openwrt/staging/blogic.git
6 years agodrm/tegra: dc: Implement legacy blending
Thierry Reding [Wed, 20 Dec 2017 08:39:14 +0000 (09:39 +0100)]
drm/tegra: dc: Implement legacy blending

This implements alpha blending on legacy display controllers (Tegra20,
Tegra30 and Tegra114). While it's theoretically possible to support the
zpos property to enable userspace to specify the Z-order of each plane
individually, this is not currently supported and the same fixed Z-
order as previously defined is used.

Reverts commit 71835caa00e8 ("drm/tegra: fb: Force alpha formats") since
the opaque formats are now supported.

Reported-by: Dmitry Osipenko <digetx@gmail.com>
Fixes: 7772fdaef939 ("drm/tegra: Support ARGB and ABGR formats")
Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Correct timeout in tegra_syncpt_wait
Dmitry Osipenko [Wed, 20 Dec 2017 15:46:14 +0000 (18:46 +0300)]
drm/tegra: Correct timeout in tegra_syncpt_wait

host1x_syncpt_wait() takes timeout value in jiffies, but DRM passes it in
milliseconds.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: gem: Correct iommu_map_sg() error checking
Dmitry Osipenko [Wed, 20 Dec 2017 15:46:13 +0000 (18:46 +0300)]
drm/tegra: gem: Correct iommu_map_sg() error checking

iommu_map_sg() doesn't return a error value, but a size of the requested
IOMMU mapping or zero in case of error.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Link DC1 to DC0 on Tegra20
Dmitry Osipenko [Wed, 20 Dec 2017 15:46:10 +0000 (18:46 +0300)]
drm/tegra: dc: Link DC1 to DC0 on Tegra20

Hardware reset isn't actually broken on Tegra20, but there is a
dependency on the first display controller to be taken out of reset for
the second to be enabled successfully. Model this dependency using a PM
device link.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
[treding@nvidia.com: minor cleanups, extend commit message]
Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Fix non-debugfs builds
Arnd Bergmann [Mon, 18 Dec 2017 13:44:46 +0000 (14:44 +0100)]
drm/tegra: Fix non-debugfs builds

The new debugfs registration fails to build when CONFIG_DEBUGFS is
disabled, because the drm_crtc structure is lacking a member in that
configuration:

drivers/gpu/drm/tegra/dc.c: In function 'tegra_dc_late_register':
drivers/gpu/drm/tegra/dc.c:1204:28: error: 'struct drm_crtc' has no member named 'debugfs_entry'

Without CONFIG_DEBUGFS, the rest of the function already degrades
into nothing, so we just avoid the one assignment.

Fixes: b95800eeef09 ("drm/tegra: dc: Register debugfs in ->late_register()")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dpaux: Keep reset defaults for hybrid pad parameters
Thierry Reding [Thu, 14 Dec 2017 12:25:37 +0000 (13:25 +0100)]
drm/tegra: dpaux: Keep reset defaults for hybrid pad parameters

Keep the reset values for the common mode voltage, output driver
impedance control and output driver current control parameters. This
fixes errors seen during SCDC communication with HDMI sinks.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Mark Tegra186 display hub PM functions __maybe_unused
Arnd Bergmann [Fri, 15 Dec 2017 12:51:52 +0000 (13:51 +0100)]
drm/tegra: Mark Tegra186 display hub PM functions __maybe_unused

The newly introduced driver has optional suspend/resume functions,
causing a warning when CONFIG_PM is disabled:

drivers/gpu/drm/tegra/hub.c:749:12: error: 'tegra_display_hub_resume' defined but not used [-Werror=unused-function]
drivers/gpu/drm/tegra/hub.c:733:12: error: 'tegra_display_hub_suspend' defined but not used [-Werror=unused-function]

This marks them __maybe_unused to shut up the warnings.

Fixes: c4755fb9064f ("drm/tegra: Add Tegra186 display hub support")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Use IOMMU groups
Thierry Reding [Thu, 12 Oct 2017 15:43:33 +0000 (17:43 +0200)]
drm/tegra: Use IOMMU groups

In order to support IOMMUs more generically and transparently handle the
ARM SMMU on Tegra186, move to using groups instead of devices for domain
attachment. An IOMMU group is a set of devices that share the same IOMMU
domain and is therefore a good match to represent what Tegra DRM needs.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agogpu: host1x: Use IOMMU groups
Thierry Reding [Tue, 14 Nov 2017 15:11:28 +0000 (16:11 +0100)]
gpu: host1x: Use IOMMU groups

Use IOMMU groups to attach the host1x device to its IOMMU domain. This
is not strictly necessary because the domain isn't shared with any other
device, but it makes the code consistent with how IOMMU is handled in
other drivers and provides an easy way to detect when no IOMMU has been
attached via device tree.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Implement zpos property
Thierry Reding [Thu, 14 Dec 2017 12:46:20 +0000 (13:46 +0100)]
drm/tegra: Implement zpos property

Implement the standard zpos property for planes on Tegra124 and later.
Earlier generations have a different blending unit that needs different
programming.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Remove redundant spinlock
Thierry Reding [Thu, 14 Dec 2017 12:50:19 +0000 (13:50 +0100)]
drm/tegra: dc: Remove redundant spinlock

The spinlock is only used to serialize accesses to the DC_CMD_INT_MASK
register. However, this register is accesses either with interrupts
masked (in tegra_crtc_atomic_enable()) or protected by the vbl_lock and
vblank_time_lock spinlocks of the DRM device. Therefore, these accesses
don't need any extra serialization and the lock can be removed.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Use direct offset to plane registers
Thierry Reding [Thu, 14 Dec 2017 12:37:53 +0000 (13:37 +0100)]
drm/tegra: dc: Use direct offset to plane registers

