| Age | Commit message (Collapse) | Author |
|---|
|
"using namespace" in a header file propagates the namespace to
the files including the header file. So it should be avoided.
This removes "using namespace" in header files in qcam.
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>
|
|
"using namespace" in a header file propagates the namespace to
the files including the header file. So it should be avoided.
This removes "using namespace" in stream_options.h
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>
|
|
"using namespace" in a header file propagates the namespace to
the files including the header file. So it should be avoided.
This removes "using namespace" in agc.hpp.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: David Plowman <david.plowman@raspberrypi.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The camera session keeps requeuing requests until the capture limit is
reached. This causes more request than the limit to complete, as there's
a queue of requests in flight. When capturing from multiple cameras
concurrently, this results in the captureDone signal being emitted for
every request completion after the limit is reached, instead of once per
camera session when reaching the limit.
Fix this by simply dropping any request that completes after the limit
is reached. We could instead avoid requeuing more requests than needed
to reach the limit, but that may cause request starvation in pipelines,
which are currently not handled consistently (or correctly).
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The FileSink class constructs stream names internally the same way that
the CameraSession does, except that it fails to add the camera name.
This results in files being written without the camera name.
This could be fixed in FileSink, but we would still duplicate code to
construct stream names. Pass the stream names map from CameraSession to
FileSink instead, and store it internally.
Fixes: 02001fecb0f5 ("cam: Turn BufferWriter into a FrameSink")
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The streamName_ map contains names for all streams, rename it to
streamNames_ to make this more explicit.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
libunwind has an API to provide symbolic names for functions. It's less
optimal than using backtrace_symbols() or libdw, as it doesn't allow
deferring the symbolic names lookup, but it can be usefull as a fallback
if no other option is available.
A sample backtrace when falling back to libunwind looks like
libcamera::VimcCameraData::init()+0xbd
libcamera::PipelineHandlerVimc::match(libcamera::DeviceEnumerator*)+0x3e0
libcamera::CameraManager::Private::createPipelineHandlers()+0x1a7
libcamera::CameraManager::Private::init()+0x98
libcamera::CameraManager::Private::run()+0x9f
libcamera::Thread::startThread()+0xee
decltype(*(std::__1::forward<libcamera::Thread*>(fp0)).*fp()) std::__1::__invoke<void (libcamera::Thread::*)(), libcamera::Thread*, void>(void (libcamera::Thread::*&&)(), libcamera::Thread*&&)+0x77
void std::__1::__thread_execute<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, void (libcamera::Thread::*)(), libcamera::Thread*, 2ul>(std::__1::tuple<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, void (libcamera::Thread::*)(), libcamera::Thread*>&, std::__1::__tuple_indices<2ul>)+0x3e
void* std::__1::__thread_proxy<std::__1::tuple<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, void (libcamera::Thread::*)(), libcamera::Thread*> >(void*)+0x62
start_thread+0xde
???
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
libunwind is an alternative to glibc's backtrace() to extract a
backtrace. Use it when available to extend backtrace support to more
platforms.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
libdw provides access to debugging information. This allows creating
better stack trace entries, with file names and line numbers, but also
with demangled symbols as the symbol name is available and can be passed
to abi::__cxa_demangle().
