Not all C libraries include support for locale objects (locale_t) and
the strto*_l() family of functions. A notable example is uClibc that can
be compiled with a hardcoded "C" locale. Compilation then fails as the
newlocale(), freelocale() and strtod_l() functions are not defined.
Fix the compilation breakage by checking for the availability of the
locale_t type, and fall back to strtod() when the type isn't available.
This may not lead to the correct result if support for locale objects
isn't available and the locale isn't hardcoded to "C", but that is such
a corner case that we will likely never encounter it.
Fixes: e8ae254970 ("libcamera: yaml_parser: Use C locale")
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
The strtod() function is locale-dependent, and thus ill-suited to parse
numbers coming from, for instance, YAML files. The YamlObject class uses
strtod_l() to fix that issue, but that function is not available with
all libc implementations. Correctly handling this problem is becoming
out of scope for the YamlObject class.
As a first step, add a strtod() helper function in the utils namespace
that copies the implementation from YamlObject, and use it in
YamlObject. The core issue will then be fixed in utils::strtod().
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Add an entry to the sensor properties for OnSemi AR0521.
The chip manual list one additional test pattern "PN9 Link integrity
test pattern" not supported by the mainline Linux driver.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The dlopen() & co. functions are provided by libdl on some systems, and
but the C library on others. Starting from version 0.62.0, meson handles
this behind the scenes when using dependency('dl'). Add a todo comment
to remember we should replace the manual implementation with a
dependency() call when updating to meson 0.62.0 or newer.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Move the simple converter implementation to a generic V4L2 M2M class
derived from the converter interface. This latter could be used by
other pipeline implementations and as base class for customized V4L2 M2M
converters.
Signed-off-by: Xavier Roumegue <xavier.roumegue@oss.nxp.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Declare a converter Abstract Base Class intended to provide generic
interfaces to hardware offering size and format conversion services on
streams. This is mainly based on the public interfaces of the current
converter class implementation found in the simple pipeline handler.
The main change is the introduction of loadConfiguration() function
which can be used by the concrete implementation to load hardware
specific runtime parameters defined by the application.
Signed-off-by: Xavier Roumegue <xavier.roumegue@oss.nxp.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The IMX519 driver reports ColorBars and ColorBarsFadeToGray but upon
inspecting the test pattern modes via v4l2-ctl utility, these test
patterns do not comply with MIPI CCS v1.1 (Section 10.1).
Hence, omit from camera-sensor-properties database for now and add a
comment for the same.
Fixes: e3b26b4c4e ("libcamera: camera_sensor: Add IMX519 sensor properties")
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Pass an IPA cookie from the pipeline handler to the IPA and eventually back to
the pipeline handler through the setDelayedControls signal. This cookie is used
to index the RPiController::Metadata object to be used for the frame.
The IPA cookie is then returned from DelayedControls when the frame with the
applied controls has been returned from the sensor, and eventually passed back
to the IPA from the signalIspPrepare signal.
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Reviewed-by: David Plowman <david.plowman@raspberrypi.com>
Tested-by: David Plowman <david.plowman@raspberrypi.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Allow the caller to provide a cookie value to DelayedControls::reset and
DelayedControls::push. This cookie value is returned from DelayedControls::get
for the frame that has the control values applied to it.
The cookie value is useful in tracking when a set of controls has been applied
to a frame. In a subsequent commit, it will be used by the Raspberry Pi IPA to
track the IPA context used when setting the control values.
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Reviewed-by: David Plowman <david.plowman@raspberrypi.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Convert ControlRingBuffer to a templated class to allow arbitrary ring buffer
array types to be defined. Rename ControlRingBuffer to RingBuffer to indicate
this.
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Reviewed-by: David Plowman <david.plowman@raspberrypi.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Switch the Raspberry Pi pipeline handler to use the DelayedControls
implementation in the RPi:: namespace. This will allow us to use Raspberry Pi
specific API changes in future commits.
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Reviewed-by: David Plowman <david.plowman@raspberrypi.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Some UVC cameras have been identified that can provide V4L2 menu
controls without any menu items.
This leads to a segfault where we try to construct a
ControlInfo(Span<>,default) with an empty span.
Convert the v4l2ControlInfo and v4l2MenuControlInfo helper functions to
return std::optional<ControlInfo> to be able to account in the caller if
the control is valid, and only add acceptable controls to the supported
control list.
Menu controls without a list of menu items are no longer added as a
valid control and a warning is logged.
This also fixes a potential crash that would have occured in the
unlikely event that a ctrl.minimum was set to less than 0.
Bug: https://bugs.libcamera.org/show_bug.cgi?id=167
Reported-by: Marian Buschsieweke <marian.buschsieweke@ovgu.de>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Implement support for raw Bayer capture at runtime, from start() to
stop(). Support of raw formats in the camera configuration is split to a
subsequent change to ease review.
