[v2,1/2] media: docs-rst: Document memory-to-memory video decoder interface

Commit Message

Tomasz Figa Oct. 22, 2018, 2:48 p.m. UTC

Due to complexity of the video decoding process, the V4L2 drivers of
stateful decoder hardware require specific sequences of V4L2 API calls
to be followed. These include capability enumeration, initialization,
decoding, seek, pause, dynamic resolution change, drain and end of
stream.

Specifics of the above have been discussed during Media Workshops at
LinuxCon Europe 2012 in Barcelona and then later Embedded Linux
Conference Europe 2014 in Düsseldorf. The de facto Codec API that
originated at those events was later implemented by the drivers we already
have merged in mainline, such as s5p-mfc or coda.

The only thing missing was the real specification included as a part of
Linux Media documentation. Fix it now and document the decoder part of
the Codec API.

Signed-off-by: Tomasz Figa <[e-mail]>
---
 Documentation/media/uapi/v4l/dev-decoder.rst  | 1082 +++++++++++++++++
 Documentation/media/uapi/v4l/devices.rst      |    1 +
 Documentation/media/uapi/v4l/pixfmt-v4l2.rst  |    5 +
 Documentation/media/uapi/v4l/v4l2.rst         |   10 +-
 .../media/uapi/v4l/vidioc-decoder-cmd.rst     |   40 +-
 Documentation/media/uapi/v4l/vidioc-g-fmt.rst |   14 +
 6 files changed, 1137 insertions(+), 15 deletions(-)
 create mode 100644 Documentation/media/uapi/v4l/dev-decoder.rst

Comments

Stanimir Varbanov Oct. 29, 2018, 9:45 a.m. UTC | #1

Hi Tomasz,

On 10/22/2018 05:48 PM, Tomasz Figa wrote:
> Due to complexity of the video decoding process, the V4L2 drivers of
> stateful decoder hardware require specific sequences of V4L2 API calls
> to be followed. These include capability enumeration, initialization,
> decoding, seek, pause, dynamic resolution change, drain and end of
> stream.
> 
> Specifics of the above have been discussed during Media Workshops at
> LinuxCon Europe 2012 in Barcelona and then later Embedded Linux
> Conference Europe 2014 in Düsseldorf. The de facto Codec API that
> originated at those events was later implemented by the drivers we already
> have merged in mainline, such as s5p-mfc or coda.
> 
> The only thing missing was the real specification included as a part of
> Linux Media documentation. Fix it now and document the decoder part of
> the Codec API.
> 
> Signed-off-by: Tomasz Figa <[e-mail]>
> ---
>  Documentation/media/uapi/v4l/dev-decoder.rst  | 1082 +++++++++++++++++
>  Documentation/media/uapi/v4l/devices.rst      |    1 +
>  Documentation/media/uapi/v4l/pixfmt-v4l2.rst  |    5 +
>  Documentation/media/uapi/v4l/v4l2.rst         |   10 +-
>  .../media/uapi/v4l/vidioc-decoder-cmd.rst     |   40 +-
>  Documentation/media/uapi/v4l/vidioc-g-fmt.rst |   14 +
>  6 files changed, 1137 insertions(+), 15 deletions(-)
>  create mode 100644 Documentation/media/uapi/v4l/dev-decoder.rst
> 
> diff --git a/Documentation/media/uapi/v4l/dev-decoder.rst b/Documentation/media/uapi/v4l/dev-decoder.rst
> new file mode 100644
> index 000000000000..09c7a6621b8e
> --- /dev/null
> +++ b/Documentation/media/uapi/v4l/dev-decoder.rst

<cut>

> +Capture setup
> +=============
> +

<cut>

> +
> +2.  **Optional.** Acquire the visible resolution via
> +    :c:func:`VIDIOC_G_SELECTION`.
> +
> +    * **Required fields:**
> +
> +      ``type``
> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> +
> +      ``target``
> +          set to ``V4L2_SEL_TGT_COMPOSE``
> +
> +    * **Return fields:**
> +
> +      ``r.left``, ``r.top``, ``r.width``, ``r.height``
> +          the visible rectangle; it must fit within the frame buffer resolution
> +          returned by :c:func:`VIDIOC_G_FMT` on ``CAPTURE``.
> +
> +    * The following selection targets are supported on ``CAPTURE``:
> +
> +      ``V4L2_SEL_TGT_CROP_BOUNDS``
> +          corresponds to the coded resolution of the stream
> +
> +      ``V4L2_SEL_TGT_CROP_DEFAULT``
> +          the rectangle covering the part of the ``CAPTURE`` buffer that
> +          contains meaningful picture data (visible area); width and height
> +          will be equal to the visible resolution of the stream
> +
> +      ``V4L2_SEL_TGT_CROP``
> +          the rectangle within the coded resolution to be output to
> +          ``CAPTURE``; defaults to ``V4L2_SEL_TGT_CROP_DEFAULT``; read-only on
> +          hardware without additional compose/scaling capabilities

Hans should correct me if I'm wrong but V4L2_SEL_TGT_CROP_xxx are
applicable over OUTPUT queue type?

> +
> +      ``V4L2_SEL_TGT_COMPOSE_BOUNDS``
> +          the maximum rectangle within a ``CAPTURE`` buffer, which the cropped
> +          frame can be output into; equal to ``V4L2_SEL_TGT_CROP`` if the
> +          hardware does not support compose/scaling
> +
> +      ``V4L2_SEL_TGT_COMPOSE_DEFAULT``
> +          equal to ``V4L2_SEL_TGT_CROP``
> +
> +      ``V4L2_SEL_TGT_COMPOSE``
> +          the rectangle inside a ``CAPTURE`` buffer into which the cropped
> +          frame is written; defaults to ``V4L2_SEL_TGT_COMPOSE_DEFAULT``;
> +          read-only on hardware without additional compose/scaling capabilities
> +
> +      ``V4L2_SEL_TGT_COMPOSE_PADDED``
> +          the rectangle inside a ``CAPTURE`` buffer which is overwritten by the
> +          hardware; equal to ``V4L2_SEL_TGT_COMPOSE`` if the hardware does not
> +          write padding pixels
> +
> +    .. warning::
> +
> +       The values are guaranteed to be meaningful only after the decoder
> +       successfully parses the stream metadata. The client must not rely on the
> +       query before that happens.
> +

Tomasz Figa Oct. 29, 2018, 10:06 a.m. UTC | #2

Hi Stanimir,

On Mon, Oct 29, 2018 at 6:45 PM Stanimir Varbanov
<[e-mail]> wrote:
>
> Hi Tomasz,
>
> On 10/22/2018 05:48 PM, Tomasz Figa wrote:
> > Due to complexity of the video decoding process, the V4L2 drivers of
> > stateful decoder hardware require specific sequences of V4L2 API calls
> > to be followed. These include capability enumeration, initialization,
> > decoding, seek, pause, dynamic resolution change, drain and end of
> > stream.
> >
> > Specifics of the above have been discussed during Media Workshops at
> > LinuxCon Europe 2012 in Barcelona and then later Embedded Linux
> > Conference Europe 2014 in Düsseldorf. The de facto Codec API that
> > originated at those events was later implemented by the drivers we already
> > have merged in mainline, such as s5p-mfc or coda.
> >
> > The only thing missing was the real specification included as a part of
> > Linux Media documentation. Fix it now and document the decoder part of
> > the Codec API.
> >
> > Signed-off-by: Tomasz Figa <[e-mail]>
> > ---
> >  Documentation/media/uapi/v4l/dev-decoder.rst  | 1082 +++++++++++++++++
> >  Documentation/media/uapi/v4l/devices.rst      |    1 +
> >  Documentation/media/uapi/v4l/pixfmt-v4l2.rst  |    5 +
> >  Documentation/media/uapi/v4l/v4l2.rst         |   10 +-
> >  .../media/uapi/v4l/vidioc-decoder-cmd.rst     |   40 +-
> >  Documentation/media/uapi/v4l/vidioc-g-fmt.rst |   14 +
> >  6 files changed, 1137 insertions(+), 15 deletions(-)
> >  create mode 100644 Documentation/media/uapi/v4l/dev-decoder.rst
> >
> > diff --git a/Documentation/media/uapi/v4l/dev-decoder.rst b/Documentation/media/uapi/v4l/dev-decoder.rst
> > new file mode 100644
> > index 000000000000..09c7a6621b8e
> > --- /dev/null
> > +++ b/Documentation/media/uapi/v4l/dev-decoder.rst
>
> <cut>
>
> > +Capture setup
> > +=============
> > +
>
> <cut>
>
> > +
> > +2.  **Optional.** Acquire the visible resolution via
> > +    :c:func:`VIDIOC_G_SELECTION`.
> > +
> > +    * **Required fields:**
> > +
> > +      ``type``
> > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> > +
> > +      ``target``
> > +          set to ``V4L2_SEL_TGT_COMPOSE``
> > +
> > +    * **Return fields:**
> > +
> > +      ``r.left``, ``r.top``, ``r.width``, ``r.height``
> > +          the visible rectangle; it must fit within the frame buffer resolution
> > +          returned by :c:func:`VIDIOC_G_FMT` on ``CAPTURE``.
> > +
> > +    * The following selection targets are supported on ``CAPTURE``:
> > +
> > +      ``V4L2_SEL_TGT_CROP_BOUNDS``
> > +          corresponds to the coded resolution of the stream
> > +
> > +      ``V4L2_SEL_TGT_CROP_DEFAULT``
> > +          the rectangle covering the part of the ``CAPTURE`` buffer that
> > +          contains meaningful picture data (visible area); width and height
> > +          will be equal to the visible resolution of the stream
> > +
> > +      ``V4L2_SEL_TGT_CROP``
> > +          the rectangle within the coded resolution to be output to
> > +          ``CAPTURE``; defaults to ``V4L2_SEL_TGT_CROP_DEFAULT``; read-only on
> > +          hardware without additional compose/scaling capabilities
>
> Hans should correct me if I'm wrong but V4L2_SEL_TGT_CROP_xxx are
> applicable over OUTPUT queue type?

There is no such restriction. CROP selection targets of an OUTPUT
queue apply to the video stream read from the buffers, COMPOSE targets
of an OUTPUT queue apply to the output of the queue and input to the
processing block (hardware) in case of mem2mem devices, then CROP
targets of a CAPTURE queue apply to the output of the processing and
SELECTION targets of a CAPTURE queue apply to the stream written to
the buffers.

For a decoder, the OUTPUT stream is just a sequence of bytes, so
selection API doesn't apply to it. The processing (decoding) produces
a video stream and so the necessary selection capabilities are exposed
on the CAPTURE queue.

Best regards,
Tomasz

Tomasz Figa Oct. 29, 2018, 10:07 a.m. UTC | #3

On Mon, Oct 29, 2018 at 7:06 PM Tomasz Figa <[e-mail]> wrote:
>
> Hi Stanimir,
>
> On Mon, Oct 29, 2018 at 6:45 PM Stanimir Varbanov
> <[e-mail]> wrote:
> >
> > Hi Tomasz,
> >
> > On 10/22/2018 05:48 PM, Tomasz Figa wrote:
> > > Due to complexity of the video decoding process, the V4L2 drivers of
> > > stateful decoder hardware require specific sequences of V4L2 API calls
> > > to be followed. These include capability enumeration, initialization,
> > > decoding, seek, pause, dynamic resolution change, drain and end of
> > > stream.
> > >
> > > Specifics of the above have been discussed during Media Workshops at
> > > LinuxCon Europe 2012 in Barcelona and then later Embedded Linux
> > > Conference Europe 2014 in Düsseldorf. The de facto Codec API that
> > > originated at those events was later implemented by the drivers we already
> > > have merged in mainline, such as s5p-mfc or coda.
> > >
> > > The only thing missing was the real specification included as a part of
> > > Linux Media documentation. Fix it now and document the decoder part of
> > > the Codec API.
> > >
> > > Signed-off-by: Tomasz Figa <[e-mail]>
> > > ---
> > >  Documentation/media/uapi/v4l/dev-decoder.rst  | 1082 +++++++++++++++++
> > >  Documentation/media/uapi/v4l/devices.rst      |    1 +
> > >  Documentation/media/uapi/v4l/pixfmt-v4l2.rst  |    5 +
> > >  Documentation/media/uapi/v4l/v4l2.rst         |   10 +-
> > >  .../media/uapi/v4l/vidioc-decoder-cmd.rst     |   40 +-
> > >  Documentation/media/uapi/v4l/vidioc-g-fmt.rst |   14 +
> > >  6 files changed, 1137 insertions(+), 15 deletions(-)
> > >  create mode 100644 Documentation/media/uapi/v4l/dev-decoder.rst
> > >
> > > diff --git a/Documentation/media/uapi/v4l/dev-decoder.rst b/Documentation/media/uapi/v4l/dev-decoder.rst
> > > new file mode 100644
> > > index 000000000000..09c7a6621b8e
> > > --- /dev/null
> > > +++ b/Documentation/media/uapi/v4l/dev-decoder.rst
> >
> > <cut>
> >
> > > +Capture setup
> > > +=============
> > > +
> >
> > <cut>
> >
> > > +
> > > +2.  **Optional.** Acquire the visible resolution via
> > > +    :c:func:`VIDIOC_G_SELECTION`.
> > > +
> > > +    * **Required fields:**
> > > +
> > > +      ``type``
> > > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> > > +
> > > +      ``target``
> > > +          set to ``V4L2_SEL_TGT_COMPOSE``
> > > +
> > > +    * **Return fields:**
> > > +
> > > +      ``r.left``, ``r.top``, ``r.width``, ``r.height``
> > > +          the visible rectangle; it must fit within the frame buffer resolution
> > > +          returned by :c:func:`VIDIOC_G_FMT` on ``CAPTURE``.
> > > +
> > > +    * The following selection targets are supported on ``CAPTURE``:
> > > +
> > > +      ``V4L2_SEL_TGT_CROP_BOUNDS``
> > > +          corresponds to the coded resolution of the stream
> > > +
> > > +      ``V4L2_SEL_TGT_CROP_DEFAULT``
> > > +          the rectangle covering the part of the ``CAPTURE`` buffer that
> > > +          contains meaningful picture data (visible area); width and height
> > > +          will be equal to the visible resolution of the stream
> > > +
> > > +      ``V4L2_SEL_TGT_CROP``
> > > +          the rectangle within the coded resolution to be output to
> > > +          ``CAPTURE``; defaults to ``V4L2_SEL_TGT_CROP_DEFAULT``; read-only on
> > > +          hardware without additional compose/scaling capabilities
> >
> > Hans should correct me if I'm wrong but V4L2_SEL_TGT_CROP_xxx are
> > applicable over OUTPUT queue type?
>
> There is no such restriction. CROP selection targets of an OUTPUT
> queue apply to the video stream read from the buffers, COMPOSE targets
> of an OUTPUT queue apply to the output of the queue and input to the
> processing block (hardware) in case of mem2mem devices, then CROP
> targets of a CAPTURE queue apply to the output of the processing and
> SELECTION targets of a CAPTURE queue apply to the stream written to

I mean, COMPOSE targets. Sorry for the noise.

> the buffers.
>
> For a decoder, the OUTPUT stream is just a sequence of bytes, so
> selection API doesn't apply to it. The processing (decoding) produces
> a video stream and so the necessary selection capabilities are exposed
> on the CAPTURE queue.
>
> Best regards,
> Tomasz

Hans Verkuil Nov. 12, 2018, 11:37 a.m. UTC | #4

Hi Tomasz,

A general note for the stateful and stateless patches: they describe specific
use-cases of the more generic Codec Interface, and as such should be one
level deeper in the section hierarchy.

I.e. instead of being section 4.6/7/8:

https://hverkuil.home.xs4all.nl/request-api/uapi/v4l/devices.html

they should be 4.5.1/2/3.

On 10/22/2018 04:48 PM, Tomasz Figa wrote:
> Due to complexity of the video decoding process, the V4L2 drivers of
> stateful decoder hardware require specific sequences of V4L2 API calls
> to be followed. These include capability enumeration, initialization,
> decoding, seek, pause, dynamic resolution change, drain and end of
> stream.
> 
> Specifics of the above have been discussed during Media Workshops at
> LinuxCon Europe 2012 in Barcelona and then later Embedded Linux
> Conference Europe 2014 in Düsseldorf. The de facto Codec API that
> originated at those events was later implemented by the drivers we already
> have merged in mainline, such as s5p-mfc or coda.
> 
> The only thing missing was the real specification included as a part of
> Linux Media documentation. Fix it now and document the decoder part of
> the Codec API.
> 
> Signed-off-by: Tomasz Figa <[e-mail]>
> ---
>  Documentation/media/uapi/v4l/dev-decoder.rst  | 1082 +++++++++++++++++
>  Documentation/media/uapi/v4l/devices.rst      |    1 +
>  Documentation/media/uapi/v4l/pixfmt-v4l2.rst  |    5 +
>  Documentation/media/uapi/v4l/v4l2.rst         |   10 +-
>  .../media/uapi/v4l/vidioc-decoder-cmd.rst     |   40 +-
>  Documentation/media/uapi/v4l/vidioc-g-fmt.rst |   14 +
>  6 files changed, 1137 insertions(+), 15 deletions(-)
>  create mode 100644 Documentation/media/uapi/v4l/dev-decoder.rst
> 
> diff --git a/Documentation/media/uapi/v4l/dev-decoder.rst b/Documentation/media/uapi/v4l/dev-decoder.rst
> new file mode 100644
> index 000000000000..09c7a6621b8e
> --- /dev/null
> +++ b/Documentation/media/uapi/v4l/dev-decoder.rst
> @@ -0,0 +1,1082 @@
> +.. -*- coding: utf-8; mode: rst -*-
> +
> +.. _decoder:
> +
> +*************************************************
> +Memory-to-memory Stateful Video Decoder Interface
> +*************************************************
> +
> +A stateful video decoder takes complete chunks of the bitstream (e.g. Annex-B
> +H.264/HEVC stream, raw VP8/9 stream) and decodes them into raw video frames in
> +display order. The decoder is expected not to require any additional information
> +from the client to process these buffers.
> +
> +Performing software parsing, processing etc. of the stream in the driver in
> +order to support this interface is strongly discouraged. In case such
> +operations are needed, use of the Stateless Video Decoder Interface (in
> +development) is strongly advised.
> +
> +Conventions and notation used in this document
> +==============================================
> +
> +1. The general V4L2 API rules apply if not specified in this document
> +   otherwise.
> +
> +2. The meaning of words “must”, “may”, “should”, etc. is as per RFC
> +   2119.

Make this a link to https://tools.ietf.org/html/rfc2119

> +
> +3. All steps not marked “optional” are required.
> +
> +4. :c:func:`VIDIOC_G_EXT_CTRLS`, :c:func:`VIDIOC_S_EXT_CTRLS` may be used

, -> and

> +   interchangeably with :c:func:`VIDIOC_G_CTRL`, :c:func:`VIDIOC_S_CTRL`,

, -> and

> +   unless specified otherwise.
> +
> +5. Single-plane API (see spec) and applicable structures may be used

plane -> planar

Replace 'spec' with a reference to
https://hverkuil.home.xs4all.nl/spec/uapi/v4l/planar-apis.html#planar-apis

> +   interchangeably with Multi-plane API, unless specified otherwise,

plane -> planar

> +   depending on decoder capabilities and following the general V4L2
> +   guidelines.
> +
> +6. i = [a..b]: sequence of integers from a to b, inclusive, i.e. i =
> +   [0..2]: i = 0, 1, 2.
> +
> +7. Given an ``OUTPUT`` buffer A, A’ represents a buffer on the ``CAPTURE``

A, A' -> A, then A'

> +   queue containing data (decoded frame/stream) that resulted from processing

'(decoded frame/stream)': not clear what is meant here. Can't you just drop it?
I think the meaning is clear without that text.

> +   buffer A.
> +
> +.. _decoder-glossary:
> +
> +Glossary
> +========
> +
> +CAPTURE
> +   the destination buffer queue; for decoder, the queue of buffers containing