Traditionally, windows were accessed indirectly, through a register
selection window that required a global register to be programmed with
the index of the window to access. Since the global register could be
written from modesetting functions as well as the interrupt handler
concurrently, accesses had to be serialized using a lock. Using direct
accesses to the window registers the lock can be avoided.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Support more formats
Thierry Reding [Tue, 14 Nov 2017 15:07:40 +0000 (16:07 +0100)]
drm/tegra: dc: Support more formats

Also, split up formats into per-SoC lists because not all generations
support all of them. Note that the list is now exhaustive for all RGB
formats, but not for YUV and indexed formats.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: fb: Force alpha formats
Thierry Reding [Tue, 14 Nov 2017 15:09:30 +0000 (16:09 +0100)]
drm/tegra: fb: Force alpha formats

Tegra20 and Tegra30 don't actually support the 24-bit RGB formats that
don't have an alpha component. In order to allow the fbdev emulation to
run on those chips, force the 32-bit RGBA formats.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dpaux: Add Tegra186 support
Thierry Reding [Thu, 12 Oct 2017 15:34:11 +0000 (17:34 +0200)]
drm/tegra: dpaux: Add Tegra186 support

DPAUX is the same as on previous generations. Supporting it is as simple
as adding the compatible string so that the driver will bind to any of
the devices.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dpaux: Implement runtime PM
Thierry Reding [Thu, 12 Oct 2017 15:32:52 +0000 (17:32 +0200)]
drm/tegra: dpaux: Implement runtime PM

Move clock and reset management into runtime PM callbacks and hook them
up. This cleans up the code structure so that power management code does
not clutter up the rest.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: sor: Support HDMI 2.0 modes
Thierry Reding [Thu, 12 Oct 2017 17:14:21 +0000 (19:14 +0200)]
drm/tegra: sor: Support HDMI 2.0 modes

In addition to using the SCDC helpers to enable support for scrambling
for HDMI 2.0 modes, take into account the high pixel clocks when
programming some of the registers.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: sor: Add Tegra186 support
Thierry Reding [Thu, 12 Oct 2017 17:12:57 +0000 (19:12 +0200)]
drm/tegra: sor: Add Tegra186 support

The SOR found on Tegra186 is very similar to the one found on Tegra210
and earlier. However, due to some changes in the display architecture,
some programming sequences have changed and some register have moved
around.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: sor: Parameterize register offsets
Thierry Reding [Thu, 12 Oct 2017 17:04:17 +0000 (19:04 +0200)]
drm/tegra: sor: Parameterize register offsets

Future Tegra generations have an increased number of display controllers
that can drive individual SORs. In order to support that, the offset and
layout of some registers has changed in backwards-incompatible ways. Use
parameterized register offsets to support this.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Support ARGB and ABGR formats
Thierry Reding [Thu, 12 Oct 2017 15:30:55 +0000 (17:30 +0200)]
drm/tegra: Support ARGB and ABGR formats

These formats can easily be supported on all generations of Tegra.

Note that the XRGB and XBGR formats that we supported were in fact using
the ARGB and ABGR Tegra formats. This happened to work in cases where no
alpha was being considered. This change is also a fix for those formats.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Add Tegra186 support
Thierry Reding [Wed, 30 Aug 2017 15:42:54 +0000 (17:42 +0200)]
drm/tegra: dc: Add Tegra186 support

The display architecture has changed in several signifcant ways with the
new Tegra186 SoC. Display controllers are a completely different design,
but have been given a frontend that simulates the register interface for
earlier chips.

Unfortunately the frontend isn't completely backwards compatible, so the
driver needs parameterization to take the changes into account.

One big change is that the total number of display controllers has been
increased to three. At the same time the number of planes available has
remained constant. However, planes can now be freely assigned between
the display controllers, giving applications more flexibility in making
the best use of the available resources.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Add Tegra186 display hub support
Thierry Reding [Mon, 13 Nov 2017 10:08:13 +0000 (11:08 +0100)]
drm/tegra: Add Tegra186 display hub support

The display architecture has changed in several significant ways with
the new Tegra186 SoC. Shared between all display controllers is a set
of common resources referred to as the display hub. The hub generates
accesses to memory and feeds them into various composition pipelines,
each of which being a window that can be assigned to arbitrary heads.

Atomic state is subclassed in order to track the global bandwidth
requirements and select and adjust the hub clocks appropriately. The
plane code is shared to a large degree with earlier SoC generations,
except where the programming differs.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Move common plane code to separate file
Thierry Reding [Fri, 10 Nov 2017 14:27:25 +0000 (15:27 +0100)]
drm/tegra: Move common plane code to separate file

Subsequent patches will add support for Tegra186 which has a different
architecture and needs different plane code but which can share a lot of
code with earlier Tegra support.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Move state definition to header
Thierry Reding [Fri, 10 Nov 2017 15:00:05 +0000 (16:00 +0100)]
drm/tegra: dc: Move state definition to header

Move the display controller state definition to the header file so that
it can be referenced by other files.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Remove duplicate plane funcs
Thierry Reding [Wed, 30 Aug 2017 16:04:12 +0000 (18:04 +0200)]
drm/tegra: dc: Remove duplicate plane funcs

Both tegra_overlay_plane_funcs is identical to tegra_plane_funcs. Get
rid of the duplicate and use one set of function pointers for all
planes.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Remove tegra_overlay_plane_destroy()
Thierry Reding [Wed, 30 Aug 2017 16:02:51 +0000 (18:02 +0200)]
drm/tegra: dc: Remove tegra_overlay_plane_destroy()

