cascardo/linux.git
7 years agoMerge branch 'drm-etnaviv-next' of git://git.pengutronix.de/git/lst/linux into drm...
Dave Airlie [Wed, 28 Sep 2016 01:24:05 +0000 (11:24 +1000)]
Merge branch 'drm-etnaviv-next' of git://git.pengutronix.de/git/lst/linux into drm-next

Notable changes:

- Cleanups from Fabio to some error paths and proper error propagation.

- Lots of refactoring and new code to support the new MMU version 2,
still relatively unoptimized and doesn't yet provide better process
isolation than MMUv1, but enough to get newer cores up and running.

- New hardware support: GC3000, as found on the NXP i.MX6 QuadPlus SoC.

* 'drm-etnaviv-next' of git://git.pengutronix.de/git/lst/linux: (25 commits)
  drm/etnaviv: mark whole context as lost in recover worker
  drm/etnaviv: record correct cmdbuf IOVA in dump
  drm/etnaviv: space out IOVA layout for cmdbufs on MMUv2
  drm/etnaviv: fix up model and revision for GC2000+
  drm/etnaviv: implement IOMMUv2 translation
  drm/etnaviv: handle MMU exception in IRQ handler
  drm/etnaviv: add flushing logic for MMUv2
  drm/etnaviv: add function to construct MMUv2 init buffer
  drm/etnaviv: map cmdbuf through MMU on version 2
  drm/etnaviv: split out iova search and MMU reaping logic
  drm/etnaviv: split out FE start
  drm/etnaviv: split out wait for gpu idle
  drm/etnaviv: move gpu_va() to etnaviv mmu
  drm/etnaviv: remove unused iommu_v2 header
  drm/etnaviv: move IOMMU domain allocation into etnaviv MMU
  drm/etnaviv: indirect IOMMU restore through etnaviv MMU
  drm/etnaviv: move linear window setup into etnaviv_iommuv1_restore
  drm/etnaviv: rename etnaviv_iommu_domain_restore to etnaviv_iommuv1_restore
  drm/etnaviv: only check if the cmdbuf is inside the linear window on MMUv1
  drm/etnaviv: only try to use the linear window on MMUv1
  ...

7 years agoMerge tag 'sunxi-drm-fixes-for-4.9' of https://git.kernel.org/pub/scm/linux/kernel...
Dave Airlie [Wed, 28 Sep 2016 01:23:15 +0000 (11:23 +1000)]
Merge tag 'sunxi-drm-fixes-for-4.9' of https://git./linux/kernel/git/mripard/linux into drm-next

Allwinner sun4i DRM fixes for 4.9

A few fixes for the sun4i drm driver that range, including some fixes that
might prevent multiple planes from working depending on the sequence where
they are enabled.

* tag 'sunxi-drm-fixes-for-4.9' of https://git.kernel.org/pub/scm/linux/kernel/git/mripard/linux:
  drm/sun4i: Fix the high buffer address mask
  drm/sun4i: tv: Check mode pointer
  drm/sun4i: Fix formats usable by the primary plane
  drm/sun4i: dotclock: Round to closest clock rate
  drm/sun4i: Fix sparse warnings
  drm/sun4i: dotclock: Allow divider = 127
  drm/sun4i: dotclock: Fix clock rate read back calcation
  drm/sun4i: backend: remove redundant dev_err call in sun4i_backend_bind()

7 years agoMerge branch 'for-next' of ssh://people.freedesktop.org/~seanpaul/dogwood into drm...
Dave Airlie [Wed, 28 Sep 2016 01:22:27 +0000 (11:22 +1000)]
Merge branch 'for-next' of ssh://people.freedesktop.org/~seanpaul/dogwood into drm-next

I've included some improvements to PSR from myself, as well as a great
series from Tomasz to clean up and tighten up vblank/flip handling.

The last patch is one from Tomeu that has been floating around for a
while, and since rockchip is one of the beneficiaries, I figured this
would be a good place to pick it up.

* 'for-next' of ssh://people.freedesktop.org/~seanpaul/dogwood:
  drm/rockchip: Balance irq refcount on failure
  drm/rockchip: Kill vop_plane_state
  drm/rockchip: Always signal event in next vblank after cfg_done
  drm/rockchip: Do not enable vblank without event
  drm/rockchip: Replace custom wait_for_vblanks with helper
  drm/rockchip: Unreference framebuffers from flip work
  drm/rockchip: Avoid race with vblank count increment
  drm/rockchip: Get rid of some unnecessary code
  drm/rockchip: Clear interrupt status bits before enabling
  drm/rockchip: Fix up bug in psr state machine
  drm/bridge: analogix_dp: Remove duplicated code
  drm/rockchip: Reduce psr flush time to 100ms
  drm/rockchip: Don't key off vblank for psr

7 years agoMerge tag 'tilcdc-4.9-3.1' of https://github.com/jsarha/linux into drm-next
Dave Airlie [Wed, 28 Sep 2016 01:17:26 +0000 (11:17 +1000)]
Merge tag 'tilcdc-4.9-3.1' of https://github.com/jsarha/linux into drm-next

Second attempt for 3rd drm/tilcdc pull request for v4.9.

* tag 'tilcdc-4.9-3.1' of https://github.com/jsarha/linux:
  drm/tilcdc: fix wrong error handling
  drm/tilcdc: Return directly after a failed kfree_table_init() in tilcdc_convert_slave_node()
  drm/tilcdc: Remove "default" from blue-and-red-wiring property binding
  drm/tilcdc: Fix non static symbol warning
  drm/tilcdc: mark symbols static where possible
  drm/tilcdc: add missing header dependencies
  drm/tilcdc: WARN if CRTC is touched without CRTC lock
  drm/tilcdc: Take CRTC lock when calling tilcdc_crtc_disable()
  drm/tilcdc: Remove unnecessary tilcdc_crtc_disable() from tilcdc_unload()
  drm/tilcdc: Flush flip-work workqueue before drm_flip_work_cleanup()
  drm/tilcdc: Clean up LCDC functional clock rate setting code
  drm/tilcdc: Take crtc modeset lock while updating the crtc clock rate

7 years agoMerge branch 'sti-drm-next-2016-09-20' of https://github.com/vinceab/linux into drm...
Dave Airlie [Wed, 28 Sep 2016 01:14:58 +0000 (11:14 +1000)]
Merge branch 'sti-drm-next-2016-09-20' of https://github.com/vinceab/linux into drm-next

Here are some patches for drm-next.
It contains:
- minor fixes for typo and warning.
- sparse and coccicheck warning fixes
- bunch of patches fixing issues found while testing drm/sti with an
atomic version of weston
- the removal of the support of stih415-416 sti platform

* 'sti-drm-next-2016-09-20' of https://github.com/vinceab/linux:
  drm/sti: remove stih415-416 platform support
  drm/sti: fix compositor debugfs creation
  drm/sti: use valid video mode
  drm/sti: in crtc_atomic_flush, enable only planes of this crtc
  drm/sti: use vtg array instead of vtg_main/aux
  drm/sti: use different notifier_block for each pipe
  drm/sti: fix atomic_disable check
  drm/sti: run gdp init sequence only once
  drm/sti: run hqvdp init sequence only once
  drm/sti: fix debug logs
  drm/sti: dpms function missing for HDMI connector
  drm/sti: Fix sparse warnings
  drm: sti: fix coccicheck warnings
  drm: sti: Replace drm_fb_get_bpp_depth() with drm_format_plane_cpp()

7 years agoMerge branch 'msm-next' of git://people.freedesktop.org/~robclark/linux into drm...
Dave Airlie [Wed, 28 Sep 2016 01:09:59 +0000 (11:09 +1000)]
Merge branch 'msm-next' of git://people.freedesktop.org/~robclark/linux into drm-next

A bit smaller pull-req this time around.  Some continued DT binding
cleanup to get the corresponding dts bits merged upstream (through
other trees).  And explicit fence-fd support for submit ioctl.

