Re: [PATCH v2 1/7] media: uvcvideo: Use pr_cont() macro

From: Ricardo Ribalda
Date: Wed Nov 04 2020 - 17:32:10 EST


Hi

On Wed, Nov 4, 2020 at 10:51 PM Joe Perches <joe@xxxxxxxxxxx> wrote:
>
> On Wed, 2020-11-04 at 23:42 +0200, Laurent Pinchart wrote:
> > Hi Joe,
>
> Hi Laurent.
>
> > On Wed, Nov 04, 2020 at 11:29:30AM -0800, Joe Perches wrote:
> > > On Wed, 2020-11-04 at 19:07 +0100, Ricardo Ribalda wrote:
> > > > Replace all the uses of printk(KERN_CONT ... with pr_cont().
> > >
> > > Perhaps remove the uvc_printk macro and uses and use the more
> > > common pr_fmt and pr_<level> mechanisms.
> >
> > I'd actually go for dev_* instead, to give some context. It's fairly
> > common to have multiple UVC devices connected to a system, so printing
> > the device name would be useful. It can still be wrapped with
> > uvc_printk() if we want to wrap the cast from uvc_device to a struct
> > device (we should actually try to get the device corresponding to the
> > USB interface where available, so we should use uvc_streaming->intf->dev
> > where possible, and fallback to uvc_device->udev->dev otherwise), or
> > drop the wrapper completely.
>
> Of course yes. I was not going to look around and update the existing
> call sites to find whatever controlling uvc_device * or other struct *
> to a real device that exists though.
>
> It's not even clear from the changes that an appropriate pointer to
> some struct exists in all the functions.
>
> That's work for someone that knows the actual subsystem and I do not.

I have updated my tree with the dev_ variants

https://github.com/ribalda/linux/commit/b8785fd8efb4f2e5bbf5d0f2df3e0d69a5439015

will send it to the tree when I get more feedback from the other patches.

Thanks!

>
> cheers, Joe
>


--
Ricardo Ribalda