Re: [PATCH v2 1/3] usb: gadget: uvc: stop pump thread on video disable

[Greg KH <gregkh@linuxfoundation.org>]

On Thu, Oct 05, 2023 at 11:48:05AM +0300, Laurent Pinchart wrote:
> On Thu, Oct 05, 2023 at 10:40:10AM +0200, Greg KH wrote:
> > On Thu, Oct 05, 2023 at 11:17:16AM +0300, Laurent Pinchart wrote:
> > > Hi Michael,
> > > 
> > > Thank you for the patch.
> > > 
> > > On Mon, Sep 11, 2023 at 04:05:28PM +0200, Michael Grzeschik wrote:
> > > > Since the uvc-video gadget driver is using the v4l2 interface,
> > > > the streamon and streamoff can be triggered at any times. To ensure
> > > > that the pump worker will be closed as soon the userspace is
> > > > calling streamoff we synchronize the state of the gadget ensuring
> > > > the pump worker to bail out.
> > > 
> > > I'm sorry but I really dislike this. Not only does the patch fail to
> > > ensure real synchronization, as the uvcg_video_pump() function still
> > > runs asynchronously, it messes up the usage of the state field that now
> > > tracks the state both from a host point of view (which it was doing so
> > > far, updating the state based on callbacks from the UDC), and from a
> > > gadget userspace point of view. This lacks clarity and is confusing.
> > > Furthermore, the commit message doesn't even explain what issue is being
> > > fixed here.
> > > 
> > > Greg, I think this series has been merged too soon :-(
> > 
> > Ok, I'll go revert them now, thanks for the review.
> 
> Or we can wait a day for Michael to reply, in case this can quickly be
> fixed on top for v6.7. I'm now reading on the loooon discussion from v1,
> and reviewing the other pending patches that try to tackle the same
> issue.

I'd rather take a patchset that everyone agrees with, reverting was easy
and now done.

thanks,

greg k-h