In raw mode, the ISP is bypassed. There is no need to provide parameter
buffers, and the ISP will not generate statistics. This requires
multiple changes in the buffer handling:
- The params and stats buffers don't need to be allocated, and the
corresponding video nodes don't need to be started or stopped.
- The IPA module fillParamsBuffer() operation must not be called in
queueRequestDevice(). As a result, the IPA module thus doesn't emit
the paramsBufferReady signal. The main and self path video buffers
must thus be queued directly in queueRequestDevice().
- The tryCompleteRequest() function must not to wait until the params
buffer has been dequeued.
- When the frame buffer has been captured, the IPA module
processStatsBuffer() operation must be called directly to fill request
metadata.
Signed-off-by: Florian Sylvestre <fsylvestre@baylibre.com>
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Tested-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Unlike RkISP1Path::generateConfiguration(), the validate() function
doesn't take the camera sensor resolution into account but only
considers the absolute minimum and maximum sizes supported by the ISP to
validate the stream size. Fix it by using the same logic as when
generating the configuration.
Instead of passing the sensor resolution to the validate() function,
pass the CameraSensor pointer to prepare for subsequent changes that
will require access to more camera sensor data. While at it, update the
generateConfiguration() function similarly for the same reason.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Query the driver for the output formats and sizes that it supports,
instead of hardcoding them. This allows future-proofing for formats that
are supported by some but not all versions of the driver.
As the rkisp1 driver currently does not support VIDIOC_ENUM_FRAMESIZES,
fallback to the hardcoded list of supported formats and framesizes. This
feature will be added to the driver in parallel, though we cannot
guarantee that users will have a new enough kernel for it.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
The StreamRole enum has enumerators such as 'Raw' that are too generic
to be in the global libcamera namespace. Turn it into a scoped enum to
avoid namespace clashes, and update users accordingly.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
libcamera prints stream role values in log messages. To be more
user-friendly, add a specialization of operator<<() to print the role
name as a string instead of a numerical value.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Currently, the Android HAL does not work on rkisp1-based devices because
required FrameDurationLimits metadata is missing from the IPA
implementation.
This change sets FrameDurationLimits for rkisp1 based on the existing
ipu3 implementation, using the sensor's reported range of vertical
blanking intervals with the minimum reported horizontal blanking
interval.
Signed-off-by: Nicholas Roth <nicholas@rothemail.net>
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The RkISP1 pipeline and IPA module allows for the CameraSensorInfo to be
empty, probably to accommodate some sensor used in a test platform that
does not provide the mandatory libcamera requirements.
As the \todo item in the IPA reports, there is a possibility that the
received CameraSensorInfo is empty and it should be checked before
accessing it, but currently such requirement is not enforced in the
code.
This allows to assume all the test platforms in use have now
successfully moved their sensor driver to comply with the minimum
requirements and provide a populated CameraSensorInfo to the IPA.
As the safety check is not enforced, and as we don't want to allow
faulty sensors to send empty CameraSensorInfo to the IPA, remove the
\todo item in the IPA and fail hard in the pipeline handler if the
sensor does not comply with libcamera requirements.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The RkISP1 implementation of IPA::configure() still uses the legacy
interface where sensor controls (and eventually lens controls) are
passed from the pipeline handler to the IPA in a map.
Since the introduction of mojom-based IPA interface definition, it is
possible to define custom data types and use them in the interface
definition between the pipeline handler and the IPA.
Align the RkISP1 IPA::configure() implementation with the one in the
IPU3 IPA module by using a custom data type instead of relying on a map
to pass controls to the IPA.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The CameraSensor class validates that the sensor driver in use supports
the controls required for IPA modules to work correctly.
For in-tree IPA modules, whose pipeline handlers already use
CameraSensor there's no need to validate such controls again.
Remove controls validation from the IPU3 and RkISP1 IPA modules and rely
on CameraSensor doing that at initialization time.
The list of mandatory controls is expanded to add V4L2_CID_ANALOGUE_GAIN
without which IPA modules cannot function.
The new requirement only applies to RAW sensors, platforms like UVC and
Simple are not impacted by this change.
While at it, expand the sensor driver requirements documentation to
include V4L2_ANALOGUE_GAIN in the list of mandatory controls a sensor
driver has to support.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The BufferMask enum provides a way of identifying which stream a frame buffer
belongs to. This enum is defined in the raspberrypi.mojom interface file.
However, the IPA does not need these enum definitions to mmap buffers that it
uses.
Move this enum out of the raspberrypi.mojom interface file and put it into
the RPi namespace visible only to the pipeline handler. This removes the
need to include the auto-generated IPA interface header in the RPi::Stream
definition.
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: David Plowman <david.plowman@raspberrypi.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Report the sensor timestamp in metadata. Use the timestamp from the
first buffer. Accuracy could be improved by using the frame start event
from the CSI-2 receiver, but the kernel driver doesn't support it yet.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Add to the formats map all the supported ISI video capture
stream formats.