This function is a simple wrapper around tegra_plane_destroy(), so it
can be dropped.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Remove duplicate plane funcs
Thierry Reding [Wed, 30 Aug 2017 16:04:12 +0000 (18:04 +0200)]
drm/tegra: dc: Remove duplicate plane funcs

Both tegra_primary_plane_funcs and tegra_cursor_plane_funcs are
identical. Get rid of the duplicate and use one set of function pointers
for all planes.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Remove tegra_primary_plane_destroy()
Thierry Reding [Wed, 30 Aug 2017 16:02:51 +0000 (18:02 +0200)]
drm/tegra: dc: Remove tegra_primary_plane_destroy()

This function is a simple wrapper around tegra_plane_destroy(), so it
can be dropped.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Remove custom page-flip handler
Thierry Reding [Thu, 12 Oct 2017 15:40:46 +0000 (17:40 +0200)]
drm/tegra: Remove custom page-flip handler

Tegra display hardware has GO bits and meets all the requirements to use
drm_crtc_arm_vblank_event(). Use it instead and get rid of the hand-
rolled implementation.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Use atomic commit helpers
Thierry Reding [Thu, 12 Oct 2017 15:40:46 +0000 (17:40 +0200)]
drm/tegra: Use atomic commit helpers

There's no reason not to use them, and they already get all the
semantics right, so rip out all of the custom code and replace it by the
helpers.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Support background color
Thierry Reding [Mon, 13 Nov 2017 10:20:48 +0000 (11:20 +0100)]
drm/tegra: dc: Support background color

Starting with Tegra124, the interface to set the background color (the
value generated for pixels that are not sourced from any window) is via
a different register. Earlier generations called this the border color.
Reverse the feature flag and assume that IP revisions that don't have
support for background color will support border color instead.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: vic: Properly align arguments
Thierry Reding [Wed, 8 Nov 2017 12:22:17 +0000 (13:22 +0100)]
drm/tegra: vic: Properly align arguments

Properly align function arguments on subsequent lines with the first
argument on the first line.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Do not wrap lines unnecessarily
Thierry Reding [Thu, 9 Nov 2017 12:17:11 +0000 (13:17 +0100)]
drm/tegra: Do not wrap lines unnecessarily

The tegra_drm_alloc() function signature fits on a single line, no need
to wrap it.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: sor: Register debugfs in ->late_register()
Thierry Reding [Wed, 8 Nov 2017 12:20:01 +0000 (13:20 +0100)]
drm/tegra: sor: Register debugfs in ->late_register()

The ->late_register() and ->early_unregister() callbacks are called at
the right time to make sure userspace only accesses interfaces when it
should. Move debugfs registration and unregistration to these callback
functions to avoid potential races with userspace.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: sor: Root debugfs files at the connector
Thierry Reding [Thu, 12 Oct 2017 17:07:54 +0000 (19:07 +0200)]
drm/tegra: sor: Root debugfs files at the connector

Rather create new files within the top-level DRM device's debugfs node,
add the SOR specific files to the connector's debugfs node. This avoids
the need to come up with subdirectory names and is also more intuitive.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: hdmi: Register debugfs in ->late_register()
Thierry Reding [Wed, 8 Nov 2017 12:18:31 +0000 (13:18 +0100)]
drm/tegra: hdmi: Register debugfs in ->late_register()

The ->late_register() and ->early_unregister() callbacks are called at
the right time to make sure userspace only accesses interfaces when it
should. Move debugfs registration and unregistration to these callback
functions to avoid potential races with userspace.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dsi: Register debugfs in ->late_register()
Thierry Reding [Wed, 8 Nov 2017 12:12:44 +0000 (13:12 +0100)]
drm/tegra: dsi: Register debugfs in ->late_register()

The ->late_register() and ->early_unregister() callbacks are called at
the right time to make sure userspace only accesses interfaces when it
should. Move debugfs registration and unregistration to these callback
functions to avoid potential races with userspace.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Register debugfs in ->late_register()
Thierry Reding [Wed, 8 Nov 2017 12:40:54 +0000 (13:40 +0100)]
drm/tegra: dc: Register debugfs in ->late_register()

The ->late_register() and ->early_unregister() callbacks are called at
the right time to make sure userspace only accesses interfaces when it
should. Move debugfs registration and unregistration to these callback
functions to avoid potential races with userspace.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Reshuffle some code
Thierry Reding [Wed, 8 Nov 2017 12:32:05 +0000 (13:32 +0100)]
drm/tegra: dc: Reshuffle some code

Reshuffle some code so that functions are defined closer to where they
are used.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: sor: Move register definitions into a table
Thierry Reding [Fri, 10 Nov 2017 11:21:51 +0000 (12:21 +0100)]
drm/tegra: sor: Move register definitions into a table

After commit 932f6529139e ("drm/tegra: sor: Trace register accesses"),
the debugfs register dump implementation causes excessive stack usage
and can result in build warnings. To fix this, move the register
definitions into a table and iterate over the table while dumping the
registers to debugfs.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: hdmi: Move register definitions into a table
Thierry Reding [Fri, 10 Nov 2017 11:20:46 +0000 (12:20 +0100)]
drm/tegra: hdmi: Move register definitions into a table

After commit 07a8aab89927 ("drm/tegra: hdmi: Trace register accesses"),
the debugfs register dump implementation causes excessive stack usage
and can result in build warnings. To fix this, move the register
definitions into a table and iterate over the table while dumping the
registers to debugfs.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dsi: Move register definitions into a table
Thierry Reding [Fri, 10 Nov 2017 11:18:22 +0000 (12:18 +0100)]
drm/tegra: dsi: Move register definitions into a table

