Header comment blocks used to contain the file name. Considered as
useless information, the names have been removed, with the last ones
supposed to be dropped in commit d3bf27180e ("libcamera: Drop
remaining file names from header comment blocks"). A few have however
been forgotten, and more crept back since. Remove them.
While at it, fix one typo in a header comment block by replacing
'MaliC55 with Mali-C55', and add a missing blank line in
src/ipa/rpi/pisp/pisp.cpp.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
In Qt 6.9.0 the qtmochelpers.h header introduced a construct that gcc 12
and gcc 13 incorrectly flag as variable shadowing:
In file included from src/apps/qcam/qcam.p/moc_viewfinder_gl.cpp:12:
/usr/include/qt6/QtCore/qtmochelpers.h: In instantiation of ‘constexpr void QtMocHelpers::detail::UintDataStorage<std::integer_sequence<int, Idx ...>, T ...>::forEach(F&&) const [with F = QtMocHelpers::UintData<QtMocHelpers::SignalData<void(libcamera::FrameBuffer*)> >::copyTo<{anonymous}::qt_meta_tag_ZN12ViewFinderGLE_t, QtMocHelpers::MetaObjectContents<24, 10, 60, 3> >(QtMocHelpers::MetaObjectContents<24, 10, 60, 3>&, size_t, uint&) const::<lambda(const auto:39&)>; int ...Idx = {0}; T = {QtMocHelpers::SignalData<void(libcamera::FrameBuffer*)>}]’:
/usr/include/qt6/QtCore/qtmochelpers.h:255:21: required from ‘constexpr const QtMocHelpers::MetaObjectContents<24, 10, 60, 3> ViewFinderGL::qt_staticMetaObjectContent<{anonymous}::qt_meta_tag_ZN12ViewFinderGLE_t>’
src/apps/qcam/qcam.p/../../../../../../src/apps/qcam/viewfinder_gl.h:32:2: required from ‘constexpr const auto ViewFinderGL::qt_staticMetaObjectStaticContent<{anonymous}::qt_meta_tag_ZN12ViewFinderGLE_t>’
src/apps/qcam/qcam.p/moc_viewfinder_gl.cpp:63:5: required from here
src/apps/qcam/qcam.p/../../../../../../src/apps/qcam/viewfinder_gl.h:32:2: in ‘constexpr’ expansion of ‘ViewFinderGL::qt_create_metaobjectdata<{anonymous}::qt_meta_tag_ZN12ViewFinderGLE_t>()’
src/apps/qcam/qcam.p/moc_viewfinder_gl.cpp:58:87: in ‘constexpr’ expansion of ‘QtMocHelpers::metaObjectData<ViewFinderGL, {anonymous}::qt_meta_tag_ZN12ViewFinderGLE_t, StringRefStorage<char [13], char [15], char [1], char [24], char [7]>, UintData<SignalData<void(libcamera::FrameBuffer*)> >, UintData<>, UintData<> >(0, qt_stringData, qt_methods, qt_properties, qt_enums, QtMocHelpers::UintData<>(), const QtMocHelpers::detail::UintDataBlock<0, 0>{uint [1](), uint [1]()})’
/usr/include/qt6/QtCore/qtmochelpers.h:563:36: in ‘constexpr’ expansion of ‘(& methods)->QtMocHelpers::UintData<QtMocHelpers::SignalData<void(libcamera::FrameBuffer*)> >::copyTo<{anonymous}::qt_meta_tag_ZN12ViewFinderGLE_t, QtMocHelpers::MetaObjectContents<24, 10, 60, 3> >(result, ((size_t)dataoffset), metatypeoffset)’
/usr/include/qt6/QtCore/qtmochelpers.h:201:57: error: declaration of ‘entry’ shadows a member of ‘QtMocHelpers::detail::UintDataStorage<std::integer_sequence<int, 0>, QtMocHelpers::SignalData<void(libcamera::FrameBuffer*)> >’ [-Werror=shadow]
201 | [[maybe_unused]] auto invoke = [&f](const auto &entry) { f(entry.entry); return 0; };
| ~~~~~~~~~~~~^~~~~
/usr/include/qt6/QtCore/qtmochelpers.h:180:7: note: shadowed declaration is here
180 | T entry;
| ^~~~~
There is little we can do but silence the warning. Do so selectively
based on the Qt and gcc versions, to still detect variable shadowing
with compilers unaffected by the issue.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Umang Jain <uajain@igalia.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Device::openCard() in the cam DRM helpers looks for a /dev/dri/card*
device that can be opened and that doesn't fail when asked about
DRM_CAP_DUMB_BUFFER capability (regardless whether the capability is
supported by the device).
