The libcamera internal headers are located in src/libcamera/include/.
The directory is added to the compiler headers search path with a meson
include_directories() directive, and internal headers are included with
(e.g. for the internal semaphore.h header)
#include "semaphore.h"
All was well, until libcxx decided to implement the C++20
synchronization library. The __threading_support header gained a
#include <semaphore.h>
to include the pthread's semaphore support. As include_directories()
adds src/libcamera/include/ to the compiler search path with -I, the
internal semaphore.h is included instead of the pthread version.
Needless to say, the compiler isn't happy.
Three options have been considered to fix this issue:
- Use -iquote instead of -I. The -iquote option instructs gcc to only
consider the header search path for headers included with the ""
version. Meson unfortunately doesn't support this option.
- Rename the internal semaphore.h header. This was deemed to be the
beginning of a long whack-a-mole game, where namespace clashes with
system libraries would appear over time (possibly dependent on
particular system configurations) and would need to be constantly
fixed.
- Move the internal headers to another directory to create a unique
namespace through path components. This causes lots of churn in all
the existing source files through the all project.
The first option would be best, but isn't available to us due to missing
support in meson. Even if -iquote support was added, we would need to
fix the problem before a new version of meson containing the required
support would be released.
The third option is thus the only practical solution available. Bite the
bullet, and do it, moving headers to include/libcamera/internal/.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Jacopo Mondi <jacopo@jmondi.org>
Add a MediaDevice::version() function to expose the media device API
version number. This can be useful to support driver API changes in
pipeline handlers.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Extend MediaDevice to inherit from the Loggable interface to support a
logPrefix which presents the device node path, and the driver name.
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The conversion of pointers to integers is implementation defined and
differs between g++ and clang++ when utilising a uint64_t type.
#include <iostream>
int main(int argc, char **argv)
{
void *ptr = reinterpret_cast<void *>(0xf1234567);
uint64_t u64 = reinterpret_cast<uint64_t>(ptr);
uint64_t uint64 = reinterpret_cast<uintptr_t>(ptr);
std::cout << "ptr " << ptr
<< " ptr -> u64 0x" << std::hex << u64
<< " ptr -> uintptr_t -> u64 0x" << std::hex << uint64
<< std::endl;
return 0;
}
When compiled with g++ for a 32-bit platform produces the following
unexpected output:
ptr 0xf1234567 ptr -> u64 0xfffffffff1234567 ptr -> uintptr_t -> u64 0xf1234567
The standards states:
"A pointer can be explicitly converted to any integral type large enough
to hold all values of its type. The mapping function is
implementation-defined. [Note: It is intended to be unsurprising to
those who know the addressing structure of the underlying machine. — end
note]"
And as such the g++ implementation appears to be little more surprising
than expected in this situation.
The MediaDevice passes pointers to the kernel via the struct
media_v2_topology in which pointers are cast using a uint64 type (__u64),
which is affected by the sign extension described above when BIT(32) is
set and causes an invalid address to be given to the kernel.
Ensure that we cast using uintptr_t which is not affected by the sign
extension issue.
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Remove spaces between empty curly brackets (replacing { } with {}) to
comply with the coding style. Fix one other coding style violation on
the lines touched by those fixes.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Now that our usage of C compatibility header is documented, use them
consistently through the source code.
While at it, group the C and C++ include statements as defined in the
coding style, and fix a handful of #include ordering issues.
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>
Add lock() and unlock() which are backed by lockf() and allow an
instance of libcamera to claim exclusive access to a media device.
These functions are the base of allowing multiple user of libcamera to
coexist in the system without stepping on each other's toes.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
All external callers to open() and close() have been refactored and
there is no need to expose these functions anymore, make them private.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To gain better control of when a file descriptor is open to the media
device and reduce the work needed in pipeline handler implementations,
handle the file descriptor in acquire() and release().
This changes the current behavior where a file descriptor is only open
when requested by the pipeline handler to that one is always open as
long a media device is acquired. This new behavior is desired to allow
implementing exclusive access to a pipeline handler between processes.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
There is no reason to reread the MEDIA_IOC_DEVICE_INFO information every
time the media device is opened. Move it populate() where it will be
read once together the other information about the media device.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Remove the need for the caller to open and close the media device when
populating the MediaDevice. This is done as an effort to make the usage
of the MediaDevice less error prone and the interface stricter.
