[libcamera-devel] [PATCH 2/2] libcamera: pipeline: simple: Initialize V4L2DeviceFormat before use

Laurent Pinchart laurent.pinchart at ideasonboard.com
Wed Oct 21 20:37:28 CEST 2020


Hi Jacopo,

On Wed, Oct 21, 2020 at 06:40:44PM +0200, Jacopo Mondi wrote:
> On Wed, Oct 21, 2020 at 07:30:53PM +0300, Laurent Pinchart wrote:
> > On Wed, Oct 21, 2020 at 03:13:55PM +0200, Jacopo Mondi wrote:
> > > On Wed, Oct 21, 2020 at 05:47:44AM +0300, Laurent Pinchart wrote:
> > > > The V4L2DeviceFormat has no default constructor. Zero it before use when
> > >
> > > Isn't it implicitly generated by the compiler ?
> >
> > It is, but the implicitly-defined constructor won't help us here:
> >
> > "If the implicitly-declared default constructor is not defined as
> > deleted, it is defined (that is, a function body is generated and
> > compiled) by the compiler if odr-used, and it has the same effect as a
> > user-defined constructor with empty body and empty initializer list."
> >
> > (https://en.cppreference.com/w/cpp/language/default_constructor is an
> > interesting read)
> 
> Ah, thanks for the pointer!
> 
> > > > setting formats. Failure to do so leaves the planes uninitialized,
> > > > potentially causing memory corruption.
> > >
> > > If not, isn't it better to add it in the class ?
> >
> > I had considered that, but wasn't sure, and was hoping nobody would ask
> > ;-)
> 
> I'm sorry :)
> 
> But once fixed in the constructor, it's fixed for good, isn't it ?

Yes, which is why it's probably best. I'll work on it.

> > It could make sense to replace the planes array and planesCount field
> > with an std::vector. A bit of a shame to have to use dynamic allocation
> > when we know there will never be more than 3 planes though. I wish there
> > was a dynamic array container, with fixed memory preallocation and
> > dynamic size up to the number of elements.
> 
> Can't we use std::array<3> in that case ? Although std::array<> is
> horrible to use :(

We can use std::array<>, but it won't initialize the contents of the
array by default. Furthermore, arrays have a fixed size, they don't
embed and additional field that reports the number of elements actually
used.

> > Any preference between using a vector, and keeping it as-is with a
> > constructor that sets numPlanes to 0 ?
> 
> To me a constructor sounds quite right, but other options are fine
> too.
> 
> > > > Signed-off-by: Laurent Pinchart <laurent.pinchart at ideasonboard.com>
> > > > ---
> > > >  src/libcamera/pipeline/simple/converter.cpp | 6 +++---
> > > >  1 file changed, 3 insertions(+), 3 deletions(-)
> > > >
> > > > diff --git a/src/libcamera/pipeline/simple/converter.cpp b/src/libcamera/pipeline/simple/converter.cpp
> > > > index b4ee021f6210..b3705abcd626 100644
> > > > --- a/src/libcamera/pipeline/simple/converter.cpp
> > > > +++ b/src/libcamera/pipeline/simple/converter.cpp
> > > > @@ -72,7 +72,7 @@ std::vector<PixelFormat> SimpleConverter::formats(PixelFormat input)
> > > >  	 * Set the format on the input side (V4L2 output) of the converter to
> > > >  	 * enumerate the conversion capabilities on its output (V4L2 capture).
> > > >  	 */
> > > > -	V4L2DeviceFormat format;
> > > > +	V4L2DeviceFormat format = {};
> > > >  	format.fourcc = m2m_->output()->toV4L2PixelFormat(input);
> > > >  	format.size = { 1, 1 };
> > > >
> > > > @@ -103,7 +103,7 @@ SizeRange SimpleConverter::sizes(const Size &input)
> > > >  	 * Set the size on the input side (V4L2 output) of the converter to
> > > >  	 * enumerate the scaling capabilities on its output (V4L2 capture).
> > > >  	 */
> > > > -	V4L2DeviceFormat format;
> > > > +	V4L2DeviceFormat format = {};
> > > >  	format.fourcc = V4L2PixelFormat();
> > > >  	format.size = input;
> > > >
> > > > @@ -142,7 +142,7 @@ SizeRange SimpleConverter::sizes(const Size &input)
> > > >  int SimpleConverter::configure(PixelFormat inputFormat, const Size &inputSize,
> > > >  			       StreamConfiguration *cfg)
> > > >  {
> > > > -	V4L2DeviceFormat format;
> > > > +	V4L2DeviceFormat format = {};
> > > >  	int ret;
> > > >
> > > >  	V4L2PixelFormat videoFormat = m2m_->output()->toV4L2PixelFormat(inputFormat);

-- 
Regards,

Laurent Pinchart


More information about the libcamera-devel mailing list