| Age | Commit message (Collapse) | Author |
|---|
|
V4L2VideDevice::createBuffer() creates the same number of
FrameBuffer::Planes as V4L2 format planes. Therefore, if the v4l2 format
single is single-planar format, the created number of
FrameBuffer::Planes is 1. It should rather create the same number of
FrameBuffer::Planes as the color format planes.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The plane length is the length of the plane size. The buffer length
to be allocated for a plane is the offset and the length of
FrameBuffer::Plane.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
FrameBuffer::Plane has offset info now. This uses the offset
in mapping FrameBuffer in MainWindow.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
IPABuffer is represented by FrameBuffer. FrameBuffer::Plane has
now an offset. This uses the offset variable to map the IPABuffer.
The mapped IPABuffer is represented and managed as MappedFrameBuffer.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
This fixes the way of mapping FrameBuffer in FrameSink by
using offset.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
MappedBuffer::maps()
MappedBuffer::maps() returns std::vector<MappedBuffer::Plane>.
Plane has the address, but the address points the beginning of the
buffer containing the plane.
This makes the Plane point the beginning of the plane. So
MappedBuffer::maps()[i].data() returns the address of i-th plane.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
This adds offset to FrameBuffer::Plane. It enables representing frame
buffers that store planes in the same dmabuf at different offsets, as
for instance required by the V4L2 NV12 pixel format.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Make the build type selectable between debug and plain through the debug
USE flag.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
Add two new use flags, doc and test, to select building of documentation
and tests instead of unconditionally disabling them.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The libcamera build system requires openssl to sign IPA modules, and
gnutls is required at runtime to verify signatures. Add dependencies to
both.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The Gentoo Python3 supported versions are now 3.6 to 3.10. Update the
package accordingly.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
checkstyle.py uses single-quoted strings in most locations already.
There are a few locations where this wouldn't be convenient (when the
string itself contains a single quote, which would then require
escaping), but there are also a few other locations where double quotes
are used when single quotes would work fine. Change those to standardize
on single-quoted strings.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The CaseSensitive option requires clang-format 12 which is not widely
available.
Remove the option, to fix clang-format uses until a solution which
supports clang-format version specific configurations can be merged.
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The dot package is identified as a required package for the Documentation, and
already marked as a required dependency with meson.
If meson can not find 'dot', then the documentation will not be built,
so we can explicitly enable it in our Doxyfile.
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The CLASS_DIAGRAMS option has become obsolete, and we already define the
default value.
Remove it.
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The LIBCAMERA_O_PTR() define adds the ';' at the end of the
templated call to _o().
While this works for the only current user in camera_manager.cpp, even
the statement there adds another semi-colon following it.
The addition of the semi-colon in the define unnecessarily prohibits the
macro from being used in places other than the end of a statement.
Remove it.
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The documentation suggests to use CameraConfiguration::operator[] to
access the StreamConfiguration it contains, but as CameraConfiguration
instances are generated by the Camera class and are returned wrapped in
a unique_ptr<>, the usage of operator[] would require an awkward syntax such
as (*config)[i].
Better to suggest the usage of the CameraConfiguration::at() function
instead to access the StreamConfigurations.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
This adds getter functions of stride, offset and size to CameraBuffer
interface.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
CameraBuffer implementation maps a given buffer_handle_t in
constructor. Mapping is redundant to only know the plane info like
stride and offset. Mapping should be executed later in the first
plane() call.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
buffer_handle_t doesn't provide sufficient info to map a buffer
properly. cros::CameraBufferManager enables handling the buffer on
ChromeOS, but no way is provided for other platforms.
Therefore, we put the assumption that planes are in the same buffer
and they are consecutive. This modifies the way of mapping in
generic_camera_buffer with the assumption.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Pipeline managers sets a default value to StreamConfiguration::size. The
original fixation code was attempting to use it, but as it was truncating
the caps to its first structure it would never actually find a best match.
In this patch, instead of truncating, we weight various matches using the
product of the width and height delta. We also split delta from ranges
apart and prefer fixed size over them as ranges are not reliable.
This patch also removes the related todo, as it seems that libcamera core
won't go further then providing this default value and won't be sorting the
format and size lists.
Signed-off-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Tested-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
It's not allowed in GStreamer to push events while holding the object
lock. This reduce the scope into which we hold the object lock. In
fact we don't need to protect against gst_task_resume() concurrency
when we stop the task as resume only do something if the task is paused.
This fixes a deadlock when running multiple instances of libcamerasrc
and closing one of the streaming window.
Signed-off-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Tested-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
It's not allowed to have multiple instances of CameraManager. This
requirement is not easy for GStreamer were the device monitor and
the camerasrc, or two camerasrc instances don't usually have any
interaction between each other. Fix this by implementing a minimalist
singleton around CameraManager constructor and start()/stop()
operations.
