Add Multi-Inno MI0283QT binding

Add device-tree binding documentation for the MI0283QT display panel.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Rob Herring <robh@kernel.org>

dt-bindings: display/panel: Add common rotation property

Display panels can be oriented many ways, especially in the embedded
world. The rotation property is a way to describe this orientation.
The counter clockwise direction is chosen because that's what fbdev
and drm use.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Thierry Reding <treding@nvidia.com>
Acked-by: Rob Herring <robh@kernel.org>

of: Add vendor prefix for Multi-Inno

Multi-Inno Technology Co.,Ltd is a Hong Kong based company offering
LCD, LCD module products and complete panel solutions.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Rob Herring <robh@kernel.org>

drm/tinydrm: Add MIPI DBI support

Add support for MIPI DBI compatible controllers.
Interface type C option 1 and 3 are supported (SPI).

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Thierry Reding <treding@nvidia.com>

drm/tinydrm: Add helper functions

Add common functionality needed by many tinydrm drivers.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Thierry Reding <treding@nvidia.com>

drm: Add DRM support for tiny LCD displays

tinydrm provides helpers for very simple displays that can use
CMA backed framebuffers and need flushing on changes.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Thierry Reding <treding@nvidia.com>

drm/amd/amdgpu: post card if there is real hw resetting performed

Check whether we need to post rather than whether the asic is
posted.  There are some cases (e.g., GPU reset or resume from
hibernate) where we need to force post even if the asic has
been posted.

Signed-off-by: Jim Qu <Jim.Qu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

Merge tag 'drm-intel-next-fixes-2017-02-17' of git://anongit.freedesktop.org/git/drm-intel into drm-next

i915 and GVT fixes for v4.11 merge window

* tag 'drm-intel-next-fixes-2017-02-17' of git://anongit.freedesktop.org/git/drm-intel: (32 commits)
  drm/i915: Fix not finding the VBT when it overlaps with OPREGION_ASLE_EXT
  drm/i915: Pass timeout==0 on to i915_gem_object_wait_fence()
  drm/i915/gvt: Disable access to stolen memory as a guest
  drm/i915: Avoid spurious WARNs about the wrong pipe in the PPS code
  drm/i915: Check for timeout completion when waiting for the rq to submitted
  drm/i915: A hotfix for making aliasing PPGTT work for GVT-g
  drm/i915: Restore context and pd for ringbuffer submission after reset
  drm/i915: Let execlist_update_context() cover !FULL_PPGTT mode.
  drm/i915/lspcon: Fix resume time initialization due to unasserted HPD
  drm/i915/gen9+: Enable hotplug detection early
  drm/i915: Reject set-tiling-ioctl with stride==0 and a tiling mode
  drm/i915: Recreate internal objects with single page segments if dmar fails
  drm/i915/gvt: return error code if dma map iova failed
  drm/i915/gvt: optimize the inhibit context mmio load
  drm/i915/gvt: add sprite plane flip done support.
  drm/i915/gvt: add missing display part reset for vGPU reset
  drm/i915/gvt: Fix shadow context descriptor
  drm/i915/gvt: Fix alignment for GTT allocation
  drm/i915/gvt: fix crash at function release_shadow_wa_ctx
  drm/i915/gvt: enable IOMMU for gvt
  ...

Merge branch 'linux-4.11' of git://github.com/skeggsb/linux into drm-next

- Rework of the secure boot code, in preparation for GP10x secure boot.
- Improvements to channel recovery
- Initial power budget code
- Some preparation for an upcoming MMU rework (probably 4.12)
- Misc other fixes.

* 'linux-4.11' of git://github.com/skeggsb/linux: (88 commits)
  drm/nouveau/tmr: provide backtrace when a timeout is hit
  drm/nouveau/pci/g92: Fix rearm
  drm/nouveau/drm/therm/fan: add a fallback if no fan control is specified in the vbios
  drm/nouveau/hwmon: expose power_max and power_crit
  drm/nouveau/iccsense: Parse max and crit power level
  drm/nouveau/bios/power_budget: Add basic power budget parsing
  drm/nouveau/fifo/gk104-: preempt recovery
  drm/nouveau/fifo/gk104-: trigger mmu fault before attempting engine recovery
  drm/nouveau/fifo/gk104-: ACK SCHED_ERROR before attempting CTXSW_TIMEOUT recovery
  drm/nouveau/fifo/gk104-: directly use new recovery code for ctxsw timeout
  drm/nouveau/fifo/gk104-: directly use new recovery code for mmu faults
  drm/nouveau/fifo/gk104-: reset all engines a killed channel is still active on
  drm/nouveau/fifo/gk104-: refactor recovery code
  drm/nouveau/fifo/gk104-: better detection of chid when parsing engine status
  drm/nouveau/fifo/gk104-: separate out engine status parsing
  drm/nouveau/fifo: add an api for initiating channel recovery
  drm/nouveau/top: add function to translate subdev index to mmu fault id
  drm/nouveau/gr/gf100-: implement chsw_load() method
  drm/nouveau/gr: implement chsw_load() method
  drm/nouveau/core: add engine method to assist in determining chsw direction
  ...

