In order to clear the V4L2VideoDevice cache, we must call
V4L2VideoDevice::releaseBuffers() when stopping the camera. To do this
without releasing the allocated buffers, we have to rework the internal
buffer allocation logic.
Remove calls to V4L2VideoDevice::importBuffers() and releaseBuffers()
from within the RPi::Stream class. Instead, move these calls to the
PipelineHandlerBase class, where we can call releaseBuffers(), i.e. clear
the V4L2VideoDevice cache unconditionally on stop without de-allocating
the internal buffers.
The code in Stream::clearBuffers() can be then moved into releaseBuffers()
which will actually de-allocate the internal buffers when required.
Closes: https://gitlab.freedesktop.org/camera/libcamera/-/issues/265
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Tested-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com> # rpi4
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Signed-off-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
If the first frame of a stream is bad, the IPA will not get called with
frame == 0, leaving activeState.agc expo/again uninitialized. This causes
the agc algorithm to set a very low gain and exposure on the next run
(where it will hit the if (!stats->valid) {} path) resulting in starting
with a black image.
Fix this by using a valid flag instead of checking for frame == 0.
The entire activeState gets cleared to 0 on configure() resetting the new
valid flag.
Signed-off-by: Hans de Goede <johannes.goede@oss.qualcomm.com>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
On most imx sensors the formula seems to be:
again_float = 512.0 / (512.0 - again_ctrl_value)
So the minimum again of 0 makes sense and actually translates to a gain of 1.0.
And the max gain of 400 leads to 512.0 / 112.0 = 4.57 which is about typical for
a maximum again. Since a minimum again value of 0 is actually normal, the special
handling of againMin == 0 is undesirable and this is actually causing problems
on these IMX sensors. again10 correctly gets set to 0 which is less than the
adjusted againMin which causes the AGC code to not work properly.
Fix this by dropping the special handling of againMin == 0.
Signed-off-by: Vasiliy Doylov <nekocwd@mainlining.org>
Reviewed-by: Hans de Goede <johannes.goede@oss.qualcomm.com>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Save the result of `Camera::generateConfiguration()` in a local variable,
and only commit it to the member variable `config_` if it passes all checks.
This removes the need for explicit `config_.reset()` calls in error paths,
hence making things simpler.
Signed-off-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Stefan Klug <stefan.klug@ideasonboard.com>
The ControlList associated with a Request are created with
the libcamera::controls::controls id map.
The cam application, when parsing controls from a script
assigns to the Request's control list a newly created one, which is
initialized without a valid idmap or info map.
This causes IPA that run in isolated mode to fail, as the IPC
serialization/deserialization code requires all ControlList to have
a valid idmap or info map associated.
Instead of overwriting the Request::controls() list, merge it with
the one created by cam when parsing the capture script.
Closes: https://gitlab.freedesktop.org/camera/libcamera/-/issues/295
Fixes: 568865a6c1 ("cam: Use script parser to set controls")
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The list of controls part of a Request is initialized with the
ControlInfoMap of the Camera the Request is created from.
Applications can re-assign the controls list in a Request
which could cause issues during serialization.
Validate that the ControlList in a Request is valid when
the Request is queued to the Camera by inspecting the
ControlInfoMap pointer.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Reviewed-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Address a long standing \todo item that suggested to implement a
read-only interface for the Request::metadata() accessor and deflect to
the internal implementation for the read-write accessor used by pipeline
handlers.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Acked-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
The control lists associated with a Request is created with the global
libcamera::controls::controls id map.
This is fine as the idmap/info map used to contruct a ControlList are not
used for any control validation purposes by the libcamera code.
However creating a ControlList with the camera ControlInfoMap has two
advantages:
1) The idmap can be extracted from the info map, but not the other way
around
2) The control list is constructed with a valid map of info to the
controls it can actually supports, instead of the global map of
libcamera controls
Initialize the ControlList part of a Request with the Camera control
info map, as the association between a Request and a Camera is permanent
and valid for the whole lifetime of a Request.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
The existing RkISP1Params helper class allows the RkISP1 IPA to handle
both the extensible parameters format and the legacy fixed-size format.