Signed-off-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Tested-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
This deadlock occurs when a buffer is holding the last reference on
the allocator. In gst_libcamera_allocator_release() we must drop the
object lock before dropping the last ref of that object since the
destructor will lock it again causing deadlock.
This was notice while switching camera or resolution in Cheese software.
Signed-off-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Tested-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
ASan needs to be loaded first before gstreamer is loaded. This was not
possible, so verify_asan_link_order was disabled. Better way to tackle
this issue was disabling forks on the gstreamer side.
verify_asan_link_order=0 disables the check on ASan side which checks if
ASan was loaded before any other shared objects. Since, gstreamer spawns
a child helper process while building the registry, we needed to disable
this check. But with gst_registry_fork_set_enabled() it is possible to
disable spawning this child helper process, so this ensures that ASan is
loaded before any other shared object is loaded.
Signed-off-by: Vedant Paranjape <vedantparanjape160201@gmail.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
This patch simplifies memory management, i.e., by replacing bare
pointers with g_autoptr or g_autofree according to use case.
While at it also update time representation of timeout variable with
GST_SECOND.
Signed-off-by: Vedant Paranjape <vedantparanjape160201@gmail.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Fix a minor spelling error.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The BufferWriter class has been replaced by the FileSink.
Adapt the reference, to match the source code.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The CameraManager object should be deleted when it is no longer used to
prevent it from leaking.
When the application closes, the memory will be released, but it would
show up in reports from memory validation tools such as valgrind if not
handled correctly.
Recommend best-practices in the guide and ensure it is automatically
cleaned up when the CameraManager goes out of scope.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The build and run instructions read as if meson is the only way to
compile the application. Although this is recommended, it is not
required - adapt the language to fit better.
While here, fix the naming of the simple_cam meson object to match
the other usages of 'simple-cam'.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
This TODO has no real meaning, and isn't an action. Remove it.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The Gstreamer gstlibcamerasrc element operates better with the
glimagesink, as it can perform direct renders or hardware accellerated
image format conversion.
Further extend the sample commands to demonstrate how to specify a fixed
width and height parameter to the pipeline.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The compiler generated constructor does not initialize the
ControlInfoMap::idmap_ field.
Fix this by explicitly initializing the field in the class
declaration.
Reported-by: Coverity CID=354657
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The IPASessionConfiguration now has the grid configuration stored. Use
it it at prepare() and process() calls in AWB and pass it as a reference
to the private functions when needed.
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The IPASessionConfiguration now has the grid configuration stored. Use
it at process() call in AGC and pass it as a reference to the private
functions when needed.
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The Doxygen LATEX_SOURCE_CODE, RTF_SOURCE_CODE and
DOCBOOK_PROGRAMLISTING options are deprecated, and generate warnings
with the Doxygen's master branch. Drop them.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The MappedBuffer structure is a custom container that binds a data
pointer with a length. This is exactly what Span is. Use it instead.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
Now that the interface is properly used by the AGC class, move it into
ipa::ipu3::algorithms and let the loops do the calls.
As we need to exchange the exposure_ and gain_ by passing them through the
FrameContext, use the calculated values in setControls() function to
ease the reading.
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Now that the interface is properly used by the AWB class, move it into
ipa::ipu3::algorithms and let the loops do the calls.
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
In preparation for using the AGC through the new algorithm interfaces,
convert the existing code to use the new function types.
Now that the process call is rewritten, re-enable the compiler flag to
warn when a function declaration hides virtual functions from a base class
(-Woverloaded-virtual).
We never use converged_ so remove its declaration. The controls may not
need to be updated at each call, but it should be decided on the context
side and not by a specific call by using a lock status in the Agc
structure for instance.
As the params_ local variable is not useful anymore, remove it here
too.
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
When the stats are received, pass them with the context to the existing
AWB algorithm. IPAFrameContext now has a new structure to store the
gains calculated by the AWB algorithm.
When an EventFillParams event is received, call prepare() and set the new
gains accordingly in the params structure.
There is no more a need for the IPU3Awb::initialise() function, as the
params are always set in prepare().
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Introduce a new algorithm to manage the tone mapping handling of the
IPU3.
The initial algorithm is chosen to configure the gamma contrast curve
which moves the implementation out of AWB for simplicity. As it is
initialised with a default gamma value of 1.1, there is no need to use
the default table at initialisation anymore.
This demonstrates the way to use process() call when the EventStatReady
comes in. The function calculates the LUT in the context of a frame, and
when prepare() is called, the parameters are filled with the updated
values.
AGC is modified to take the new process interface into account.
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Implement a new modular framework for algorithms with a common context
structure that is passed to each algorithm through a common API.
This patch:
- removes all the local references from IPAIPU3 and uses IPAContext
- implements the list of pointers and the loop at configure call on each
algorithm
- loops in fillParams on each prepare() call on the algorithm list
- loops in prepareStats on each process() call on the algorithm list
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Introduce three functions in the Algorithm class to manage algorithms:
- configure which is called when IPA is configured only
- prepare called on EventFillParams event at each frame when the request
is queued
- process called on EventStatReady event at each frame completion when
the statistics have been generated.
