The drm_sched_job_unschedulable trace point can access
entity->dependency after it was cleared by the callback
installed in drm_sched_entity_add_dependency_cb, causing:
BUG: kernel NULL pointer dereference, address: 0000000000000020
[...]
Workqueue: comp_1.1.0 drm_sched_run_job_work [gpu_sched]
RIP: 0010:trace_event_raw_event_drm_sched_job_unschedulable+0x70/0xd0 [gpu_sched]
To fix this we either need to keep a reference to the fence before
setting up the callbacks, or move the trace_drm_sched_job_unschedulable
calls into drm_sched_entity_add_dependency_cb where they can be
done earlier.
Fixes: 76d97c870f ("drm/sched: Trace dependencies for GPU jobs")
Signed-off-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Signed-off-by: Philipp Stanner <phasta@kernel.org>
Link: https://lore.kernel.org/r/20250901124032.1955-1-pierre-eric.pelloux-prayer@amd.com
Depending on which display that is connected to the controller, an
"1" means either a black or a white pixel.
The supported formats (R1/R2/XRGB8888) expects the pixels
to map against (4bit):
00 => Black
01 => Dark Gray
10 => Light Gray
11 => White
If this is not what the display map against, make it possible to invert
the pixels.
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Marcus Folkesson <marcus.folkesson@gmail.com>
Link: https://lore.kernel.org/r/20250721-st7571-format-v2-4-159f4134098c@gmail.com
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Depending on which display that is connected to the controller, an "1"
means either a black or a white pixel.
The supported formats (R1/R2/XRGB8888) expects the pixels
to map against (4bit):
00 => Black
01 => Dark Gray
10 => Light Gray
11 => White
If this is not what the display map against, the controller has support
to invert these values.
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Marcus Folkesson <marcus.folkesson@gmail.com>
Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
Link: https://lore.kernel.org/r/20250721-st7571-format-v2-2-159f4134098c@gmail.com
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Several RISC-V boards feature Imagination GPUs that are compatible with
the PowerVR driver. An example is the IMG BXM-4-64 GPU on the Lichee Pi
4A board. This commit adjusts the driver's Kconfig dependencies to allow
the PowerVR driver to be compiled on the RISC-V architecture.
By enabling compilation on RISC-V, we expand support for these GPUs,
providing graphics acceleration capabilities and enhancing hardware
compatibility on RISC-V platforms.
The RISC-V support is restricted to 64-bit systems (RISCV && 64BIT) as
the driver currently has an implicit dependency on a 64-bit platform.
Add a dependency on MMU to fix a build warning on RISC-V configurations
without an MMU.
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
Reviewed-by: Matt Coster <matt.coster@imgtec.com>
Signed-off-by: Michal Wilczynski <m.wilczynski@samsung.com>
Link: https://lore.kernel.org/r/20250822-apr_14_for_sending-v13-4-af656f7cc6c3@samsung.com
Signed-off-by: Matt Coster <matt.coster@imgtec.com>
Rework the PowerVR Rogue GPU binding to use an explicit, per variant
style for defining power domain properties and add support for the
T-HEAD TH1520 SoC's GPU.
To improve clarity and precision, the binding is refactored so that
power domain items are listed explicitly for each variant [1]. The
previous method relied on an implicit, positional mapping between the
`power-domains` and `power-domain-names` properties. This change
replaces the generic rules with self contained if/then blocks for each
GPU variant, making the relationship between power domains and their
names explicit and unambiguous.
The generic if block for img,img-rogue, which previously required
power-domains and power-domain-names for all variants, is removed.
Instead, each specific GPU variant now defines its own power domain
requirements within a self-contained if/then block, making the schema
more explicit.
This new structure is then used to add support for the
`thead,th1520-gpu`. While its BXM-4-64 IP has two conceptual power
domains, the TH1520 SoC integrates them behind a single power gate. The
new binding models this with a specific rule that enforces a single
`power-domains` entry and disallows the `power-domain-names` property.