decoder -> decoders

> +   decoded frames; for encoder, the queue of buffers containing encoded

encoder -> encoders

> +   bitstream; ``V4L2_BUF_TYPE_VIDEO_CAPTURE```` or

```` -> ``

> +   ``V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE``; data are captured from the hardware

are -> is

> +   into ``CAPTURE`` buffers
> +
> +client
> +   application client communicating with the decoder or encoder implementing
> +   this interface
> +
> +coded format
> +   encoded/compressed video bitstream format (e.g. H.264, VP8, etc.); see
> +   also: raw format
> +
> +coded height
> +   height for given coded resolution
> +
> +coded resolution
> +   stream resolution in pixels aligned to codec and hardware requirements;
> +   typically visible resolution rounded up to full macroblocks;
> +   see also: visible resolution
> +
> +coded width
> +   width for given coded resolution
> +
> +decode order
> +   the order in which frames are decoded; may differ from display order if the
> +   coded format includes a feature of frame reordering; for decoders,
> +   ``OUTPUT`` buffers must be queued by the client in decode order; for
> +   encoders ``CAPTURE`` buffers must be returned by the encoder in decode order
> +
> +destination
> +   data resulting from the decode process; ``CAPTURE``

Do you mean:

see ``CAPTURE``

Right now there is just a 'CAPTURE' word after the ';', which looks weird.

> +
> +display order
> +   the order in which frames must be displayed; for encoders, ``OUTPUT``
> +   buffers must be queued by the client in display order; for decoders,
> +   ``CAPTURE`` buffers must be returned by the decoder in display order
> +
> +DPB
> +   Decoded Picture Buffer; an H.264 term for a buffer that stores a decoded
> +   raw frame available for reference in further decoding steps.
> +
> +EOS
> +   end of stream
> +
> +IDR
> +   Instantaneous Decoder Refresh; a type of a keyframe in H.264-encoded stream,
> +   which clears the list of earlier reference frames (DPBs)
> +
> +keyframe
> +   an encoded frame that does not reference frames decoded earlier, i.e.
> +   can be decoded fully on its own.
> +
> +macroblock
> +   a processing unit in image and video compression formats based on linear
> +   block transforms (e.g. H.264, VP8, VP9); codec-specific, but for most of
> +   popular codecs the size is 16x16 samples (pixels)
> +
> +OUTPUT
> +   the source buffer queue; for decoders, the queue of buffers containing
> +   encoded bitstream; for encoders, the queue of buffers containing raw frames;
> +   ``V4L2_BUF_TYPE_VIDEO_OUTPUT`` or ``V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE``; the
> +   hardware is fed with data from ``OUTPUT`` buffers
> +
> +PPS
> +   Picture Parameter Set; a type of metadata entity in H.264 bitstream
> +
> +raw format
> +   uncompressed format containing raw pixel data (e.g. YUV, RGB formats)
> +
> +resume point
> +   a point in the bitstream from which decoding may start/continue, without
> +   any previous state/data present, e.g.: a keyframe (VP8/VP9) or
> +   SPS/PPS/IDR sequence (H.264); a resume point is required to start decode
> +   of a new stream, or to resume decoding after a seek
> +
> +source
> +   data fed to the decoder or encoder; ``OUTPUT``

Should be: 'see ``OUTPUT``', I think.

> +
> +source height
> +   height in pixels for given source resolution; relevant to encoders only
> +
> +source resolution
> +   resolution in pixels of source frames being source to the encoder and
> +   subject to further cropping to the bounds of visible resolution; relevant to
> +   encoders only
> +
> +source width
> +   width in pixels for given source resolution; relevant to encoders only

I would drop these three terms: they are not used in this document since this
describes a decoder and not an encoder.

> +
> +SPS
> +   Sequence Parameter Set; a type of metadata entity in H.264 bitstream
> +
> +stream metadata
> +   additional (non-visual) information contained inside encoded bitstream;
> +   for example: coded resolution, visible resolution, codec profile
> +
> +visible height
> +   height for given visible resolution; display height
> +
> +visible resolution
> +   stream resolution of the visible picture, in pixels, to be used for
> +   display purposes; must be smaller or equal to coded resolution;
> +   display resolution
> +
> +visible width
> +   width for given visible resolution; display width
> +
> +State machine
> +=============
> +
> +.. kernel-render:: DOT
> +   :alt: DOT digraph of decoder state machine
> +   :caption: Decoder state machine
> +
> +   digraph decoder_state_machine {
> +       node [shape = doublecircle, label="Decoding"] Decoding;
> +
> +       node [shape = circle, label="Initialization"] Initialization;
> +       node [shape = circle, label="Capture\nsetup"] CaptureSetup;
> +       node [shape = circle, label="Dynamic\nresolution\nchange"] ResChange;
> +       node [shape = circle, label="Stopped"] Stopped;
> +       node [shape = circle, label="Drain"] Drain;
> +       node [shape = circle, label="Seek"] Seek;
> +       node [shape = circle, label="End of stream"] EoS;
> +
> +       node [shape = point]; qi
> +       qi -> Initialization [ label = "open()" ];
> +
> +       Initialization -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
> +
> +       CaptureSetup -> Stopped [ label = "CAPTURE\nbuffers\nready" ];
> +
> +       Decoding -> ResChange [ label = "Stream\nresolution\nchange" ];
> +       Decoding -> Drain [ label = "V4L2_DEC_CMD_STOP" ];
> +       Decoding -> EoS [ label = "EoS mark\nin the stream" ];
> +       Decoding -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +       Decoding -> Stopped [ label = "VIDIOC_STREAMOFF(CAPTURE)" ];
> +       Decoding -> Decoding;
> +
> +       ResChange -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
> +       ResChange -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +
> +       EoS -> Drain [ label = "Implicit\ndrain" ];
> +
> +       Drain -> Stopped [ label = "All CAPTURE\nbuffers dequeued\nor\nVIDIOC_STREAMOFF(CAPTURE)" ];
> +       Drain -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +
> +       Seek -> Decoding [ label = "VIDIOC_STREAMON(OUTPUT)" ];
> +       Seek -> Initialization [ label = "VIDIOC_REQBUFS(OUTPUT, 0)" ];
> +
> +       Stopped -> Decoding [ label = "V4L2_DEC_CMD_START\nor\nVIDIOC_STREAMON(CAPTURE)" ];
> +       Stopped -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +   }
> +
> +Querying capabilities
> +=====================
> +
> +1. To enumerate the set of coded formats supported by the decoder, the
> +   client may call :c:func:`VIDIOC_ENUM_FMT` on ``OUTPUT``.
> +
> +   * The full set of supported formats will be returned, regardless of the
> +     format set on ``CAPTURE``.
> +
> +2. To enumerate the set of supported raw formats, the client may call
> +   :c:func:`VIDIOC_ENUM_FMT` on ``CAPTURE``.
> +
> +   * Only the formats supported for the format currently active on ``OUTPUT``
> +     will be returned.
> +
> +   * In order to enumerate raw formats supported by a given coded format,
> +     the client must first set that coded format on ``OUTPUT`` and then
> +     enumerate formats on ``CAPTURE``.
> +
> +3. The client may use :c:func:`VIDIOC_ENUM_FRAMESIZES` to detect supported
> +   resolutions for a given format, passing desired pixel format in
> +   :c:type:`v4l2_frmsizeenum` ``pixel_format``.
> +
> +   * Values returned by :c:func:`VIDIOC_ENUM_FRAMESIZES` for a coded pixel
> +     formats will include all possible coded resolutions supported by the

formats -> format

> +     decoder for given coded pixel format.
> +
> +   * Values returned by :c:func:`VIDIOC_ENUM_FRAMESIZES` for a raw pixel format
> +     will include all possible frame buffer resolutions supported by the
> +     decoder for given raw pixel format and the coded format currently set on
> +     ``OUTPUT``.
> +
> +4. Supported profiles and levels for given format, if applicable, may be

given format -> the coded format currently set on ``OUTPUT``

> +   queried using their respective controls via :c:func:`VIDIOC_QUERYCTRL`.
> +
> +Initialization
> +==============
> +
> +1. **Optional.** Enumerate supported ``OUTPUT`` formats and resolutions. See
> +   `Querying capabilities` above.

I would drop this first step. It's obvious and it actually has nothing to do
with initialization as such.

> +
> +2. Set the coded format on ``OUTPUT`` via :c:func:`VIDIOC_S_FMT`
> +
> +   * **Required fields:**
> +
> +     ``type``
> +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> +
> +     ``pixelformat``
> +         a coded pixel format
> +
> +     ``width``, ``height``
> +         required only if cannot be parsed from the stream for the given
> +         coded format; optional otherwise - set to zero to ignore
> +
> +     ``sizeimage``
> +         desired size of ``OUTPUT`` buffers; the decoder may adjust it to
> +         match hardware requirements
> +
> +     other fields
> +         follow standard semantics
> +
> +   * **Return fields:**
> +
> +     ``sizeimage``
> +         adjusted size of ``CAPTURE`` buffers

CAPTURE -> OUTPUT

> +
> +   * If width and height are set to non-zero values, the ``CAPTURE`` format
> +     will be updated with an appropriate frame buffer resolution instantly.
> +     However, for coded formats that include stream resolution information,
> +     after the decoder is done parsing the information from the stream, it will
> +     update the ``CAPTURE`` format with new values and signal a source change
> +     event.

What if the initial width and height specified by userspace matches the parsed
width and height? Do you still get a source change event? I think you should
always get this event since there are other parameters that depend on the parsing
of the meta data.

But that should be made explicit here.

> +
> +   .. warning::

I'd call this a note rather than a warning.

> +
> +      Changing the ``OUTPUT`` format may change the currently set ``CAPTURE``
> +      format. The decoder will derive a new ``CAPTURE`` format from the
> +      ``OUTPUT`` format being set, including resolution, colorimetry
> +      parameters, etc. If the client needs a specific ``CAPTURE`` format, it
> +      must adjust it afterwards.
> +
> +3.  **Optional.** Query the minimum number of buffers required for ``OUTPUT``
> +    queue via :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to
> +    use more buffers than the minimum required by hardware/format.

Why is this useful? As far as I can tell only the s5p-mfc *encoder* supports
this control, so this seems pointless. And since the output queue gets a bitstream
I don't see any reason for reading this control in a decoder.

> +
> +    * **Required fields:**
> +
> +      ``id``
> +          set to ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
> +
> +    * **Return fields:**
> +
> +      ``value``
> +          the minimum number of ``OUTPUT`` buffers required for the currently
> +          set format
> +
> +4.  Allocate source (bitstream) buffers via :c:func:`VIDIOC_REQBUFS` on
> +    ``OUTPUT``.
> +
> +    * **Required fields:**
> +
> +      ``count``
> +          requested number of buffers to allocate; greater than zero
> +
> +      ``type``
> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> +
> +      ``memory``
> +          follows standard semantics
> +
> +    * **Return fields:**
> +
> +      ``count``
> +          the actual number of buffers allocated
> +
> +    .. warning::
> +
> +       The actual number of allocated buffers may differ from the ``count``
> +       given. The client must check the updated value of ``count`` after the
> +       call returns.
> +
> +    .. note::
> +
> +       To allocate more than the minimum number of buffers (for pipeline
> +       depth), the client may query the ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
> +       control to get the minimum number of buffers required by the
> +       decoder/format, and pass the obtained value plus the number of
> +       additional buffers needed in the ``count`` field to
> +       :c:func:`VIDIOC_REQBUFS`.

As mentioned above, this makes no sense for stateful decoders IMHO.

> +
> +    Alternatively, :c:func:`VIDIOC_CREATE_BUFS` on the ``OUTPUT`` queue can be
> +    used to have more control over buffer allocation.
> +
> +    * **Required fields:**
> +
> +      ``count``
> +          requested number of buffers to allocate; greater than zero
> +
> +      ``type``
> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> +
> +      ``memory``
> +          follows standard semantics
> +
> +      ``format``
> +          follows standard semantics
> +
> +    * **Return fields:**
> +
> +      ``count``
> +          adjusted to the number of allocated buffers
> +
> +    .. warning::
> +
> +       The actual number of allocated buffers may differ from the ``count``
> +       given. The client must check the updated value of ``count`` after the
> +       call returns.
> +
> +5.  Start streaming on the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`.
> +
> +6.  **This step only applies to coded formats that contain resolution information
> +    in the stream.** Continue queuing/dequeuing bitstream buffers to/from the
> +    ``OUTPUT`` queue via :c:func:`VIDIOC_QBUF` and :c:func:`VIDIOC_DQBUF`. The
> +    buffers will be processed and returned to the client in order, until
> +    required metadata to configure the ``CAPTURE`` queue are found. This is
> +    indicated by the decoder sending a ``V4L2_EVENT_SOURCE_CHANGE`` event with
> +    ``V4L2_EVENT_SRC_CH_RESOLUTION`` source change type.
> +
> +    * It is not an error if the first buffer does not contain enough data for
> +      this to occur. Processing of the buffers will continue as long as more
> +      data is needed.
> +
> +    * If data in a buffer that triggers the event is required to decode the
> +      first frame, it will not be returned to the client, until the
> +      initialization sequence completes and the frame is decoded.
> +
> +    * If the client sets width and height of the ``OUTPUT`` format to 0,
> +      calling :c:func:`VIDIOC_G_FMT`, :c:func:`VIDIOC_S_FMT` or
> +      :c:func:`VIDIOC_TRY_FMT` on the ``CAPTURE`` queue will return the
> +      ``-EACCES`` error code, until the decoder configures ``CAPTURE`` format
> +      according to stream metadata.
> +
> +    .. important::
> +
> +       Any client query issued after the decoder queues the event will return
> +       values applying to the just parsed stream, including queue formats,
> +       selection rectangles and controls.
> +
> +    .. note::
> +
> +       A client capable of acquiring stream parameters from the bitstream on
> +       its own may attempt to set the width and height of the ``OUTPUT`` format
> +       to non-zero values matching the coded size of the stream, skip this step
> +       and continue with the `Capture setup` sequence. However, it must not
> +       rely on any driver queries regarding stream parameters, such as
> +       selection rectangles and controls, since the decoder has not parsed them
> +       from the stream yet. If the values configured by the client do not match
> +       those parsed by the decoder, a `Dynamic resolution change` will be
> +       triggered to reconfigure them.
> +
> +    .. note::
> +
> +       No decoded frames are produced during this phase.
> +
> +7.  Continue with the `Capture setup` sequence.
> +
> +Capture setup
> +=============
> +
> +1.  Call :c:func:`VIDIOC_G_FMT` on the ``CAPTURE`` queue to get format for the
> +    destination buffers parsed/decoded from the bitstream.
> +
> +    * **Required fields:**
> +
> +      ``type``
> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> +
> +    * **Return fields:**
> +
> +      ``width``, ``height``
> +          frame buffer resolution for the decoded frames
> +
> +      ``pixelformat``
> +          pixel format for decoded frames
> +
> +      ``num_planes`` (for _MPLANE ``type`` only)
> +          number of planes for pixelformat
> +
> +      ``sizeimage``, ``bytesperline``
> +          as per standard semantics; matching frame buffer format
> +
> +    .. note::
> +
> +       The value of ``pixelformat`` may be any pixel format supported by the
> +       decoder for the current stream. The decoder should choose a
> +       preferred/optimal format for the default configuration. For example, a
> +       YUV format may be preferred over an RGB format if an additional
> +       conversion step would be required for the latter.
> +
> +2.  **Optional.** Acquire the visible resolution via
> +    :c:func:`VIDIOC_G_SELECTION`.
> +
> +    * **Required fields:**
> +
> +      ``type``
> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> +
> +      ``target``
> +          set to ``V4L2_SEL_TGT_COMPOSE``
> +
> +    * **Return fields:**
> +
> +      ``r.left``, ``r.top``, ``r.width``, ``r.height``
> +          the visible rectangle; it must fit within the frame buffer resolution
> +          returned by :c:func:`VIDIOC_G_FMT` on ``CAPTURE``.
> +
> +    * The following selection targets are supported on ``CAPTURE``:
> +
> +      ``V4L2_SEL_TGT_CROP_BOUNDS``
> +          corresponds to the coded resolution of the stream
> +
> +      ``V4L2_SEL_TGT_CROP_DEFAULT``
> +          the rectangle covering the part of the ``CAPTURE`` buffer that
> +          contains meaningful picture data (visible area); width and height
> +          will be equal to the visible resolution of the stream
> +
> +      ``V4L2_SEL_TGT_CROP``
> +          the rectangle within the coded resolution to be output to
> +          ``CAPTURE``; defaults to ``V4L2_SEL_TGT_CROP_DEFAULT``; read-only on
> +          hardware without additional compose/scaling capabilities
> +
> +      ``V4L2_SEL_TGT_COMPOSE_BOUNDS``
> +          the maximum rectangle within a ``CAPTURE`` buffer, which the cropped
> +          frame can be output into; equal to ``V4L2_SEL_TGT_CROP`` if the
> +          hardware does not support compose/scaling
> +
> +      ``V4L2_SEL_TGT_COMPOSE_DEFAULT``
> +          equal to ``V4L2_SEL_TGT_CROP``
> +
> +      ``V4L2_SEL_TGT_COMPOSE``
> +          the rectangle inside a ``CAPTURE`` buffer into which the cropped
> +          frame is written; defaults to ``V4L2_SEL_TGT_COMPOSE_DEFAULT``;
> +          read-only on hardware without additional compose/scaling capabilities
> +
> +      ``V4L2_SEL_TGT_COMPOSE_PADDED``
> +          the rectangle inside a ``CAPTURE`` buffer which is overwritten by the
> +          hardware; equal to ``V4L2_SEL_TGT_COMPOSE`` if the hardware does not
> +          write padding pixels
> +
> +    .. warning::
> +
> +       The values are guaranteed to be meaningful only after the decoder
> +       successfully parses the stream metadata. The client must not rely on the
> +       query before that happens.
> +
> +3.  Query the minimum number of buffers required for the ``CAPTURE`` queue via
> +    :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to use more
> +    buffers than the minimum required by hardware/format.

Is this step optional or required? Can it change when a resolution change occurs?
How does this relate to the checks for the minimum number of buffers that REQBUFS
does?

The 'This is useful if' sentence suggests that it is optional, but I think that
sentence just confuses the issue.

> +
> +    * **Required fields:**
> +
> +      ``id``
> +          set to ``V4L2_CID_MIN_BUFFERS_FOR_CAPTURE``
> +
> +    * **Return fields:**
> +
> +      ``value``
> +          minimum number of buffers required to decode the stream parsed in
> +          this initialization sequence.
> +
> +    .. note::
> +
> +       The minimum number of buffers must be at least the number required to
> +       successfully decode the current stream. This may for example be the
> +       required DPB size for an H.264 stream given the parsed stream
> +       configuration (resolution, level).
> +
> +    .. warning::
> +
> +       The value is guaranteed to be meaningful only after the decoder
> +       successfully parses the stream metadata. The client must not rely on the
> +       query before that happens.
> +
> +4.  **Optional.** Enumerate ``CAPTURE`` formats via :c:func:`VIDIOC_ENUM_FMT` on
> +    the ``CAPTURE`` queue. Once the stream information is parsed and known, the
> +    client may use this ioctl to discover which raw formats are supported for
> +    given stream and select one of them via :c:func:`VIDIOC_S_FMT`.

Can the list returned here differ from the list returned in the 'Querying capabilities'
step? If so, then I assume it will always be a subset of what was returned in
the 'Querying' step?

> +
> +    .. important::
> +
> +       The decoder will return only formats supported for the currently
> +       established coded format, as per the ``OUTPUT`` format and/or stream
> +       metadata parsed in this initialization sequence, even if more formats
> +       may be supported by the decoder in general.
> +
> +       For example, a decoder may support YUV and RGB formats for resolutions
> +       1920x1088 and lower, but only YUV for higher resolutions (due to
> +       hardware limitations). After parsing a resolution of 1920x1088 or lower,
> +       :c:func:`VIDIOC_ENUM_FMT` may return a set of YUV and RGB pixel formats,
> +       but after parsing resolution higher than 1920x1088, the decoder will not
> +       return RGB, unsupported for this resolution.
> +
> +       However, subsequent resolution change event triggered after
> +       discovering a resolution change within the same stream may switch
> +       the stream into a lower resolution and :c:func:`VIDIOC_ENUM_FMT`
> +       would return RGB formats again in that case.
> +
> +5.  **Optional.** Set the ``CAPTURE`` format via :c:func:`VIDIOC_S_FMT` on the
> +    ``CAPTURE`` queue. The client may choose a different format than
> +    selected/suggested by the decoder in :c:func:`VIDIOC_G_FMT`.
> +
> +    * **Required fields:**
> +
> +      ``type``
> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> +
> +      ``pixelformat``
> +          a raw pixel format
> +
> +    .. note::
> +
> +       The client may use :c:func:`VIDIOC_ENUM_FMT` after receiving the
> +       ``V4L2_EVENT_SOURCE_CHANGE`` event to find out the set of raw formats
> +       supported for the stream.

Isn't this a duplicate of step 4? I think this note can be dropped.

> +
> +6.  If all the following conditions are met, the client may resume the decoding
> +    instantly:
> +
> +    * ``sizeimage`` of the new format (determined in previous steps) is less
> +      than or equal to the size of currently allocated buffers,
> +
> +    * the number of buffers currently allocated is greater than or equal to the
> +      minimum number of buffers acquired in previous steps. To fulfill this
> +      requirement, the client may use :c:func:`VIDIOC_CREATE_BUFS` to add new
> +      buffers.
> +
> +    In such case, the remaining steps do not apply and the client may resume
> +    the decoding by one of the following actions:
> +
> +    * if the ``CAPTURE`` queue is streaming, call :c:func:`VIDIOC_DECODER_CMD`
> +      with the ``V4L2_DEC_CMD_START`` command,
> +
> +    * if the ``CAPTURE`` queue is not streaming, call :c:func:`VIDIOC_STREAMON`
> +      on the ``CAPTURE`` queue.
> +
> +    However, if the client intends to change the buffer set, to lower
> +    memory usage or for any other reasons, it may be achieved by following
> +    the steps below.
> +
> +7.  **If the** ``CAPTURE`` **queue is streaming,** keep queuing and dequeuing
> +    buffers on the ``CAPTURE`` queue until a buffer marked with the
> +    ``V4L2_BUF_FLAG_LAST`` flag is dequeued.
> +
> +8.  **If the** ``CAPTURE`` **queue is streaming,** call :c:func:`VIDIOC_STREAMOFF`
> +    on the ``CAPTURE`` queue to stop streaming.
> +
> +    .. warning::
> +
> +       The ``OUTPUT`` queue must remain streaming. Calling
> +       :c:func:`VIDIOC_STREAMOFF` on it would abort the sequence and trigger a
> +       seek.
> +
> +9.  **If the** ``CAPTURE`` **queue has buffers allocated,** free the ``CAPTURE``
> +    buffers using :c:func:`VIDIOC_REQBUFS`.
> +
> +    * **Required fields:**
> +
> +      ``count``
> +          set to 0
> +
> +      ``type``
> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> +
> +      ``memory``
> +          follows standard semantics
> +
> +10. Allocate ``CAPTURE`` buffers via :c:func:`VIDIOC_REQBUFS` on the
> +    ``CAPTURE`` queue.
> +
> +    * **Required fields:**
> +
> +      ``count``
> +          requested number of buffers to allocate; greater than zero
> +
> +      ``type``
> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> +
> +      ``memory``
> +          follows standard semantics
> +
> +    * **Return fields:**
> +
> +      ``count``
> +          actual number of buffers allocated
> +
> +    .. warning::
> +
> +       The actual number of allocated buffers may differ from the ``count``
> +       given. The client must check the updated value of ``count`` after the
> +       call returns.
> +
> +    .. note::
> +
> +       To allocate more than the minimum number of buffers (for pipeline
> +       depth), the client may query the ``V4L2_CID_MIN_BUFFERS_FOR_CAPTURE``
> +       control to get the minimum number of buffers required, and pass the
> +       obtained value plus the number of additional buffers needed in the
> +       ``count`` field to :c:func:`VIDIOC_REQBUFS`.

Same question as before: is it optional or required to obtain the value of this
control? And can't the driver just set the min_buffers_needed field in the capture
vb2_queue to the minimum number of buffers that are required?

Should you be allowed to allocate buffers at all if the capture format isn't
known? I.e. width/height is still 0. It makes no sense to call REQBUFS since
there is no format size known that REQBUFS can use.

> +
> +    Alternatively, :c:func:`VIDIOC_CREATE_BUFS` on the ``CAPTURE`` queue can be
> +    used to have more control over buffer allocation. For example, by
> +    allocating buffers larger than the current ``CAPTURE`` format, future
> +    resolution changes can be accommodated.
> +
> +    * **Required fields:**
> +
> +      ``count``
> +          requested number of buffers to allocate; greater than zero
> +
> +      ``type``
> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> +
> +      ``memory``
> +          follows standard semantics
> +
> +      ``format``
> +          a format representing the maximum framebuffer resolution to be
> +          accommodated by newly allocated buffers
> +
> +    * **Return fields:**
> +
> +      ``count``
> +          adjusted to the number of allocated buffers
> +
> +    .. warning::
> +
> +       The actual number of allocated buffers may differ from the ``count``
> +       given. The client must check the updated value of ``count`` after the
> +       call returns.
> +
> +    .. note::
> +
> +       To allocate buffers for a format different than parsed from the stream
> +       metadata, the client must proceed as follows, before the metadata
> +       parsing is initiated:
> +
> +       * set width and height of the ``OUTPUT`` format to desired coded resolution to
> +         let the decoder configure the ``CAPTURE`` format appropriately,
> +
> +       * query the ``CAPTURE`` format using :c:func:`VIDIOC_G_FMT` and save it
> +         until this step.
> +
> +       The format obtained in the query may be then used with
> +       :c:func:`VIDIOC_CREATE_BUFS` in this step to allocate the buffers.
> +
> +11. Call :c:func:`VIDIOC_STREAMON` on the ``CAPTURE`` queue to start decoding
> +    frames.
> +
> +Decoding
> +========
> +
> +This state is reached after the `Capture setup` sequence finishes succesfully.
> +In this state, the client queues and dequeues buffers to both queues via
> +:c:func:`VIDIOC_QBUF` and :c:func:`VIDIOC_DQBUF`, following the standard
> +semantics.
> +
> +The contents of the source ``OUTPUT`` buffers depend on the active coded pixel
> +format and may be affected by codec-specific extended controls, as stated in
> +the documentation of each format.
> +
> +Both queues operate independently, following the standard behavior of V4L2
> +buffer queues and memory-to-memory devices. In addition, the order of decoded
> +frames dequeued from the ``CAPTURE`` queue may differ from the order of queuing
> +coded frames to the ``OUTPUT`` queue, due to properties of the selected coded
> +format, e.g. frame reordering.
> +
> +The client must not assume any direct relationship between ``CAPTURE``
> +and ``OUTPUT`` buffers and any specific timing of buffers becoming
> +available to dequeue. Specifically,
> +
> +* a buffer queued to ``OUTPUT`` may result in no buffers being produced
> +  on ``CAPTURE`` (e.g. if it does not contain encoded data, or if only
> +  metadata syntax structures are present in it),
> +
> +* a buffer queued to ``OUTPUT`` may result in more than 1 buffer produced
> +  on ``CAPTURE`` (if the encoded data contained more than one frame, or if
> +  returning a decoded frame allowed the decoder to return a frame that
> +  preceded it in decode, but succeeded it in the display order),
> +
> +* a buffer queued to ``OUTPUT`` may result in a buffer being produced on
> +  ``CAPTURE`` later into decode process, and/or after processing further
> +  ``OUTPUT`` buffers, or be returned out of order, e.g. if display
> +  reordering is used,
> +
> +* buffers may become available on the ``CAPTURE`` queue without additional
> +  buffers queued to ``OUTPUT`` (e.g. during drain or ``EOS``), because of the
> +  ``OUTPUT`` buffers queued in the past whose decoding results are only
> +  available at later time, due to specifics of the decoding process.
> +
> +.. note::
> +
> +   To allow matching decoded ``CAPTURE`` buffers with ``OUTPUT`` buffers they
> +   originated from, the client can set the ``timestamp`` field of the
> +   :c:type:`v4l2_buffer` struct when queuing an ``OUTPUT`` buffer. The
> +   ``CAPTURE`` buffer(s), which resulted from decoding that ``OUTPUT`` buffer
> +   will have their ``timestamp`` field set to the same value when dequeued.
> +
> +   In addition to the straighforward case of one ``OUTPUT`` buffer producing
> +   one ``CAPTURE`` buffer, the following cases are defined:
> +
> +   * one ``OUTPUT`` buffer generates multiple ``CAPTURE`` buffers: the same
> +     ``OUTPUT`` timestamp will be copied to multiple ``CAPTURE`` buffers,
> +
> +   * multiple ``OUTPUT`` buffers generate one ``CAPTURE`` buffer: timestamp of
> +     the ``OUTPUT`` buffer queued last will be copied,
> +
> +   * the decoding order differs from the display order (i.e. the
> +     ``CAPTURE`` buffers are out-of-order compared to the ``OUTPUT`` buffers):
> +     ``CAPTURE`` timestamps will not retain the order of ``OUTPUT`` timestamps
> +     and thus monotonicity of the timestamps cannot be guaranteed.

Should stateful codecs be required to support 'tags'? See:

https://www.mail-archive.com/[e-mail]/msg136314.html

To be honest, I'm inclined to require this for all m2m devices eventually.

> +
> +During the decoding, the decoder may initiate one of the special sequences, as
> +listed below. The sequences will result in the decoder returning all the
> +``CAPTURE`` buffers that originated from all the ``OUTPUT`` buffers processed
> +before the sequence started. Last of the buffers will have the
> +``V4L2_BUF_FLAG_LAST`` flag set. To determine the sequence to follow, the client
> +must check if there is any pending event and,
> +
> +* if a ``V4L2_EVENT_SOURCE_CHANGE`` event is pending, the `Dynamic resolution
> +  change` sequence needs to be followed,
> +
> +* if a ``V4L2_EVENT_EOS`` event is pending, the `End of stream` sequence needs
> +  to be followed.
> +
> +Some of the sequences can be intermixed with each other and need to be handled
> +as they happen. The exact operation is documented for each sequence.
> +
> +Seek
> +====
> +
> +Seek is controlled by the ``OUTPUT`` queue, as it is the source of coded data.
> +The seek does not require any specific operation on the ``CAPTURE`` queue, but
> +it may be affected as per normal decoder operation.
> +
> +1. Stop the ``OUTPUT`` queue to begin the seek sequence via
> +   :c:func:`VIDIOC_STREAMOFF`.
> +
> +   * **Required fields:**
> +
> +     ``type``
> +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> +
> +   * The decoder will drop all the pending ``OUTPUT`` buffers and they must be
> +     treated as returned to the client (following standard semantics).
> +
> +2. Restart the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`
> +
> +   * **Required fields:**
> +
> +     ``type``
> +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> +
> +   * The decoder will start accepting new source bitstream buffers after the
> +     call returns.
> +
> +3. Start queuing buffers containing coded data after the seek to the ``OUTPUT``
> +   queue until a suitable resume point is found.
> +
> +   .. note::
> +
> +      There is no requirement to begin queuing coded data starting exactly
> +      from a resume point (e.g. SPS or a keyframe). Any queued ``OUTPUT``
> +      buffers will be processed and returned to the client until a suitable
> +      resume point is found.  While looking for a resume point, the decoder
> +      should not produce any decoded frames into ``CAPTURE`` buffers.
> +
> +      Some hardware is known to mishandle seeks to a non-resume point. Such an
> +      operation may result in an unspecified number of corrupted decoded frames
> +      being made available on the ``CAPTURE`` queue. Drivers must ensure that
> +      no fatal decoding errors or crashes occur, and implement any necessary
> +      handling and workarounds for hardware issues related to seek operations.

Is there a requirement that those corrupted frames have V4L2_BUF_FLAG_ERROR set?
I.e., can userspace detect those currupted frames?

> +
> +   .. warning::
> +
> +      In case of the H.264 codec, the client must take care not to seek over a
> +      change of SPS/PPS. Even though the target frame could be a keyframe, the
> +      stale SPS/PPS inside decoder state would lead to undefined results when
> +      decoding. Although the decoder must handle such case without a crash or a
> +      fatal decode error, the client must not expect a sensible decode output.
> +
> +4. After a resume point is found, the decoder will start returning ``CAPTURE``
> +   buffers containing decoded frames.
> +
> +.. important::
> +
> +   A seek may result in the `Dynamic resolution change` sequence being
> +   iniitated, due to the seek target having decoding parameters different from
> +   the part of the stream decoded before the seek. The sequence must be handled
> +   as per normal decoder operation.
> +
> +.. warning::
> +
> +   It is not specified when the ``CAPTURE`` queue starts producing buffers
> +   containing decoded data from the ``OUTPUT`` buffers queued after the seek,
> +   as it operates independently from the ``OUTPUT`` queue.
> +
> +   The decoder may return a number of remaining ``CAPTURE`` buffers containing
> +   decoded frames originating from the ``OUTPUT`` buffers queued before the
> +   seek sequence is performed.
> +
> +   The ``VIDIOC_STREAMOFF`` operation discards any remaining queued
> +   ``OUTPUT`` buffers, which means that not all of the ``OUTPUT`` buffers
> +   queued before the seek sequence may have matching ``CAPTURE`` buffers
> +   produced.  For example, given the sequence of operations on the
> +   ``OUTPUT`` queue:
> +
> +     QBUF(A), QBUF(B), STREAMOFF(), STREAMON(), QBUF(G), QBUF(H),
> +
> +   any of the following results on the ``CAPTURE`` queue is allowed:
> +
> +     {A’, B’, G’, H’}, {A’, G’, H’}, {G’, H’}.

Isn't it the case that if you would want to avoid that, then you should call
DECODER_STOP, wait for the last buffer on the CAPTURE queue, then seek and
call DECODER_START. If you do that, then you should always get {A’, B’, G’, H’}.
(basically following the Drain sequence).

Admittedly, you typically want to do an instantaneous seek, so this is probably
not what you want to do normally.

It might help to have this documented in a separate note.

> +
> +.. note::
> +
> +   To achieve instantaneous seek, the client may restart streaming on the
> +   ``CAPTURE`` queue too to discard decoded, but not yet dequeued buffers.
> +
> +Dynamic resolution change
> +=========================
> +
> +Streams that include resolution metadata in the bitstream may require switching
> +to a different resolution during the decoding.
> +
> +The sequence starts when the decoder detects a coded frame with one or more of
> +the following parameters different from previously established (and reflected
> +by corresponding queries):
> +
> +* coded resolution (``OUTPUT`` width and height),
> +
> +* visible resolution (selection rectangles),
> +
> +* the minimum number of buffers needed for decoding.
> +
> +Whenever that happens, the decoder must proceed as follows:
> +
> +1.  After encountering a resolution change in the stream, the decoder sends a
> +    ``V4L2_EVENT_SOURCE_CHANGE`` event with source change type set to
> +    ``V4L2_EVENT_SRC_CH_RESOLUTION``.
> +
> +    .. important::
> +
> +       Any client query issued after the decoder queues the event will return
> +       values applying to the stream after the resolution change, including
> +       queue formats, selection rectangles and controls.
> +
> +2.  The decoder will then process and decode all remaining buffers from before
> +    the resolution change point.
> +
> +    * The last buffer from before the change must be marked with the
> +      ``V4L2_BUF_FLAG_LAST`` flag, similarly to the `Drain` sequence above.
> +
> +    .. warning::
> +
> +       The last buffer may be empty (with :c:type:`v4l2_buffer` ``bytesused``
> +       = 0) and in such case it must be ignored by the client, as it does not
> +       contain a decoded frame.
> +
> +    .. note::
> +
> +       Any attempt to dequeue more buffers beyond the buffer marked with
> +       ``V4L2_BUF_FLAG_LAST`` will result in a -EPIPE error from
> +       :c:func:`VIDIOC_DQBUF`.
> +
> +The client must continue the sequence as described below to continue the
> +decoding process.
> +
> +1.  Dequeue the source change event.
> +
> +    .. important::
> +
> +       A source change triggers an implicit decoder drain, similar to the
> +       explicit `Drain` sequence. The decoder is stopped after it completes.
> +       The decoding process must be resumed with either a pair of calls to
> +       :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
> +       ``CAPTURE`` queue, or a call to :c:func:`VIDIOC_DECODER_CMD` with the
> +       ``V4L2_DEC_CMD_START`` command.
> +
> +2.  Continue with the `Capture setup` sequence.
> +
> +.. note::
> +
> +   During the resolution change sequence, the ``OUTPUT`` queue must remain
> +   streaming. Calling :c:func:`VIDIOC_STREAMOFF` on the ``OUTPUT`` queue would
> +   abort the sequence and initiate a seek.
> +
> +   In principle, the ``OUTPUT`` queue operates separately from the ``CAPTURE``
> +   queue and this remains true for the duration of the entire resolution change
> +   sequence as well.
> +
> +   The client should, for best performance and simplicity, keep queuing/dequeuing
> +   buffers to/from the ``OUTPUT`` queue even while processing this sequence.
> +
> +Drain
> +=====
> +
> +To ensure that all queued ``OUTPUT`` buffers have been processed and related
> +``CAPTURE`` buffers output to the client, the client must follow the drain
> +sequence described below. After the drain sequence ends, the client has
> +received all decoded frames for all ``OUTPUT`` buffers queued before the
> +sequence was started.
> +
> +1. Begin drain by issuing :c:func:`VIDIOC_DECODER_CMD`.
> +
> +   * **Required fields:**
> +
> +     ``cmd``
> +         set to ``V4L2_DEC_CMD_STOP``
> +
> +     ``flags``
> +         set to 0
> +
> +     ``pts``
> +         set to 0
> +
> +   .. warning::
> +
> +   The sentence can be only initiated if both ``OUTPUT`` and ``CAPTURE`` queues

'sentence'? You mean 'decoder command'?

> +   are streaming. For compatibility reasons, the call to
> +   :c:func:`VIDIOC_DECODER_CMD` will not fail even if any of the queues is not
> +   streaming, but at the same time it will not initiate the `Drain` sequence
> +   and so the steps described below would not be applicable.
> +
> +2. Any ``OUTPUT`` buffers queued by the client before the
> +   :c:func:`VIDIOC_DECODER_CMD` was issued will be processed and decoded as
> +   normal. The client must continue to handle both queues independently,
> +   similarly to normal decode operation. This includes,
> +
> +   * handling any operations triggered as a result of processing those buffers,
> +     such as the `Dynamic resolution change` sequence, before continuing with
> +     the drain sequence,
> +
> +   * queuing and dequeuing ``CAPTURE`` buffers, until a buffer marked with the
> +     ``V4L2_BUF_FLAG_LAST`` flag is dequeued,
> +
> +     .. warning::
> +
> +        The last buffer may be empty (with :c:type:`v4l2_buffer`
> +        ``bytesused`` = 0) and in such case it must be ignored by the client,
> +        as it does not contain a decoded frame.
> +
> +     .. note::
> +
> +        Any attempt to dequeue more buffers beyond the buffer marked with
> +        ``V4L2_BUF_FLAG_LAST`` will result in a -EPIPE error from
> +        :c:func:`VIDIOC_DQBUF`.
> +
> +   * dequeuing processed ``OUTPUT`` buffers, until all the buffers queued
> +     before the ``V4L2_DEC_CMD_STOP`` command are dequeued.
> +
> +   * dequeuing the ``V4L2_EVENT_EOS`` event, if the client subscribed to it.
> +
> +   .. note::
> +
> +      For backwards compatibility, the decoder will signal a ``V4L2_EVENT_EOS``
> +      event when the last the last frame has been decoded and all frames are

'the last the last' -> the last

> +      ready to be dequeued. It is a deprecated behavior and the client must not
> +      rely on it. The ``V4L2_BUF_FLAG_LAST`` buffer flag should be used
> +      instead.
> +
> +3. Once all the ``OUTPUT`` buffers queued before the ``V4L2_DEC_CMD_STOP`` call
> +   and the last ``CAPTURE`` buffer are dequeued, the decoder is stopped and it

This sentence is a bit confusing. This is better IMHO:

3. Once all the ``OUTPUT`` buffers queued before the ``V4L2_DEC_CMD_STOP`` call
   are dequeued and the last ``CAPTURE`` buffer is dequeued, the decoder is stopped and it

> +   will accept, but not process any newly queued ``OUTPUT`` buffers until the

process any -> process, any

> +   client issues any of the following operations:
> +
> +   * ``V4L2_DEC_CMD_START`` - the decoder will resume the operation normally,
> +
> +   * a pair of :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
> +     ``CAPTURE`` queue - the decoder will resume the operation normally,
> +     however any ``CAPTURE`` buffers still in the queue will be returned to the
> +     client,
> +
> +   * a pair of :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
> +     ``OUTPUT`` queue - any pending source buffers will be returned to the
> +     client and the `Seek` sequence will be triggered.
> +
> +.. note::
> +
> +   Once the drain sequence is initiated, the client needs to drive it to
> +   completion, as described by the steps above, unless it aborts the process by
> +   issuing :c:func:`VIDIOC_STREAMOFF` on any of the ``OUTPUT`` or ``CAPTURE``
> +   queues.  The client is not allowed to issue ``V4L2_DEC_CMD_START`` or
> +   ``V4L2_DEC_CMD_STOP`` again while the drain sequence is in progress and they
> +   will fail with -EBUSY error code if attempted.
> +
> +   Although mandatory, the availability of decoder commands may be queried
> +   using :c:func:`VIDIOC_TRY_DECODER_CMD`.
> +
> +End of stream
> +=============
> +
> +If the decoder encounters an end of stream marking in the stream, the decoder
> +will initiate the `Drain` sequence, which the client must handle as described
> +above, skipping the initial :c:func:`VIDIOC_DECODER_CMD`.
> +
> +Commit points
> +=============
> +
> +Setting formats and allocating buffers trigger changes in the behavior of the
> +decoder.
> +
> +1. Setting the format on the ``OUTPUT`` queue may change the set of formats
> +   supported/advertised on the ``CAPTURE`` queue. In particular, it also means
> +   that the ``CAPTURE`` format may be reset and the client must not rely on the
> +   previously set format being preserved.
> +
> +2. Enumerating formats on the ``CAPTURE`` queue always returns only formats
> +   supported for the current ``OUTPUT`` format.
> +
> +3. Setting the format on the ``CAPTURE`` queue does not change the list of
> +   formats available on the ``OUTPUT`` queue. An attempt to set the ``CAPTURE``
> +   format that is not supported for the currently selected ``OUTPUT`` format
> +   will result in the decoder adjusting the requested ``CAPTURE`` format to a
> +   supported one.
> +
> +4. Enumerating formats on the ``OUTPUT`` queue always returns the full set of
> +   supported coded formats, irrespectively of the current ``CAPTURE`` format.
> +
> +5. While buffers are allocated on the ``OUTPUT`` queue, the client must not
> +   change the format on the queue. Drivers will return the -EBUSY error code
> +   for any such format change attempt.
> +
> +To summarize, setting formats and allocation must always start with the
> +``OUTPUT`` queue and the ``OUTPUT`` queue is the master that governs the
> +set of supported formats for the ``CAPTURE`` queue.
> diff --git a/Documentation/media/uapi/v4l/devices.rst b/Documentation/media/uapi/v4l/devices.rst
> index fb7f8c26cf09..12d43fe711cf 100644
> --- a/Documentation/media/uapi/v4l/devices.rst
> +++ b/Documentation/media/uapi/v4l/devices.rst
> @@ -15,6 +15,7 @@ Interfaces
>      dev-output
>      dev-osd
>      dev-codec
> +    dev-decoder
>      dev-effect
>      dev-raw-vbi
>      dev-sliced-vbi
> diff --git a/Documentation/media/uapi/v4l/pixfmt-v4l2.rst b/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
> index 826f2305da01..ca5f2270a829 100644
> --- a/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
> +++ b/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
> @@ -32,6 +32,11 @@ Single-planar format structure
>  	to a multiple of the scale factor of any smaller planes. For
>  	example when the image format is YUV 4:2:0, ``width`` and
>  	``height`` must be multiples of two.
> +
> +	For compressed formats that contain the resolution information encoded
> +	inside the stream, when fed to a stateful mem2mem decoder, the fields
> +	may be zero to rely on the decoder to detect the right values. For more
> +	details see :ref:`decoder` and format descriptions.
>      * - __u32
>        - ``pixelformat``
>        - The pixel format or type of compression, set by the application.
> diff --git a/Documentation/media/uapi/v4l/v4l2.rst b/Documentation/media/uapi/v4l/v4l2.rst
> index b89e5621ae69..65dc096199ad 100644
> --- a/Documentation/media/uapi/v4l/v4l2.rst
> +++ b/Documentation/media/uapi/v4l/v4l2.rst
> @@ -53,6 +53,10 @@ Authors, in alphabetical order:
>  
>    - Original author of the V4L2 API and documentation.
>  
> +- Figa, Tomasz <[e-mail]>
> +
> +  - Documented the memory-to-memory decoder interface.
> +
>  - H Schimek, Michael <[e-mail]>
>  
>    - Original author of the V4L2 API and documentation.
> @@ -61,6 +65,10 @@ Authors, in alphabetical order:
>  
>    - Documented the Digital Video timings API.
>  
> +- Osciak, Pawel <[e-mail]>
> +
> +  - Documented the memory-to-memory decoder interface.
> +
>  - Osciak, Pawel <[e-mail]>
>  
>    - Designed and documented the multi-planar API.
> @@ -85,7 +93,7 @@ Authors, in alphabetical order:
>  
>    - Designed and documented the VIDIOC_LOG_STATUS ioctl, the extended control ioctls, major parts of the sliced VBI API, the MPEG encoder and decoder APIs and the DV Timings API.
>  
> -**Copyright** |copy| 1999-2016: Bill Dirks, Michael H. Schimek, Hans Verkuil, Martin Rubli, Andy Walls, Muralidharan Karicheri, Mauro Carvalho Chehab, Pawel Osciak, Sakari Ailus & Antti Palosaari.
> +**Copyright** |copy| 1999-2018: Bill Dirks, Michael H. Schimek, Hans Verkuil, Martin Rubli, Andy Walls, Muralidharan Karicheri, Mauro Carvalho Chehab, Pawel Osciak, Sakari Ailus & Antti Palosaari, Tomasz Figa
>  
>  Except when explicitly stated as GPL, programming examples within this
