From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from lb3-smtp-cloud2.xs4all.net ([194.109.24.29]:42716 "EHLO lb3-smtp-cloud2.xs4all.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754238AbdCFJmu (ORCPT ); Mon, 6 Mar 2017 04:42:50 -0500 Subject: Re: [PATCH v2 2/3] v4l: Clearly document interactions between formats, controls and buffers To: Laurent Pinchart References: <20170228150320.10104-1-laurent.pinchart+renesas@ideasonboard.com> <20170228150320.10104-3-laurent.pinchart+renesas@ideasonboard.com> <30dd7931-7c7c-70ed-1a62-00756406761c@xs4all.nl> <3393968.bzjHRkn7sW@avalon> Cc: Laurent Pinchart , linux-media@vger.kernel.org, linux-renesas-soc@vger.kernel.org, Mauro Carvalho Chehab , Hans Verkuil , Sakari Ailus From: Hans Verkuil Message-ID: <661a092f-bc28-81a6-0b0a-cd94a365e1e8@xs4all.nl> Date: Mon, 6 Mar 2017 10:41:17 +0100 MIME-Version: 1.0 In-Reply-To: <3393968.bzjHRkn7sW@avalon> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-media-owner@vger.kernel.org List-ID: On 05/03/17 15:35, Laurent Pinchart wrote: > Hi Hans, > > On Saturday 04 Mar 2017 11:53:45 Hans Verkuil wrote: >> On 28/02/17 16:03, Laurent Pinchart wrote: >>> V4L2 exposes parameters that influence buffers sizes through the format >>> ioctls (VIDIOC_G_FMT, VIDIOC_TRY_FMT and VIDIO_S_FMT). Other parameters >> >> S_SELECTION should be mentioned here as well (more about that later). > > I'll update that according to the conversation we had on IRC about this topic. > >>> not part of the format structure may also influence buffer sizes or >>> buffer layout in general. One existing such parameter is rotation, which >>> is implemented by the VIDIOC_ROTATE control and thus exposed through the >>> V4L2 control ioctls. >>> >>> The interaction between those parameters and buffers is currently only >>> partially specified by the V4L2 API. In particular interactions between >>> controls and buffers isn't specified at all. The behaviour of the >>> VIDIOC_S_FMT ioctl when buffers are allocated is also not fully >>> specified. >>> >>> This commit clearly defines and documents the interactions between >>> formats, controls and buffers. >>> >>> The preparatory discussions for the documentation change considered >>> completely disallowing controls that change the buffer size or layout, >>> in favour of extending the format API with a new ioctl that would bundle >>> those controls with format information. The idea has been rejected, as >>> this would essentially be a restricted version of the upcoming request >>> API that wouldn't bring any additional value. >>> >>> Another option we have considered was to mandate the use of the request >>> API to modify controls that influence buffer size or layout. This has >>> also been rejected on the grounds that requiring the request API to >>> change rotation even when streaming is stopped would significantly >>> complicate implementation of drivers and usage of the V4L2 API for >>> applications. >>> >>> Applications will however be required to use the upcoming request API to >>> change at runtime formats or controls that influence the buffer size or >>> layout, because of the need to synchronize buffers with the formats and >>> controls. Otherwise there would be no way to interpret the content of a >>> buffer correctly. >>> >>> Signed-off-by: Laurent Pinchart >>> >>> --- >>> >>> Documentation/media/uapi/v4l/buffer.rst | 88 ++++++++++++++++++++++++++++ >>> 1 file changed, 88 insertions(+) >>> >>> diff --git a/Documentation/media/uapi/v4l/buffer.rst >>> b/Documentation/media/uapi/v4l/buffer.rst index >>> ac58966ccb9b..5c58db98ab7a 100644 >>> --- a/Documentation/media/uapi/v4l/buffer.rst >>> +++ b/Documentation/media/uapi/v4l/buffer.rst >>> @@ -34,6 +34,94 @@ flags are copied from the OUTPUT video buffer to the >>> CAPTURE video> >>> buffer. >>> >>> +Interactions between formats, controls and buffers >>> +================================================== >>> + >>> +V4L2 exposes parameters that influence the buffer size, or the way data >>> is >>> +laid out in the buffer. Those parameters are exposed through both formats >>> and +controls. One example of such a control is the ``V4L2_CID_ROTATE`` >>> control +that modifies the direction in which pixels are stored in the >>> buffer, as well +as the buffer size when the selected format includes >>> padding at the end of +lines. >>> + >>> +The set of information needed to interpret the content of a buffer (e.g. >>> the +pixel format, the line stride, the tiling orientation or the >>> rotation) is +collectively referred to in the rest of this section as the >>> buffer layout. + >>> +Modifying formats or controls that influence the buffer size or layout >>> require +the stream to be stopped. Any attempt at such a modification >>> while the stream +is active shall cause the format or control set ioctl >>> to return the ``EBUSY`` +error code. >> >> This is not what happens today: it's not the streaming part that causes >> EBUSY to be returned but whether or not buffers