Re: [PATCH v8 13/13] [media] v4l: Document explicit synchronization behavior
On 03/13/2018 08:33 PM, Hans Verkuil wrote: > On 03/09/2018 09:49 AM, Gustavo Padovan wrote: >> From: Gustavo Padovan>> >> Add section to VIDIOC_QBUF and VIDIOC_QUERY_BUF about it >> >> v6: - Close some gaps in the docs (Hans) >> >> v5: >> - Remove V4L2_CAP_ORDERED >> - Add doc about V4L2_FMT_FLAG_UNORDERED >> >> v4: >> - Document ordering behavior for in-fences >> - Document V4L2_CAP_ORDERED capability >> - Remove doc about OUT_FENCE event >> - Document immediate return of out-fence in QBUF >> >> v3: >> - make the out_fence refer to the current buffer (Hans) >> - Note what happens when the IN_FENCE is not set (Hans) >> >> v2: >> - mention that fences are files (Hans) >> - rework for the new API >> >> Signed-off-by: Gustavo Padovan >> --- >> Documentation/media/uapi/v4l/vidioc-qbuf.rst | 55 >> +++- >> Documentation/media/uapi/v4l/vidioc-querybuf.rst | 12 -- >> 2 files changed, 63 insertions(+), 4 deletions(-) >> >> diff --git a/Documentation/media/uapi/v4l/vidioc-qbuf.rst >> b/Documentation/media/uapi/v4l/vidioc-qbuf.rst >> index 9e448a4aa3aa..371d84966e34 100644 >> --- a/Documentation/media/uapi/v4l/vidioc-qbuf.rst >> +++ b/Documentation/media/uapi/v4l/vidioc-qbuf.rst >> @@ -54,7 +54,7 @@ When the buffer is intended for output (``type`` is >> or ``V4L2_BUF_TYPE_VBI_OUTPUT``) applications must also initialize the >> ``bytesused``, ``field`` and ``timestamp`` fields, see :ref:`buffer` >> for details. Applications must also set ``flags`` to 0. The >> -``reserved2`` and ``reserved`` fields must be set to 0. When using the >> +``reserved`` field must be set to 0. When using the >> :ref:`multi-planar API `, the ``m.planes`` field must >> contain a userspace pointer to a filled-in array of struct >> :c:type:`v4l2_plane` and the ``length`` field must be set >> @@ -118,6 +118,59 @@ immediately with an ``EAGAIN`` error code when no >> buffer is available. >> The struct :c:type:`v4l2_buffer` structure is specified in >> :ref:`buffer`. >> >> +Explicit Synchronization >> + >> + >> +Explicit Synchronization allows us to control the synchronization of >> +shared buffers from userspace by passing fences to the kernel and/or >> +receiving them from it. Fences passed to the kernel are named in-fences and >> +the kernel should wait on them to signal before using the buffer. On the >> other >> +side, the kernel can create out-fences for the buffers it queues to the >> +drivers. Out-fences signal when the driver is finished with buffer, i.e., >> the >> +buffer is ready. The fences are represented as a file and passed as a file >> +descriptor to userspace. >> + >> +The in-fences are communicated to the kernel at the ``VIDIOC_QBUF`` ioctl >> +using the ``V4L2_BUF_FLAG_IN_FENCE`` buffer flag and the `fence_fd` field. >> If >> +an in-fence needs to be passed to the kernel, `fence_fd` should be set to >> the >> +fence file descriptor number and the ``V4L2_BUF_FLAG_IN_FENCE`` should be >> set >> +as well. Setting one but not the other will cause ``VIDIOC_QBUF`` to return >> +with an error. > > This sentence is confusing since it is not clear what 'one' and 'the other' > refer > to. Be specific here. I think it should be 'Setting V4L2_BUF_FLAG_IN_FENCE > but not > fence_fd'. Ignore this comment. > > The fence_fd field will be ignored if the >> +``V4L2_BUF_FLAG_IN_FENCE`` is not set. Looking at the code, I don't think this is correct. You get an error if IN_FENCE is not set but fence_fd is. Removing this sentence would fix this and the previous sentence would make a lot more sense. >> + >> +The videobuf2-core will guarantee that all buffers queued with an in-fence >> will >> +be queued to the drivers in the same order. Fences may signal out of order, >> so >> +this guarantee at videobuf2 is necessary to not change ordering. So when >> +waiting on a fence to signal all buffers queued after will be also block >> until > > after -> afterwards > will be also block -> will also be blocked > >> +that fence signal. > > signal -> signals What is missing in the documentation is what happens when you mix in-fence buffers and buffers without an in-fence. Regards, Hans
