RPe is runtime-determined by PCODE and caching it caused stale values,
leading to incorrect GuC SLPC parameter settings.
Drop the cached rpe_freq field and query fresh values from hardware
on each use to ensure GuC SLPC parameters reflect current RPe.
v2: Remove cached RPe frequency field (Rodrigo)
v3: Remove extra variable (Vinay)
Modify function name (Vinay)
v4: Maintain a separate function for PVC (Rodrigo)
v5: Avoid RPn update while fetching RPe frequency (Rodrigo)
v6: Split platform-specific RPe comments (Vinay)
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/5166
Signed-off-by: Sk Anirban <sk.anirban@intel.com>
Reviewed-by: Vinay Belgaumkar <vinay.belgaumkar@intel.com>
Link: https://patch.msgid.link/20251112185153.3593145-5-sk.anirban@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Replace the technically inaccurate drm_format_info() with the
accurate drm_get_format_info() in the SKL+ cursor DDB
allocation code.
Since we're only interested in the linear modifier here, the two
functions do actually return the same information. But let's not
use drm_format_info() to avoid setting a bad example.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20251107181126.5743-7-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Use drm_get_format_info() instead of drm_format_info() to populate
fb->format during the BIOS FB readout. The difference being that
drm_get_format_info() knows about compressed formats whereas
drm_format_info() doesn't.
This doesn't actually matter in practice since the BIOS FB should
never be compressed, but no reason we shouldn't use the more accurate
function here anyway.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20251107181126.5743-4-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Pass the format info into plane->max_stride() from the
caller instead of doing yet another drm_format_info()
lookup on the spot.
drm_format_info() is both rather expensive, and technically
incorrect since it doesn't return the correct format info
for compressed formats (though that doesn't actually matter
for the current .max_stride() implementations since they
are just interested in the cpp value).
Most callers already have the format info available. The
only exception is intel_dumb_fb_max_stride() where we shall
use the actually correct drm_get_format_info() variant.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20251107181126.5743-3-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Wrap intel_plane_fb_max_stride() in intel_dumb_fb_max_stride()
for the purposes of dumb fb creation. I want to change
intel_plane_fb_max_stride() to take a 'struct drm_format_info'
instead of the 'u32 pixel_format' so we need an excplicit format
info lookup in the dumb fb path and I don't really want to have
that in i915_gem_dumb_create() directly.
This makes intel_plane_fb_max_stride() internal to the display
code again, and thus we can pass in struct intel_display instead
of struct drm_device.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patch.msgid.link/20251107181126.5743-2-ville.syrjala@linux.intel.com
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Add the missing fields of the nvfw_hs_load_header_v2 struct, so that the
struct matches the actual contents of the firmware images.
nvfw_hs_load_header_v2 is a struct that defines a header for some firmware
images used by Nouveau. The current structure definition is incomplete;
it omits the last two fields because they are unused.
To maintain consistency between Nouveau, OpenRM, and Nova, and to
make it easier to support possible future images, we should fully define
the struct. Also add a __counted_by tag for the flex array.
Signed-off-by: Timur Tabi <ttabi@nvidia.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Link: https://patch.msgid.link/20251010223957.1078525-1-ttabi@nvidia.com
From GP100 onwards it's not possible to initialise comptag RAM without
PMU firmware, which nouveau has no support for.
As such, this code is essentially a no-op and will always revert to the
equivalent non-compressed kind due to comptag allocation failure. It's
also broken for the needs of VM_BIND/Vulkan.
Remove the code entirely to make way for supporting compression on GPUs
that support GSM-RM.
Signed-off-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Mohamed Ahmed <mohamedahmedegypt2001@gmail.com>
Signed-off-by: Mary Guillemard <mary@mary.zone>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Reviewed-by: James Jones <jajones@nvidia.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Link: https://patch.msgid.link/20251110-nouveau-compv6-v6-3-83b05475f57c@mary.zone
Some driver components, like eudebug or ccs-mode, can't be used
when VFs are enabled. Add functions to allow those components
to block the PF from enabling VFs for the requested duration.
Introduce trivial counter to allow lockdown or exclusive access
that can be used in the scenarios where we can't follow the strict
owner semantics as required by the rw_semaphore implementation.
Before enabling VFs, the PF will try to arm the "vfs_enabling"
guard for the exclusive access. This will fail if there are
some lockdown requests already initiated by the other components.
For testing purposes, add debugfs file which will call these new
functions from the file's open/close hooks.
Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Christoph Manszewski <christoph.manszewski@intel.com>
Reviewed-by: Christoph Manszewski <christoph.manszewski@intel.com>
Link: https://patch.msgid.link/20251109162451.4779-1-michal.wajdeczko@intel.com
The sparse array used for error decoding from is unnecessarily big. It
should be better handled by a switch statement that will also allow us
to more easily improve this code.