The rework also revealed and fixes a potential memory leak in
MediaDevice::populate() where resources would not be deleted if the
second MEDIA_IOC_G_TOPOLOGY would fail.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Include the header file corresponding to the source file in the very
first position. This complies with the Google C++ coding style
guideliens, and helps ensuring that the headers are self-contained.
Three bugs are already caught by this change (missing includes or
forward declarations) in device_enumerator.h, event_dispatcher_poll.h
and pipeline_handler.h. Fix them.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
The documentation style for the Doxygen comment blocks is inconsistent
in the library. Document the expectations and update all existing
comment blocks to match.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Jacopo Mondi <jacopo@jmondi.org>
The local variable ifaceId generates a compiler warning that it is used
without initialisation.
The loop, and check against the loop iterator, and the
topology.num_interfaces actually ensures that this isn't the case, and
that ifaceId can only be used when set to a valid ID.
The media_v2_interface.id field documents the ID as simply "Unique ID
for the interface" but does not specify if zero is a valid ID or not. As
such we don't further check the pre-initialised ID for validity and keep
the existing check on the iterator count to determine if the ID had been
discovered successfully.
Fixes: d4af90d729 ("libcamera: media_device: Create entities with
major and minor numbers")
Acked-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Rapid growth of the library have resulted in slightly different wording
to document that a function returns 0 on success or a negative error
code otherwise. Align all different variations.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Prior to kernel v4.19, the MEDIA_IOC_G_TOPOLOGY ioctl didn't expose
entity flags. Fallback to calling MEDIA_IOC_ENUM_ENTITIES for each
entity to retrieve the flags in that case.
Fixes: 67d313240c ("libcamera: pipeline: uvcvideo: create a V4L2Device for the default video entity")
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The signal is emitted when the hardware device corresponding to the
media device is unplugged. This will trigger the full unplug handling
chain.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Pipeline handlers such as the upcoming ucvideo handler need the model
name to properly name the Camera objects they create. Store the model
name when querying the media device info and handle it in a similar
fashion as the driver name which is already retrieved.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add a function to the MediaLink class to set the state of a link to
enabled or disabled. The function makes use of an internal MediaDevice
method, which is defined private and only accessible by the MediaLink
setEnabled() function itself.
Also add to MediaDevice a function to reset all links registered in the
media graph.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Add three overloaded functions 'link()' to retrieve a link between two
pads. Each overloaded implementation exposes a different method to
identify the source and sink pads.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Add a MediaDevice member field to the MediaObject class hierarcy.
Each media object now has a reference to the media device it belongs to,
and which it has been created by.
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Add back the 'struct' keyword for structure types.
C++ allows omitting the 'struct' keywork. Add it back to make clear
we're dealing with structures and not class types.
While at there re-sort fields of media_v2_* type in populate() function
to reflect the declaration order in the kernel headers.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Do not compare an usigned int with -1 to avoid going through cast.
Also align function parameters and long assignement lines while at there.
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
This silences valgrind that otherwise warns about usage of uninitialized
values. While not strictly required as the kernel should fill the whole
arrays in MEDIA_IOC_G_TOPOLOGY, the extra cost, in a non-critical path,
is negligible compared to the ability to run without valgrind warnings.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Mention in the valid() method documentation, that a media graph has to
be populated to be valid.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Extend the MediaEntity object with device node major and minor numbers,
and retrieve them from the media graph using interfaces. They will be
used by the DeviceEnumerator to retrieve the devnode path.
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Most errors recorded during graph parsing are logged but not propagated
to the caller. Fix this and delete objects that are created but not
successfully added to the graph to avoid memory leaks. As the error code
returned from the addObject() and populate*() functions doesn't matter
much, turn them into bool functions.
Additionally, add a way to query whether the media graph was valid, and
clear objects before populating the graph to avoid leaking them.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The MediaDevice object implements handling and configuration of the media
graph associated with a media device.
The class allows enumeration of all pads, links and entities registered in
the media graph.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