The existing AGC implementation already has a function named process(),
though it has different arguments. Adding the new virtual process()
interface causes a compiler warning due to the AGC implementation
overloading a virtual function, even though the overload can be resolved
correctly.
Temporarily disable the warning in this commit to maintain bisection
until the AGC is converted to the new interface.
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
An increasing amount of data and information needs to be shared between
the components that build up to implement image processing algorithms.
Create a context structure which will allow us to work towards calling
algorithms in a modular way, and sharing information between the modules.
The IPA context is a global context set at configure time
(IPASessionConfiguration) and a per-frame context (IPAFrameContext) used
while streaming.
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The abstract Algorithm class was originally placed in libipa as an
attempt define a generic algorithm container. This was a little
optimistic and pushed a bit far too early.
Move the Algorithm class into the IPU3 which is the only user of the
class, as we adapt it to support modular algorithm components for the
IPU3.
Not documenting the namespace may cause issues with Doxygen in libipa.
The file libipa.cpp is thus created as an empty file for now, but we
can leverage it in the future to add more global libipa documentation,
and possibly code too.
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
IPCMessage::payload() converts the IPCMessage into an IPCUnixSocket
payload. However, if IPCMessage is constructed with one of the
following constructors -
IPCMessage::IPCMessage(),
IPCMessage::IPCMessage(uint32_t cmd)
IPCMessage::IPCMessage(const Header &header)
The data_ vector of IPCMessage is empty and uninitialised. In that
case, IPCMessage::payload will try to memcpy() an empty data_ vector
which can lead to invoking memcpy() with a nullptr parameter, which
is then identified by the address sanity checker.. Add a non-empty
data_ vector check to avoid it.
The issue is noticed by running a test manually, testing the vimc
IPA code paths in isolated mode. It is only noticed when the test
is compiled with -Db_sanitize=address,undefined meson built-in option.
ipc_pipe.cpp:110:8: runtime error: null pointer passed as argument 2, which is declared to never be null
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
In IPCUnixSocket, a payload can be sent/received with empty fd vector,
which leads to passing a nullptr in memcpy() in both sendData()
and recvData(). Add a null check for fd vector's data pointer
to avoid invoking memcpy() with nullptr.
The issue is noticed by running a test manually testing the vimc
IPA code paths in isolated mode. It is only noticed when the test
is compiled with -Db_sanitize=address,undefined meson built-in option.
ipc_unixsocket.cpp:268:8: runtime error: null pointer passed as argument 2, which is declared to never be null
ipc_unixsocket.cpp:312:8: runtime error: null pointer passed as argument 1, which is declared to never be null
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
|
|
Regarding (de)serialization in isolated IPA calls, we have four layers:
- struct
- byte vector + fd vector
- IPCMessage
- IPC payload
The proxy handles the upper three layers (with help from the
IPADataSerializer), and passes an IPCMessage to the IPC mechanism
(implemented as an IPCPipe), which sends an IPC payload to its worker
counterpart.
When a FileDescriptor is involved, previously it was only a
FileDescriptor in the first layer; in the lower three it was an int. To
reduce the risk of potential fd leaks in the future, keep the
FileDescriptor as-is throughout the upper three layers. Only the IPC
mechanism will deal with ints, if it so wishes, when it does the actual
IPC. IPCPipeUnixSocket does deal with ints for sending fds, so the
conversion between IPCMessage and IPCUnixSocket::Payload converts
between FileDescriptor and int.
Additionally, change the data portion of the serialized form of
FileDescriptor to a 32-bit unsigned integer, for alightnment purposes
and in preparation for conversion to an index into the fd array.
Also update the deserializer of FrameBuffer::Plane accordingly.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Tested-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
|
|
Development and usage of isolated IPA's has identified bugs in the
transmission of the file descriptors between the pipeline handler and
the IPA through.
Add tests to identify these bugs and prevent regressions.
This tests basically checks for two things:
- Camera reconfigurations without stopping CameraManager
- Fd leaks across IPA IPC boundary [1]
Currently, it uses vimc, but can be easily changed to using another
platform (for e.g. IPU3) by changing kCamId_ and kIpaProxyName_.
The test performs kNumOfReconfigures_ (currently set to 10)
reconfigurations of the camera. Each reconfiguration runs
start(), capture(100ms), stop() of the camera. Hence, the test
runs approximately for 1 second.
For checking the fd leaks, the test monitors the /proc/$PROXY_PID/fd
directory for open fds. It compares the number of open fds after each
run to the number of open fds before the first run. If those two are
found to be mis-matched, the test shall report failure.
The test validates IPA IPC code paths which are used when the IPA is run
in isolated mode. The CameraTest is constructed with the isolate flag
set to enforce this.
[1] https://bugs.libcamera.org/show_bug.cgi?id=63
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