There can be matching devices that are not display devices. This can
lead to selection of such a device and inability to use KMS output with
the `cam' application. The ultimate goal is to display something on the
device and later the KMS sink will fail if there is no connector
attached to the device (although it can actually fail earlier, when
trying to set DRM_CLIENT_CAP_ATOMIC capability if this is not
supported).
Let's avoid selecting devices without connectors, CRTCs or encoders.
The added check makes the original check for DRM_CAP_DUMB_BUFFER API
most likely unnecessary, let's remove it.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Mattijs Korpershoek <mkorpershoek@kernel.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The current `ControlValue` mechanism does not support arrays
of strings, the assignment in the removed snippet will in fact
trigger an assertion failure in `ControlValue::set()` because
`sizeof(std::string) != ControlValueSize[ControlTypeString]`.
Fixes: b35f04b3c1 ("cam: capture_script: Support parsing array controls")
Signed-off-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
With the newly introduced `SDLTexture1Plane` it is easy to handle
any single-plane format that has an SDL equivalent. So use it for
more YUV and RGB formats.
The mapping of RGB formats is not entirely straightforward because
`SDL_PIXELFORMAT_ZZZ...888...` defines a format where the order of
the components is endian dependent, while libcamera's `ZZZ...888...`
formats are derived from the matching DRM formats, and the RGB formats
in question are defined to be little-endian there. So the
endian-independent `SDL_PIXELFORMAT_{ZZZ24,ZZZZ32}` are used.
Signed-off-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
`SDLTextureYUYV` uses `SDL_PIXELFORMAT_YUY2`, which is a single plane
format. To support other single plane formats, replace `SDLTextureYUYV`
with `SDLTexture1Plane` that can be instantiated with an arbitrary SDL
pixel format and that uses `SDL_UpdateTexture()` to update the texture
using exactly a single plane.
Signed-off-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
In commit ee2b011b65 ("apps: cam: Try raw role if default viewfinder
role fails"), the viewfinder role is specified as the default if no role
is yet chosen.
This was unfortunately added by directly accessing the vector rather
than extending the size when the vector is empty. Fix the code to push
the default viewfinder role on to the back of the vector, increasing the
size appropriately.
Fixes: ee2b011b65 ("apps: cam: Try raw role if default viewfinder role fails")
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Tested-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
cam currently defaults to the viewfinder role when no role is specified.
This means that on platforms that only support the raw role (such as a
raw sensor with no softISP on a simple pipeline platform),
generateConfiguration() would return nullptr and cam would bail out.
At least this is what is supposed to happen based on the little
documentation that we have written regarding generateConfiguration(),
specifically in the application writer's guide, which is probably the
most influential piece of documentation:
The ``Camera::generateConfiguration()`` function accepts a list of
desired roles and generates a ``CameraConfiguration`` with the best
stream parameters configuration for each of the requested roles. If the
camera can handle the requested roles, it returns an initialized
``CameraConfiguration`` and a null pointer if it can't.
Currently the simple pipeline handler will return a raw configuration
anyway (if it only supports raw) even if a non-raw role was requested.
Thus cam receives a raw configuration instead of a nullptr when no role
is specified and viewfinder is requested.
However, in the near future, support for raw streams with softISP on the
simple pipeline handler will be merged. This will notably change the
behavior of the simple pipeline handler to return nullptr if a non-raw
role was requested on a platform that only supports raw. This is proper
behavior according to documentation, but changes cam's behavior as it
used to capture fine with no parameters but will no longer be able to.
Technically this is an issue with the roles API, as we are mixing
roles in the sense of "configuration hints" (eg. viewfinder vs recording
vs still capture) with roles in the sense of "platform capabilities"
(raw vs everything else). In the long term the proper solution is to
rework the roles API.