>  part can be used and distributed without restrictions.
> diff --git a/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst b/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
> index 85c916b0ce07..2f73fe22a9cd 100644
> --- a/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
> +++ b/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
> @@ -49,14 +49,16 @@ The ``cmd`` field must contain the command code. Some commands use the
>  
>  A :ref:`write() <func-write>` or :ref:`VIDIOC_STREAMON`
>  call sends an implicit START command to the decoder if it has not been
> -started yet.
> +started yet. Applies to both queues of mem2mem decoders.
>  
>  A :ref:`close() <func-close>` or :ref:`VIDIOC_STREAMOFF <VIDIOC_STREAMON>`
>  call of a streaming file descriptor sends an implicit immediate STOP
> -command to the decoder, and all buffered data is discarded.
> +command to the decoder, and all buffered data is discarded. Applies to both
> +queues of mem2mem decoders.
>  
> -These ioctls are optional, not all drivers may support them. They were
> -introduced in Linux 3.3.
> +In principle, these ioctls are optional, not all drivers may support them. They were
> +introduced in Linux 3.3. They are, however, mandatory for stateful mem2mem decoders
> +(as further documented in :ref:`decoder`).
>  
>  
>  .. tabularcolumns:: |p{1.1cm}|p{2.4cm}|p{1.2cm}|p{1.6cm}|p{10.6cm}|
> @@ -160,26 +162,36 @@ introduced in Linux 3.3.
>  	``V4L2_DEC_CMD_RESUME`` for that. This command has one flag:
>  	``V4L2_DEC_CMD_START_MUTE_AUDIO``. If set, then audio will be
>  	muted when playing back at a non-standard speed.
> +
> +	For stateful mem2mem decoders, the command may be also used to restart
> +	the decoder in case of an implicit stop initiated by the decoder
> +	itself, without the ``V4L2_DEC_CMD_STOP`` being called explicitly.
> +	No flags or other arguments are accepted in case of mem2mem decoders.
> +	See :ref:`decoder` for more details.
>      * - ``V4L2_DEC_CMD_STOP``
>        - 1
>        - Stop the decoder. When the decoder is already stopped, this
>  	command does nothing. This command has two flags: if
>  	``V4L2_DEC_CMD_STOP_TO_BLACK`` is set, then the decoder will set
>  	the picture to black after it stopped decoding. Otherwise the last
> -	image will repeat. mem2mem decoders will stop producing new frames
> -	altogether. They will send a ``V4L2_EVENT_EOS`` event when the
> -	last frame has been decoded and all frames are ready to be
> -	dequeued and will set the ``V4L2_BUF_FLAG_LAST`` buffer flag on
> -	the last buffer of the capture queue to indicate there will be no
> -	new buffers produced to dequeue. This buffer may be empty,
> -	indicated by the driver setting the ``bytesused`` field to 0. Once
> -	the ``V4L2_BUF_FLAG_LAST`` flag was set, the
> -	:ref:`VIDIOC_DQBUF <VIDIOC_QBUF>` ioctl will not block anymore,
> -	but return an ``EPIPE`` error code. If
> +	image will repeat. If
>  	``V4L2_DEC_CMD_STOP_IMMEDIATELY`` is set, then the decoder stops
>  	immediately (ignoring the ``pts`` value), otherwise it will keep
>  	decoding until timestamp >= pts or until the last of the pending
>  	data from its internal buffers was decoded.
> +
> +	A stateful mem2mem decoder will proceed with decoding the source
> +	buffers pending before the command is issued and then stop producing
> +	new frames. It will send a ``V4L2_EVENT_EOS`` event when the last frame
> +	has been decoded and all frames are ready to be dequeued and will set
> +	the ``V4L2_BUF_FLAG_LAST`` buffer flag on the last buffer of the
> +	capture queue to indicate there will be no new buffers produced to
> +	dequeue. This buffer may be empty, indicated by the driver setting the
> +	``bytesused`` field to 0. Once the buffer with the
> +	``V4L2_BUF_FLAG_LAST`` flag set was dequeued, the :ref:`VIDIOC_DQBUF
> +	<VIDIOC_QBUF>` ioctl will not block anymore, but return an ``EPIPE``
> +	error code. No flags or other arguments are accepted in case of mem2mem
> +	decoders.  See :ref:`decoder` for more details.
>      * - ``V4L2_DEC_CMD_PAUSE``
>        - 2
>        - Pause the decoder. When the decoder has not been started yet, the
> diff --git a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
> index 3ead350e099f..0fc0b78a943e 100644
> --- a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
> +++ b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
> @@ -53,6 +53,13 @@ devices that is either the struct
>  member. When the requested buffer type is not supported drivers return
>  an ``EINVAL`` error code.
>  
> +A stateful mem2mem decoder will not allow operations on the
> +``V4L2_BUF_TYPE_VIDEO_CAPTURE`` or ``V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE``
> +buffer type until the corresponding ``V4L2_BUF_TYPE_VIDEO_OUTPUT`` or
> +``V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE`` buffer type is configured. If such an
> +operation is attempted, drivers return an ``EACCES`` error code. Refer to
> +:ref:`decoder` for more details.

This isn't right. EACCES is returned as long as the output format resolution is
unknown. If it is set explicitly, then this will work without an error.

> +
>  To change the current format parameters applications initialize the
>  ``type`` field and all fields of the respective ``fmt`` union member.
>  For details see the documentation of the various devices types in
> @@ -145,6 +152,13 @@ On success 0 is returned, on error -1 and the ``errno`` variable is set
>  appropriately. The generic error codes are described at the
>  :ref:`Generic Error Codes <gen-errors>` chapter.
>  
> +EACCES
> +    The format is not accessible until another buffer type is configured.
> +    Relevant for the V4L2_BUF_TYPE_VIDEO_CAPTURE and
> +    V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE buffer types of mem2mem decoders, which
> +    require the format of V4L2_BUF_TYPE_VIDEO_OUTPUT or
> +    V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE buffer type to be configured first.

Ditto.

> +
>  EINVAL
>      The struct :c:type:`v4l2_format` ``type`` field is
>      invalid or the requested buffer type not supported.
> 

Regards,

	Hans

Stanimir Varbanov Nov. 12, 2018, 3:04 p.m. UTC | #5

Hi Tomasz,

On 10/22/18 5:48 PM, Tomasz Figa wrote:
> Due to complexity of the video decoding process, the V4L2 drivers of
> stateful decoder hardware require specific sequences of V4L2 API calls
> to be followed. These include capability enumeration, initialization,
> decoding, seek, pause, dynamic resolution change, drain and end of
> stream.
> 
> Specifics of the above have been discussed during Media Workshops at
> LinuxCon Europe 2012 in Barcelona and then later Embedded Linux
> Conference Europe 2014 in Düsseldorf. The de facto Codec API that
> originated at those events was later implemented by the drivers we already
> have merged in mainline, such as s5p-mfc or coda.
> 
> The only thing missing was the real specification included as a part of
> Linux Media documentation. Fix it now and document the decoder part of
> the Codec API.
> 
> Signed-off-by: Tomasz Figa <[e-mail]>
> ---
>  Documentation/media/uapi/v4l/dev-decoder.rst  | 1082 +++++++++++++++++
>  Documentation/media/uapi/v4l/devices.rst      |    1 +
>  Documentation/media/uapi/v4l/pixfmt-v4l2.rst  |    5 +
>  Documentation/media/uapi/v4l/v4l2.rst         |   10 +-
>  .../media/uapi/v4l/vidioc-decoder-cmd.rst     |   40 +-
>  Documentation/media/uapi/v4l/vidioc-g-fmt.rst |   14 +
>  6 files changed, 1137 insertions(+), 15 deletions(-)
>  create mode 100644 Documentation/media/uapi/v4l/dev-decoder.rst
> 
> diff --git a/Documentation/media/uapi/v4l/dev-decoder.rst b/Documentation/media/uapi/v4l/dev-decoder.rst
> new file mode 100644
> index 000000000000..09c7a6621b8e
> --- /dev/null
> +++ b/Documentation/media/uapi/v4l/dev-decoder.rst


> +State machine
> +=============
> +
> +.. kernel-render:: DOT
> +   :alt: DOT digraph of decoder state machine
> +   :caption: Decoder state machine
> +
> +   digraph decoder_state_machine {
> +       node [shape = doublecircle, label="Decoding"] Decoding;
> +
> +       node [shape = circle, label="Initialization"] Initialization;
> +       node [shape = circle, label="Capture\nsetup"] CaptureSetup;
> +       node [shape = circle, label="Dynamic\nresolution\nchange"] ResChange;
> +       node [shape = circle, label="Stopped"] Stopped;
> +       node [shape = circle, label="Drain"] Drain;
> +       node [shape = circle, label="Seek"] Seek;
> +       node [shape = circle, label="End of stream"] EoS;
> +
> +       node [shape = point]; qi
> +       qi -> Initialization [ label = "open()" ];
> +
> +       Initialization -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
> +
> +       CaptureSetup -> Stopped [ label = "CAPTURE\nbuffers\nready" ];
> +
> +       Decoding -> ResChange [ label = "Stream\nresolution\nchange" ];
> +       Decoding -> Drain [ label = "V4L2_DEC_CMD_STOP" ];
> +       Decoding -> EoS [ label = "EoS mark\nin the stream" ];
> +       Decoding -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +       Decoding -> Stopped [ label = "VIDIOC_STREAMOFF(CAPTURE)" ];
> +       Decoding -> Decoding;
> +
> +       ResChange -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
> +       ResChange -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +
> +       EoS -> Drain [ label = "Implicit\ndrain" ];
> +
> +       Drain -> Stopped [ label = "All CAPTURE\nbuffers dequeued\nor\nVIDIOC_STREAMOFF(CAPTURE)" ];
> +       Drain -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +
> +       Seek -> Decoding [ label = "VIDIOC_STREAMON(OUTPUT)" ];
> +       Seek -> Initialization [ label = "VIDIOC_REQBUFS(OUTPUT, 0)" ];

Shouldn't this be [ label = "VIDIOC_STREAMOFF(CAPTURE)" ], for me it is
looks more natural for v4l2?

For example I want to exit immediately from decoding state with calls to
streamoff(OUTPUT) and streamoff(CAPTURE). This could be when you press
ctrl-c while playing video, in this case I don't expect EoS nor buffers
draining.

> +
> +       Stopped -> Decoding [ label = "V4L2_DEC_CMD_START\nor\nVIDIOC_STREAMON(CAPTURE)" ];
> +       Stopped -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +   }
> +

Hans Verkuil Nov. 15, 2018, 2:34 p.m. UTC | #6

On 10/22/2018 04:48 PM, Tomasz Figa wrote:
> Due to complexity of the video decoding process, the V4L2 drivers of
> stateful decoder hardware require specific sequences of V4L2 API calls
> to be followed. These include capability enumeration, initialization,
> decoding, seek, pause, dynamic resolution change, drain and end of
> stream.
> 
> Specifics of the above have been discussed during Media Workshops at
> LinuxCon Europe 2012 in Barcelona and then later Embedded Linux
> Conference Europe 2014 in Düsseldorf. The de facto Codec API that
> originated at those events was later implemented by the drivers we already
> have merged in mainline, such as s5p-mfc or coda.
> 
> The only thing missing was the real specification included as a part of
> Linux Media documentation. Fix it now and document the decoder part of
> the Codec API.
> 
> Signed-off-by: Tomasz Figa <[e-mail]>
> ---
>  Documentation/media/uapi/v4l/dev-decoder.rst  | 1082 +++++++++++++++++
>  Documentation/media/uapi/v4l/devices.rst      |    1 +
>  Documentation/media/uapi/v4l/pixfmt-v4l2.rst  |    5 +
>  Documentation/media/uapi/v4l/v4l2.rst         |   10 +-
>  .../media/uapi/v4l/vidioc-decoder-cmd.rst     |   40 +-
>  Documentation/media/uapi/v4l/vidioc-g-fmt.rst |   14 +
>  6 files changed, 1137 insertions(+), 15 deletions(-)
>  create mode 100644 Documentation/media/uapi/v4l/dev-decoder.rst
> 
> diff --git a/Documentation/media/uapi/v4l/dev-decoder.rst b/Documentation/media/uapi/v4l/dev-decoder.rst
> new file mode 100644
> index 000000000000..09c7a6621b8e
> --- /dev/null
> +++ b/Documentation/media/uapi/v4l/dev-decoder.rst
> @@ -0,0 +1,1082 @@
> +.. -*- coding: utf-8; mode: rst -*-
> +
> +.. _decoder:
> +
> +*************************************************
> +Memory-to-memory Stateful Video Decoder Interface
> +*************************************************
> +
> +A stateful video decoder takes complete chunks of the bitstream (e.g. Annex-B
> +H.264/HEVC stream, raw VP8/9 stream) and decodes them into raw video frames in
> +display order. The decoder is expected not to require any additional information
> +from the client to process these buffers.
> +
> +Performing software parsing, processing etc. of the stream in the driver in
> +order to support this interface is strongly discouraged. In case such
> +operations are needed, use of the Stateless Video Decoder Interface (in
> +development) is strongly advised.
> +
> +Conventions and notation used in this document
> +==============================================
> +
> +1. The general V4L2 API rules apply if not specified in this document
> +   otherwise.
> +
> +2. The meaning of words “must”, “may”, “should”, etc. is as per RFC
> +   2119.
> +
> +3. All steps not marked “optional” are required.
> +
> +4. :c:func:`VIDIOC_G_EXT_CTRLS`, :c:func:`VIDIOC_S_EXT_CTRLS` may be used
> +   interchangeably with :c:func:`VIDIOC_G_CTRL`, :c:func:`VIDIOC_S_CTRL`,
> +   unless specified otherwise.
> +
> +5. Single-plane API (see spec) and applicable structures may be used
> +   interchangeably with Multi-plane API, unless specified otherwise,
> +   depending on decoder capabilities and following the general V4L2
> +   guidelines.
> +
> +6. i = [a..b]: sequence of integers from a to b, inclusive, i.e. i =
> +   [0..2]: i = 0, 1, 2.
> +
> +7. Given an ``OUTPUT`` buffer A, A’ represents a buffer on the ``CAPTURE``
> +   queue containing data (decoded frame/stream) that resulted from processing
> +   buffer A.
> +
> +.. _decoder-glossary:
> +
> +Glossary
> +========
> +
> +CAPTURE
> +   the destination buffer queue; for decoder, the queue of buffers containing
> +   decoded frames; for encoder, the queue of buffers containing encoded
> +   bitstream; ``V4L2_BUF_TYPE_VIDEO_CAPTURE```` or
> +   ``V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE``; data are captured from the hardware
> +   into ``CAPTURE`` buffers
> +
> +client
> +   application client communicating with the decoder or encoder implementing
> +   this interface
> +
> +coded format
> +   encoded/compressed video bitstream format (e.g. H.264, VP8, etc.); see
> +   also: raw format
> +
> +coded height
> +   height for given coded resolution
> +
> +coded resolution
> +   stream resolution in pixels aligned to codec and hardware requirements;
> +   typically visible resolution rounded up to full macroblocks;
> +   see also: visible resolution
> +
> +coded width
> +   width for given coded resolution
> +
> +decode order
> +   the order in which frames are decoded; may differ from display order if the
> +   coded format includes a feature of frame reordering; for decoders,
> +   ``OUTPUT`` buffers must be queued by the client in decode order; for
> +   encoders ``CAPTURE`` buffers must be returned by the encoder in decode order
> +
> +destination
> +   data resulting from the decode process; ``CAPTURE``
> +
> +display order
> +   the order in which frames must be displayed; for encoders, ``OUTPUT``
> +   buffers must be queued by the client in display order; for decoders,
> +   ``CAPTURE`` buffers must be returned by the decoder in display order
> +
> +DPB
> +   Decoded Picture Buffer; an H.264 term for a buffer that stores a decoded
> +   raw frame available for reference in further decoding steps.
> +
> +EOS
> +   end of stream
> +
> +IDR
> +   Instantaneous Decoder Refresh; a type of a keyframe in H.264-encoded stream,
> +   which clears the list of earlier reference frames (DPBs)
> +
> +keyframe
> +   an encoded frame that does not reference frames decoded earlier, i.e.
> +   can be decoded fully on its own.
> +
> +macroblock
> +   a processing unit in image and video compression formats based on linear
> +   block transforms (e.g. H.264, VP8, VP9); codec-specific, but for most of
> +   popular codecs the size is 16x16 samples (pixels)
> +
> +OUTPUT
> +   the source buffer queue; for decoders, the queue of buffers containing
> +   encoded bitstream; for encoders, the queue of buffers containing raw frames;
> +   ``V4L2_BUF_TYPE_VIDEO_OUTPUT`` or ``V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE``; the
> +   hardware is fed with data from ``OUTPUT`` buffers
> +
> +PPS
> +   Picture Parameter Set; a type of metadata entity in H.264 bitstream
> +
> +raw format
> +   uncompressed format containing raw pixel data (e.g. YUV, RGB formats)
> +
> +resume point
> +   a point in the bitstream from which decoding may start/continue, without
> +   any previous state/data present, e.g.: a keyframe (VP8/VP9) or
> +   SPS/PPS/IDR sequence (H.264); a resume point is required to start decode
> +   of a new stream, or to resume decoding after a seek
> +
> +source
> +   data fed to the decoder or encoder; ``OUTPUT``
> +
> +source height
> +   height in pixels for given source resolution; relevant to encoders only
> +
> +source resolution
> +   resolution in pixels of source frames being source to the encoder and
> +   subject to further cropping to the bounds of visible resolution; relevant to
> +   encoders only
> +
> +source width
> +   width in pixels for given source resolution; relevant to encoders only
> +
> +SPS
> +   Sequence Parameter Set; a type of metadata entity in H.264 bitstream
> +
> +stream metadata
> +   additional (non-visual) information contained inside encoded bitstream;
> +   for example: coded resolution, visible resolution, codec profile
> +
> +visible height
> +   height for given visible resolution; display height
> +
> +visible resolution
> +   stream resolution of the visible picture, in pixels, to be used for
> +   display purposes; must be smaller or equal to coded resolution;
> +   display resolution
> +
> +visible width
> +   width for given visible resolution; display width
> +
> +State machine
> +=============
> +
> +.. kernel-render:: DOT
> +   :alt: DOT digraph of decoder state machine
> +   :caption: Decoder state machine
> +
> +   digraph decoder_state_machine {
> +       node [shape = doublecircle, label="Decoding"] Decoding;
> +
> +       node [shape = circle, label="Initialization"] Initialization;
> +       node [shape = circle, label="Capture\nsetup"] CaptureSetup;
> +       node [shape = circle, label="Dynamic\nresolution\nchange"] ResChange;
> +       node [shape = circle, label="Stopped"] Stopped;
> +       node [shape = circle, label="Drain"] Drain;
> +       node [shape = circle, label="Seek"] Seek;
> +       node [shape = circle, label="End of stream"] EoS;
> +
> +       node [shape = point]; qi
> +       qi -> Initialization [ label = "open()" ];
> +
> +       Initialization -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
> +
> +       CaptureSetup -> Stopped [ label = "CAPTURE\nbuffers\nready" ];
> +
> +       Decoding -> ResChange [ label = "Stream\nresolution\nchange" ];
> +       Decoding -> Drain [ label = "V4L2_DEC_CMD_STOP" ];
> +       Decoding -> EoS [ label = "EoS mark\nin the stream" ];
> +       Decoding -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +       Decoding -> Stopped [ label = "VIDIOC_STREAMOFF(CAPTURE)" ];
> +       Decoding -> Decoding;
> +
> +       ResChange -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
> +       ResChange -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +
> +       EoS -> Drain [ label = "Implicit\ndrain" ];
> +
> +       Drain -> Stopped [ label = "All CAPTURE\nbuffers dequeued\nor\nVIDIOC_STREAMOFF(CAPTURE)" ];
> +       Drain -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +
> +       Seek -> Decoding [ label = "VIDIOC_STREAMON(OUTPUT)" ];
> +       Seek -> Initialization [ label = "VIDIOC_REQBUFS(OUTPUT, 0)" ];
> +
> +       Stopped -> Decoding [ label = "V4L2_DEC_CMD_START\nor\nVIDIOC_STREAMON(CAPTURE)" ];
> +       Stopped -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> +   }
> +
> +Querying capabilities
> +=====================
> +
> +1. To enumerate the set of coded formats supported by the decoder, the
> +   client may call :c:func:`VIDIOC_ENUM_FMT` on ``OUTPUT``.
> +
> +   * The full set of supported formats will be returned, regardless of the
> +     format set on ``CAPTURE``.
> +
> +2. To enumerate the set of supported raw formats, the client may call
> +   :c:func:`VIDIOC_ENUM_FMT` on ``CAPTURE``.
> +
> +   * Only the formats supported for the format currently active on ``OUTPUT``
> +     will be returned.
> +
> +   * In order to enumerate raw formats supported by a given coded format,
> +     the client must first set that coded format on ``OUTPUT`` and then
> +     enumerate formats on ``CAPTURE``.
> +
> +3. The client may use :c:func:`VIDIOC_ENUM_FRAMESIZES` to detect supported
> +   resolutions for a given format, passing desired pixel format in
> +   :c:type:`v4l2_frmsizeenum` ``pixel_format``.
> +
> +   * Values returned by :c:func:`VIDIOC_ENUM_FRAMESIZES` for a coded pixel
> +     formats will include all possible coded resolutions supported by the
> +     decoder for given coded pixel format.
> +
> +   * Values returned by :c:func:`VIDIOC_ENUM_FRAMESIZES` for a raw pixel format
> +     will include all possible frame buffer resolutions supported by the
> +     decoder for given raw pixel format and the coded format currently set on
> +     ``OUTPUT``.
> +
> +4. Supported profiles and levels for given format, if applicable, may be
> +   queried using their respective controls via :c:func:`VIDIOC_QUERYCTRL`.
> +
> +Initialization
> +==============
> +
> +1. **Optional.** Enumerate supported ``OUTPUT`` formats and resolutions. See
> +   `Querying capabilities` above.
> +
> +2. Set the coded format on ``OUTPUT`` via :c:func:`VIDIOC_S_FMT`
> +
> +   * **Required fields:**
> +
> +     ``type``
> +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> +
> +     ``pixelformat``
> +         a coded pixel format
> +
> +     ``width``, ``height``
> +         required only if cannot be parsed from the stream for the given
> +         coded format; optional otherwise - set to zero to ignore
> +
> +     ``sizeimage``
> +         desired size of ``OUTPUT`` buffers; the decoder may adjust it to
> +         match hardware requirements
> +
> +     other fields
> +         follow standard semantics
> +
> +   * **Return fields:**
> +
> +     ``sizeimage``
> +         adjusted size of ``CAPTURE`` buffers
> +
> +   * If width and height are set to non-zero values, the ``CAPTURE`` format
> +     will be updated with an appropriate frame buffer resolution instantly.
> +     However, for coded formats that include stream resolution information,
> +     after the decoder is done parsing the information from the stream, it will
> +     update the ``CAPTURE`` format with new values and signal a source change
> +     event.
> +
> +   .. warning::
> +
> +      Changing the ``OUTPUT`` format may change the currently set ``CAPTURE``
> +      format. The decoder will derive a new ``CAPTURE`` format from the
> +      ``OUTPUT`` format being set, including resolution, colorimetry
> +      parameters, etc. If the client needs a specific ``CAPTURE`` format, it
> +      must adjust it afterwards.
> +
> +3.  **Optional.** Query the minimum number of buffers required for ``OUTPUT``
> +    queue via :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to
> +    use more buffers than the minimum required by hardware/format.
> +
> +    * **Required fields:**
> +
> +      ``id``
> +          set to ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
> +
> +    * **Return fields:**
> +
> +      ``value``
> +          the minimum number of ``OUTPUT`` buffers required for the currently
> +          set format
> +
> +4.  Allocate source (bitstream) buffers via :c:func:`VIDIOC_REQBUFS` on
> +    ``OUTPUT``.
> +
> +    * **Required fields:**
> +
> +      ``count``
> +          requested number of buffers to allocate; greater than zero
> +
> +      ``type``
> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> +
> +      ``memory``
> +          follows standard semantics
> +
> +    * **Return fields:**
> +
> +      ``count``
> +          the actual number of buffers allocated
> +
> +    .. warning::
> +
> +       The actual number of allocated buffers may differ from the ``count``
> +       given. The client must check the updated value of ``count`` after the
> +       call returns.
> +
> +    .. note::
> +
> +       To allocate more than the minimum number of buffers (for pipeline
> +       depth), the client may query the ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
> +       control to get the minimum number of buffers required by the
> +       decoder/format, and pass the obtained value plus the number of
> +       additional buffers needed in the ``count`` field to
> +       :c:func:`VIDIOC_REQBUFS`.
> +
> +    Alternatively, :c:func:`VIDIOC_CREATE_BUFS` on the ``OUTPUT`` queue can be
> +    used to have more control over buffer allocation.
> +
> +    * **Required fields:**
> +
> +      ``count``
> +          requested number of buffers to allocate; greater than zero
> +
> +      ``type``
> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> +
> +      ``memory``
> +          follows standard semantics
> +
> +      ``format``
> +          follows standard semantics
> +
> +    * **Return fields:**
> +
> +      ``count``
> +          adjusted to the number of allocated buffers
> +
> +    .. warning::
> +
> +       The actual number of allocated buffers may differ from the ``count``
> +       given. The client must check the updated value of ``count`` after the
> +       call returns.
> +
> +5.  Start streaming on the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`.
> +
> +6.  **This step only applies to coded formats that contain resolution information
> +    in the stream.**

As far as I know all codecs have resolution/metadata in the stream.

As discussed in the "[PATCH vicodec v4 0/3] Add support to more pixel formats in
vicodec" thread, it is easiest to assume that there is always metadata.

Perhaps there should be a single mention somewhere that such codecs are not
supported at the moment, but to be frank how can you decode a stream without
it containing such essential information? You are much more likely to implement
such a codec as a stateless codec.

So I would just drop this sentence here (and perhaps at other places in this
document or the encoder document as well).

Regards,

	Hans

Nicolas Dufresne Nov. 17, 2018, 4:31 a.m. UTC | #7

Le jeudi 15 novembre 2018 à 15:34 +0100, Hans Verkuil a écrit :
> On 10/22/2018 04:48 PM, Tomasz Figa wrote:
> > Due to complexity of the video decoding process, the V4L2 drivers of
> > stateful decoder hardware require specific sequences of V4L2 API calls
> > to be followed. These include capability enumeration, initialization,
> > decoding, seek, pause, dynamic resolution change, drain and end of
> > stream.
> > 
> > Specifics of the above have been discussed during Media Workshops at
> > LinuxCon Europe 2012 in Barcelona and then later Embedded Linux
> > Conference Europe 2014 in Düsseldorf. The de facto Codec API that
> > originated at those events was later implemented by the drivers we already
> > have merged in mainline, such as s5p-mfc or coda.
> > 
> > The only thing missing was the real specification included as a part of
> > Linux Media documentation. Fix it now and document the decoder part of
> > the Codec API.
> > 
> > Signed-off-by: Tomasz Figa <[e-mail]>
> > ---
> >  Documentation/media/uapi/v4l/dev-decoder.rst  | 1082 +++++++++++++++++
> >  Documentation/media/uapi/v4l/devices.rst      |    1 +
> >  Documentation/media/uapi/v4l/pixfmt-v4l2.rst  |    5 +
> >  Documentation/media/uapi/v4l/v4l2.rst         |   10 +-
> >  .../media/uapi/v4l/vidioc-decoder-cmd.rst     |   40 +-
> >  Documentation/media/uapi/v4l/vidioc-g-fmt.rst |   14 +
> >  6 files changed, 1137 insertions(+), 15 deletions(-)
> >  create mode 100644 Documentation/media/uapi/v4l/dev-decoder.rst
> > 
> > diff --git a/Documentation/media/uapi/v4l/dev-decoder.rst b/Documentation/media/uapi/v4l/dev-decoder.rst
> > new file mode 100644
> > index 000000000000..09c7a6621b8e
> > --- /dev/null
> > +++ b/Documentation/media/uapi/v4l/dev-decoder.rst
> > @@ -0,0 +1,1082 @@
> > +.. -*- coding: utf-8; mode: rst -*-
> > +
> > +.. _decoder:
> > +
> > +*************************************************
> > +Memory-to-memory Stateful Video Decoder Interface
> > +*************************************************
> > +
> > +A stateful video decoder takes complete chunks of the bitstream (e.g. Annex-B
> > +H.264/HEVC stream, raw VP8/9 stream) and decodes them into raw video frames in
> > +display order. The decoder is expected not to require any additional information
> > +from the client to process these buffers.
> > +
> > +Performing software parsing, processing etc. of the stream in the driver in
> > +order to support this interface is strongly discouraged. In case such
> > +operations are needed, use of the Stateless Video Decoder Interface (in
> > +development) is strongly advised.
> > +
> > +Conventions and notation used in this document
> > +==============================================
> > +
> > +1. The general V4L2 API rules apply if not specified in this document
> > +   otherwise.
> > +
> > +2. The meaning of words “must”, “may”, “should”, etc. is as per RFC
> > +   2119.
> > +
> > +3. All steps not marked “optional” are required.
> > +
> > +4. :c:func:`VIDIOC_G_EXT_CTRLS`, :c:func:`VIDIOC_S_EXT_CTRLS` may be used
> > +   interchangeably with :c:func:`VIDIOC_G_CTRL`, :c:func:`VIDIOC_S_CTRL`,
> > +   unless specified otherwise.
> > +
> > +5. Single-plane API (see spec) and applicable structures may be used
> > +   interchangeably with Multi-plane API, unless specified otherwise,
> > +   depending on decoder capabilities and following the general V4L2
> > +   guidelines.
> > +
> > +6. i = [a..b]: sequence of integers from a to b, inclusive, i.e. i =
> > +   [0..2]: i = 0, 1, 2.
> > +
> > +7. Given an ``OUTPUT`` buffer A, A’ represents a buffer on the ``CAPTURE``
> > +   queue containing data (decoded frame/stream) that resulted from processing
> > +   buffer A.
> > +
> > +.. _decoder-glossary:
> > +
> > +Glossary
> > +========
> > +
> > +CAPTURE
> > +   the destination buffer queue; for decoder, the queue of buffers containing
> > +   decoded frames; for encoder, the queue of buffers containing encoded
> > +   bitstream; ``V4L2_BUF_TYPE_VIDEO_CAPTURE```` or
> > +   ``V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE``; data are captured from the hardware
> > +   into ``CAPTURE`` buffers
> > +
> > +client
> > +   application client communicating with the decoder or encoder implementing
> > +   this interface
> > +
> > +coded format
> > +   encoded/compressed video bitstream format (e.g. H.264, VP8, etc.); see
> > +   also: raw format
> > +
> > +coded height
> > +   height for given coded resolution
> > +
> > +coded resolution
> > +   stream resolution in pixels aligned to codec and hardware requirements;
> > +   typically visible resolution rounded up to full macroblocks;
> > +   see also: visible resolution
> > +
> > +coded width
> > +   width for given coded resolution
> > +
> > +decode order
> > +   the order in which frames are decoded; may differ from display order if the
> > +   coded format includes a feature of frame reordering; for decoders,
> > +   ``OUTPUT`` buffers must be queued by the client in decode order; for
> > +   encoders ``CAPTURE`` buffers must be returned by the encoder in decode order
> > +
> > +destination
> > +   data resulting from the decode process; ``CAPTURE``
> > +
> > +display order
> > +   the order in which frames must be displayed; for encoders, ``OUTPUT``
> > +   buffers must be queued by the client in display order; for decoders,
> > +   ``CAPTURE`` buffers must be returned by the decoder in display order
> > +
> > +DPB
> > +   Decoded Picture Buffer; an H.264 term for a buffer that stores a decoded
> > +   raw frame available for reference in further decoding steps.
> > +
> > +EOS
> > +   end of stream
> > +
> > +IDR
> > +   Instantaneous Decoder Refresh; a type of a keyframe in H.264-encoded stream,
> > +   which clears the list of earlier reference frames (DPBs)
> > +
> > +keyframe
> > +   an encoded frame that does not reference frames decoded earlier, i.e.
> > +   can be decoded fully on its own.
> > +
> > +macroblock
> > +   a processing unit in image and video compression formats based on linear
> > +   block transforms (e.g. H.264, VP8, VP9); codec-specific, but for most of
> > +   popular codecs the size is 16x16 samples (pixels)
> > +
> > +OUTPUT
> > +   the source buffer queue; for decoders, the queue of buffers containing
> > +   encoded bitstream; for encoders, the queue of buffers containing raw frames;
> > +   ``V4L2_BUF_TYPE_VIDEO_OUTPUT`` or ``V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE``; the
> > +   hardware is fed with data from ``OUTPUT`` buffers
> > +
> > +PPS
> > +   Picture Parameter Set; a type of metadata entity in H.264 bitstream
> > +
> > +raw format
> > +   uncompressed format containing raw pixel data (e.g. YUV, RGB formats)
> > +
> > +resume point
> > +   a point in the bitstream from which decoding may start/continue, without
> > +   any previous state/data present, e.g.: a keyframe (VP8/VP9) or
> > +   SPS/PPS/IDR sequence (H.264); a resume point is required to start decode
> > +   of a new stream, or to resume decoding after a seek
> > +
> > +source
> > +   data fed to the decoder or encoder; ``OUTPUT``
> > +
> > +source height
> > +   height in pixels for given source resolution; relevant to encoders only
> > +
> > +source resolution
> > +   resolution in pixels of source frames being source to the encoder and
> > +   subject to further cropping to the bounds of visible resolution; relevant to
> > +   encoders only
> > +
> > +source width
> > +   width in pixels for given source resolution; relevant to encoders only
> > +
> > +SPS
> > +   Sequence Parameter Set; a type of metadata entity in H.264 bitstream
> > +
> > +stream metadata
> > +   additional (non-visual) information contained inside encoded bitstream;
> > +   for example: coded resolution, visible resolution, codec profile
> > +
> > +visible height
> > +   height for given visible resolution; display height
> > +
> > +visible resolution
> > +   stream resolution of the visible picture, in pixels, to be used for
> > +   display purposes; must be smaller or equal to coded resolution;
> > +   display resolution
> > +
> > +visible width
> > +   width for given visible resolution; display width
> > +
> > +State machine
> > +=============
> > +
> > +.. kernel-render:: DOT
> > +   :alt: DOT digraph of decoder state machine
> > +   :caption: Decoder state machine
> > +
> > +   digraph decoder_state_machine {
> > +       node [shape = doublecircle, label="Decoding"] Decoding;
> > +
> > +       node [shape = circle, label="Initialization"] Initialization;
> > +       node [shape = circle, label="Capture\nsetup"] CaptureSetup;
> > +       node [shape = circle, label="Dynamic\nresolution\nchange"] ResChange;
> > +       node [shape = circle, label="Stopped"] Stopped;
> > +       node [shape = circle, label="Drain"] Drain;
> > +       node [shape = circle, label="Seek"] Seek;
> > +       node [shape = circle, label="End of stream"] EoS;
> > +
> > +       node [shape = point]; qi
> > +       qi -> Initialization [ label = "open()" ];
> > +
> > +       Initialization -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
> > +
> > +       CaptureSetup -> Stopped [ label = "CAPTURE\nbuffers\nready" ];
> > +
> > +       Decoding -> ResChange [ label = "Stream\nresolution\nchange" ];
> > +       Decoding -> Drain [ label = "V4L2_DEC_CMD_STOP" ];
> > +       Decoding -> EoS [ label = "EoS mark\nin the stream" ];
> > +       Decoding -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> > +       Decoding -> Stopped [ label = "VIDIOC_STREAMOFF(CAPTURE)" ];
> > +       Decoding -> Decoding;
> > +
> > +       ResChange -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
> > +       ResChange -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> > +
> > +       EoS -> Drain [ label = "Implicit\ndrain" ];
> > +
> > +       Drain -> Stopped [ label = "All CAPTURE\nbuffers dequeued\nor\nVIDIOC_STREAMOFF(CAPTURE)" ];
> > +       Drain -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> > +
> > +       Seek -> Decoding [ label = "VIDIOC_STREAMON(OUTPUT)" ];
> > +       Seek -> Initialization [ label = "VIDIOC_REQBUFS(OUTPUT, 0)" ];
> > +
> > +       Stopped -> Decoding [ label = "V4L2_DEC_CMD_START\nor\nVIDIOC_STREAMON(CAPTURE)" ];
> > +       Stopped -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
> > +   }
> > +
> > +Querying capabilities
> > +=====================
> > +
> > +1. To enumerate the set of coded formats supported by the decoder, the
> > +   client may call :c:func:`VIDIOC_ENUM_FMT` on ``OUTPUT``.
> > +
> > +   * The full set of supported formats will be returned, regardless of the
> > +     format set on ``CAPTURE``.
> > +
> > +2. To enumerate the set of supported raw formats, the client may call
> > +   :c:func:`VIDIOC_ENUM_FMT` on ``CAPTURE``.
> > +
> > +   * Only the formats supported for the format currently active on ``OUTPUT``
> > +     will be returned.
> > +
> > +   * In order to enumerate raw formats supported by a given coded format,
> > +     the client must first set that coded format on ``OUTPUT`` and then
> > +     enumerate formats on ``CAPTURE``.
> > +
> > +3. The client may use :c:func:`VIDIOC_ENUM_FRAMESIZES` to detect supported
> > +   resolutions for a given format, passing desired pixel format in
> > +   :c:type:`v4l2_frmsizeenum` ``pixel_format``.
> > +
> > +   * Values returned by :c:func:`VIDIOC_ENUM_FRAMESIZES` for a coded pixel
> > +     formats will include all possible coded resolutions supported by the
> > +     decoder for given coded pixel format.
> > +
> > +   * Values returned by :c:func:`VIDIOC_ENUM_FRAMESIZES` for a raw pixel format
> > +     will include all possible frame buffer resolutions supported by the
> > +     decoder for given raw pixel format and the coded format currently set on
> > +     ``OUTPUT``.
> > +
> > +4. Supported profiles and levels for given format, if applicable, may be
> > +   queried using their respective controls via :c:func:`VIDIOC_QUERYCTRL`.
> > +
> > +Initialization
> > +==============
> > +
> > +1. **Optional.** Enumerate supported ``OUTPUT`` formats and resolutions. See
> > +   `Querying capabilities` above.
> > +
> > +2. Set the coded format on ``OUTPUT`` via :c:func:`VIDIOC_S_FMT`
> > +
> > +   * **Required fields:**
> > +
> > +     ``type``
> > +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> > +
> > +     ``pixelformat``
> > +         a coded pixel format
> > +
> > +     ``width``, ``height``
> > +         required only if cannot be parsed from the stream for the given
> > +         coded format; optional otherwise - set to zero to ignore
> > +
> > +     ``sizeimage``
> > +         desired size of ``OUTPUT`` buffers; the decoder may adjust it to
> > +         match hardware requirements
> > +
> > +     other fields
> > +         follow standard semantics
> > +
> > +   * **Return fields:**
> > +
> > +     ``sizeimage``
> > +         adjusted size of ``CAPTURE`` buffers
> > +
> > +   * If width and height are set to non-zero values, the ``CAPTURE`` format
> > +     will be updated with an appropriate frame buffer resolution instantly.
> > +     However, for coded formats that include stream resolution information,
> > +     after the decoder is done parsing the information from the stream, it will
> > +     update the ``CAPTURE`` format with new values and signal a source change
> > +     event.
> > +
> > +   .. warning::
> > +
> > +      Changing the ``OUTPUT`` format may change the currently set ``CAPTURE``
> > +      format. The decoder will derive a new ``CAPTURE`` format from the
> > +      ``OUTPUT`` format being set, including resolution, colorimetry
> > +      parameters, etc. If the client needs a specific ``CAPTURE`` format, it
> > +      must adjust it afterwards.
> > +
> > +3.  **Optional.** Query the minimum number of buffers required for ``OUTPUT``
> > +    queue via :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to
> > +    use more buffers than the minimum required by hardware/format.
> > +
> > +    * **Required fields:**
> > +
> > +      ``id``
> > +          set to ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
> > +
> > +    * **Return fields:**
> > +
> > +      ``value``
> > +          the minimum number of ``OUTPUT`` buffers required for the currently
> > +          set format
> > +
> > +4.  Allocate source (bitstream) buffers via :c:func:`VIDIOC_REQBUFS` on
> > +    ``OUTPUT``.
> > +
> > +    * **Required fields:**
> > +
> > +      ``count``
> > +          requested number of buffers to allocate; greater than zero
> > +
> > +      ``type``
> > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> > +
> > +      ``memory``
> > +          follows standard semantics
> > +
> > +    * **Return fields:**
> > +
> > +      ``count``
> > +          the actual number of buffers allocated
> > +
> > +    .. warning::
> > +
> > +       The actual number of allocated buffers may differ from the ``count``
> > +       given. The client must check the updated value of ``count`` after the
> > +       call returns.
> > +
> > +    .. note::
> > +
> > +       To allocate more than the minimum number of buffers (for pipeline
> > +       depth), the client may query the ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
> > +       control to get the minimum number of buffers required by the
> > +       decoder/format, and pass the obtained value plus the number of
> > +       additional buffers needed in the ``count`` field to
> > +       :c:func:`VIDIOC_REQBUFS`.
> > +
> > +    Alternatively, :c:func:`VIDIOC_CREATE_BUFS` on the ``OUTPUT`` queue can be
> > +    used to have more control over buffer allocation.
> > +
> > +    * **Required fields:**
> > +
> > +      ``count``
> > +          requested number of buffers to allocate; greater than zero
> > +
> > +      ``type``
> > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> > +
> > +      ``memory``
> > +          follows standard semantics
> > +
> > +      ``format``
> > +          follows standard semantics
> > +
> > +    * **Return fields:**
> > +
> > +      ``count``
> > +          adjusted to the number of allocated buffers
> > +
> > +    .. warning::
> > +
> > +       The actual number of allocated buffers may differ from the ``count``
> > +       given. The client must check the updated value of ``count`` after the
> > +       call returns.
> > +
> > +5.  Start streaming on the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`.
> > +
> > +6.  **This step only applies to coded formats that contain resolution information
> > +    in the stream.**
> 
> As far as I know all codecs have resolution/metadata in the stream.

Was this comment about what we currently support in V4L2 interface ? In
real life, there is CODEC that works only with out-of-band codec data.
A well known one is AVC1 (and HVC1). In this mode, the AVC H264 does
not have start code, and the headers are not allowed in the bitstream
itself. This format is much more efficient to process then AVC Annex B,
since you can just read the NAL size and jump over instead of scanning
for start code. This is the format used in the very popular ISOMP4
container.

The other sign that these codecs do exist is the recurence of the
notion of codec data in pretty much every codec abstraction there is
(ffmpeg, GStreamer, Android Media Codec)
> 
> As discussed in the "[PATCH vicodec v4 0/3] Add support to more pixel formats in
> vicodec" thread, it is easiest to assume that there is always metadata.
> 
> Perhaps there should be a single mention somewhere that such codecs are not
> supported at the moment, but to be frank how can you decode a stream without
> it containing such essential information? You are much more likely to implement
> such a codec as a stateless codec.

That is I believe a miss-interpretation of what a stateless codec is.
It's not because you have to set one blob of CODEC data after S_FMT on
a specific control that this CODEC becomes stateless. The fact is not
supported now is just become we didn't have come accross HW that
supports these.

FFMPEG offers a state full software codec, and you still have this
codec_data blob for many of the format. They also only supports AVC1,
they parser will always convert the stream to that, because it's just
more efficient format. Android Media Codec works similarly. What keeps
them stateful, is that you don't need to parse it, you don't even need
to know what these data contains. They are blobs placed in the
container that you pass as-is to the decoder.

> 
> So I would just drop this sentence here (and perhaps at other places in this
> document or the encoder document as well).
> 
> Regards,
> 
> 	Hans

Hans Verkuil Nov. 17, 2018, 11:43 a.m. UTC | #8

On 11/17/2018 05:31 AM, Nicolas Dufresne wrote:
> Le jeudi 15 novembre 2018 à 15:34 +0100, Hans Verkuil a écrit :
>> On 10/22/2018 04:48 PM, Tomasz Figa wrote:
>>> Due to complexity of the video decoding process, the V4L2 drivers of
>>> stateful decoder hardware require specific sequences of V4L2 API calls
>>> to be followed. These include capability enumeration, initialization,
>>> decoding, seek, pause, dynamic resolution change, drain and end of
>>> stream.
>>>
>>> Specifics of the above have been discussed during Media Workshops at
>>> LinuxCon Europe 2012 in Barcelona and then later Embedded Linux
>>> Conference Europe 2014 in Düsseldorf. The de facto Codec API that
>>> originated at those events was later implemented by the drivers we already
>>> have merged in mainline, such as s5p-mfc or coda.
>>>
>>> The only thing missing was the real specification included as a part of
>>> Linux Media documentation. Fix it now and document the decoder part of
>>> the Codec API.
>>>
>>> Signed-off-by: Tomasz Figa <[e-mail]>
>>> ---
>>>  Documentation/media/uapi/v4l/dev-decoder.rst  | 1082 +++++++++++++++++
>>>  Documentation/media/uapi/v4l/devices.rst      |    1 +
>>>  Documentation/media/uapi/v4l/pixfmt-v4l2.rst  |    5 +
>>>  Documentation/media/uapi/v4l/v4l2.rst         |   10 +-
>>>  .../media/uapi/v4l/vidioc-decoder-cmd.rst     |   40 +-
>>>  Documentation/media/uapi/v4l/vidioc-g-fmt.rst |   14 +
>>>  6 files changed, 1137 insertions(+), 15 deletions(-)
>>>  create mode 100644 Documentation/media/uapi/v4l/dev-decoder.rst
>>>
>>> diff --git a/Documentation/media/uapi/v4l/dev-decoder.rst b/Documentation/media/uapi/v4l/dev-decoder.rst
>>> new file mode 100644
>>> index 000000000000..09c7a6621b8e
>>> --- /dev/null
>>> +++ b/Documentation/media/uapi/v4l/dev-decoder.rst
>>> @@ -0,0 +1,1082 @@
>>> +.. -*- coding: utf-8; mode: rst -*-
>>> +
>>> +.. _decoder:
>>> +
>>> +*************************************************
>>> +Memory-to-memory Stateful Video Decoder Interface
>>> +*************************************************
>>> +
>>> +A stateful video decoder takes complete chunks of the bitstream (e.g. Annex-B
>>> +H.264/HEVC stream, raw VP8/9 stream) and decodes them into raw video frames in
>>> +display order. The decoder is expected not to require any additional information
>>> +from the client to process these buffers.
>>> +
>>> +Performing software parsing, processing etc. of the stream in the driver in
>>> +order to support this interface is strongly discouraged. In case such
>>> +operations are needed, use of the Stateless Video Decoder Interface (in
>>> +development) is strongly advised.
>>> +
>>> +Conventions and notation used in this document
>>> +==============================================
>>> +
>>> +1. The general V4L2 API rules apply if not specified in this document
>>> +   otherwise.
>>> +
>>> +2. The meaning of words “must”, “may”, “should”, etc. is as per RFC
>>> +   2119.
>>> +
>>> +3. All steps not marked “optional” are required.
>>> +
>>> +4. :c:func:`VIDIOC_G_EXT_CTRLS`, :c:func:`VIDIOC_S_EXT_CTRLS` may be used
>>> +   interchangeably with :c:func:`VIDIOC_G_CTRL`, :c:func:`VIDIOC_S_CTRL`,
>>> +   unless specified otherwise.
>>> +
>>> +5. Single-plane API (see spec) and applicable structures may be used
>>> +   interchangeably with Multi-plane API, unless specified otherwise,