* 'msm-next' of git://people.freedesktop.org/~robclark/linux:
  drm/msm: bump kernel api version for explicit fencing
  drm/msm: submit support for out-fences
  drm/msm: move fence allocation out of msm_gpu_submit()
  drm/msm: submit support for in-fences
  drm/msm: extend the submit ioctl to pass in flags
  drm/msm/mdp5: Set rotation property initial value to DRM_ROTATE_0 insted of 0
  drm/msm/hdmi: don't print error when adding i2c adapter fails
  drm/msm/mdp4: mark symbols static where possible
  drm/msm: Remove call to reservation_object_test_signaled_rcu before wait
  drm/msm/hdmi: Clean up HDMI gpio DT bindings
  drm/msm/mdp4: Fix issue with LCDC/LVDS port parsing

7 years agoMerge tag 'topic/drm-misc-2016-09-25' of git://anongit.freedesktop.org/drm-intel...
Dave Airlie [Wed, 28 Sep 2016 00:28:23 +0000 (10:28 +1000)]
Merge tag 'topic/drm-misc-2016-09-25' of git://anongit.freedesktop.org/drm-intel into drm-next

- more core cleanup patches to prep drm_file to be used for
  kernel-internal contexts (David Herrmann)
- more split-up+docs for drm_crtc.c
- lots of small fixes and polish all over

* tag 'topic/drm-misc-2016-09-25' of git://anongit.freedesktop.org/drm-intel: (37 commits)
  drm: bridge: analogix/dp: mark symbols static where possible
  drm/bochs: mark bochs_connector_get_modes() static
  drm/bridge: analogix_dp: Improve panel on time
  drm/bridge: analogix_dp: Don't read EDID if panel present
  drm/bridge: analogix_dp: Remove duplicated code
  Revert "drm/i2c: tda998x: don't register the connector"
  drm: Fix plane type uabi breakage
  dma-buf/sync_file: free fences array in num_fences is 1
  drm/i2c: tda998x: don't register the connector
  drm: Don't swallow error codes in drm_dev_alloc()
  drm: Distinguish no name from ENOMEM in set_unique()
  drm: Remove dirty property from docs
  drm/doc: Document color space handling
  drm: Extract drm_color_mgmt.[hc]
  drm/doc: Polish plane composition property docs
  drm: Conslidate blending properties in drm_blend.[hc]
  drm/doc: Polish for drm_plane.[hc]
  drm: Extract drm_plane.[hc]
  drm/tilcdc: Add atomic and crtc headers to crtc.c
  drm: Fix typo in encoder docs
  ...

7 years agoMerge tag 'drm-amdkfd-next-2016-09-19' of git://people.freedesktop.org/~gabbayo/linux...
Dave Airlie [Wed, 28 Sep 2016 00:25:42 +0000 (10:25 +1000)]
Merge tag 'drm-amdkfd-next-2016-09-19' of git://people.freedesktop.org/~gabbayo/linux into drm-next

This is amdkfd's pull request for kernel 4.9. It contains a fix to a possible
infinite loop bug and a couple of other minor "cleaning" patches.

* tag 'drm-amdkfd-next-2016-09-19' of git://people.freedesktop.org/~gabbayo/linux:
  drm/amdkfd: Pass 'struct queue_propertices' by reference
  drm/amdkfd: Unify multiple calls to pr_debug() into one
  drm/amdkfd: Fix possible infinite loop
  drm/amdkfd: Reuse function to find a process through pasid
  drm/amdkfd: Add some missing memset zero'ing in queue init func
  drm/amdkfd: Tidy up kfd_generate_gpu_id() uint64_t bitshift unpack

7 years agodrm: bridge: analogix/dp: mark symbols static where possible
Baoyou Xie [Sun, 25 Sep 2016 07:32:44 +0000 (15:32 +0800)]
drm: bridge: analogix/dp: mark symbols static where possible

We get 2 warnings when building kernel with W=1:
drivers/gpu/drm/bridge/analogix/analogix_dp_core.c:1053:5: warning: no previous prototype for 'analogix_dp_get_modes' [-Wmissing-prototypes]
drivers/gpu/drm/bridge/analogix/analogix_dp_core.c:1097:1: warning: no previous prototype for 'analogix_dp_detect' [-Wmissing-prototypes]

In fact, both functions are only used in the file in which they are
declared and don't need a declaration, but can be made static.
So this patch marks both functions with 'static'.

Signed-off-by: Baoyou Xie <baoyou.xie@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1474788764-6069-1-git-send-email-baoyou.xie@linaro.org
7 years agodrm/bochs: mark bochs_connector_get_modes() static
Baoyou Xie [Sun, 25 Sep 2016 07:25:34 +0000 (15:25 +0800)]
drm/bochs: mark bochs_connector_get_modes() static

We get 1 warning when building kernel with W=1:
drivers/gpu/drm/bochs/bochs_kms.c:181:5: warning: no previous prototype for 'bochs_connector_get_modes' [-Wmissing-prototypes]

In fact, this function is only used in the file in which it is
declared and don't need a declaration, but can be made static.
So this patch marks it 'static'.

Signed-off-by: Baoyou Xie <baoyou.xie@linaro.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1474788334-11007-1-git-send-email-baoyou.xie@linaro.org
7 years agodrm/bridge: analogix_dp: Improve panel on time
Sean Paul [Wed, 7 Sep 2016 11:23:15 +0000 (07:23 -0400)]
drm/bridge: analogix_dp: Improve panel on time

In order to reduce the time required to turn on the panel, this patch
makes 2 assumptions:
1- In detect(): if there's a panel, we're connected.
2- In get_modes(): if there's a panel, let the panel driver decide if
   it should prepare/unprepare in order to get the modes.

The first is straightforward, and shouldn't need further explanation. The
second should eliminate the prepare/unprepare delays from get_modes() in
most cases, since panels generally hardcode their modes in the driver as
opposed to reading EDID. If a panel does need to read EDID, it should be
responsible for ensuring it's in a state in which it can.

Signed-off-by: Sean Paul <seanpaul@chromium.org>
7 years agodrm/bridge: analogix_dp: Don't read EDID if panel present
Sean Paul [Wed, 24 Aug 2016 14:58:26 +0000 (10:58 -0400)]
drm/bridge: analogix_dp: Don't read EDID if panel present

If there's a panel connected to the analogix_dp bridge, rely on
the panel driver for modes, rather than reading EDID *and* calling
get_modes() on the panel.

This allows panels with a valid EDID to read it in the panel driver
(e.g. simple_panel), and panels with invalid EDID to homebrew modes
in their get_modes implementation.

Signed-off-by: Sean Paul <seanpaul@chromium.org>
7 years agodrm/bridge: analogix_dp: Remove duplicated code
Tomeu Vizoso [Wed, 24 Aug 2016 14:54:41 +0000 (10:54 -0400)]
drm/bridge: analogix_dp: Remove duplicated code

Remove code for reading the EDID and DPCD fields and use the helpers
instead.

Besides the obvious code reduction, other helpers are being added to the
core that could be used in this driver and will be good to be able to
use them instead of duplicating them.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Tested-by: Javier Martinez Canillas <javier@osg.samsung.com>
Tested-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Cc: Javier Martinez Canillas <javier@osg.samsung.com>
Cc: Mika Kahola <mika.kahola@intel.com>
Cc: Yakir Yang <ykk@rock-chips.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Archit Taneja <architt@codeaurora.org>
7 years agodrm/tilcdc: fix wrong error handling
Daniel Schultz [Fri, 23 Sep 2016 10:52:49 +0000 (12:52 +0200)]
drm/tilcdc: fix wrong error handling

When 'component_bind_all' fails it should not try to unbind components
in the error handling. This will produce a null pointer kernel panic when
no component exist.

This patch changes the order of the error handling. Now, it will only
unbind components if the are bound. Otherwise, the module will jump to
an error label below.

Signed-off-by: Daniel Schultz <d.schultz@phytec.de>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Jyri Sarha <jsarha@ti.com>
7 years agodrm/tilcdc: Return directly after a failed kfree_table_init() in tilcdc_convert_slave...
Markus Elfring [Thu, 22 Sep 2016 07:29:23 +0000 (09:29 +0200)]
drm/tilcdc: Return directly after a failed kfree_table_init() in tilcdc_convert_slave_node()

Return directly after a memory allocation failed in this function
at the beginning.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Jyri Sarha <jsarha@ti.com>
7 years agoRevert "drm/i2c: tda998x: don't register the connector"
Sean Paul [Fri, 23 Sep 2016 07:07:52 +0000 (00:07 -0700)]
Revert "drm/i2c: tda998x: don't register the connector"

This reverts commit 6a2925ea12006911c8180a89feda6d040873ed18.

commit 6a2925ea12006911c8180a89feda6d040873ed18
Author: Brian Starkey <brian.starkey@arm.com>
Date:   Mon Jul 25 11:55:48 2016 +0100

    drm/i2c: tda998x: don't register the connector

[seanpaul]
Patch isn't fully baked, and still causing problems. Revert
until this is sorted.

Acked-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
7 years agodrm: Fix plane type uabi breakage
Daniel Vetter [Fri, 23 Sep 2016 06:35:25 +0000 (08:35 +0200)]
drm: Fix plane type uabi breakage

Turns out assuming that only stuff in uabi is uabi is a bit naive, and
we have a bunch of properties for which the enum values are placed in
random headers. A proper fix would be to split out uapi include
headers, but meanwhile sprinkle at least some warning over them.

Fixes: 532b36712ddf ("drm/doc: Polish for drm_plane.[hc]")
Cc: Archit Taneja <architt@codeaurora.org>
Cc: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1474612525-9488-1-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/tilcdc: Remove "default" from blue-and-red-wiring property binding
Jyri Sarha [Mon, 12 Sep 2016 15:45:30 +0000 (18:45 +0300)]
drm/tilcdc: Remove "default" from blue-and-red-wiring property binding

Remove "default" keyword from blue-and-red-wiring devicetree property
binding document. The code does not support and there is no intention
to support it.

Reported-by: Rob Herring <robh@kernel.org>
Signed-off-by: Jyri Sarha <jsarha@ti.com>
7 years agodrm/tilcdc: Fix non static symbol warning
Wei Yongjun [Sat, 10 Sep 2016 12:32:57 +0000 (12:32 +0000)]
drm/tilcdc: Fix non static symbol warning

Fixes the following sparse warning:

drivers/gpu/drm/tilcdc/tilcdc_drv.c:64:5: warning:
 symbol 'tilcdc_atomic_check' was not declared. Should it be static?

Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Jyri Sarha <jsarha@ti.com>
7 years agodrm/tilcdc: mark symbols static where possible
Baoyou Xie [Thu, 8 Sep 2016 11:29:24 +0000 (19:29 +0800)]
drm/tilcdc: mark symbols static where possible

We get 3 warnings when building kernel with W=1:
drivers/gpu/drm/tilcdc/tilcdc_slave_compat.c:142:29: warning: no previous prototype for 'tilcdc_get_overlay' [-Wmissing-prototypes]
drivers/gpu/drm/tilcdc/tilcdc_slave_compat.c:198:13: warning: no previous prototype for 'tilcdc_convert_slave_node' [-Wmissing-prototypes]
drivers/gpu/drm/tilcdc/tilcdc_slave_compat.c:264:12: warning: no previous prototype for 'tilcdc_slave_compat_init' [-Wmissing-prototypes]

In fact, these functions are only used in the file in which they are
declared and don't need a declaration, but can be made static.
So this patch marks these functions with 'static'.

Signed-off-by: Baoyou Xie <baoyou.xie@linaro.org>
Signed-off-by: Jyri Sarha <jsarha@ti.com>
7 years agodrm/tilcdc: add missing header dependencies
Baoyou Xie [Thu, 8 Sep 2016 11:08:56 +0000 (19:08 +0800)]
drm/tilcdc: add missing header dependencies

We get 4 warnings when building kernel with W=1:
drivers/gpu/drm/tilcdc/tilcdc_tfp410.c:393:12: warning: no previous prototype for 'tilcdc_tfp410_init' [-Wmissing-prototypes]
drivers/gpu/drm/tilcdc/tilcdc_tfp410.c:398:13: warning: no previous prototype for 'tilcdc_tfp410_fini' [-Wmissing-prototypes]
drivers/gpu/drm/tilcdc/tilcdc_panel.c:443:12: warning: no previous prototype for 'tilcdc_panel_init' [-Wmissing-prototypes]
drivers/gpu/drm/tilcdc/tilcdc_panel.c:448:13: warning: no previous prototype for 'tilcdc_panel_fini' [-Wmissing-prototypes]

In fact, these functions are declared in
drivers/gpu/drm/tilcdc/tilcdc_tfp410.h,
drivers/gpu/drm/tilcdc/tilcdc_panel.h,
so this patch adds missing header dependencies.

Signed-off-by: Baoyou Xie <baoyou.xie@linaro.org>
Signed-off-by: Jyri Sarha <jsarha@ti.com>
7 years agodma-buf/sync_file: free fences array in num_fences is 1
Gustavo Padovan [Wed, 21 Sep 2016 07:20:19 +0000 (10:20 +0300)]
dma-buf/sync_file: free fences array in num_fences is 1

When merging sync_files there is a case when we can end up with only one
fence in the merged sync_file: when all fences belong to the same
timeline.

So for this case a fence_array is not created instead we just assigned the
fence to sync_file->fence. Then we do not use the fences array anymore nor
does free it.

This patch frees the array.

Reported-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1474442419-6040-1-git-send-email-gustavo@padovan.org
7 years agodrm/i2c: tda998x: don't register the connector
Brian Starkey [Mon, 25 Jul 2016 10:55:48 +0000 (11:55 +0100)]
drm/i2c: tda998x: don't register the connector

The connector shouldn't be registered until the rest of the whole device
is set up, so that consistent state is presented to userspace.

As drm_dev_register() now registers all of the connectors anyway,
there's no need to explicitly do it in individual drivers so remove
the calls to drm_connector_register()/drm_connector_unregister().

This allows componentised drivers to use tda998x without having racy
initialisation.

Signed-off-by: Brian Starkey <brian.starkey@arm.com>
Reviewed-by: Liviu Dudau <Liviu.Dudau@arm.com>
Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1469444148-19003-1-git-send-email-brian.starkey@arm.com
Link: http://patchwork.freedesktop.org/patch/msgid/545C835D.1050008@arm.com
7 years agodrm: Don't swallow error codes in drm_dev_alloc()
Tom Gundersen [Wed, 21 Sep 2016 14:59:19 +0000 (16:59 +0200)]
drm: Don't swallow error codes in drm_dev_alloc()

There are many reasons other than ENOMEM that drm_dev_init() can
fail. Return ERR_PTR rather than NULL to be able to distinguish
these in the caller.

Signed-off-by: Tom Gundersen <teg@jklm.no>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20160921145919.13754-2-teg@jklm.no
7 years agodrm: Distinguish no name from ENOMEM in set_unique()
Tom Gundersen [Wed, 21 Sep 2016 14:59:18 +0000 (16:59 +0200)]
drm: Distinguish no name from ENOMEM in set_unique()

If passing name == NULL to drm_drv_set_unique() we now get -ENOMEM
as kstrdup() returns NULL. Instead check for this explicitly and
return -EINVAL if no name is provided.

Signed-off-by: Tom Gundersen <teg@jklm.no>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20160921145919.13754-1-teg@jklm.no
7 years agodrm/sun4i: Fix the high buffer address mask
Maxime Ripard [Mon, 19 Sep 2016 19:53:08 +0000 (21:53 +0200)]
drm/sun4i: Fix the high buffer address mask

The highest 3bits of the 4 layers buffers are all part of the same
register. However, our mask computation was wrong, leading to all the
lowest register bits being removed when we use regmap_update_bits, which
will lead to the buffers being set to some random part of the RAM.

Fix our mask.

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
7 years agodrm/sun4i: tv: Check mode pointer
Maxime Ripard [Mon, 19 Sep 2016 20:17:51 +0000 (22:17 +0200)]
drm/sun4i: tv: Check mode pointer

The drm_mode_create call might return NULL in case of a failure, and the
current code doesn't check for that. Make sure it does.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
7 years agodrm/sun4i: Fix formats usable by the primary plane
Maxime Ripard [Mon, 19 Sep 2016 20:17:50 +0000 (22:17 +0200)]
drm/sun4i: Fix formats usable by the primary plane

Even though all our planes can support the ARGB formats, the lowest plane
(ie the primary plane) cannot use the alpha component, otherwise it will
just result in the transparent area being entirely black.

Since some applications will still require the ARGB format, let's force the
format to XRGB to drop the alpha component entirely.

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
7 years agodrm: Remove dirty property from docs
Daniel Vetter [Wed, 21 Sep 2016 08:59:30 +0000 (10:59 +0200)]
drm: Remove dirty property from docs

We removed it in

commit 6ab10b76ff6252bd9be0849c40f5865e39a29961
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Fri Aug 12 22:48:45 2016 +0200

    drm/kms: Nuke dirty_info property

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1474448370-32227-7-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/doc: Document color space handling
Daniel Vetter [Wed, 21 Sep 2016 08:59:29 +0000 (10:59 +0200)]
drm/doc: Document color space handling

Again move it from the unmaintainable csv into DOC free-form overview
sections.

v2: Types Lionel&Sean spotted.

Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1474448370-32227-6-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm: Extract drm_color_mgmt.[hc]
Daniel Vetter [Wed, 21 Sep 2016 08:59:28 +0000 (10:59 +0200)]
drm: Extract drm_color_mgmt.[hc]

For both the new degamm/lut/gamma atomic combo, and the old legacy
gamma tables.

Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1474448370-32227-5-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/doc: Polish plane composition property docs
Daniel Vetter [Wed, 21 Sep 2016 08:59:27 +0000 (10:59 +0200)]
drm/doc: Polish plane composition property docs

Try to spec a bit more precisely how they all fit together, now that
at least the code is for all the additional properties is in one
place.

Also remove the entries for the standardized properties from the
table, because that thing is supremely unmaintaineable.

v2: Fix typos Sean spotted.

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: Sean Paul <seanpaul@chromium.org>
Cc: Benjamin Gaignard <benjamin.gaignard@linaro.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1474448370-32227-4-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm: Conslidate blending properties in drm_blend.[hc]
Daniel Vetter [Wed, 21 Sep 2016 08:59:26 +0000 (10:59 +0200)]
drm: Conslidate blending properties in drm_blend.[hc]

Imo zpos, rotatation, blending eq (once we have it) and all that
should be in drm_blend.c, since those are all about how exactly the
pixels are rendered onto the CRTC's visible area. Also noticed that
one exported function accidentally ended up in drm_crtc_internal.h,
move it to the right place too.

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1474448370-32227-3-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm/doc: Polish for drm_plane.[hc]
Daniel Vetter [Wed, 21 Sep 2016 08:59:25 +0000 (10:59 +0200)]
drm/doc: Polish for drm_plane.[hc]

Big thing is untangling and carefully documenting the different uapi
types of planes. I also sprinkled a few more cross references around
to make this easier to discover.

As usual, remove the kerneldoc for internal functions which are not
exported. Aside: We should probably go OCD on all the ioctl handlers
and consistenly give them an _ioctl postfix.

