| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
The v4l2_subdev_link_validate() helper function is meant to be used as a
drop-in implementation of a V4L2 subdev entity .link_validate() handler.
It supports subdev-to-subdev links only, and complains if one end of the
link is not a subdev. This forces drivers that have video output devices
connected to subdevs to implement a custom .link_validate() handler,
calling v4l2_subdev_link_validate() for the subdev-to-subdev links, and
performing manual link validation for the video-to-subdev links.
Video devices embed a media entity, and therefore also have a
.link_validate() operation. For video capture devices, the operation
should be manually implemented by drivers for validate the
subdev-to-video links. For video output devices, on the other hand, that
operation is never called, as link validation is performed in the
context of the sink entity.
As a result, we end up forcing drivers to implement a custom
.link_validate() handler for subdevs connected to video output devices,
when the video devices provide an operation that could be used for that
purpose.
To improve that situation, make v4l2_subdev_link_validate() delegate
link validation to the source's .link_validate() operation when the link
source is a video device and the link sink is a subdev. This allows
broader usage of v4l2_subdev_link_validate(), and simplifies drivers by
making video device link validation easy to implement in the video
device .link_validate(), regardless of whether the video device is an
output device or a capture device.
Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
Acked-by: Sakari Ailus <[email protected]>
|
|
The v4l2_subdev_link_validate() function prints a one-time warning if it
gets called on a link whose source or sink is not a subdev. As links get
validated in the context of their sink, a call to the helper when the
link's sink is not a subdev indicates that the driver has set its
.link_validate() handler to v4l2_subdev_link_validate() on a non-subdev
entity, which is a clear driver bug. On the other hand, the link's
source not being a subdev indicates that the helper is used for a subdev
connected to a video output device, which is a lesser issue, if only
because this is currently common practice.
There are no drivers left in the kernel that use
v4l2_subdev_link_validate() in a context where it may get called on a
non-subdev sink. Replace the pr_warn_once() with a WARN_ON_ONCE() in
this case to make sure that new offenders won't be introduced.
A subsequent change will improve the v4l2_subdev_link_validate() helper
to properly support validating video device to subdev links.
Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
Acked-by: Sakari Ailus <[email protected]>
|
|
The v4l2_subdev_link_validate_get_format() function warns if the pad
given as argument belongs to a non-subdev entity. This can't happen, as
the function is called from v4l2_subdev_link_validate() only (indirectly
through v4l2_subdev_link_validate_locked()), and that function ensures
that both ends of the link are subdevs.
Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
|
|
At the moment v4l2_subdev_s_stream_helper() only works for subdevices
that support routing. As enable/disable_streams now also works for
subdevices without routing, improve v4l2_subdev_s_stream_helper() to do
the same.
Reviewed-by: Laurent Pinchart <[email protected]>
Tested-by: Umang Jain <[email protected]>
Signed-off-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
v4l2_subdev_enable/disable_streams()
At the moment the v4l2_subdev_enable/disable_streams() functions call
fallback helpers to handle the case where the subdev only implements
.s_stream(), and the main function handles the case where the subdev
implements streams (V4L2_SUBDEV_FL_STREAMS, which implies
.enable/disable_streams()).
What is missing is support for subdevs which do not implement streams
support, but do implement .enable/disable_streams(). Example cases of
these subdevices are single-stream cameras, where using
.enable/disable_streams() is not required but helps us remove the users
of the legacy .s_stream(), and subdevices with multiple source pads (but
single stream per pad), where .enable/disable_streams() allows the
subdevice to control the enable/disable state per pad.
The two single-streams cases (.s_stream() and .enable/disable_streams())
are very similar, and with small changes we can change the
v4l2_subdev_enable/disable_streams() functions to support all three
cases, without needing separate fallback functions.
A few potentially problematic details, though:
- For the single-streams cases we use sd->enabled_pads field, which
limits the number of pads for the subdevice to 64. For simplicity I
added the check for this limitation to the beginning of the function,
and it also applies to the streams case.
- The fallback functions only allowed the target pad to be a source pad.