>>> +   depending on decoder capabilities and following the general V4L2
>>> +   guidelines.
>>> +
>>> +6. i = [a..b]: sequence of integers from a to b, inclusive, i.e. i =
>>> +   [0..2]: i = 0, 1, 2.
>>> +
>>> +7. Given an ``OUTPUT`` buffer A, A’ represents a buffer on the ``CAPTURE``
>>> +   queue containing data (decoded frame/stream) that resulted from processing
>>> +   buffer A.
>>> +
>>> +.. _decoder-glossary:
>>> +
>>> +Glossary
>>> +========
>>> +
>>> +CAPTURE
>>> +   the destination buffer queue; for decoder, the queue of buffers containing
>>> +   decoded frames; for encoder, the queue of buffers containing encoded
>>> +   bitstream; ``V4L2_BUF_TYPE_VIDEO_CAPTURE```` or
>>> +   ``V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE``; data are captured from the hardware
>>> +   into ``CAPTURE`` buffers
>>> +
>>> +client
>>> +   application client communicating with the decoder or encoder implementing
>>> +   this interface
>>> +
>>> +coded format
>>> +   encoded/compressed video bitstream format (e.g. H.264, VP8, etc.); see
>>> +   also: raw format
>>> +
>>> +coded height
>>> +   height for given coded resolution
>>> +
>>> +coded resolution
>>> +   stream resolution in pixels aligned to codec and hardware requirements;
>>> +   typically visible resolution rounded up to full macroblocks;
>>> +   see also: visible resolution
>>> +
>>> +coded width
>>> +   width for given coded resolution
>>> +
>>> +decode order
>>> +   the order in which frames are decoded; may differ from display order if the
>>> +   coded format includes a feature of frame reordering; for decoders,
>>> +   ``OUTPUT`` buffers must be queued by the client in decode order; for
>>> +   encoders ``CAPTURE`` buffers must be returned by the encoder in decode order
>>> +
>>> +destination
>>> +   data resulting from the decode process; ``CAPTURE``
>>> +
>>> +display order
>>> +   the order in which frames must be displayed; for encoders, ``OUTPUT``
>>> +   buffers must be queued by the client in display order; for decoders,
>>> +   ``CAPTURE`` buffers must be returned by the decoder in display order
>>> +
>>> +DPB
>>> +   Decoded Picture Buffer; an H.264 term for a buffer that stores a decoded
>>> +   raw frame available for reference in further decoding steps.
>>> +
>>> +EOS
>>> +   end of stream
>>> +
>>> +IDR
>>> +   Instantaneous Decoder Refresh; a type of a keyframe in H.264-encoded stream,
>>> +   which clears the list of earlier reference frames (DPBs)
>>> +
>>> +keyframe
>>> +   an encoded frame that does not reference frames decoded earlier, i.e.
>>> +   can be decoded fully on its own.
>>> +
>>> +macroblock
>>> +   a processing unit in image and video compression formats based on linear
>>> +   block transforms (e.g. H.264, VP8, VP9); codec-specific, but for most of
>>> +   popular codecs the size is 16x16 samples (pixels)
>>> +
>>> +OUTPUT
>>> +   the source buffer queue; for decoders, the queue of buffers containing
>>> +   encoded bitstream; for encoders, the queue of buffers containing raw frames;
>>> +   ``V4L2_BUF_TYPE_VIDEO_OUTPUT`` or ``V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE``; the
>>> +   hardware is fed with data from ``OUTPUT`` buffers
>>> +
>>> +PPS
>>> +   Picture Parameter Set; a type of metadata entity in H.264 bitstream
>>> +
>>> +raw format
>>> +   uncompressed format containing raw pixel data (e.g. YUV, RGB formats)
>>> +
>>> +resume point
>>> +   a point in the bitstream from which decoding may start/continue, without
>>> +   any previous state/data present, e.g.: a keyframe (VP8/VP9) or
>>> +   SPS/PPS/IDR sequence (H.264); a resume point is required to start decode
>>> +   of a new stream, or to resume decoding after a seek
>>> +
>>> +source
>>> +   data fed to the decoder or encoder; ``OUTPUT``
>>> +
>>> +source height
>>> +   height in pixels for given source resolution; relevant to encoders only
>>> +
>>> +source resolution
>>> +   resolution in pixels of source frames being source to the encoder and
>>> +   subject to further cropping to the bounds of visible resolution; relevant to
>>> +   encoders only
>>> +
>>> +source width
>>> +   width in pixels for given source resolution; relevant to encoders only
>>> +
>>> +SPS
>>> +   Sequence Parameter Set; a type of metadata entity in H.264 bitstream
>>> +
>>> +stream metadata
>>> +   additional (non-visual) information contained inside encoded bitstream;
>>> +   for example: coded resolution, visible resolution, codec profile
>>> +
>>> +visible height
>>> +   height for given visible resolution; display height
>>> +
>>> +visible resolution
>>> +   stream resolution of the visible picture, in pixels, to be used for
>>> +   display purposes; must be smaller or equal to coded resolution;
>>> +   display resolution
>>> +
>>> +visible width
>>> +   width for given visible resolution; display width
>>> +
>>> +State machine
>>> +=============
>>> +
>>> +.. kernel-render:: DOT
>>> +   :alt: DOT digraph of decoder state machine
>>> +   :caption: Decoder state machine
>>> +
>>> +   digraph decoder_state_machine {
>>> +       node [shape = doublecircle, label="Decoding"] Decoding;
>>> +
>>> +       node [shape = circle, label="Initialization"] Initialization;
>>> +       node [shape = circle, label="Capture\nsetup"] CaptureSetup;
>>> +       node [shape = circle, label="Dynamic\nresolution\nchange"] ResChange;
>>> +       node [shape = circle, label="Stopped"] Stopped;
>>> +       node [shape = circle, label="Drain"] Drain;
>>> +       node [shape = circle, label="Seek"] Seek;
>>> +       node [shape = circle, label="End of stream"] EoS;
>>> +
>>> +       node [shape = point]; qi
>>> +       qi -> Initialization [ label = "open()" ];
>>> +
>>> +       Initialization -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
>>> +
>>> +       CaptureSetup -> Stopped [ label = "CAPTURE\nbuffers\nready" ];
>>> +
>>> +       Decoding -> ResChange [ label = "Stream\nresolution\nchange" ];
>>> +       Decoding -> Drain [ label = "V4L2_DEC_CMD_STOP" ];
>>> +       Decoding -> EoS [ label = "EoS mark\nin the stream" ];
>>> +       Decoding -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
>>> +       Decoding -> Stopped [ label = "VIDIOC_STREAMOFF(CAPTURE)" ];
>>> +       Decoding -> Decoding;
>>> +
>>> +       ResChange -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
>>> +       ResChange -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
>>> +
>>> +       EoS -> Drain [ label = "Implicit\ndrain" ];
>>> +
>>> +       Drain -> Stopped [ label = "All CAPTURE\nbuffers dequeued\nor\nVIDIOC_STREAMOFF(CAPTURE)" ];
>>> +       Drain -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
>>> +
>>> +       Seek -> Decoding [ label = "VIDIOC_STREAMON(OUTPUT)" ];
>>> +       Seek -> Initialization [ label = "VIDIOC_REQBUFS(OUTPUT, 0)" ];
>>> +
>>> +       Stopped -> Decoding [ label = "V4L2_DEC_CMD_START\nor\nVIDIOC_STREAMON(CAPTURE)" ];
>>> +       Stopped -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
>>> +   }
>>> +
>>> +Querying capabilities
>>> +=====================
>>> +
>>> +1. To enumerate the set of coded formats supported by the decoder, the
>>> +   client may call :c:func:`VIDIOC_ENUM_FMT` on ``OUTPUT``.
>>> +
>>> +   * The full set of supported formats will be returned, regardless of the
>>> +     format set on ``CAPTURE``.
>>> +
>>> +2. To enumerate the set of supported raw formats, the client may call
>>> +   :c:func:`VIDIOC_ENUM_FMT` on ``CAPTURE``.
>>> +
>>> +   * Only the formats supported for the format currently active on ``OUTPUT``
>>> +     will be returned.
>>> +
>>> +   * In order to enumerate raw formats supported by a given coded format,
>>> +     the client must first set that coded format on ``OUTPUT`` and then
>>> +     enumerate formats on ``CAPTURE``.
>>> +
>>> +3. The client may use :c:func:`VIDIOC_ENUM_FRAMESIZES` to detect supported
>>> +   resolutions for a given format, passing desired pixel format in
>>> +   :c:type:`v4l2_frmsizeenum` ``pixel_format``.
>>> +
>>> +   * Values returned by :c:func:`VIDIOC_ENUM_FRAMESIZES` for a coded pixel
>>> +     formats will include all possible coded resolutions supported by the
>>> +     decoder for given coded pixel format.
>>> +
>>> +   * Values returned by :c:func:`VIDIOC_ENUM_FRAMESIZES` for a raw pixel format
>>> +     will include all possible frame buffer resolutions supported by the
>>> +     decoder for given raw pixel format and the coded format currently set on
>>> +     ``OUTPUT``.
>>> +
>>> +4. Supported profiles and levels for given format, if applicable, may be
>>> +   queried using their respective controls via :c:func:`VIDIOC_QUERYCTRL`.
>>> +
>>> +Initialization
>>> +==============
>>> +
>>> +1. **Optional.** Enumerate supported ``OUTPUT`` formats and resolutions. See
>>> +   `Querying capabilities` above.
>>> +
>>> +2. Set the coded format on ``OUTPUT`` via :c:func:`VIDIOC_S_FMT`
>>> +
>>> +   * **Required fields:**
>>> +
>>> +     ``type``
>>> +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
>>> +
>>> +     ``pixelformat``
>>> +         a coded pixel format
>>> +
>>> +     ``width``, ``height``
>>> +         required only if cannot be parsed from the stream for the given
>>> +         coded format; optional otherwise - set to zero to ignore
>>> +
>>> +     ``sizeimage``
>>> +         desired size of ``OUTPUT`` buffers; the decoder may adjust it to
>>> +         match hardware requirements
>>> +
>>> +     other fields
>>> +         follow standard semantics
>>> +
>>> +   * **Return fields:**
>>> +
>>> +     ``sizeimage``
>>> +         adjusted size of ``CAPTURE`` buffers
>>> +
>>> +   * If width and height are set to non-zero values, the ``CAPTURE`` format
>>> +     will be updated with an appropriate frame buffer resolution instantly.
>>> +     However, for coded formats that include stream resolution information,
>>> +     after the decoder is done parsing the information from the stream, it will
>>> +     update the ``CAPTURE`` format with new values and signal a source change
>>> +     event.
>>> +
>>> +   .. warning::
>>> +
>>> +      Changing the ``OUTPUT`` format may change the currently set ``CAPTURE``
>>> +      format. The decoder will derive a new ``CAPTURE`` format from the
>>> +      ``OUTPUT`` format being set, including resolution, colorimetry
>>> +      parameters, etc. If the client needs a specific ``CAPTURE`` format, it
>>> +      must adjust it afterwards.
>>> +
>>> +3.  **Optional.** Query the minimum number of buffers required for ``OUTPUT``
>>> +    queue via :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to
>>> +    use more buffers than the minimum required by hardware/format.
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``id``
>>> +          set to ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
>>> +
>>> +    * **Return fields:**
>>> +
>>> +      ``value``
>>> +          the minimum number of ``OUTPUT`` buffers required for the currently
>>> +          set format
>>> +
>>> +4.  Allocate source (bitstream) buffers via :c:func:`VIDIOC_REQBUFS` on
>>> +    ``OUTPUT``.
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``count``
>>> +          requested number of buffers to allocate; greater than zero
>>> +
>>> +      ``type``
>>> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
>>> +
>>> +      ``memory``
>>> +          follows standard semantics
>>> +
>>> +    * **Return fields:**
>>> +
>>> +      ``count``
>>> +          the actual number of buffers allocated
>>> +
>>> +    .. warning::
>>> +
>>> +       The actual number of allocated buffers may differ from the ``count``
>>> +       given. The client must check the updated value of ``count`` after the
>>> +       call returns.
>>> +
>>> +    .. note::
>>> +
>>> +       To allocate more than the minimum number of buffers (for pipeline
>>> +       depth), the client may query the ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
>>> +       control to get the minimum number of buffers required by the
>>> +       decoder/format, and pass the obtained value plus the number of
>>> +       additional buffers needed in the ``count`` field to
>>> +       :c:func:`VIDIOC_REQBUFS`.
>>> +
>>> +    Alternatively, :c:func:`VIDIOC_CREATE_BUFS` on the ``OUTPUT`` queue can be
>>> +    used to have more control over buffer allocation.
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``count``
>>> +          requested number of buffers to allocate; greater than zero
>>> +
>>> +      ``type``
>>> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
>>> +
>>> +      ``memory``
>>> +          follows standard semantics
>>> +
>>> +      ``format``
>>> +          follows standard semantics
>>> +
>>> +    * **Return fields:**
>>> +
>>> +      ``count``
>>> +          adjusted to the number of allocated buffers
>>> +
>>> +    .. warning::
>>> +
>>> +       The actual number of allocated buffers may differ from the ``count``
>>> +       given. The client must check the updated value of ``count`` after the
>>> +       call returns.
>>> +
>>> +5.  Start streaming on the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`.
>>> +
>>> +6.  **This step only applies to coded formats that contain resolution information
>>> +    in the stream.**
>>
>> As far as I know all codecs have resolution/metadata in the stream.
> 
> Was this comment about what we currently support in V4L2 interface ? In

Yes, I was talking about all V4L2 codecs.

> real life, there is CODEC that works only with out-of-band codec data.
> A well known one is AVC1 (and HVC1). In this mode, the AVC H264 does
> not have start code, and the headers are not allowed in the bitstream
> itself. This format is much more efficient to process then AVC Annex B,
> since you can just read the NAL size and jump over instead of scanning
> for start code. This is the format used in the very popular ISOMP4
> container.

How would such a codec handle resolution changes? Or is that not allowed?

> 
> The other sign that these codecs do exist is the recurence of the
> notion of codec data in pretty much every codec abstraction there is
> (ffmpeg, GStreamer, Android Media Codec)
>>
>> As discussed in the "[PATCH vicodec v4 0/3] Add support to more pixel formats in
>> vicodec" thread, it is easiest to assume that there is always metadata.
>>
>> Perhaps there should be a single mention somewhere that such codecs are not
>> supported at the moment, but to be frank how can you decode a stream without
>> it containing such essential information? You are much more likely to implement
>> such a codec as a stateless codec.
> 
> That is I believe a miss-interpretation of what a stateless codec is.
> It's not because you have to set one blob of CODEC data after S_FMT on
> a specific control that this CODEC becomes stateless. The fact is not
> supported now is just become we didn't have come accross HW that
> supports these.
> 
> FFMPEG offers a state full software codec, and you still have this
> codec_data blob for many of the format. They also only supports AVC1,
> they parser will always convert the stream to that, because it's just
> more efficient format. Android Media Codec works similarly. What keeps
> them stateful, is that you don't need to parse it, you don't even need
> to know what these data contains. They are blobs placed in the
> container that you pass as-is to the decoder.

This is all useful information, thank you.

I think I stand by my recommendation to add a statement that we don't
support such codecs at the moment. When we add support for such a codec,
then we will need to update the spec with the right rules.

I think it is difficult to make assumptions without having seen how a
typical stateful HW codec would handle this.

Regards,

	Hans

> 
>>
>> So I would just drop this sentence here (and perhaps at other places in this
>> document or the encoder document as well).
>>
>> Regards,
>>
>> 	Hans
>

Nicolas Dufresne Nov. 18, 2018, 1:25 a.m. UTC | #9

Le samedi 17 novembre 2018 à 12:43 +0100, Hans Verkuil a écrit :
> > > As far as I know all codecs have resolution/metadata in the stream.
> > 
> > Was this comment about what we currently support in V4L2 interface ? In
> 
> Yes, I was talking about all V4L2 codecs.
> 
> > real life, there is CODEC that works only with out-of-band codec data.
> > A well known one is AVC1 (and HVC1). In this mode, the AVC H264 does
> > not have start code, and the headers are not allowed in the bitstream
> > itself. This format is much more efficient to process then AVC Annex B,
> > since you can just read the NAL size and jump over instead of scanning
> > for start code. This is the format used in the very popular ISOMP4
> > container.
> 
> How would such a codec handle resolution changes? Or is that not allowed?

That's a good question. It is of course allowed, but you'd need the
request API if you want to queue this change (i.e. if you want a
resolution change using the events). Meanwhile, one can always  do
CMD_STOP, wait for the decoder to be drained, pass the new codec data,
push frames / restart streaming.

The former is the only mode implemented in GStreamer, so I'll let
Tomasz comment more on his thought how this could work.

One of the main issue with the resolution change mechanism (even
without codec data) is that you have no guaranty that the input buffer
(buffers on V4L2 OUTPUT queue) are large enough to fit a full encoded
frame of the new resolution. So userspace should have some formula to
calculate the required size, and use the CMD_STOP / drain method to be
able to re-allocate these buffers. So the newest resolution change
method, to be implemented correctly require more work (but is all
doable).

Nicolas

Tomasz Figa Jan. 22, 2019, 10:02 a.m. UTC | #10

On Mon, Nov 12, 2018 at 8:37 PM Hans Verkuil <[e-mail]> wrote:
>
> Hi Tomasz,
>
> A general note for the stateful and stateless patches: they describe specific
> use-cases of the more generic Codec Interface, and as such should be one
> level deeper in the section hierarchy.

I wonder what exactly this Codec Interface is. Is it a historical name
for mem2mem? If so, perhaps it would make sense to rename it?

>
> I.e. instead of being section 4.6/7/8:
>
> https://hverkuil.home.xs4all.nl/request-api/uapi/v4l/devices.html
>
> they should be 4.5.1/2/3.
>

FYI, the first RFC started like that, but it only made the spec
difficult to navigate and the section numbers too long.

Still, no strong opinion. I'm okay moving it there, if you think it's better.

> On 10/22/2018 04:48 PM, Tomasz Figa wrote:
> > Due to complexity of the video decoding process, the V4L2 drivers of
> > stateful decoder hardware require specific sequences of V4L2 API calls
> > to be followed. These include capability enumeration, initialization,
> > decoding, seek, pause, dynamic resolution change, drain and end of
> > stream.
[snipping any comments that I agree with]
> > +
> > +source height
> > +   height in pixels for given source resolution; relevant to encoders only
> > +
> > +source resolution
> > +   resolution in pixels of source frames being source to the encoder and
> > +   subject to further cropping to the bounds of visible resolution; relevant to
> > +   encoders only
> > +
> > +source width
> > +   width in pixels for given source resolution; relevant to encoders only
>
> I would drop these three terms: they are not used in this document since this
> describes a decoder and not an encoder.
>

The glossary is shared between encoder and decoder, as suggested in
previous round of review.

[snip]
> > +
> > +   * If width and height are set to non-zero values, the ``CAPTURE`` format
> > +     will be updated with an appropriate frame buffer resolution instantly.
> > +     However, for coded formats that include stream resolution information,
> > +     after the decoder is done parsing the information from the stream, it will
> > +     update the ``CAPTURE`` format with new values and signal a source change
> > +     event.
>
> What if the initial width and height specified by userspace matches the parsed
> width and height? Do you still get a source change event? I think you should
> always get this event since there are other parameters that depend on the parsing
> of the meta data.
>
> But that should be made explicit here.
>

Yes, the change event should happen always after the driver determines
the format of the stream. Will specify it explicitly.

> > +
> > +   .. warning::
>
> I'd call this a note rather than a warning.
>

I think it deserves at least the "important" level, since it informs
about the side effects of the call, affecting any actions that the
client might have done before.


> > +
> > +      Changing the ``OUTPUT`` format may change the currently set ``CAPTURE``
> > +      format. The decoder will derive a new ``CAPTURE`` format from the
> > +      ``OUTPUT`` format being set, including resolution, colorimetry
> > +      parameters, etc. If the client needs a specific ``CAPTURE`` format, it
> > +      must adjust it afterwards.
> > +
> > +3.  **Optional.** Query the minimum number of buffers required for ``OUTPUT``
> > +    queue via :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to
> > +    use more buffers than the minimum required by hardware/format.
>
> Why is this useful? As far as I can tell only the s5p-mfc *encoder* supports
> this control, so this seems pointless. And since the output queue gets a bitstream
> I don't see any reason for reading this control in a decoder.
>

Indeed, querying this for bitstream buffers probably doesn't make much
sense. I'll remove it.

> > +
> > +    * **Required fields:**
> > +
> > +      ``id``
> > +          set to ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
> > +
> > +    * **Return fields:**
> > +
> > +      ``value``
> > +          the minimum number of ``OUTPUT`` buffers required for the currently
> > +          set format
> > +
> > +4.  Allocate source (bitstream) buffers via :c:func:`VIDIOC_REQBUFS` on
> > +    ``OUTPUT``.
> > +
> > +    * **Required fields:**
> > +
> > +      ``count``
> > +          requested number of buffers to allocate; greater than zero
> > +
> > +      ``type``
> > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> > +
> > +      ``memory``
> > +          follows standard semantics
> > +
> > +    * **Return fields:**
> > +
> > +      ``count``
> > +          the actual number of buffers allocated
> > +
> > +    .. warning::
> > +
> > +       The actual number of allocated buffers may differ from the ``count``
> > +       given. The client must check the updated value of ``count`` after the
> > +       call returns.
> > +
> > +    .. note::
> > +
> > +       To allocate more than the minimum number of buffers (for pipeline
> > +       depth), the client may query the ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
> > +       control to get the minimum number of buffers required by the
> > +       decoder/format, and pass the obtained value plus the number of
> > +       additional buffers needed in the ``count`` field to
> > +       :c:func:`VIDIOC_REQBUFS`.
>
> As mentioned above, this makes no sense for stateful decoders IMHO.
>

Ack.

> > +
> > +    Alternatively, :c:func:`VIDIOC_CREATE_BUFS` on the ``OUTPUT`` queue can be
> > +    used to have more control over buffer allocation.
> > +
> > +    * **Required fields:**
> > +
> > +      ``count``
> > +          requested number of buffers to allocate; greater than zero
> > +
> > +      ``type``
> > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> > +
> > +      ``memory``
> > +          follows standard semantics
> > +
> > +      ``format``
> > +          follows standard semantics
> > +
> > +    * **Return fields:**
> > +
> > +      ``count``
> > +          adjusted to the number of allocated buffers
> > +
> > +    .. warning::
> > +
> > +       The actual number of allocated buffers may differ from the ``count``
> > +       given. The client must check the updated value of ``count`` after the
> > +       call returns.
> > +
> > +5.  Start streaming on the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`.
> > +
> > +6.  **This step only applies to coded formats that contain resolution information
> > +    in the stream.** Continue queuing/dequeuing bitstream buffers to/from the
> > +    ``OUTPUT`` queue via :c:func:`VIDIOC_QBUF` and :c:func:`VIDIOC_DQBUF`. The
> > +    buffers will be processed and returned to the client in order, until
> > +    required metadata to configure the ``CAPTURE`` queue are found. This is
> > +    indicated by the decoder sending a ``V4L2_EVENT_SOURCE_CHANGE`` event with
> > +    ``V4L2_EVENT_SRC_CH_RESOLUTION`` source change type.
> > +
> > +    * It is not an error if the first buffer does not contain enough data for
> > +      this to occur. Processing of the buffers will continue as long as more
> > +      data is needed.
> > +
> > +    * If data in a buffer that triggers the event is required to decode the
> > +      first frame, it will not be returned to the client, until the
> > +      initialization sequence completes and the frame is decoded.
> > +
> > +    * If the client sets width and height of the ``OUTPUT`` format to 0,
> > +      calling :c:func:`VIDIOC_G_FMT`, :c:func:`VIDIOC_S_FMT` or
> > +      :c:func:`VIDIOC_TRY_FMT` on the ``CAPTURE`` queue will return the
> > +      ``-EACCES`` error code, until the decoder configures ``CAPTURE`` format
> > +      according to stream metadata.
> > +
> > +    .. important::
> > +
> > +       Any client query issued after the decoder queues the event will return
> > +       values applying to the just parsed stream, including queue formats,
> > +       selection rectangles and controls.
> > +
> > +    .. note::
> > +
> > +       A client capable of acquiring stream parameters from the bitstream on
> > +       its own may attempt to set the width and height of the ``OUTPUT`` format
> > +       to non-zero values matching the coded size of the stream, skip this step
> > +       and continue with the `Capture setup` sequence. However, it must not
> > +       rely on any driver queries regarding stream parameters, such as
> > +       selection rectangles and controls, since the decoder has not parsed them
> > +       from the stream yet. If the values configured by the client do not match
> > +       those parsed by the decoder, a `Dynamic resolution change` will be
> > +       triggered to reconfigure them.
> > +
> > +    .. note::
> > +
> > +       No decoded frames are produced during this phase.
> > +
> > +7.  Continue with the `Capture setup` sequence.
> > +
> > +Capture setup
> > +=============
> > +
> > +1.  Call :c:func:`VIDIOC_G_FMT` on the ``CAPTURE`` queue to get format for the
> > +    destination buffers parsed/decoded from the bitstream.
> > +
> > +    * **Required fields:**
> > +
> > +      ``type``
> > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> > +
> > +    * **Return fields:**
> > +
> > +      ``width``, ``height``
> > +          frame buffer resolution for the decoded frames
> > +
> > +      ``pixelformat``
> > +          pixel format for decoded frames
> > +
> > +      ``num_planes`` (for _MPLANE ``type`` only)
> > +          number of planes for pixelformat
> > +
> > +      ``sizeimage``, ``bytesperline``
> > +          as per standard semantics; matching frame buffer format
> > +
> > +    .. note::
> > +
> > +       The value of ``pixelformat`` may be any pixel format supported by the
> > +       decoder for the current stream. The decoder should choose a
> > +       preferred/optimal format for the default configuration. For example, a
> > +       YUV format may be preferred over an RGB format if an additional
> > +       conversion step would be required for the latter.
> > +
> > +2.  **Optional.** Acquire the visible resolution via
> > +    :c:func:`VIDIOC_G_SELECTION`.
> > +
> > +    * **Required fields:**
> > +
> > +      ``type``
> > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> > +
> > +      ``target``
> > +          set to ``V4L2_SEL_TGT_COMPOSE``
> > +
> > +    * **Return fields:**
> > +
> > +      ``r.left``, ``r.top``, ``r.width``, ``r.height``
> > +          the visible rectangle; it must fit within the frame buffer resolution
> > +          returned by :c:func:`VIDIOC_G_FMT` on ``CAPTURE``.
> > +
> > +    * The following selection targets are supported on ``CAPTURE``:
> > +
> > +      ``V4L2_SEL_TGT_CROP_BOUNDS``
> > +          corresponds to the coded resolution of the stream
> > +
> > +      ``V4L2_SEL_TGT_CROP_DEFAULT``
> > +          the rectangle covering the part of the ``CAPTURE`` buffer that
> > +          contains meaningful picture data (visible area); width and height
> > +          will be equal to the visible resolution of the stream
> > +
> > +      ``V4L2_SEL_TGT_CROP``
> > +          the rectangle within the coded resolution to be output to
> > +          ``CAPTURE``; defaults to ``V4L2_SEL_TGT_CROP_DEFAULT``; read-only on
> > +          hardware without additional compose/scaling capabilities
> > +
> > +      ``V4L2_SEL_TGT_COMPOSE_BOUNDS``
> > +          the maximum rectangle within a ``CAPTURE`` buffer, which the cropped
> > +          frame can be output into; equal to ``V4L2_SEL_TGT_CROP`` if the
> > +          hardware does not support compose/scaling
> > +
> > +      ``V4L2_SEL_TGT_COMPOSE_DEFAULT``
> > +          equal to ``V4L2_SEL_TGT_CROP``
> > +
> > +      ``V4L2_SEL_TGT_COMPOSE``
> > +          the rectangle inside a ``CAPTURE`` buffer into which the cropped
> > +          frame is written; defaults to ``V4L2_SEL_TGT_COMPOSE_DEFAULT``;
> > +          read-only on hardware without additional compose/scaling capabilities
> > +
> > +      ``V4L2_SEL_TGT_COMPOSE_PADDED``
> > +          the rectangle inside a ``CAPTURE`` buffer which is overwritten by the
> > +          hardware; equal to ``V4L2_SEL_TGT_COMPOSE`` if the hardware does not
> > +          write padding pixels
> > +
> > +    .. warning::
> > +
> > +       The values are guaranteed to be meaningful only after the decoder
> > +       successfully parses the stream metadata. The client must not rely on the
> > +       query before that happens.
> > +
> > +3.  Query the minimum number of buffers required for the ``CAPTURE`` queue via
> > +    :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to use more
> > +    buffers than the minimum required by hardware/format.
>
> Is this step optional or required? Can it change when a resolution change occurs?

Probably not with a simple resolution change, but a case when a stream
is changed on the fly would trigger what we call "resolution change"
here, but what would effectively be a "source change" and that could
include a change in the number of required CAPTURE buffers.

> How does this relate to the checks for the minimum number of buffers that REQBUFS
> does?

The control returns the minimum that REQBUFS would allow, so the
application can add few more buffers on top of that and improve the
pipelining.

>
> The 'This is useful if' sentence suggests that it is optional, but I think that
> sentence just confuses the issue.
>

It used to be optional and I didn't rephrase it after turning it into
mandatory. How about:

    This enables the client to request more buffers
    than the minimum required by hardware/format and achieve better pipelining.

> > +
> > +    * **Required fields:**
> > +
> > +      ``id``
> > +          set to ``V4L2_CID_MIN_BUFFERS_FOR_CAPTURE``
> > +
> > +    * **Return fields:**
> > +
> > +      ``value``
> > +          minimum number of buffers required to decode the stream parsed in
> > +          this initialization sequence.
> > +
> > +    .. note::
> > +
> > +       The minimum number of buffers must be at least the number required to
> > +       successfully decode the current stream. This may for example be the
> > +       required DPB size for an H.264 stream given the parsed stream
> > +       configuration (resolution, level).
> > +
> > +    .. warning::
> > +
> > +       The value is guaranteed to be meaningful only after the decoder
> > +       successfully parses the stream metadata. The client must not rely on the
> > +       query before that happens.
> > +
> > +4.  **Optional.** Enumerate ``CAPTURE`` formats via :c:func:`VIDIOC_ENUM_FMT` on
> > +    the ``CAPTURE`` queue. Once the stream information is parsed and known, the
> > +    client may use this ioctl to discover which raw formats are supported for
> > +    given stream and select one of them via :c:func:`VIDIOC_S_FMT`.
>
> Can the list returned here differ from the list returned in the 'Querying capabilities'
> step? If so, then I assume it will always be a subset of what was returned in
> the 'Querying' step?

Depends on whether you're considering just VIDIOC_ENUM_FMT or also
VIDIOC_G_FMT and VIDIOC_ENUM_FRAMESIZES.

The initial VIDIOC_ENUM_FMT has no way to account for any resolution
constraints of the formats, so the list would include all raw pixel
formats that the decoder can handle with selected coded pixel format.
However, the list can be further narrowed down by using
VIDIOC_ENUM_FRAMESIZES, to restrict each raw format only to the
resolutions it can handle.

The VIDIOC_ENUM_FMT call in this sequence (after getting the stream
information) would have the knowledge about the resolution, so the
list returned here would only include the formats that can be actually
handled. It should match the result of the initial query using both
VIDIOC_ENUM_FMT and VIDIOC_ENUM_FRAMESIZES.

>
> > +
> > +    .. important::
> > +
> > +       The decoder will return only formats supported for the currently
> > +       established coded format, as per the ``OUTPUT`` format and/or stream
> > +       metadata parsed in this initialization sequence, even if more formats
> > +       may be supported by the decoder in general.
> > +
> > +       For example, a decoder may support YUV and RGB formats for resolutions
> > +       1920x1088 and lower, but only YUV for higher resolutions (due to
> > +       hardware limitations). After parsing a resolution of 1920x1088 or lower,
> > +       :c:func:`VIDIOC_ENUM_FMT` may return a set of YUV and RGB pixel formats,
> > +       but after parsing resolution higher than 1920x1088, the decoder will not
> > +       return RGB, unsupported for this resolution.
> > +
> > +       However, subsequent resolution change event triggered after
> > +       discovering a resolution change within the same stream may switch
> > +       the stream into a lower resolution and :c:func:`VIDIOC_ENUM_FMT`
> > +       would return RGB formats again in that case.
> > +
> > +5.  **Optional.** Set the ``CAPTURE`` format via :c:func:`VIDIOC_S_FMT` on the
> > +    ``CAPTURE`` queue. The client may choose a different format than
> > +    selected/suggested by the decoder in :c:func:`VIDIOC_G_FMT`.
> > +
> > +    * **Required fields:**
> > +
> > +      ``type``
> > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> > +
> > +      ``pixelformat``
> > +          a raw pixel format
> > +
> > +    .. note::
> > +
> > +       The client may use :c:func:`VIDIOC_ENUM_FMT` after receiving the
> > +       ``V4L2_EVENT_SOURCE_CHANGE`` event to find out the set of raw formats
> > +       supported for the stream.
>
> Isn't this a duplicate of step 4? I think this note can be dropped.
>

Ack.

> > +
> > +6.  If all the following conditions are met, the client may resume the decoding
> > +    instantly:
> > +
> > +    * ``sizeimage`` of the new format (determined in previous steps) is less
> > +      than or equal to the size of currently allocated buffers,
> > +
> > +    * the number of buffers currently allocated is greater than or equal to the
> > +      minimum number of buffers acquired in previous steps. To fulfill this
> > +      requirement, the client may use :c:func:`VIDIOC_CREATE_BUFS` to add new
> > +      buffers.
> > +
> > +    In such case, the remaining steps do not apply and the client may resume
> > +    the decoding by one of the following actions:
> > +
> > +    * if the ``CAPTURE`` queue is streaming, call :c:func:`VIDIOC_DECODER_CMD`
> > +      with the ``V4L2_DEC_CMD_START`` command,
> > +
> > +    * if the ``CAPTURE`` queue is not streaming, call :c:func:`VIDIOC_STREAMON`
> > +      on the ``CAPTURE`` queue.
> > +
> > +    However, if the client intends to change the buffer set, to lower
> > +    memory usage or for any other reasons, it may be achieved by following
> > +    the steps below.
> > +
> > +7.  **If the** ``CAPTURE`` **queue is streaming,** keep queuing and dequeuing
> > +    buffers on the ``CAPTURE`` queue until a buffer marked with the
> > +    ``V4L2_BUF_FLAG_LAST`` flag is dequeued.
> > +
> > +8.  **If the** ``CAPTURE`` **queue is streaming,** call :c:func:`VIDIOC_STREAMOFF`
> > +    on the ``CAPTURE`` queue to stop streaming.
> > +
> > +    .. warning::
> > +
> > +       The ``OUTPUT`` queue must remain streaming. Calling
> > +       :c:func:`VIDIOC_STREAMOFF` on it would abort the sequence and trigger a
> > +       seek.
> > +
> > +9.  **If the** ``CAPTURE`` **queue has buffers allocated,** free the ``CAPTURE``
> > +    buffers using :c:func:`VIDIOC_REQBUFS`.
> > +
> > +    * **Required fields:**
> > +
> > +      ``count``
> > +          set to 0
> > +
> > +      ``type``
> > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> > +
> > +      ``memory``
> > +          follows standard semantics
> > +
> > +10. Allocate ``CAPTURE`` buffers via :c:func:`VIDIOC_REQBUFS` on the
> > +    ``CAPTURE`` queue.
> > +
> > +    * **Required fields:**
> > +
> > +      ``count``
> > +          requested number of buffers to allocate; greater than zero
> > +
> > +      ``type``
> > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> > +
> > +      ``memory``
> > +          follows standard semantics
> > +
> > +    * **Return fields:**
> > +
> > +      ``count``
> > +          actual number of buffers allocated
> > +
> > +    .. warning::
> > +
> > +       The actual number of allocated buffers may differ from the ``count``
> > +       given. The client must check the updated value of ``count`` after the
> > +       call returns.
> > +
> > +    .. note::
> > +
> > +       To allocate more than the minimum number of buffers (for pipeline
> > +       depth), the client may query the ``V4L2_CID_MIN_BUFFERS_FOR_CAPTURE``
> > +       control to get the minimum number of buffers required, and pass the
> > +       obtained value plus the number of additional buffers needed in the
> > +       ``count`` field to :c:func:`VIDIOC_REQBUFS`.
>
> Same question as before: is it optional or required to obtain the value of this
> control? And can't the driver just set the min_buffers_needed field in the capture
> vb2_queue to the minimum number of buffers that are required?

min_buffers_needed is the number of buffers that must be queued before
start_streaming() can be called, so not relevant here. The control is
about the number of buffers to be allocated. Although the drivers must
ensure that REQBUFS allocates the absolute minimum number of buffers
for the decoding to be able to progress, the pipeline depth is
something that the applications should control (e.g. depending on the
consumers of the decoded buffers) and this is allowed by this control.

>
> Should you be allowed to allocate buffers at all if the capture format isn't
> known? I.e. width/height is still 0. It makes no sense to call REQBUFS since
> there is no format size known that REQBUFS can use.
>

Indeed, REQBUFS(CAPTURE) must not be allowed before the stream
information is known (regardless of whether it comes from the OUTPUT
format or is parsed from the stream). Let me add this to the related
note in the Initialization sequence, which already includes
VIDIOC_*_FMT.

For the Capture setup sequence, though, it's expected to happen when
the stream information is already known, so I wouldn't change the
description here.

> > +
> > +    Alternatively, :c:func:`VIDIOC_CREATE_BUFS` on the ``CAPTURE`` queue can be
> > +    used to have more control over buffer allocation. For example, by
> > +    allocating buffers larger than the current ``CAPTURE`` format, future
> > +    resolution changes can be accommodated.
> > +
> > +    * **Required fields:**
> > +
> > +      ``count``
> > +          requested number of buffers to allocate; greater than zero
> > +
> > +      ``type``
> > +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
> > +
> > +      ``memory``
> > +          follows standard semantics
> > +
> > +      ``format``
> > +          a format representing the maximum framebuffer resolution to be
> > +          accommodated by newly allocated buffers
> > +
> > +    * **Return fields:**
> > +
> > +      ``count``
> > +          adjusted to the number of allocated buffers
> > +
> > +    .. warning::
> > +
> > +       The actual number of allocated buffers may differ from the ``count``
> > +       given. The client must check the updated value of ``count`` after the
> > +       call returns.
> > +
> > +    .. note::
> > +
> > +       To allocate buffers for a format different than parsed from the stream
> > +       metadata, the client must proceed as follows, before the metadata
> > +       parsing is initiated:
> > +
> > +       * set width and height of the ``OUTPUT`` format to desired coded resolution to
> > +         let the decoder configure the ``CAPTURE`` format appropriately,
> > +
> > +       * query the ``CAPTURE`` format using :c:func:`VIDIOC_G_FMT` and save it
> > +         until this step.
> > +
> > +       The format obtained in the query may be then used with
> > +       :c:func:`VIDIOC_CREATE_BUFS` in this step to allocate the buffers.
> > +
> > +11. Call :c:func:`VIDIOC_STREAMON` on the ``CAPTURE`` queue to start decoding
> > +    frames.
> > +
> > +Decoding
> > +========
> > +
> > +This state is reached after the `Capture setup` sequence finishes succesfully.
> > +In this state, the client queues and dequeues buffers to both queues via
> > +:c:func:`VIDIOC_QBUF` and :c:func:`VIDIOC_DQBUF`, following the standard
> > +semantics.
> > +
> > +The contents of the source ``OUTPUT`` buffers depend on the active coded pixel
> > +format and may be affected by codec-specific extended controls, as stated in
> > +the documentation of each format.
> > +
> > +Both queues operate independently, following the standard behavior of V4L2
> > +buffer queues and memory-to-memory devices. In addition, the order of decoded
> > +frames dequeued from the ``CAPTURE`` queue may differ from the order of queuing
> > +coded frames to the ``OUTPUT`` queue, due to properties of the selected coded
> > +format, e.g. frame reordering.
> > +
> > +The client must not assume any direct relationship between ``CAPTURE``
> > +and ``OUTPUT`` buffers and any specific timing of buffers becoming
> > +available to dequeue. Specifically,
> > +
> > +* a buffer queued to ``OUTPUT`` may result in no buffers being produced
> > +  on ``CAPTURE`` (e.g. if it does not contain encoded data, or if only
> > +  metadata syntax structures are present in it),
> > +
> > +* a buffer queued to ``OUTPUT`` may result in more than 1 buffer produced
> > +  on ``CAPTURE`` (if the encoded data contained more than one frame, or if
> > +  returning a decoded frame allowed the decoder to return a frame that
> > +  preceded it in decode, but succeeded it in the display order),
> > +
> > +* a buffer queued to ``OUTPUT`` may result in a buffer being produced on
> > +  ``CAPTURE`` later into decode process, and/or after processing further
> > +  ``OUTPUT`` buffers, or be returned out of order, e.g. if display
> > +  reordering is used,
> > +
> > +* buffers may become available on the ``CAPTURE`` queue without additional
> > +  buffers queued to ``OUTPUT`` (e.g. during drain or ``EOS``), because of the
> > +  ``OUTPUT`` buffers queued in the past whose decoding results are only
> > +  available at later time, due to specifics of the decoding process.
> > +
> > +.. note::
> > +
> > +   To allow matching decoded ``CAPTURE`` buffers with ``OUTPUT`` buffers they
> > +   originated from, the client can set the ``timestamp`` field of the
> > +   :c:type:`v4l2_buffer` struct when queuing an ``OUTPUT`` buffer. The
> > +   ``CAPTURE`` buffer(s), which resulted from decoding that ``OUTPUT`` buffer
> > +   will have their ``timestamp`` field set to the same value when dequeued.
> > +
> > +   In addition to the straighforward case of one ``OUTPUT`` buffer producing
> > +   one ``CAPTURE`` buffer, the following cases are defined:
> > +
> > +   * one ``OUTPUT`` buffer generates multiple ``CAPTURE`` buffers: the same
> > +     ``OUTPUT`` timestamp will be copied to multiple ``CAPTURE`` buffers,
> > +
> > +   * multiple ``OUTPUT`` buffers generate one ``CAPTURE`` buffer: timestamp of
> > +     the ``OUTPUT`` buffer queued last will be copied,
> > +
> > +   * the decoding order differs from the display order (i.e. the
> > +     ``CAPTURE`` buffers are out-of-order compared to the ``OUTPUT`` buffers):
> > +     ``CAPTURE`` timestamps will not retain the order of ``OUTPUT`` timestamps
> > +     and thus monotonicity of the timestamps cannot be guaranteed.
>
> Should stateful codecs be required to support 'tags'? See:
>
> https://www.mail-archive.com/[e-mail]/msg136314.html
>
> To be honest, I'm inclined to require this for all m2m devices eventually.
>

I guess this goes outside of the scope now, since we deferred tags.

Other than that, I would indeed make all m2m devices support tags,
since it shouldn't differ from the timestamp copy feature as we have
now.

> > +
> > +During the decoding, the decoder may initiate one of the special sequences, as
> > +listed below. The sequences will result in the decoder returning all the
> > +``CAPTURE`` buffers that originated from all the ``OUTPUT`` buffers processed
> > +before the sequence started. Last of the buffers will have the
> > +``V4L2_BUF_FLAG_LAST`` flag set. To determine the sequence to follow, the client
> > +must check if there is any pending event and,
> > +
> > +* if a ``V4L2_EVENT_SOURCE_CHANGE`` event is pending, the `Dynamic resolution
> > +  change` sequence needs to be followed,
> > +
> > +* if a ``V4L2_EVENT_EOS`` event is pending, the `End of stream` sequence needs
> > +  to be followed.
> > +
> > +Some of the sequences can be intermixed with each other and need to be handled
> > +as they happen. The exact operation is documented for each sequence.
> > +
> > +Seek
> > +====
> > +
> > +Seek is controlled by the ``OUTPUT`` queue, as it is the source of coded data.
> > +The seek does not require any specific operation on the ``CAPTURE`` queue, but
> > +it may be affected as per normal decoder operation.
> > +
> > +1. Stop the ``OUTPUT`` queue to begin the seek sequence via
> > +   :c:func:`VIDIOC_STREAMOFF`.
> > +
> > +   * **Required fields:**
> > +
> > +     ``type``
> > +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> > +
> > +   * The decoder will drop all the pending ``OUTPUT`` buffers and they must be
> > +     treated as returned to the client (following standard semantics).
> > +
> > +2. Restart the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`
> > +
> > +   * **Required fields:**
> > +
> > +     ``type``
> > +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> > +
> > +   * The decoder will start accepting new source bitstream buffers after the
> > +     call returns.
> > +
> > +3. Start queuing buffers containing coded data after the seek to the ``OUTPUT``
> > +   queue until a suitable resume point is found.
> > +
> > +   .. note::
> > +
> > +      There is no requirement to begin queuing coded data starting exactly
> > +      from a resume point (e.g. SPS or a keyframe). Any queued ``OUTPUT``
> > +      buffers will be processed and returned to the client until a suitable
> > +      resume point is found.  While looking for a resume point, the decoder
> > +      should not produce any decoded frames into ``CAPTURE`` buffers.
> > +
> > +      Some hardware is known to mishandle seeks to a non-resume point. Such an
> > +      operation may result in an unspecified number of corrupted decoded frames
> > +      being made available on the ``CAPTURE`` queue. Drivers must ensure that
> > +      no fatal decoding errors or crashes occur, and implement any necessary
> > +      handling and workarounds for hardware issues related to seek operations.
>
> Is there a requirement that those corrupted frames have V4L2_BUF_FLAG_ERROR set?
> I.e., can userspace detect those currupted frames?