Link: https://lore.kernel.org/all/4d79c8dd-c5fb-442c-ac65-37e7176b0cdd@linaro.org/ [1]
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Reviewed-by: Matt Coster <matt.coster@imgtec.com>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Michal Wilczynski <m.wilczynski@samsung.com>
Link: https://lore.kernel.org/r/20250822-apr_14_for_sending-v13-2-af656f7cc6c3@samsung.com
Signed-off-by: Matt Coster <matt.coster@imgtec.com>
Update the Imagination PVR DRM driver to leverage the pwrseq framework
for managing the complex power sequence of the GPU on the T-HEAD TH1520
SoC.
To cleanly separate platform-specific logic from the generic driver,
this patch introduces an `init` callback to the `pwr_power_sequence_ops`
struct. This allows for different power management strategies to be
selected at probe time based on the device's compatible string.
A `pvr_device_data` struct, associated with each compatible in the
of_device_id table, points to the appropriate ops table (manual or
pwrseq).
At probe time, the driver now calls the `->init()` op. For pwrseq-based
platforms, this callback calls `devm_pwrseq_get("gpu-power")`, deferring
probe if the sequencer is not yet available. For other platforms, it
falls back to the existing manual clock and reset handling. The runtime
PM callbacks continue to call the appropriate functions via the ops
table.
Signed-off-by: Michal Wilczynski <m.wilczynski@samsung.com>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Matt Coster <matt.coster@imgtec.com>
Link: https://lore.kernel.org/r/20250822-apr_14_for_sending-v13-1-af656f7cc6c3@samsung.com
Signed-off-by: Matt Coster <matt.coster@imgtec.com>
Commit a7352c8494 ("dt-bindings: npu: rockchip,rknn: Add bindings") adds
the device-tree binding rockchip,rk3588-rknn-core.yaml, whereas the commit
ed98261b41 ("accel/rocket: Add a new driver for Rockchip's NPU") adds the
section DRM ACCEL DRIVER FOR ROCKCHIP NPU in MAINTAINERS with a file entry
referring to rockchip,rknn-core.yaml. Note that the file entry is missing
the part rk3588, compared to the added file above, which it intends to
refer to.
Adjust the file entry to the intended file name.
Fixes: ed98261b41 ("accel/rocket: Add a new driver for Rockchip's NPU")
Signed-off-by: Lukas Bulwahn <lukas.bulwahn@redhat.com>
Signed-off-by: Tomeu Vizoso <tomeu@tomeuvizoso.net>
Link: https://lore.kernel.org/r/20250826063248.32153-1-lukas.bulwahn@redhat.com
The problem is that pm_runtime_get_sync() can return 1 on success so
checking for zero doesn't work. Use the pm_runtime_resume_and_get()
function instead. The pm_runtime_resume_and_get() function does
additional cleanup as well so that's a bonus as well.
Fixes: 0810d5ad88 ("accel/rocket: Add job submission IOCTL")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Tomeu Vizoso <tomeu@tomeuvizoso.net>
Link: https://lore.kernel.org/r/aKcRW6fsRP_o5C_y@stanley.mountain
With the current dependency on only DRM, a config of
CONFIG_DRM_ACCEL_ROCKET=y
is possible, but of course wrong, because without DRM_ACCEL the build-
system will never even enter drivers/accel/* .
So depend on DRM_ACCEL instead of just DRM.
Fixes: ed98261b41 ("accel/rocket: Add a new driver for Rockchip's NPU")
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Tomeu Vizoso <tomeu@tomeuvizoso.net>
Link: https://lore.kernel.org/r/20250814113519.1551855-3-heiko@sntech.de
Remove fixed PPI lane count setup. The R-Car DSI host is capable
of operating in 1..4 DSI lane mode. Remove the hard-coded 4-lane
configuration from PPI register settings and instead configure
the PPI lane count according to lane count information already
obtained by this driver instance.
Configure TXSETR register to match PPI lane count. The R-Car V4H
Reference Manual R19UH0186EJ0121 Rev.1.21 section 67.2.2.3 Tx Set
Register (TXSETR), field LANECNT description indicates that the
TXSETR register LANECNT bitfield lane count must be configured
such, that it matches lane count configuration in PPISETR register
DLEN bitfield. Make sure the LANECNT and DLEN bitfields are
configured to match.