Acked-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1474448370-32227-2-git-send-email-daniel.vetter@ffwll.ch
7 years agodrm: Extract drm_plane.[hc]
Daniel Vetter [Wed, 21 Sep 2016 08:59:24 +0000 (10:59 +0200)]
drm: Extract drm_plane.[hc]

Just pure code movement, cleanup and polish will happen in later
patches.

v2: Don't forget all the ioctl! To extract those cleanly I decided to
put check_src_coords into drm_framebuffer.c (and give it a
drm_framebuffer_ prefix), since that just checks framebuffer
constraints.

v3: rebase over PAGE_FLIP_TARGET.

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
[seanpaul]
This patch as posted on the list was rebased on:

commit 6f00975c619064a18c23fd3aced325ae165a73b9
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Sat Aug 20 12:22:11 2016 +0200

    drm: Reject page_flip for !DRIVER_MODESET

so as a result of moving the page_flip ioctl, this fix has
been rolled into this patch.

Signed-off-by: Sean Paul <seanpaul@chromium.org>
7 years agodrm/tilcdc: Add atomic and crtc headers to crtc.c
Sean Paul [Wed, 21 Sep 2016 13:14:53 +0000 (06:14 -0700)]
drm/tilcdc: Add atomic and crtc headers to crtc.c

Also reorder alphabetically and fix up drm_flip_work header.

Reviewed-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
7 years agodrm/rockchip: Balance irq refcount on failure
Sean Paul [Fri, 16 Sep 2016 18:22:03 +0000 (14:22 -0400)]
drm/rockchip: Balance irq refcount on failure

If create_crtc fails in vop bind, ensure the irq refcount is zeroed
back out before exiting.

Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
7 years agodrm/rockchip: Kill vop_plane_state
Tomasz Figa [Wed, 14 Sep 2016 12:55:01 +0000 (21:55 +0900)]
drm/rockchip: Kill vop_plane_state

After changes introduced by last patches, there is no useful data stored
in vop_plane_state struct.  Let's remove it and make the driver use
generic plane state alone.

Signed-off-by: Tomasz Figa <tfiga@chromium.org>
7 years agodrm/rockchip: Always signal event in next vblank after cfg_done
Tomasz Figa [Wed, 14 Sep 2016 12:55:00 +0000 (21:55 +0900)]
drm/rockchip: Always signal event in next vblank after cfg_done

This patch makes the driver send the pending vblank event in next vblank
following the commit, relying on vblank signalling improvements done in
previous patches. This gives us vblank events that always represent the
real moment of changes hitting on the screen (which was the case only
for complete FB changes before) and lets us remove the manual window
update check.

Signed-off-by: Tomasz Figa <tfiga@chromium.org>
7 years agodrm/rockchip: Do not enable vblank without event
Tomasz Figa [Wed, 14 Sep 2016 12:54:59 +0000 (21:54 +0900)]
drm/rockchip: Do not enable vblank without event

Originally we needed to enable vblank for any atomic commit to kick the
PSR machine, but that was changed and we no longer need to do so from
a vblank interrupt. Let's return to original behavior of enabling
vblank only if it is really necessary.

This essentially reverts commit 5b6804034ae9 ("drm/rockchip: Enable
vblank without event").

Signed-off-by: Tomasz Figa <tfiga@chromium.org>
7 years agodrm/rockchip: Replace custom wait_for_vblanks with helper
Tomasz Figa [Wed, 14 Sep 2016 12:54:58 +0000 (21:54 +0900)]
drm/rockchip: Replace custom wait_for_vblanks with helper

Currently the driver uses a custom function to wait for flip to complete
after an atomic commit. It was needed before because of two problems:
 - there is no hardware vblank counter, so the original helper would
   have a race condition with the vblank interrupt,
 - the driver didn't support unreferencing cursor framebuffers
   asynchronously to the commit, which was what the helper expected.
Since both problems have been solved by previous patches, we can now
make the driver use the generic helper and remove custom waiting code.

Signed-off-by: Tomasz Figa <tfiga@chromium.org>
7 years agodrm/rockchip: Unreference framebuffers from flip work
Tomasz Figa [Wed, 14 Sep 2016 12:54:57 +0000 (21:54 +0900)]
drm/rockchip: Unreference framebuffers from flip work

Currently the driver waits for vblank and then unreferences old
framebuffers from atomic commit code path. This is however breaking the
legacy cursor API, which requires the updates to be fully asynchronous.
Instead of just adding a special case for cursor, we can have actually
smaller amount of code to unreference any changed framebuffer from a
flip work.

Signed-off-by: Tomasz Figa <tfiga@chromium.org>
7 years agodrm/rockchip: Avoid race with vblank count increment
Tomasz Figa [Wed, 14 Sep 2016 12:54:56 +0000 (21:54 +0900)]
drm/rockchip: Avoid race with vblank count increment

Since VOP does not have a hardware vblank count register, the ongoing
commit might be racing with a requested vblank interrupt, which would
increment the software vblank counter before the changes being committed
actually happen.

To avoid this, we can extend .atomic_flush(), so after it sets cfg_done
bit, it polls the vblank interrupt bit until it's inactive to make sure
that any old vblank interrupt gets to the handler and then uses
synchronize_irq(vop->irq) to make sure the handler finishes running.

The polling case should happen very rarely, but even if, the total wait
time should be relatively low and in practice almost equal to the vop
hardirq handler running time.

Signed-off-by: Tomasz Figa <tfiga@chromium.org>
7 years agodrm/rockchip: Get rid of some unnecessary code
Tomasz Figa [Wed, 14 Sep 2016 12:54:55 +0000 (21:54 +0900)]
drm/rockchip: Get rid of some unnecessary code

Current code implements prepare_fb and cleanup_fb callbacks only to
grab/release fb references, which is already done by atomic framework
when creating/destryoing plane state. Let's remove these
unused bits.

Signed-off-by: Tomasz Figa <tfiga@chromium.org>
7 years agodrm/rockchip: Clear interrupt status bits before enabling
Tomasz Figa [Wed, 14 Sep 2016 12:54:54 +0000 (21:54 +0900)]
drm/rockchip: Clear interrupt status bits before enabling

The enable register only masks the raw status bits to signal CPU
interrupt only for enabled interrupts. The status bits are activated
regardless of the enable register. This means that we might have an old
interrupt event queued, which we are not interested in. To avoid getting
a spurious interrupt signalled, we have to clear the old bit before we
update the enable register.

Signed-off-by: Tomasz Figa <tfiga@chromium.org>
7 years agodrm/rockchip: Fix up bug in psr state machine
Sean Paul [Tue, 6 Sep 2016 18:11:53 +0000 (14:11 -0400)]
drm/rockchip: Fix up bug in psr state machine

The ->set() callback would always be called when transitioning
from FLUSH->DISABLE since we assign state to psr->state right
above the skip condition.

Reported-by: Daniel Kurtz <djkurtz@chromium.org>
Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
7 years agodrm/bridge: analogix_dp: Remove duplicated code
Tomeu Vizoso [Fri, 5 Aug 2016 12:59:03 +0000 (14:59 +0200)]
drm/bridge: analogix_dp: Remove duplicated code

Remove code for reading the EDID and DPCD fields and use the helpers
instead.

Besides the obvious code reduction, other helpers are being added to the
core that could be used in this driver and will be good to be able to
use them instead of duplicating them.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Tested-by: Javier Martinez Canillas <javier@osg.samsung.com>
Tested-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Yakir Yang <ykk@rock-chips.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Cc: Javier Martinez Canillas <javier@osg.samsung.com>
Cc: Mika Kahola <mika.kahola@intel.com>
Cc: Yakir Yang <ykk@rock-chips.com>
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Archit Taneja <architt@codeaurora.org>
7 years agodrm/rockchip: Reduce psr flush time to 100ms
Sean Paul [Thu, 18 Aug 2016 19:03:22 +0000 (12:03 -0700)]
drm/rockchip: Reduce psr flush time to 100ms

3 seconds is a bit too conservative, drop this to 100ms for
better power savings.

Reviewed-by: Yakir Yang <ykk@rock-chips.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
7 years agodrm/rockchip: Don't key off vblank for psr
Sean Paul [Thu, 18 Aug 2016 19:01:46 +0000 (12:01 -0700)]
drm/rockchip: Don't key off vblank for psr

Instead of keying off vblank for psr, just flush every time
we get an atomic update. This ensures that cursor updates
will properly disable psr (without turning vblank on/off),
and unifies the paths between fb_dirty and atomic psr
enable/disable.

Reviewed-by: Yakir Yang <ykk@rock-chips.com>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
7 years agodrm: Fix typo in encoder docs
Dhinakaran Pandiyan [Mon, 19 Sep 2016 22:40:48 +0000 (15:40 -0700)]
drm: Fix typo in encoder docs

Corrected typo in bridge and encoder comparison. Also, added a one-line
encoder description from the previous documentation.

Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1474324848-6446-1-git-send-email-dhinakaran.pandiyan@intel.com
7 years agoGPU-DRM: Replace a kzalloc() call by kcalloc() in drm_legacy_addbufs_sg()
Markus Elfring [Mon, 19 Sep 2016 15:30:31 +0000 (17:30 +0200)]
GPU-DRM: Replace a kzalloc() call by kcalloc() in drm_legacy_addbufs_sg()

The script "checkpatch.pl" can point information out like the following.

