The vsp1_pipe object context variables for div_size and
current_partition allowed state to be maintained through processing the
partitions during processing.

Now that the partition tables are calculated during stream on, there is
no requirement to store these variables in the pipe object.

Utilise local variables for the processing as required.
Signed-off-by: NKieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NHans Verkuil <hansverk@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Previously the active window and partition sizes for each partition were
calculated for each partition every frame. This data is constant and
only needs to be calculated once at the start of the stream.

Extend the vsp1_pipe object to dynamically store the number of partitions
required and pre-calculate the partition sizes into this table.
Signed-off-by: NKieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NHans Verkuil <hansverk@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Separate the code change from the function move so that code changes can
be clearly identified. This commit has no functional change.

The partition algorithm functions will be changed, and
vsp1_video_pipeline_setup_partitions() will call vsp1_video_partition().
To prepare for that, move the function without any code change.
Signed-off-by: NKieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NHans Verkuil <hansverk@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Presently any received buffers are only released back to vb2 if
vsp1_video_stop_streaming() is called. If vsp1_video_start_streaming()
encounters an error, we will be warned by the vb2 handlers that buffers
have not been returned.

Move the buffer cleanup code to its own function to prevent duplication
and call from both vsp1_video_stop_streaming() and the error path in
vsp1_video_start_streaming().
Signed-off-by: NKieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NHans Verkuil <hansverk@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

On Gen2 hardware the VSP1 is a bus master and accesses the display list
and video buffers through DMA directly. On Gen3 hardware, however,
memory accesses go through a separate IP core called FCP.

The VSP1 driver unconditionally maps DMA buffers through the VSP device.
While this doesn't cause any practical issue so far, DMA mappings will
be incorrect as soon as we will enable IOMMU support for the FCP on Gen3
platforms, resulting in IOMMU faults.

Fix this by mapping all buffers through the FCP device if present, and
through the VSP1 device as usual otherwise.
Suggested-by: NMagnus Damm <magnus.damm@gmail.com>
[Cache the bus master device]
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: NKieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Signed-off-by: NKieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Acked-by: NMauro Cavalho Chehab <mchehab@s-opensource.com>

The HGT is a Histogram Generator Two-Dimensions. It computes a weighted
frequency histograms for hue and saturation areas over a configurable
region of the image with optional subsampling.
Signed-off-by: NNiklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

The HGO is a Histogram Generator One-Dimension. It computes per-channel
histograms over a configurable region of the image with optional
subsampling.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Histogram generators are single-pad entities that branch as leaf nodes
at any point in the pipeline. Make sure that pipeline traversal and
routing configuration support them correctly.

Support for the actual HGO and HGT operation will come later.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Some WPF instances, on Gen3 devices, can perform 90° rotation when
writing frames to memory. Implement support for this using the
V4L2_CID_ROTATE control.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Fix all multi-line comments to comply with the kernel coding style.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

With multiple inputs through the BRU it is feasible for the streams to
race each other at stream-on.

Multiple VIDIOC_STREAMON calls racing each other could have process
N-1 skipping over the pipeline setup section and then start the pipeline
early, if videobuf2 has already enqueued buffers to the driver for
process N but not called the .start_streaming() operation yet

In the case of the video pipelines, this
can present two serious issues.

 1) A null-dereference if the pipe->dl is committed at the same time as
    the vsp1_video_setup_pipeline() is processing

 2) A hardware hang, where a display list is committed without having
    called vsp1_video_setup_pipeline() first

Repair this issue, by ensuring that only the stream which configures the
pipeline is able to start it.
Signed-off-by: NKieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

v4l2_fh_init is already done. So call the v4l2_fh_exit in error
condition before returing from the function.
Signed-off-by: NShailendra Verma <shailendra.v@samsung.com>
Reviewed-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Use the vb2 ioctl handler directly.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Tested-by: NJacopo Mondi <jacopo@jmondi.org>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

The media_entity_pipeline_start() and media_entity_pipeline_stop()
functions are renamed as media_pipeline_start() and media_pipeline_stop(),
respectively. The reason is two-fold: the pipeline struct is, rightly,
already called media_pipeline (rather than media_entity_pipeline) and what
this really is about is a pipeline. A pipeline consists of entities ---
and, well, other objects embedded in these entities.