It is not very clear to me why this check was needed, but it was not
needed in the streams case. However, I doubt the
v4l2_subdev_enable/disable_streams() code has ever been tested with
sink pads, so to be on the safe side, I added the same check
to the v4l2_subdev_enable/disable_streams() functions.
Reviewed-by: Laurent Pinchart <[email protected]>
Tested-by: Umang Jain <[email protected]>
Signed-off-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Add two internal helper functions, v4l2_subdev_collect_streams() and
v4l2_subdev_set_streams_enabled(), which allows us to refactor
v4l2_subdev_enable/disable_streams() functions.
This (I think) makes the code a bit easier to read, and lets us more
easily add new functionality in the helper functions in the following
patch.
Reviewed-by: Laurent Pinchart <[email protected]>
Tested-by: Umang Jain <[email protected]>
Signed-off-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
We support camera privacy leds with the .s_stream() operation, in
call_s_stream(), but we don't have that support when the subdevice
implements .enable/disable_streams() operations.
Add the support by enabling the led when the first stream for a
subdevice is enabled, and disabling the led then the last stream is
disabled.
Reviewed-by: Umang Jain <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Tested-by: Umang Jain <[email protected]>
Signed-off-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Add a helper function which returns whether the subdevice is streaming,
i.e. if .s_stream or .enable_streams has been called successfully.
Reviewed-by: Umang Jain <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Tested-by: Umang Jain <[email protected]>
Signed-off-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
v4l2_subdev_enable/disable_streams_fallback() supports falling back to
.s_stream() for subdevs with a single source pad. It also tracks the
enabled streams for that one pad in the sd->enabled_streams field.
Tracking the enabled streams with sd->enabled_streams does not make
sense, as with .s_stream() there can only be a single stream per pad.
Thus, as the v4l2_subdev_enable/disable_streams_fallback() only supports
a single source pad, all we really need is a boolean which tells whether
streaming has been enabled on this pad or not.
However, as we only need a true/false state for a pad (instead of
tracking which streams have been enabled for a pad), we can easily
extend the fallback mechanism to support multiple source pads as we only
need to keep track of which pads have been enabled.
Change the sd->enabled_streams field to sd->enabled_pads, which is a
64-bit bitmask tracking the enabled source pads. With this change we can
remove the restriction that
v4l2_subdev_enable/disable_streams_fallback() only supports a single
source pad.
Reviewed-by: Laurent Pinchart <[email protected]>
Tested-by: Umang Jain <[email protected]>
Signed-off-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
call_s_stream() uses sd->enabled_streams to track whether streaming has
already been enabled. However,
v4l2_subdev_enable/disable_streams_fallback(), which was the original
user of this field, already uses it, and
v4l2_subdev_enable/disable_streams_fallback() will call call_s_stream().
This leads to a conflict as both functions set the field. Afaics, both
functions set the field to the same value, so it won't cause a runtime
bug, but it's still wrong and if we, e.g., change how
v4l2_subdev_enable/disable_streams_fallback() operates we might easily
cause bugs.
Fix this by adding a new field, 's_stream_enabled', for
call_s_stream().
Reviewed-by: Umang Jain <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Tested-by: Umang Jain <[email protected]>
Signed-off-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Add some checks to verify that the subdev driver implements required
features.
A subdevice that supports streams (V4L2_SUBDEV_FL_STREAMS) must do one
of the following:
- Implement neither .enable/disable_streams() nor .s_stream(), if the
subdev is part of a video driver that uses an internal method to
enable the subdev.
- Implement only .enable/disable_streams(), if support for legacy
sink-side subdevices is not needed.
- Implement .enable/disable_streams() and .s_stream(), if support for
legacy sink-side subdevices is needed.
At the moment the framework doesn't check this requirement. Add the
check.
Reviewed-by: Umang Jain <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Tested-by: Umang Jain <[email protected]>
Signed-off-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Use v4l2_subdev_has_op() in v4l2_subdev_enable/disable_streams() instead
of open coding the same.
Reviewed-by: Umang Jain <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Tested-by: Umang Jain <[email protected]>
Signed-off-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Add helper functions to enable and disable the privacy led. This moves
the #if from the call site to the privacy led functions, and makes
adding privacy led support to v4l2_subdev_enable/disable_streams()
cleaner.