After commit 75af8fa7fd47 ("drm/tegra: dsi: Trace register accesses"),
the debugfs register dump implementation causes excessive stack usage
and can result in build warnings. To fix this, move the register
definitions into a table and iterate over the table while dumping the
registers to debugfs.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: dc: Move register definitions into a table
Thierry Reding [Fri, 10 Nov 2017 11:13:22 +0000 (12:13 +0100)]
drm/tegra: dc: Move register definitions into a table

After commit 67e04d1ab19b ("drm/tegra: dc: Trace register accesses"),
the debugfs register dump implementation causes excessive stack usage
and can result in build warnings. To fix this, move the register
definitions into a table and iterate over the table while dumping the
registers to debugfs.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodt-bindings: display: tegra: Update SOR for Tegra186
Thierry Reding [Wed, 30 Aug 2017 10:04:07 +0000 (12:04 +0200)]
dt-bindings: display: tegra: Update SOR for Tegra186

Update the SOR bindings for Tegra186, in which a new property is
required to identify the instance of the SOR interface and the clock
tree has slightly changed as well.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agogpu: host1x: Cleanup on initialization failure
Thierry Reding [Wed, 8 Nov 2017 12:08:21 +0000 (13:08 +0100)]
gpu: host1x: Cleanup on initialization failure

When an error happens during the initialization of one of the sub-
devices, make sure to properly cleanup all sub-devices that have been
initialized up to that point.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agogpu: host1x: Rewrite conditional for better readability
Thierry Reding [Mon, 31 Jul 2017 09:18:44 +0000 (11:18 +0200)]
gpu: host1x: Rewrite conditional for better readability

The current check is slightly difficult to read, rewrite it to improve
that a little.

Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/tegra: Sanitize format modifiers
Thierry Reding [Thu, 12 Oct 2017 14:39:20 +0000 (16:39 +0200)]
drm/tegra: Sanitize format modifiers

The existing format modifier definitions were merged prematurely, and
recent work has unveiled that the definitions are suboptimal in several
ways:

  - The format specifiers, except for one, are not Tegra specific, but
    the names don't reflect that.
  - The number space is split into two, reserving 32 bits for some
    "parameter" which most of the modifiers are not going to have.
  - Symbolic names for the modifiers are not using the standard
    DRM_FORMAT_MOD_* prefix, which makes them awkward to use.
  - The vendor prefix NV is somewhat ambiguous.

Fortunately, nobody's started using these modifiers, so we can still fix
the above issues. Do so by using the standard prefix. Also, remove TEGRA
from the name of those modifiers that exist on NVIDIA GPUs as well. In
case of the block linear modifiers, make the "parameter" smaller (4
bits, though only 6 values are valid) and don't let that leak into any
of the other modifiers.

Finally, also use the more canonical NVIDIA instead of the ambiguous NV
prefix.

Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/fourcc: Fix fourcc_mod_code() definition
Thierry Reding [Wed, 1 Nov 2017 14:20:04 +0000 (15:20 +0100)]
drm/fourcc: Fix fourcc_mod_code() definition

Avoid a compiler warnings when the val parameter is an expression.

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Thierry Reding <treding@nvidia.com>
6 years agodrm/qxl: reapply cursor after resetting primary
Ray Strode [Mon, 27 Nov 2017 21:50:10 +0000 (16:50 -0500)]
drm/qxl: reapply cursor after resetting primary

QXL associates mouse state with its primary plane.

Destroying a primary plane and putting a new one in place has the side
effect of destroying the cursor as well.

This commit changes the driver to reapply the cursor any time a new
primary is created. It achieves this by keeping a reference to the
cursor bo on the qxl_crtc struct.

This fix is very similar to

commit 4532b241a4b7 ("drm/qxl: reapply cursor after SetCrtc calls")

which got implicitly reverted as part of implementing the atomic
modeset feature.

Cc: Gerd Hoffmann <kraxel@redhat.com>
Cc: Dave Airlie <airlied@redhat.com>
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1512097
Fixes: 1277eed5fecb ("drm: qxl: Atomic phase 1: convert cursor to universal plane")
Cc: stable@vger.kernel.org
Signed-off-by: Ray Strode <rstrode@redhat.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
6 years agodrm/qxl: unref cursor bo when finished with it
Ray Strode [Mon, 27 Nov 2017 21:50:09 +0000 (16:50 -0500)]
drm/qxl: unref cursor bo when finished with it

qxl_cursor_atomic_update allocs a bo for the cursor that
it never frees up at the end of the function.

This commit fixes that.

Signed-off-by: Ray Strode <rstrode@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
6 years agoMerge tag 'drm-intel-next-2017-12-01' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Thu, 7 Dec 2017 22:39:14 +0000 (08:39 +1000)]
Merge tag 'drm-intel-next-2017-12-01' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

[airlied: fix conflict in intel_dsi.c]

drm-intel-next-2017-12-01:

- Init clock gate fix (Ville)
- Execlists event handling corrections (Chris, Michel)
- Improvements on GPU Cache invalidation and context switch (Chris)
- More perf OA changes (Lionel)
- More selftests improvements and fixes (Chris, Matthew)
- Clean-up on modules parameters (Chris)
- Clean-up around old ringbuffer submission and hw semaphore on old platforms (Chris)
- More Cannonlake stabilization effort (David, James)
- Display planes clean-up and improvements (Ville)
- New PMU interface for perf queries... (Tvrtko)
- ... and other subsequent PMU changes and fixes (Tvrtko, Chris)
- Remove success dmesg noise from rotation (Chris)
- New DMC for Kabylake (Anusha)
- Fixes around atomic commits (Daniel)
- GuC updates and fixes (Sagar, Michal, Chris)
- Couple gmbus/i2c fixes (Ville)
- Use exponential backoff for all our wait_for() (Chris)
- Fixes for i915/fbdev (Chris)
- Backlight fixes (Arnd)
- Updates on shrinker (Chris)
- Make Hotplug enable more robuts (Chris)
- Disable huge pages (TPH) on lack of a needed workaround (Joonas)
- New GuC images for SKL, KBL, BXT (Sagar)
- Add HW Workaround for Geminilake performance (Valtteri)
- Fixes for PPS timings (Imre)
- More IPS fixes (Maarten)
- Many fixes for Display Port on gen2-gen4 (Ville)
- Retry GPU reset making the recover from hang more robust (Chris)