With libdw, the backtrace previously generated by backtrace_symbols()
src/cam/../libcamera/libcamera.so(_ZN9libcamera14VimcCameraData4initEv+0xbd) [0x7f7dbb73222d]
src/cam/../libcamera/libcamera.so(_ZN9libcamera19PipelineHandlerVimc5matchEPNS_16DeviceEnumeratorE+0x3e0) [0x7f7dbb731c40]
src/cam/../libcamera/libcamera.so(_ZN9libcamera13CameraManager7Private22createPipelineHandlersEv+0x1a7) [0x7f7dbb5ea027]
src/cam/../libcamera/libcamera.so(_ZN9libcamera13CameraManager7Private4initEv+0x98) [0x7f7dbb5e9dc8]
src/cam/../libcamera/libcamera.so(_ZN9libcamera13CameraManager7Private3runEv+0x9f) [0x7f7dbb5e9c5f]
src/cam/../libcamera/base/libcamera-base.so(_ZN9libcamera6Thread11startThreadEv+0xee) [0x7f7dbb3e95be]
src/cam/../libcamera/base/libcamera-base.so(+0x4f9d7) [0x7f7dbb3ec9d7]
src/cam/../libcamera/base/libcamera-base.so(+0x4f90e) [0x7f7dbb3ec90e]
src/cam/../libcamera/base/libcamera-base.so(+0x4f2c2) [0x7f7dbb3ec2c2]
/lib64/libpthread.so.0(+0x7e8e) [0x7f7dbab65e8e]
/lib64/libc.so.6(clone+0x3f) [0x7f7dbb10b26f]
becomes
libcamera::VimcCameraData::init()+0xbd (src/libcamera/libcamera.so [0x00007f9de605b22d])
libcamera::PipelineHandlerVimc::match(libcamera::DeviceEnumerator*)+0x3e0 (src/libcamera/libcamera.so [0x00007f9de605ac40])
libcamera::CameraManager::Private::createPipelineHandlers()+0x1a7 (src/libcamera/libcamera.so [0x00007f9de5f13027])
libcamera::CameraManager::Private::init()+0x98 (src/libcamera/libcamera.so [0x00007f9de5f12dc8])
libcamera::CameraManager::Private::run()+0x9f (src/libcamera/libcamera.so [0x00007f9de5f12c5f])
libcamera::Thread::startThread()+0xee (src/libcamera/base/libcamera-base.so [0x00007f9de5d125be])
decltype(*(std::__1::forward<libcamera::Thread*>(fp0)).*fp()) std::__1::__invoke<void (libcamera::Thread::*)(), libcamera::Thread*, void>(void (libcamera::Thread::*&&)(), libcamera::Thread*&&)+0x77 (src/libcamera/base/libcamera-base.so [0x00007f9de5d159d7])
void std::__1::__thread_execute<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, void (libcamera::Thread::*)(), libcamera::Thread*, 2ul>(std::__1::tuple<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, void (libcamera::Thread::*)(), libcamera::Thread*>&, std::__1::__tuple_indices<2ul>)+0x3e (src/libcamera/base/libcamera-base.so [0x00007f9de5d1590e])
void* std::__1::__thread_proxy<std::__1::tuple<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, void (libcamera::Thread::*)(), libcamera::Thread*> >(void*)+0x62 (src/libcamera/base/libcamera-base.so [0x00007f9de5d152c2])
start_thread+0xde (/var/tmp/portage/sys-libs/glibc-2.33-r1/work/glibc-2.33/nptl/pthread_create.c:482)
__clone+0x3f (../sysdeps/unix/sysv/linux/x86_64/clone.S:97)
The stack entries related to libcamera are missing source file name and
line information, which will be investigated separately, but this is
still an improvement.
Use libdw when available, falling back to backtrace_symbols() otherwise,
or if libdw fails for any reason.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
Create a new class to abstract generation and access to call stack
backtraces. The current implementation depends on the glibc backtrace()
implementation and is copied from the logger. Future development will
bring support for libunwind, transparently for the users of the class.
The logger backtrace implementation is dropped, replaced by usage of the
new Backtrace class.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The Size class has new helpers that can simplify the code in the IPU3
pipeline handler. Use them.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
Add four new member functions to the Size class (two in-place and two
const) to grow and shrink a Size by given margins.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
The libcamera Android Camera HAL generates camera configurations for the
StillCapture, Raw and ViewFinder stream roles. But there is only a check
if the configuration generation failed, for the StillCapture stream role.
This could lead to a NULL pointer dereference if a pipeline handler fails
to generate a default configuration for one of the other two stream roles.
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The problem is happening because we seem to add a CameraStream
associated buffer (depending on the CameraStream::Type) to the Request,
in CameraDevice::processCaptureRequest().