Reviewed-by: Umang Jain <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Tested-by: Umang Jain <[email protected]>
Signed-off-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Add trivial S_ROUTING IOCTL support for drivers where routing is static.
Essentially this means returning the same information G_ROUTING call would
have done.
Signed-off-by: Sakari Ailus <[email protected]>
Co-developed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Julien Massot <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Return the routes set using S_ROUTING back to the user. Also reflect this
in documentation.
Signed-off-by: Sakari Ailus <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
The len_routes field is used to tell the size of the routes array in
struct v4l2_subdev_routing. This way the number of routes returned from
S_ROUTING IOCTL may be larger than the number of routes provided, in case
there are more routes returned by the driver.
Note that this uAPI is still disabled in the code, so this change can
safely be done. Anyone who manually patched the code to enable this uAPI
must update their code.
The patch also increases the number of reserved fields in struct
v4l2_subdev_routing.
Signed-off-by: Sakari Ailus <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Add two new functions, v4l2_subdev_lock_states() and
v4l2_subdev_unclock_states(), to acquire and release the state of two
sub-devices. They differ from calling v4l2_subdev_{un,}lock_state() so
that if the two states share the same lock, the lock is acquired only
once.
Also use the new functions in v4l2_subdev_link_validate().
Signed-off-by: Sakari Ailus <[email protected]>
Reviewed-by: Julien Massot <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Turn on the privacy LED only if streamon succeeds. This can be done after
enabling streaming on the sensor.
Fixes: b6e10ff6c23d ("media: v4l2-core: Make the v4l2-core code enable/disable the privacy LED if present")
Signed-off-by: Sakari Ailus <[email protected]>
Reviewed-by: Hans de Goede <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
Reviewed-by: Umang Jain <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
When support for streams was added to the V4L2 subdev API, the
v4l2_subdev_crop structure was extended with a stream field, but the
field was not handled in the core code that translates the
VIDIOC_SUBDEV_[GS]_CROP ioctls to the selection API. Fix it.
Fixes: 2f91e10ee6fd ("media: subdev: add stream based configuration")
Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
After the conversion of dv timing calls to use a pad argument is done,
remove the old callbacks. Update the subdev ioctl handlers to use the
new callbacks.
Signed-off-by: Paweł Anikiel <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Currently, subdev dv timing calls (i.e. g/s/query_dv_timings) are video
ops without a pad argument. This is a problem if the subdevice can have
different dv timings for each pad (e.g. a DisplayPort receiver with
multiple virtual channels).
To solve this, change these calls to include a pad argument, and put
them into pad ops. Keep the old ones temporarily to make the switch
easier.
Signed-off-by: Paweł Anikiel <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Subdev states store all standard pad configuration data, except for
frame intervals. Fix it by adding interval fields in the
v4l2_subdev_pad_config and v4l2_subdev_stream_config structures, with
corresponding accessor functions and a helper function to implement the
.get_frame_interval() operation.
Signed-off-by: Laurent Pinchart <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Due to a historical mishap, the v4l2_subdev_frame_interval structure
is the only part of the V4L2 subdev userspace API that doesn't contain a
'which' field. This prevents trying frame intervals using the subdev
'TRY' state mechanism.
Adding a 'which' field is simple as the structure has 8 reserved fields.
This would however break userspace as the field is currently set to 0,
corresponding to V4L2_SUBDEV_FORMAT_TRY, while the corresponding ioctls
currently operate on the 'ACTIVE' state. We thus need to add a new
subdev client cap, V4L2_SUBDEV_CLIENT_CAP_INTERVAL_USES_WHICH, to
indicate that userspace is aware of this new field.
All drivers that implement the subdev .get_frame_interval() and
.set_frame_interval() operations are updated to return -EINVAL when
operating on the TRY state, preserving the current behaviour.
While at it, fix a bad copy&paste in the documentation of the struct
v4l2_subdev_frame_interval_enum 'which' field.
Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Philipp Zabel <[email protected]> # for imx-media
Reviewed-by: Hans Verkuil <[email protected]>
Reviewed-by: Luca Ceresoli <[email protected]> # for tegra-video
Reviewed-by: Mauro Carvalho Chehab <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
The subdev .[gs]_frame_interval are video operations, but they operate
on pads (and even on streams). Not only is this confusing, it causes
practical issues for drivers as the operations don't receive a subdev
state pointer, requiring manual state handling.