* tag 'drm-intel-next-2017-12-01' of git://anongit.freedesktop.org/drm/drm-intel: (101 commits)
  drm/i915: Update DRIVER_DATE to 20171201
  drm/i915/cnl: Mask previous DDI - PLL mapping
  drm/i915: Remove unsafe i915.enable_rc6
  drm/i915: Sleep and retry a GPU reset if at first we don't succeed
  drm/i915: Interlaced DP output doesn't work on VLV/CHV
  drm/i915: Pass crtc state to intel_pipe_{enable,disable}()
  drm/i915: Wait for pipe to start on i830 as well
  drm/i915: Fix vblank timestamp/frame counter jumps on gen2
  drm/i915: Fix deadlock in i830_disable_pipe()
  drm/i915: Fix has_audio readout for DDI A
  drm/i915: Don't add the "force audio" property to DP connectors that don't support audio
  drm/i915: Disable DP audio for g4x
  drm/i915/selftests: Wake the device before executing requests on the GPU
  drm/i915: Set fake_vma.size as well as fake_vma.node.size for capture
  drm/i915: Tidy up signed/unsigned comparison
  drm/i915: Enable IPS with only sprite plane visible too, v4.
  drm/i915: Make ips_enabled a property depending on whether IPS is enabled, v3.
  drm/i915: Avoid PPS HW/SW state mismatch due to rounding
  drm/i915: Skip switch-to-kernel-context on suspend when wedged
  drm/i915/glk: Apply WaProgramL3SqcReg1DefaultForPerf for GLK too
  ...

6 years agoMerge tag 'drm-misc-next-2017-12-07' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Thu, 7 Dec 2017 22:15:54 +0000 (08:15 +1000)]
Merge tag 'drm-misc-next-2017-12-07' of git://anongit.freedesktop.org/drm/drm-misc into drm-next

UAPI Changes:

- Add "panel orientation" property to DRM to indicate orientation of the
panel vs the device's casing (Hans de Goede)

Core Changes:

- misc doc and bug fixes

Driver Changes:

- sun4i: Many improvements to the DE driver like multi-plane support and
YUV formats (Jernej Skrabec)

* tag 'drm-misc-next-2017-12-07' of git://anongit.freedesktop.org/drm/drm-misc: (50 commits)
  drm/sun4i: Fix uninitialized variables in vi layer
  drm/fb-helper: Fix potential NULL pointer dereference
  gpu: drm: stm: Adopt SPDX identifiers
  gpu: drm: sti: Adopt SPDX identifiers
  drm/fsl-dcu: Use drm_mode_config_helper_suspend/resume()
  drm/sun4i: Wire in DE2 YUV support
  drm/sun4i: Expand DE2 scaler lib with YUV support
  drm/sun4i: Add DE2 definitions for YUV formats
  drm/sun4i: Add DE2 CSC library
  drm/sun4i: Add CCSC property to DE2 configuration
  drm/sun4i: Add support for HW scaling to DE2
  drm/sun4i: Add scaler configuration to DE2 mixers
  drm/sun4i: Add support for DE2 VI planes
  drm/sun4i: Reorganize UI layer code in DE2
  drm/sun4i: Add support for all HW supported DE2 RGB formats
  drm/sun4i: Add multi plane support to DE2 driver
  drm/sun4i: Move interlace related code in DE2
  drm/sun4i: Move channel size related code in DE2
  drm/sun4i: Move line width setting in DE2
  drm/sun4i: Use values calculated by atomic check
  ...

6 years agodrm/sun4i: Fix uninitialized variables in vi layer
Jernej Skrabec [Wed, 6 Dec 2017 15:26:03 +0000 (16:26 +0100)]
drm/sun4i: Fix uninitialized variables in vi layer

min_scale and max_scale in sun8i_vi_layer_atomic_check() can be used
without initialization.

Fix that.

Fixes: b862a648de3b ("drm/sun4i: Add support for HW scaling to DE2")
Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171206152603.25937-1-jernej.skrabec@siol.net
6 years agodrm/fb-helper: Fix potential NULL pointer dereference
Gustavo A. R. Silva [Tue, 5 Dec 2017 17:46:28 +0000 (11:46 -0600)]
drm/fb-helper: Fix potential NULL pointer dereference

fb_helper is being dereferenced before it is null checked,
hence there is a potential null pointer dereference.

Fix this by moving the pointer dereference after fb_helper
has been null checked.

This issue was detected with the help of Coccinelle.