However, when the camera stops, all the current buffers are marked with
FrameMetadata::FrameCancelled and proceed to completion. But the buffer
associated with the CameraStream (that was previously added to the
request) has now been cleared out with a part of streams_.clear(), even
before the camera stop() has been invoked. Any access to those request
buffers after they have been cleared, will result in a crash.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
Creating a CameraBuffer instance doesn't map memory. Fix the error
message accordingly.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Hirokazu Honda <hiroh@chromium.org>
|
|
When the pipeline handler start() method is supplied with a NULL list
of controls, we send an empty control list to the IPA. When the IPA is
running in isolated mode the control list goes through the data
serializer, for which it must be marked correctly as a list of
"controls::controls", otherwise the IPA process will abort.
The IPA has a similar problem returning a control list in its
configure() method. We must be careful to initialise it properly even
when empty.
Signed-off-by: David Plowman <david.plowman@raspberrypi.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Naushir Patuck <naush@raspberrypi.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The intel-ipu3.h public interface from the kernel does not define how to
parse the statistics for a cell. This had to be identified by a process
of reverse engineering, and later identifying the structures from [0]
leading to our custom definition of struct Ipu3AwbCell.
[0]
https://chromium.googlesource.com/chromiumos/platform/arc-camera/+/refs/heads/master/hal/intel/include/ia_imaging/awb_public.h
To improve the kernel interface, a proposal has been made to the
linux-kernel [1] to incorporate the memory layout for each cell into the
intel-ipu3 header directly.
[1]
https://lore.kernel.org/linux-media/20211005202019.253353-1-jeanmichel.hautbois@ideasonboard.com/
Update our local copy of the intel-ipu3.h to match the proposal and
change the AGC and AWB algorithms to reference that structure directly,
allowing us to remove the deprecated custom Ipu3AwbCell definition.
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 we know how the AWB statistics are formatted, use a simplified
loop in processBrightness() to parse the green values and get the
histogram.
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 pixels output by the camera normally include a black level, because
sensors do not always report a signal level of '0' for black. Pixels at
or below this level should be considered black and to achieve that, we
need to substract an offset to all the pixels. This can be taken into
account by reading the lowest value of a special region on sensors which
is not exposed to light. This provides a substracting factor to be
able to adjust the expected black levels in the resulting images.
For a camera outputting 10-bit pixel values (in the range 0 to 1023) a
typical black level might be 64. It is a fixed value, obtained by
capturing a raw frame with minimum exposure and gain fixed to 1.0 while
covering the sensor (the darker the better). We consider it good enough
as a very first approximation, until we measure it during a tuning
process and include it in a configuration file
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 statistics buffer 'ipu3_uapi_awb_raw_buffer' stores the ImgU
calculation results in a buffer aligned horizontally to a multiple of 4
cells. The AWB loop should take care of it to add the proper offset
between lines and avoid any staircase effect.
It is no longer required to pass the grid configuration context to the
private functions called from process() which simplifies the code flow.
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 gains have a precision u3.13, range [0, 8[ which means that a gain
multiplier value of 1.0 is represented as a multiplication by 8192 in
the ImgU. Correct the gains as this was misunderstood in the first
place.
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 algorithm uses the statistics of a cell only if there is not too
much saturated pixels in it. The grey world algorithm works fine when
there are a limited number of outliers.
Consider a valid zone to be at least 80% of unsaturated cells in it.
This value could very well be configurable, and make the algorithm more
or less tolerant.
While at it, implement it in a configure() call as it will not change
during execution, and cache the cellsPerZone values estimated with
std::round as we are using cmath.
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 loops over the width and height of the image when calculating the
BDS grid parameters are nested, but they're actually independent. Split
them to reduce the complexity.
While at it, split out the constants to documented const expressions
for the grid sizes.
Signed-off-by: Laurent Pinchart <laurent.pinchart@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>
|
|
Until now, the limits used to calculate the grid based on the Bayer Down
Scaler configuration where taken from the kernel documentation [0].
While testing and understanding the format of the ImgU statistics, it
appears that the ones defined in CrOS [1] are the correct ones. Use
those.