To improve the situation, turn the operations into pad operations, and
extend them to receive a state pointer like other pad operations.
While at it, rename the operations to .[gs]et_frame_interval at the same
time to match the naming scheme of other pad operations. This isn't
strictly necessary, but given that all drivers using those operations
need to be modified, handling the rename separately would generate more
churn for very little gain (if at all).
Signed-off-by: Laurent Pinchart <[email protected]>
Reviewed-by: Hans Verkuil <[email protected]>
Reviewed-by: Philipp Zabel <[email protected]> # for imx-media
Reviewed-by: Luca Ceresoli <[email protected]> # for tegra-video
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Graciously handle an uninitialised (but still zeroed) sub-device in
v4l2_subdev_cleanup(). The list_empty() check there is unnecessary, too,
so replace that by cheking whether the lists's next field is NULL.
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
The subdev .init_cfg() operation is affected by two issues:
- It has long been extended to initialize a whole v4l2_subdev_state
instead of just a v4l2_subdev_pad_config, but its name has stuck
around.
- Despite operating on a whole subdev state and not being directly
exposed to the subdev users (either in-kernel or through the userspace
API), .init_cfg() is categorized as a subdev pad operation.
This participates in making the subdev API confusing for new developers.
Fix it by renaming the operation to .init_state(), and make it a subdev
internal operation.
Signed-off-by: Laurent Pinchart <[email protected]>
Acked-by: Michael Riesch <[email protected]> # for imx415
Acked-by: Shuah Khan <[email protected]> # for vimc
Reviewed-by: Philipp Zabel <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
[Sakari Ailus: Resolved a conflict in Renesas vsp1 driver.]
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Return NULL from sub-device pad state access functions
(v4l2_subdev_state_get_{format,crop,compose}) for non-existent pads. While
this behaviour differs from older set of pad state information access
functions, we've had a WARN_ON() there for a long time and callers also do
validate the pad index nowadays. Therefore problems are not expected.
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
Compile sub-device state information access functions
v4l2_subdev_state_get_{format,crop,compose} unconditionally as they are
now also used by plain V4L2 drivers.
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
The sub-device state access functions take three arguments: sub-device
state, pad and stream. The stream is not relevant for the majority of
drivers and having to specify 0 for the stream is considered a nuisance.
Provide a two-argument macros for these state access functions to cover
the needs of stream-unaware users.
Signed-off-by: Sakari Ailus <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
Now that v4l2_subdev_state_get_format() always returns format, don't call
alternative v4l2_subdev_get_pad_format() anymore.
Signed-off-by: Sakari Ailus <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
Rename the sub-devices state information access functions, removing
"_stream" from them and replacing "format" by "ffmt". This makes them
shorter and so more convenient to use. No other sets of functions will be
needed to access this information.
Signed-off-by: Sakari Ailus <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
There are two sets of functions that return information from sub-device
state, one for stream-unaware users and another for stream-aware users.
Add support for stream-aware functions to return format, crop and compose
information from pad-based array that are functionally equivalent to the
old, stream-unaware ones.
Also check state is non-NULL, in order to guard against old drivers
potentially calling this with NULL state for active formats or selection
rectangles.
Signed-off-by: Sakari Ailus <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
Store the sub-device in the sub-device state. This will be needed in e.g.
validating pad number when retrieving information for non-stream-aware
users. There are expected to be more needs for this in the future.