Fixes: c777990fb45b ("drm/fb-helper: Handle function NULL argument")
Signed-off-by: Gustavo A. R. Silva <garsilva@embeddedor.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20171205174628.GA31189@embeddedor.com
6 years agoMerge branch 'drm-next-4.16' of git://people.freedesktop.org/~agd5f/linux into drm...
Dave Airlie [Wed, 6 Dec 2017 20:28:22 +0000 (06:28 +1000)]
Merge branch 'drm-next-4.16' of git://people.freedesktop.org/~agd5f/linux into drm-next

First feature request for 4.16.  Highlights:
- RV and Vega header cleanups
- TTM operation context support
- 48 bit GPUVM fixes for Vega/RV
- More smatch fixes
- ECC support for vega10
- Resizeable BAR support
- Multi-display sync support in DC
- SR-IOV fixes
- Various scheduler improvements
- GPU reset fixes and vram lost tracking
- Clean up DC/powerplay interfaces
- DCN display fixes
- Various DC fixes

* 'drm-next-4.16' of git://people.freedesktop.org/~agd5f/linux: (291 commits)
  drm/radeon: Use drm_fb_helper_lastclose() and _poll_changed()
  drm/amdgpu: Use drm_fb_helper_lastclose() and _poll_changed()
  drm/amd/display: Use drm_fb_helper_poll_changed()
  drm/ttm: swap consecutive allocated pooled pages v4
  drm/amdgpu: fix amdgpu_sync_resv v2
  drm/ttm: swap consecutive allocated cached pages v3
  drm/amd/amdgpu: set gtt size according to system memory size only
  drm/amdgpu: Get rid of dep_sync as a seperate object.
  drm/amdgpu: allow specifying vm_block_size for multi level PDs v2
  drm/amdgpu: move validation of the VM size into the VM code
  drm/amdgpu: allow non pot VM size values
  drm/amdgpu: choose number of VM levels based on VM size
  drm/amdgpu: unify VM size handling of Vega10 with older generation
  drm/amdgpu: fix amdgpu_vm_num_entries
  drm/amdgpu: fix VM PD addr shift
  drm/amdgpu: correct vce4.0 fw config for SRIOV (V2)
  drm/amd/display: Don't call dm_log_to_buffer directly in dc_conn_log
  drm/amd/display: Add dm_logger_append_va API
  drm/ttm: Use a static string instead of an array of char *
  drm/amd/display: remove usage of legacy_cursor_update
  ...

6 years agodrm/radeon: Use drm_fb_helper_lastclose() and _poll_changed()
Noralf Trønnes [Tue, 5 Dec 2017 18:25:02 +0000 (19:25 +0100)]
drm/radeon: Use drm_fb_helper_lastclose() and _poll_changed()

This driver can use drm_fb_helper_lastclose() in its .lastclose function.
It can also use drm_fb_helper_output_poll_changed() as its
.output_poll_changed callback.

Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: Use drm_fb_helper_lastclose() and _poll_changed()
Noralf Trønnes [Tue, 5 Dec 2017 18:24:55 +0000 (19:24 +0100)]
drm/amdgpu: Use drm_fb_helper_lastclose() and _poll_changed()

This driver can use drm_fb_helper_lastclose() in its .lastclose function.
It can also use drm_fb_helper_output_poll_changed() as its
.output_poll_changed callback.
Remove the unused driver implementations.

Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Use drm_fb_helper_poll_changed()
Noralf Trønnes [Tue, 5 Dec 2017 18:24:54 +0000 (19:24 +0100)]
drm/amd/display: Use drm_fb_helper_poll_changed()

This driver can use drm_fb_helper_output_poll_changed() as its
.output_poll_changed callback.

Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: "Christian König" <christian.koenig@amd.com>
Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/ttm: swap consecutive allocated pooled pages v4
Christian König [Mon, 4 Dec 2017 10:26:14 +0000 (11:26 +0100)]
drm/ttm: swap consecutive allocated pooled pages v4

When we detect consecutive allocation of pages swap them to avoid
accidentally freeing them as huge page.

v2: use swap
v3: check if it's really the first allocated page
v4: don't touch the loop variable

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Roger He <Hongbo.He@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: fix amdgpu_sync_resv v2
Christian König [Fri, 24 Nov 2017 10:41:52 +0000 (11:41 +0100)]
drm/amdgpu: fix amdgpu_sync_resv v2

Fixes a bug introduced by AMDGPU_GEM_CREATE_EXPLICIT_SYNC. We still need
to wait for pipelined moves in the shared fences list.

v2: fix typo

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Andres Rodriguez <andresx7@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/ttm: swap consecutive allocated cached pages v3
Christian König [Mon, 4 Dec 2017 10:17:54 +0000 (11:17 +0100)]
drm/ttm: swap consecutive allocated cached pages v3

When we detect consecutive allocation of pages swap them to avoid
accidentally freeing them as huge page.

v2: use swap
v3: check if it's really the first allocated page

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Roger He <Hongbo.He@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/amdgpu: set gtt size according to system memory size only
Roger He [Wed, 29 Nov 2017 09:12:03 +0000 (17:12 +0800)]
drm/amd/amdgpu: set gtt size according to system memory size only

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Roger He <Hongbo.He@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: Get rid of dep_sync as a seperate object.
Andrey Grodzovsky [Mon, 13 Nov 2017 19:47:52 +0000 (14:47 -0500)]
drm/amdgpu: Get rid of dep_sync as a seperate object.

Instead mark fence as explicit in it's amdgpu_sync_entry.

v2:
Fix use after free bug and add new parameter description.
Signed-off-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: allow specifying vm_block_size for multi level PDs v2
Christian König [Mon, 27 Nov 2017 15:22:05 +0000 (16:22 +0100)]
drm/amdgpu: allow specifying vm_block_size for multi level PDs v2

This patch allows specifying the vm_block_size even when multi level
page directories are active.

v2: fix signed/unsigned compare warning

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: move validation of the VM size into the VM code
Christian König [Thu, 23 Nov 2017 11:57:18 +0000 (12:57 +0100)]
drm/amdgpu: move validation of the VM size into the VM code

This moves validation of the VM size parameter into amdgpu_vm_adjust_size().