With the introduction of v4l2-isp.h in the Linux kernel the part of
the RkISP1Params helper class that handles extensible parameters can
be generalized so that other IPA modules can use the same helpers
to populate a parameters buffer compatible with v4l2-isp.h.
Generalize the RkISP1Params class to a new libipa component named
V4L2Params and derive the existing RkISP1Params from it, leaving
in the RkISP1-specific implementation the handling of the legacy format.
Deriving RkISP1Params from V4L2Params requires changing the size
associated to each block to include the size of v4l2_params_block_header
in the ipa:rkisp1::kBlockTypeInfo map as the V4L2Params::block()
implementation doesn't account for that as RkIS1Params::block()
implementation did.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Tested-by: Antoine Bouyer <antoine.bouyer@nxp.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Import the v4l2-isp.h header from the Linux kernel version v6.19-rc1 at
revision e36dbd1cf3df ("media: uapi: Introduce V4L2 generic ISP types").
Create the include/linux/media/ directory so that header files exported
from the kernel which include this file do not need to be adjusted when
imported in libcamera.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Tested-by: Antoine Bouyer <antoine.bouyer@nxp.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Import stddef.h from the Linux kernel sources v6.19-rc1 at revision
c2a756891bb4 ("uapi: wrap compiler_types.h in an ifdef instead of the
implicit strip")
The stddef.h header is imported to provide the __counted_by() symbol
which is not available in the header version shipped with, in example,
Debian 12. __counted_by() will be used by the v4l2-isp.h header which
will be imported next.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Acked-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Add a table with the list of camera sensors currently supported by
libcamera. Similar to the ISP feature matrix, this too is a living
document, and will be updated as support for more sensors is added.
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Jai Luthra <jai.luthra@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Add a new living document presenting a matrix of image processing
feature support across all currently supported platforms in libcamera.
This will hopefuly help in answering questions like is HDR supported
with Software ISP or is Auto Focus supported on Raspberry Pi?
This matrix will be regularly updated as new features and platforms are
added.
Acked-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Stefan Klug <stefan.klug@ideasonboard.com>
Signed-off-by: Jai Luthra <jai.luthra@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
It is possible that same device is processed multiple times, leading to
multiple `MediaDevice`s being instantiated, mostly likely leading to
a fatal error:
Trying to register a camera with a duplicated ID xyzw...
There is a time window after the `udev_monitor` has been created in `init()`
and the first (and only) enumeration done in `enumerate()`. If e.g. a UVC
camera is connected in this time frame, then it is possible that it will be
processed both by the `udevNotify()` and the initial `enumerate()`, leading
to the fatal error. This can be reproduced as follows:
1. $ gdb --args cam -m
2. (gdb) break libcamera::DeviceEnumeratorUdev::enumerate
3. (gdb) run
4. when the breakpoint is hit, connect a usb camera
5. (gdb) continue
6. observe fatal error
To address this, keep track of the devnums of all devices reported by
udev, and reject devices with already known devnums. This ensures that
the same device won't be reported multiple times (assuming that udev
reports "add" / "remove" events in the correct order).
Closes: https://gitlab.freedesktop.org/camera/libcamera/-/issues/293
Signed-off-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
The AWB gains reported in the metadata are currently scaled against the
maximum gain, but in fact the gain itself is already stored as a
floating point gain value.
Remove the incorrect scaling and report the gains directly.
Fixes: a0b97475b1 ("ipa: simple: Report the ColourGains in metadata")
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Meson already takes care of passing the proper absolute or relative
paths to commands. There is no need do more path manipulation.
So simplify the script by using the paths as-is. This also fixes the
path manipulation issue that prevented libcamera from building as a
subproject.