Re: [PATCH v8 13/13] [media] v4l: Document explicit synchronization behavior
On 03/09/2018 09:49 AM, Gustavo Padovan wrote: > From: Gustavo Padovan> > Add section to VIDIOC_QBUF and VIDIOC_QUERY_BUF about it > > v6: - Close some gaps in the docs (Hans) > > v5: > - Remove V4L2_CAP_ORDERED > - Add doc about V4L2_FMT_FLAG_UNORDERED > > v4: > - Document ordering behavior for in-fences > - Document V4L2_CAP_ORDERED capability > - Remove doc about OUT_FENCE event > - Document immediate return of out-fence in QBUF > > v3: > - make the out_fence refer to the current buffer (Hans) > - Note what happens when the IN_FENCE is not set (Hans) > > v2: > - mention that fences are files (Hans) > - rework for the new API > > Signed-off-by: Gustavo Padovan > --- > Documentation/media/uapi/v4l/vidioc-qbuf.rst | 55 > +++- > Documentation/media/uapi/v4l/vidioc-querybuf.rst | 12 -- > 2 files changed, 63 insertions(+), 4 deletions(-) > > diff --git a/Documentation/media/uapi/v4l/vidioc-qbuf.rst > b/Documentation/media/uapi/v4l/vidioc-qbuf.rst > index 9e448a4aa3aa..371d84966e34 100644 > --- a/Documentation/media/uapi/v4l/vidioc-qbuf.rst > +++ b/Documentation/media/uapi/v4l/vidioc-qbuf.rst > @@ -54,7 +54,7 @@ When the buffer is intended for output (``type`` is > or ``V4L2_BUF_TYPE_VBI_OUTPUT``) applications must also initialize the > ``bytesused``, ``field`` and ``timestamp`` fields, see :ref:`buffer` > for details. Applications must also set ``flags`` to 0. The > -``reserved2`` and ``reserved`` fields must be set to 0. When using the > +``reserved`` field must be set to 0. When using the > :ref:`multi-planar API `, the ``m.planes`` field must > contain a userspace pointer to a filled-in array of struct > :c:type:`v4l2_plane` and the ``length`` field must be set > @@ -118,6 +118,59 @@ immediately with an ``EAGAIN`` error code when no buffer > is available. > The struct :c:type:`v4l2_buffer` structure is specified in > :ref:`buffer`. > > +Explicit Synchronization > + > + > +Explicit Synchronization allows us to control the synchronization of > +shared buffers from userspace by passing fences to the kernel and/or > +receiving them from it. Fences passed to the kernel are named in-fences and > +the kernel should wait on them to signal before using the buffer. On the > other > +side, the kernel can create out-fences for the buffers it queues to the > +drivers. Out-fences signal when the driver is finished with buffer, i.e., the > +buffer is ready. The fences are represented as a file and passed as a file > +descriptor to userspace. > + > +The in-fences are communicated to the kernel at the ``VIDIOC_QBUF`` ioctl > +using the ``V4L2_BUF_FLAG_IN_FENCE`` buffer flag and the `fence_fd` field. If > +an in-fence needs to be passed to the kernel, `fence_fd` should be set to the > +fence file descriptor number and the ``V4L2_BUF_FLAG_IN_FENCE`` should be set > +as well. Setting one but not the other will cause ``VIDIOC_QBUF`` to return > +with an error. This sentence is confusing since it is not clear what 'one' and 'the other' refer to. Be specific here. I think it should be 'Setting V4L2_BUF_FLAG_IN_FENCE but not fence_fd'. The fence_fd field will be ignored if the > +``V4L2_BUF_FLAG_IN_FENCE`` is not set. > + > +The videobuf2-core will guarantee that all buffers queued with an in-fence > will > +be queued to the drivers in the same order. Fences may signal out of order, > so > +this guarantee at videobuf2 is necessary to not change ordering. So when > +waiting on a fence to signal all buffers queued after will