Fixes: 155358310f ("drm: rcar-du: Add R-Car DSI driver")
Cc: stable@vger.kernel.org
Signed-off-by: Marek Vasut <marek.vasut+renesas@mailbox.org>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen+renesas@ideasonboard.com>
Link: https://lore.kernel.org/r/20250813210840.97621-1-marek.vasut+renesas@mailbox.org
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The OVR_REG_FLD_MOD function takes the start and end bits as parameter
and will generate a mask out of them.
This makes it difficult to share the masks between callers, since we now
need two arguments and to keep them consistent.
Let's change OVR_REG_FLD_MOD to take the mask as an argument instead,
and let the caller create the mask. Eventually, this mask will be moved
to a define.
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://lore.kernel.org/r/20250827-drm-tidss-field-api-v3-13-7689b664cc63@kernel.org
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The VP_REG_FLD_MOD function takes the start and end bits as parameter
and will generate a mask out of them.
This makes it difficult to share the masks between callers, since we now
need two arguments and to keep them consistent.
Let's change VP_REG_FLD_MOD to take the mask as an argument instead, and
let the caller create the mask. Eventually, this mask will be moved to a
define.
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://lore.kernel.org/r/20250827-drm-tidss-field-api-v3-12-7689b664cc63@kernel.org
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The VP_REG_GET function takes the start and end bits as parameter and
will generate a mask out of them.
This makes it difficult to share the masks between callers, since we now
need two arguments and to keep them consistent.
Let's change VP_REG_GET to take the mask as an argument instead, and let
the caller create the mask. Eventually, this mask will be moved to a
define.
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://lore.kernel.org/r/20250827-drm-tidss-field-api-v3-11-7689b664cc63@kernel.org
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The VID_REG_FLD_MOD function takes the start and end bits as parameter
and will generate a mask out of them.
This makes it difficult to share the masks between callers, since we now
need two arguments and to keep them consistent.
Let's change VID_REG_FLD_MOD to take the mask as an argument instead,
and let the caller create the mask. Eventually, this mask will be moved
to a define.
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://lore.kernel.org/r/20250827-drm-tidss-field-api-v3-10-7689b664cc63@kernel.org
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The VID_REG_GET function takes the start and end bits as parameter and
will generate a mask out of them.
This makes it difficult to share the masks between callers, since we now
need two arguments and to keep them consistent.
Let's change VID_REG_GET to take the mask as an argument instead, and
let the caller create the mask. Eventually, this mask will be moved to a
define.
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://lore.kernel.org/r/20250827-drm-tidss-field-api-v3-9-7689b664cc63@kernel.org
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The REG_FLD_MOD function takes the start and end bits as parameter and
will generate a mask out of them.
This makes it difficult to share the masks between callers, since we now
need two arguments and to keep them consistent.
Let's change REG_FLD_MOD to take the mask as an argument instead, and
let the caller create the mask. Eventually, this mask will be moved to a
define.
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://lore.kernel.org/r/20250827-drm-tidss-field-api-v3-8-7689b664cc63@kernel.org
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The REG_GET function takes the start and end bits as parameter and will
generate a mask out of them.
This makes it difficult to share the masks between callers, since we now
need two arguments and to keep them consistent.
Let's change REG_GET to take the mask as an argument instead, and let
the caller create the mask. Eventually, this mask will be moved to a
define.
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://lore.kernel.org/r/20250827-drm-tidss-field-api-v3-7-7689b664cc63@kernel.org
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The dispc driver uses upper-cased, inlined, functions to provide
macro-like accessors to the dispc registers.
This is confusing, since upper-case is usually used by macros, and that
pattern will create gcc errors later on in this series.
Let's switch to macros to make it more consistent, and prevent those
errors down the line.
Signed-off-by: Maxime Ripard <mripard@kernel.org>
Link: https://lore.kernel.org/r/20250827-drm-tidss-field-api-v3-2-7689b664cc63@kernel.org
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
The simple display pipe is obsolete and the atomic helpers allow for
more control over the rendering process. As such, this patch replaces
the old simple display pipe system with the newer atomic helpers.