Fixes: 19371dee41 ("utils: gen-shader-headers: Add a utility to generate headers from shaders")
Signed-off-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The current definition of the stream-role property of libcamerasrc
element does not allow for the Raw (0) value to be assigned.
That property is used to query the pipeline handlers through
generateConfiguration(roles) to retrieve the formats exposed to
the user. Not being able to specify the raw stream-role, bayer
formats are not reported so can not be used for caps negotiation.
Adding value stream-role=raw enables usage of bayer GStreamer pipelines
like:
`gst-launch-1.0 libcamerasrc camera-name=<name> src::stream-role=raw !
video/x-bayer, format=bggr16le ! ...`
Signed-off-by: Julien Vuillaumier <julien.vuillaumier@nxp.com>
Reviewed-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Update videodev2.h to Linux kernel version v6.18.
The downstream defined RPi CFE stats and parameters formats have found
their way in upstream so we can drop the local definitions.
Also, new definitions for the Amlogic C3 ISP formats and Renesas CRU
formats have been added upstream.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Update media-bus-format.h to Linux kernel version v6.18.
Mainline has introduced two new media bus codes whose definition
conflicts with the downstream definition of
MEDIA_BUS_FMT_RGB202020_1X60.
Update the definition of MEDIA_BUS_FMT_RGB202020_1X60 not to conflict
with the newly added symbol.
MEDIA_BUS_FMT_RGB202020_1X60, which is used by the Mali C55 ISP,
will land in v6.19 with value 0x1028, so this change actually
aligns the downstream definition with the forthcoming upstream
support for Mali C55.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Acked-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Update drm_fourcc.h to Linux kernel version v6.18.
As the upstream drm_fourcc.h version doesn't include definitions
for RAW Bayer formats, some of the symbols we defined
downstream now conflict with new symbols defined in mainline.
In particular, mainline has added:
#define DRM_FORMAT_MOD_VENDOR_MTK 0x0b
#define DRM_FORMAT_MOD_VENDOR_APPLE 0x0c
Which conflict with the downstream definitions:
#define DRM_FORMAT_MOD_VENDOR_MIPI 0x0b
#define DRM_FORMAT_MOD_VENDOR_RPI 0x0c
In order not to break the library ABI, maintain the downstream symbols
definitions as they are even if they conflict.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Acked-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
When a raw stream is requested, whether alone or together with a
processed stream, its buffers must be handled outside the software ISP
machinery. They serve as output buffers, even when a processed stream
is produced. But when both processed and raw streams are requested, the
buffer can be completed only after the processing is finished, to make
sure it's untouched by the application as long as the processing runs.
At most one raw stream and at most one processed stream are supported
and can be combined. An example of producing both raw and processed
files using `cam' application:
cam -c1 -C100 -Ffile# \
-s role=viewfinder,width=1920,height=1080,pixelformat=RGB888 \
-s role=raw,width=3280,height=2464,pixelformat=SRGGB8
Note the difference in viewfinder and raw stream sizes due to the fact
that debayering requires enlarging the image width, which enforces
selecting a larger sensor resolution in this case.
In order to track whether a raw stream is requested and which one it is,
SimpleCameraData::rawStream_ member variable is introduced.
This is the final step to make raw streams working.
Reviewed-by: Umang Jain <uajain@igalia.com>
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
If software ISP is enabled then metadata is required in the simple
pipeline. But this doesn't apply if the software ISP is not actually
used, for example when only a raw stream is produced. Then the pipeline
waits for metadata that never comes.
This patch fixes the problem by requiring metadata only when software
ISP is used.
Reviewed-by: Umang Jain <uajain@igalia.com>
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
When software ISP is enabled, we want to be able to provide a raw stream
in addition to the processed stream. For this purpose, we need two
streams. If only the processed stream is requested, it doesn't harm to
allocate two.
The number of streams is determined as a camera property in the pipeline
matching. To be able to produce both raw and processed output, two
streams must be provided. The actual number of streams needed (one or
two) is determined only in SimplePipelineHandler::validate().