>

I think the question is whether the kernel driver can actually detect
those corrupted frames. We can't guarantee reporting errors to the
userspace, if the hardware doesn't actually report them.

Could we perhaps keep this an open question and possibly address with
some extension that could be an opt in for the decoders that can
report errors?

> > +
> > +   .. warning::
> > +
> > +      In case of the H.264 codec, the client must take care not to seek over a
> > +      change of SPS/PPS. Even though the target frame could be a keyframe, the
> > +      stale SPS/PPS inside decoder state would lead to undefined results when
> > +      decoding. Although the decoder must handle such case without a crash or a
> > +      fatal decode error, the client must not expect a sensible decode output.
> > +
> > +4. After a resume point is found, the decoder will start returning ``CAPTURE``
> > +   buffers containing decoded frames.
> > +
> > +.. important::
> > +
> > +   A seek may result in the `Dynamic resolution change` sequence being
> > +   iniitated, due to the seek target having decoding parameters different from
> > +   the part of the stream decoded before the seek. The sequence must be handled
> > +   as per normal decoder operation.
> > +
> > +.. warning::
> > +
> > +   It is not specified when the ``CAPTURE`` queue starts producing buffers
> > +   containing decoded data from the ``OUTPUT`` buffers queued after the seek,
> > +   as it operates independently from the ``OUTPUT`` queue.
> > +
> > +   The decoder may return a number of remaining ``CAPTURE`` buffers containing
> > +   decoded frames originating from the ``OUTPUT`` buffers queued before the
> > +   seek sequence is performed.
> > +
> > +   The ``VIDIOC_STREAMOFF`` operation discards any remaining queued
> > +   ``OUTPUT`` buffers, which means that not all of the ``OUTPUT`` buffers
> > +   queued before the seek sequence may have matching ``CAPTURE`` buffers
> > +   produced.  For example, given the sequence of operations on the
> > +   ``OUTPUT`` queue:
> > +
> > +     QBUF(A), QBUF(B), STREAMOFF(), STREAMON(), QBUF(G), QBUF(H),
> > +
> > +   any of the following results on the ``CAPTURE`` queue is allowed:
> > +
> > +     {A’, B’, G’, H’}, {A’, G’, H’}, {G’, H’}.
>
> Isn't it the case that if you would want to avoid that, then you should call
> DECODER_STOP, wait for the last buffer on the CAPTURE queue, then seek and
> call DECODER_START. If you do that, then you should always get {A’, B’, G’, H’}.
> (basically following the Drain sequence).

Yes, it is, but I think it depends on the application needs. Here we
just give a primitive to change the place in the stream that's being
decoded (or change the stream on the fly).

Actually, with the timestamp copy, I guess we wouldn't even need to do
the DECODER_STOP, as we could just discard the CAPTURE buffers until
we get one that matches the timestamp of the first OUTPUT buffer after
the seek.

>
> Admittedly, you typically want to do an instantaneous seek, so this is probably
> not what you want to do normally.
>
> It might help to have this documented in a separate note.
>

The instantaneous seek is documented below. I'm not sure if there is
any practical need to document the other case, but I could add a
sentence like below to the warning above. What do you think?

   The ``VIDIOC_STREAMOFF`` operation discards any remaining queued
   ``OUTPUT`` buffers, which means that not all of the ``OUTPUT`` buffers
   queued before the seek sequence may have matching ``CAPTURE`` buffers
   produced.  For example, given the sequence of operations on the
   ``OUTPUT`` queue:

     QBUF(A), QBUF(B), STREAMOFF(), STREAMON(), QBUF(G), QBUF(H),

   any of the following results on the ``CAPTURE`` queue is allowed:

     {A’, B’, G’, H’}, {A’, G’, H’}, {G’, H’}.

   To determine the CAPTURE buffer containing the first decoded frame
after the seek,
   the client may observe the timestamps to match the CAPTURE and OUTPUT buffers
   or use V4L2_DEC_CMD_STOP and V4L2_DEC_CMD_START to drain the decoder.

> > +
> > +.. note::
> > +
> > +   To achieve instantaneous seek, the client may restart streaming on the
> > +   ``CAPTURE`` queue too to discard decoded, but not yet dequeued buffers.
> > +
> > +Dynamic resolution change
> > +=========================
> > +
> > +Streams that include resolution metadata in the bitstream may require switching
> > +to a different resolution during the decoding.
> > +
> > +The sequence starts when the decoder detects a coded frame with one or more of
> > +the following parameters different from previously established (and reflected
> > +by corresponding queries):
> > +
> > +* coded resolution (``OUTPUT`` width and height),
> > +
> > +* visible resolution (selection rectangles),
> > +
> > +* the minimum number of buffers needed for decoding.
> > +
> > +Whenever that happens, the decoder must proceed as follows:
> > +
> > +1.  After encountering a resolution change in the stream, the decoder sends a
> > +    ``V4L2_EVENT_SOURCE_CHANGE`` event with source change type set to
> > +    ``V4L2_EVENT_SRC_CH_RESOLUTION``.
> > +
> > +    .. important::
> > +
> > +       Any client query issued after the decoder queues the event will return
> > +       values applying to the stream after the resolution change, including
> > +       queue formats, selection rectangles and controls.
> > +
> > +2.  The decoder will then process and decode all remaining buffers from before
> > +    the resolution change point.
> > +
> > +    * The last buffer from before the change must be marked with the
> > +      ``V4L2_BUF_FLAG_LAST`` flag, similarly to the `Drain` sequence above.
> > +
> > +    .. warning::
> > +
> > +       The last buffer may be empty (with :c:type:`v4l2_buffer` ``bytesused``
> > +       = 0) and in such case it must be ignored by the client, as it does not
> > +       contain a decoded frame.
> > +
> > +    .. note::
> > +
> > +       Any attempt to dequeue more buffers beyond the buffer marked with
> > +       ``V4L2_BUF_FLAG_LAST`` will result in a -EPIPE error from
> > +       :c:func:`VIDIOC_DQBUF`.
> > +
> > +The client must continue the sequence as described below to continue the
> > +decoding process.
> > +
> > +1.  Dequeue the source change event.
> > +
> > +    .. important::
> > +
> > +       A source change triggers an implicit decoder drain, similar to the
> > +       explicit `Drain` sequence. The decoder is stopped after it completes.
> > +       The decoding process must be resumed with either a pair of calls to
> > +       :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
> > +       ``CAPTURE`` queue, or a call to :c:func:`VIDIOC_DECODER_CMD` with the
> > +       ``V4L2_DEC_CMD_START`` command.
> > +
> > +2.  Continue with the `Capture setup` sequence.
> > +
> > +.. note::
> > +
> > +   During the resolution change sequence, the ``OUTPUT`` queue must remain
> > +   streaming. Calling :c:func:`VIDIOC_STREAMOFF` on the ``OUTPUT`` queue would
> > +   abort the sequence and initiate a seek.
> > +
> > +   In principle, the ``OUTPUT`` queue operates separately from the ``CAPTURE``
> > +   queue and this remains true for the duration of the entire resolution change
> > +   sequence as well.
> > +
> > +   The client should, for best performance and simplicity, keep queuing/dequeuing
> > +   buffers to/from the ``OUTPUT`` queue even while processing this sequence.
> > +
> > +Drain
> > +=====
> > +
> > +To ensure that all queued ``OUTPUT`` buffers have been processed and related
> > +``CAPTURE`` buffers output to the client, the client must follow the drain
> > +sequence described below. After the drain sequence ends, the client has
> > +received all decoded frames for all ``OUTPUT`` buffers queued before the
> > +sequence was started.
> > +
> > +1. Begin drain by issuing :c:func:`VIDIOC_DECODER_CMD`.
> > +
> > +   * **Required fields:**
> > +
> > +     ``cmd``
> > +         set to ``V4L2_DEC_CMD_STOP``
> > +
> > +     ``flags``
> > +         set to 0
> > +
> > +     ``pts``
> > +         set to 0
> > +
> > +   .. warning::
> > +
> > +   The sentence can be only initiated if both ``OUTPUT`` and ``CAPTURE`` queues
>
> 'sentence'? You mean 'decoder command'?

Sequence. :)

>
> > +   are streaming. For compatibility reasons, the call to
> > +   :c:func:`VIDIOC_DECODER_CMD` will not fail even if any of the queues is not
> > +   streaming, but at the same time it will not initiate the `Drain` sequence
> > +   and so the steps described below would not be applicable.
> > +
> > +2. Any ``OUTPUT`` buffers queued by the client before the
> > +   :c:func:`VIDIOC_DECODER_CMD` was issued will be processed and decoded as
> > +   normal. The client must continue to handle both queues independently,
> > +   similarly to normal decode operation. This includes,
> > +
> > +   * handling any operations triggered as a result of processing those buffers,
> > +     such as the `Dynamic resolution change` sequence, before continuing with
> > +     the drain sequence,
> > +
> > +   * queuing and dequeuing ``CAPTURE`` buffers, until a buffer marked with the
> > +     ``V4L2_BUF_FLAG_LAST`` flag is dequeued,
> > +
> > +     .. warning::
> > +
> > +        The last buffer may be empty (with :c:type:`v4l2_buffer`
> > +        ``bytesused`` = 0) and in such case it must be ignored by the client,
> > +        as it does not contain a decoded frame.
> > +
> > +     .. note::
> > +
> > +        Any attempt to dequeue more buffers beyond the buffer marked with
> > +        ``V4L2_BUF_FLAG_LAST`` will result in a -EPIPE error from
> > +        :c:func:`VIDIOC_DQBUF`.
> > +
> > +   * dequeuing processed ``OUTPUT`` buffers, until all the buffers queued
> > +     before the ``V4L2_DEC_CMD_STOP`` command are dequeued.
> > +
> > +   * dequeuing the ``V4L2_EVENT_EOS`` event, if the client subscribed to it.
> > +
> > +   .. note::
> > +
> > +      For backwards compatibility, the decoder will signal a ``V4L2_EVENT_EOS``
> > +      event when the last the last frame has been decoded and all frames are
>
> 'the last the last' -> the last

Ack.

>
> > +      ready to be dequeued. It is a deprecated behavior and the client must not
> > +      rely on it. The ``V4L2_BUF_FLAG_LAST`` buffer flag should be used
> > +      instead.
> > +
> > +3. Once all the ``OUTPUT`` buffers queued before the ``V4L2_DEC_CMD_STOP`` call
> > +   and the last ``CAPTURE`` buffer are dequeued, the decoder is stopped and it
>
> This sentence is a bit confusing. This is better IMHO:
>
> 3. Once all the ``OUTPUT`` buffers queued before the ``V4L2_DEC_CMD_STOP`` call
>    are dequeued and the last ``CAPTURE`` buffer is dequeued, the decoder is stopped and it
>

Ack.

> > +   will accept, but not process any newly queued ``OUTPUT`` buffers until the
>
> process any -> process, any
>

Ack.

> > +   client issues any of the following operations:
> > +
> > +   * ``V4L2_DEC_CMD_START`` - the decoder will resume the operation normally,
> > +
> > +   * a pair of :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
> > +     ``CAPTURE`` queue - the decoder will resume the operation normally,
> > +     however any ``CAPTURE`` buffers still in the queue will be returned to the
> > +     client,
> > +
> > +   * a pair of :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
> > +     ``OUTPUT`` queue - any pending source buffers will be returned to the
> > +     client and the `Seek` sequence will be triggered.
> > +
> > +.. note::
> > +
> > +   Once the drain sequence is initiated, the client needs to drive it to
> > +   completion, as described by the steps above, unless it aborts the process by
> > +   issuing :c:func:`VIDIOC_STREAMOFF` on any of the ``OUTPUT`` or ``CAPTURE``
> > +   queues.  The client is not allowed to issue ``V4L2_DEC_CMD_START`` or
> > +   ``V4L2_DEC_CMD_STOP`` again while the drain sequence is in progress and they
> > +   will fail with -EBUSY error code if attempted.
> > +
> > +   Although mandatory, the availability of decoder commands may be queried
> > +   using :c:func:`VIDIOC_TRY_DECODER_CMD`.
> > +
> > +End of stream
> > +=============
> > +
> > +If the decoder encounters an end of stream marking in the stream, the decoder
> > +will initiate the `Drain` sequence, which the client must handle as described
> > +above, skipping the initial :c:func:`VIDIOC_DECODER_CMD`.
> > +
> > +Commit points
> > +=============
> > +
> > +Setting formats and allocating buffers trigger changes in the behavior of the
> > +decoder.
> > +
> > +1. Setting the format on the ``OUTPUT`` queue may change the set of formats
> > +   supported/advertised on the ``CAPTURE`` queue. In particular, it also means
> > +   that the ``CAPTURE`` format may be reset and the client must not rely on the
> > +   previously set format being preserved.
> > +
> > +2. Enumerating formats on the ``CAPTURE`` queue always returns only formats
> > +   supported for the current ``OUTPUT`` format.
> > +
> > +3. Setting the format on the ``CAPTURE`` queue does not change the list of
> > +   formats available on the ``OUTPUT`` queue. An attempt to set the ``CAPTURE``
> > +   format that is not supported for the currently selected ``OUTPUT`` format
> > +   will result in the decoder adjusting the requested ``CAPTURE`` format to a
> > +   supported one.
> > +
> > +4. Enumerating formats on the ``OUTPUT`` queue always returns the full set of
> > +   supported coded formats, irrespectively of the current ``CAPTURE`` format.
> > +
> > +5. While buffers are allocated on the ``OUTPUT`` queue, the client must not
> > +   change the format on the queue. Drivers will return the -EBUSY error code
> > +   for any such format change attempt.
> > +
> > +To summarize, setting formats and allocation must always start with the
> > +``OUTPUT`` queue and the ``OUTPUT`` queue is the master that governs the
> > +set of supported formats for the ``CAPTURE`` queue.
> > diff --git a/Documentation/media/uapi/v4l/devices.rst b/Documentation/media/uapi/v4l/devices.rst
> > index fb7f8c26cf09..12d43fe711cf 100644
> > --- a/Documentation/media/uapi/v4l/devices.rst
> > +++ b/Documentation/media/uapi/v4l/devices.rst
> > @@ -15,6 +15,7 @@ Interfaces
> >      dev-output
> >      dev-osd
> >      dev-codec
> > +    dev-decoder
> >      dev-effect
> >      dev-raw-vbi
> >      dev-sliced-vbi
> > diff --git a/Documentation/media/uapi/v4l/pixfmt-v4l2.rst b/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
> > index 826f2305da01..ca5f2270a829 100644
> > --- a/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
> > +++ b/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
> > @@ -32,6 +32,11 @@ Single-planar format structure
> >       to a multiple of the scale factor of any smaller planes. For
> >       example when the image format is YUV 4:2:0, ``width`` and
> >       ``height`` must be multiples of two.
> > +
> > +     For compressed formats that contain the resolution information encoded
> > +     inside the stream, when fed to a stateful mem2mem decoder, the fields
> > +     may be zero to rely on the decoder to detect the right values. For more
> > +     details see :ref:`decoder` and format descriptions.
> >      * - __u32
> >        - ``pixelformat``
> >        - The pixel format or type of compression, set by the application.
> > diff --git a/Documentation/media/uapi/v4l/v4l2.rst b/Documentation/media/uapi/v4l/v4l2.rst
> > index b89e5621ae69..65dc096199ad 100644
> > --- a/Documentation/media/uapi/v4l/v4l2.rst
> > +++ b/Documentation/media/uapi/v4l/v4l2.rst
> > @@ -53,6 +53,10 @@ Authors, in alphabetical order:
> >
> >    - Original author of the V4L2 API and documentation.
> >
> > +- Figa, Tomasz <[e-mail]>
> > +
> > +  - Documented the memory-to-memory decoder interface.
> > +
> >  - H Schimek, Michael <[e-mail]>
> >
> >    - Original author of the V4L2 API and documentation.
> > @@ -61,6 +65,10 @@ Authors, in alphabetical order:
> >
> >    - Documented the Digital Video timings API.
> >
> > +- Osciak, Pawel <[e-mail]>
> > +
> > +  - Documented the memory-to-memory decoder interface.
> > +
> >  - Osciak, Pawel <[e-mail]>
> >
> >    - Designed and documented the multi-planar API.
> > @@ -85,7 +93,7 @@ Authors, in alphabetical order:
> >
> >    - Designed and documented the VIDIOC_LOG_STATUS ioctl, the extended control ioctls, major parts of the sliced VBI API, the MPEG encoder and decoder APIs and the DV Timings API.
> >
> > -**Copyright** |copy| 1999-2016: Bill Dirks, Michael H. Schimek, Hans Verkuil, Martin Rubli, Andy Walls, Muralidharan Karicheri, Mauro Carvalho Chehab, Pawel Osciak, Sakari Ailus & Antti Palosaari.
> > +**Copyright** |copy| 1999-2018: Bill Dirks, Michael H. Schimek, Hans Verkuil, Martin Rubli, Andy Walls, Muralidharan Karicheri, Mauro Carvalho Chehab, Pawel Osciak, Sakari Ailus & Antti Palosaari, Tomasz Figa
> >
> >  Except when explicitly stated as GPL, programming examples within this
> >  part can be used and distributed without restrictions.
> > diff --git a/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst b/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
> > index 85c916b0ce07..2f73fe22a9cd 100644
> > --- a/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
> > +++ b/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
> > @@ -49,14 +49,16 @@ The ``cmd`` field must contain the command code. Some commands use the
> >
> >  A :ref:`write() <func-write>` or :ref:`VIDIOC_STREAMON`
> >  call sends an implicit START command to the decoder if it has not been
> > -started yet.
> > +started yet. Applies to both queues of mem2mem decoders.
> >
> >  A :ref:`close() <func-close>` or :ref:`VIDIOC_STREAMOFF <VIDIOC_STREAMON>`
> >  call of a streaming file descriptor sends an implicit immediate STOP
> > -command to the decoder, and all buffered data is discarded.
> > +command to the decoder, and all buffered data is discarded. Applies to both
> > +queues of mem2mem decoders.
> >
> > -These ioctls are optional, not all drivers may support them. They were
> > -introduced in Linux 3.3.
> > +In principle, these ioctls are optional, not all drivers may support them. They were
> > +introduced in Linux 3.3. They are, however, mandatory for stateful mem2mem decoders
> > +(as further documented in :ref:`decoder`).
> >
> >
> >  .. tabularcolumns:: |p{1.1cm}|p{2.4cm}|p{1.2cm}|p{1.6cm}|p{10.6cm}|
> > @@ -160,26 +162,36 @@ introduced in Linux 3.3.
> >       ``V4L2_DEC_CMD_RESUME`` for that. This command has one flag:
> >       ``V4L2_DEC_CMD_START_MUTE_AUDIO``. If set, then audio will be
> >       muted when playing back at a non-standard speed.
> > +
> > +     For stateful mem2mem decoders, the command may be also used to restart
> > +     the decoder in case of an implicit stop initiated by the decoder
> > +     itself, without the ``V4L2_DEC_CMD_STOP`` being called explicitly.
> > +     No flags or other arguments are accepted in case of mem2mem decoders.
> > +     See :ref:`decoder` for more details.
> >      * - ``V4L2_DEC_CMD_STOP``
> >        - 1
> >        - Stop the decoder. When the decoder is already stopped, this
> >       command does nothing. This command has two flags: if
> >       ``V4L2_DEC_CMD_STOP_TO_BLACK`` is set, then the decoder will set
> >       the picture to black after it stopped decoding. Otherwise the last
> > -     image will repeat. mem2mem decoders will stop producing new frames
> > -     altogether. They will send a ``V4L2_EVENT_EOS`` event when the
> > -     last frame has been decoded and all frames are ready to be
> > -     dequeued and will set the ``V4L2_BUF_FLAG_LAST`` buffer flag on
> > -     the last buffer of the capture queue to indicate there will be no
> > -     new buffers produced to dequeue. This buffer may be empty,
> > -     indicated by the driver setting the ``bytesused`` field to 0. Once
> > -     the ``V4L2_BUF_FLAG_LAST`` flag was set, the
> > -     :ref:`VIDIOC_DQBUF <VIDIOC_QBUF>` ioctl will not block anymore,
> > -     but return an ``EPIPE`` error code. If
> > +     image will repeat. If
> >       ``V4L2_DEC_CMD_STOP_IMMEDIATELY`` is set, then the decoder stops
> >       immediately (ignoring the ``pts`` value), otherwise it will keep
> >       decoding until timestamp >= pts or until the last of the pending
> >       data from its internal buffers was decoded.
> > +
> > +     A stateful mem2mem decoder will proceed with decoding the source
> > +     buffers pending before the command is issued and then stop producing
> > +     new frames. It will send a ``V4L2_EVENT_EOS`` event when the last frame
> > +     has been decoded and all frames are ready to be dequeued and will set
> > +     the ``V4L2_BUF_FLAG_LAST`` buffer flag on the last buffer of the
> > +     capture queue to indicate there will be no new buffers produced to
> > +     dequeue. This buffer may be empty, indicated by the driver setting the
> > +     ``bytesused`` field to 0. Once the buffer with the
> > +     ``V4L2_BUF_FLAG_LAST`` flag set was dequeued, the :ref:`VIDIOC_DQBUF
> > +     <VIDIOC_QBUF>` ioctl will not block anymore, but return an ``EPIPE``
> > +     error code. No flags or other arguments are accepted in case of mem2mem
> > +     decoders.  See :ref:`decoder` for more details.
> >      * - ``V4L2_DEC_CMD_PAUSE``
> >        - 2
> >        - Pause the decoder. When the decoder has not been started yet, the
> > diff --git a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
> > index 3ead350e099f..0fc0b78a943e 100644
> > --- a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
> > +++ b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
> > @@ -53,6 +53,13 @@ devices that is either the struct
> >  member. When the requested buffer type is not supported drivers return
> >  an ``EINVAL`` error code.
> >
> > +A stateful mem2mem decoder will not allow operations on the
> > +``V4L2_BUF_TYPE_VIDEO_CAPTURE`` or ``V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE``
> > +buffer type until the corresponding ``V4L2_BUF_TYPE_VIDEO_OUTPUT`` or
> > +``V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE`` buffer type is configured. If such an
> > +operation is attempted, drivers return an ``EACCES`` error code. Refer to
> > +:ref:`decoder` for more details.
>
> This isn't right. EACCES is returned as long as the output format resolution is
> unknown. If it is set explicitly, then this will work without an error.

I think that's what is written above. The stream resolution is known
when the applicable OUTPUT queue is configured, which lets the driver
determine the format constraints on the applicable CAPTURE queue. If
it's not clear, could you help rephrasing?

>
> > +
> >  To change the current format parameters applications initialize the
> >  ``type`` field and all fields of the respective ``fmt`` union member.
> >  For details see the documentation of the various devices types in
> > @@ -145,6 +152,13 @@ On success 0 is returned, on error -1 and the ``errno`` variable is set
> >  appropriately. The generic error codes are described at the
> >  :ref:`Generic Error Codes <gen-errors>` chapter.
> >
> > +EACCES
> > +    The format is not accessible until another buffer type is configured.
> > +    Relevant for the V4L2_BUF_TYPE_VIDEO_CAPTURE and
> > +    V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE buffer types of mem2mem decoders, which
> > +    require the format of V4L2_BUF_TYPE_VIDEO_OUTPUT or
> > +    V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE buffer type to be configured first.
>
> Ditto.

Ditto.

Best regards,
Tomasz

Hans Verkuil Jan. 22, 2019, 2:47 p.m. UTC | #11

On 01/22/19 11:02, Tomasz Figa wrote:
> On Mon, Nov 12, 2018 at 8:37 PM Hans Verkuil <[e-mail]> wrote:
>>
>> Hi Tomasz,
>>
>> A general note for the stateful and stateless patches: they describe specific
>> use-cases of the more generic Codec Interface, and as such should be one
>> level deeper in the section hierarchy.
> 
> I wonder what exactly this Codec Interface is. Is it a historical name
> for mem2mem? If so, perhaps it would make sense to rename it?

Yeah, it should be renamed to "Video Memory-to-Memory Interface", and the
codecs are just specific instances of such an interface.

> 
>>
>> I.e. instead of being section 4.6/7/8:
>>
>> https://hverkuil.home.xs4all.nl/request-api/uapi/v4l/devices.html
>>
>> they should be 4.5.1/2/3.
>>
> 
> FYI, the first RFC started like that, but it only made the spec
> difficult to navigate and the section numbers too long.
> 
> Still, no strong opinion. I'm okay moving it there, if you think it's better.

It should be moved and the interface name should be renamed. It makes a lot
more sense with those changes.

I've posted a patch for this.

> 
>> On 10/22/2018 04:48 PM, Tomasz Figa wrote:
>>> Due to complexity of the video decoding process, the V4L2 drivers of
>>> stateful decoder hardware require specific sequences of V4L2 API calls
>>> to be followed. These include capability enumeration, initialization,
>>> decoding, seek, pause, dynamic resolution change, drain and end of
>>> stream.
> [snipping any comments that I agree with]
>>> +
>>> +source height
>>> +   height in pixels for given source resolution; relevant to encoders only
>>> +
>>> +source resolution
>>> +   resolution in pixels of source frames being source to the encoder and
>>> +   subject to further cropping to the bounds of visible resolution; relevant to
>>> +   encoders only
>>> +
>>> +source width
>>> +   width in pixels for given source resolution; relevant to encoders only
>>
>> I would drop these three terms: they are not used in this document since this
>> describes a decoder and not an encoder.
>>
> 
> The glossary is shared between encoder and decoder, as suggested in
> previous round of review.
> 
> [snip]
>>> +
>>> +   * If width and height are set to non-zero values, the ``CAPTURE`` format
>>> +     will be updated with an appropriate frame buffer resolution instantly.
>>> +     However, for coded formats that include stream resolution information,
>>> +     after the decoder is done parsing the information from the stream, it will
>>> +     update the ``CAPTURE`` format with new values and signal a source change
>>> +     event.
>>
>> What if the initial width and height specified by userspace matches the parsed
>> width and height? Do you still get a source change event? I think you should
>> always get this event since there are other parameters that depend on the parsing
>> of the meta data.
>>
>> But that should be made explicit here.
>>
> 
> Yes, the change event should happen always after the driver determines
> the format of the stream. Will specify it explicitly.
> 
>>> +
>>> +   .. warning::
>>
>> I'd call this a note rather than a warning.
>>
> 
> I think it deserves at least the "important" level, since it informs
> about the side effects of the call, affecting any actions that the
> client might have done before.
> 
> 
>>> +
>>> +      Changing the ``OUTPUT`` format may change the currently set ``CAPTURE``
>>> +      format. The decoder will derive a new ``CAPTURE`` format from the
>>> +      ``OUTPUT`` format being set, including resolution, colorimetry
>>> +      parameters, etc. If the client needs a specific ``CAPTURE`` format, it
>>> +      must adjust it afterwards.
>>> +
>>> +3.  **Optional.** Query the minimum number of buffers required for ``OUTPUT``
>>> +    queue via :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to
>>> +    use more buffers than the minimum required by hardware/format.
>>
>> Why is this useful? As far as I can tell only the s5p-mfc *encoder* supports
>> this control, so this seems pointless. And since the output queue gets a bitstream
>> I don't see any reason for reading this control in a decoder.
>>
> 
> Indeed, querying this for bitstream buffers probably doesn't make much
> sense. I'll remove it.
> 
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``id``
>>> +          set to ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
>>> +
>>> +    * **Return fields:**
>>> +
>>> +      ``value``
>>> +          the minimum number of ``OUTPUT`` buffers required for the currently
>>> +          set format
>>> +
>>> +4.  Allocate source (bitstream) buffers via :c:func:`VIDIOC_REQBUFS` on
>>> +    ``OUTPUT``.
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``count``
>>> +          requested number of buffers to allocate; greater than zero
>>> +
>>> +      ``type``
>>> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
>>> +
>>> +      ``memory``
>>> +          follows standard semantics
>>> +
>>> +    * **Return fields:**
>>> +
>>> +      ``count``
>>> +          the actual number of buffers allocated
>>> +
>>> +    .. warning::
>>> +
>>> +       The actual number of allocated buffers may differ from the ``count``
>>> +       given. The client must check the updated value of ``count`` after the
>>> +       call returns.
>>> +
>>> +    .. note::
>>> +
>>> +       To allocate more than the minimum number of buffers (for pipeline
>>> +       depth), the client may query the ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
>>> +       control to get the minimum number of buffers required by the
>>> +       decoder/format, and pass the obtained value plus the number of
>>> +       additional buffers needed in the ``count`` field to
>>> +       :c:func:`VIDIOC_REQBUFS`.
>>
>> As mentioned above, this makes no sense for stateful decoders IMHO.
>>
> 
> Ack.
> 
>>> +
>>> +    Alternatively, :c:func:`VIDIOC_CREATE_BUFS` on the ``OUTPUT`` queue can be
>>> +    used to have more control over buffer allocation.
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``count``
>>> +          requested number of buffers to allocate; greater than zero
>>> +
>>> +      ``type``
>>> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
>>> +
>>> +      ``memory``
>>> +          follows standard semantics
>>> +
>>> +      ``format``
>>> +          follows standard semantics
>>> +
>>> +    * **Return fields:**
>>> +
>>> +      ``count``
>>> +          adjusted to the number of allocated buffers
>>> +
>>> +    .. warning::
>>> +
>>> +       The actual number of allocated buffers may differ from the ``count``
>>> +       given. The client must check the updated value of ``count`` after the
>>> +       call returns.
>>> +
>>> +5.  Start streaming on the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`.
>>> +
>>> +6.  **This step only applies to coded formats that contain resolution information
>>> +    in the stream.** Continue queuing/dequeuing bitstream buffers to/from the
>>> +    ``OUTPUT`` queue via :c:func:`VIDIOC_QBUF` and :c:func:`VIDIOC_DQBUF`. The
>>> +    buffers will be processed and returned to the client in order, until
>>> +    required metadata to configure the ``CAPTURE`` queue are found. This is
>>> +    indicated by the decoder sending a ``V4L2_EVENT_SOURCE_CHANGE`` event with
>>> +    ``V4L2_EVENT_SRC_CH_RESOLUTION`` source change type.
>>> +
>>> +    * It is not an error if the first buffer does not contain enough data for
>>> +      this to occur. Processing of the buffers will continue as long as more
>>> +      data is needed.
>>> +
>>> +    * If data in a buffer that triggers the event is required to decode the
>>> +      first frame, it will not be returned to the client, until the
>>> +      initialization sequence completes and the frame is decoded.
>>> +
>>> +    * If the client sets width and height of the ``OUTPUT`` format to 0,
>>> +      calling :c:func:`VIDIOC_G_FMT`, :c:func:`VIDIOC_S_FMT` or
>>> +      :c:func:`VIDIOC_TRY_FMT` on the ``CAPTURE`` queue will return the
>>> +      ``-EACCES`` error code, until the decoder configures ``CAPTURE`` format
>>> +      according to stream metadata.
>>> +
>>> +    .. important::
>>> +
>>> +       Any client query issued after the decoder queues the event will return
>>> +       values applying to the just parsed stream, including queue formats,
>>> +       selection rectangles and controls.
>>> +
>>> +    .. note::
>>> +
>>> +       A client capable of acquiring stream parameters from the bitstream on
>>> +       its own may attempt to set the width and height of the ``OUTPUT`` format
>>> +       to non-zero values matching the coded size of the stream, skip this step
>>> +       and continue with the `Capture setup` sequence. However, it must not
>>> +       rely on any driver queries regarding stream parameters, such as
>>> +       selection rectangles and controls, since the decoder has not parsed them
>>> +       from the stream yet. If the values configured by the client do not match
>>> +       those parsed by the decoder, a `Dynamic resolution change` will be
>>> +       triggered to reconfigure them.
>>> +
>>> +    .. note::
>>> +
>>> +       No decoded frames are produced during this phase.
>>> +
>>> +7.  Continue with the `Capture setup` sequence.
>>> +
>>> +Capture setup
>>> +=============
>>> +
>>> +1.  Call :c:func:`VIDIOC_G_FMT` on the ``CAPTURE`` queue to get format for the
>>> +    destination buffers parsed/decoded from the bitstream.
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``type``
>>> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
>>> +
>>> +    * **Return fields:**
>>> +
>>> +      ``width``, ``height``
>>> +          frame buffer resolution for the decoded frames
>>> +
>>> +      ``pixelformat``
>>> +          pixel format for decoded frames
>>> +
>>> +      ``num_planes`` (for _MPLANE ``type`` only)
>>> +          number of planes for pixelformat
>>> +
>>> +      ``sizeimage``, ``bytesperline``
>>> +          as per standard semantics; matching frame buffer format
>>> +
>>> +    .. note::
>>> +
>>> +       The value of ``pixelformat`` may be any pixel format supported by the
>>> +       decoder for the current stream. The decoder should choose a
>>> +       preferred/optimal format for the default configuration. For example, a
>>> +       YUV format may be preferred over an RGB format if an additional
>>> +       conversion step would be required for the latter.
>>> +
>>> +2.  **Optional.** Acquire the visible resolution via
>>> +    :c:func:`VIDIOC_G_SELECTION`.
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``type``
>>> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
>>> +
>>> +      ``target``
>>> +          set to ``V4L2_SEL_TGT_COMPOSE``
>>> +
>>> +    * **Return fields:**
>>> +
>>> +      ``r.left``, ``r.top``, ``r.width``, ``r.height``
>>> +          the visible rectangle; it must fit within the frame buffer resolution
>>> +          returned by :c:func:`VIDIOC_G_FMT` on ``CAPTURE``.
>>> +
>>> +    * The following selection targets are supported on ``CAPTURE``:
>>> +
>>> +      ``V4L2_SEL_TGT_CROP_BOUNDS``
>>> +          corresponds to the coded resolution of the stream
>>> +
>>> +      ``V4L2_SEL_TGT_CROP_DEFAULT``
>>> +          the rectangle covering the part of the ``CAPTURE`` buffer that
>>> +          contains meaningful picture data (visible area); width and height
>>> +          will be equal to the visible resolution of the stream
>>> +
>>> +      ``V4L2_SEL_TGT_CROP``
>>> +          the rectangle within the coded resolution to be output to
>>> +          ``CAPTURE``; defaults to ``V4L2_SEL_TGT_CROP_DEFAULT``; read-only on
>>> +          hardware without additional compose/scaling capabilities
>>> +
>>> +      ``V4L2_SEL_TGT_COMPOSE_BOUNDS``
>>> +          the maximum rectangle within a ``CAPTURE`` buffer, which the cropped
>>> +          frame can be output into; equal to ``V4L2_SEL_TGT_CROP`` if the
>>> +          hardware does not support compose/scaling
>>> +
>>> +      ``V4L2_SEL_TGT_COMPOSE_DEFAULT``
>>> +          equal to ``V4L2_SEL_TGT_CROP``
>>> +
>>> +      ``V4L2_SEL_TGT_COMPOSE``
>>> +          the rectangle inside a ``CAPTURE`` buffer into which the cropped
>>> +          frame is written; defaults to ``V4L2_SEL_TGT_COMPOSE_DEFAULT``;
>>> +          read-only on hardware without additional compose/scaling capabilities
>>> +
>>> +      ``V4L2_SEL_TGT_COMPOSE_PADDED``
>>> +          the rectangle inside a ``CAPTURE`` buffer which is overwritten by the
>>> +          hardware; equal to ``V4L2_SEL_TGT_COMPOSE`` if the hardware does not
>>> +          write padding pixels
>>> +
>>> +    .. warning::
>>> +
>>> +       The values are guaranteed to be meaningful only after the decoder
>>> +       successfully parses the stream metadata. The client must not rely on the
>>> +       query before that happens.
>>> +
>>> +3.  Query the minimum number of buffers required for the ``CAPTURE`` queue via
>>> +    :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to use more
>>> +    buffers than the minimum required by hardware/format.
>>
>> Is this step optional or required? Can it change when a resolution change occurs?
> 
> Probably not with a simple resolution change, but a case when a stream
> is changed on the fly would trigger what we call "resolution change"
> here, but what would effectively be a "source change" and that could
> include a change in the number of required CAPTURE buffers.
> 
>> How does this relate to the checks for the minimum number of buffers that REQBUFS
>> does?
> 
> The control returns the minimum that REQBUFS would allow, so the
> application can add few more buffers on top of that and improve the
> pipelining.
> 
>>
>> The 'This is useful if' sentence suggests that it is optional, but I think that
>> sentence just confuses the issue.
>>
> 
> It used to be optional and I didn't rephrase it after turning it into
> mandatory. How about:
> 
>     This enables the client to request more buffers
>     than the minimum required by hardware/format and achieve better pipelining.

Hmm, OK. It'll do, I guess. I never liked these MIN_BUFFERS controls, I wish they
would return something like the recommended number of buffers that will give you
decent performance.

> 
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``id``
>>> +          set to ``V4L2_CID_MIN_BUFFERS_FOR_CAPTURE``
>>> +
>>> +    * **Return fields:**
>>> +
>>> +      ``value``
>>> +          minimum number of buffers required to decode the stream parsed in
>>> +          this initialization sequence.
>>> +
>>> +    .. note::
>>> +
>>> +       The minimum number of buffers must be at least the number required to
>>> +       successfully decode the current stream. This may for example be the
>>> +       required DPB size for an H.264 stream given the parsed stream
>>> +       configuration (resolution, level).
>>> +
>>> +    .. warning::
>>> +
>>> +       The value is guaranteed to be meaningful only after the decoder
>>> +       successfully parses the stream metadata. The client must not rely on the
>>> +       query before that happens.
>>> +
>>> +4.  **Optional.** Enumerate ``CAPTURE`` formats via :c:func:`VIDIOC_ENUM_FMT` on
>>> +    the ``CAPTURE`` queue. Once the stream information is parsed and known, the
>>> +    client may use this ioctl to discover which raw formats are supported for
>>> +    given stream and select one of them via :c:func:`VIDIOC_S_FMT`.
>>
>> Can the list returned here differ from the list returned in the 'Querying capabilities'
>> step? If so, then I assume it will always be a subset of what was returned in
>> the 'Querying' step?
> 
> Depends on whether you're considering just VIDIOC_ENUM_FMT or also
> VIDIOC_G_FMT and VIDIOC_ENUM_FRAMESIZES.
> 
> The initial VIDIOC_ENUM_FMT has no way to account for any resolution
> constraints of the formats, so the list would include all raw pixel
> formats that the decoder can handle with selected coded pixel format.
> However, the list can be further narrowed down by using
> VIDIOC_ENUM_FRAMESIZES, to restrict each raw format only to the
> resolutions it can handle.
> 
> The VIDIOC_ENUM_FMT call in this sequence (after getting the stream
> information) would have the knowledge about the resolution, so the
> list returned here would only include the formats that can be actually
> handled. It should match the result of the initial query using both
> VIDIOC_ENUM_FMT and VIDIOC_ENUM_FRAMESIZES.

Right, so this will be a subset of the initial query taking the resolution
into account.

> 
>>
>>> +
>>> +    .. important::
>>> +
>>> +       The decoder will return only formats supported for the currently
>>> +       established coded format, as per the ``OUTPUT`` format and/or stream
>>> +       metadata parsed in this initialization sequence, even if more formats
>>> +       may be supported by the decoder in general.
>>> +
>>> +       For example, a decoder may support YUV and RGB formats for resolutions
>>> +       1920x1088 and lower, but only YUV for higher resolutions (due to
>>> +       hardware limitations). After parsing a resolution of 1920x1088 or lower,
>>> +       :c:func:`VIDIOC_ENUM_FMT` may return a set of YUV and RGB pixel formats,
>>> +       but after parsing resolution higher than 1920x1088, the decoder will not
>>> +       return RGB, unsupported for this resolution.
>>> +
>>> +       However, subsequent resolution change event triggered after
>>> +       discovering a resolution change within the same stream may switch
>>> +       the stream into a lower resolution and :c:func:`VIDIOC_ENUM_FMT`
>>> +       would return RGB formats again in that case.
>>> +
>>> +5.  **Optional.** Set the ``CAPTURE`` format via :c:func:`VIDIOC_S_FMT` on the
>>> +    ``CAPTURE`` queue. The client may choose a different format than
>>> +    selected/suggested by the decoder in :c:func:`VIDIOC_G_FMT`.
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``type``
>>> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
>>> +
>>> +      ``pixelformat``
>>> +          a raw pixel format
>>> +
>>> +    .. note::
>>> +
>>> +       The client may use :c:func:`VIDIOC_ENUM_FMT` after receiving the
>>> +       ``V4L2_EVENT_SOURCE_CHANGE`` event to find out the set of raw formats
>>> +       supported for the stream.
>>
>> Isn't this a duplicate of step 4? I think this note can be dropped.
>>
> 
> Ack.
> 
>>> +
>>> +6.  If all the following conditions are met, the client may resume the decoding
>>> +    instantly:
>>> +
>>> +    * ``sizeimage`` of the new format (determined in previous steps) is less
>>> +      than or equal to the size of currently allocated buffers,
>>> +
>>> +    * the number of buffers currently allocated is greater than or equal to the
>>> +      minimum number of buffers acquired in previous steps. To fulfill this
>>> +      requirement, the client may use :c:func:`VIDIOC_CREATE_BUFS` to add new
>>> +      buffers.
>>> +
>>> +    In such case, the remaining steps do not apply and the client may resume
>>> +    the decoding by one of the following actions:
>>> +
>>> +    * if the ``CAPTURE`` queue is streaming, call :c:func:`VIDIOC_DECODER_CMD`
>>> +      with the ``V4L2_DEC_CMD_START`` command,
>>> +
>>> +    * if the ``CAPTURE`` queue is not streaming, call :c:func:`VIDIOC_STREAMON`
>>> +      on the ``CAPTURE`` queue.
>>> +
>>> +    However, if the client intends to change the buffer set, to lower
>>> +    memory usage or for any other reasons, it may be achieved by following
>>> +    the steps below.
>>> +
>>> +7.  **If the** ``CAPTURE`` **queue is streaming,** keep queuing and dequeuing
>>> +    buffers on the ``CAPTURE`` queue until a buffer marked with the
>>> +    ``V4L2_BUF_FLAG_LAST`` flag is dequeued.
>>> +
>>> +8.  **If the** ``CAPTURE`` **queue is streaming,** call :c:func:`VIDIOC_STREAMOFF`
>>> +    on the ``CAPTURE`` queue to stop streaming.
>>> +
>>> +    .. warning::
>>> +
>>> +       The ``OUTPUT`` queue must remain streaming. Calling
>>> +       :c:func:`VIDIOC_STREAMOFF` on it would abort the sequence and trigger a
>>> +       seek.
>>> +
>>> +9.  **If the** ``CAPTURE`` **queue has buffers allocated,** free the ``CAPTURE``
>>> +    buffers using :c:func:`VIDIOC_REQBUFS`.
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``count``
>>> +          set to 0
>>> +
>>> +      ``type``
>>> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
>>> +
>>> +      ``memory``
>>> +          follows standard semantics
>>> +
>>> +10. Allocate ``CAPTURE`` buffers via :c:func:`VIDIOC_REQBUFS` on the
>>> +    ``CAPTURE`` queue.
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``count``
>>> +          requested number of buffers to allocate; greater than zero
>>> +
>>> +      ``type``
>>> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
>>> +
>>> +      ``memory``
>>> +          follows standard semantics
>>> +
>>> +    * **Return fields:**
>>> +
>>> +      ``count``
>>> +          actual number of buffers allocated
>>> +
>>> +    .. warning::
>>> +
>>> +       The actual number of allocated buffers may differ from the ``count``
>>> +       given. The client must check the updated value of ``count`` after the
>>> +       call returns.
>>> +
>>> +    .. note::
>>> +
>>> +       To allocate more than the minimum number of buffers (for pipeline
>>> +       depth), the client may query the ``V4L2_CID_MIN_BUFFERS_FOR_CAPTURE``
>>> +       control to get the minimum number of buffers required, and pass the
>>> +       obtained value plus the number of additional buffers needed in the
>>> +       ``count`` field to :c:func:`VIDIOC_REQBUFS`.
>>
>> Same question as before: is it optional or required to obtain the value of this
>> control? And can't the driver just set the min_buffers_needed field in the capture
>> vb2_queue to the minimum number of buffers that are required?
> 
> min_buffers_needed is the number of buffers that must be queued before
> start_streaming() can be called, so not relevant here.

Yeah, sorry about that. I clearly misunderstood this.

 The control is
> about the number of buffers to be allocated. Although the drivers must
> ensure that REQBUFS allocates the absolute minimum number of buffers
> for the decoding to be able to progress, the pipeline depth is
> something that the applications should control (e.g. depending on the
> consumers of the decoded buffers) and this is allowed by this control.
> 
>>
>> Should you be allowed to allocate buffers at all if the capture format isn't
>> known? I.e. width/height is still 0. It makes no sense to call REQBUFS since
>> there is no format size known that REQBUFS can use.
>>
> 
> Indeed, REQBUFS(CAPTURE) must not be allowed before the stream
> information is known (regardless of whether it comes from the OUTPUT
> format or is parsed from the stream). Let me add this to the related
> note in the Initialization sequence, which already includes
> VIDIOC_*_FMT.
> 
> For the Capture setup sequence, though, it's expected to happen when
> the stream information is already known, so I wouldn't change the
> description here.

Makes sense.

> 
>>> +
>>> +    Alternatively, :c:func:`VIDIOC_CREATE_BUFS` on the ``CAPTURE`` queue can be
>>> +    used to have more control over buffer allocation. For example, by
>>> +    allocating buffers larger than the current ``CAPTURE`` format, future
>>> +    resolution changes can be accommodated.
>>> +
>>> +    * **Required fields:**
>>> +
>>> +      ``count``
>>> +          requested number of buffers to allocate; greater than zero
>>> +
>>> +      ``type``
>>> +          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
>>> +
>>> +      ``memory``
>>> +          follows standard semantics
>>> +
>>> +      ``format``
>>> +          a format representing the maximum framebuffer resolution to be
>>> +          accommodated by newly allocated buffers
>>> +
>>> +    * **Return fields:**
>>> +
>>> +      ``count``
>>> +          adjusted to the number of allocated buffers
>>> +
>>> +    .. warning::
>>> +
>>> +       The actual number of allocated buffers may differ from the ``count``
>>> +       given. The client must check the updated value of ``count`` after the
>>> +       call returns.
>>> +
>>> +    .. note::
>>> +
>>> +       To allocate buffers for a format different than parsed from the stream
>>> +       metadata, the client must proceed as follows, before the metadata
>>> +       parsing is initiated:
>>> +
>>> +       * set width and height of the ``OUTPUT`` format to desired coded resolution to
>>> +         let the decoder configure the ``CAPTURE`` format appropriately,
>>> +
>>> +       * query the ``CAPTURE`` format using :c:func:`VIDIOC_G_FMT` and save it
>>> +         until this step.
>>> +
>>> +       The format obtained in the query may be then used with
>>> +       :c:func:`VIDIOC_CREATE_BUFS` in this step to allocate the buffers.
>>> +
>>> +11. Call :c:func:`VIDIOC_STREAMON` on the ``CAPTURE`` queue to start decoding