Signed-off-by: Sakari Ailus <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media
Pull media updates from Mauro Carvalho Chehab:
- the old V4L2 core videobuf kAPI was finally removed. All media
drivers should now be using VB2 kAPI
- new automotive driver: mgb4
- new platform video driver: npcm-video
- new sensor driver: mt9m114
- new TI driver used in conjunction with Cadence CSI2RX IP to bridge
TI-specific parts
- ir-rx51 was removed and the N900 DT binding was moved to the
pwm-ir-tx generic driver
- drop atomisp-specific ov5693, using the upstream driver instead
- the camss driver has gained RDI3 support for VFE 17x
- the atomisp driver now detects ISP2400 or ISP2401 at run time. No
need to set it up at build time anymore
- lots of driver fixes, cleanups and improvements
* tag 'media/v6.7-1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media: (377 commits)
media: nuvoton: VIDEO_NPCM_VCD_ECE should depend on ARCH_NPCM
media: venus: Fix firmware path for resources
media: venus: hfi_cmds: Replace one-element array with flex-array member and use __counted_by
media: venus: hfi_parser: Add check to keep the number of codecs within range
media: venus: hfi: add checks to handle capabilities from firmware
media: venus: hfi: fix the check to handle session buffer requirement
media: venus: hfi: add checks to perform sanity on queue pointers
media: platform: cadence: select MIPI_DPHY dependency
media: MAINTAINERS: Fix path for J721E CSI2RX bindings
media: cec: meson: always include meson sub-directory in Makefile
media: videobuf2: Fix IS_ERR checking in vb2_dc_put_userptr()
media: platform: mtk-mdp3: fix uninitialized variable in mdp_path_config()
media: mediatek: vcodec: using encoder device to alloc/free encoder memory
media: imx-jpeg: notify source chagne event when the first picture parsed
media: cx231xx: Use EP5_BUF_SIZE macro
media: siano: Drop unnecessary error check for debugfs_create_dir/file()
media: mediatek: vcodec: Handle invalid encoder vsi
media: aspeed: Drop unnecessary error check for debugfs_create_file()
Documentation: media: buffer.rst: fix V4L2_BUF_FLAG_PREPARED
Documentation: media: gen-errors.rst: fix confusing ENOTTY description
...
|
|
disabled
Since the stream API is still experimental it is currently locked away
behind the internal, default disabled, v4l2_subdev_enable_streams_api flag.
Advertising V4L2_SUBDEV_CAP_STREAMS when the streams API is disabled
confuses userspace. E.g. it causes the following libcamera error:
ERROR SimplePipeline simple.cpp:1497 Failed to reset routes for
/dev/v4l-subdev1: Inappropriate ioctl for device
Don't report V4L2_SUBDEV_CAP_STREAMS when the streams API is disabled
to avoid problems like this.
Reported-by: Dennis Bonke <[email protected]>
Fixes: 9a6b5bf4c1bb ("media: add V4L2_SUBDEV_CAP_STREAMS")
Cc: [email protected] # for >= 6.3
Signed-off-by: Hans de Goede <[email protected]>
Acked-by: Sakari Ailus <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Print debug level information on returned frame descriptors.
Signed-off-by: Sakari Ailus <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
Clear frame descriptor before calling transmitter's get_frame_desc() op.
Also remove the corresponding memset() calls from drivers.
Signed-off-by: Sakari Ailus <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
The subdev .s_stream() operation must not be called to start an already
started subdev, or stop an already stopped one. This requirement has
never been formally documented. Fix it, and catch possible offenders
with a WARN_ON() in the call_s_stream() wrapper.
Signed-off-by: Laurent Pinchart <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Hans Verkuil <[email protected]>
|
|
It is often useful to see when streaming for a device is being enabled
or disabled. Add debug prints for this to v4l2_subdev_enable_streams()
and v4l2_subdev_disable_streams().
Signed-off-by: Tomi Valkeinen <[email protected]>
Reviewed-by: Jacopo Mondi <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
The routing parameter of v4l2_subdev_set_routing_with_fmt() is missing
'const'. Add it.
Signed-off-by: Tomi Valkeinen <[email protected]>
Reviewed-by: Jacopo Mondi <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
Now that the kernel drivers have been fixed to initialize the stream
field, and we have the client capability which the userspace uses to say
it has initialized the stream field, we can drop the implicit zeroing of
the stream field in the various check functions.
Signed-off-by: Tomi Valkeinen <[email protected]>
Reviewed-by: Jacopo Mondi <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
Support matching V4L2 async sub-devices based on particular fwnode
endpoint. This makes it possible to instantiate multiple V4L2 sub-devices
based on given fwnode endpoints from a single device, based on driver
needs.