In theory, software ISP could produce multiple processed streams but
this is out of scope of this patch series. Hence two streams are
sufficient at the moment.
When software ISP is not enabled, the camera won't be able to produce
multiple streams (assuming there's no hardware converter) and only
single stream should be allocated as before. The simple pipeline
handler assumes there's a linear pipeline from the camera sensor to a
video capture device, and only supports a single stream. Branches in
the hardware pipeline that would allow capturing multiple streams from
the same camera sensor are not supported. We have no plan to change
that, as a device that can produce multiple streams will likely be
better supported by a dedicated pipeline handler.
Reviewed-by: Umang Jain <uajain@igalia.com>
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
When a raw stream is requested, either alone or together with a
processed stream, it can be produced without conversion. Let's amend
the corresponding check on the number of configurations, so that the
mere presence of a raw stream doesn't enforce conversion.
Reviewed-by: Umang Jain <uajain@igalia.com>
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
SimpleCameraConfiguration::validate() looks for the best configuration.
As part of enabling raw stream support, the method must consider raw
streams in addition to the processed streams.
Raw streams are adjusted from the capture format and size.
Configuration validation computes the maximum size of all the requested
streams and compares it to the output sizes. When e.g. only a raw
stream is requested then this may result in an invalid adjustment of its
size. This is because the output sizes are computed for processed
streams and may be smaller than capture sizes. If a raw stream with the
capture size is requested, it may then be wrongly adjusted to a larger
size because the output sizes, which are irrelevant for raw streams
anyway, are smaller than the requested capture size. The problem is
resolved by tracking raw and processed streams maximum sizes separately
and comparing raw stream sizes against capture rather than output sizes.
Note that with both processed and raw streams, the requested sizes must
be mutually matching, including resizing due to debayer requirements.
For example, the following `cam' setup is valid for imx219
cam -s role=viewfinder,width=1920,height=1080 \
-s role=raw,width=3280,height=2464
rather than
cam -s role=viewfinder,width=1920,height=1080 \
-s role=raw,width=1920,height=1080
due to the resolution of 1924x1080 actually selected for debayering to
1920x1080. If the resolutions don't match mutually or don't match the
available sizes, validation adjusts them.
Setting up the right configurations is still not enough to make the raw
streams working. Buffer handling must be changed in the simple
pipeline, which is addressed in followup patches.
Co-developed-by: Umang Jain <uajain@igalia.com>
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Umang Jain <uajain@igalia.com>
Signed-off-by: Umang Jain <uajain@igalia.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Let's handle both processed and/or raw output configurations. In
addition to the already handled processed formats and sizes, this patch
adds handling of raw formats and sizes, which correspond to the capture
formats and sizes.
When creating stream configurations, raw or processed formats are
selected according to the requested stream roles.
This is another preparatory patch without making raw outputs working.
Reviewed-by: Umang Jain <uajain@igalia.com>
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
In order to support raw streams, we need to add raw formats to software
ISP configurations. In this preparatory patch, the raw formats are
excluded from output configurations for conversions.
Reviewed-by: Umang Jain <uajain@igalia.com>
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
In order to initialise and deinitialise gpuisp we need to be able to setup
EGL in the same thread as Debayer::process() happens in.
This requires extending the Debayer object to provide start and stop
methods which are triggered through invokeMethod in the same way as
process() is.
Introduce start() and stop() methods to the Debayer class. Trigger those
methods as described above via invokeMethod. The debayer_egl class will
take care of initialising and de-initialising as necessary. Debayer CPU
sees no functional change.
Per feedback from Barnabas the stop method is using blocking
synchronisation and thus we drop ispWorkerThread_.removeMessages().
[bod: Made method blocking not queued per Robert's bugfixes]
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Move the initialisation of Bayer params and CCM to a new constructor in the
Debayer class.
Ensure we call the base class constructor from DebayerCpu's constructor in
the expected constructor order Debayer then DebayerCpu.
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