This allows to populate the list of stream formats for all the non-RAW
use cases, as the ISI can perform colorspace conversion between YUV and
RGB.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add a pipeline handler for the ISI capture interface found on
several versions of the i.MX8 SoC generation.
The pipeline handler supports capturing multiple streams from the same
camera in YUV, RGB or RAW formats. The number of streams is limited by
the number of ISI pipelines, and is currently hardcoded to 2 as the code
has been tested on the i.MX8MP only. Further development will make this
dynamic to support other SoCs.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
This reverts commit 30d704732b.
It turns out that this commit causes some regressions and is in fact
unnecessary because the related commit "libcamera: v4l2_videodevice:
Guard against releasing unallocated buffers"
(a2bdff6d0b) fixes the problem
completely (if the buffers were never allocated, the video device
avoids trying to free them even if the pipeline handler asks).
The reason for the regressions is that in this new (broken) scheme we
would never call clearBuffers() on all the streams if the internal
buffers were never allocated (i.e. buffersAllocated_ is never
set). This causes the stream's bufferMap_ list to get longer and
longer if there are multiple back-to-back calls to configure, and
dev_->importBuffers() will ultimately to fail.
So either we need to think more carefully about how to stop the
pipeline handler from freeing buffers that it doesn't own, or we just
leave it as the other commit resolves the problem on its own. In the
interim, simply reverting this commit certainly seems like the best
solution.
Fixes: 30d704732b ("pipeline: raspberrypi: Do not unconditionally free buffers on close")
Signed-off-by: David Plowman <david.plowman@raspberrypi.com>
Reviewed-by: Naushir Patuck <naush@raspberrypi.com>
Acked-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The V4L2VideoDevice ensures that all sequence numbers for applications
commence at zero from the libcamera perspective. This should be the
behaviour expected by kernel drivers, but this is not always the case.
This is handled internally to ensure consistency, and a warning is
printed if the device does not start from zero. It was expected that the
Warning would help highlight where kernel drivers should be fixed, but
this has led to several false positive reports of failures where people
have been concerned that this warning is a cause for unrelated issues.
Lower the log level print to 'Info', to reduce the apparent severity of
this warning. Info is likely more appropriate that Debug to continue to
facilitate awareness of kernel drivers that could be improved, while not
appearing to be a fault.
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The tryCompleteRequest() function looks up the RkISP1FrameInfo that all
but one of its callers already look up. Remove the double look up by
passing the info pointer to tryCompleteRequest().
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
The PipelineHandler::generateConfiguration() function allocates a
CameraConfiguration instance and returns it. The ownership of the
instance is transferred to the caller. This is a perfect match for a
std::unique_ptr<>, which the Camera::generateConfiguration() function
already returns. Update PipelineHandler::generateConfiguration() to
match it. This fixes a memory leak in one of the error return paths in
the IPU3 pipeline handler.
While at it, update the Camera::generateConfiguration() function
documentation to drop the sentence that describes the ownership
transfer, as that is implied by usage of std::unique_ptr<>.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
releaseBuffers() unconditionally calls ioctl(REQBUFS, 0) to release device
buffer allocations through the close() and class destructor functions. If
another libcamera process is running concurrently with a different sensor, it
would cause the ioctl to fail in the kernel because the buffer queue is owned
owned by the other process. This in turn would cause libcamera to generate an
error log message.
Fix this by ensuring the releaseBuffers() only calls ioctl(REQBUFS, 0) if there
have been buffers previously allocated by the device. This is done by testing
the presense of the V4L2BufferCache in the object.
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Tested-by: David Plowman <david.plowman@raspberrypi.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Currently, libdl is always required. Sometimes, the C++ runtime includes
a dynamic loader, which is the case for Android. Additionally, Android
11 does not have libdl. This fix allows libcamera to build on Android.
Signed-off-by: Nicholas Roth <nicholas@rothemail.net>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Each declaration of a LogCategory will create a new LogCategory, and
will be stored in an unordered_set Logger::categories_. This means that
when a plugin .so is unloaded and loaded, as happens when destructing
and creating a CamereManager, we'll get duplicate categories.
The Logger::registerCategory docs say "Log categories must have unique
names. If a category with the same name already exists this function
performs no operation.". The code does not comply with this.
We solve the issue with two changes:
Change the unordered_set to a vector for simplicity, as there's no need
for an unordered_set.
Instead of using the LogCategory constructor to create new categories in
_LOG_CATEGORY() macro, use a factory method. The factory method will
return either an existing LogCategory if one exists with the given name,
or a newly created one.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
LogCategory just stores the char * that was given to it in the
constructor, i.e. it refers to memory "outside" LogCategory. If the
LogCategory is defined in a .so that is unloaded, then it leads to the
LogCategory pointing to freed memory, causing a crash.
Fix this by taking a copy of the name by using a std::string instead of
just storing the pointer.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
For the purpose of testing serializing/deserializing Flags in function
parameters, add an enum class TestFlags and Flags<TestFlags> to some
function parameters, both for input and output and Signals.
While at it, update the ipa_interface_test.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
