Query to find out how many compute units on a GPU.
Useful for OpenCL usermode drivers.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Signed-off-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Signed-off-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

v2: agd5f: simplify patch
Signed-off-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

And also domain to prefered_domains. That matches better
what those values represent.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Cc: Marek Olšák <maraeo@gmail.com>
Reviewed-by: NMarek Olšák <marek.olsak@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

The underlying reason for the crashes seems to be fixed now.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

We never check the return value anyway and if the
index isn't valid would crash way before calling
the functions.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

When we set the valid bit on invalid GART entries they are
loaded into the TLB when an adjacent entry is loaded. This
poisons the TLB with invalid entries which are sometimes
not correctly removed on TLB flush.

For stable inclusion the patch probably needs to be modified a bit.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Make sure that a hdmi deep color mode can't exceed the max tmds
clock limit of a hdmi sink if such a limit is defined by edid.

If requested deep color bpc would exceed the limit given the mode
to be set, try to degrade gracefully to lower supported deep color
bpc or to standard 8 bpc if needed.
Signed-off-by: NMario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

HDMI deep color setup must know which modes are supported if
it needs to degrade gracefully, as only 12 bpc / dc_36 is
guaranteed, but 10 bpc / dc_30 is optional. The maximum bpc
is not sufficient for this.
Signed-off-by: NMario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

Hawaii has the same version of VCE as other CIK parts.
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org

Replace occurrences of "v & 0xffffffff" with lower_32_bits(v)
when it's next to an upper_32_bits(v). Also remove unnecessary
"upper_32_bits(v) & 0xffffffff" code snippets.
Signed-off-by: NChristian König <christian.koenig@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>

This panel is used by nyan-big and can be supported by the simple-panel
driver.
Signed-off-by: NStéphane Marchesin <marcheu@chromium.org>
[treding@nvidia.com: add device tree binding document]
Signed-off-by: NThierry Reding <treding@nvidia.com>

According to the DP specification the disparity of the first symbol
should always be negative. It is therefore safe to assume that panels
will conform to that and therefore parameterizing this field should
never be necessary.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Instead of always enabling all four lanes, enable only the number probed
from the link.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Power on only those lanes required for the specified link.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Interlaced mode is currently not supported on the SOR, so don't program
any associated registers.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Use the speed probed from the link at runtime rather than relying on a
hardcoded default.
Signed-off-by: NThierry Reding <treding@nvidia.com>

The number of HBLANK and VBLANK symbols can be computed at runtime so
that they can be set appropriately depending on the video mode and DP
link.

These values are used by the packet generation logic to determine how
many audio samples can be transferred during the blanking intervals.
Signed-off-by: NThierry Reding <treding@nvidia.com>

The currently hardcoded link parameters don't work on all eDP panels, so
compute the parameters at runtime depending on the mode and panel type
to allow the driver to cope with a wider variety of panels.

Note that the number of bits per pixel of the panel is still hardcoded,
but this can be addressed in a separate patch.

This is largely based on a patch by Stéphane Marchesin but the algorithm
was largely rewritten to be more readable and concise.
Signed-off-by: NStéphane Marchesin <marcheu@chromium.org>
Signed-off-by: NThierry Reding <treding@nvidia.com>

Lanes are powered up in decreasing order. Power them down in increasing
order for consistency.
Signed-off-by: NStéphane Marchesin <marcheu@chromium.org>
Signed-off-by: NThierry Reding <treding@nvidia.com>

The comment above mentions link A/B but this isn't what the code does,
so let's fix that.
Signed-off-by: NStéphane Marchesin <marcheu@chromium.org>
Signed-off-by: NThierry Reding <treding@nvidia.com>

The code currently rounds up the clock to the next MHZ, which is
rounding up a 69.5MHz clock to 70MHz on my machine. This in turn
prevents the display from syncing. Removing this rounding fixes eDP
for me.
Signed-off-by: NStéphane Marchesin <marcheu@chromium.org>
Signed-off-by: NThierry Reding <treding@nvidia.com>

Other output drivers set up debugfs slightly differently. Bring the SOR
driver in line with those for consistency.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Removing only the root directory will fail when there are still files in
it. Instead of manually removing all files, remove the whole directory
recursively.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Doing so allows the hotplug events generated by the connector to be
properly handled by the DRM poll helpers.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Calling the drm_helper_hpd_irq_event() helper can sleep, so instead of
invoking it directly from the interrupt handler, schedule a work queue
and run it from there.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Enable hardware cursor support on Tegra124. Earlier generations support
the hardware cursor to some degree as well, but not in a way that can be
generically exposed.
Signed-off-by: NThierry Reding <treding@nvidia.com>

The DRM core can now cope with drivers that don't have an associated
struct drm_bus, so the host1x implementation is no longer useful.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Describe how devices are registered using the drm_*_init() functions.
Adding this to docbook requires a largish set of changes to the comments
in drm_{pci,usb,platform}.c since they are doxygen-style rather than
proper kernel-doc and therefore mess with the docbook generation.

While at it, mark usage of drm_put_dev() as discouraged in favour of
calling drm_dev_unregister() and drm_dev_unref() directly.
Acked-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: NThierry Reding <treding@nvidia.com>

Add a helper function that allows drivers to statically set the unique
name of the device. This will allow platform and USB drivers to get rid
of their DRM bus implementations and directly use drm_dev_alloc() and
drm_dev_register().
Reviewed-by: NDaniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: NDavid Herrmann <dh.herrmann@gmail.com>
Signed-off-by: NThierry Reding <treding@nvidia.com>

The internal host1x_{,un}register_client() functions can potentially be
confused with public the host1x_client_{,un}register() functions.

Rename them to host1x_{add,del}_client() to remove some of the possible
confusion.
Signed-off-by: NThierry Reding <treding@nvidia.com>

The function is never used outside of the source file and therefore can
be locally scoped.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Tegra124 is mostly backwards-compatible with Tegra114. However, Tegra124
supports a few more features (e.g. interlacing, ...). Introduce a new
compatible string and TMDS tables to cope with these differences.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Accessing the CRC debugfs file will hang the system if the SOR is not
enabled, so make sure that it is stays enabled until the CRC has been
read.
Signed-off-by: NThierry Reding <treding@nvidia.com>

In some cases the pixel clock used to not be correct, which is why it
had to be recomputed. It turns out that the reason why it wasn't correct
is that it was used wrongly. If used correctly there's not need for the
recomputation.
Signed-off-by: NThierry Reding <treding@nvidia.com>

The shift clock divider is highly dependent on the type of output, so
push computation of it down into the output drivers. The old code used
to work merely by accident.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Program the shift clock divider in tegra_crtc_setup_clk() since that's
where the divider is computed, so passing it around can be avoided.
Signed-off-by: NThierry Reding <treding@nvidia.com>

Assert the DSI controller's reset when the driver is unloaded to reduce
power consumption and to put the controller into a known state for
subsequent driver reloads.
Signed-off-by: NThierry Reding <treding@nvidia.com>

When disabling the DSI controller, the code wasn't really doing what it
was supposed to.
Signed-off-by: NThierry Reding <treding@nvidia.com>