[0] https://www.kernel.org/doc/html/latest/userspace-api/media/v4l/pixfmt-meta-intel-ipu3.html?highlight=v4l2_meta_fmt_ipu3_params#intel-ipu3-imgu-uapi-data-types
[1] https://chromium.googlesource.com/chromiumos/platform/arc-camera/+/refs/heads/master/hal/intel/include/ia_imaging/awb_public.h
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 variables mix the terms cell, region and zone. It can confuse the
reader, and make the algorithm more difficult to follow. Rename the
local variables to be consistent with their definitions:
- Cells are defined in Pixels
- Zones are defined in Cells
There is no "region" as such, so replace it with the correct term.
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 pixel component sums for the Accumulator are inconsistent with other
similar structures such as the IPAFrameContext::awb::gains. Group the
red, green, and blue sums together in a struct and store them as
uint64_t to reduce potential architectural differences.
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 IspStatsRegion structure was introduced as an attempt to prepare for
a generic AWB algorithm structure. The structure name by itself is not
explicit and it is too optimistic to try and make a generic one for now.
Its role is to accumulate the pixels in a given zone. Rename it to
accumulator, and remove the uncounted field at the same time. It is
always possible to know how many pixels are not relevant for the
algorithm by calculating total-counted. The uncounted field was only
declared and not used. Amend the documentation accordingly.
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 structure Ipu3AwbCell describes the AWB stats layout on the kernel
side. We will need it to be used by the AGC algorithm to be introduced
later, so let's make it visible from ipa::ipu3::algorithms and not only
for the AWB class.
The IspStatsRegion will be needed by AGC too, so let's move it in the
same namespace 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>
|
|
CameraStream always sets the format of processor output buffer to
MJPEG. This fixes the issue.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
|
|
The cause is that we read out delayed values using a frame's sequence
number (DelayedControls::get). But we fill the values up
(DelayedControls::applyControls) incrementing writeCount by only one
even if the sequence number has jumped by several since last
time. This is exactly what happens when frames are being dropped.
So the fix is to increment writeCount by "as much as the sequence
number has jumped since last time", which means that we just follow
the sequence number directly.
Bug: https://bugs.libcamera.org/show_bug.cgi?id=74
Signed-off-by: David Plowman <david.plowman@raspberrypi.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Naushir Patuck <naush@raspberrypi.com>
Tested-by: Naushir Patuck <naush@raspberrypi.com>
Tested-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
It can be useful, for diagnosis purpose, to know what plane and CRTC the
KMS sink auto-selects. Print the display pipeline configuration at start
time.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
|
|
The key and value of the test pattern mode are originally the index of
v4l2 control and the corresponding test pattern mode control value.
This key and value are useful in the initialization for reporting
available test pattern modes. However, the map of the reversed key and
value is much more useful in applying a requested test pattern mode.
Reverses the key and value of the map as the initialization is one
time but the test pattern mode request will be multiple times.
Signed-off-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Currently, the gst_libcamera_device_new() function assumes that a call to
Camera::generateConfiguration() will always succeed, but that may not be
the case and the return value must to be checked.
Otherwise, this could lead to a NULL pointer dereference if the pipeline
handler fails to generate a config for the VideoRecording stream role.
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
There is a possibility that an out-of-order completion of capture
request happens by calling process_capture_result() directly on error
paths. The framework expects that errors should be notified as soon as
possible, but the request completion order should remain intact.
An existing instance of this is abortRequest(), which sends the capture
results on flushing state, without considering order-of-completion.
Since we have a queue of Camera3RequestDescriptor tracking each
capture request placed by framework to libcamera HAL, we should be only
sending back capture results from a single location, by inspecting
the queue. As per the patch, this now happens in
CameraDevice::sendCaptureResults().
Each descriptor is now equipped with its own status to denote whether
the capture request is complete and ready to be send back to the
framework or needs to be waited upon. This ensures that the order of
completion is respected for the requests.
Since we are fixing out-of-order request completion in abortRequest(),
change the function to read from the Camera3RequestDescriptor directly,
instead of camera3_capture_request_t. The descriptor should have all the
information necessary to set the request buffers' state to error.
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
abortRequest() and notifyError() do not modify any members of
CameraDevice hence, these functions can be const.
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
The descriptors_ map holds Camera3RequestDescriptor(s) which are
per-capture requests placed by the framework to libcamera HAL.