Add a CASE_ERR() macro to keep the table compact and use it instead of
the 256-entries array, which saves some space:
$ bloat-o-meter xe_pcode.o.old xe_pcode.o
add/remove: 0/1 grow/shrink: 2/0 up/down: 190/-4096 (-3906)
Function old new delta
__pcode_mailbox_rw 363 465 +102
__pcode_mailbox_rw.cold 58 146 +88
err_decode 4096 - -4096
Total: Before=7890, After=3984, chg -49.51%
Reviewed-by: Raag Jadav <raag.jadav@intel.com>
Link: https://patch.msgid.link/20251110-pcode-errmap-v2-1-cb18c8f54238@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
The TIDSS hardware does not have independent maximum or minimum pixel
clock limits for each video port. Instead, these limits are determined
by the SoC's clock architecture. Previously, this constraint was
modeled using the 'max_pclk_khz' and 'min_pclk_khz' fields in
'dispc_features', but this approach is static and does not account for
the dynamic behavior of PLLs.
This patch removes the 'max_pclk_khz' and 'min_pclk_khz' fields from
'dispc_features'. The correct way to check if a requested mode's pixel
clock is supported is by using 'clk_round_rate()' in the 'mode_valid()'
hook. If the best frequency match for the mode clock falls within the
supported tolerance, it is approved. TIDSS supports a 5% pixel clock
tolerance, which is now reflected in the validation logic.
This change allows existing DSS-compatible drivers to be reused across
SoCs that only differ in their pixel clock characteristics. The
validation uses 'clk_round_rate()' for each mode, which may introduce
additional delay (about 3.5 ms for 30 modes), but this is generally
negligible. Users desiring faster validation may bypass these calls
selectively, for example, checking only the highest resolution mode,
as shown here[1].
[1]: https://lore.kernel.org/all/20250704094851.182131-3-j-choudhary@ti.com/
Tested-by: Michael Walle <mwalle@kernel.org>
Reviewed-by: Devarsh Thakkar <devarsht@ti.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Jayesh Choudhary <j-choudhary@ti.com>
Signed-off-by: Swamil Jain <s-jain1@ti.com>
Link: https://patch.msgid.link/20251104151422.307162-2-s-jain1@ti.com
[Tomi: dropped 'inline' from check_pixel_clock]
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
On DE2 and DE3, UI scalers are located right after VI scalers. So in
order to calculate proper UI scaler base address, number of VI scalers
must be known. In practice, it is same as number of VI channels, but it
doesn't need to be.
Let's make a quirk for this number. Code for configuring channels and
associated functions won't have access to vi_num quirk anymore after
rework for independent planes.
Reviewed-by: Chen-Yu Tsai <wens@kernel.org>
Tested-by: Ryan Walklin <ryan@testtoast.com>
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Link: https://patch.msgid.link/20251104180942.61538-27-jernej.skrabec@gmail.com
Signed-off-by: Chen-Yu Tsai <wens@kernel.org>
Determination if FCC unit can be used for VI layer alpha depends on
number of VI channels. This info won't be available anymore in future
to VI layer driver because of DE33 way of allocating planes from same
pool to different mixers.
While order is slightly changed, it doesn't affect anything due to
double buffering of registers. New order keeps related registers
together and quirk separate.
Reviewed-by: Chen-Yu Tsai <wens@kernel.org>
Tested-by: Ryan Walklin <ryan@testtoast.com>
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Link: https://patch.msgid.link/20251104180942.61538-25-jernej.skrabec@gmail.com
Signed-off-by: Chen-Yu Tsai <wens@kernel.org>
Now that channel base calculation is straightforward, let's update VI
scaler base calculation to be simpler. At the same time, also introduce
macro to avoid magic numbers.
Note, reason why current magic value and new macro value isn't the same
is because sun8i_channel_base() already introduces offset to channel
registers. Previous value is just the difference to VI scaler registers.
However, new code calculates scaler base from channel base. This is also
easier to understand when looking into BSP driver. Macro value can be
easily found whereas old diff value was not.
Reviewed-by: Chen-Yu Tsai <wens@kernel.org>
Tested-by: Ryan Walklin <ryan@testtoast.com>
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Link: https://patch.msgid.link/20251104180942.61538-24-jernej.skrabec@gmail.com
Signed-off-by: Chen-Yu Tsai <wens@kernel.org>
With upcoming DE33 support, layer management must be decoupled from
other operations like blender configuration. There are two reasons:
- DE33 will have separate driver for planes and thus it will be harder
to manage different register spaces
- Architecturaly it's better to split access by modules. Blender is now
exclusively managed by mixer.
Reviewed-by: Chen-Yu Tsai <wens@kernel.org>
Tested-by: Ryan Walklin <ryan@testtoast.com>
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Link: https://patch.msgid.link/20251104180942.61538-8-jernej.skrabec@gmail.com
Signed-off-by: Chen-Yu Tsai <wens@kernel.org>