>>> +    frames.
>>> +
>>> +Decoding
>>> +========
>>> +
>>> +This state is reached after the `Capture setup` sequence finishes succesfully.
>>> +In this state, the client queues and dequeues buffers to both queues via
>>> +:c:func:`VIDIOC_QBUF` and :c:func:`VIDIOC_DQBUF`, following the standard
>>> +semantics.
>>> +
>>> +The contents of the source ``OUTPUT`` buffers depend on the active coded pixel
>>> +format and may be affected by codec-specific extended controls, as stated in
>>> +the documentation of each format.
>>> +
>>> +Both queues operate independently, following the standard behavior of V4L2
>>> +buffer queues and memory-to-memory devices. In addition, the order of decoded
>>> +frames dequeued from the ``CAPTURE`` queue may differ from the order of queuing
>>> +coded frames to the ``OUTPUT`` queue, due to properties of the selected coded
>>> +format, e.g. frame reordering.
>>> +
>>> +The client must not assume any direct relationship between ``CAPTURE``
>>> +and ``OUTPUT`` buffers and any specific timing of buffers becoming
>>> +available to dequeue. Specifically,
>>> +
>>> +* a buffer queued to ``OUTPUT`` may result in no buffers being produced
>>> +  on ``CAPTURE`` (e.g. if it does not contain encoded data, or if only
>>> +  metadata syntax structures are present in it),
>>> +
>>> +* a buffer queued to ``OUTPUT`` may result in more than 1 buffer produced
>>> +  on ``CAPTURE`` (if the encoded data contained more than one frame, or if
>>> +  returning a decoded frame allowed the decoder to return a frame that
>>> +  preceded it in decode, but succeeded it in the display order),
>>> +
>>> +* a buffer queued to ``OUTPUT`` may result in a buffer being produced on
>>> +  ``CAPTURE`` later into decode process, and/or after processing further
>>> +  ``OUTPUT`` buffers, or be returned out of order, e.g. if display
>>> +  reordering is used,
>>> +
>>> +* buffers may become available on the ``CAPTURE`` queue without additional
>>> +  buffers queued to ``OUTPUT`` (e.g. during drain or ``EOS``), because of the
>>> +  ``OUTPUT`` buffers queued in the past whose decoding results are only
>>> +  available at later time, due to specifics of the decoding process.
>>> +
>>> +.. note::
>>> +
>>> +   To allow matching decoded ``CAPTURE`` buffers with ``OUTPUT`` buffers they
>>> +   originated from, the client can set the ``timestamp`` field of the
>>> +   :c:type:`v4l2_buffer` struct when queuing an ``OUTPUT`` buffer. The
>>> +   ``CAPTURE`` buffer(s), which resulted from decoding that ``OUTPUT`` buffer
>>> +   will have their ``timestamp`` field set to the same value when dequeued.
>>> +
>>> +   In addition to the straighforward case of one ``OUTPUT`` buffer producing

straighforward -> straightforward

>>> +   one ``CAPTURE`` buffer, the following cases are defined:
>>> +
>>> +   * one ``OUTPUT`` buffer generates multiple ``CAPTURE`` buffers: the same
>>> +     ``OUTPUT`` timestamp will be copied to multiple ``CAPTURE`` buffers,
>>> +
>>> +   * multiple ``OUTPUT`` buffers generate one ``CAPTURE`` buffer: timestamp of
>>> +     the ``OUTPUT`` buffer queued last will be copied,
>>> +
>>> +   * the decoding order differs from the display order (i.e. the
>>> +     ``CAPTURE`` buffers are out-of-order compared to the ``OUTPUT`` buffers):
>>> +     ``CAPTURE`` timestamps will not retain the order of ``OUTPUT`` timestamps
>>> +     and thus monotonicity of the timestamps cannot be guaranteed.

I think this last point should be rewritten. The timestamp is just a value that
is copied, there are no monotonicity requirements for m2m devices in general.

>>
>> Should stateful codecs be required to support 'tags'? See:
>>
>> https://www.mail-archive.com/[e-mail]/msg136314.html
>>
>> To be honest, I'm inclined to require this for all m2m devices eventually.
>>
> 
> I guess this goes outside of the scope now, since we deferred tags.
> 
> Other than that, I would indeed make all m2m devices support tags,
> since it shouldn't differ from the timestamp copy feature as we have
> now.

I agree.

> 
>>> +
>>> +During the decoding, the decoder may initiate one of the special sequences, as
>>> +listed below. The sequences will result in the decoder returning all the
>>> +``CAPTURE`` buffers that originated from all the ``OUTPUT`` buffers processed
>>> +before the sequence started. Last of the buffers will have the
>>> +``V4L2_BUF_FLAG_LAST`` flag set. To determine the sequence to follow, the client
>>> +must check if there is any pending event and,
>>> +
>>> +* if a ``V4L2_EVENT_SOURCE_CHANGE`` event is pending, the `Dynamic resolution
>>> +  change` sequence needs to be followed,
>>> +
>>> +* if a ``V4L2_EVENT_EOS`` event is pending, the `End of stream` sequence needs
>>> +  to be followed.
>>> +
>>> +Some of the sequences can be intermixed with each other and need to be handled
>>> +as they happen. The exact operation is documented for each sequence.
>>> +
>>> +Seek
>>> +====
>>> +
>>> +Seek is controlled by the ``OUTPUT`` queue, as it is the source of coded data.
>>> +The seek does not require any specific operation on the ``CAPTURE`` queue, but
>>> +it may be affected as per normal decoder operation.
>>> +
>>> +1. Stop the ``OUTPUT`` queue to begin the seek sequence via
>>> +   :c:func:`VIDIOC_STREAMOFF`.
>>> +
>>> +   * **Required fields:**
>>> +
>>> +     ``type``
>>> +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
>>> +
>>> +   * The decoder will drop all the pending ``OUTPUT`` buffers and they must be
>>> +     treated as returned to the client (following standard semantics).
>>> +
>>> +2. Restart the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`
>>> +
>>> +   * **Required fields:**
>>> +
>>> +     ``type``
>>> +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
>>> +
>>> +   * The decoder will start accepting new source bitstream buffers after the
>>> +     call returns.
>>> +
>>> +3. Start queuing buffers containing coded data after the seek to the ``OUTPUT``
>>> +   queue until a suitable resume point is found.
>>> +
>>> +   .. note::
>>> +
>>> +      There is no requirement to begin queuing coded data starting exactly
>>> +      from a resume point (e.g. SPS or a keyframe). Any queued ``OUTPUT``
>>> +      buffers will be processed and returned to the client until a suitable
>>> +      resume point is found.  While looking for a resume point, the decoder
>>> +      should not produce any decoded frames into ``CAPTURE`` buffers.
>>> +
>>> +      Some hardware is known to mishandle seeks to a non-resume point. Such an
>>> +      operation may result in an unspecified number of corrupted decoded frames
>>> +      being made available on the ``CAPTURE`` queue. Drivers must ensure that
>>> +      no fatal decoding errors or crashes occur, and implement any necessary
>>> +      handling and workarounds for hardware issues related to seek operations.
>>
>> Is there a requirement that those corrupted frames have V4L2_BUF_FLAG_ERROR set?
>> I.e., can userspace detect those currupted frames?
>>
> 
> I think the question is whether the kernel driver can actually detect
> those corrupted frames. We can't guarantee reporting errors to the
> userspace, if the hardware doesn't actually report them.
> 
> Could we perhaps keep this an open question and possibly address with
> some extension that could be an opt in for the decoders that can
> report errors?

Hmm, how about: If the hardware can detect such corrupted decoded frames, then
it shall set V4L2_BUF_FLAG_ERROR.

> 
>>> +
>>> +   .. warning::
>>> +
>>> +      In case of the H.264 codec, the client must take care not to seek over a
>>> +      change of SPS/PPS. Even though the target frame could be a keyframe, the
>>> +      stale SPS/PPS inside decoder state would lead to undefined results when
>>> +      decoding. Although the decoder must handle such case without a crash or a
>>> +      fatal decode error, the client must not expect a sensible decode output.
>>> +
>>> +4. After a resume point is found, the decoder will start returning ``CAPTURE``
>>> +   buffers containing decoded frames.
>>> +
>>> +.. important::
>>> +
>>> +   A seek may result in the `Dynamic resolution change` sequence being
>>> +   iniitated, due to the seek target having decoding parameters different from
>>> +   the part of the stream decoded before the seek. The sequence must be handled
>>> +   as per normal decoder operation.
>>> +
>>> +.. warning::
>>> +
>>> +   It is not specified when the ``CAPTURE`` queue starts producing buffers
>>> +   containing decoded data from the ``OUTPUT`` buffers queued after the seek,
>>> +   as it operates independently from the ``OUTPUT`` queue.
>>> +
>>> +   The decoder may return a number of remaining ``CAPTURE`` buffers containing
>>> +   decoded frames originating from the ``OUTPUT`` buffers queued before the
>>> +   seek sequence is performed.
>>> +
>>> +   The ``VIDIOC_STREAMOFF`` operation discards any remaining queued
>>> +   ``OUTPUT`` buffers, which means that not all of the ``OUTPUT`` buffers
>>> +   queued before the seek sequence may have matching ``CAPTURE`` buffers
>>> +   produced.  For example, given the sequence of operations on the
>>> +   ``OUTPUT`` queue:
>>> +
>>> +     QBUF(A), QBUF(B), STREAMOFF(), STREAMON(), QBUF(G), QBUF(H),
>>> +
>>> +   any of the following results on the ``CAPTURE`` queue is allowed:
>>> +
>>> +     {A’, B’, G’, H’}, {A’, G’, H’}, {G’, H’}.
>>
>> Isn't it the case that if you would want to avoid that, then you should call
>> DECODER_STOP, wait for the last buffer on the CAPTURE queue, then seek and
>> call DECODER_START. If you do that, then you should always get {A’, B’, G’, H’}.
>> (basically following the Drain sequence).
> 
> Yes, it is, but I think it depends on the application needs. Here we
> just give a primitive to change the place in the stream that's being
> decoded (or change the stream on the fly).
> 
> Actually, with the timestamp copy, I guess we wouldn't even need to do
> the DECODER_STOP, as we could just discard the CAPTURE buffers until
> we get one that matches the timestamp of the first OUTPUT buffer after
> the seek.
> 
>>
>> Admittedly, you typically want to do an instantaneous seek, so this is probably
>> not what you want to do normally.
>>
>> It might help to have this documented in a separate note.
>>
> 
> The instantaneous seek is documented below. I'm not sure if there is
> any practical need to document the other case, but I could add a
> sentence like below to the warning above. What do you think?
> 
>    The ``VIDIOC_STREAMOFF`` operation discards any remaining queued
>    ``OUTPUT`` buffers, which means that not all of the ``OUTPUT`` buffers
>    queued before the seek sequence may have matching ``CAPTURE`` buffers
>    produced.  For example, given the sequence of operations on the
>    ``OUTPUT`` queue:
> 
>      QBUF(A), QBUF(B), STREAMOFF(), STREAMON(), QBUF(G), QBUF(H),
> 
>    any of the following results on the ``CAPTURE`` queue is allowed:

is allowed -> are allowed

> 
>      {A’, B’, G’, H’}, {A’, G’, H’}, {G’, H’}.
> 
>    To determine the CAPTURE buffer containing the first decoded frame
> after the seek,
>    the client may observe the timestamps to match the CAPTURE and OUTPUT buffers
>    or use V4L2_DEC_CMD_STOP and V4L2_DEC_CMD_START to drain the decoder.

Ack.

> 
>>> +
>>> +.. note::
>>> +
>>> +   To achieve instantaneous seek, the client may restart streaming on the
>>> +   ``CAPTURE`` queue too to discard decoded, but not yet dequeued buffers.
>>> +
>>> +Dynamic resolution change
>>> +=========================
>>> +
>>> +Streams that include resolution metadata in the bitstream may require switching
>>> +to a different resolution during the decoding.
>>> +
>>> +The sequence starts when the decoder detects a coded frame with one or more of
>>> +the following parameters different from previously established (and reflected
>>> +by corresponding queries):
>>> +
>>> +* coded resolution (``OUTPUT`` width and height),
>>> +
>>> +* visible resolution (selection rectangles),
>>> +
>>> +* the minimum number of buffers needed for decoding.
>>> +
>>> +Whenever that happens, the decoder must proceed as follows:
>>> +
>>> +1.  After encountering a resolution change in the stream, the decoder sends a
>>> +    ``V4L2_EVENT_SOURCE_CHANGE`` event with source change type set to
>>> +    ``V4L2_EVENT_SRC_CH_RESOLUTION``.
>>> +
>>> +    .. important::
>>> +
>>> +       Any client query issued after the decoder queues the event will return
>>> +       values applying to the stream after the resolution change, including
>>> +       queue formats, selection rectangles and controls.
>>> +
>>> +2.  The decoder will then process and decode all remaining buffers from before
>>> +    the resolution change point.
>>> +
>>> +    * The last buffer from before the change must be marked with the
>>> +      ``V4L2_BUF_FLAG_LAST`` flag, similarly to the `Drain` sequence above.
>>> +
>>> +    .. warning::
>>> +
>>> +       The last buffer may be empty (with :c:type:`v4l2_buffer` ``bytesused``
>>> +       = 0) and in such case it must be ignored by the client, as it does not
>>> +       contain a decoded frame.
>>> +
>>> +    .. note::
>>> +
>>> +       Any attempt to dequeue more buffers beyond the buffer marked with
>>> +       ``V4L2_BUF_FLAG_LAST`` will result in a -EPIPE error from
>>> +       :c:func:`VIDIOC_DQBUF`.
>>> +
>>> +The client must continue the sequence as described below to continue the
>>> +decoding process.
>>> +
>>> +1.  Dequeue the source change event.
>>> +
>>> +    .. important::
>>> +
>>> +       A source change triggers an implicit decoder drain, similar to the
>>> +       explicit `Drain` sequence. The decoder is stopped after it completes.
>>> +       The decoding process must be resumed with either a pair of calls to
>>> +       :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
>>> +       ``CAPTURE`` queue, or a call to :c:func:`VIDIOC_DECODER_CMD` with the
>>> +       ``V4L2_DEC_CMD_START`` command.
>>> +
>>> +2.  Continue with the `Capture setup` sequence.
>>> +
>>> +.. note::
>>> +
>>> +   During the resolution change sequence, the ``OUTPUT`` queue must remain
>>> +   streaming. Calling :c:func:`VIDIOC_STREAMOFF` on the ``OUTPUT`` queue would
>>> +   abort the sequence and initiate a seek.
>>> +
>>> +   In principle, the ``OUTPUT`` queue operates separately from the ``CAPTURE``
>>> +   queue and this remains true for the duration of the entire resolution change
>>> +   sequence as well.
>>> +
>>> +   The client should, for best performance and simplicity, keep queuing/dequeuing
>>> +   buffers to/from the ``OUTPUT`` queue even while processing this sequence.
>>> +
>>> +Drain
>>> +=====
>>> +
>>> +To ensure that all queued ``OUTPUT`` buffers have been processed and related
>>> +``CAPTURE`` buffers output to the client, the client must follow the drain
>>> +sequence described below. After the drain sequence ends, the client has
>>> +received all decoded frames for all ``OUTPUT`` buffers queued before the
>>> +sequence was started.
>>> +
>>> +1. Begin drain by issuing :c:func:`VIDIOC_DECODER_CMD`.
>>> +
>>> +   * **Required fields:**
>>> +
>>> +     ``cmd``
>>> +         set to ``V4L2_DEC_CMD_STOP``
>>> +
>>> +     ``flags``
>>> +         set to 0
>>> +
>>> +     ``pts``
>>> +         set to 0
>>> +
>>> +   .. warning::
>>> +
>>> +   The sentence can be only initiated if both ``OUTPUT`` and ``CAPTURE`` queues
>>
>> 'sentence'? You mean 'decoder command'?
> 
> Sequence. :)

Ah, that makes a lot more sense!

> 
>>
>>> +   are streaming. For compatibility reasons, the call to
>>> +   :c:func:`VIDIOC_DECODER_CMD` will not fail even if any of the queues is not
>>> +   streaming, but at the same time it will not initiate the `Drain` sequence
>>> +   and so the steps described below would not be applicable.
>>> +
>>> +2. Any ``OUTPUT`` buffers queued by the client before the
>>> +   :c:func:`VIDIOC_DECODER_CMD` was issued will be processed and decoded as
>>> +   normal. The client must continue to handle both queues independently,
>>> +   similarly to normal decode operation. This includes,
>>> +
>>> +   * handling any operations triggered as a result of processing those buffers,
>>> +     such as the `Dynamic resolution change` sequence, before continuing with
>>> +     the drain sequence,
>>> +
>>> +   * queuing and dequeuing ``CAPTURE`` buffers, until a buffer marked with the
>>> +     ``V4L2_BUF_FLAG_LAST`` flag is dequeued,
>>> +
>>> +     .. warning::
>>> +
>>> +        The last buffer may be empty (with :c:type:`v4l2_buffer`
>>> +        ``bytesused`` = 0) and in such case it must be ignored by the client,
>>> +        as it does not contain a decoded frame.
>>> +
>>> +     .. note::
>>> +
>>> +        Any attempt to dequeue more buffers beyond the buffer marked with
>>> +        ``V4L2_BUF_FLAG_LAST`` will result in a -EPIPE error from
>>> +        :c:func:`VIDIOC_DQBUF`.
>>> +
>>> +   * dequeuing processed ``OUTPUT`` buffers, until all the buffers queued
>>> +     before the ``V4L2_DEC_CMD_STOP`` command are dequeued.
>>> +
>>> +   * dequeuing the ``V4L2_EVENT_EOS`` event, if the client subscribed to it.
>>> +
>>> +   .. note::
>>> +
>>> +      For backwards compatibility, the decoder will signal a ``V4L2_EVENT_EOS``
>>> +      event when the last the last frame has been decoded and all frames are
>>
>> 'the last the last' -> the last
> 
> Ack.
> 
>>
>>> +      ready to be dequeued. It is a deprecated behavior and the client must not
>>> +      rely on it. The ``V4L2_BUF_FLAG_LAST`` buffer flag should be used
>>> +      instead.
>>> +
>>> +3. Once all the ``OUTPUT`` buffers queued before the ``V4L2_DEC_CMD_STOP`` call
>>> +   and the last ``CAPTURE`` buffer are dequeued, the decoder is stopped and it
>>
>> This sentence is a bit confusing. This is better IMHO:
>>
>> 3. Once all the ``OUTPUT`` buffers queued before the ``V4L2_DEC_CMD_STOP`` call
>>    are dequeued and the last ``CAPTURE`` buffer is dequeued, the decoder is stopped and it
>>
> 
> Ack.
> 
>>> +   will accept, but not process any newly queued ``OUTPUT`` buffers until the
>>
>> process any -> process, any
>>
> 
> Ack.
> 
>>> +   client issues any of the following operations:
>>> +
>>> +   * ``V4L2_DEC_CMD_START`` - the decoder will resume the operation normally,
>>> +
>>> +   * a pair of :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
>>> +     ``CAPTURE`` queue - the decoder will resume the operation normally,
>>> +     however any ``CAPTURE`` buffers still in the queue will be returned to the
>>> +     client,
>>> +
>>> +   * a pair of :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
>>> +     ``OUTPUT`` queue - any pending source buffers will be returned to the
>>> +     client and the `Seek` sequence will be triggered.
>>> +
>>> +.. note::
>>> +
>>> +   Once the drain sequence is initiated, the client needs to drive it to
>>> +   completion, as described by the steps above, unless it aborts the process by
>>> +   issuing :c:func:`VIDIOC_STREAMOFF` on any of the ``OUTPUT`` or ``CAPTURE``
>>> +   queues.  The client is not allowed to issue ``V4L2_DEC_CMD_START`` or
>>> +   ``V4L2_DEC_CMD_STOP`` again while the drain sequence is in progress and they
>>> +   will fail with -EBUSY error code if attempted.
>>> +
>>> +   Although mandatory, the availability of decoder commands may be queried
>>> +   using :c:func:`VIDIOC_TRY_DECODER_CMD`.
>>> +
>>> +End of stream
>>> +=============
>>> +
>>> +If the decoder encounters an end of stream marking in the stream, the decoder
>>> +will initiate the `Drain` sequence, which the client must handle as described
>>> +above, skipping the initial :c:func:`VIDIOC_DECODER_CMD`.
>>> +
>>> +Commit points
>>> +=============
>>> +
>>> +Setting formats and allocating buffers trigger changes in the behavior of the
>>> +decoder.
>>> +
>>> +1. Setting the format on the ``OUTPUT`` queue may change the set of formats
>>> +   supported/advertised on the ``CAPTURE`` queue. In particular, it also means
>>> +   that the ``CAPTURE`` format may be reset and the client must not rely on the
>>> +   previously set format being preserved.
>>> +
>>> +2. Enumerating formats on the ``CAPTURE`` queue always returns only formats
>>> +   supported for the current ``OUTPUT`` format.
>>> +
>>> +3. Setting the format on the ``CAPTURE`` queue does not change the list of
>>> +   formats available on the ``OUTPUT`` queue. An attempt to set the ``CAPTURE``
>>> +   format that is not supported for the currently selected ``OUTPUT`` format
>>> +   will result in the decoder adjusting the requested ``CAPTURE`` format to a
>>> +   supported one.
>>> +
>>> +4. Enumerating formats on the ``OUTPUT`` queue always returns the full set of
>>> +   supported coded formats, irrespectively of the current ``CAPTURE`` format.
>>> +
>>> +5. While buffers are allocated on the ``OUTPUT`` queue, the client must not
>>> +   change the format on the queue. Drivers will return the -EBUSY error code
>>> +   for any such format change attempt.
>>> +
>>> +To summarize, setting formats and allocation must always start with the
>>> +``OUTPUT`` queue and the ``OUTPUT`` queue is the master that governs the
>>> +set of supported formats for the ``CAPTURE`` queue.
>>> diff --git a/Documentation/media/uapi/v4l/devices.rst b/Documentation/media/uapi/v4l/devices.rst
>>> index fb7f8c26cf09..12d43fe711cf 100644
>>> --- a/Documentation/media/uapi/v4l/devices.rst
>>> +++ b/Documentation/media/uapi/v4l/devices.rst
>>> @@ -15,6 +15,7 @@ Interfaces
>>>      dev-output
>>>      dev-osd
>>>      dev-codec
>>> +    dev-decoder
>>>      dev-effect
>>>      dev-raw-vbi
>>>      dev-sliced-vbi
>>> diff --git a/Documentation/media/uapi/v4l/pixfmt-v4l2.rst b/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
>>> index 826f2305da01..ca5f2270a829 100644
>>> --- a/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
>>> +++ b/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
>>> @@ -32,6 +32,11 @@ Single-planar format structure
>>>       to a multiple of the scale factor of any smaller planes. For
>>>       example when the image format is YUV 4:2:0, ``width`` and
>>>       ``height`` must be multiples of two.
>>> +
>>> +     For compressed formats that contain the resolution information encoded
>>> +     inside the stream, when fed to a stateful mem2mem decoder, the fields
>>> +     may be zero to rely on the decoder to detect the right values. For more
>>> +     details see :ref:`decoder` and format descriptions.
>>>      * - __u32
>>>        - ``pixelformat``
>>>        - The pixel format or type of compression, set by the application.
>>> diff --git a/Documentation/media/uapi/v4l/v4l2.rst b/Documentation/media/uapi/v4l/v4l2.rst
>>> index b89e5621ae69..65dc096199ad 100644
>>> --- a/Documentation/media/uapi/v4l/v4l2.rst
>>> +++ b/Documentation/media/uapi/v4l/v4l2.rst
>>> @@ -53,6 +53,10 @@ Authors, in alphabetical order:
>>>
>>>    - Original author of the V4L2 API and documentation.
>>>
>>> +- Figa, Tomasz <[e-mail]>
>>> +
>>> +  - Documented the memory-to-memory decoder interface.
>>> +
>>>  - H Schimek, Michael <[e-mail]>
>>>
>>>    - Original author of the V4L2 API and documentation.
>>> @@ -61,6 +65,10 @@ Authors, in alphabetical order:
>>>
>>>    - Documented the Digital Video timings API.
>>>
>>> +- Osciak, Pawel <[e-mail]>
>>> +
>>> +  - Documented the memory-to-memory decoder interface.
>>> +
>>>  - Osciak, Pawel <[e-mail]>
>>>
>>>    - Designed and documented the multi-planar API.
>>> @@ -85,7 +93,7 @@ Authors, in alphabetical order:
>>>
>>>    - Designed and documented the VIDIOC_LOG_STATUS ioctl, the extended control ioctls, major parts of the sliced VBI API, the MPEG encoder and decoder APIs and the DV Timings API.
>>>
>>> -**Copyright** |copy| 1999-2016: Bill Dirks, Michael H. Schimek, Hans Verkuil, Martin Rubli, Andy Walls, Muralidharan Karicheri, Mauro Carvalho Chehab, Pawel Osciak, Sakari Ailus & Antti Palosaari.
>>> +**Copyright** |copy| 1999-2018: Bill Dirks, Michael H. Schimek, Hans Verkuil, Martin Rubli, Andy Walls, Muralidharan Karicheri, Mauro Carvalho Chehab, Pawel Osciak, Sakari Ailus & Antti Palosaari, Tomasz Figa
>>>
>>>  Except when explicitly stated as GPL, programming examples within this
>>>  part can be used and distributed without restrictions.
>>> diff --git a/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst b/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
>>> index 85c916b0ce07..2f73fe22a9cd 100644
>>> --- a/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
>>> +++ b/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
>>> @@ -49,14 +49,16 @@ The ``cmd`` field must contain the command code. Some commands use the
>>>
>>>  A :ref:`write() <func-write>` or :ref:`VIDIOC_STREAMON`
>>>  call sends an implicit START command to the decoder if it has not been
>>> -started yet.
>>> +started yet. Applies to both queues of mem2mem decoders.
>>>
>>>  A :ref:`close() <func-close>` or :ref:`VIDIOC_STREAMOFF <VIDIOC_STREAMON>`
>>>  call of a streaming file descriptor sends an implicit immediate STOP
>>> -command to the decoder, and all buffered data is discarded.
>>> +command to the decoder, and all buffered data is discarded. Applies to both
>>> +queues of mem2mem decoders.
>>>
>>> -These ioctls are optional, not all drivers may support them. They were
>>> -introduced in Linux 3.3.
>>> +In principle, these ioctls are optional, not all drivers may support them. They were
>>> +introduced in Linux 3.3. They are, however, mandatory for stateful mem2mem decoders
>>> +(as further documented in :ref:`decoder`).
>>>
>>>
>>>  .. tabularcolumns:: |p{1.1cm}|p{2.4cm}|p{1.2cm}|p{1.6cm}|p{10.6cm}|
>>> @@ -160,26 +162,36 @@ introduced in Linux 3.3.
>>>       ``V4L2_DEC_CMD_RESUME`` for that. This command has one flag:
>>>       ``V4L2_DEC_CMD_START_MUTE_AUDIO``. If set, then audio will be
>>>       muted when playing back at a non-standard speed.
>>> +
>>> +     For stateful mem2mem decoders, the command may be also used to restart
>>> +     the decoder in case of an implicit stop initiated by the decoder
>>> +     itself, without the ``V4L2_DEC_CMD_STOP`` being called explicitly.
>>> +     No flags or other arguments are accepted in case of mem2mem decoders.
>>> +     See :ref:`decoder` for more details.
>>>      * - ``V4L2_DEC_CMD_STOP``
>>>        - 1
>>>        - Stop the decoder. When the decoder is already stopped, this
>>>       command does nothing. This command has two flags: if
>>>       ``V4L2_DEC_CMD_STOP_TO_BLACK`` is set, then the decoder will set
>>>       the picture to black after it stopped decoding. Otherwise the last
>>> -     image will repeat. mem2mem decoders will stop producing new frames
>>> -     altogether. They will send a ``V4L2_EVENT_EOS`` event when the
>>> -     last frame has been decoded and all frames are ready to be
>>> -     dequeued and will set the ``V4L2_BUF_FLAG_LAST`` buffer flag on
>>> -     the last buffer of the capture queue to indicate there will be no
>>> -     new buffers produced to dequeue. This buffer may be empty,
>>> -     indicated by the driver setting the ``bytesused`` field to 0. Once
>>> -     the ``V4L2_BUF_FLAG_LAST`` flag was set, the
>>> -     :ref:`VIDIOC_DQBUF <VIDIOC_QBUF>` ioctl will not block anymore,
>>> -     but return an ``EPIPE`` error code. If
>>> +     image will repeat. If
>>>       ``V4L2_DEC_CMD_STOP_IMMEDIATELY`` is set, then the decoder stops
>>>       immediately (ignoring the ``pts`` value), otherwise it will keep
>>>       decoding until timestamp >= pts or until the last of the pending
>>>       data from its internal buffers was decoded.
>>> +
>>> +     A stateful mem2mem decoder will proceed with decoding the source
>>> +     buffers pending before the command is issued and then stop producing
>>> +     new frames. It will send a ``V4L2_EVENT_EOS`` event when the last frame
>>> +     has been decoded and all frames are ready to be dequeued and will set
>>> +     the ``V4L2_BUF_FLAG_LAST`` buffer flag on the last buffer of the
>>> +     capture queue to indicate there will be no new buffers produced to
>>> +     dequeue. This buffer may be empty, indicated by the driver setting the
>>> +     ``bytesused`` field to 0. Once the buffer with the
>>> +     ``V4L2_BUF_FLAG_LAST`` flag set was dequeued, the :ref:`VIDIOC_DQBUF
>>> +     <VIDIOC_QBUF>` ioctl will not block anymore, but return an ``EPIPE``
>>> +     error code. No flags or other arguments are accepted in case of mem2mem
>>> +     decoders.  See :ref:`decoder` for more details.
>>>      * - ``V4L2_DEC_CMD_PAUSE``
>>>        - 2
>>>        - Pause the decoder. When the decoder has not been started yet, the
>>> diff --git a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
>>> index 3ead350e099f..0fc0b78a943e 100644
>>> --- a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
>>> +++ b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
>>> @@ -53,6 +53,13 @@ devices that is either the struct
>>>  member. When the requested buffer type is not supported drivers return
>>>  an ``EINVAL`` error code.
>>>
>>> +A stateful mem2mem decoder will not allow operations on the
>>> +``V4L2_BUF_TYPE_VIDEO_CAPTURE`` or ``V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE``
>>> +buffer type until the corresponding ``V4L2_BUF_TYPE_VIDEO_OUTPUT`` or
>>> +``V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE`` buffer type is configured. If such an
>>> +operation is attempted, drivers return an ``EACCES`` error code. Refer to
>>> +:ref:`decoder` for more details.
>>
>> This isn't right. EACCES is returned as long as the output format resolution is
>> unknown. If it is set explicitly, then this will work without an error.

Ah, sorry, I phrased that poorly. Let me try again:

This isn't right. EACCES is returned for CAPTURE operations as long as the
output format resolution is unknown or the CAPTURE format is explicitly set.
If the CAPTURE format is set explicitly, then this will work without an error.

> 
> I think that's what is written above. The stream resolution is known
> when the applicable OUTPUT queue is configured, which lets the driver
> determine the format constraints on the applicable CAPTURE queue. If
> it's not clear, could you help rephrasing?
> 
>>
>>> +
>>>  To change the current format parameters applications initialize the
>>>  ``type`` field and all fields of the respective ``fmt`` union member.
>>>  For details see the documentation of the various devices types in
>>> @@ -145,6 +152,13 @@ On success 0 is returned, on error -1 and the ``errno`` variable is set
>>>  appropriately. The generic error codes are described at the
>>>  :ref:`Generic Error Codes <gen-errors>` chapter.
>>>
>>> +EACCES
>>> +    The format is not accessible until another buffer type is configured.
>>> +    Relevant for the V4L2_BUF_TYPE_VIDEO_CAPTURE and
>>> +    V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE buffer types of mem2mem decoders, which
>>> +    require the format of V4L2_BUF_TYPE_VIDEO_OUTPUT or
>>> +    V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE buffer type to be configured first.
>>
>> Ditto.
> 
> Ditto.
> 
> Best regards,
> Tomasz
> 

Regards,

	Hans

Tomasz Figa Jan. 23, 2019, 5:27 a.m. UTC | #12

On Tue, Jan 22, 2019 at 11:47 PM Hans Verkuil <[e-mail]> wrote:
>
> On 01/22/19 11:02, Tomasz Figa wrote:
> > On Mon, Nov 12, 2018 at 8:37 PM Hans Verkuil <[e-mail]> wrote:
> >>
> >> Hi Tomasz,
> >>
> >> A general note for the stateful and stateless patches: they describe specific
> >> use-cases of the more generic Codec Interface, and as such should be one
> >> level deeper in the section hierarchy.
> >
> > I wonder what exactly this Codec Interface is. Is it a historical name
> > for mem2mem? If so, perhaps it would make sense to rename it?
>
> Yeah, it should be renamed to "Video Memory-to-Memory Interface", and the
> codecs are just specific instances of such an interface.
>

Ack.

> >
> >>
> >> I.e. instead of being section 4.6/7/8:
> >>
> >> https://hverkuil.home.xs4all.nl/request-api/uapi/v4l/devices.html
> >>
> >> they should be 4.5.1/2/3.
> >>
> >
> > FYI, the first RFC started like that, but it only made the spec
> > difficult to navigate and the section numbers too long.
> >
> > Still, no strong opinion. I'm okay moving it there, if you think it's better.
>
> It should be moved and the interface name should be renamed. It makes a lot
> more sense with those changes.
>
> I've posted a patch for this.
>

Thanks. I've rebased on top of it.

[snip]
> >>> +3.  Query the minimum number of buffers required for the ``CAPTURE`` queue via
> >>> +    :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to use more
> >>> +    buffers than the minimum required by hardware/format.
> >>
> >> Is this step optional or required? Can it change when a resolution change occurs?
> >
> > Probably not with a simple resolution change, but a case when a stream
> > is changed on the fly would trigger what we call "resolution change"
> > here, but what would effectively be a "source change" and that could
> > include a change in the number of required CAPTURE buffers.
> >
> >> How does this relate to the checks for the minimum number of buffers that REQBUFS
> >> does?
> >
> > The control returns the minimum that REQBUFS would allow, so the
> > application can add few more buffers on top of that and improve the
> > pipelining.
> >
> >>
> >> The 'This is useful if' sentence suggests that it is optional, but I think that
> >> sentence just confuses the issue.
> >>
> >
> > It used to be optional and I didn't rephrase it after turning it into
> > mandatory. How about:
> >
> >     This enables the client to request more buffers
> >     than the minimum required by hardware/format and achieve better pipelining.
>
> Hmm, OK. It'll do, I guess. I never liked these MIN_BUFFERS controls, I wish they
> would return something like the recommended number of buffers that will give you
> decent performance.
>

The problem here is that the kernel doesn't know what is decent for
the application, since it doesn't know how the results of the decoding
are used. Over-allocating would result to a waste of memory, which
could then make it less than decent for memory-constrained
applications.

> >
> >>> +
> >>> +    * **Required fields:**
> >>> +
> >>> +      ``id``
> >>> +          set to ``V4L2_CID_MIN_BUFFERS_FOR_CAPTURE``
> >>> +
> >>> +    * **Return fields:**
> >>> +
> >>> +      ``value``
> >>> +          minimum number of buffers required to decode the stream parsed in
> >>> +          this initialization sequence.
> >>> +
> >>> +    .. note::
> >>> +
> >>> +       The minimum number of buffers must be at least the number required to
> >>> +       successfully decode the current stream. This may for example be the
> >>> +       required DPB size for an H.264 stream given the parsed stream
> >>> +       configuration (resolution, level).
> >>> +
> >>> +    .. warning::
> >>> +
> >>> +       The value is guaranteed to be meaningful only after the decoder
> >>> +       successfully parses the stream metadata. The client must not rely on the
> >>> +       query before that happens.
> >>> +
> >>> +4.  **Optional.** Enumerate ``CAPTURE`` formats via :c:func:`VIDIOC_ENUM_FMT` on
> >>> +    the ``CAPTURE`` queue. Once the stream information is parsed and known, the
> >>> +    client may use this ioctl to discover which raw formats are supported for
> >>> +    given stream and select one of them via :c:func:`VIDIOC_S_FMT`.
> >>
> >> Can the list returned here differ from the list returned in the 'Querying capabilities'
> >> step? If so, then I assume it will always be a subset of what was returned in
> >> the 'Querying' step?
> >
> > Depends on whether you're considering just VIDIOC_ENUM_FMT or also
> > VIDIOC_G_FMT and VIDIOC_ENUM_FRAMESIZES.
> >
> > The initial VIDIOC_ENUM_FMT has no way to account for any resolution
> > constraints of the formats, so the list would include all raw pixel
> > formats that the decoder can handle with selected coded pixel format.
> > However, the list can be further narrowed down by using
> > VIDIOC_ENUM_FRAMESIZES, to restrict each raw format only to the
> > resolutions it can handle.
> >
> > The VIDIOC_ENUM_FMT call in this sequence (after getting the stream
> > information) would have the knowledge about the resolution, so the
> > list returned here would only include the formats that can be actually
> > handled. It should match the result of the initial query using both
> > VIDIOC_ENUM_FMT and VIDIOC_ENUM_FRAMESIZES.
>
> Right, so this will be a subset of the initial query taking the resolution
> into account.
>

Do you think it could be worth adding a note about this?

[snip]
> >>> +Decoding
> >>> +========
> >>> +
> >>> +This state is reached after the `Capture setup` sequence finishes succesfully.
> >>> +In this state, the client queues and dequeues buffers to both queues via
> >>> +:c:func:`VIDIOC_QBUF` and :c:func:`VIDIOC_DQBUF`, following the standard
> >>> +semantics.
> >>> +
> >>> +The contents of the source ``OUTPUT`` buffers depend on the active coded pixel
> >>> +format and may be affected by codec-specific extended controls, as stated in
> >>> +the documentation of each format.
> >>> +
> >>> +Both queues operate independently, following the standard behavior of V4L2
> >>> +buffer queues and memory-to-memory devices. In addition, the order of decoded
> >>> +frames dequeued from the ``CAPTURE`` queue may differ from the order of queuing
> >>> +coded frames to the ``OUTPUT`` queue, due to properties of the selected coded
> >>> +format, e.g. frame reordering.
> >>> +
> >>> +The client must not assume any direct relationship between ``CAPTURE``
> >>> +and ``OUTPUT`` buffers and any specific timing of buffers becoming
> >>> +available to dequeue. Specifically,
> >>> +
> >>> +* a buffer queued to ``OUTPUT`` may result in no buffers being produced
> >>> +  on ``CAPTURE`` (e.g. if it does not contain encoded data, or if only
> >>> +  metadata syntax structures are present in it),
> >>> +
> >>> +* a buffer queued to ``OUTPUT`` may result in more than 1 buffer produced
> >>> +  on ``CAPTURE`` (if the encoded data contained more than one frame, or if
> >>> +  returning a decoded frame allowed the decoder to return a frame that
> >>> +  preceded it in decode, but succeeded it in the display order),
> >>> +
> >>> +* a buffer queued to ``OUTPUT`` may result in a buffer being produced on
> >>> +  ``CAPTURE`` later into decode process, and/or after processing further
> >>> +  ``OUTPUT`` buffers, or be returned out of order, e.g. if display
> >>> +  reordering is used,
> >>> +
> >>> +* buffers may become available on the ``CAPTURE`` queue without additional
> >>> +  buffers queued to ``OUTPUT`` (e.g. during drain or ``EOS``), because of the
> >>> +  ``OUTPUT`` buffers queued in the past whose decoding results are only
> >>> +  available at later time, due to specifics of the decoding process.
> >>> +
> >>> +.. note::
> >>> +
> >>> +   To allow matching decoded ``CAPTURE`` buffers with ``OUTPUT`` buffers they
> >>> +   originated from, the client can set the ``timestamp`` field of the
> >>> +   :c:type:`v4l2_buffer` struct when queuing an ``OUTPUT`` buffer. The
> >>> +   ``CAPTURE`` buffer(s), which resulted from decoding that ``OUTPUT`` buffer
> >>> +   will have their ``timestamp`` field set to the same value when dequeued.
> >>> +
> >>> +   In addition to the straighforward case of one ``OUTPUT`` buffer producing
>
> straighforward -> straightforward
>

Ack.

> >>> +   one ``CAPTURE`` buffer, the following cases are defined:
> >>> +
> >>> +   * one ``OUTPUT`` buffer generates multiple ``CAPTURE`` buffers: the same
> >>> +     ``OUTPUT`` timestamp will be copied to multiple ``CAPTURE`` buffers,
> >>> +
> >>> +   * multiple ``OUTPUT`` buffers generate one ``CAPTURE`` buffer: timestamp of
> >>> +     the ``OUTPUT`` buffer queued last will be copied,
> >>> +
> >>> +   * the decoding order differs from the display order (i.e. the
> >>> +     ``CAPTURE`` buffers are out-of-order compared to the ``OUTPUT`` buffers):
> >>> +     ``CAPTURE`` timestamps will not retain the order of ``OUTPUT`` timestamps
> >>> +     and thus monotonicity of the timestamps cannot be guaranteed.
>
> I think this last point should be rewritten. The timestamp is just a value that
> is copied, there are no monotonicity requirements for m2m devices in general.
>

Actually I just realized the last point might not even be achievable
for some of the decoders (s5p-mfc, mtk-vcodec), as they don't report
which frame originates from which bitstream buffer and the driver just
picks the most recently consumed OUTPUT buffer to copy the timestamp
from. (s5p-mfc actually "forgets" to set the timestamp in some cases
too...)

I need to think a bit more about this.

[snip]
> >>> +Seek
> >>> +====
> >>> +
> >>> +Seek is controlled by the ``OUTPUT`` queue, as it is the source of coded data.
> >>> +The seek does not require any specific operation on the ``CAPTURE`` queue, but
> >>> +it may be affected as per normal decoder operation.
> >>> +
> >>> +1. Stop the ``OUTPUT`` queue to begin the seek sequence via
> >>> +   :c:func:`VIDIOC_STREAMOFF`.
> >>> +
> >>> +   * **Required fields:**
> >>> +
> >>> +     ``type``
> >>> +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> >>> +
> >>> +   * The decoder will drop all the pending ``OUTPUT`` buffers and they must be
> >>> +     treated as returned to the client (following standard semantics).
> >>> +
> >>> +2. Restart the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`
> >>> +
> >>> +   * **Required fields:**
> >>> +
> >>> +     ``type``
> >>> +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
> >>> +
> >>> +   * The decoder will start accepting new source bitstream buffers after the
> >>> +     call returns.
> >>> +
> >>> +3. Start queuing buffers containing coded data after the seek to the ``OUTPUT``
> >>> +   queue until a suitable resume point is found.
> >>> +
> >>> +   .. note::
> >>> +
> >>> +      There is no requirement to begin queuing coded data starting exactly
> >>> +      from a resume point (e.g. SPS or a keyframe). Any queued ``OUTPUT``
> >>> +      buffers will be processed and returned to the client until a suitable
> >>> +      resume point is found.  While looking for a resume point, the decoder
> >>> +      should not produce any decoded frames into ``CAPTURE`` buffers.
> >>> +
> >>> +      Some hardware is known to mishandle seeks to a non-resume point. Such an
> >>> +      operation may result in an unspecified number of corrupted decoded frames
> >>> +      being made available on the ``CAPTURE`` queue. Drivers must ensure that
> >>> +      no fatal decoding errors or crashes occur, and implement any necessary
> >>> +      handling and workarounds for hardware issues related to seek operations.
> >>
> >> Is there a requirement that those corrupted frames have V4L2_BUF_FLAG_ERROR set?
> >> I.e., can userspace detect those currupted frames?
> >>
> >
> > I think the question is whether the kernel driver can actually detect
> > those corrupted frames. We can't guarantee reporting errors to the
> > userspace, if the hardware doesn't actually report them.
> >
> > Could we perhaps keep this an open question and possibly address with
> > some extension that could be an opt in for the decoders that can
> > report errors?
>
> Hmm, how about: If the hardware can detect such corrupted decoded frames, then
> it shall set V4L2_BUF_FLAG_ERROR.
>

Sounds good to me.

Actually, let me add a paragraph about error handling in the decoding
section, since it's a general problem, not limited to seeking. I can
then refer to it from the Seek sequence.

> >
> >>> +
> >>> +   .. warning::
> >>> +
> >>> +      In case of the H.264 codec, the client must take care not to seek over a
> >>> +      change of SPS/PPS. Even though the target frame could be a keyframe, the
> >>> +      stale SPS/PPS inside decoder state would lead to undefined results when
> >>> +      decoding. Although the decoder must handle such case without a crash or a
> >>> +      fatal decode error, the client must not expect a sensible decode output.
> >>> +
> >>> +4. After a resume point is found, the decoder will start returning ``CAPTURE``
> >>> +   buffers containing decoded frames.
> >>> +
> >>> +.. important::
> >>> +
> >>> +   A seek may result in the `Dynamic resolution change` sequence being
> >>> +   iniitated, due to the seek target having decoding parameters different from
> >>> +   the part of the stream decoded before the seek. The sequence must be handled
> >>> +   as per normal decoder operation.
> >>> +
> >>> +.. warning::
> >>> +
> >>> +   It is not specified when the ``CAPTURE`` queue starts producing buffers
> >>> +   containing decoded data from the ``OUTPUT`` buffers queued after the seek,
> >>> +   as it operates independently from the ``OUTPUT`` queue.
> >>> +
> >>> +   The decoder may return a number of remaining ``CAPTURE`` buffers containing
> >>> +   decoded frames originating from the ``OUTPUT`` buffers queued before the
> >>> +   seek sequence is performed.
> >>> +
> >>> +   The ``VIDIOC_STREAMOFF`` operation discards any remaining queued
> >>> +   ``OUTPUT`` buffers, which means that not all of the ``OUTPUT`` buffers
> >>> +   queued before the seek sequence may have matching ``CAPTURE`` buffers
> >>> +   produced.  For example, given the sequence of operations on the
> >>> +   ``OUTPUT`` queue:
> >>> +
> >>> +     QBUF(A), QBUF(B), STREAMOFF(), STREAMON(), QBUF(G), QBUF(H),
> >>> +
> >>> +   any of the following results on the ``CAPTURE`` queue is allowed:
> >>> +
> >>> +     {A’, B’, G’, H’}, {A’, G’, H’}, {G’, H’}.
> >>
> >> Isn't it the case that if you would want to avoid that, then you should call
> >> DECODER_STOP, wait for the last buffer on the CAPTURE queue, then seek and
> >> call DECODER_START. If you do that, then you should always get {A’, B’, G’, H’}.
> >> (basically following the Drain sequence).
> >
> > Yes, it is, but I think it depends on the application needs. Here we
> > just give a primitive to change the place in the stream that's being
> > decoded (or change the stream on the fly).
> >
> > Actually, with the timestamp copy, I guess we wouldn't even need to do
> > the DECODER_STOP, as we could just discard the CAPTURE buffers until
> > we get one that matches the timestamp of the first OUTPUT buffer after
> > the seek.
> >
> >>
> >> Admittedly, you typically want to do an instantaneous seek, so this is probably
> >> not what you want to do normally.
> >>
> >> It might help to have this documented in a separate note.
> >>
> >
> > The instantaneous seek is documented below. I'm not sure if there is
> > any practical need to document the other case, but I could add a
> > sentence like below to the warning above. What do you think?
> >
> >    The ``VIDIOC_STREAMOFF`` operation discards any remaining queued
> >    ``OUTPUT`` buffers, which means that not all of the ``OUTPUT`` buffers
> >    queued before the seek sequence may have matching ``CAPTURE`` buffers
> >    produced.  For example, given the sequence of operations on the
> >    ``OUTPUT`` queue:
> >
> >      QBUF(A), QBUF(B), STREAMOFF(), STREAMON(), QBUF(G), QBUF(H),
> >
> >    any of the following results on the ``CAPTURE`` queue is allowed:
>
> is allowed -> are allowed
>

Only one can happen at given time, so I think singular is correct
here? (i.e. Any [...] is allowed)

[snip]
> >>> diff --git a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
> >>> index 3ead350e099f..0fc0b78a943e 100644
> >>> --- a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
> >>> +++ b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
> >>> @@ -53,6 +53,13 @@ devices that is either the struct
> >>>  member. When the requested buffer type is not supported drivers return
> >>>  an ``EINVAL`` error code.
> >>>
> >>> +A stateful mem2mem decoder will not allow operations on the
> >>> +``V4L2_BUF_TYPE_VIDEO_CAPTURE`` or ``V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE``
> >>> +buffer type until the corresponding ``V4L2_BUF_TYPE_VIDEO_OUTPUT`` or
> >>> +``V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE`` buffer type is configured. If such an
> >>> +operation is attempted, drivers return an ``EACCES`` error code. Refer to
> >>> +:ref:`decoder` for more details.
> >>
> >> This isn't right. EACCES is returned as long as the output format resolution is
> >> unknown. If it is set explicitly, then this will work without an error.
>
