Stop reaching into other components' registers directly and route those
operations through the component that owns them.
Move the timestamp/coherency helpers into panthor_gpu, add a doorbell
helper, and update call sites accordingly. This keeps register knowledge
local to each block and avoids spreading cross-component register
accesses across the driver.
This is a preparatory cleanup for using per-component iomem bases.
v3:
- Pick up Ack from Boris and R-bs from Liviu and Steve
v2:
- Fix incorrect spelling of timestamp helpers
- Fix unintended trailing backslash
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Acked-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Karunika Choo <karunika.choo@arm.com>
Tested-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
Link: https://patch.msgid.link/20260427155934.416502-4-karunika.choo@arm.com
Split the panthor register definitions into per-component headers for
the GPU, MMU, firmware, power and generic hardware registers.
This makes the register layout easier to follow and prepares the driver
for component-local iomem mappings by grouping definitions with the code
that owns them. The old monolithic panthor_regs.h header can then be
dropped.
No functional change intended.
v3:
- Pick up Ack from Boris and R-bs from Liviu and Steve
v2:
- Merge GPU_ID definitions into panthor_gpu_regs.h
- deleted panthor_hw_regs.h
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Acked-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Karunika Choo <karunika.choo@arm.com>
Tested-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
Link: https://patch.msgid.link/20260427155934.416502-3-karunika.choo@arm.com
Convert the Panthor register access helpers to take an iomem pointer
instead of a panthor_device pointer.
This makes the helpers usable with block-local registers instead of
routing all accesses to go through ptdev->iomem. It is a preparatory
change for splitting the register space by components and for moving
callers away from cross-component register accesses.
No functional change intended.
v3:
- Pick up R-bs from Liviu and Steve
v2:
- Pick up Ack from Boris.
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Acked-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Karunika Choo <karunika.choo@arm.com>
Tested-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
Link: https://patch.msgid.link/20260427155934.416502-2-karunika.choo@arm.com
As I will be departing from my current position, I will
no longer maintain the driver, and remove myself from
ivpu's maintainers entry. Andrzej will support Karol
in maintaining the driver.
Signed-off-by: Maciej Falkowski <maciej.falkowski@linux.intel.com>
Reviewed-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Reviewed-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Signed-off-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Link: https://patch.msgid.link/20260429083958.2148777-1-maciej.falkowski@linux.intel.com
shmob_drm_connector_create() can init the connector in two ways, based on
the 'if (sdev->pdata)':
1. manually in shmob_drm_connector_create(), or
2. delegating to drm_bridge_connector_init()
Whichever branch is taken, drm_connector_attach_encoder() is called
immediately after to attach the connector to the encoder.
Now drm_bridge_connector_init() calls drm_connector_attach_encoder() on the
connector so it is not needed anymore in case 2 and should be removed, but
it is still needed in case 1. Move drm_connector_attach_encoder() from the
common path to inside shmob_drm_connector_create() in order to get back to
a single drm_connector_attach_encoder() in both cases.
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://patch.msgid.link/20260423115550.444930-7-luca.ceresoli@bootlin.com
Signed-off-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
drm_connector_attach_encoder() is now called by
drm_bridge_connector_init().
Reviewed-by: Liu Ying <victor.liu@nxp.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://patch.msgid.link/20260423115550.444930-5-luca.ceresoli@bootlin.com
Signed-off-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
Currently all users of the bridge-connector must call
drm_connector_attach_encoder() immediately after a successful
drm_bridge_connector_init().
This is an unnecessary burden for users. Move the call to the end of
drm_bridge_connector_init() so all callers can be simplified.
Update documentation accordingly, rewording a bit the whole paragraph which
would otherwise become poorly readable due to the growing list of actions.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Link: https://patch.msgid.link/20260423-drm-bridge-connector-attach_encoder-v2-1-2ae6ca69b390@bootlin.com
Signed-off-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
When Nouveau boots with GSP-RM, it bypasses several traditional code
paths to allow GSP-RM to handle those features. In particular,
some VBIOS parsing is skipped, and a side effect is that the VBIOS
is not exposed in the DRM debugfs entries.
Fix this by updating the drm BIOS struct (nvbios) with the VBIOS data
from the nvkm BIOS struct (nvkm_bios). This happens normally in
NVInitVBIOS(), but that function is skipped when booting with GSP-RM.
Signed-off-by: Timur Tabi <ttabi@nvidia.com>
Link: https://patch.msgid.link/20260428202825.1123719-1-ttabi@nvidia.com
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
We have a helper drm_printf_indent() for tab indenting the prints. It
makes the actual strings more readable, and highlights the indented
parts better in source.
Reviewed-by: Chaitanya Kumar Borah <chaitanya.kumar.borah@intel.com>
Link: https://patch.msgid.link/20260408082211.3040194-2-jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
We have a helper drm_printf_indent() for tab indenting the prints. It
makes the actual strings more readable, and highlights the indented
parts better in source.
Reviewed-by: Chaitanya Kumar Borah <chaitanya.kumar.borah@intel.com>
Link: https://patch.msgid.link/20260408082211.3040194-1-jani.nikula@intel.com
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Add configurable frequency limits to allow users to constrain the NPU
operating frequency range for power and thermal management. This support
requires firmware API version 3.34.0 or newer.
New sysfs interface:
The freq/ subdirectory contains the following attributes:
- hw_min_freq: Minimum frequency supported by hardware (read-only)
- hw_max_freq: Maximum frequency supported by hardware (read-only)
- hw_efficient_freq: Hardware's optimal operating frequency (read-only)
- current_freq: Current NPU frequency in MHz (read-only)
- set_min_freq: Configure minimum operating frequency (50XX+ devices)
- set_max_freq: Configure maximum operating frequency (50XX+ devices)
Legacy attributes npu_max_frequency_mhz and npu_current_frequency_mhz
are maintained for backward compatibility.
Implementation details:
- Frequency configuration is communicated to firmware via JSM messages
- User-specified frequency values are clamped to hardware limits
- Power-efficient frequency (pn_ratio) is adjusted dynamically to stay
within the configured range
- Frequency configuration is initialized during device boot
- The JSM API header is updated to version 3.34.0 to support the new
VPU_JSM_MSG_FREQ_CONFIG firmware message
Added description for the sysfs attributes in the Documentation/ABI.
Signed-off-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Reviewed-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Signed-off-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Link: https://patch.msgid.link/20260408150152.2093638-1-andrzej.kacprowski@linux.intel.com
Currently if a user enqueue a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.
This lack of consistentcy cannot be addressed without refactoring the API.
alloc_workqueue() treats all queues as per-CPU by default, while unbound
workqueues must opt-in via WQ_UNBOUND.
This default is suboptimal: most workloads benefit from unbound queues,
allowing the scheduler to place worker threads where they’re needed and
reducing noise when CPUs are isolated.
This change adds a new WQ_PERCPU flag to explicitly request
alloc_workqueue() to be per-cpu when WQ_UNBOUND has not been specified.
With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
must now use WQ_PERCPU.
Once migration is complete, WQ_UNBOUND can be removed and unbound will
become the implicit default.
Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
Link: https://patch.msgid.link/20251031102020.95349-3-marco.crivellari@suse.com
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