WARNING: Prefer kcalloc over kzalloc with multiply

Thus fix the affected source code place.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/b340d1ca-f114-a523-e6d1-afe7fd6c3826@users.sourceforge.net
7 years agoGPU-DRM: Replace a kzalloc() call by kcalloc() in drm_legacy_addbufs_agp()
Markus Elfring [Mon, 19 Sep 2016 15:24:20 +0000 (17:24 +0200)]
GPU-DRM: Replace a kzalloc() call by kcalloc() in drm_legacy_addbufs_agp()

The script "checkpatch.pl" can point information out like the following.

WARNING: Prefer kcalloc over kzalloc with multiply

Thus fix the affected source code place.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/606c22f3-5da5-2e84-783d-bfe5289b4a01@users.sourceforge.net
7 years agoGPU-DRM: Replace two kzalloc() calls by kcalloc() in drm_legacy_addbufs_pci()
Markus Elfring [Mon, 19 Sep 2016 15:17:34 +0000 (17:17 +0200)]
GPU-DRM: Replace two kzalloc() calls by kcalloc() in drm_legacy_addbufs_pci()

The script "checkpatch.pl" can point information out like the following.

WARNING: Prefer kcalloc over kzalloc with multiply

Thus fix the affected source code places.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/51f88f44-130d-48ac-f531-32ca9ae3919f@users.sourceforge.net
7 years agoGPU-DRM: Use kmalloc_array() in drm_legacy_addbufs_pci()
Markus Elfring [Mon, 19 Sep 2016 15:07:06 +0000 (17:07 +0200)]
GPU-DRM: Use kmalloc_array() in drm_legacy_addbufs_pci()

A multiplication for the size determination of a memory allocation
indicated that an array data structure should be processed.
Thus use the corresponding function "kmalloc_array".

This issue was detected by using the Coccinelle software.

Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/b71c8fee-8e84-9f1b-8569-f1ae8b879cc5@users.sourceforge.net
7 years agodrm/msm: bump kernel api version for explicit fencing
Rob Clark [Fri, 16 Sep 2016 22:37:44 +0000 (18:37 -0400)]
drm/msm: bump kernel api version for explicit fencing

Signed-off-by: Rob Clark <robdclark@gmail.com>
7 years agoMerge branch 'for-next' of http://git.agner.ch/git/linux-drm-fsl-dcu into drm-next
Dave Airlie [Tue, 20 Sep 2016 20:28:25 +0000 (06:28 +1000)]
Merge branch 'for-next' of git.agner.ch/git/linux-drm-fsl-dcu into drm-next

fsl-dcu fixes.

* 'for-next' of http://git.agner.ch/git/linux-drm-fsl-dcu:
  drm/fsl-dcu: disable clock on error path
  drm/fsl-dcu: use PTR_ERR_OR_ZERO() to simplify the code
  drm/fsl-dcu: fix endian issue when using clk_register_divider

7 years agodrm/sti: remove stih415-416 platform support
Vincent Abriou [Tue, 20 Sep 2016 13:03:33 +0000 (15:03 +0200)]
drm/sti: remove stih415-416 platform support

stih415 and stih416 platform are obsolete and no more supported.
Only stih407 and stih410 platform are maintained.

Signed-off-by: Vincent Abriou <vincent.abriou@st.com>
Acked-by: Benjamin Gaignard <benjamin.gaignard@linaro.org>
Acked-by: Peter Griffin <peter.griffin@linaro.org>
7 years agodma-buf/sync_file: fix documentation error
Emilio López [Mon, 19 Sep 2016 04:21:20 +0000 (01:21 -0300)]
dma-buf/sync_file: fix documentation error

The ioctl name and description on the documentation block don't
match the ioctl being defined. This was probably overlooked while
renaming the ioctls during the sync file destaging. This patch
provides a more accurate description of what the ioctl actually does.

Signed-off-by: Emilio López <emilio.lopez@collabora.co.uk>
Reviewed-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20160919042120.6280-1-emilio.lopez@collabora.co.uk
7 years agodma-buf/sync_file: Increment refcount of fence when all are signaled.
Rafael Antognolli [Thu, 15 Sep 2016 19:14:25 +0000 (12:14 -0700)]
dma-buf/sync_file: Increment refcount of fence when all are signaled.

When we merge several fences, if all of them are signaled already, we
still keep one of them. So instead of using add_fence(), which will not
increase the refcount of signaled fences, we should explicitly call
fence_get() for the fence we are keeping.

This patch fixes a kernel panic that can be triggered by creating a fence
that is expired (or increasing the timeline until it expires), then
creating a merged fence out of it, and deleting the merged fence. This
will make the original expired fence's refcount go to zero.

Testcase: igt/sw_sync/sync_expired_merge
Signed-off-by: Rafael Antognolli <rafael.antognolli@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
Link: http://patchwork.freedesktop.org/patch/msgid/1473966865-4508-1-git-send-email-rafael.antognolli@intel.com
7 years agodrm/sti: fix compositor debugfs creation
Vincent Abriou [Thu, 15 Sep 2016 15:11:07 +0000 (17:11 +0200)]
drm/sti: fix compositor debugfs creation

Fix typo and issue while creating the vid and mixer debugfs entries.

Signed-off-by: Vincent Abriou <vincent.abriou@st.com>
Acked-by: Benjamin Gaignard <benjamin.gaignard@linaro.org>
7 years agodrm/sti: use valid video mode
Fabien Dessenne [Tue, 6 Sep 2016 07:42:53 +0000 (09:42 +0200)]
drm/sti: use valid video mode

In atomic mode the crtc_xxx (eg crtc_hdisplay) members of the mode
structure may be unset before calling atomic_check/commit for planes.
Instead of, use xxx members which are actually set.

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Acked-by: Vincent Abriou <vincent.abriou@st.com>
7 years agodrm/sti: in crtc_atomic_flush, enable only planes of this crtc
Fabien Dessenne [Tue, 6 Sep 2016 07:42:39 +0000 (09:42 +0200)]
drm/sti: in crtc_atomic_flush, enable only planes of this crtc

crtc_atomic_flush performs some additional processing, like plane
enable at mixer level.
Enable only the planes attached to the CRTC.

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Acked-by: Vincent Abriou <vincent.abriou@st.com>
7 years agodrm/sti: use vtg array instead of vtg_main/aux
Fabien Dessenne [Tue, 6 Sep 2016 07:42:25 +0000 (09:42 +0200)]
drm/sti: use vtg array instead of vtg_main/aux

This is more generic and more consistent with the other members of the
sti_compositor struct.

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Acked-by: Vincent Abriou <vincent.abriou@st.com>
7 years agodrm/sti: use different notifier_block for each pipe
Fabien Dessenne [Tue, 6 Sep 2016 07:42:13 +0000 (09:42 +0200)]
drm/sti: use different notifier_block for each pipe

Each pipe shall have its own notifier block to manage the vblank event.
This fixes issues where a client registered on given pipe is later
abusively notified of events on the other pipe.

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Acked-by: Vincent Abriou <vincent.abriou@st.com>
7 years agodrm/sti: fix atomic_disable check
Fabien Dessenne [Tue, 6 Sep 2016 07:41:48 +0000 (09:41 +0200)]
drm/sti: fix atomic_disable check

When a drm_plane is being disabled, its ->crtc member is set to NULL
before the .atomic_disable() func is called.
To get the crtc of the plane, read old_state->crtc instead of
drm_plane->crtc

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Acked-by: Vincent Abriou <vincent.abriou@st.com>
7 years agodrm/sti: run gdp init sequence only once
Fabien Dessenne [Tue, 6 Sep 2016 07:42:00 +0000 (09:42 +0200)]
drm/sti: run gdp init sequence only once

Do not rely on plane->status to define whether this is the first update
but rather check for gdp->vtg.
This avoids multiple and unwanted calls to sti_vtg_register_client()
which breaks the kernel scheduler.

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Acked-by: Vincent Abriou <vincent.abriou@st.com>
7 years agodrm/sti: run hqvdp init sequence only once
Fabien Dessenne [Wed, 24 Aug 2016 10:12:37 +0000 (12:12 +0200)]
drm/sti: run hqvdp init sequence only once

Do not rely on plane->status to define whether this is the first update
but rather check for hqvdp->xp70_initialized bit status.
This avoids multiple and unwanted calls to sti_vtg_register_client()
which breaks the kernel scheduler.

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Acked-by: Vincent Abriou <vincent.abriou@st.com>
7 years agodrm/sti: fix debug logs
Fabien Dessenne [Tue, 6 Sep 2016 07:41:35 +0000 (09:41 +0200)]
drm/sti: fix debug logs

Add some missing \n in logs.