> Ah, sorry, I phrased that poorly. Let me try again:
>
> This isn't right. EACCES is returned for CAPTURE operations as long as the
> output format resolution is unknown or the CAPTURE format is explicitly set.
> If the CAPTURE format is set explicitly, then this will work without an error.

We don't allow directly setting CAPTURE format explicitly either,
because the driver wouldn't have anything to validate the format
against. We allow the client to do it indirectly, by setting the width
and height of the OUTPUT format, which unblocks the CAPTURE format
operations, because the driver can then validate against the OUTPUT
(coded) format.

Best regards,
Tomasz

Hans Verkuil Jan. 23, 2019, 8:10 a.m. UTC | #13

On 01/23/2019 06:27 AM, Tomasz Figa wrote:
> On Tue, Jan 22, 2019 at 11:47 PM Hans Verkuil <[e-mail]> wrote:
>>
>> On 01/22/19 11:02, Tomasz Figa wrote:
>>> On Mon, Nov 12, 2018 at 8:37 PM Hans Verkuil <[e-mail]> wrote:
>>>>
>>>> Hi Tomasz,
>>>>
>>>> A general note for the stateful and stateless patches: they describe specific
>>>> use-cases of the more generic Codec Interface, and as such should be one
>>>> level deeper in the section hierarchy.
>>>
>>> I wonder what exactly this Codec Interface is. Is it a historical name
>>> for mem2mem? If so, perhaps it would make sense to rename it?
>>
>> Yeah, it should be renamed to "Video Memory-to-Memory Interface", and the
>> codecs are just specific instances of such an interface.
>>
> 
> Ack.
> 
>>>
>>>>
>>>> I.e. instead of being section 4.6/7/8:
>>>>
>>>> https://hverkuil.home.xs4all.nl/request-api/uapi/v4l/devices.html
>>>>
>>>> they should be 4.5.1/2/3.
>>>>
>>>
>>> FYI, the first RFC started like that, but it only made the spec
>>> difficult to navigate and the section numbers too long.
>>>
>>> Still, no strong opinion. I'm okay moving it there, if you think it's better.
>>
>> It should be moved and the interface name should be renamed. It makes a lot
>> more sense with those changes.
>>
>> I've posted a patch for this.
>>
> 
> Thanks. I've rebased on top of it.
> 
> [snip]
>>>>> +3.  Query the minimum number of buffers required for the ``CAPTURE`` queue via
>>>>> +    :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to use more
>>>>> +    buffers than the minimum required by hardware/format.
>>>>
>>>> Is this step optional or required? Can it change when a resolution change occurs?
>>>
>>> Probably not with a simple resolution change, but a case when a stream
>>> is changed on the fly would trigger what we call "resolution change"
>>> here, but what would effectively be a "source change" and that could
>>> include a change in the number of required CAPTURE buffers.
>>>
>>>> How does this relate to the checks for the minimum number of buffers that REQBUFS
>>>> does?
>>>
>>> The control returns the minimum that REQBUFS would allow, so the
>>> application can add few more buffers on top of that and improve the
>>> pipelining.
>>>
>>>>
>>>> The 'This is useful if' sentence suggests that it is optional, but I think that
>>>> sentence just confuses the issue.
>>>>
>>>
>>> It used to be optional and I didn't rephrase it after turning it into
>>> mandatory. How about:
>>>
>>>     This enables the client to request more buffers
>>>     than the minimum required by hardware/format and achieve better pipelining.
>>
>> Hmm, OK. It'll do, I guess. I never liked these MIN_BUFFERS controls, I wish they
>> would return something like the recommended number of buffers that will give you
>> decent performance.
>>
> 
> The problem here is that the kernel doesn't know what is decent for
> the application, since it doesn't know how the results of the decoding
> are used. Over-allocating would result to a waste of memory, which
> could then make it less than decent for memory-constrained
> applications.
> 
>>>
>>>>> +
>>>>> +    * **Required fields:**
>>>>> +
>>>>> +      ``id``
>>>>> +          set to ``V4L2_CID_MIN_BUFFERS_FOR_CAPTURE``
>>>>> +
>>>>> +    * **Return fields:**
>>>>> +
>>>>> +      ``value``
>>>>> +          minimum number of buffers required to decode the stream parsed in
>>>>> +          this initialization sequence.
>>>>> +
>>>>> +    .. note::
>>>>> +
>>>>> +       The minimum number of buffers must be at least the number required to
>>>>> +       successfully decode the current stream. This may for example be the
>>>>> +       required DPB size for an H.264 stream given the parsed stream
>>>>> +       configuration (resolution, level).
>>>>> +
>>>>> +    .. warning::
>>>>> +
>>>>> +       The value is guaranteed to be meaningful only after the decoder
>>>>> +       successfully parses the stream metadata. The client must not rely on the
>>>>> +       query before that happens.
>>>>> +
>>>>> +4.  **Optional.** Enumerate ``CAPTURE`` formats via :c:func:`VIDIOC_ENUM_FMT` on
>>>>> +    the ``CAPTURE`` queue. Once the stream information is parsed and known, the
>>>>> +    client may use this ioctl to discover which raw formats are supported for
>>>>> +    given stream and select one of them via :c:func:`VIDIOC_S_FMT`.
>>>>
>>>> Can the list returned here differ from the list returned in the 'Querying capabilities'
>>>> step? If so, then I assume it will always be a subset of what was returned in
>>>> the 'Querying' step?
>>>
>>> Depends on whether you're considering just VIDIOC_ENUM_FMT or also
>>> VIDIOC_G_FMT and VIDIOC_ENUM_FRAMESIZES.
>>>
>>> The initial VIDIOC_ENUM_FMT has no way to account for any resolution
>>> constraints of the formats, so the list would include all raw pixel
>>> formats that the decoder can handle with selected coded pixel format.
>>> However, the list can be further narrowed down by using
>>> VIDIOC_ENUM_FRAMESIZES, to restrict each raw format only to the
>>> resolutions it can handle.
>>>
>>> The VIDIOC_ENUM_FMT call in this sequence (after getting the stream
>>> information) would have the knowledge about the resolution, so the
>>> list returned here would only include the formats that can be actually
>>> handled. It should match the result of the initial query using both
>>> VIDIOC_ENUM_FMT and VIDIOC_ENUM_FRAMESIZES.
>>
>> Right, so this will be a subset of the initial query taking the resolution
>> into account.
>>
> 
> Do you think it could be worth adding a note about this?

Yes, it helps clarify things.

> 
> [snip]
>>>>> +Decoding
>>>>> +========
>>>>> +
>>>>> +This state is reached after the `Capture setup` sequence finishes succesfully.
>>>>> +In this state, the client queues and dequeues buffers to both queues via
>>>>> +:c:func:`VIDIOC_QBUF` and :c:func:`VIDIOC_DQBUF`, following the standard
>>>>> +semantics.
>>>>> +
>>>>> +The contents of the source ``OUTPUT`` buffers depend on the active coded pixel
>>>>> +format and may be affected by codec-specific extended controls, as stated in
>>>>> +the documentation of each format.
>>>>> +
>>>>> +Both queues operate independently, following the standard behavior of V4L2
>>>>> +buffer queues and memory-to-memory devices. In addition, the order of decoded
>>>>> +frames dequeued from the ``CAPTURE`` queue may differ from the order of queuing
>>>>> +coded frames to the ``OUTPUT`` queue, due to properties of the selected coded
>>>>> +format, e.g. frame reordering.
>>>>> +
>>>>> +The client must not assume any direct relationship between ``CAPTURE``
>>>>> +and ``OUTPUT`` buffers and any specific timing of buffers becoming
>>>>> +available to dequeue. Specifically,
>>>>> +
>>>>> +* a buffer queued to ``OUTPUT`` may result in no buffers being produced
>>>>> +  on ``CAPTURE`` (e.g. if it does not contain encoded data, or if only
>>>>> +  metadata syntax structures are present in it),
>>>>> +
>>>>> +* a buffer queued to ``OUTPUT`` may result in more than 1 buffer produced
>>>>> +  on ``CAPTURE`` (if the encoded data contained more than one frame, or if
>>>>> +  returning a decoded frame allowed the decoder to return a frame that
>>>>> +  preceded it in decode, but succeeded it in the display order),
>>>>> +
>>>>> +* a buffer queued to ``OUTPUT`` may result in a buffer being produced on
>>>>> +  ``CAPTURE`` later into decode process, and/or after processing further
>>>>> +  ``OUTPUT`` buffers, or be returned out of order, e.g. if display
>>>>> +  reordering is used,
>>>>> +
>>>>> +* buffers may become available on the ``CAPTURE`` queue without additional
>>>>> +  buffers queued to ``OUTPUT`` (e.g. during drain or ``EOS``), because of the
>>>>> +  ``OUTPUT`` buffers queued in the past whose decoding results are only
>>>>> +  available at later time, due to specifics of the decoding process.
>>>>> +
>>>>> +.. note::
>>>>> +
>>>>> +   To allow matching decoded ``CAPTURE`` buffers with ``OUTPUT`` buffers they
>>>>> +   originated from, the client can set the ``timestamp`` field of the
>>>>> +   :c:type:`v4l2_buffer` struct when queuing an ``OUTPUT`` buffer. The
>>>>> +   ``CAPTURE`` buffer(s), which resulted from decoding that ``OUTPUT`` buffer
>>>>> +   will have their ``timestamp`` field set to the same value when dequeued.
>>>>> +
>>>>> +   In addition to the straighforward case of one ``OUTPUT`` buffer producing
>>
>> straighforward -> straightforward
>>
> 
> Ack.
> 
>>>>> +   one ``CAPTURE`` buffer, the following cases are defined:
>>>>> +
>>>>> +   * one ``OUTPUT`` buffer generates multiple ``CAPTURE`` buffers: the same
>>>>> +     ``OUTPUT`` timestamp will be copied to multiple ``CAPTURE`` buffers,
>>>>> +
>>>>> +   * multiple ``OUTPUT`` buffers generate one ``CAPTURE`` buffer: timestamp of
>>>>> +     the ``OUTPUT`` buffer queued last will be copied,
>>>>> +
>>>>> +   * the decoding order differs from the display order (i.e. the
>>>>> +     ``CAPTURE`` buffers are out-of-order compared to the ``OUTPUT`` buffers):
>>>>> +     ``CAPTURE`` timestamps will not retain the order of ``OUTPUT`` timestamps
>>>>> +     and thus monotonicity of the timestamps cannot be guaranteed.
>>
>> I think this last point should be rewritten. The timestamp is just a value that
>> is copied, there are no monotonicity requirements for m2m devices in general.
>>
> 
> Actually I just realized the last point might not even be achievable
> for some of the decoders (s5p-mfc, mtk-vcodec), as they don't report
> which frame originates from which bitstream buffer and the driver just
> picks the most recently consumed OUTPUT buffer to copy the timestamp
> from. (s5p-mfc actually "forgets" to set the timestamp in some cases
> too...)
> 
> I need to think a bit more about this.
> 
> [snip]
>>>>> +Seek
>>>>> +====
>>>>> +
>>>>> +Seek is controlled by the ``OUTPUT`` queue, as it is the source of coded data.
>>>>> +The seek does not require any specific operation on the ``CAPTURE`` queue, but
>>>>> +it may be affected as per normal decoder operation.
>>>>> +
>>>>> +1. Stop the ``OUTPUT`` queue to begin the seek sequence via
>>>>> +   :c:func:`VIDIOC_STREAMOFF`.
>>>>> +
>>>>> +   * **Required fields:**
>>>>> +
>>>>> +     ``type``
>>>>> +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
>>>>> +
>>>>> +   * The decoder will drop all the pending ``OUTPUT`` buffers and they must be
>>>>> +     treated as returned to the client (following standard semantics).
>>>>> +
>>>>> +2. Restart the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`
>>>>> +
>>>>> +   * **Required fields:**
>>>>> +
>>>>> +     ``type``
>>>>> +         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
>>>>> +
>>>>> +   * The decoder will start accepting new source bitstream buffers after the
>>>>> +     call returns.
>>>>> +
>>>>> +3. Start queuing buffers containing coded data after the seek to the ``OUTPUT``
>>>>> +   queue until a suitable resume point is found.
>>>>> +
>>>>> +   .. note::
>>>>> +
>>>>> +      There is no requirement to begin queuing coded data starting exactly
>>>>> +      from a resume point (e.g. SPS or a keyframe). Any queued ``OUTPUT``
>>>>> +      buffers will be processed and returned to the client until a suitable
>>>>> +      resume point is found.  While looking for a resume point, the decoder
>>>>> +      should not produce any decoded frames into ``CAPTURE`` buffers.
>>>>> +
>>>>> +      Some hardware is known to mishandle seeks to a non-resume point. Such an
>>>>> +      operation may result in an unspecified number of corrupted decoded frames
>>>>> +      being made available on the ``CAPTURE`` queue. Drivers must ensure that
>>>>> +      no fatal decoding errors or crashes occur, and implement any necessary
>>>>> +      handling and workarounds for hardware issues related to seek operations.
>>>>
>>>> Is there a requirement that those corrupted frames have V4L2_BUF_FLAG_ERROR set?
>>>> I.e., can userspace detect those currupted frames?
>>>>
>>>
>>> I think the question is whether the kernel driver can actually detect
>>> those corrupted frames. We can't guarantee reporting errors to the
>>> userspace, if the hardware doesn't actually report them.
>>>
>>> Could we perhaps keep this an open question and possibly address with
>>> some extension that could be an opt in for the decoders that can
>>> report errors?
>>
>> Hmm, how about: If the hardware can detect such corrupted decoded frames, then
>> it shall set V4L2_BUF_FLAG_ERROR.
>>
> 
> Sounds good to me.
> 
> Actually, let me add a paragraph about error handling in the decoding
> section, since it's a general problem, not limited to seeking. I can
> then refer to it from the Seek sequence.
> 
>>>
>>>>> +
>>>>> +   .. warning::
>>>>> +
>>>>> +      In case of the H.264 codec, the client must take care not to seek over a
>>>>> +      change of SPS/PPS. Even though the target frame could be a keyframe, the
>>>>> +      stale SPS/PPS inside decoder state would lead to undefined results when
>>>>> +      decoding. Although the decoder must handle such case without a crash or a
>>>>> +      fatal decode error, the client must not expect a sensible decode output.
>>>>> +
>>>>> +4. After a resume point is found, the decoder will start returning ``CAPTURE``
>>>>> +   buffers containing decoded frames.
>>>>> +
>>>>> +.. important::
>>>>> +
>>>>> +   A seek may result in the `Dynamic resolution change` sequence being
>>>>> +   iniitated, due to the seek target having decoding parameters different from
>>>>> +   the part of the stream decoded before the seek. The sequence must be handled
>>>>> +   as per normal decoder operation.
>>>>> +
>>>>> +.. warning::
>>>>> +
>>>>> +   It is not specified when the ``CAPTURE`` queue starts producing buffers
>>>>> +   containing decoded data from the ``OUTPUT`` buffers queued after the seek,
>>>>> +   as it operates independently from the ``OUTPUT`` queue.
>>>>> +
>>>>> +   The decoder may return a number of remaining ``CAPTURE`` buffers containing
>>>>> +   decoded frames originating from the ``OUTPUT`` buffers queued before the
>>>>> +   seek sequence is performed.
>>>>> +
>>>>> +   The ``VIDIOC_STREAMOFF`` operation discards any remaining queued
>>>>> +   ``OUTPUT`` buffers, which means that not all of the ``OUTPUT`` buffers
>>>>> +   queued before the seek sequence may have matching ``CAPTURE`` buffers
>>>>> +   produced.  For example, given the sequence of operations on the
>>>>> +   ``OUTPUT`` queue:
>>>>> +
>>>>> +     QBUF(A), QBUF(B), STREAMOFF(), STREAMON(), QBUF(G), QBUF(H),
>>>>> +
>>>>> +   any of the following results on the ``CAPTURE`` queue is allowed:
>>>>> +
>>>>> +     {A’, B’, G’, H’}, {A’, G’, H’}, {G’, H’}.
>>>>
>>>> Isn't it the case that if you would want to avoid that, then you should call
>>>> DECODER_STOP, wait for the last buffer on the CAPTURE queue, then seek and
>>>> call DECODER_START. If you do that, then you should always get {A’, B’, G’, H’}.
>>>> (basically following the Drain sequence).
>>>
>>> Yes, it is, but I think it depends on the application needs. Here we
>>> just give a primitive to change the place in the stream that's being
>>> decoded (or change the stream on the fly).
>>>
>>> Actually, with the timestamp copy, I guess we wouldn't even need to do
>>> the DECODER_STOP, as we could just discard the CAPTURE buffers until
>>> we get one that matches the timestamp of the first OUTPUT buffer after
>>> the seek.
>>>
>>>>
>>>> Admittedly, you typically want to do an instantaneous seek, so this is probably
>>>> not what you want to do normally.
>>>>
>>>> It might help to have this documented in a separate note.
>>>>
>>>
>>> The instantaneous seek is documented below. I'm not sure if there is
>>> any practical need to document the other case, but I could add a
>>> sentence like below to the warning above. What do you think?
>>>
>>>    The ``VIDIOC_STREAMOFF`` operation discards any remaining queued
>>>    ``OUTPUT`` buffers, which means that not all of the ``OUTPUT`` buffers
>>>    queued before the seek sequence may have matching ``CAPTURE`` buffers
>>>    produced.  For example, given the sequence of operations on the
>>>    ``OUTPUT`` queue:
>>>
>>>      QBUF(A), QBUF(B), STREAMOFF(), STREAMON(), QBUF(G), QBUF(H),
>>>
>>>    any of the following results on the ``CAPTURE`` queue is allowed:
>>
>> is allowed -> are allowed
>>
> 
> Only one can happen at given time, so I think singular is correct
> here? (i.e. Any [...] is allowed)

I think you are right.

> 
> [snip]
>>>>> diff --git a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
>>>>> index 3ead350e099f..0fc0b78a943e 100644
>>>>> --- a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
>>>>> +++ b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
>>>>> @@ -53,6 +53,13 @@ devices that is either the struct
>>>>>  member. When the requested buffer type is not supported drivers return
>>>>>  an ``EINVAL`` error code.
>>>>>
>>>>> +A stateful mem2mem decoder will not allow operations on the
>>>>> +``V4L2_BUF_TYPE_VIDEO_CAPTURE`` or ``V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE``
>>>>> +buffer type until the corresponding ``V4L2_BUF_TYPE_VIDEO_OUTPUT`` or
>>>>> +``V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE`` buffer type is configured. If such an
>>>>> +operation is attempted, drivers return an ``EACCES`` error code. Refer to
>>>>> +:ref:`decoder` for more details.
>>>>
>>>> This isn't right. EACCES is returned as long as the output format resolution is
>>>> unknown. If it is set explicitly, then this will work without an error.
>>
>> Ah, sorry, I phrased that poorly. Let me try again:
>>
>> This isn't right. EACCES is returned for CAPTURE operations as long as the
>> output format resolution is unknown or the CAPTURE format is explicitly set.
>> If the CAPTURE format is set explicitly, then this will work without an error.
> 
> We don't allow directly setting CAPTURE format explicitly either,
> because the driver wouldn't have anything to validate the format
> against. We allow the client to do it indirectly, by setting the width
> and height of the OUTPUT format, which unblocks the CAPTURE format
> operations, because the driver can then validate against the OUTPUT
> (coded) format.

You are completely right, just forget what I said.

Regards,

	Hans

> 
> Best regards,
> Tomasz
>

Tomasz Figa Jan. 24, 2019, 9:06 a.m. UTC | #14

On Wed, Jan 23, 2019 at 2:27 PM Tomasz Figa <[e-mail]> wrote:
>
> On Tue, Jan 22, 2019 at 11:47 PM Hans Verkuil <[e-mail]> wrote:
> >
> > On 01/22/19 11:02, Tomasz Figa wrote:
[snip]
> > >>> +   one ``CAPTURE`` buffer, the following cases are defined:
> > >>> +
> > >>> +   * one ``OUTPUT`` buffer generates multiple ``CAPTURE`` buffers: the same
> > >>> +     ``OUTPUT`` timestamp will be copied to multiple ``CAPTURE`` buffers,
> > >>> +
> > >>> +   * multiple ``OUTPUT`` buffers generate one ``CAPTURE`` buffer: timestamp of
> > >>> +     the ``OUTPUT`` buffer queued last will be copied,
> > >>> +
> > >>> +   * the decoding order differs from the display order (i.e. the
> > >>> +     ``CAPTURE`` buffers are out-of-order compared to the ``OUTPUT`` buffers):
> > >>> +     ``CAPTURE`` timestamps will not retain the order of ``OUTPUT`` timestamps
> > >>> +     and thus monotonicity of the timestamps cannot be guaranteed.
> >
> > I think this last point should be rewritten. The timestamp is just a value that
> > is copied, there are no monotonicity requirements for m2m devices in general.
> >
>
> Actually I just realized the last point might not even be achievable
> for some of the decoders (s5p-mfc, mtk-vcodec), as they don't report
> which frame originates from which bitstream buffer and the driver just
> picks the most recently consumed OUTPUT buffer to copy the timestamp
> from. (s5p-mfc actually "forgets" to set the timestamp in some cases
> too...)
>
> I need to think a bit more about this.

Actually I misread the code. Both s5p-mfc and mtk-vcodec seem to
correctly match the buffers.

Best regards,
Tomasz

Nicolas Dufresne Jan. 24, 2019, 7:55 p.m. UTC | #15

Le jeudi 24 janvier 2019 à 18:06 +0900, Tomasz Figa a écrit :
> > Actually I just realized the last point might not even be achievable
> > for some of the decoders (s5p-mfc, mtk-vcodec), as they don't report
> > which frame originates from which bitstream buffer and the driver just
> > picks the most recently consumed OUTPUT buffer to copy the timestamp
> > from. (s5p-mfc actually "forgets" to set the timestamp in some cases
> > too...)
> > 
> > I need to think a bit more about this.
> 
> Actually I misread the code. Both s5p-mfc and mtk-vcodec seem to
> correctly match the buffers.

Ok good, since otherwise it would have been a regression in MFC driver.
This timestamp passing thing could in theory be made optional though,
it lives under some COPY_TIMESTAMP kind of flag. What that means though
is that a driver without such a capability would need to signal dropped
frames using some other mean.

In userspace, the main use is to match the produced frame against a
userspace specific list of frames. At least this seems to be the case
in Gst and Chromium, since the userspace list contains a superset of
the metadata found in the v4l2_buffer.

Now, using the produced timestamp, userspace can deduce frame that the
driver should have produced but didn't (could be a deadline case codec,
or simply the frames where corrupted). It's quite normal for a codec to
just keep parsing until it finally find something it can decode.

That's at least one way to do it, but there is other possible
mechanism. The sequence number could be used, or even producing buffers
with the ERROR flag set. What matters is just to give userspace a way
to clear these frames, which would simply grow userspace memory usage
over time.

Nicolas

Tomasz Figa Jan. 25, 2019, 3:27 a.m. UTC | #16

On Fri, Jan 25, 2019 at 4:55 AM Nicolas Dufresne <[e-mail]> wrote:
>
> Le jeudi 24 janvier 2019 à 18:06 +0900, Tomasz Figa a écrit :
> > > Actually I just realized the last point might not even be achievable
> > > for some of the decoders (s5p-mfc, mtk-vcodec), as they don't report
> > > which frame originates from which bitstream buffer and the driver just
> > > picks the most recently consumed OUTPUT buffer to copy the timestamp
> > > from. (s5p-mfc actually "forgets" to set the timestamp in some cases
> > > too...)
> > >
> > > I need to think a bit more about this.
> >
> > Actually I misread the code. Both s5p-mfc and mtk-vcodec seem to
> > correctly match the buffers.
>
> Ok good, since otherwise it would have been a regression in MFC driver.
> This timestamp passing thing could in theory be made optional though,
> it lives under some COPY_TIMESTAMP kind of flag. What that means though
> is that a driver without such a capability would need to signal dropped
> frames using some other mean.
>
> In userspace, the main use is to match the produced frame against a
> userspace specific list of frames. At least this seems to be the case
> in Gst and Chromium, since the userspace list contains a superset of
> the metadata found in the v4l2_buffer.
>
> Now, using the produced timestamp, userspace can deduce frame that the
> driver should have produced but didn't (could be a deadline case codec,
> or simply the frames where corrupted). It's quite normal for a codec to
> just keep parsing until it finally find something it can decode.
>
> That's at least one way to do it, but there is other possible
> mechanism. The sequence number could be used, or even producing buffers
> with the ERROR flag set. What matters is just to give userspace a way
> to clear these frames, which would simply grow userspace memory usage
> over time.

Is it just me or we were missing some consistent error handling then?

I feel like the drivers should definitely return the bitstream buffers
with the ERROR flag, if there is a decode failure of data in the
buffer. Still, that could become more complicated if there is more
than 1 frame in that piece of bitstream, but only 1 frame is corrupted
(or whatever).

Another case is when the bitstream, even if corrupted, is still enough
to produce some output. My intuition tells me that such CAPTURE buffer
should be then returned with the ERROR flag. That wouldn't still be
enough for any more sophisticated userspace error concealment, but
could still let the userspace know to perhaps drop the frame.

Best regards,
Tomasz

Nicolas Dufresne Jan. 30, 2019, 4:02 a.m. UTC | #17

Le vendredi 25 janvier 2019 à 12:27 +0900, Tomasz Figa a écrit :
> On Fri, Jan 25, 2019 at 4:55 AM Nicolas Dufresne <[e-mail]> wrote:
> > Le jeudi 24 janvier 2019 à 18:06 +0900, Tomasz Figa a écrit :
> > > > Actually I just realized the last point might not even be achievable
> > > > for some of the decoders (s5p-mfc, mtk-vcodec), as they don't report
> > > > which frame originates from which bitstream buffer and the driver just
> > > > picks the most recently consumed OUTPUT buffer to copy the timestamp
> > > > from. (s5p-mfc actually "forgets" to set the timestamp in some cases
> > > > too...)
> > > > 
> > > > I need to think a bit more about this.
> > > 
> > > Actually I misread the code. Both s5p-mfc and mtk-vcodec seem to
> > > correctly match the buffers.
> > 
> > Ok good, since otherwise it would have been a regression in MFC driver.
> > This timestamp passing thing could in theory be made optional though,
> > it lives under some COPY_TIMESTAMP kind of flag. What that means though
> > is that a driver without such a capability would need to signal dropped
> > frames using some other mean.
> > 
> > In userspace, the main use is to match the produced frame against a
> > userspace specific list of frames. At least this seems to be the case
> > in Gst and Chromium, since the userspace list contains a superset of
> > the metadata found in the v4l2_buffer.
> > 
> > Now, using the produced timestamp, userspace can deduce frame that the
> > driver should have produced but didn't (could be a deadline case codec,
> > or simply the frames where corrupted). It's quite normal for a codec to
> > just keep parsing until it finally find something it can decode.
> > 
> > That's at least one way to do it, but there is other possible
> > mechanism. The sequence number could be used, or even producing buffers
> > with the ERROR flag set. What matters is just to give userspace a way
> > to clear these frames, which would simply grow userspace memory usage
> > over time.
> 
> Is it just me or we were missing some consistent error handling then?
> 
> I feel like the drivers should definitely return the bitstream buffers
> with the ERROR flag, if there is a decode failure of data in the
> buffer. Still, that could become more complicated if there is more
> than 1 frame in that piece of bitstream, but only 1 frame is corrupted
> (or whatever).

I agree, but it might be more difficult then it looks (even FFMPEG does
not do that). I believe the code that is processing the bitstream in
stateful codecs is mostly unrelated from the code actually doing the
decoding. So what might happen is that the decoding part will never
actually allocate a buffer for the skipped / corrupted part of the
bitstream. Also, the notion of a skipped frame is not always evident in
when parsing H264 or HEVC NALs. There is still a full page of text just
to explain how to detect that start of a new frame.

Yet, it would be interesting to study the firmwares we have and see
what they provide that would help making decode errors more explicit.

> 
> Another case is when the bitstream, even if corrupted, is still enough
> to produce some output. My intuition tells me that such CAPTURE buffer
> should be then returned with the ERROR flag. That wouldn't still be
> enough for any more sophisticated userspace error concealment, but
> could still let the userspace know to perhaps drop the frame.

You mean if a frame was concealed (typically the frame was decoded from
a closed by reference instead of the expected reference). That is
something signalled by FFPEG. We should document this possibility. I
actually have something implemented in GStreamer. Basically if we have
the ERROR flag with a payload size smaller then expected, I drop the
frame and produce a drop event message, while if I have a frame with
ERROR flag but of the right payload size, I assume it is corrupted, and
simply flag it as corrupted, leaving to the application the decision to
display it or not. This is a case that used to happen with some UVC
cameras (though some have been fixed, and the UVC camera should drop
smaller payload size buffers now).

> 
> Best regards,
> Tomasz

Tomasz Figa Feb. 6, 2019, 5:35 a.m. UTC | #18

On Wed, Jan 30, 2019 at 1:02 PM Nicolas Dufresne <[e-mail]> wrote:
>
> Le vendredi 25 janvier 2019 à 12:27 +0900, Tomasz Figa a écrit :
> > On Fri, Jan 25, 2019 at 4:55 AM Nicolas Dufresne <[e-mail]> wrote:
> > > Le jeudi 24 janvier 2019 à 18:06 +0900, Tomasz Figa a écrit :
> > > > > Actually I just realized the last point might not even be achievable
> > > > > for some of the decoders (s5p-mfc, mtk-vcodec), as they don't report
> > > > > which frame originates from which bitstream buffer and the driver just
> > > > > picks the most recently consumed OUTPUT buffer to copy the timestamp
> > > > > from. (s5p-mfc actually "forgets" to set the timestamp in some cases
> > > > > too...)
> > > > >
> > > > > I need to think a bit more about this.
> > > >
> > > > Actually I misread the code. Both s5p-mfc and mtk-vcodec seem to
> > > > correctly match the buffers.
> > >
> > > Ok good, since otherwise it would have been a regression in MFC driver.
> > > This timestamp passing thing could in theory be made optional though,
> > > it lives under some COPY_TIMESTAMP kind of flag. What that means though
> > > is that a driver without such a capability would need to signal dropped
> > > frames using some other mean.
> > >
> > > In userspace, the main use is to match the produced frame against a
> > > userspace specific list of frames. At least this seems to be the case
> > > in Gst and Chromium, since the userspace list contains a superset of
> > > the metadata found in the v4l2_buffer.
> > >
> > > Now, using the produced timestamp, userspace can deduce frame that the
> > > driver should have produced but didn't (could be a deadline case codec,
> > > or simply the frames where corrupted). It's quite normal for a codec to
> > > just keep parsing until it finally find something it can decode.
> > >
> > > That's at least one way to do it, but there is other possible
> > > mechanism. The sequence number could be used, or even producing buffers
> > > with the ERROR flag set. What matters is just to give userspace a way
> > > to clear these frames, which would simply grow userspace memory usage
> > > over time.
> >
> > Is it just me or we were missing some consistent error handling then?
> >
> > I feel like the drivers should definitely return the bitstream buffers
> > with the ERROR flag, if there is a decode failure of data in the
> > buffer. Still, that could become more complicated if there is more
> > than 1 frame in that piece of bitstream, but only 1 frame is corrupted
> > (or whatever).
>
> I agree, but it might be more difficult then it looks (even FFMPEG does
> not do that). I believe the code that is processing the bitstream in
> stateful codecs is mostly unrelated from the code actually doing the
> decoding. So what might happen is that the decoding part will never
> actually allocate a buffer for the skipped / corrupted part of the
> bitstream. Also, the notion of a skipped frame is not always evident in
> when parsing H264 or HEVC NALs. There is still a full page of text just
> to explain how to detect that start of a new frame.

Right. I don't think we can guarantee that we can always correlate the
errors with exact buffers and so I phrased the paragraph about errors
in v3 in a bit more conservative way:

See the snapshot hosted by Hans (thanks!):
https://hverkuil.home.xs4all.nl/codec-api/uapi/v4l/dev-decoder.html#decoding

>
> Yet, it would be interesting to study the firmwares we have and see
> what they provide that would help making decode errors more explicit.
>

Agreed.

> >
> > Another case is when the bitstream, even if corrupted, is still enough
> > to produce some output. My intuition tells me that such CAPTURE buffer
> > should be then returned with the ERROR flag. That wouldn't still be
> > enough for any more sophisticated userspace error concealment, but
> > could still let the userspace know to perhaps drop the frame.
>
> You mean if a frame was concealed (typically the frame was decoded from
> a closed by reference instead of the expected reference). That is
> something signalled by FFPEG. We should document this possibility. I
> actually have something implemented in GStreamer. Basically if we have
> the ERROR flag with a payload size smaller then expected, I drop the
> frame and produce a drop event message, while if I have a frame with
> ERROR flag but of the right payload size, I assume it is corrupted, and
> simply flag it as corrupted, leaving to the application the decision to
> display it or not. This is a case that used to happen with some UVC
> cameras (though some have been fixed, and the UVC camera should drop
> smaller payload size buffers now).

I think it's a behavior that makes the most sense indeed.

Technically one could also consider the case of 0 < bytesused <
sizeimage, which could mean that only a part of the frame is in the
buffer. An application could try to blend it with previous frame using
some concealing algorithms. I haven't seen an app that could do such
thing, though.

Best regards,
Tomasz

Tomasz Figa April 9, 2019, 9:47 a.m. UTC | #19

On Wed, Feb 6, 2019 at 2:35 PM Tomasz Figa <[e-mail]> wrote:
>
> On Wed, Jan 30, 2019 at 1:02 PM Nicolas Dufresne <[e-mail]> wrote:
> >
> > Le vendredi 25 janvier 2019 à 12:27 +0900, Tomasz Figa a écrit :
> > > On Fri, Jan 25, 2019 at 4:55 AM Nicolas Dufresne <[e-mail]> wrote:
> > > > Le jeudi 24 janvier 2019 à 18:06 +0900, Tomasz Figa a écrit :
> > > > > > Actually I just realized the last point might not even be achievable
> > > > > > for some of the decoders (s5p-mfc, mtk-vcodec), as they don't report
> > > > > > which frame originates from which bitstream buffer and the driver just
> > > > > > picks the most recently consumed OUTPUT buffer to copy the timestamp
> > > > > > from. (s5p-mfc actually "forgets" to set the timestamp in some cases
> > > > > > too...)
> > > > > >
> > > > > > I need to think a bit more about this.
> > > > >
> > > > > Actually I misread the code. Both s5p-mfc and mtk-vcodec seem to
> > > > > correctly match the buffers.
> > > >
> > > > Ok good, since otherwise it would have been a regression in MFC driver.
> > > > This timestamp passing thing could in theory be made optional though,
> > > > it lives under some COPY_TIMESTAMP kind of flag. What that means though
> > > > is that a driver without such a capability would need to signal dropped
> > > > frames using some other mean.
> > > >
> > > > In userspace, the main use is to match the produced frame against a
> > > > userspace specific list of frames. At least this seems to be the case
> > > > in Gst and Chromium, since the userspace list contains a superset of
> > > > the metadata found in the v4l2_buffer.
> > > >
> > > > Now, using the produced timestamp, userspace can deduce frame that the
> > > > driver should have produced but didn't (could be a deadline case codec,
> > > > or simply the frames where corrupted). It's quite normal for a codec to
> > > > just keep parsing until it finally find something it can decode.
> > > >
> > > > That's at least one way to do it, but there is other possible
> > > > mechanism. The sequence number could be used, or even producing buffers
> > > > with the ERROR flag set. What matters is just to give userspace a way
> > > > to clear these frames, which would simply grow userspace memory usage
> > > > over time.
> > >
> > > Is it just me or we were missing some consistent error handling then?
> > >
> > > I feel like the drivers should definitely return the bitstream buffers
> > > with the ERROR flag, if there is a decode failure of data in the
> > > buffer. Still, that could become more complicated if there is more
> > > than 1 frame in that piece of bitstream, but only 1 frame is corrupted
> > > (or whatever).
> >
> > I agree, but it might be more difficult then it looks (even FFMPEG does
> > not do that). I believe the code that is processing the bitstream in
> > stateful codecs is mostly unrelated from the code actually doing the
> > decoding. So what might happen is that the decoding part will never
> > actually allocate a buffer for the skipped / corrupted part of the
> > bitstream. Also, the notion of a skipped frame is not always evident in
> > when parsing H264 or HEVC NALs. There is still a full page of text just
> > to explain how to detect that start of a new frame.
>
> Right. I don't think we can guarantee that we can always correlate the
> errors with exact buffers and so I phrased the paragraph about errors
> in v3 in a bit more conservative way:
>
> See the snapshot hosted by Hans (thanks!):
> https://hverkuil.home.xs4all.nl/codec-api/uapi/v4l/dev-decoder.html#decoding
>
> >
> > Yet, it would be interesting to study the firmwares we have and see
> > what they provide that would help making decode errors more explicit.
> >
>
> Agreed.
>
> > >
> > > Another case is when the bitstream, even if corrupted, is still enough
> > > to produce some output. My intuition tells me that such CAPTURE buffer
> > > should be then returned with the ERROR flag. That wouldn't still be
> > > enough for any more sophisticated userspace error concealment, but
> > > could still let the userspace know to perhaps drop the frame.
> >
> > You mean if a frame was concealed (typically the frame was decoded from
> > a closed by reference instead of the expected reference). That is
> > something signalled by FFPEG. We should document this possibility. I
> > actually have something implemented in GStreamer. Basically if we have
> > the ERROR flag with a payload size smaller then expected, I drop the
> > frame and produce a drop event message, while if I have a frame with
> > ERROR flag but of the right payload size, I assume it is corrupted, and
> > simply flag it as corrupted, leaving to the application the decision to
> > display it or not. This is a case that used to happen with some UVC
> > cameras (though some have been fixed, and the UVC camera should drop
> > smaller payload size buffers now).
>
> I think it's a behavior that makes the most sense indeed.
>
> Technically one could also consider the case of 0 < bytesused <
> sizeimage, which could mean that only a part of the frame is in the
> buffer. An application could try to blend it with previous frame using
> some concealing algorithms. I haven't seen an app that could do such
> thing, though.

Actually some interesting thought on this. I don't think the existing
drivers would return any CAPTURE buffers on errors right now, but just
return the OUTPUT buffer that failed to be decoded. Should we change
this so that always one CAPTURE buffer with the ERROR flag is
returned, to signal the application that a frame was potentially
dropped?

Best regards,
Tomasz

Hans Verkuil April 10, 2019, 9:26 a.m. UTC | #20

On 4/9/19 11:47 AM, Tomasz Figa wrote:
> On Wed, Feb 6, 2019 at 2:35 PM Tomasz Figa <[e-mail]> wrote:
>>
>> On Wed, Jan 30, 2019 at 1:02 PM Nicolas Dufresne <[e-mail]> wrote:
>>>
>>> Le vendredi 25 janvier 2019 à 12:27 +0900, Tomasz Figa a écrit :
>>>> On Fri, Jan 25, 2019 at 4:55 AM Nicolas Dufresne <[e-mail]> wrote:
>>>>> Le jeudi 24 janvier 2019 à 18:06 +0900, Tomasz Figa a écrit :
>>>>>>> Actually I just realized the last point might not even be achievable
>>>>>>> for some of the decoders (s5p-mfc, mtk-vcodec), as they don't report
>>>>>>> which frame originates from which bitstream buffer and the driver just
>>>>>>> picks the most recently consumed OUTPUT buffer to copy the timestamp
>>>>>>> from. (s5p-mfc actually "forgets" to set the timestamp in some cases
>>>>>>> too...)
>>>>>>>
>>>>>>> I need to think a bit more about this.
>>>>>>
>>>>>> Actually I misread the code. Both s5p-mfc and mtk-vcodec seem to
>>>>>> correctly match the buffers.
>>>>>
>>>>> Ok good, since otherwise it would have been a regression in MFC driver.
>>>>> This timestamp passing thing could in theory be made optional though,
>>>>> it lives under some COPY_TIMESTAMP kind of flag. What that means though
>>>>> is that a driver without such a capability would need to signal dropped
>>>>> frames using some other mean.
>>>>>
>>>>> In userspace, the main use is to match the produced frame against a
>>>>> userspace specific list of frames. At least this seems to be the case
>>>>> in Gst and Chromium, since the userspace list contains a superset of
>>>>> the metadata found in the v4l2_buffer.
>>>>>
>>>>> Now, using the produced timestamp, userspace can deduce frame that the
>>>>> driver should have produced but didn't (could be a deadline case codec,
>>>>> or simply the frames where corrupted). It's quite normal for a codec to
>>>>> just keep parsing until it finally find something it can decode.
>>>>>
>>>>> That's at least one way to do it, but there is other possible
>>>>> mechanism. The sequence number could be used, or even producing buffers
>>>>> with the ERROR flag set. What matters is just to give userspace a way
>>>>> to clear these frames, which would simply grow userspace memory usage
>>>>> over time.
>>>>
>>>> Is it just me or we were missing some consistent error handling then?
>>>>
>>>> I feel like the drivers should definitely return the bitstream buffers
>>>> with the ERROR flag, if there is a decode failure of data in the
>>>> buffer. Still, that could become more complicated if there is more
>>>> than 1 frame in that piece of bitstream, but only 1 frame is corrupted
>>>> (or whatever).
>>>
>>> I agree, but it might be more difficult then it looks (even FFMPEG does
>>> not do that). I believe the code that is processing the bitstream in
>>> stateful codecs is mostly unrelated from the code actually doing the
>>> decoding. So what might happen is that the decoding part will never
>>> actually allocate a buffer for the skipped / corrupted part of the
>>> bitstream. Also, the notion of a skipped frame is not always evident in
>>> when parsing H264 or HEVC NALs. There is still a full page of text just
>>> to explain how to detect that start of a new frame.
>>
>> Right. I don't think we can guarantee that we can always correlate the
>> errors with exact buffers and so I phrased the paragraph about errors
>> in v3 in a bit more conservative way:
>>
>> See the snapshot hosted by Hans (thanks!):
>> https://hverkuil.home.xs4all.nl/codec-api/uapi/v4l/dev-decoder.html#decoding
>>
>>>
>>> Yet, it would be interesting to study the firmwares we have and see
>>> what they provide that would help making decode errors more explicit.
>>>
>>
>> Agreed.
>>
>>>>
>>>> Another case is when the bitstream, even if corrupted, is still enough
>>>> to produce some output. My intuition tells me that such CAPTURE buffer
>>>> should be then returned with the ERROR flag. That wouldn't still be
>>>> enough for any more sophisticated userspace error concealment, but
>>>> could still let the userspace know to perhaps drop the frame.
>>>
>>> You mean if a frame was concealed (typically the frame was decoded from
>>> a closed by reference instead of the expected reference). That is
>>> something signalled by FFPEG. We should document this possibility. I
>>> actually have something implemented in GStreamer. Basically if we have
>>> the ERROR flag with a payload size smaller then expected, I drop the
>>> frame and produce a drop event message, while if I have a frame with
>>> ERROR flag but of the right payload size, I assume it is corrupted, and
>>> simply flag it as corrupted, leaving to the application the decision to
>>> display it or not. This is a case that used to happen with some UVC
>>> cameras (though some have been fixed, and the UVC camera should drop
>>> smaller payload size buffers now).
>>
>> I think it's a behavior that makes the most sense indeed.
>>
>> Technically one could also consider the case of 0 < bytesused <
>> sizeimage, which could mean that only a part of the frame is in the
>> buffer. An application could try to blend it with previous frame using
>> some concealing algorithms. I haven't seen an app that could do such
>> thing, though.
> 
> Actually some interesting thought on this. I don't think the existing
> drivers would return any CAPTURE buffers on errors right now, but just
> return the OUTPUT buffer that failed to be decoded. Should we change
> this so that always one CAPTURE buffer with the ERROR flag is
> returned, to signal the application that a frame was potentially
> dropped?

It's what vicodec does. Since device_run is called with both an OUTPUT
and a CAPTURE buffer it would make sense to return both buffers with the
ERROR flag if decoding (or encoding for that matter) fails.

But I would also be fine if this is seen as driver specific. Since an
OUTPUT buffer can decode to multiple CAPTURE buffers it may not be
all that useful in practice.

Regards,

	Hans

Patch

diff --git a/Documentation/media/uapi/v4l/dev-decoder.rst b/Documentation/media/uapi/v4l/dev-decoder.rst
new file mode 100644
index 000000000000..09c7a6621b8e
--- /dev/null
+++ b/Documentation/media/uapi/v4l/dev-decoder.rst
@@ -0,0 +1,1082 @@ 
+.. -*- coding: utf-8; mode: rst -*-
+
+.. _decoder:
+
+*************************************************
+Memory-to-memory Stateful Video Decoder Interface
+*************************************************
+
+A stateful video decoder takes complete chunks of the bitstream (e.g. Annex-B
+H.264/HEVC stream, raw VP8/9 stream) and decodes them into raw video frames in
+display order. The decoder is expected not to require any additional information
+from the client to process these buffers.
+
+Performing software parsing, processing etc. of the stream in the driver in
+order to support this interface is strongly discouraged. In case such
+operations are needed, use of the Stateless Video Decoder Interface (in
+development) is strongly advised.
+
+Conventions and notation used in this document
+==============================================
+
+1. The general V4L2 API rules apply if not specified in this document
+   otherwise.
+
+2. The meaning of words “must”, “may”, “should”, etc. is as per RFC
+   2119.
+
+3. All steps not marked “optional” are required.
+
+4. :c:func:`VIDIOC_G_EXT_CTRLS`, :c:func:`VIDIOC_S_EXT_CTRLS` may be used
+   interchangeably with :c:func:`VIDIOC_G_CTRL`, :c:func:`VIDIOC_S_CTRL`,
+   unless specified otherwise.
+
+5. Single-plane API (see spec) and applicable structures may be used
+   interchangeably with Multi-plane API, unless specified otherwise,
+   depending on decoder capabilities and following the general V4L2
+   guidelines.
+
+6. i = [a..b]: sequence of integers from a to b, inclusive, i.e. i =
+   [0..2]: i = 0, 1, 2.
+
+7. Given an ``OUTPUT`` buffer A, A’ represents a buffer on the ``CAPTURE``
+   queue containing data (decoded frame/stream) that resulted from processing
+   buffer A.
+
+.. _decoder-glossary:
+
+Glossary
+========
+
+CAPTURE
+   the destination buffer queue; for decoder, the queue of buffers containing
+   decoded frames; for encoder, the queue of buffers containing encoded
+   bitstream; ``V4L2_BUF_TYPE_VIDEO_CAPTURE```` or
+   ``V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE``; data are captured from the hardware
+   into ``CAPTURE`` buffers
+
+client
+   application client communicating with the decoder or encoder implementing
+   this interface
+
