Re: [PATCH v4 29/36] media: imx: mipi-csi2: enable setting and getting of frame rates

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Tue, 2017-03-14 at 08:34 +0100, Hans Verkuil wrote:
> On 03/13/2017 10:03 PM, Sakari Ailus wrote:
> > Hi Steve,
> > 
> > On Mon, Mar 13, 2017 at 11:06:22AM -0700, Steve Longerbeam wrote:
> >>
> >>
> >> On 03/13/2017 06:55 AM, Philipp Zabel wrote:
> >>> On Mon, 2017-03-13 at 13:27 +0000, Russell King - ARM Linux wrote:
> >>>> On Mon, Mar 13, 2017 at 03:16:48PM +0200, Sakari Ailus wrote:
> >>>>> The vast majority of existing drivers do not implement them nor the user
> >>>>> space expects having to set them. Making that mandatory would break existing
> >>>>> user space.
> >>>>>
> >>>>> In addition, that does not belong to link validation either: link validation
> >>>>> should only include static properties of the link that are required for
> >>>>> correct hardware operation. Frame rate is not such property: hardware that
> >>>>> supports the MC interface generally does not recognise such concept (with
> >>>>> the exception of some sensors). Additionally, it is dynamic: the frame rate
> >>>>> can change during streaming, making its validation at streamon time useless.
> >>>>
> >>>> So how do we configure the CSI, which can do frame skipping?
> >>>>
> >>>> With what you're proposing, it means it's possible to configure the
> >>>> camera sensor source pad to do 50fps.  Configure the CSI sink pad to
> >>>> an arbitary value, such as 30fps, and configure the CSI source pad to
> >>>> 15fps.
> >>>>
> >>>> What you actually get out of the CSI is 25fps, which bears very little
> >>>> with the actual values used on the CSI source pad.
> >>>>
> >>>> You could say "CSI should ask the camera sensor" - well, that's fine
> >>>> if it's immediately downstream, but otherwise we'd need to go walking
> >>>> down the graph to find something that resembles its source - there may
> >>>> be mux and CSI2 interface subdev blocks in that path.  Or we just accept
> >>>> that frame rates are completely arbitary and bear no useful meaning what
> >>>> so ever.
> >>>
> >>> Which would include the frame interval returned by VIDIOC_G_PARM on the
> >>> connected video device, as that gets its information from the CSI output
> >>> pad's frame interval.
> >>>
> >>
> >> I'm kinda in the middle on this topic. I agree with Sakari that
> >> frame rate can fluctuate, but that should only be temporary. If
> >> the frame rate permanently shifts from what a subdev reports via
> >> g_frame_interval, then that is a system problem. So I agree with
> >> Phillip and Russell that a link validation of frame interval still
> >> makes sense.
> >>
> >> But I also have to agree with Sakari that a subdev that has no
> >> control over frame rate has no business implementing those ops.
> >>
> >> And then I agree with Russell that for subdevs that do have control
> >> over frame rate, they would have to walk the graph to find the frame
> >> rate source.
> >>
> >> So we're stuck in a broken situation: either the subdevs have to walk
> >> the graph to find the source of frame rate, or s_frame_interval
> >> would have to be mandatory and validated between pads, same as set_fmt.
> > 
> > It's not broken; what we are missing though is documentation on how to
> > control devices that can change the frame rate i.e. presumably drop frames
> > occasionally.
> > 
> > If you're doing something that hasn't been done before, it may be that new
> > documentation needs to be written to accomodate that use case. As we have an
> > existing interface (VIDIOC_SUBDEV_[GS]_FRAME_INTERVAL) it does make sense
> > to use that. What is not possible, though, is to mandate its use in link
> > validation everywhere.
> > 
> > If you had a hardware limitation that would require that the frame rate is
> > constant, then we'd need to handle that in link validation for that
> > particular piece of hardware. But there really is no case for doing that for
> > everything else.
> > 
> 
> General note: I would strongly recommend that g/s_parm support is removed in
> v4l2_subdev in favor of g/s_frame_interval.
> 
> g/s_parm is an abomination...

Agreed. Just in this specific case I was talking about G_PARM on
the /dev/video node, not the v4l2_subdev nodes. This is currently used
by non-subdev-aware userspace to obtain the framerate from the video
capture device.

> There seem to be only a few i2c drivers that use g/s_parm, so this shouldn't
> be a lot of work.
> 
> Having two APIs for the same thing is always very bad.
> 
> Regards,
> 
> 	Hans
> 

regards
Philipp

_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel



[Index of Archives]     [Linux Driver Backports]     [DMA Engine]     [Linux GPIO]     [Linux SPI]     [Video for Linux]     [Linux USB Devel]     [Linux Coverity]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]
  Powered by Linux