Signed-off-by: Fabien Dessenne <fabien.dessenne@st.com>
Acked-by: Vincent Abriou <vincent.abriou@st.com>
7 years agodrm/sti: dpms function missing for HDMI connector
Vincent Abriou [Wed, 14 Sep 2016 08:14:23 +0000 (10:14 +0200)]
drm/sti: dpms function missing for HDMI connector

Signed-off-by: Vincent Abriou <vincent.abriou@st.com>
Acked-by: Benjamin Gaignard <benjamin.gaignard@linaro.org>
7 years agodrm/sti: Fix sparse warnings
Ville Syrjälä [Mon, 19 Sep 2016 13:33:53 +0000 (16:33 +0300)]
drm/sti: Fix sparse warnings

drm/sti/sti_mixer.c:361:6: warning: symbol 'sti_mixer_set_matrix' was not declared. Should it be static?
drm/sti/sti_gdp.c:476:5: warning: symbol 'sti_gdp_field_cb' was not declared. Should it be static?
drm/sti/sti_gdp.c:885:24: warning: symbol 'sti_gdp_plane_helpers_funcs' was not declared. Should it be static?
drm/sti/sti_cursor.c:348:24: warning: symbol 'sti_cursor_plane_helpers_funcs' was not declared. Should it be static?
drm/sti/sti_compositor.c:28:28: warning: symbol 'stih407_compositor_data' was not declared. Should it be static?
drm/sti/sti_compositor.c:49:28: warning: symbol 'stih416_compositor_data' was not declared. Should it be static?
drm/sti/sti_vtg.c:75:1: warning: symbol 'vtg_lookup' was not declared. Should it be static?
drm/sti/sti_vtg.c:476:24: warning: symbol 'sti_vtg_driver' was not declared. Should it be static?
drm/sti/sti_dvo.c:109:5: warning: symbol 'dvo_awg_generate_code' was not declared. Should it be static?
drm/sti/sti_dvo.c:602:24: warning: symbol 'sti_dvo_driver' was not declared. Should it be static?
drm/sti/sti_vtac.c:209:24: warning: symbol 'sti_vtac_driver' was not declared. Should it be static?
drm/sti/sti_tvout.c:914:24: warning: symbol 'sti_tvout_driver' was not declared. Should it be static?
drm/sti/sti_hqvdp.c:786:5: warning: symbol 'sti_hqvdp_vtg_cb' was not declared. Should it be static?
drm/sti/sti_hqvdp.c:1253:24: warning: symbol 'sti_hqvdp_plane_helpers_funcs' was not declared. Should it be static?
drm/sti/sti_hqvdp.c:1292:5: warning: symbol 'sti_hqvdp_bind' was not declared. Should it be static?
drm/sti/sti_hqvdp.c:1385:24: warning: symbol 'sti_hqvdp_driver' was not declared. Should it be static?
drm/sti/sti_drv.c:143:6: warning: symbol 'sti_drm_dbg_cleanup' was not declared. Should it be static?

Cc: Benjamin Gaignard <benjamin.gaignard@linaro.org>
Cc: Vincent Abriou <vincent.abriou@st.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Acked-by: Vincent Abriou <vincent.abriou@st.com>
7 years agodrm: sti: fix coccicheck warnings
Benjamin Gaignard [Wed, 24 Aug 2016 07:35:20 +0000 (09:35 +0200)]
drm: sti: fix coccicheck warnings

Fix four warnings shown by coccicheck.

Signed-off-by: Benjamin Gaignard <benjamin.gaignard@linaro.org>
Acked-by: Vincent Abriou <vincent.abriou@st.com>
7 years agodrm: sti: Replace drm_fb_get_bpp_depth() with drm_format_plane_cpp()
Laurent Pinchart [Wed, 8 Jun 2016 23:32:11 +0000 (02:32 +0300)]
drm: sti: Replace drm_fb_get_bpp_depth() with drm_format_plane_cpp()

The driver needs the number of bytes per pixel, not the bpp and depth
info meant for fbdev compatibility. Use the right API.

Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Vincent Abriou <vincent.abriou@st.com>
7 years agodrm/sun4i: dotclock: Round to closest clock rate
Chen-Yu Tsai [Thu, 15 Sep 2016 15:14:02 +0000 (23:14 +0800)]
drm/sun4i: dotclock: Round to closest clock rate

With display pixel clocks we want to have the closest possible clock
rate, to minimize timing and refresh rate skews. Whether the actual
clock rate is higher or lower than the requested rate is less important.

Also check candidates against the requested rate, rather than the
ideal parent rate, the varying dividers also influence the difference
between the requested rate and the rounded rate.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
7 years agoMerge tag 'imx-drm-next-2016-09-19' of git://git.pengutronix.de/git/pza/linux into...
Dave Airlie [Mon, 19 Sep 2016 20:24:26 +0000 (06:24 +1000)]
Merge tag 'imx-drm-next-2016-09-19' of git://git.pengutronix.de/git/pza/linux into drm-next

imx-drm active plane reconfiguration, cleanup, FSU/IC/IRT/VDIC support

- add active plane reconfiguration support (v4),
  use the atomic_disable callback
- stop calling disable_plane manually in the plane destroy path
- let mode cleanup destroy mode objects on driver unbind
- drop deprecated load/unload drm_driver ops
- add exclusive fence to plane state, so the atomic helper can
  wait on them, remove the open-coded fence wait from imx-drm
- add low level deinterlacer (VDIC) support
- add support for channel linking via the frame synchronisation unit (FSU)
- add queued image conversion support for memory-to-memory scaling, rotation,
  and color space conversion, using IC and IRT.

* tag 'imx-drm-next-2016-09-19' of git://git.pengutronix.de/git/pza/linux:
  gpu: ipu-v3: Add queued image conversion support
  gpu: ipu-v3: Add ipu_rot_mode_is_irt()
  gpu: ipu-v3: fix a possible NULL dereference
  drm/imx: parallel-display: detach bridge or panel on unbind
  drm/imx: imx-ldb: detach bridge on unbind
  drm/imx: imx-ldb: detach panel on unbind
  gpu: ipu-v3: Add FSU channel linking support
  gpu: ipu-v3: Add Video Deinterlacer unit
  drm/imx: add exclusive fence to plane state
  drm/imx: fold ipu_plane_disable into ipu_disable_plane
  drm/imx: don't destroy mode objects manually on driver unbind
  drm/imx: drop deprecated load/unload drm_driver ops
  drm/imx: don't call disable_plane in plane destroy path
  drm/imx: Add active plane reconfiguration support
  drm/imx: Use DRM_PLANE_COMMIT_NO_DISABLE_AFTER_MODESET flag
  drm/imx: ipuv3-crtc: Use the callback ->atomic_disable instead of ->disable
  gpu: ipu-v3: Do not wait for DMFC FIFO to clear when disabling DMFC channel

7 years agoMerge tag 'drm-intel-next-2016-09-19' of git://anongit.freedesktop.org/drm-intel...
Dave Airlie [Mon, 19 Sep 2016 20:23:22 +0000 (06:23 +1000)]
Merge tag 'drm-intel-next-2016-09-19' of git://anongit.freedesktop.org/drm-intel into drm-next

- refactor the sseu code (Imre)
- refine guc dmesg output (Dave Gordon)
- more vgpu work
- more skl wm fixes (Lyude)
- refactor dpll code in prep for upfront link training (Jim Bride et al)
- consolidate all platform feature checks into intel_device_info (Carlos Santa)
- refactor elsp/execlist submission as prep for re-submission after hang
  recovery and eventually scheduling (Chris Wilson)
- allow synchronous gpu reset handling, to remove tricky/impossible/fragile
  error recovery code (Chris Wilson)
- prep work for nonblocking (execlist) submission, using fences to track
  depencies and drive elsp submission (Chris Wilson)
- partial error recover/resubmission of non-guilty batches after hangs (Chris Wilson)
- full dma-buf implicit fencing support (Chris Wilson)
- dp link training fixes (Jim, Dhinkaran, Navare, ...)
- obey dp branch device pixel rate/bpc/clock limits (Mika Kahola), needed for
  many vga dongles
- bunch of small cleanups and polish all over, as usual

[airlied: printing macros collided]

* tag 'drm-intel-next-2016-09-19' of git://anongit.freedesktop.org/drm-intel: (163 commits)
  drm/i915: Update DRIVER_DATE to 20160919
  drm: Fix DisplayPort branch device ID kernel-doc
  drm/i915: use NULL for NULL pointers
  drm/i915: do not use 'false' as a NULL pointer
  drm/i915: make intel_dp_compute_bpp static
  drm: Add DP branch device info on debugfs
  drm/i915: Update bits per component for display info
  drm/i915: Check pixel rate for DP to VGA dongle
  drm/i915: Read DP branch device SW revision
  drm/i915: Read DP branch device HW revision
  drm/i915: Cleanup DisplayPort AUX channel initialization
  drm: Read DP branch device id
  drm: Helper to read max bits per component
  drm: Helper to read max clock rate
  drm: Drop VGA from bpc definitions
  drm: Add missing DP downstream port types
  drm/i915: Add ddb size field to device info structure
  drm/i915/guc: general tidying up (submission)
  drm/i915/guc: general tidying up (loader)
  drm/i915: clarify PMINTRMSK/pm_intr_keep usage
  ...