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: allow non pot VM size values
Christian König [Thu, 23 Nov 2017 10:23:18 +0000 (11:23 +0100)]
drm/amdgpu: allow non pot VM size values

The VM size actually doesn't need to be a power of two.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: choose number of VM levels based on VM size
Christian König [Thu, 23 Nov 2017 10:16:05 +0000 (11:16 +0100)]
drm/amdgpu: choose number of VM levels based on VM size

This allows us limiting the VM size for testing even of Vega10.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: unify VM size handling of Vega10 with older generation
Christian König [Wed, 22 Nov 2017 16:00:35 +0000 (17:00 +0100)]
drm/amdgpu: unify VM size handling of Vega10 with older generation

One function to rule them all.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: fix amdgpu_vm_num_entries
Christian König [Mon, 20 Nov 2017 13:29:01 +0000 (14:29 +0100)]
drm/amdgpu: fix amdgpu_vm_num_entries

The block size only affects the leave nodes, everything else is fixed.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: fix VM PD addr shift
Christian König [Mon, 27 Nov 2017 13:01:51 +0000 (14:01 +0100)]
drm/amdgpu: fix VM PD addr shift

The block size only affects the leave nodes, everything else is fixed.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu: correct vce4.0 fw config for SRIOV (V2)
Frank Min [Mon, 6 Nov 2017 07:34:55 +0000 (15:34 +0800)]
drm/amdgpu: correct vce4.0 fw config for SRIOV (V2)

1. program vce 4.0 fw with 48 bit address
2. correct vce 4.0 fw stack and date offset

Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Frank Min <Frank.Min@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Don't call dm_log_to_buffer directly in dc_conn_log
Michel Dänzer [Tue, 28 Nov 2017 11:06:13 +0000 (12:06 +0100)]
drm/amd/display: Don't call dm_log_to_buffer directly in dc_conn_log

dm_log_to_buffer logs unconditionally, so calling it directly resulted
in the main message being logged even when the event type isn't enabled
in the event mask.

To fix this, use the new dm_logger_append_va API.

Fixes spurious messages like

 [drm] {1920x1200, 2080x1235@154000Khz}

in dmesg when a mode is set.

v2:
* Use new dm_logger_append_va API, fixes incorrect va_list usage in v1
* Just use and decrease entry.buf_offset to get rid of the trailing
  newline

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: Add dm_logger_append_va API
Michel Dänzer [Tue, 28 Nov 2017 11:06:13 +0000 (12:06 +0100)]
drm/amd/display: Add dm_logger_append_va API

Same as dm_logger_append, except it takes a va_list instead of a
variable number of arguments. dm_logger_append is now a minimal wrapper
around dm_logger_append_va.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/ttm: Use a static string instead of an array of char *
Joe Perches [Tue, 28 Nov 2017 19:34:16 +0000 (11:34 -0800)]
drm/ttm: Use a static string instead of an array of char *

Make the object a bit smaller by using a simple string instead
of a format string and array of char *.

$ size drivers/gpu/drm/ttm/ttm_page_alloc_dma.o*
   text    data     bss     dec     hex filename
   8820     216    4136   13172    3374 drivers/gpu/drm/ttm/ttm_page_alloc_dma.o.defconfig.new
   8910     216    4136   13262    33ce drivers/gpu/drm/ttm/ttm_page_alloc_dma.o.defconfig.old
  25383    5044    4384   34811    87fb drivers/gpu/drm/ttm/ttm_page_alloc_dma.o.allyesconfig.new
  25797    5428    4384   35609    8b19 drivers/gpu/drm/ttm/ttm_page_alloc_dma.o.allyesconfig.old

Miscellanea:

o The h array had more entries than were emitted, all are now removed

Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/display: remove usage of legacy_cursor_update
Shirish S [Fri, 1 Dec 2017 06:27:04 +0000 (01:27 -0500)]
drm/amd/display: remove usage of legacy_cursor_update

Currently the atomic check code uses legacy_cursor_update
to differnetiate if the cursor plane is being requested by
the user, which is not required as we shall be updating
plane only if modeset is requested/required.

Have tested cursor plane and underlay get updated seamlessly,
without any lag or frame drops.

Signed-off-by: Shirish S <shirish.s@amd.com>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Reviewed-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup raven1 vcn header files.
Feifei Xu [Mon, 27 Nov 2017 09:57:30 +0000 (17:57 +0800)]
drm/amd/include:cleanup raven1 vcn header files.

Cleanup asic_reg/raven1/VCN folder. Remove unused vcn_1_0_default.h.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup raven1 thm header files.
Feifei Xu [Mon, 27 Nov 2017 09:32:44 +0000 (17:32 +0800)]
drm/amd/include:cleanup raven1 thm header files.

Cleanup asic_reg/raven1/THM folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup raven1 nbio header files.
Feifei Xu [Mon, 27 Nov 2017 09:29:29 +0000 (17:29 +0800)]
drm/amd/include:cleanup raven1 nbio header files.

Cleanup asic_reg/raven1/NBIO folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup raven1 mp header files.
Feifei Xu [Mon, 27 Nov 2017 09:20:55 +0000 (17:20 +0800)]
drm/amd/include:cleanup raven1 mp header files.

Cleanup asic_reg/raven1/MP folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup raven1 mmhub header files.
Feifei Xu [Mon, 27 Nov 2017 09:16:06 +0000 (17:16 +0800)]
drm/amd/include:cleanup raven1 mmhub header files.

Cleanup asic_reg/raven1/MMHUB folder.Remove unused mmhub_9_1_default.h

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup raven1 gc header files.
Feifei Xu [Mon, 27 Nov 2017 09:00:12 +0000 (17:00 +0800)]
drm/amd/include:cleanup raven1 gc header files.