be also block > until after -> afterwards will be also block -> will also be blocked > +that fence signal. signal -> signals > + > +If the in-fence signals with an error the buffer will be marked with > +``V4L2_BUF_FLAG_ERROR`` when returned to userspace at ``VIDIOC_DQBUF``. > +Even with the error the order of dequeueing the buffers are preserved. are -> is > + > +To get an out-fence back from V4L2 the ``V4L2_BUF_FLAG_OUT_FENCE`` flag > should > +be set to ask for a fence to be attached to the buffer. The out-fence fd is > +sent to userspace as a ``VIDIOC_QBUF`` return argument on the `fence_fd` > field. on -> in > + > +Note the the same `fence_fd` field is used for both sending the in-fence as > +input argument to receive the out-fence as a return argument. A buffer can input argument -> an input argument and > +have both in-fence ond out-fence. have both an in-fence and an out-fence. > + > +At streamoff the out-fences will either signal normally if the driver waits > +for the operations on the buffers to finish or signal with an error if the > +driver cancels the pending operations. Buffers with in-fences won't be queued > +to the driver if their fences signal. They will be cleaned up. > + > +The ``V4L2_FMT_FLAG_UNORDERED`` flag in ``VIDIOC_ENUM_FMT`` tells userspace > +that the when
Re: [PATCH v8 13/13] [media] v4l: Document explicit synchronization behavior
Hi Gustavo, a very small comment below On Fri, Mar 09, 2018 at 02:49:20PM -0300, Gustavo Padovan wrote: > From: Gustavo Padovan> > Add section to VIDIOC_QBUF and VIDIOC_QUERY_BUF about it > > v6: - Close some gaps in the docs (Hans) > > v5: > - Remove V4L2_CAP_ORDERED > - Add doc about V4L2_FMT_FLAG_UNORDERED > > v4: > - Document ordering behavior for in-fences > - Document V4L2_CAP_ORDERED capability > - Remove doc about OUT_FENCE event > - Document immediate return of out-fence in QBUF > > v3: > - make the out_fence refer to the current buffer (Hans) > - Note what happens when the IN_FENCE is not set (Hans) > > v2: > - mention that fences are files (Hans) > - rework for the new API > > Signed-off-by: Gustavo Padovan > --- > Documentation/media/uapi/v4l/vidioc-qbuf.rst | 55 > +++- > Documentation/media/uapi/v4l/vidioc-querybuf.rst | 12 -- > 2 files changed, 63 insertions(+), 4 deletions(-) > [snip] > +Note the the same `fence_fd` field is used for both sending the in-fence as > +input argument to receive the out-fence as a return argument. A buffer can > +have both in-fence ond out-fence. I feel like an "and" is missing here... the same `fence_fd` field is used for both sending the in-fence as input argument to receive the out-fence as a return argument the same `fence_fd` field is used for both sending the in-fence as input argument *and* to receive the out-fence as a return argument I'm not a native speaker so I might be wrong though. Thanks j signature.asc Description: PGP signature
[PATCH v8 13/13] [media] v4l: Document explicit synchronization behavior
From: Gustavo PadovanAdd section to VIDIOC_QBUF and VIDIOC_QUERY_BUF about it v6: - Close some gaps in the docs (Hans) v5: - Remove V4L2_CAP_ORDERED - Add doc about V4L2_FMT_FLAG_UNORDERED v4: - Document ordering behavior for in-fences - Document V4L2_CAP_ORDERED capability - Remove doc about OUT_FENCE event - Document immediate return of out-fence in QBUF v3: - make the out_fence refer to the current buffer (Hans) - Note what happens when the IN_FENCE is not set (Hans) v2: - mention that fences are files (Hans) - rework for the new API Signed-off-by: Gustavo Padovan --- Documentation/media/uapi/v4l/vidioc-qbuf.rst | 55 +++- Documentation/media/uapi/v4l/vidioc-querybuf.rst | 12 -- 2 files changed, 63 insertions(+), 4 deletions(-) diff --git a/Documentation/media/uapi/v4l/vidioc-qbuf.rst b/Documentation/media/uapi/v4l/vidioc-qbuf.rst index 9e448a4aa3aa..371d84966e34 100644 --- a/Documentation/media/uapi/v4l/vidioc-qbuf.rst +++ b/Documentation/media/uapi/v4l/vidioc-qbuf.rst @@ -54,7 +54,7 @@ When the buffer is intended for output (``type`` is or ``V4L2_BUF_TYPE_VBI_OUTPUT``) applications must also initialize the ``bytesused``, ``field`` and ``timestamp`` fields, see :ref:`buffer` for details. Applications must also set ``flags`` to 0. The -``reserved2`` and ``reserved`` fields must be set to 0. When using the +``reserved`` field must be set to 0. When using the :ref:`multi-planar API `, the ``m.planes`` field must contain a userspace pointer to a filled-in array of struct :c:type:`v4l2_plane` and the ``length`` field must be set @@ -118,6 +118,59 @@ immediately with an ``EAGAIN`` error code when no buffer is available. The struct :c:type:`v4l2_buffer` structure is specified in :ref:`buffer`. +Explicit Synchronization + + +Explicit Synchronization allows us to control the synchronization of +shared buffers from userspace by passing fences to the kernel and/or +receiving them from it. Fences passed to the kernel are named in-fences and +the kernel should wait on them to signal before using the buffer. On the other +side, the kernel can create out-fences for the buffers it queues to the +drivers. Out-fences signal when the driver is finished with buffer, i.e., the +buffer is ready. The fences are represented as a file and passed as a file +descriptor to userspace. + +The in-fences are communicated to the kernel at the ``VIDIOC_QBUF`` ioctl +using the ``V4L2_BUF_FLAG_IN_FENCE`` buffer flag and the `fence_fd` field. If +an in-fence needs to be passed to the kernel, `fence_fd` should be set to the +fence file descriptor number and the ``V4L2_BUF_FLAG_IN_FENCE`` should be set +as well. Setting one but not the other will cause ``VIDIOC_QBUF`` to return +with an error. The fence_fd field will be ignored if the +``V4L2_BUF_FLAG_IN_FENCE`` is not set. + +The videobuf2-core will guarantee that all buffers queued with an in-fence will +be queued to the drivers in the same order. Fences may signal out of order, so +this guarantee at videobuf2 is necessary to not change ordering. So when +waiting on a fence to signal all buffers queued after will be also block until +that fence signal. + +If the in-fence signals with an error the buffer will be marked with +``V4L2_BUF_FLAG_ERROR`` when returned to userspace at ``VIDIOC_DQBUF``. +Even with the error the order of dequeueing the buffers are preserved. + +To get an out-fence back from V4L2 the ``V4L2_BUF_FLAG_OUT_FENCE`` flag should +be set to ask for a fence to be attached to the buffer. The out-fence fd is +sent to userspace as a ``VIDIOC_QBUF`` return argument on the `fence_fd` field. + +Note the the same `fence_fd` field is used for both sending the in-fence as +input argument to receive the out-fence as a return argument. A buffer can +have both in-fence ond out-fence. + +At streamoff the out-fences will either signal normally if the driver waits +for the operations on the buffers to finish or signal with an error if the +driver cancels the pending operations. Buffers with in-fences won't be queued +to the driver if their fences signal. They will be cleaned up. + +The ``V4L2_FMT_FLAG_UNORDERED`` flag in ``VIDIOC_ENUM_FMT`` tells userspace +that the when using this format the order in which buffers are dequeued can +be different from the order in which they were queued. + +Ordering is important to fences because it can optimize the pipeline with +other drivers like a DRM/KMS display driver. For example, if a capture from the +camera is happening in an orderly manner one can send the capture buffer +out-fence to the DRM/KMS driver and rest sure that the buffers will be shown on +the screen at the correct order. If an ordered queue can not be set then such +arrangements with other drivers may not be possible. Return Value diff --git