7 years agoMerge branch 'drm-next-4.9' of git://people.freedesktop.org/~agd5f/linux into drm...
Dave Airlie [Mon, 19 Sep 2016 20:17:38 +0000 (06:17 +1000)]
Merge branch 'drm-next-4.9' of git://people.freedesktop.org/~agd5f/linux into drm-next

More radeon and amdgpu changes for 4.9. Highlights:
- Initial SI support for amdgpu (controlled by a Kconfig option)
- misc ttm cleanups
- runtimepm fixes
- S3/S4 fixes
- power improvements
- lots of code cleanups and optimizations

* 'drm-next-4.9' of git://people.freedesktop.org/~agd5f/linux: (151 commits)
  drm/ttm: remove cpu_address member from ttm_tt
  drm/radeon/radeon_device: remove unused function
  drm/amdgpu: clean function declarations in amdgpu_ttm.c up
  drm/amdgpu: use the new ring ib and dma frame size callbacks (v2)
  drm/amdgpu/vce3: add ring callbacks for ib and dma frame size
  drm/amdgpu/vce2: add ring callbacks for ib and dma frame size
  drm/amdgpu/vce: add common ring callbacks for ib and dma frame size
  drm/amdgpu/uvd6: add ring callbacks for ib and dma frame size
  drm/amdgpu/uvd5: add ring callbacks for ib and dma frame size
  drm/amdgpu/uvd4.2: add ring callbacks for ib and dma frame size
  drm/amdgpu/sdma3: add ring callbacks for ib and dma frame size
  drm/amdgpu/sdma2.4: add ring callbacks for ib and dma frame size
  drm/amdgpu/cik_sdma: add ring callbacks for ib and dma frame size
  drm/amdgpu/si_dma: add ring callbacks for ib and dma frame size
  drm/amdgpu/gfx8: add ring callbacks for ib and dma frame size
  drm/amdgpu/gfx7: add ring callbacks for ib and dma frame size
  drm/amdgpu/gfx6: add ring callbacks for ib and dma frame size
  drm/amdgpu/ring: add an interface to get dma frame and ib size
  drm/amdgpu/sdma3: drop unused functions
  drm/amdgpu/gfx6: drop gds_switch callback
  ...

7 years agodrm/sun4i: Fix sparse warnings
Ville Syrjälä [Mon, 19 Sep 2016 13:33:54 +0000 (16:33 +0300)]
drm/sun4i: Fix sparse warnings

drm/sun4i/sun4i_tv.c:181:21: warning: symbol 'ntsc_video_levels' was not declared. Should it be static?
drm/sun4i/sun4i_tv.c:185:21: warning: symbol 'pal_video_levels' was not declared. Should it be static?
drm/sun4i/sun4i_tv.c:189:21: warning: symbol 'ntsc_burst_levels' was not declared. Should it be static?
drm/sun4i/sun4i_tv.c:193:21: warning: symbol 'pal_burst_levels' was not declared. Should it be static?
drm/sun4i/sun4i_tv.c:197:20: warning: symbol 'ntsc_color_gains' was not declared. Should it be static?
drm/sun4i/sun4i_tv.c:201:20: warning: symbol 'pal_color_gains' was not declared. Should it be static?
drm/sun4i/sun4i_tv.c:205:26: warning: symbol 'ntsc_resync_parameters' was not declared. Should it be static?
drm/sun4i/sun4i_tv.c:209:26: warning: symbol 'pal_resync_parameters' was not declared. Should it be static?
drm/sun4i/sun4i_tv.c:213:16: warning: symbol 'tv_modes' was not declared. Should it be static?

Cc: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
7 years agodrm/amdkfd: Pass 'struct queue_propertices' by reference
Edward O'Callaghan [Sat, 17 Sep 2016 05:01:45 +0000 (15:01 +1000)]
drm/amdkfd: Pass 'struct queue_propertices' by reference

Allow init_queue() to take 'struct queue_properties' by reference.

Signed-off-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
7 years agodrm/amdkfd: Unify multiple calls to pr_debug() into one
Edward O'Callaghan [Sat, 17 Sep 2016 05:01:44 +0000 (15:01 +1000)]
drm/amdkfd: Unify multiple calls to pr_debug() into one

Signed-off-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
7 years agodrm/amdkfd: Fix possible infinite loop
Edward O'Callaghan [Sat, 17 Sep 2016 05:01:43 +0000 (15:01 +1000)]
drm/amdkfd: Fix possible infinite loop

When the loop predicating timeout parameter passed happens to
not be a multiple of 20 the unsigned integer will overflow and
the loop will become unbounded.

Signed-off-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
7 years agodrm/amdkfd: Reuse function to find a process through pasid
Edward O'Callaghan [Sat, 17 Sep 2016 05:01:42 +0000 (15:01 +1000)]
drm/amdkfd: Reuse function to find a process through pasid

The kfd_lookup_process_by_pasid() is just for that purpose,
so use it instead of repeating the code.

v2: return on the condition (p == NULL) instead of BUG_ON(!p).

Signed-off-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
7 years agodrm/amdkfd: Add some missing memset zero'ing in queue init func
Edward O'Callaghan [Sat, 17 Sep 2016 05:01:41 +0000 (15:01 +1000)]
drm/amdkfd: Add some missing memset zero'ing in queue init func

Signed-off-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
7 years agodrm/amdkfd: Tidy up kfd_generate_gpu_id() uint64_t bitshift unpack
Edward O'Callaghan [Sat, 17 Sep 2016 05:01:40 +0000 (15:01 +1000)]
drm/amdkfd: Tidy up kfd_generate_gpu_id() uint64_t bitshift unpack

Dereference the one time and unpack the lower and upper 32bit
portions with the proper kernel helper macros.

Signed-off-by: Edward O'Callaghan <funfunctor@folklore1984.net>
Reviewed-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
7 years agodrm/fb-helper: Fix sparse warnings
Ville Syrjälä [Mon, 19 Sep 2016 13:33:44 +0000 (16:33 +0300)]
drm/fb-helper: Fix sparse warnings

drm/drm_fb_helper.c:2306:12: warning: symbol 'drm_fb_helper_modinit' was not declared. Should it be static?

While at it, move the lefover static inline to the right place.

Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1474292035-15695-4-git-send-email-ville.syrjala@linux.intel.com
7 years agodrm/blend: Fix sparse warnings
Ville Syrjälä [Mon, 19 Sep 2016 13:33:43 +0000 (16:33 +0300)]
drm/blend: Fix sparse warnings

drm/drm_blend.c:207:5: warning: symbol 'drm_atomic_normalize_zpos' was not declared. Should it be static?

Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Benjamin Gaignard <benjamin.gaignard@linaro.org>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1474292035-15695-3-git-send-email-ville.syrjala@linux.intel.com
7 years agodrm/atomic-helper: Fix sparse warnings
Ville Syrjälä [Mon, 19 Sep 2016 13:33:42 +0000 (16:33 +0300)]
drm/atomic-helper: Fix sparse warnings

drm/drm_atomic_helper.c:1696:6: warning: symbol 'plane_crtc_active' was not declared. Should it be static?

Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1474292035-15695-2-git-send-email-ville.syrjala@linux.intel.com
7 years agodrm: fix implicit declaration build error on ia64
Jani Nikula [Fri, 16 Sep 2016 10:06:36 +0000 (13:06 +0300)]
drm: fix implicit declaration build error on ia64

   drivers/gpu/drm/drm_dp_helper.c: In function 'drm_dp_downstream_debug':
>> drivers/gpu/drm/drm_dp_helper.c:551:2: error: implicit declaration of
>> function 'seq_printf' [-Werror=implicit-function-declaration]
     seq_printf(m, "\tDP branch device present: %s\n",
     ^
>> drivers/gpu/drm/drm_dp_helper.c:559:3: error: implicit declaration of
>> function 'seq_puts' [-Werror=implicit-function-declaration]
     seq_puts(m, "\t\tType: DisplayPort\n");
     ^

Reported-by: kbuild test robot <fengguang.wu@intel.com>
References: https://lists.freedesktop.org/archives/intel-gfx/2016-September/106638.html
Fixes: 80209e5f2c42 ("drm: Add DP branch device info on debugfs")
Cc: Mika Kahola <mika.kahola@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/1474020396-14875-1-git-send-email-jani.nikula@intel.com
7 years agodrm: Move all decl for drm_edid.c to drm_edid.h
Daniel Vetter [Wed, 31 Aug 2016 16:09:06 +0000 (18:09 +0200)]
drm: Move all decl for drm_edid.c to drm_edid.h

Some were still left in drm_crtc.h. Also include drm_edid.h in the
rst files.

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20160831160913.12991-4-daniel.vetter@ffwll.ch
7 years agodrm: Extract drm_bridge.h
Daniel Vetter [Wed, 31 Aug 2016 16:09:05 +0000 (18:09 +0200)]
drm: Extract drm_bridge.h

We don't want to burry the bridge structures kerneldoc in drm_crtc.h.

Cc: Archit Taneja <archit.taneja@gmail.com>
Reviewed-by: Archit Taneja <architt@codeaurora.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20160831160913.12991-3-daniel.vetter@ffwll.ch
7 years agodrm: Move a few macros away from drm_crtc.h
Daniel Vetter [Wed, 31 Aug 2016 16:09:04 +0000 (18:09 +0200)]
drm: Move a few macros away from drm_crtc.h

Now that there's less stuff in there I noticed that I overlooked them.
Sprinkle some docs over them while at it.