are allocated. > > Note that this patch documents the recommended behaviour we want to achieve. > It can differ from the behaviour implemented today as long as today's > behaviour (at least the behaviour of the drivers that we don't consider as > non-compliant, which is the vast majority) is not contradicted by this patch. > In other words, we can recommend here a more generic versatile and powerful > behaviour as long as the more restricted behaviour implemented today is also > allowed. > >> Today we do not support changing buffer sizes on the fly, so any attempt to >> call an ioctl that would change the buffer size is blocked and EBUSY is >> returned. To be precise: drivers call vb2_is_busy() to determine this. >> >> To my knowledge all vb2-using drivers behave like this. There may be old >> drivers that do not do this (and these have a high likelyhood of being >> wrong). > > Please also note that the above description doesn't contradict the behaviour > implemented today. It explains that changing formats and controls is not > allowed while streaming, which is a subset of the restriction implemented by > many drivers through vb2_is_busy() of not allowing changes when buffers are > allocated. > > Please see below for more about this in response to your "here's the 99% case" > comment. > >>> + >>> +Controls that only influence the buffer layout can be modified at any >>> time >>> +when the stream is stopped. As they don't influence the buffer size, no >>> +special handling is needed to synchronize those controls with buffer >>> +allocation. >>> + >>> +Formats and controls that influence the buffer size interact with buffer >>> +allocation. As buffer allocation is an expensive operation, drivers >>> should >>> +allow format or controls that influence the buffer size to be changed >>> with >>> +buffers allocated. A typical ioctl sequence to modify format and controls >>> is + >>> + #. VIDIOC_STREAMOFF >>> + #. VIDIOC_S_FMT >>> + #. VIDIOC_S_EXT_CTRLS >>> + #. VIDIOC_QBUF >>> + #. VIDIOC_STREAMON >>> + >>> +Queued buffers must be large enough for the new format or controls. >>> + >>> +Drivers shall return a ``ENOSPC`` error in response to format change >>> +(:c:func:`VIDIOC_S_FMT`) or control changes (:c:func:`VIDIOC_S_CTRL` or >>> +:c:func:`VIDIOC_S_EXT_CTRLS`) if buffers too small for the new format are >>> +currently queued. As a simplification, drivers are allowed to return an >>> error +from these ioctls if any buffer is currently queued, without >>> checking the +queued buffers sizes. Drivers shall also return a >>> ``ENOSPC`` error from the +:c:func:`VIDIOC_QBUF` ioctl if the buffer >>> being queued is too small for the +current format or controls. >> >> Actually, today qbuf will return -EINVAL (from __verify_length in >> videobuf2-v4l2.c) in these cases. > > No, __verify_length() only checks that the bytesused value doesn't exceed the > buffer length (and only for output buffers, not for capture buffers). > videobuf2 isn't format-aware, so it doesn't check the buffer length against > formats and controls. True, sorry about that. However, buf_prepare *does* check for that (see uvc_buffer_prepare()) and returns -EINVAL. So my comment stands. > >> I am pretty sure you can't change that to ENOSPC since this has always been >> -EINVAL, and so changing this would break the ABI. > > As videobuf2 doesn't perform any such check today, I'm not changing the return > value :-) Sorry, no. Most drivers test this in buf_prepare (and any that do not are almost certainly buggy) and return -EINVAL. > >> Trying to change the format while buffers are allocated will return EBUSY >> today. >> >> If you are trying to document what will happen when drivers allow format >> changes on the fly then this is not at all clear from what you write here. >> >> So: >> >> If the driver does not support changing the format while buffers are queued, >> then it will return EBUSY (true for almost (?) all drivers today). If it >> does support this, then it will behave as described above, except for the >> ENOSPC error in QBUF. >> >> Note that the meaning of ENOSPC should also be explicitly documented in the >> ioctls that can return this. > > Agreed. > >>> Together, these requirements ensure that queued >>> >>> +buffers will always be large enough for the configured format and >>> controls. + >>> +Userspace applications can query the buffer size required for a given >>> format +and controls by first setting the desired control values and then >>> trying the +desired format. The :c:func:`VIDIOC_TRY_FMT` ioctl will >>> return the required +buffer size. >>> + >>> + #. VIDIOC_S_EXT_CTRLS(x) >>> + #. VIDIOC_TRY_FMT() >>> + #. VIDIOC_S_EXT_CTRLS(y) >>> + #. VIDIOC_TRY_FMT() >>> + >>> +The :c:func:`VIDIOC_CREATE_BUFS` ioctl can then be used to allocate >>> buffers +based on the queried sizes (for instance by allocating a set of >>> buffers large +enough for all the desired formats and controls, or by >>> allocating separate set +of appropriately sized buffers