As the pipeline object will be in the future moved from entities to pads
in order to support multiple pipelines through a single entity, do the
renaming now.

Similarly, functions operating on struct media_entity_graph as well as the
struct itself are renamed by dropping the "entity_" part from the prefix
of the function family and the data structure. The graph traversal which
is what the functions are about is not specifically about entities only
and will operate on pads for the same reason as the media pipeline.

The patch has been generated using the following command:

git grep -l media_entity |xargs perl -i -pe '
	s/media_entity_pipeline/media_pipeline/g;
	s/media_entity_graph/media_graph/g'

And a few manual edits related to line start alignment and line wrapping.
Signed-off-by: NSakari Ailus <sakari.ailus@linux.intel.com>
Acked-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Support both the HSV24 and HSV32 formats. From a hardware point of view
pretend the formats are RGB, the RPF and WPF will just pass the data
through without performing any processing.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Acked-by: NRicardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Acked-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Several multi-line comments added at the vsp1 patch series
violate the Kernel CodingStyle. Fix them.
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

The VYUY format isn't supported on Gen3 hardware, disable it.

Gen2 hardware supports VYUY in practice even though the documentation
doesn't advertise it, so keep it for Gen2 devices.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Adapt vsp1_video_pipeline_run() such that it can iterate each partition
required for constructing this frame's display list chain in the event
that multiple display lists are required to process in hardware.

The first display list is held as the head list object, whilst any
following parition display lists are linked to the head by means of
vsp1_dl_list_add_chain().

Linking the chained display list headers to process using the auto start
mechanism of the hardware is performed during the vsp1_dl_list_commit().
Signed-off-by: NKieran Bingham <kieran+renesas@bingham.xyz>
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: NNiklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

The partition algorithm needs to determine the capabilities of each
entity in the pipeline to identify the correct maximum partition width.

Extend the vsp1 entity operations to provide a max_width operation and
use this call to calculate the number of partitions that will be
processed by the algorithm.

Gen 2 hardware does not require multiple partitioning, and as such
will always return a single partition.
Signed-off-by: NKieran Bingham <kieran+renesas@bingham.xyz>
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Acked-by: NNiklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

The new VSP1_ENTITY_PARAMS_PARTITION configuration parameters type
covers all registers that need to be configured for every partition.
This prepares for support of image partitioning, and replaces the
.set_memory() operation as the memory registers take different values
for every partition.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Replace the current boolean parameter (full / !full) with an explicit
enum.

- VSP1_ENTITY_PARAMS_INIT for parameters to be configured at pipeline
  initialization time only (V4L2 stream on or DRM atomic update)
- VSP1_ENTITY_PARAMS_RUNTIME for all parameters that can be freely
  modified at runtime (through V4L2 controls)

This will allow future extensions when implementing image partitioning
support.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Acked-by: NNiklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

The frame-end function releases and completes the buffers on the input
and output entities of the pipe before marking the pipe->state as
'STOPPED'. This introduces a race whereby with the pipe->state still
'RUNNING', a QBUF handler can commence processing a frame before the
frame_end function has completed.

In the event that this happens, a frame queued by QBUF hangs due to the
incorrect pipe->state setting which prevents vsp1_pipeline_run from
issuing a CMD_STRCMD.

By locking the entire function we prevent this from occurring, but we
also change the locking state of the buffer release code. This has been
analysed visually as acceptable, but it must be considered that this now
causes the video->irqlock to be taken under the pipe->irqlock context.
Signed-off-by: NKieran Bingham <kieran+renesas@bingham.xyz>
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Pipelines can only be run if all their video nodes are streaming. Commit
b4dfb9b3 ("[media] v4l: vsp1: Stop the pipeline upon the first
STREAMOFF") fixed the pipeline stop sequence, but introduced a race
condition that makes it possible to run a pipeline after stopping the
stream on a video node by queuing a buffer on the other side of the
pipeline.