In the meantime, fix cam so that it will try the raw role if the default
viewfinder role returns no configuration. cam is an app that is capable
of using the raw stream, so this is appropriate behavior. If roles are
specified, then do not retry, as in this situation the user knows what
streams they can use and what they want.
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
`std::vector<bool>` is a specialization that implements a dynamic
bit vector, therefore it is not suitable to provide storage for
an array of `bool`. Hence a statically sized array is used when
parsing an array of boolean values.
Instead, use the array overload of `std::make_unique` since the
size is known beforehand.
Signed-off-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Rename the `SingleStream` test to `SimpleCapture`, and extend it
to support using multiple roles. And instantiate another test suite
from the `SimpleCapture` test that tests multiple streams in one
capture session.
Signed-off-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Prepare to add a test suite for capture operations with multiple
streams.
Modify the Capture helper class to support multiple roles and streams
in the configure() and capture() operations. The buffer count
of each stream is asserted to be the same.
Multi-stream support will be added in next patches.
Signed-off-by: Barnabás Pőcze <barnabas.pocze@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Several .cpp files in the cam application don't include their
corresponding header first, as usually done by libcamera to ensure that
headers are self-contained. Reorder headers to fix it. This shows
through a compilation error that file_sink.h is missing
libcamera/controls.h, fix it as well.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Stefan Klug <stefan.klug@ideasonboard.com>
The above two classes have very similar implementations, in fact, the
only essential difference is how many requests are queued. `CaptureBalanced`
queues a predetermined number of requests, while `CaptureUnbalanced`
queues requests without limit.
This can be addressed by introducing a "capture" and a "queue" limit
into the `Capture` class, which determine at most how many requests
can be queued, and how many request completions are expected before
stopping.
Signed-off-by: Barnabás Pőcze <pobrn@protonmail.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Just like other gtest macros, `GTEST_SKIP()` returns an object
to which a formatted message can be added using the usual `<<`
stream operator. So use it instead of printing to `std::cout`.
Signed-off-by: Barnabás Pőcze <pobrn@protonmail.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
Instead of calling `event_base_once()` every time a deferred call
is added to the loop, create an event source at construction, and
simply trigger that when a new deferred call is scheduled.
Signed-off-by: Barnabás Pőcze <pobrn@protonmail.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Deque has fast pop_front and push_back operations while making
fewer allocations for the same number of elements as an `std::list`.
So use an `std::deque` for storing the deferred calls of the loop.
Signed-off-by: Barnabás Pőcze <pobrn@protonmail.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
The compiler generated functions are not appropriate, so
delete the copy/move constructor/assignment to avoid
potential issues.
Signed-off-by: Barnabás Pőcze <pobrn@protonmail.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Using a const lvalue reference to `std::function<>` is not ideal
because it forces a copy to happen. Use an rvalue reference and
`std::move()` to avoid that.
Signed-off-by: Barnabás Pőcze <pobrn@protonmail.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
EINVAL should be returned only when the requested format is unsupported.
On errors when writing the data, let's return EIO, which is the closest
description of the situation when we don't inspect it more.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Now that there is support for retrieving the allowed directions of a
control, print this information when listing controls.
Sample output:
$ cam --list-controls -c 2
Using camera Virtual0 as cam0
Control: [inout] draft::FaceDetectMode:
- FaceDetectModeOff (0)
Control: [inout] libcamera::FrameDurationLimits: [16666..33333]
Size: 2
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>
Now that the vendor of the control can be queried, print it in
--list-controls.
Example output:
$ cam -c 1 --list-controls
Using camera platform/vimc.0 Sensor B as cam0
Control: libcamera::Brightness: [-1.000000..1.000000]
Control: libcamera::Contrast: [0.000000..2.000000]
Control: libcamera::Saturation: [0.000000..2.000000]
Signed-off-by: Paul Elder <paul.elder@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>
When DNG support is missing, the cam application ignores the .dng suffix
of the file pattern and writes raw binary data instead, without
notifying the user. This leads to confusion. Fix it by printing an error
message.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Support for DNG capture is conditioned by the availability of libtiff,
which is indicated by the HAVE_TIFF macro set by meson. The dng_writer.h
header then defines HAVE_DNG, which is used in a couple of places to
conditionally compile DNG-related code. Most of the other locations
where conditional compilation is required use HAVE_TIFF.