Signed-off-by: Sakari Ailus <[email protected]>
Tested-by: Philipp Zabel <[email protected]> # imx6qp
Tested-by: Niklas Söderlund <[email protected]> # rcar + adv746x
Tested-by: Aishwarya Kothari <[email protected]> # Apalis i.MX6Q with TC358743
Tested-by: Lad Prabhakar <[email protected]> # Renesas RZ/G2L SMARC
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
Add new ioctls to set and get subdev client capabilities. Client in this
context means the userspace application which opens the subdev device
node. The client capabilities are stored in the file handle of the
opened subdev device node, and the client must set the capabilities for
each opened subdev.
For now we only add a single flag, V4L2_SUBDEV_CLIENT_CAP_STREAMS, which
indicates that the client is streams-aware.
The reason for needing such a flag is as follows:
Many structs passed via ioctls, e.g. struct v4l2_subdev_format, contain
reserved fields (usually a single array field). These reserved fields
can be used to extend the ioctl. The userspace is required to zero the
reserved fields.
We recently added a new 'stream' field to many of these structs, and the
space for the field was taken from these reserved arrays. The assumption
was that these new 'stream' fields are always initialized to zero if the
userspace does not use them. This was a mistake, as, as mentioned above,
the userspace is required to zero the _reserved_ fields. In other words,
there is no requirement to zero this new stream field, and if the
userspace doesn't use the field (which is the case for all userspace
applications at the moment), the field may contain random data.
This shows that the way the reserved fields are defined in v4l2 is, in
my opinion, somewhat broken, but there is nothing to do about that.
To fix this issue we need a way for the userspace to tell the kernel
that the userspace has indeed set the 'stream' field, and it's fine for
the kernel to access it. This is achieved with the new ioctl, which the
userspace should usually use right after opening the subdev device node.
Signed-off-by: Tomi Valkeinen <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
A common case with subdev routing is that on the subdevice just before
the DMA engines (video nodes), no multiplexing is allowed on the source
pads, as the DMA engine can only handle a single stream.
In some other situations one might also want to do the same check on the
sink side.
Add new routing validation flags to check these:
V4L2_SUBDEV_ROUTING_NO_SINK_MULTIPLEXING and
V4L2_SUBDEV_ROUTING_NO_SOURCE_MULTIPLEXING.
Signed-off-by: Tomi Valkeinen <[email protected]>
Reviewed-by: Jacopo Mondi <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
V4L2_SUBDEV_ROUTING_NO_STREAM_MIX routing validation flag means that all
routes from a sink pad must go to the same source pad and all routes
going to the same source pad must originate from the same sink pad.
This does not cover all use cases. For example, if a device routes
all streams from a single sink pad to any of the source pads, but
streams from multiple sink pads can go to the same source pad, the
current flag is too restrictive.
Split the flag into two parts, V4L2_SUBDEV_ROUTING_NO_SINK_STREAM_MIX
and V4L2_SUBDEV_ROUTING_NO_SOURCE_STREAM_MIX, which add the restriction
only on one side of the device. Together they mean the same as
V4L2_SUBDEV_ROUTING_NO_STREAM_MIX.
Signed-off-by: Tomi Valkeinen <[email protected]>
Reviewed-by: Andy Shevchenko <[email protected]>
Reviewed-by: Jacopo Mondi <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
Route validation docs use the word 'may'. Change that to 'shall' for
emphasis.
Signed-off-by: Tomi Valkeinen <[email protected]>
Reviewed-by: Jacopo Mondi <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
Link validation currently accesses invalid pointers if the link passed to
it is not between two sub-devices. This is of course a driver bug.
Ignore the error but print a warning message, as this is how it used to
work previously.
Fixes: a6b995ed03ff ("media: subdev: use streams in v4l2_subdev_link_validate()")
Reported-by: Hans de Goede <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Tested-by: Hans de Goede <[email protected]>
Reviewed-by: Tomi Valkeinen <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
The new subdev state code has a possible deadlock scenario during link
validation when the pipeline contains subdevs that support state and
that do not support state.
The current code locks the states of the subdevs on both ends of the
link when starting the link validation, locking the sink side first,
then the source. If either (or both) of the subdevs does not support
state, nothing is done for that subdev at this point, and instead the
locking is handled the old way, i.e. the subdev's ops do the locking
internally.