Fix this by clearing the buffers ready flag when stopping the stream,
which will prevent the QBUF handler from running the pipeline.

Fixes: b4dfb9b3 ("[media] v4l: vsp1: Stop the pipeline upon the first STREAMOFF")
Reported-by: NKieran Bingham <kieran@bingham.xyz>
Tested-by: NKieran Bingham <kieran@bingham.xyz>
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Make this a proper typed array. Drop the old allocate context code since
that is no longer used.

Note that the memops functions now get a struct device pointer instead of
the struct device ** that was there initially (actually a void pointer to
a struct containing only a struct device pointer).

This code is now a lot cleaner.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Sakari Ailus <sakari.ailus@iki.fi>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Stop using alloc_ctx and just fill in the device pointer.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Acked-by: NFabien Dessenne <fabien.dessenne@st.com>
Acked-by: NBenoit Parrot <bparrot@ti.com>
Reviewed-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Controls are applied to the hardware in the configure operation of the
VSP entities, which is only called when starting the video stream. To
enable runtime modification of controls we need to call the configure
operations for every frame. Doing so is currently not safe, as most
parameters shouldn't be modified during streaming. Furthermore the
configure operation can sleep, preventing it from being called from the
frame completion interrupt handler for the next frame.

Fix this by adding an argument to the configure operation to tell
entities whether to perform a full configuration (as done now) or a
partial runtime configuration. In the latter case the operation will
only configure the subset of parameters related to runtime-configurable
controls, and won't be allowed to sleep when doing so.

Because partial reconfiguration can depend on parameters computed when
performing a full configuration, the core guarantees that the configure
operation will always be called with full and partial modes in that
order at stream start. Entities thus don't have to duplicate
configuration steps in the full and partial code paths.

This change affects the VSP driver core only, all entities return
immediately from the configure operation when called for a partial
runtime configuration. Entities will be modified one by one in further
commits.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

The device is stopped when STREAMOFF is called on the last video node in
the pipeline. This results in possible memory corruption and/or crashes,
as userspace could free buffers while the hardware is still writing to
them, and the frame completion interrupt handler could try to access
buffers that don't exist anymore.

Fix this by stopping the pipeline upon the first STREAMOFF call, not the
last.
Reported-by: NKuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

The structures are never modified, make them const.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

The frame sequence number is global to the pipeline, there's no need to
store copies in each video node.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@s-opensource.com>

Instead of embedding pipelines in the vsp1_video objects allocate them
on demand when they are needed. This fixes the streamon race condition
where pipelines objects from different video nodes could be used for the
same pipeline.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Move the pipeline initialization and cleanup functions to prepare for
the next commit. No functional code change is performed here.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

This prepares for dynamic pipeline allocation by providing a field that
can be used to store the pipeline pointer atomically under driver
control.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Pass the pipe explicitly instead of retrieving it through media
entities. This decouples device state stored in the pipeline from the
active state stored in entities, preparing for dynamic pipeline
creation.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

The primary purpose of those functions is to build the pipeline, rename
them to make this clearer.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Modules write register values to the active display list pointed to by
the pipeline. In order to support preparing display lists ahead of time,
pass them explicitly to all configuration functions.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Reuse the runtime logic to initialize the default format instead of
open-coding it. This ensures coherency between intialization and
runtime.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

The subdev s_stream operation is abused as a generic way to setup
modules at every frame. Move the code out to a new VSP1 entity configure
operation.

Most modules now have an empty s_stream operation that can be removed.
The only exception is the WPF module that needs to perform hardware
configuration when stopping the stream. The code can be simplified
accordingly as we know that that operation never fails.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Use the pad config structure part of the vsp1_entity to store all active
pad selection rectangles. This generalizes the code to operate on pad
config structures.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Don't restrict display list usage to the DRM pipeline, use them
unconditionally. This prepares the driver to support the request API.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>