| Age | Commit message (Collapse) | Author |
|---|
|
As the device pointers are deleted in the pipelines destructor it's not
a good idea to have them uninitialized.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
ISP output pad should be set to YUYV8_2X8 for non-bayer output format.
Bayer formats are not listed in RkISP1CameraConfiguration::validate(),
only non-bayer are listed, so we can set YUYV8_2X8 directly.
This will need to be changed if we add support for bayer output with
libcamera.
Signed-off-by: Helen Koike <helen.koike@collabora.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The IPAModule::loadIPAModuleInfo() function includes a *data pointer
which is used as a null-pointer comparison in the error path with a
conditional statement of "if (ret || !data)".
The data variable is not initialised, and a single error path evaluates
this as "if (true || uninitialised)".
Whilst this error path does not incorrectly utilise the uninitialised
data, as the ret evaluates to true already, it does leave a statement
which includes an uninitialised variable.
Help the static analysers by initialising the data variable when it is
defined.
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Add an error debug message when disassociating part of a process
execution context using unshare fails.
As this is currently used to isolate a child process which is
immediately terminated silently if unshare fails, add a debug printout
and propagate up the error code to make the failure more visible.
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
Fix an argument name mismatch between method declaration and
definition.
Reported-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
Allow logging to syslog, or any given ostream, or to nowhere. The
logging API is updated to accomodate these new logging destinations.
LogMessage is modified to allow this.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
When a video device is stopped all the buffers there queued are released
and their state is set to BufferCancelled.
Currently, on buffer completion, cancelled buffers are blindly re-queued
to the ImgU input or CIO2 output devices, preventing them to be
re-started succesfully in future capture sessions.
Fix that by inspecting the buffers status and skip re-queueing if
they're reported as cancelled. For the ImgU output buffer this is not
required, as cancelled request should be reported to applications in
order to report them failure of the capture operations.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
With the current IPU3 kernel driver implementation, a linked pipe shall be
used (buffers should be queued on it) in order not to block all other pipes.
Currently all links on the ImgU device are only disabled at match() time,
implying that once an ImgU pipe gets linked, it should be used until the
whole pipeline is not re-matched and links disabled again. This is a severe
limitation for applications that wants to switch between cameras using
different pipes going through a full library tear-down and reload.
Perform link disabling at configure() time as well, so that a camera
configuration operation always unlock the usage of the assigned pipe,
regardless of the previously linked ones.
Unfortunately this requires a camera start/stop sequence to always go
through a configure step, a requirement that is not enforced by the
Camera state machine.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
Add a '-s|--size' option to qcam to allow selecting the stream
resolution using a command line option.
If the sizes are not supported by the camera, they get automatically
adjusted and the user notified via an output message.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
Reserve identifiers for user-defined message types and add an operation
to the Message class to register the type identifiers.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
The SignalMessage type is undocumented. Fix it.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
To avoid depending on unitialised values, set the pixelFormat field of
the StreamConfiguration class to 0 in the constructor.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
The internal buffers between the CIO2 and ImgU are freed by the
CIO2Device::stop() method, which is called first when stopping
streaming. The ImgUDevice::stop() method is then called, and attempts to
report completion for all queued buffers, which we have just freed. The
use-after-free corrupts memory, leading to crashes.
Fix this by moving the vector of internal buffers to the IPU3CameraData
where it belongs, and free the buffers after stopping both devices.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
To avoid printing debugging messages related to stream configuration
adjustement when generating a new configuration, set the pixel format
explicitly instead of relying on the internal validate() call to do so.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
Fix a number of spelling errors and word duplications throughout the comments
within libcamera.
These were picked up with spellintian.
Also one capitalisation of the first word of a \return statement picked
up by checkstyle.py while creating this patch.
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
Add support for importing external buffers in all pipeline handlers.