CameraDevice::requestComplete() looks for the descriptor for which the
camera request has been completed and removes it from the map.
Since the requests are placed in form of FIFO and the framework expects
the order of completion to be FIFO as well, this calls for a need of
a queue rather than a std::map.
This patch still keeps the same lifetime of Camera3RequestDescriptor as
before i.e. in the requestComplete(). Previously, a descriptor was
extracted from the map and its lifetime was bound to requestComplete().
The lifetime is kept the same by manually calling .pop_front() on the
queue. In the subsequent commit, this is likely to change with a
centralized location of dropping descriptors from the queue for request
completion.
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
Use Camera3RequestDescriptor as cookie for the Capture Request.
The cookie is used to lookup descriptors map in
CameraDevice::requestComplete(). The map will be transformed to a
queue in subsequent commit.
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
Acquire fences for streams of type Mapped generated by
post-processing are not correctly handled and are currently
ignored by the camera HAL.
Fix this by adding CameraStream::waitFence(), executed before
starting the post-processing in CameraStream::process().
The change applies to all streams generated by post-processing (Mapped
and Internal) but currently acquire fences of Internal streams are
handled by the camera worker. Postpone that to post-processing time by
passing -1 to the Worker for Internal streams.
Also correct the release_fence handling for failed captures, as the
framework requires the release fences to be set to the acquire fence
value if the acquire fence has not been waited on.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Hirokazu Honda <hiroh@chromium.org>
Tested-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The libcamera and libcamera-base libraries are currently unversioned, but
donwstream users expect these to have a proper SONAME version in order to
package them.
A stable release has not yet happened because the project is still under
development and the API/ABI might change. But having a versioned SONAME
would allow distributions to package libcamera, without the need to add
any downstream patch to set the version.
Since the "0.0.0" version is already used in different places, let's also
use that as the library version. The meson build system will use the first
part of the version ("0") as the SONAME version, which is aligned with the
convention used by other projects:
$ ls /lib64/libcamera*so* -1
/lib64/libcamera-base.so
/lib64/libcamera-base.so.0
/lib64/libcamera-base.so.0.0.0
/lib64/libcamera.so
/lib64/libcamera.so.0
/lib64/libcamera.so.0.0.0
$ objdump -p /lib64/libcamera.so.0.0.0 | grep SONAME
SONAME libcamera.so.0
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The Camera3RequestDescriptor containing the capture request is added to
the descriptors_ map after a call to CameraWorker::queueRequest(). This
is a race condition since CameraWorker::queueRequest() queues requests
to libcamera::Camera asynchronously. The requests may thus complete
before they get added to descriptors_, in which case requestComplete()
will fail to lookup the request in the map.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Hirokazu Honda <hiroh@chromium.org>
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
|
|
Fix all name in all instances of the function gst_libcamera_get_camera_mananger
to gst_libcamera_get_camera_manager. Spelling of manager was incorrect.
This patch has no functional changes.
Signed-off-by: Vedant Paranjape <vedantparanjape160201@gmail.com>
Reviewed-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The idMap variable may be used uninitialized in the
ControlSerializer::deserialize<ControlList>() function as reported by
gcc 11:
../../src/libcamera/control_serializer.cpp: In member function ‘T libcamera::ControlSerializer::deserialize(libcamera::ByteStreamBuffer&) [with T = libcamera::ControlList]’:
../../src/libcamera/control_serializer.cpp:609:33: error: ‘idMap’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
609 | ControlList ctrls(*idMap);
|
This is due to a missing default case in a switch/case. Fix it by adding
the default case.
Fixes: 6b1404fc4836 ("libcamera: control_serializer: Fix usage of uninitialized variable")
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
|
|
Generation of thumbnail is not occuring currently because
ANDROID_JPEG_THUMBNAIL_SIZE is not set for request metadata passed
to PostProcessorJpeg::process(). The commit 1264628d3c92("android:
jpeg: Configure thumbnailer based on request metadata") introduced
the mechanism to retrieve the thumbanil size from request metadata,
however it didn't add the counterpart i.e. inserting the size in
the request metadata in request metadata template, at the first place.