Using both HAVE_TIFF and HAVE_DNG is confusing. HAVE_DNG would be a
better name, but as the macro is defined in dng_writer.h, it would
require all files that need to test for DNG support to include that
header. Failure to include it (directly or indirectly) would result in
the code covered by the macro to be silently disabled.
To avoid the confusion, standardize on using HAVE_TIFF everywhere and
drop HAVE_DNG.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Now that controls can be queried for array information, print it in
--list-controls when applicable.
Example output (with dummy controls added to vimc):
$ cam -c 1 --list-controls
Using camera platform/vimc.0 Sensor B as cam0
Control: ColourGains: [1.000000..4.000000]
Size: 2
Control: Brightness: [-1.000000..1.000000]
Control: AfWindows: [(0, 0)/1x1..(0, 0)/100x100]
Size: n
Control: Contrast: [0.000000..2.000000]
Control: Saturation: [0.000000..2.000000]
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>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Now that enum names can be obtained from ControlId, use that information
to print out the list of supported enum values in --list-controls.
Example output (with a dummy AwbMode ControlInfo added to vimc):
$ cam -c 1 --list-controls
Using camera platform/vimc.0 Sensor B as cam0
Control: AwbMode:
- AwbTungsten (2)
- AwbFluorescent (3)
- AwbDaylight (5)
Control: Brightness: [-1.000000..1.000000]
Control: Contrast: [0.000000..2.000000]
Control: Saturation: [0.000000..2.000000]
Signed-off-by: Paul Elder <paul.elder@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Mimic the letterbox behaviour of the Qt viewfinder by rendering the
image centered. This is done by adding a projection matrix to the vertex
shader to scale the rendered rectangle.
Another option would have been to keep using glViewport() (which would
have needed to be moved to paintGL(), as Qt resets the viewport to span
the full widget before calling). Hidpi displays would then need special
handling of the device pixel ratio, which is done automatically by Qt
when it sets the default viewport. Using a projection matrix avoids this
complication.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Tested-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
There's no need to call glClearColor() twice before drawing any GL
content. Drop the first call. This doesn't introduce any functional
change.
While at it, pass floats instead of doubles to glClearColor(), as
required by the function.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Tested-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Starting in Qt 6.7.0, vertex buffers and shader programs are unbound
just before calling QOpenGLWidget::paintGL(). This breaks rendering in
the GL viewfinder in two ways.
First, we bind the vertex buffer only once at initialization time. There
is therefore no vertex buffer mapped at rendering time, preventing both
the vertex shader from having access to the vertex and texture
coordinates.
Then, we bind the shader program only when rendering the first frame.
There is thus no shader program bound for all subsequent frames,
breaking rendering.
Fix this by binding the vertex buffer where needed, when setting
attribute buffers for the shader program, and binding the shader program
for every frame.
As we use a single vertex buffer, we could bind it at the beginning of
paintGL() and keep it bound indefinitely. That would however fail to
clearly indicate in the source code where the vertex buffer is needed,
making the code more difficult to understand as it would rely on
implicit assumptions. Release the vertex buffer explicitly when we don't
need it anymore to avoid this.
While at it, fix a coding style violation by adding missing curly
brackets.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Tested-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
When the widget's aspect ratio doesn't match the camera aspect ratio,
the viewfinder is rendered letter-boxed. The side rectangles are not
painted by the viewfinder, and Qt thus renders the parent widget
background to fill that space.
To make it black, we have two options:
- The simplest option is to set the widget's autoFillBackground property
to true. This causes Qt to paint the whole widget with its background
colour before calling paintEvent(). As the camera image typically
covers most (if not all) of the viewfinder widget, this is less
efficient.
- The more complicated option is to paint the letterbox rectangles
manually. We can additionally set the widget's WA_OpaquePaintEvent
attribute to instruct Qt to skip painting the parent widget. This
reduces CPU usage by about 1% (and may reduce GPU usage as well).
Note that the WA_OpaquePaintEvent attribute has to be disabled when we
render the stopped icon, as the icon has a transparent background.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