Use the stream memory type in the pipeline handlers during buffer
allocation to import buffers to or export buffers from the video device.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
Add and use an operation to assign to Buffer representing external
memory locations an index at queueRequest() time. The index is used to
identify the memory buffer to be queued to the video device once the
buffer will be queued in a Request.
In order to minimize relocations in the V4L2 backend, this method
provides a best-effort caching mechanisms that attempts to reuse
BufferMemory previously mapped to the buffer's dmabuf file descriptors,
if any.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
Buffer instances reference memory, which is modelled internally by a
BufferMemory instance. Store a pointer to the BufferMemory in the Buffer
class, and populate it when the buffer is queued to the camera through a
request. This is useful for applications to access the buffer memory in
the buffer or request completion handler.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
In addition to referencing buffer memory by index, add support to
referencing it using dmabuf file descriptors. This will be used to
reference buffer memory allocated outside of libcamera and import it.
The dmabuf file descriptors are stored in an array in the Buffer class,
and a new Stream::createBuffer() overload is added to construct a buffer
from dmabuf file descriptor.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
Define the memory type a Stream uses and allow application to set it
through the associated StreamConfiguration.
A Stream can use either internal or external memory allocation methods,
depending on where the data produced by the stream is actually saved.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
All interactions with the Stream's buffers currently go through the
BufferPool. In order to shorten accessing the buffers array, and eventually
restrict access to the Stream's internal buffer pool, provide operations to
access, create and destroy buffers.
It is still possible to access the pool for pipeline handlers to
populate it by exporting buffers from a video device to Stream's pool.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
The cam and qcam applications, as well as the camera capture test case,
access the Stream::bufferPool in order to know how many requests to
initially queue. As part of an effort to remove access to the buffer
pool from applications, use the buffer count from the stream
configuration instead.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
Access the number of allocated buffer for the streams through the stream
configuration instead of the stream's buffers pool.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
When stopping the stream buffers have been queued, in which case their
completion is never be notified to the user. This can lead to memory
leaks. Fix it by notifying completion of all queued buffers with the
status set to error.
As a result the base PipelineHandler implementation can be simplified,
as all requests complete as the result of stopping the stream. The
stop() method that manually completes all queued requests isn't needed
anymore.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
The Buffer class is a large beast the stores information about the
buffer memory, dynamic metadata related to the frame stored in the
buffer, and buffer reference data (in the index). In order to implement
buffer import we will need to extend this with dmabuf file descriptors,
making usage of the class even more complex.
Refactor the Buffer class by splitting the buffer memory information to
a BufferMemory class, and repurposing the Buffer class to reference a
buffer and to store dynamic metadata. The BufferMemory class becomes a
long term storage, valid and stable from the time buffer memory is
allocated to the time it is freed. The Buffer class, on the other hand,
becomes transient, is created on demand when an application requires a
buffer, is given to a request, and is deleted when the request
completes.
Buffer and BufferMemory don't need to be copied, so their copy
constructor and assignment operators are deleted.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
When starting the stream on a capture video device it is often needed to
queue all the allocated buffers. Add a helper method to do so, and
refactor the existing queueBuffer() method to make it clearer.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
Applications often have to map requests queued to a camera to external
resources. To make this easy, add a 64-bit integer cookie to the Request
class that is set when the request is created and can be retrieved at
any time, especially in the request completion handler. The cookie is
completely transparent for libcamera and is never modified.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
libcamera guarantees that requests complete in sequence. This
requirement is currently pushed down to pipeline handlers. Three out of
four of our pipeline handlers implement that requirement based on the
sole assumption that buffers will always complete in sequeuence, while
the IPU3 pipeline handler implements a more complex logic.