drm/nouveau/tmr: provide backtrace when a timeout is hit

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/pci/g92: Fix rearm

704a6c008b7942bb7f30bb43d2a6bcad7f543662 broke pci msi rearm for g92 GPUs.

g92 needs the nv46_pci_msi_rearm, where g94+ gpus used nv40_pci_msi_rearm.

Reported-by: Andrew Randrianasulu <randrianasulu@gmail.com>
Signed-off-by: Karol Herbst <karolherbst@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Cc: stable@vger.kernel.org

drm/nouveau/drm/therm/fan: add a fallback if no fan control is specified in the vbios

This seems to be absolutely necessary for a lot of NV40.

Reported-by: gsgf on IRC/freenode
Signed-off-by: Martin Peres <martin.peres@free.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/hwmon: expose power_max and power_crit

Signed-off-by: Karol Herbst <karolherbst@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/iccsense: Parse max and crit power level

Signed-off-by: Karol Herbst <karolherbst@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/bios/power_budget: Add basic power budget parsing

v2: Set entry to 0xff if not found
    Add cap entry for ver 0x30 tables
    Rework to fix memory leak
v3: More error checks
    Simplify check for invalid entries
v4: disable for ver 0x10 for now
    move assignments after the second last return

Signed-off-by: Karol Herbst <karolherbst@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: preempt recovery

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: trigger mmu fault before attempting engine recovery

Greatly improves the chances of recovering the GPU from a CTXSW_TIMEOUT.

Tested with piglit's arb_shader_image_load_store-atomicity, which causes
GR to hang in such a way that recovery failed (CTXSW_TIMEOUT continually
re-triggers).

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: ACK SCHED_ERROR before attempting CTXSW_TIMEOUT recovery

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: directly use new recovery code for ctxsw timeout

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: directly use new recovery code for mmu faults

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: reset all engines a killed channel is still active on

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: refactor recovery code

This will serve as a basis for implementing some improvements to how
we recover the GPU from channel errors.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: better detection of chid when parsing engine status

The previous commit simply changes the interface, but should result in
the same behaviour as previously.  This commit has been split out from
it as it can result in a different channel being selected.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gk104-: separate out engine status parsing

We'll be wanting to reuse this logic in more places.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo: add an api for initiating channel recovery

This will be used by callers outside of fifo interrupt handlers.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/top: add function to translate subdev index to mmu fault id

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr/gf100-: implement chsw_load() method

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/gr: implement chsw_load() method

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/core: add engine method to assist in determining chsw direction

FIFO gives us load/save/switch status, and we need to be able to determine
which direction a "switch" is failing during channel recovery.

In order to do this, we apparently need to query the engine itself.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: check for dead channel before trying to idle

This prevents *very* long waits while attempting to destroy channels
after a fault has occurred.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau: request notifications for channels that have been killed

These will be used to improve error recovery behaviour.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/gf100-: provide notification to user if channel is killed

There are instances (such as non-recoverable GPU page faults) where
NVKM decides that a channel's context is no longer viable, and will
be removed from the runlist.

This commit notifies the owner of the channel when this happens, so
it has the opportunity to take some kind of recovery action instead
of hanging.

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo/g84-: rename non-stall interrupt event

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/fifo: tidy up channel creation event code

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/core: increase maximum number of notifies that a client can request

Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drm/nouveau/drm/nouveau/led: prevent a possible use-after-free

If the led class registration fails, we free drm->led but do not reset
it to NULL, which means that the suspend/resume/fini function will act
as if everything went well in init() and will likely crash the kernel.

This patch adds the missing drm->led = NULL.

Reported-by: Emmanuel Pescosta <emmanuelpescosta099@gmail.com>
Signed-off-by: Martin Peres <martin.peres@free.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>