Reviewed-by: Sean Paul <seanpaul@chromium.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20160831160913.12991-2-daniel.vetter@ffwll.ch
7 years agodrm: cleanup drm_core_{init,exit}()
David Herrmann [Thu, 1 Sep 2016 12:48:37 +0000 (14:48 +0200)]
drm: cleanup drm_core_{init,exit}()

Various cleanups to the DRM core initialization and exit handlers:

 - Register chrdev last: Once register_chrdev() returns, open() will
   succeed on the given chrdevs. This is usually not an issue, as no
   chardevs are registered, yet. However, nodes can be created by
   user-space via mknod(2), even though such major/minor combinations are
   unknown to the kernel. Avoid calling into drm_stub_open() in those
   cases.
   Again, drm_stub_open() would just bail out as the inode is unknown,
   but it's really non-obvious if you hack on drm_stub_open().

 - Unify error-paths into just one label. All the error-path helpers can
   be called even though the constructors were not called yet, or failed.
   Hence, just call all cleanups unconditionally.

 - Call into drm_global_release(). This is a no-op, but provides
   debugging helpers in case there're GLOBALS left on module unload. This
   function was unused until now.

 - Use DRM_ERROR() instead of printk(), and also print the error-code on
   failure (even if it is static!).

 - Don't throw away error-codes of register_chrdev()!

 - Don't hardcode -1 as errno. This is just plain wrong.

 - Order exit-handlers in the exact reverse order of initialization
   (except if the order actually matters for syncing-reasons, which is
   not the case here, though).

v2:
 - Call drm_core_exit() directly from the init-error-handler. Requires to
   drop __exit annotation, though.

Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20160901124837.680-7-dh.herrmann@gmail.com
7 years agodrm: drop obsolete drm_core.h
David Herrmann [Thu, 1 Sep 2016 12:48:36 +0000 (14:48 +0200)]
drm: drop obsolete drm_core.h

The drm_core.h header contains a set of constants meant to be used
throughout DRM. However, as it turns out, they're each used just once and
don't bring any benefit. They're also grossly mis-named and lack
name-spacing. This patch inlines them, or moves them into drm_internal.h
as appropriate:

 - CORE_AUTHOR and CORE_DESC are inlined into corresponding MODULE_*()
   macros. It's just confusing having to follow 2 pointers when trying to
   find the definition of these fields. Grep'ping for MODULE_AUTHOR()
   should reveal the full information, if there's no strong reason not to.

 - CORE_NAME, CORE_DATE, CORE_MAJOR, CORE_MINOR, and CORE_PATCHLEVEL are
   inlined into the sysfs 'version' attribute. They're stripped
   everywhere else (which is just some printk() statements). CORE_NAME
   just doesn't make *any* sense, as we hard-code it in many places,
   anyway. The other constants are outdated and just serve
   binary-compatibility purposes. Hence, inline them in 'version' sysfs
   attribute (we might even try dropping it..).

 - DRM_IF_MAJOR and DRM_IF_MINOR are moved into drm_internal.h as they're
   only used by the global ioctl handlers. Furthermore, versioning
   interfaces breaks backports and as such is deprecated, anyway. We just
   keep them for historic reasons. I doubt anyone will ever modify them
   again.

Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20160901124837.680-6-dh.herrmann@gmail.com
7 years agodrm: use drm_file to tag vm-bos
David Herrmann [Thu, 1 Sep 2016 12:48:33 +0000 (14:48 +0200)]
drm: use drm_file to tag vm-bos

Rather than using "struct file*", use "struct drm_file*" as tag VM tag for
BOs. This will pave the way for "struct drm_file*" without any "struct
file*" back-pointer.

Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20160901124837.680-3-dh.herrmann@gmail.com
7 years agodrm: remove redundant drm_file->uid
David Herrmann [Thu, 1 Sep 2016 12:48:32 +0000 (14:48 +0200)]
drm: remove redundant drm_file->uid

Each DRM file-context caches the EUID of the process that opened the file.
It is used exclusively for debugging purposes in /proc/dri/ and friends.

Note, however, that we can already fetch the EUID from
priv->pid->task->creds. The pointer-chasing will not hurt us, since it is
only about debugging, anyway.

Since we already are in an rcu-read-side, we can use __task_cred() rather
than task_cred_xxx().

Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20160901124837.680-2-dh.herrmann@gmail.com
7 years agodrm/i915: Update DRIVER_DATE to 20160919
Daniel Vetter [Mon, 19 Sep 2016 07:26:08 +0000 (09:26 +0200)]
drm/i915: Update DRIVER_DATE to 20160919

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
7 years agoMerge tag 'topic/drm-misc-2016-09-19' of git://anongit.freedesktop.org/drm-intel...
Dave Airlie [Mon, 19 Sep 2016 07:16:02 +0000 (17:16 +1000)]
Merge tag 'topic/drm-misc-2016-09-19' of git://anongit.freedesktop.org/drm-intel into drm-next

Just random misc stuff that Sean/Sumit&Archit picked up while I relaxed.
Well except for one commit:

* tag 'topic/drm-misc-2016-09-19' of git://anongit.freedesktop.org/drm-intel:
  drm: Only use compat ioctl for addfb2 on X86/IA64
  drm/qxl: squash lines for simple wrapper functions
  drm/bridge: analogix_dp: squash lines for simple wrapper functions
  drm/radeon: squash lines for simple wrapper functions
  drm/amdgpu: squash lines for simple wrapper functions
  dma-buf/sync-file: Avoid enable fence signaling if poll(.timeout=0)
  drm/fence: allow fence waiting to be interrupted by userspace
  drm: Move property validation to a helper, v2.
  drm/bridge: adv7511: add support for the 2nd chip

7 years agogpu: ipu-v3: Add queued image conversion support
Steve Longerbeam [Sat, 17 Sep 2016 19:33:58 +0000 (12:33 -0700)]
gpu: ipu-v3: Add queued image conversion support

This patch implements image conversion support using the IC tasks, with
tiling to support scaling to and from images up to 4096x4096. Image
rotation is also supported. Image conversion requests are added to
a run queue under the IC tasks.

The internal API is subsystem agnostic (no V4L2 dependency except
for the use of V4L2 fourcc pixel formats).

Callers prepare for image conversion by calling
ipu_image_convert_prepare(), which initializes the parameters of
the conversion. The caller passes in the ipu and IC task to use for
the conversion, the input and output image formats, a rotation mode,
and a completion callback and completion context pointer:

struct ipu_image_converter_ctx *
ipu_image_convert_prepare(struct ipu_soc *ipu, enum ipu_ic_task ic_task,
                          struct ipu_image *in, struct ipu_image *out,
                          enum ipu_rotate_mode rot_mode,
                          ipu_image_converter_cb_t complete,
                          void *complete_context);

A new conversion context is created that is added to an IC task
context queue. The caller is given the new conversion context,
which can then be passed to the further APIs:

int ipu_image_convert_queue(struct ipu_image_converter_run *run);

This queues the given image conversion request run to a run queue,
and starts the conversion immediately if the run queue is empty. Only
the physaddr's of the input and output image buffers are needed,
since the conversion context was created previously with
ipu_image_convert_prepare(). When the conversion completes, the run
pointer is returned to the completion callback.

void ipu_image_convert_abort(struct ipu_image_converter_ctx *ctx);

This will abort any active or pending conversions for this context.
Any currently active or pending runs belonging to this context are
returned via the completion callback with an error status.

void ipu_image_convert_unprepare(struct ipu_image_converter_ctx *ctx);

Unprepares the conversion context. Any active or pending runs will
be aborted by calling ipu_image_convert_abort().

Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
7 years agogpu: ipu-v3: Add ipu_rot_mode_is_irt()
Steve Longerbeam [Sat, 17 Sep 2016 19:33:57 +0000 (12:33 -0700)]
gpu: ipu-v3: Add ipu_rot_mode_is_irt()

Add a macro that returns boolean true if the given ipu_rotate_mode
requires the use of the Image Rotator.

Signed-off-by: Steve Longerbeam <steve_longerbeam@mentor.com>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
7 years agogpu: ipu-v3: fix a possible NULL dereference
LABBE Corentin [Wed, 24 Aug 2016 08:17:17 +0000 (10:17 +0200)]
gpu: ipu-v3: fix a possible NULL dereference

of_match_device could return NULL, and so cause a NULL pointer
dereference later.

For fixing this problem, we use of_device_get_match_data(), this will
simplify the code a little by using a standard function for
getting the match data.

Testing the return value of of_device_get_match_data is also necessary
for avoiding a second NULL deref later on devtype.

Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
7 years agodrm/sun4i: dotclock: Allow divider = 127
Chen-Yu Tsai [Thu, 15 Sep 2016 15:14:01 +0000 (23:14 +0800)]
drm/sun4i: dotclock: Allow divider = 127

The dot clock divider is 7 bits wide, and the divider range is 1 ~ 127,
or 6 ~ 127 if phase offsets are used. The 0 register value also
represents a divider of 1 or bypass.

Make the end condition of the for loop inclusive of 127 in the
round_rate callback.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>