for each use >>> case). >>> + >>> +To simplify their implementation, drivers may also require buffers to be >>> +reallocated in order to change formats or controls that influence the >>> buffer +size. In that case, to perform such changes, userspace >>> applications shall +first stop the video stream with the >>> :c:func:`VIDIOC_STREAMOFF` ioctl if it +is running and free all buffers >>> with the :c:func:`VIDIOC_REQBUFS` ioctl if +they are allocated. The >>> format or controls can then be modified, and buffers +shall then be >>> reallocated and the stream restarted. A typical ioctl sequence +is >>> + >>> + #. VIDIOC_STREAMOFF >>> + #. VIDIOC_REQBUFS(0) >>> + #. VIDIOC_S_FMT >>> + #. VIDIOC_S_EXT_CTRLS >>> + #. VIDIOC_REQBUFS(n) >>> + #. VIDIOC_QBUF >>> + #. VIDIOC_STREAMON >>> + >>> +The second :c:func:`VIDIOC_REQBUFS` call will take the new format and >>> control +value into account to compute the buffer size to allocate. >>> Applications can +also retrieve the size by calling the >>> :c:func:`VIDIOC_G_FMT` ioctl if needed. + >>> +When reallocation is required, any attempt to modify format or controls >>> that +influences the buffer size while buffers are allocated shall cause >>> the format +or control set ioctl to return the ``EBUSY`` error code. >> >> Ah, here you describe the 99% situation. This should come first. You've been >> working on the 1% that allows changing formats while buffers are allocated >> so that feels all-important to you, but in reality almost all drivers use >> the 'simplified' implementation. Describe that first, then go on describing >> what will happen for your driver. That will make much more sense to me (as >> you can tell from the preceding comments) and I'm sure to the end-user as >> well. > > This was actually my first, non published version. After discussing it with > Sakari, we decided to change the description. As Sakari mentioned, the > CREATE_BUFS ioctl becomes quite pointless if we don't allow changes with > buffers allocated. We obviously can't mandate it, as most drivers don't allow > it today, but Sakari and I thought we should recommend it. That's what I tried > to convey in this patch by describing the recommended option first. > > The approach I took here was to say > > - formats and controls can't be changed while streaming in any case > - when buffers are allocated but the stream is off, drivers can allow changes > - drivers are not required to allow changes while buffers are allocated > > What's your opinion about that ? I disagree. Drivers needs explicit support to be able to change format when buffers are allocated, but streaming is off: they need to check if the allocated buffers are big enough for the new format. BTW: we need a vb2 helper function for that. No driver does that today. Very few drivers have any need for that, it's only the more complex SoC drivers that would need this, and even then it is of limited functionality: it is really the request API that would make this truly useful. So you should really describe the common situation, then the complex one. Anything else is very confusing for the average reader (heck, it confused me!). The recommended way to organize documentation is to start with the simple, common case, then work your way up to the more complex situations. That should be done here as well. > >> What should be mentioned here as well is that S_SELECTION can also >> implicitly change the format, specifically if there is no scaler or if the >> scaler has limitations. >> >> Also there are a few ioctls that can reset selections and formats when >> called: S_INPUT/S_OUTPUT, S_STD and S_DV_TIMINGS. >> >> I think this should be mentioned here. It can be just in passing, no need to >> go in-depth on that. As long as people are aware of it. > > Agreed, I'll mention all of these. > >> The documentation of S_STD and S_DV_TIMINGS should also be updated saying >> that if the new std or timings differ from the existing std or timings, >> then the format will also change and selection rectangles will be reset to >> the defaults. Setting the std/dv_timings with the current std or timings >> will not do anything: there are applications that do this even when >> streaming so this should be allowed. >> >> Unfortunately, this is not documented but it really should. >> >> If you don't have time to update the S_STD/DV_TIMINGS ioctls, then let me >> know and I will do that. > > I likely won't have time to do that before leaving for holidays so I'd > appreciate if you could update the documentation of those ioctls. I would also > appreciate if we could get this patch merged (or rather a more recent version > thereof) without being blocked by the S_STD/DV_TIMINGS documentation update > :-) No problem. > >>> + >>> + >>> >>> .. c:type:: v4l2_buffer >>> >>> struct v4l2_buffer > > I'll post a v3 now that addresses all your comments except the one about > describing the most restrictive option first. I'm not completely against > Hmm, missing text here? Regards, Hans