As the code is mainly the same, merely replaced with the new atomic
system, there should be no change in functionality.
Signed-off-by: Ruben Wauters <rubenru09@aol.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://lore.kernel.org/r/20250818193553.2162-1-rubenru09@aol.com
Userspace jobs have drm_file.client_id as a unique identifier
as job's owners. For kernel jobs, we can allocate arbitrary
values - the risk of overlap with userspace ids is small (given
that it's a u64 value).
In the unlikely case the overlap happens, it'll only impact
trace events.
Since this ID is traced in the gpu_scheduler trace events, this
allows to determine the source of each job sent to the hardware.
To make grepping easier, the IDs are defined as they will appear
in the trace output.
Signed-off-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
Link: https://lore.kernel.org/r/20250604122827.2191-1-pierre-eric.pelloux-prayer@amd.com
Change the 'ret' variable from u32 to int to store -EINVAL. Storing the
negative error codes in unsigned type, doesn't cause an issue at runtime
but it's ugly as pants.
Additionally, assigning -EINVAL to u32 ret (i.e., u32 ret = -EINVAL) may
trigger a GCC warning when the -Wsign-conversion flag is enabled.
Fixes: aac243092b ("accel/amdxdna: Add command execution")
Signed-off-by: Qianfeng Rong <rongqianfeng@vivo.com>
Reviewed-by: Lizhi Hou <lizhi.hou@amd.com>
Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Link: https://lore.kernel.org/r/20250828033917.113364-1-rongqianfeng@vivo.com
Since commit e7fa80e293 ("drm_gem: add mutex to drm_gem_object.gpuva")
it is possible for test_prepare_array() to exceed a stack frame size of
2048 bytes depending on the exact configuration of the kernel.
drivers/gpu/drm/tests/drm_exec_test.c: In function ‘test_prepare_array’:
drivers/gpu/drm/tests/drm_exec_test.c:171:1: error: the frame size of 2128 bytes is larger than 2048 bytes [-Werror=frame-larger-than=]
171 | }
| ^
cc1: all warnings being treated as errors
make[6]: *** [scripts/Makefile.build:287: drivers/gpu/drm/tests/drm_exec_test.o] Error 1
make[6]: *** Waiting for unfinished jobs....
In order to fix this, allocate the GEM objects in test_prepare_array()
with kzalloc(), rather than placing them on the stack.
Cc: Alice Ryhl <aliceryhl@google.com>
Cc: Christian König <christian.koenig@amd.com>
Fixes: e7fa80e293 ("drm_gem: add mutex to drm_gem_object.gpuva")
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alice Ryhl <aliceryhl@google.com>
Reviewed-by: Nirmoy Das <nirmoyd@nvidia.com>
Link: https://lore.kernel.org/r/20250829075633.2306-1-dakr@kernel.org
[ Use kunit_kzalloc() instead of kzalloc(). - Danilo ]
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
The per-fd reset counter tracks GPU resets caused by jobs submitted
through a specific file descriptor. However, there's a race condition
where the file descriptor can be closed while jobs are still running,
leading to potential access to freed memory when updating the reset
counter.
Ensure that the per-fd reset counter is only updated when the file
descriptor is still valid, preventing use-after-free scenarios during
GPU reset handling.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Link: https://lore.kernel.org/r/20250826-v3d-queue-lock-v3-6-979efc43e490@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Each V3D queue works independently and all the dependencies between the
jobs are handled through the DRM scheduler. Therefore, there is no need
to use one single lock for all queues. Using it, creates unnecessary
contention between different queues that can operate independently.
Replace the global spinlock with per-queue locks to improve parallelism
and reduce contention between different V3D queues (BIN, RENDER, TFU,
CSD). This allows independent queues to operate concurrently while
maintaining proper synchronization within each queue.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Melissa Wen <mwen@igalia.com>
Link: https://lore.kernel.org/r/20250826-v3d-queue-lock-v3-3-979efc43e490@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