Cleanup asic_reg/raven1/GC folder. Remove unused files:
    gc_9_1_default.h  gc_9_1_sh_mask.h

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup raven1 dcn header files.
Feifei Xu [Mon, 27 Nov 2017 10:59:10 +0000 (18:59 +0800)]
drm/amd/include:cleanup raven1 dcn header files.

Cleanup asic_reg/raven1/DCN folder.Remove unused
dcn_1_0_default.h.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup raven1 sdma header files.
Feifei Xu [Mon, 27 Nov 2017 10:40:15 +0000 (18:40 +0800)]
drm/amd/include:cleanup raven1 sdma header files.

Cleanup asic_reg/raven1/SDMA0 folder.Remove unused sdma0_4_1_sh_mask.h.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Junwei Zhang <Jerry.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 header files.
Feifei Xu [Fri, 24 Nov 2017 04:31:36 +0000 (12:31 +0800)]
drm/amd/include:cleanup vega10 header files.

Remove asic_reg/vega10 folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 osssys header files.
Feifei Xu [Fri, 24 Nov 2017 02:46:24 +0000 (10:46 +0800)]
drm/amd/include:cleanup vega10 osssys header files.

Cleanup asic_reg/vega10/OSSSYS folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 smuio header files.
Feifei Xu [Thu, 23 Nov 2017 07:09:51 +0000 (15:09 +0800)]
drm/amd/include:cleanup vega10 smuio header files.

Cleanup asic_reg/vega10/SMUIO folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 nbif header files.
Feifei Xu [Thu, 23 Nov 2017 07:02:23 +0000 (15:02 +0800)]
drm/amd/include:cleanup vega10 nbif header files.

Cleanup asic_reg/vega10/NBIF folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 nbio header files.
Feifei Xu [Thu, 23 Nov 2017 06:54:48 +0000 (14:54 +0800)]
drm/amd/include:cleanup vega10 nbio header files.

Cleanup asic_reg/vega10/NBIO folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 mmhub header files.
Feifei Xu [Thu, 23 Nov 2017 06:30:43 +0000 (14:30 +0800)]
drm/amd/include:cleanup vega10 mmhub header files.

Cleanup asic_reg/vega10/MMHUB folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 gc header files.
Feifei Xu [Fri, 24 Nov 2017 02:29:00 +0000 (10:29 +0800)]
drm/amd/include:cleanup vega10 gc header files.

Cleanup asic_reg/vega10/GC folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 vce header files.
Feifei Xu [Thu, 23 Nov 2017 06:08:34 +0000 (14:08 +0800)]
drm/amd/include:cleanup vega10 vce header files.

Cleanup asic_reg/vega10/VCE folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 uvd header files.
Feifei Xu [Thu, 23 Nov 2017 03:09:07 +0000 (11:09 +0800)]
drm/amd/include:cleanup vega10 uvd header files.

Cleanup asic_reg/vega10/UVD folder,remove unused uvd_7_0_default.h.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 dce header files.
Feifei Xu [Thu, 23 Nov 2017 10:18:14 +0000 (18:18 +0800)]
drm/amd/include:cleanup vega10 dce header files.

Cleanup asic_reg/vega10/DC folder.Remove dce_12_0_default.h.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include: cleanup vega10 umc header files.
Feifei Xu [Wed, 22 Nov 2017 07:23:20 +0000 (15:23 +0800)]
drm/amd/include: cleanup vega10 umc header files.

Remove asic/vega10/UMC folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 thm header files.
Feifei Xu [Wed, 22 Nov 2017 06:57:05 +0000 (14:57 +0800)]
drm/amd/include:cleanup vega10 thm header files.

Cleanup asic_reg/vega10/THM folder.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 athub header files.
Feifei Xu [Thu, 16 Nov 2017 09:50:10 +0000 (17:50 +0800)]
drm/amd/include:cleanup vega10 athub header files.

Cleanup asic_reg/vega10/ATHUB folder,remove unused files.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 mp header files.
Feifei Xu [Wed, 15 Nov 2017 10:39:21 +0000 (18:39 +0800)]
drm/amd/include:cleanup vega10 mp header files.

Cleanup asic_reg/vega10/MP folder, remove mp_9_0_default.h

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 hdp header files.
Feifei Xu [Wed, 15 Nov 2017 10:09:33 +0000 (18:09 +0800)]
drm/amd/include:cleanup vega10 hdp header files.

Cleanup asic_reg/vega10/HDP folder, remove hdp_4_0_default.h

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amd/include:cleanup vega10 sdma0/1 header files.
Feifei Xu [Wed, 15 Nov 2017 08:01:30 +0000 (16:01 +0800)]
drm/amd/include:cleanup vega10 sdma0/1 header files.

To remove include/asic_reg/vega10 folder,create IP folders sdma0/1.
This patch cleanup asic_reg/vega10/SDMA folders.

Signed-off-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
6 years agodrm/amdgpu:partially revert 1cfd8e237f0318e330190ac21d63c58ae6a1f66c
Monk Liu [Tue, 21 Nov 2017 05:29:14 +0000 (13:29 +0800)]
drm/amdgpu:partially revert 1cfd8e237f0318e330190ac21d63c58ae6a1f66c

found RING0 test fail after S3 resume regression, which is
introduced by 1cfd8e237f0318e330190ac21d63c58ae6a1f66c

Because after suspend VRAM will be cleared, so driver must
unpin the GART table(resident in VRAM) during suspend so it
can be evicted to system ram and must correspondingly pin it
during resume so the GART table could be restored to VRAM.

Signed-off-by: Monk Liu <Monk.Liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>