It turns out that the logic can be moved to the base PipelineHandler
class with support from the Request class. Do so to simplify the
pipeline handlers.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
There's no need to check if buffers have been allocated before freeing
them as the BufferPool::destroyBuffers() method is a no-op when no
buffers have been allocated. Document this fact explicitly, and remove
the buffer count check.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
For a historical reason that isn't fully understood, the request
completion handler in the Camera class moves all buffers away from the
request's buffer map to a local variable before emitting the request
completion signal. There's no reason to do so, and it makes it
impossible for requests to access buffers in their destructor. Fix it.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
C++11 does not support std::clamp(), add a custom implementation in
utils.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
processing
An event notifier may be unregistered from its activated signal. This
can cause the notifiers set entry in notifiers_ to be deleted while
processNotifiers() is looping over the notifiers_ map, leading to
problems.
To fix this, add a flag to the EventNotifierPoll class to indicate that
event processing is in progress. If the flag is set, the notifiers_
entry is not deleted during notifier unregistration, but will be deleted
by the event processing loop.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
The second argument to std::array is the size of the array, not of the
elements it contains. Fix this by turning the std::array into a simple
array of const char pointers.
Fixes: 099815b85377ac68 ("libcamera: ipa_module: add isOpenSource")
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
|
|
The return value of a read() call is mistakenly checked for nonzero
rather than less than zero. Fix this.
Fixes: df23ab95f3d7 ("libcamera: process: fix compilation on Chromium OS")
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
|
|
Commit 3d20beca6616 ("libcamera: Add Process and ProcessManager
classes") causes the build to fail in the Chromium OS build environment,
because the return values of some function calls marked with the
__warn_unused_result__ attribute are ignored. Fix this.
Fixes: 3d20beca6616 ("libcamera: Add Process and ProcessManager classes")
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Add the dummy IPA that needs isolation to meson. At the same time, clean
up the IPA meson to facilitate adding more IPAs.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Add a dummy IPA that needs to be isolated.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Make IPAManager isolate an IPA in a Proxy if the IPA's license is not
open source, before returning the IPA to the caller. For now, only use
the default Linux IPA proxy, and only LGPL 2.1+ is considered open
source.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Add a skeletal default linux IPA proxy. It currently lacks the IPA proxy
protocol itself.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Add an IPAProxy class whose implementations will act as a proxy between a
pipeline handler and an isolated IPA interface. Also add an IPAProxyFactory
that will construct the IPAProxy implementations as necessary.
Update Doxygen to ignore the directory where IPAProxy implementations will
reside.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Add a Process class to abstract a process, and a ProcessManager singleton
to monitor and manage the processes.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Add a method to IPAModule to check if the module is open source.
This uses the license field of the member IPAModuleInfo.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Add a field to IPAModuleInfo to contain the license of the module.
This license field will be used to determine whether the IPA module
should be run in an isolated process or not. If the license is open
source, then the IPA module will be allowed to run without process
isolation, if the user enables it. If the license is not open source,
then the IPA module will be run with process isolation.
Update the dummy IPA and IPA test to conform to the new struct layout.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Currently the log file and the log level can only be set via environment
variables, but applications may also want to set the log file and the
log level at run time. Provide an API for this.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The SlotBase implementation stores the receiver object pointer as a void
pointer internally. The pointer is then cast back to an Object pointer
when the receiver object class derives from Object. When the receiver is
an object that inherits from both the Object class and other classes,
the Object data members may not be stored at the beginning of the object
memory. The cast back to an Object pointer is thus incorrect.
Fix this by casting the receiver object pointer to an Object pointer
where the type of the receiver object is known, and pass it along with
the receiver void pointer to the SlotBase class. The SlotBase class
stores both pointers internally, and doesn't need the isObject_ field
anymore as the same information is obtained from checking if the Object
pointer is null.
To avoid confusing the two pointers, use the same naming scheme through
the whole implementation: "obj" points to a receiver object as an
unknown type, and "object" to the receiver object cast to an Object. The
latter is null when the receiver object doesn't inherit from the Object
class.