The patch fixes this issue by setting ANDROID_JPEG_THUMBNAIL_SIZE in
the request metadata template populated by
CameraCapabilities::requestTemplatePreview(). The value for
ANDROID_JPEG_THUMBNAIL_SIZE is set to be the first non-zero size
reported by static metadata ANDROID_JPEG_AVAILABLE_THUMBNAIL_SIZES.
Fixes: 1264628d3c92("android: jpeg: Configure thumbnailer based on request metadata")
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Hirokazu Honda <hiroh@chromium.org>
|
|
CameraMetadata's constructor take in number of entries and number of
bytes to be allocated for those entries. However, CameraMetadata is
already capable of resizing its container on the fly, in case more
entries are added to it. Hence, the numbers passed in during the
construction acts as hint values for initialization.
Clarify this in CameraCapabilities::requestTemplatePreview() and
remove the \todo, as the arguments and the \todo gives the perspective
that we need to be quite accurate with the numbers of entries / bytes,
which is not the case.
Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Hirokazu Honda <hiroh@chromium.org>
|
|
Two independent instances of the ControlSerializer class are in use at
the IPC boundaries, one in the Proxy class that serializes data from the
pipeline handler to the IPA, and one in the ProxyWorker which serializes
data in the opposite direction.
Each instance operates autonomously, without any centralized point of
control, and each one assigns a numerical handle to each ControlInfoMap
it serializes. This creates a risk of potential collision on the handle
values, as both instances will use the same numerical space and
are not aware of what handles has been already used by the instance "on
the other side".
To fix that, partition the handles numerical space by initializing the
control serializer with a seed according to the role of the component
that creates the serializer and increment the handle number by 2, to
avoid any collision risk.
While this is temporary and rather hacky solution, it solves an issue
with isolated IPA modules without too much complexity added.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The ControlInfo class was originally designed to only transport
the control's minimum and maximum values which represent the control's
valid limits.
Later the default value of the control has been added to the ControlInfo
class, but the control serializer implementation has not been updated
accordingly.
This causes issues in IPA modules making use of ControlInfo::def() as,
when running in isolation, they would receive 0.
Fix that by serializing and deserializing the additional ControlValue
and update the protocol description accordingly.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
When a ControlList is deserialized, the code searches for a valid
ControlInfoMap in the local cache and use its id map to initialize the
list. If no valid ControlInfoMap is found, as it's usually the case
for lists transporting libcamera controls and properties, the globally
defined controls::controls id map is used unconditionally.
This breaks the deserialization of libcamera properties, for which a
wrong idmap is used at construction time.
As the serialization header now transports an id_map_type field, store
the idmap type at serialization time, and re-use it at
deserialization time to identify the correct id map.
Also make the validation stricter by imposing to list of V4L2 controls to
have an associated ControlInfoMap available, as there is no globally
defined idmap for such controls.
To be able to retrieve the idmap associated with a ControlList, add an
accessor function to the ControlList class.
It might be worth in future using a ControlInfoMap to initialize the
deserialized ControlList to implement controls validation against their
limit. As such validation is not implemented at the moment, maintain the
current behaviour and initialize the control list with an idmap.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The IPA::configure() function has an IPAConfigInfo parameters which
contains a map of numerical indexes to ControlInfoMap instances.
This is a leftover of the old IPA protocol, where it was not possible to
specify a rich interface as it is possible today and each entity
ControlInfoMap was indexed by a numerical id and stored in a map.
Now that the IPA interface allows to specify parameters by name, drop the
map and send the sensor's control info map only.
If we'll need more ControlInfoMap to be shared with the IPA, a new parameter
can be added to IPAConfigInfo.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
|
|
The utils.h header #include is separate from the rest of the group for
no reason. Move it to where it should be.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
|
|
Returning a non-managed pointer can cause leaks. Use a unique_ptr<>
instead to avoid possible future issues.
The std::move() for the planes argument to the FrameBuffer constructor
is dropped as it's misleading. FrameBuffer has no constructor that takes
an rvalue reference to planes, so the vector was copied despite the
move. This only clarifies the intent, no functional change is
introduced.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Hirokazu Honda <hiroh@chromium.org>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
|