The issue arises when the sink doesn't support state, but source does,
so the validation code locks the source for the duration of the
validation, and then the sink is locked only when the get_fmt op is
called. So lockdep sees the source locked first, then the sink.
Later, when the streaming is started, the sink's s_stream op is called,
which probably takes the subdev's lock. The op then calls the source's
s_stream, which takes the source's lock. So, the sink is locked first,
then the source.
Note that link validation and stream starting is not done at the same
time, so an actual deadlock should never happen. However, it's still a
clear bug.
Fix this by locking the subdev states only if both subdevs support
state. In other words, we have two scenarios:
1. Both subdevs support state. Lock sink first, then source, and keep
the locks while validating the link.
2. At least one of the subdevs do not support state. Take the lock only
for the duration of the operation (get_fmt or looking at the
routing), and release after the op is done.
Obviously 1. is better, as we have a more consistent view of the states
of the subdevs during validation. 2. is how it has been so far, so it's
no worse than this used to be.
Signed-off-by: Tomi Valkeinen <[email protected]>
Signed-off-by: Sakari Ailus <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media
Pull media updates from Mauro Carvalho Chehab:
- Removal of several VB1-only deprecated drivers: cpia2, fsl-viu, meye,
stkwebcam, tm6000, vpfe_capture and zr364xx
- saa7146 recovered from staging/deprecated. We opted to give ti a
chance, and, instead of deprecating it, the intention is to write
patches migrating it from VB1 to VB2.
- av7110 returned from staging/deprecated/ to staging/ as we're not
planning on dropping it any time soon
- media controller API has gained experimental support for G_ROUTING
and streams API. No drivers use it right now. We're planning to add
one after -rc1, giving some time to experience the API and eventually
have changes during the next development cycle
- New sensor drivers: imx296, imx415, ov8858
- Atomisp had lots of changes, specially on its sensor's interface,
making atomisp sensor drivers closer to normal sensor drivers
- media controller kAPI has gained some helpers to traverse pipelines
- uvcvideo now better support power line control
- lots of bug fixes, cleanups and driver improvements
* tag 'media/v6.3-1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media: (296 commits)
media: imx-mipi-csis: Check csis_fmt validity before use
media: v4l2-subdev.c: clear stream field
media: v4l2-ctrls-api.c: move ctrl->is_new = 1 to the correct line
media: Revert "media: saa7146: deprecate hexium_gemini/orion, mxb and ttpci"
media: Revert "media: av7110: move to staging/media/deprecated/saa7146"
media: imx-pxp: convert to regmap
media: imx-pxp: Use non-threaded IRQ
media: imx-pxp: Introduce pxp_read() and pxp_write() wrappers
media: imx-pxp: Implement frame size enumeration
media: imx-pxp: Pass pixel format value to find_format()
media: imx-pxp: Add media controller support
media: imx-pxp: Don't set bus_info manually in .querycap()
media: imx-pxp: Sort headers alphabetically
media: imx-pxp: add support for i.MX7D
media: imx-pxp: make data_path_ctrl0 platform dependent
media: imx-pxp: disable LUT block
media: imx-pxp: explicitly disable unused blocks
media: imx-pxp: extract helper function to setup data path
media: imx-pxp: detect PXP version
media: dt-bindings: media: fsl-pxp: convert to yaml
...
|
|
Both userspace and kernelspace can pass structs with an uninitialized
'stream' field. Since the check_state() function checks for a non-zero
stream field, suddenly these calls will fails with -EINVAL.
So check in the wrapper functions in v4l2-subdev.c (which are used by both
the kernel and userspace API) if V4L2_SUBDEV_FL_STREAMS is set, and if not,
then zero the stream field.
Currently no drivers set V4L2_SUBDEV_FL_STREAMS, so the stream field will
always be set to 0.
This patch might well be reverted in the future when streams support is
fully enabled and we finalized the userspace API support for this feature.
Signed-off-by: Hans Verkuil <[email protected]>
Reviewed-by: Laurent Pinchart <[email protected]>
Signed-off-by: Mauro Carvalho Chehab <[email protected]>
|