To further clarify the code, remove direct access to the SlotBase "obj"
and "object" fields as much as possible. They are replaced by two new
methods :
- SlotBase::disconnect() to disconnect a signal from the slot's receiver
object
- SlotBase::match() to test if an object pointer matches the slot
The match() method is a template method with a specialisation for the
Object type, to compare either the obj or the object pointer depending
on the type of the parameter. This is required as the Object destructor
calls the SignalBase::disconnect() method for signal connected to the
object, and passes a pointer to Object to that method, while the actual
object may have a different address due to the issue explained above.
The pointer must thus be compared with the stored Object pointer in that
case, not to the pointer to the receiver object.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
Allow signals to cross thread boundaries by posting them to the
recipient through messages instead of calling the slot directly when the
recipient lives in a different thread.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
Create a new Message class to model a message that can be passed to an
object living in another thread. Only an invalid message type is
currently defined, more messages will be added in the future.
The Thread class is extended with a messages queue, and the Object class
with thread affinity.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
The new Thread class wraps std::thread in order to integrate it with the
Object, Signal and EventDispatcher classes. By default new threads run
an internal event loop, and their run() method can be overloaded to
provide a custom thread loop.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
|
|
Commit b817bcec6b53 ("libcamera: Auto generate version information")
causes the build to fail in the Chromium OS build environment, because
git update-index tries to take a lock (ie. write) in the git repo that
is outside of the build directory.
The solution is to simply skip git update-index if we are building in
the Chromium OS build environment, and this decision is made if the
build directory is not a subdirectory of the source directory.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Commit b817bcec6b53 ("libcamera: Auto generate version information")
generates version information in order to automatically include it
various locations (Sphinx and Doxygen documentation, libcamera::version
variable available at runtime, and version.h available at compile time).
Unfortunately this causes lots of unnecessary rebuilds when modifying
the git tree state, which hinders development.
The problem is caused by the generated version.h being listed as a
dependency for the whole libcamera. This is required as meson (to the
best of my knowledge) doesn't provide a way to explicitly specify the
dependency of a single object file (camera_manager.o in this case, as
camera_manager.cpp is the only consumer of the generated version string)
on the custom target used to generate version.h. The dependency can't be
automatically detected at build time, like dependencies on normal
headers that are generated by parsing the source, because the version.h
header may not exist yet. The build could then fail in a racy way.
This change attempts at solving the issue by generating a version.cpp
instead of a version.h to set the git-based version. This minimises the
number of files that need to be rebuild when then git tree state
changes, while retaining the main purpose of the original automatic
version generation, the ability to access the git-based version string
at runtime. We however lose the ability to access git-based version
information at build time in an application building against libcamera,
but there is no expected use case for this.
The version string is moved from the libcamera namespace to the
CameraManager class in order to avoid including version.h inside
libcamera (in version.cpp and in camera_manager.cpp), which would create
dependencies causing more rebuild steps, as described above.
On the other hand, major, minor and patch level version numbers are
useful at build time. This commit changes the generation of version.h in
order to add three macros named LIBCAMERA_VERSION_MAJOR,
LIBCAMERA_VERSION_MINOR and LIBCAMERA_VERSION_PATCH for this purpose.
version.h is not included by any other libcamera header or source file,
and thus doesn't force a rebuild of the library.
The Sphinx and Doxygen documentation keep their git-based version
information, which is set during the configuration of the build and then
doesn't track git commits. We may want to investigate how to improve
this, but given that git-based version for the documentation has very
few use cases outside of tagging nightly builds, this isn't considered
an issue at the moment.
The documentation install directory now uses the base version string, in
order to avoid increasing the number of documentation directories
needlessly. This shouldn't cause any issue as the API should not change
without a change to the version number.
The version number generation and handling code now also standardises
the version variables to not start with a 'v' prefix in meson, in order
to simplify their handling. The prefix is added when generating the
relevant files.
Note that we go back to specifying the fallback version in the main
meson.build, in the call to the project() function. For the time being I
believe this should be a good compromise to avoid unnecessary
recompilation, and moving the fallback version to a different file for
tarball releases can be built on top of this.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