+coded format
+   encoded/compressed video bitstream format (e.g. H.264, VP8, etc.); see
+   also: raw format
+
+coded height
+   height for given coded resolution
+
+coded resolution
+   stream resolution in pixels aligned to codec and hardware requirements;
+   typically visible resolution rounded up to full macroblocks;
+   see also: visible resolution
+
+coded width
+   width for given coded resolution
+
+decode order
+   the order in which frames are decoded; may differ from display order if the
+   coded format includes a feature of frame reordering; for decoders,
+   ``OUTPUT`` buffers must be queued by the client in decode order; for
+   encoders ``CAPTURE`` buffers must be returned by the encoder in decode order
+
+destination
+   data resulting from the decode process; ``CAPTURE``
+
+display order
+   the order in which frames must be displayed; for encoders, ``OUTPUT``
+   buffers must be queued by the client in display order; for decoders,
+   ``CAPTURE`` buffers must be returned by the decoder in display order
+
+DPB
+   Decoded Picture Buffer; an H.264 term for a buffer that stores a decoded
+   raw frame available for reference in further decoding steps.
+
+EOS
+   end of stream
+
+IDR
+   Instantaneous Decoder Refresh; a type of a keyframe in H.264-encoded stream,
+   which clears the list of earlier reference frames (DPBs)
+
+keyframe
+   an encoded frame that does not reference frames decoded earlier, i.e.
+   can be decoded fully on its own.
+
+macroblock
+   a processing unit in image and video compression formats based on linear
+   block transforms (e.g. H.264, VP8, VP9); codec-specific, but for most of
+   popular codecs the size is 16x16 samples (pixels)
+
+OUTPUT
+   the source buffer queue; for decoders, the queue of buffers containing
+   encoded bitstream; for encoders, the queue of buffers containing raw frames;
+   ``V4L2_BUF_TYPE_VIDEO_OUTPUT`` or ``V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE``; the
+   hardware is fed with data from ``OUTPUT`` buffers
+
+PPS
+   Picture Parameter Set; a type of metadata entity in H.264 bitstream
+
+raw format
+   uncompressed format containing raw pixel data (e.g. YUV, RGB formats)
+
+resume point
+   a point in the bitstream from which decoding may start/continue, without
+   any previous state/data present, e.g.: a keyframe (VP8/VP9) or
+   SPS/PPS/IDR sequence (H.264); a resume point is required to start decode
+   of a new stream, or to resume decoding after a seek
+
+source
+   data fed to the decoder or encoder; ``OUTPUT``
+
+source height
+   height in pixels for given source resolution; relevant to encoders only
+
+source resolution
+   resolution in pixels of source frames being source to the encoder and
+   subject to further cropping to the bounds of visible resolution; relevant to
+   encoders only
+
+source width
+   width in pixels for given source resolution; relevant to encoders only
+
+SPS
+   Sequence Parameter Set; a type of metadata entity in H.264 bitstream
+
+stream metadata
+   additional (non-visual) information contained inside encoded bitstream;
+   for example: coded resolution, visible resolution, codec profile
+
+visible height
+   height for given visible resolution; display height
+
+visible resolution
+   stream resolution of the visible picture, in pixels, to be used for
+   display purposes; must be smaller or equal to coded resolution;
+   display resolution
+
+visible width
+   width for given visible resolution; display width
+
+State machine
+=============
+
+.. kernel-render:: DOT
+   :alt: DOT digraph of decoder state machine
+   :caption: Decoder state machine
+
+   digraph decoder_state_machine {
+       node [shape = doublecircle, label="Decoding"] Decoding;
+
+       node [shape = circle, label="Initialization"] Initialization;
+       node [shape = circle, label="Capture\nsetup"] CaptureSetup;
+       node [shape = circle, label="Dynamic\nresolution\nchange"] ResChange;
+       node [shape = circle, label="Stopped"] Stopped;
+       node [shape = circle, label="Drain"] Drain;
+       node [shape = circle, label="Seek"] Seek;
+       node [shape = circle, label="End of stream"] EoS;
+
+       node [shape = point]; qi
+       qi -> Initialization [ label = "open()" ];
+
+       Initialization -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
+
+       CaptureSetup -> Stopped [ label = "CAPTURE\nbuffers\nready" ];
+
+       Decoding -> ResChange [ label = "Stream\nresolution\nchange" ];
+       Decoding -> Drain [ label = "V4L2_DEC_CMD_STOP" ];
+       Decoding -> EoS [ label = "EoS mark\nin the stream" ];
+       Decoding -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
+       Decoding -> Stopped [ label = "VIDIOC_STREAMOFF(CAPTURE)" ];
+       Decoding -> Decoding;
+
+       ResChange -> CaptureSetup [ label = "CAPTURE\nformat\nestablished" ];
+       ResChange -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
+
+       EoS -> Drain [ label = "Implicit\ndrain" ];
+
+       Drain -> Stopped [ label = "All CAPTURE\nbuffers dequeued\nor\nVIDIOC_STREAMOFF(CAPTURE)" ];
+       Drain -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
+
+       Seek -> Decoding [ label = "VIDIOC_STREAMON(OUTPUT)" ];
+       Seek -> Initialization [ label = "VIDIOC_REQBUFS(OUTPUT, 0)" ];
+
+       Stopped -> Decoding [ label = "V4L2_DEC_CMD_START\nor\nVIDIOC_STREAMON(CAPTURE)" ];
+       Stopped -> Seek [ label = "VIDIOC_STREAMOFF(OUTPUT)" ];
+   }
+
+Querying capabilities
+=====================
+
+1. To enumerate the set of coded formats supported by the decoder, the
+   client may call :c:func:`VIDIOC_ENUM_FMT` on ``OUTPUT``.
+
+   * The full set of supported formats will be returned, regardless of the
+     format set on ``CAPTURE``.
+
+2. To enumerate the set of supported raw formats, the client may call
+   :c:func:`VIDIOC_ENUM_FMT` on ``CAPTURE``.
+
+   * Only the formats supported for the format currently active on ``OUTPUT``
+     will be returned.
+
+   * In order to enumerate raw formats supported by a given coded format,
+     the client must first set that coded format on ``OUTPUT`` and then
+     enumerate formats on ``CAPTURE``.
+
+3. The client may use :c:func:`VIDIOC_ENUM_FRAMESIZES` to detect supported
+   resolutions for a given format, passing desired pixel format in
+   :c:type:`v4l2_frmsizeenum` ``pixel_format``.
+
+   * Values returned by :c:func:`VIDIOC_ENUM_FRAMESIZES` for a coded pixel
+     formats will include all possible coded resolutions supported by the
+     decoder for given coded pixel format.
+
+   * Values returned by :c:func:`VIDIOC_ENUM_FRAMESIZES` for a raw pixel format
+     will include all possible frame buffer resolutions supported by the
+     decoder for given raw pixel format and the coded format currently set on
+     ``OUTPUT``.
+
+4. Supported profiles and levels for given format, if applicable, may be
+   queried using their respective controls via :c:func:`VIDIOC_QUERYCTRL`.
+
+Initialization
+==============
+
+1. **Optional.** Enumerate supported ``OUTPUT`` formats and resolutions. See
+   `Querying capabilities` above.
+
+2. Set the coded format on ``OUTPUT`` via :c:func:`VIDIOC_S_FMT`
+
+   * **Required fields:**
+
+     ``type``
+         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
+
+     ``pixelformat``
+         a coded pixel format
+
+     ``width``, ``height``
+         required only if cannot be parsed from the stream for the given
+         coded format; optional otherwise - set to zero to ignore
+
+     ``sizeimage``
+         desired size of ``OUTPUT`` buffers; the decoder may adjust it to
+         match hardware requirements
+
+     other fields
+         follow standard semantics
+
+   * **Return fields:**
+
+     ``sizeimage``
+         adjusted size of ``CAPTURE`` buffers
+
+   * If width and height are set to non-zero values, the ``CAPTURE`` format
+     will be updated with an appropriate frame buffer resolution instantly.
+     However, for coded formats that include stream resolution information,
+     after the decoder is done parsing the information from the stream, it will
+     update the ``CAPTURE`` format with new values and signal a source change
+     event.
+
+   .. warning::
+
+      Changing the ``OUTPUT`` format may change the currently set ``CAPTURE``
+      format. The decoder will derive a new ``CAPTURE`` format from the
+      ``OUTPUT`` format being set, including resolution, colorimetry
+      parameters, etc. If the client needs a specific ``CAPTURE`` format, it
+      must adjust it afterwards.
+
+3.  **Optional.** Query the minimum number of buffers required for ``OUTPUT``
+    queue via :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to
+    use more buffers than the minimum required by hardware/format.
+
+    * **Required fields:**
+
+      ``id``
+          set to ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
+
+    * **Return fields:**
+
+      ``value``
+          the minimum number of ``OUTPUT`` buffers required for the currently
+          set format
+
+4.  Allocate source (bitstream) buffers via :c:func:`VIDIOC_REQBUFS` on
+    ``OUTPUT``.
+
+    * **Required fields:**
+
+      ``count``
+          requested number of buffers to allocate; greater than zero
+
+      ``type``
+          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
+
+      ``memory``
+          follows standard semantics
+
+    * **Return fields:**
+
+      ``count``
+          the actual number of buffers allocated
+
+    .. warning::
+
+       The actual number of allocated buffers may differ from the ``count``
+       given. The client must check the updated value of ``count`` after the
+       call returns.
+
+    .. note::
+
+       To allocate more than the minimum number of buffers (for pipeline
+       depth), the client may query the ``V4L2_CID_MIN_BUFFERS_FOR_OUTPUT``
+       control to get the minimum number of buffers required by the
+       decoder/format, and pass the obtained value plus the number of
+       additional buffers needed in the ``count`` field to
+       :c:func:`VIDIOC_REQBUFS`.
+
+    Alternatively, :c:func:`VIDIOC_CREATE_BUFS` on the ``OUTPUT`` queue can be
+    used to have more control over buffer allocation.
+
+    * **Required fields:**
+
+      ``count``
+          requested number of buffers to allocate; greater than zero
+
+      ``type``
+          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
+
+      ``memory``
+          follows standard semantics
+
+      ``format``
+          follows standard semantics
+
+    * **Return fields:**
+
+      ``count``
+          adjusted to the number of allocated buffers
+
+    .. warning::
+
+       The actual number of allocated buffers may differ from the ``count``
+       given. The client must check the updated value of ``count`` after the
+       call returns.
+
+5.  Start streaming on the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`.
+
+6.  **This step only applies to coded formats that contain resolution information
+    in the stream.** Continue queuing/dequeuing bitstream buffers to/from the
+    ``OUTPUT`` queue via :c:func:`VIDIOC_QBUF` and :c:func:`VIDIOC_DQBUF`. The
+    buffers will be processed and returned to the client in order, until
+    required metadata to configure the ``CAPTURE`` queue are found. This is
+    indicated by the decoder sending a ``V4L2_EVENT_SOURCE_CHANGE`` event with
+    ``V4L2_EVENT_SRC_CH_RESOLUTION`` source change type.
+
+    * It is not an error if the first buffer does not contain enough data for
+      this to occur. Processing of the buffers will continue as long as more
+      data is needed.
+
+    * If data in a buffer that triggers the event is required to decode the
+      first frame, it will not be returned to the client, until the
+      initialization sequence completes and the frame is decoded.
+
+    * If the client sets width and height of the ``OUTPUT`` format to 0,
+      calling :c:func:`VIDIOC_G_FMT`, :c:func:`VIDIOC_S_FMT` or
+      :c:func:`VIDIOC_TRY_FMT` on the ``CAPTURE`` queue will return the
+      ``-EACCES`` error code, until the decoder configures ``CAPTURE`` format
+      according to stream metadata.
+
+    .. important::
+
+       Any client query issued after the decoder queues the event will return
+       values applying to the just parsed stream, including queue formats,
+       selection rectangles and controls.
+
+    .. note::
+
+       A client capable of acquiring stream parameters from the bitstream on
+       its own may attempt to set the width and height of the ``OUTPUT`` format
+       to non-zero values matching the coded size of the stream, skip this step
+       and continue with the `Capture setup` sequence. However, it must not
+       rely on any driver queries regarding stream parameters, such as
+       selection rectangles and controls, since the decoder has not parsed them
+       from the stream yet. If the values configured by the client do not match
+       those parsed by the decoder, a `Dynamic resolution change` will be
+       triggered to reconfigure them.
+
+    .. note::
+
+       No decoded frames are produced during this phase.
+
+7.  Continue with the `Capture setup` sequence.
+
+Capture setup
+=============
+
+1.  Call :c:func:`VIDIOC_G_FMT` on the ``CAPTURE`` queue to get format for the
+    destination buffers parsed/decoded from the bitstream.
+
+    * **Required fields:**
+
+      ``type``
+          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
+
+    * **Return fields:**
+
+      ``width``, ``height``
+          frame buffer resolution for the decoded frames
+
+      ``pixelformat``
+          pixel format for decoded frames
+
+      ``num_planes`` (for _MPLANE ``type`` only)
+          number of planes for pixelformat
+
+      ``sizeimage``, ``bytesperline``
+          as per standard semantics; matching frame buffer format
+
+    .. note::
+
+       The value of ``pixelformat`` may be any pixel format supported by the
+       decoder for the current stream. The decoder should choose a
+       preferred/optimal format for the default configuration. For example, a
+       YUV format may be preferred over an RGB format if an additional
+       conversion step would be required for the latter.
+
+2.  **Optional.** Acquire the visible resolution via
+    :c:func:`VIDIOC_G_SELECTION`.
+
+    * **Required fields:**
+
+      ``type``
+          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
+
+      ``target``
+          set to ``V4L2_SEL_TGT_COMPOSE``
+
+    * **Return fields:**
+
+      ``r.left``, ``r.top``, ``r.width``, ``r.height``
+          the visible rectangle; it must fit within the frame buffer resolution
+          returned by :c:func:`VIDIOC_G_FMT` on ``CAPTURE``.
+
+    * The following selection targets are supported on ``CAPTURE``:
+
+      ``V4L2_SEL_TGT_CROP_BOUNDS``
+          corresponds to the coded resolution of the stream
+
+      ``V4L2_SEL_TGT_CROP_DEFAULT``
+          the rectangle covering the part of the ``CAPTURE`` buffer that
+          contains meaningful picture data (visible area); width and height
+          will be equal to the visible resolution of the stream
+
+      ``V4L2_SEL_TGT_CROP``
+          the rectangle within the coded resolution to be output to
+          ``CAPTURE``; defaults to ``V4L2_SEL_TGT_CROP_DEFAULT``; read-only on
+          hardware without additional compose/scaling capabilities
+
+      ``V4L2_SEL_TGT_COMPOSE_BOUNDS``
+          the maximum rectangle within a ``CAPTURE`` buffer, which the cropped
+          frame can be output into; equal to ``V4L2_SEL_TGT_CROP`` if the
+          hardware does not support compose/scaling
+
+      ``V4L2_SEL_TGT_COMPOSE_DEFAULT``
+          equal to ``V4L2_SEL_TGT_CROP``
+
+      ``V4L2_SEL_TGT_COMPOSE``
+          the rectangle inside a ``CAPTURE`` buffer into which the cropped
+          frame is written; defaults to ``V4L2_SEL_TGT_COMPOSE_DEFAULT``;
+          read-only on hardware without additional compose/scaling capabilities
+
+      ``V4L2_SEL_TGT_COMPOSE_PADDED``
+          the rectangle inside a ``CAPTURE`` buffer which is overwritten by the
+          hardware; equal to ``V4L2_SEL_TGT_COMPOSE`` if the hardware does not
+          write padding pixels
+
+    .. warning::
+
+       The values are guaranteed to be meaningful only after the decoder
+       successfully parses the stream metadata. The client must not rely on the
+       query before that happens.
+
+3.  Query the minimum number of buffers required for the ``CAPTURE`` queue via
+    :c:func:`VIDIOC_G_CTRL`. This is useful if the client intends to use more
+    buffers than the minimum required by hardware/format.
+
+    * **Required fields:**
+
+      ``id``
+          set to ``V4L2_CID_MIN_BUFFERS_FOR_CAPTURE``
+
+    * **Return fields:**
+
+      ``value``
+          minimum number of buffers required to decode the stream parsed in
+          this initialization sequence.
+
+    .. note::
+
+       The minimum number of buffers must be at least the number required to
+       successfully decode the current stream. This may for example be the
+       required DPB size for an H.264 stream given the parsed stream
+       configuration (resolution, level).
+
+    .. warning::
+
+       The value is guaranteed to be meaningful only after the decoder
+       successfully parses the stream metadata. The client must not rely on the
+       query before that happens.
+
+4.  **Optional.** Enumerate ``CAPTURE`` formats via :c:func:`VIDIOC_ENUM_FMT` on
+    the ``CAPTURE`` queue. Once the stream information is parsed and known, the
+    client may use this ioctl to discover which raw formats are supported for
+    given stream and select one of them via :c:func:`VIDIOC_S_FMT`.
+
+    .. important::
+
+       The decoder will return only formats supported for the currently
+       established coded format, as per the ``OUTPUT`` format and/or stream
+       metadata parsed in this initialization sequence, even if more formats
+       may be supported by the decoder in general.
+
+       For example, a decoder may support YUV and RGB formats for resolutions
+       1920x1088 and lower, but only YUV for higher resolutions (due to
+       hardware limitations). After parsing a resolution of 1920x1088 or lower,
+       :c:func:`VIDIOC_ENUM_FMT` may return a set of YUV and RGB pixel formats,
+       but after parsing resolution higher than 1920x1088, the decoder will not
+       return RGB, unsupported for this resolution.
+
+       However, subsequent resolution change event triggered after
+       discovering a resolution change within the same stream may switch
+       the stream into a lower resolution and :c:func:`VIDIOC_ENUM_FMT`
+       would return RGB formats again in that case.
+
+5.  **Optional.** Set the ``CAPTURE`` format via :c:func:`VIDIOC_S_FMT` on the
+    ``CAPTURE`` queue. The client may choose a different format than
+    selected/suggested by the decoder in :c:func:`VIDIOC_G_FMT`.
+
+    * **Required fields:**
+
+      ``type``
+          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
+
+      ``pixelformat``
+          a raw pixel format
+
+    .. note::
+
+       The client may use :c:func:`VIDIOC_ENUM_FMT` after receiving the
+       ``V4L2_EVENT_SOURCE_CHANGE`` event to find out the set of raw formats
+       supported for the stream.
+
+6.  If all the following conditions are met, the client may resume the decoding
+    instantly:
+
+    * ``sizeimage`` of the new format (determined in previous steps) is less
+      than or equal to the size of currently allocated buffers,
+
+    * the number of buffers currently allocated is greater than or equal to the
+      minimum number of buffers acquired in previous steps. To fulfill this
+      requirement, the client may use :c:func:`VIDIOC_CREATE_BUFS` to add new
+      buffers.
+
+    In such case, the remaining steps do not apply and the client may resume
+    the decoding by one of the following actions:
+
+    * if the ``CAPTURE`` queue is streaming, call :c:func:`VIDIOC_DECODER_CMD`
+      with the ``V4L2_DEC_CMD_START`` command,
+
+    * if the ``CAPTURE`` queue is not streaming, call :c:func:`VIDIOC_STREAMON`
+      on the ``CAPTURE`` queue.
+
+    However, if the client intends to change the buffer set, to lower
+    memory usage or for any other reasons, it may be achieved by following
+    the steps below.
+
+7.  **If the** ``CAPTURE`` **queue is streaming,** keep queuing and dequeuing
+    buffers on the ``CAPTURE`` queue until a buffer marked with the
+    ``V4L2_BUF_FLAG_LAST`` flag is dequeued.
+
+8.  **If the** ``CAPTURE`` **queue is streaming,** call :c:func:`VIDIOC_STREAMOFF`
+    on the ``CAPTURE`` queue to stop streaming.
+
+    .. warning::
+
+       The ``OUTPUT`` queue must remain streaming. Calling
+       :c:func:`VIDIOC_STREAMOFF` on it would abort the sequence and trigger a
+       seek.
+
+9.  **If the** ``CAPTURE`` **queue has buffers allocated,** free the ``CAPTURE``
+    buffers using :c:func:`VIDIOC_REQBUFS`.
+
+    * **Required fields:**
+
+      ``count``
+          set to 0
+
+      ``type``
+          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
+
+      ``memory``
+          follows standard semantics
+
+10. Allocate ``CAPTURE`` buffers via :c:func:`VIDIOC_REQBUFS` on the
+    ``CAPTURE`` queue.
+
+    * **Required fields:**
+
+      ``count``
+          requested number of buffers to allocate; greater than zero
+
+      ``type``
+          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
+
+      ``memory``
+          follows standard semantics
+
+    * **Return fields:**
+
+      ``count``
+          actual number of buffers allocated
+
+    .. warning::
+
+       The actual number of allocated buffers may differ from the ``count``
+       given. The client must check the updated value of ``count`` after the
+       call returns.
+
+    .. note::
+
+       To allocate more than the minimum number of buffers (for pipeline
+       depth), the client may query the ``V4L2_CID_MIN_BUFFERS_FOR_CAPTURE``
+       control to get the minimum number of buffers required, and pass the
+       obtained value plus the number of additional buffers needed in the
+       ``count`` field to :c:func:`VIDIOC_REQBUFS`.
+
+    Alternatively, :c:func:`VIDIOC_CREATE_BUFS` on the ``CAPTURE`` queue can be
+    used to have more control over buffer allocation. For example, by
+    allocating buffers larger than the current ``CAPTURE`` format, future
+    resolution changes can be accommodated.
+
+    * **Required fields:**
+
+      ``count``
+          requested number of buffers to allocate; greater than zero
+
+      ``type``
+          a ``V4L2_BUF_TYPE_*`` enum appropriate for ``CAPTURE``
+
+      ``memory``
+          follows standard semantics
+
+      ``format``
+          a format representing the maximum framebuffer resolution to be
+          accommodated by newly allocated buffers
+
+    * **Return fields:**
+
+      ``count``
+          adjusted to the number of allocated buffers
+
+    .. warning::
+
+       The actual number of allocated buffers may differ from the ``count``
+       given. The client must check the updated value of ``count`` after the
+       call returns.
+
+    .. note::
+
+       To allocate buffers for a format different than parsed from the stream
+       metadata, the client must proceed as follows, before the metadata
+       parsing is initiated:
+
+       * set width and height of the ``OUTPUT`` format to desired coded resolution to
+         let the decoder configure the ``CAPTURE`` format appropriately,
+
+       * query the ``CAPTURE`` format using :c:func:`VIDIOC_G_FMT` and save it
+         until this step.
+
+       The format obtained in the query may be then used with
+       :c:func:`VIDIOC_CREATE_BUFS` in this step to allocate the buffers.
+
+11. Call :c:func:`VIDIOC_STREAMON` on the ``CAPTURE`` queue to start decoding
+    frames.
+
+Decoding
+========
+
+This state is reached after the `Capture setup` sequence finishes succesfully.
+In this state, the client queues and dequeues buffers to both queues via
+:c:func:`VIDIOC_QBUF` and :c:func:`VIDIOC_DQBUF`, following the standard
+semantics.
+
+The contents of the source ``OUTPUT`` buffers depend on the active coded pixel
+format and may be affected by codec-specific extended controls, as stated in
+the documentation of each format.
+
+Both queues operate independently, following the standard behavior of V4L2
+buffer queues and memory-to-memory devices. In addition, the order of decoded
+frames dequeued from the ``CAPTURE`` queue may differ from the order of queuing
+coded frames to the ``OUTPUT`` queue, due to properties of the selected coded
+format, e.g. frame reordering.
+
+The client must not assume any direct relationship between ``CAPTURE``
+and ``OUTPUT`` buffers and any specific timing of buffers becoming
+available to dequeue. Specifically,
+
+* a buffer queued to ``OUTPUT`` may result in no buffers being produced
+  on ``CAPTURE`` (e.g. if it does not contain encoded data, or if only
+  metadata syntax structures are present in it),
+
+* a buffer queued to ``OUTPUT`` may result in more than 1 buffer produced
+  on ``CAPTURE`` (if the encoded data contained more than one frame, or if
+  returning a decoded frame allowed the decoder to return a frame that
+  preceded it in decode, but succeeded it in the display order),
+
+* a buffer queued to ``OUTPUT`` may result in a buffer being produced on
+  ``CAPTURE`` later into decode process, and/or after processing further
+  ``OUTPUT`` buffers, or be returned out of order, e.g. if display
+  reordering is used,
+
+* buffers may become available on the ``CAPTURE`` queue without additional
+  buffers queued to ``OUTPUT`` (e.g. during drain or ``EOS``), because of the
+  ``OUTPUT`` buffers queued in the past whose decoding results are only
+  available at later time, due to specifics of the decoding process.
+
+.. note::
+
+   To allow matching decoded ``CAPTURE`` buffers with ``OUTPUT`` buffers they
+   originated from, the client can set the ``timestamp`` field of the
+   :c:type:`v4l2_buffer` struct when queuing an ``OUTPUT`` buffer. The
+   ``CAPTURE`` buffer(s), which resulted from decoding that ``OUTPUT`` buffer
+   will have their ``timestamp`` field set to the same value when dequeued.
+
+   In addition to the straighforward case of one ``OUTPUT`` buffer producing
+   one ``CAPTURE`` buffer, the following cases are defined:
+
+   * one ``OUTPUT`` buffer generates multiple ``CAPTURE`` buffers: the same
+     ``OUTPUT`` timestamp will be copied to multiple ``CAPTURE`` buffers,
+
+   * multiple ``OUTPUT`` buffers generate one ``CAPTURE`` buffer: timestamp of
+     the ``OUTPUT`` buffer queued last will be copied,
+
+   * the decoding order differs from the display order (i.e. the
+     ``CAPTURE`` buffers are out-of-order compared to the ``OUTPUT`` buffers):
+     ``CAPTURE`` timestamps will not retain the order of ``OUTPUT`` timestamps
+     and thus monotonicity of the timestamps cannot be guaranteed.
+
+During the decoding, the decoder may initiate one of the special sequences, as
+listed below. The sequences will result in the decoder returning all the
+``CAPTURE`` buffers that originated from all the ``OUTPUT`` buffers processed
+before the sequence started. Last of the buffers will have the
+``V4L2_BUF_FLAG_LAST`` flag set. To determine the sequence to follow, the client
+must check if there is any pending event and,
+
+* if a ``V4L2_EVENT_SOURCE_CHANGE`` event is pending, the `Dynamic resolution
+  change` sequence needs to be followed,
+
+* if a ``V4L2_EVENT_EOS`` event is pending, the `End of stream` sequence needs
+  to be followed.
+
+Some of the sequences can be intermixed with each other and need to be handled
+as they happen. The exact operation is documented for each sequence.
+
+Seek
+====
+
+Seek is controlled by the ``OUTPUT`` queue, as it is the source of coded data.
+The seek does not require any specific operation on the ``CAPTURE`` queue, but
+it may be affected as per normal decoder operation.
+
+1. Stop the ``OUTPUT`` queue to begin the seek sequence via
+   :c:func:`VIDIOC_STREAMOFF`.
+
+   * **Required fields:**
+
+     ``type``
+         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
+
+   * The decoder will drop all the pending ``OUTPUT`` buffers and they must be
+     treated as returned to the client (following standard semantics).
+
+2. Restart the ``OUTPUT`` queue via :c:func:`VIDIOC_STREAMON`
+
+   * **Required fields:**
+
+     ``type``
+         a ``V4L2_BUF_TYPE_*`` enum appropriate for ``OUTPUT``
+
+   * The decoder will start accepting new source bitstream buffers after the
+     call returns.
+
+3. Start queuing buffers containing coded data after the seek to the ``OUTPUT``
+   queue until a suitable resume point is found.
+
+   .. note::
+
+      There is no requirement to begin queuing coded data starting exactly
+      from a resume point (e.g. SPS or a keyframe). Any queued ``OUTPUT``
+      buffers will be processed and returned to the client until a suitable
+      resume point is found.  While looking for a resume point, the decoder
+      should not produce any decoded frames into ``CAPTURE`` buffers.
+
+      Some hardware is known to mishandle seeks to a non-resume point. Such an
+      operation may result in an unspecified number of corrupted decoded frames
+      being made available on the ``CAPTURE`` queue. Drivers must ensure that
+      no fatal decoding errors or crashes occur, and implement any necessary
+      handling and workarounds for hardware issues related to seek operations.
+
+   .. warning::
+
+      In case of the H.264 codec, the client must take care not to seek over a
+      change of SPS/PPS. Even though the target frame could be a keyframe, the
+      stale SPS/PPS inside decoder state would lead to undefined results when
+      decoding. Although the decoder must handle such case without a crash or a
+      fatal decode error, the client must not expect a sensible decode output.
+
+4. After a resume point is found, the decoder will start returning ``CAPTURE``
+   buffers containing decoded frames.
+
+.. important::
+
+   A seek may result in the `Dynamic resolution change` sequence being
+   iniitated, due to the seek target having decoding parameters different from
+   the part of the stream decoded before the seek. The sequence must be handled
+   as per normal decoder operation.
+
+.. warning::
+
+   It is not specified when the ``CAPTURE`` queue starts producing buffers
+   containing decoded data from the ``OUTPUT`` buffers queued after the seek,
+   as it operates independently from the ``OUTPUT`` queue.
+
+   The decoder may return a number of remaining ``CAPTURE`` buffers containing
+   decoded frames originating from the ``OUTPUT`` buffers queued before the
+   seek sequence is performed.
+
+   The ``VIDIOC_STREAMOFF`` operation discards any remaining queued
+   ``OUTPUT`` buffers, which means that not all of the ``OUTPUT`` buffers
+   queued before the seek sequence may have matching ``CAPTURE`` buffers
+   produced.  For example, given the sequence of operations on the
+   ``OUTPUT`` queue:
+
+     QBUF(A), QBUF(B), STREAMOFF(), STREAMON(), QBUF(G), QBUF(H),
+
+   any of the following results on the ``CAPTURE`` queue is allowed:
+
+     {A’, B’, G’, H’}, {A’, G’, H’}, {G’, H’}.
+
+.. note::
+
+   To achieve instantaneous seek, the client may restart streaming on the
+   ``CAPTURE`` queue too to discard decoded, but not yet dequeued buffers.
+
+Dynamic resolution change
+=========================
+
+Streams that include resolution metadata in the bitstream may require switching
+to a different resolution during the decoding.
+
+The sequence starts when the decoder detects a coded frame with one or more of
+the following parameters different from previously established (and reflected
+by corresponding queries):
+
+* coded resolution (``OUTPUT`` width and height),
+
+* visible resolution (selection rectangles),
+
+* the minimum number of buffers needed for decoding.
+
+Whenever that happens, the decoder must proceed as follows:
+
+1.  After encountering a resolution change in the stream, the decoder sends a
+    ``V4L2_EVENT_SOURCE_CHANGE`` event with source change type set to
+    ``V4L2_EVENT_SRC_CH_RESOLUTION``.
+
+    .. important::
+
+       Any client query issued after the decoder queues the event will return
+       values applying to the stream after the resolution change, including
+       queue formats, selection rectangles and controls.
+
+2.  The decoder will then process and decode all remaining buffers from before
+    the resolution change point.
+
+    * The last buffer from before the change must be marked with the
+      ``V4L2_BUF_FLAG_LAST`` flag, similarly to the `Drain` sequence above.
+
+    .. warning::
+
+       The last buffer may be empty (with :c:type:`v4l2_buffer` ``bytesused``
+       = 0) and in such case it must be ignored by the client, as it does not
+       contain a decoded frame.
+
+    .. note::
+
+       Any attempt to dequeue more buffers beyond the buffer marked with
+       ``V4L2_BUF_FLAG_LAST`` will result in a -EPIPE error from
+       :c:func:`VIDIOC_DQBUF`.
+
+The client must continue the sequence as described below to continue the
+decoding process.
+
+1.  Dequeue the source change event.
+
+    .. important::
+
+       A source change triggers an implicit decoder drain, similar to the
+       explicit `Drain` sequence. The decoder is stopped after it completes.
+       The decoding process must be resumed with either a pair of calls to
+       :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
+       ``CAPTURE`` queue, or a call to :c:func:`VIDIOC_DECODER_CMD` with the
+       ``V4L2_DEC_CMD_START`` command.
+
+2.  Continue with the `Capture setup` sequence.
+
+.. note::
+
+   During the resolution change sequence, the ``OUTPUT`` queue must remain
+   streaming. Calling :c:func:`VIDIOC_STREAMOFF` on the ``OUTPUT`` queue would
+   abort the sequence and initiate a seek.
+
+   In principle, the ``OUTPUT`` queue operates separately from the ``CAPTURE``
+   queue and this remains true for the duration of the entire resolution change
+   sequence as well.
+
+   The client should, for best performance and simplicity, keep queuing/dequeuing
+   buffers to/from the ``OUTPUT`` queue even while processing this sequence.
+
+Drain
+=====
+
+To ensure that all queued ``OUTPUT`` buffers have been processed and related
+``CAPTURE`` buffers output to the client, the client must follow the drain
+sequence described below. After the drain sequence ends, the client has
+received all decoded frames for all ``OUTPUT`` buffers queued before the
+sequence was started.
+
+1. Begin drain by issuing :c:func:`VIDIOC_DECODER_CMD`.
+
+   * **Required fields:**
+
+     ``cmd``
+         set to ``V4L2_DEC_CMD_STOP``
+
+     ``flags``
+         set to 0
+
+     ``pts``
+         set to 0
+
+   .. warning::
+
+   The sentence can be only initiated if both ``OUTPUT`` and ``CAPTURE`` queues
+   are streaming. For compatibility reasons, the call to
+   :c:func:`VIDIOC_DECODER_CMD` will not fail even if any of the queues is not
+   streaming, but at the same time it will not initiate the `Drain` sequence
+   and so the steps described below would not be applicable.
+
+2. Any ``OUTPUT`` buffers queued by the client before the
+   :c:func:`VIDIOC_DECODER_CMD` was issued will be processed and decoded as
+   normal. The client must continue to handle both queues independently,
+   similarly to normal decode operation. This includes,
+
+   * handling any operations triggered as a result of processing those buffers,
+     such as the `Dynamic resolution change` sequence, before continuing with
+     the drain sequence,
+
+   * queuing and dequeuing ``CAPTURE`` buffers, until a buffer marked with the
+     ``V4L2_BUF_FLAG_LAST`` flag is dequeued,
+
+     .. warning::
+
+        The last buffer may be empty (with :c:type:`v4l2_buffer`
+        ``bytesused`` = 0) and in such case it must be ignored by the client,
+        as it does not contain a decoded frame.
+
+     .. note::
+
+        Any attempt to dequeue more buffers beyond the buffer marked with
+        ``V4L2_BUF_FLAG_LAST`` will result in a -EPIPE error from
+        :c:func:`VIDIOC_DQBUF`.
+
+   * dequeuing processed ``OUTPUT`` buffers, until all the buffers queued
+     before the ``V4L2_DEC_CMD_STOP`` command are dequeued.
+
+   * dequeuing the ``V4L2_EVENT_EOS`` event, if the client subscribed to it.
+
+   .. note::
+
+      For backwards compatibility, the decoder will signal a ``V4L2_EVENT_EOS``
+      event when the last the last frame has been decoded and all frames are
+      ready to be dequeued. It is a deprecated behavior and the client must not
+      rely on it. The ``V4L2_BUF_FLAG_LAST`` buffer flag should be used
+      instead.
+
+3. Once all the ``OUTPUT`` buffers queued before the ``V4L2_DEC_CMD_STOP`` call
+   and the last ``CAPTURE`` buffer are dequeued, the decoder is stopped and it
+   will accept, but not process any newly queued ``OUTPUT`` buffers until the
+   client issues any of the following operations:
+
+   * ``V4L2_DEC_CMD_START`` - the decoder will resume the operation normally,
+
+   * a pair of :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
+     ``CAPTURE`` queue - the decoder will resume the operation normally,
+     however any ``CAPTURE`` buffers still in the queue will be returned to the
+     client,
+
+   * a pair of :c:func:`VIDIOC_STREAMOFF` and :c:func:`VIDIOC_STREAMON` on the
+     ``OUTPUT`` queue - any pending source buffers will be returned to the
+     client and the `Seek` sequence will be triggered.
+
+.. note::
+
+   Once the drain sequence is initiated, the client needs to drive it to
+   completion, as described by the steps above, unless it aborts the process by
+   issuing :c:func:`VIDIOC_STREAMOFF` on any of the ``OUTPUT`` or ``CAPTURE``
+   queues.  The client is not allowed to issue ``V4L2_DEC_CMD_START`` or
+   ``V4L2_DEC_CMD_STOP`` again while the drain sequence is in progress and they
+   will fail with -EBUSY error code if attempted.
+
+   Although mandatory, the availability of decoder commands may be queried
+   using :c:func:`VIDIOC_TRY_DECODER_CMD`.
+
+End of stream
+=============
+
+If the decoder encounters an end of stream marking in the stream, the decoder
+will initiate the `Drain` sequence, which the client must handle as described
+above, skipping the initial :c:func:`VIDIOC_DECODER_CMD`.
+
+Commit points
+=============
+
+Setting formats and allocating buffers trigger changes in the behavior of the
+decoder.
+
+1. Setting the format on the ``OUTPUT`` queue may change the set of formats
+   supported/advertised on the ``CAPTURE`` queue. In particular, it also means
+   that the ``CAPTURE`` format may be reset and the client must not rely on the
+   previously set format being preserved.
+
+2. Enumerating formats on the ``CAPTURE`` queue always returns only formats
+   supported for the current ``OUTPUT`` format.
+
+3. Setting the format on the ``CAPTURE`` queue does not change the list of
+   formats available on the ``OUTPUT`` queue. An attempt to set the ``CAPTURE``
+   format that is not supported for the currently selected ``OUTPUT`` format
+   will result in the decoder adjusting the requested ``CAPTURE`` format to a
+   supported one.
+
+4. Enumerating formats on the ``OUTPUT`` queue always returns the full set of
+   supported coded formats, irrespectively of the current ``CAPTURE`` format.
+
+5. While buffers are allocated on the ``OUTPUT`` queue, the client must not
+   change the format on the queue. Drivers will return the -EBUSY error code
+   for any such format change attempt.
+
+To summarize, setting formats and allocation must always start with the
+``OUTPUT`` queue and the ``OUTPUT`` queue is the master that governs the
+set of supported formats for the ``CAPTURE`` queue.
diff --git a/Documentation/media/uapi/v4l/devices.rst b/Documentation/media/uapi/v4l/devices.rst
index fb7f8c26cf09..12d43fe711cf 100644
--- a/Documentation/media/uapi/v4l/devices.rst
+++ b/Documentation/media/uapi/v4l/devices.rst
@@ -15,6 +15,7 @@  Interfaces
     dev-output
     dev-osd
     dev-codec
+    dev-decoder
     dev-effect
     dev-raw-vbi
     dev-sliced-vbi
diff --git a/Documentation/media/uapi/v4l/pixfmt-v4l2.rst b/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
index 826f2305da01..ca5f2270a829 100644
--- a/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
+++ b/Documentation/media/uapi/v4l/pixfmt-v4l2.rst
@@ -32,6 +32,11 @@  Single-planar format structure
 	to a multiple of the scale factor of any smaller planes. For
 	example when the image format is YUV 4:2:0, ``width`` and
 	``height`` must be multiples of two.
+
+	For compressed formats that contain the resolution information encoded
+	inside the stream, when fed to a stateful mem2mem decoder, the fields
+	may be zero to rely on the decoder to detect the right values. For more
+	details see :ref:`decoder` and format descriptions.
     * - __u32
       - ``pixelformat``
       - The pixel format or type of compression, set by the application.
diff --git a/Documentation/media/uapi/v4l/v4l2.rst b/Documentation/media/uapi/v4l/v4l2.rst
index b89e5621ae69..65dc096199ad 100644
--- a/Documentation/media/uapi/v4l/v4l2.rst
+++ b/Documentation/media/uapi/v4l/v4l2.rst
@@ -53,6 +53,10 @@  Authors, in alphabetical order:
 
   - Original author of the V4L2 API and documentation.
 
+- Figa, Tomasz <[e-mail]>
+
+  - Documented the memory-to-memory decoder interface.
+
 - H Schimek, Michael <[e-mail]>
 
   - Original author of the V4L2 API and documentation.
@@ -61,6 +65,10 @@  Authors, in alphabetical order:
 
   - Documented the Digital Video timings API.
 
+- Osciak, Pawel <[e-mail]>
+
+  - Documented the memory-to-memory decoder interface.
+
 - Osciak, Pawel <[e-mail]>
 
   - Designed and documented the multi-planar API.
@@ -85,7 +93,7 @@  Authors, in alphabetical order:
 
   - Designed and documented the VIDIOC_LOG_STATUS ioctl, the extended control ioctls, major parts of the sliced VBI API, the MPEG encoder and decoder APIs and the DV Timings API.
 
-**Copyright** |copy| 1999-2016: Bill Dirks, Michael H. Schimek, Hans Verkuil, Martin Rubli, Andy Walls, Muralidharan Karicheri, Mauro Carvalho Chehab, Pawel Osciak, Sakari Ailus & Antti Palosaari.
+**Copyright** |copy| 1999-2018: Bill Dirks, Michael H. Schimek, Hans Verkuil, Martin Rubli, Andy Walls, Muralidharan Karicheri, Mauro Carvalho Chehab, Pawel Osciak, Sakari Ailus & Antti Palosaari, Tomasz Figa
 
 Except when explicitly stated as GPL, programming examples within this
 part can be used and distributed without restrictions.
diff --git a/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst b/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
index 85c916b0ce07..2f73fe22a9cd 100644
--- a/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
+++ b/Documentation/media/uapi/v4l/vidioc-decoder-cmd.rst
@@ -49,14 +49,16 @@  The ``cmd`` field must contain the command code. Some commands use the
 
 A :ref:`write() <func-write>` or :ref:`VIDIOC_STREAMON`
 call sends an implicit START command to the decoder if it has not been
-started yet.
+started yet. Applies to both queues of mem2mem decoders.
 
 A :ref:`close() <func-close>` or :ref:`VIDIOC_STREAMOFF <VIDIOC_STREAMON>`
 call of a streaming file descriptor sends an implicit immediate STOP
-command to the decoder, and all buffered data is discarded.
+command to the decoder, and all buffered data is discarded. Applies to both
+queues of mem2mem decoders.
 
-These ioctls are optional, not all drivers may support them. They were
-introduced in Linux 3.3.
+In principle, these ioctls are optional, not all drivers may support them. They were
+introduced in Linux 3.3. They are, however, mandatory for stateful mem2mem decoders
+(as further documented in :ref:`decoder`).
 
 
 .. tabularcolumns:: |p{1.1cm}|p{2.4cm}|p{1.2cm}|p{1.6cm}|p{10.6cm}|
@@ -160,26 +162,36 @@  introduced in Linux 3.3.
 	``V4L2_DEC_CMD_RESUME`` for that. This command has one flag:
 	``V4L2_DEC_CMD_START_MUTE_AUDIO``. If set, then audio will be
 	muted when playing back at a non-standard speed.
+
+	For stateful mem2mem decoders, the command may be also used to restart
+	the decoder in case of an implicit stop initiated by the decoder
+	itself, without the ``V4L2_DEC_CMD_STOP`` being called explicitly.
+	No flags or other arguments are accepted in case of mem2mem decoders.
+	See :ref:`decoder` for more details.
     * - ``V4L2_DEC_CMD_STOP``
       - 1
       - Stop the decoder. When the decoder is already stopped, this
 	command does nothing. This command has two flags: if
 	``V4L2_DEC_CMD_STOP_TO_BLACK`` is set, then the decoder will set
 	the picture to black after it stopped decoding. Otherwise the last
-	image will repeat. mem2mem decoders will stop producing new frames
-	altogether. They will send a ``V4L2_EVENT_EOS`` event when the
-	last frame has been decoded and all frames are ready to be
-	dequeued and will set the ``V4L2_BUF_FLAG_LAST`` buffer flag on
-	the last buffer of the capture queue to indicate there will be no
-	new buffers produced to dequeue. This buffer may be empty,
-	indicated by the driver setting the ``bytesused`` field to 0. Once
-	the ``V4L2_BUF_FLAG_LAST`` flag was set, the
-	:ref:`VIDIOC_DQBUF <VIDIOC_QBUF>` ioctl will not block anymore,
-	but return an ``EPIPE`` error code. If
+	image will repeat. If
 	``V4L2_DEC_CMD_STOP_IMMEDIATELY`` is set, then the decoder stops
 	immediately (ignoring the ``pts`` value), otherwise it will keep
 	decoding until timestamp >= pts or until the last of the pending
 	data from its internal buffers was decoded.
+
+	A stateful mem2mem decoder will proceed with decoding the source
+	buffers pending before the command is issued and then stop producing
+	new frames. It will send a ``V4L2_EVENT_EOS`` event when the last frame
+	has been decoded and all frames are ready to be dequeued and will set
+	the ``V4L2_BUF_FLAG_LAST`` buffer flag on the last buffer of the
+	capture queue to indicate there will be no new buffers produced to
+	dequeue. This buffer may be empty, indicated by the driver setting the
+	``bytesused`` field to 0. Once the buffer with the
+	``V4L2_BUF_FLAG_LAST`` flag set was dequeued, the :ref:`VIDIOC_DQBUF
+	<VIDIOC_QBUF>` ioctl will not block anymore, but return an ``EPIPE``
+	error code. No flags or other arguments are accepted in case of mem2mem
+	decoders.  See :ref:`decoder` for more details.
     * - ``V4L2_DEC_CMD_PAUSE``
       - 2
       - Pause the decoder. When the decoder has not been started yet, the
diff --git a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
index 3ead350e099f..0fc0b78a943e 100644
--- a/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
+++ b/Documentation/media/uapi/v4l/vidioc-g-fmt.rst
@@ -53,6 +53,13 @@  devices that is either the struct
 member. When the requested buffer type is not supported drivers return
 an ``EINVAL`` error code.
 
+A stateful mem2mem decoder will not allow operations on the
+``V4L2_BUF_TYPE_VIDEO_CAPTURE`` or ``V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE``
+buffer type until the corresponding ``V4L2_BUF_TYPE_VIDEO_OUTPUT`` or
+``V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE`` buffer type is configured. If such an
+operation is attempted, drivers return an ``EACCES`` error code. Refer to
+:ref:`decoder` for more details.
+
 To change the current format parameters applications initialize the
 ``type`` field and all fields of the respective ``fmt`` union member.
 For details see the documentation of the various devices types in
@@ -145,6 +152,13 @@  On success 0 is returned, on error -1 and the ``errno`` variable is set
 appropriately. The generic error codes are described at the
 :ref:`Generic Error Codes <gen-errors>` chapter.
 
+EACCES
+    The format is not accessible until another buffer type is configured.
+    Relevant for the V4L2_BUF_TYPE_VIDEO_CAPTURE and
+    V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE buffer types of mem2mem decoders, which
+    require the format of V4L2_BUF_TYPE_VIDEO_OUTPUT or
+    V4L2_BUF_TYPE_VIDEO_OUTPUT_MPLANE buffer type to be configured first.
+
 EINVAL
     The struct :c:type:`v4l2_format` ``type`` field is
     invalid or the requested buffer type not supported.