Re: next-20151222 - compile failure in drivers/media/usb/uvc/uvc_driver.c

[Mauro Carvalho Chehab <mchehab@osg.samsung.com>]

Em Wed, 23 Dec 2015 01:07:31 +0200
Laurent Pinchart <laurent.pinchart@ideasonboard.com> escreveu:

> Hi Mauro,
> 
> On Tuesday 22 December 2015 16:33:50 Mauro Carvalho Chehab wrote:
> > Em Tue, 22 Dec 2015 20:06:38 +0200 Laurent Pinchart escreveu:
> > > On Tuesday 22 December 2015 09:40:43 Javier Martinez Canillas wrote:
> > > > On 12/22/2015 07:18 AM, Valdis Kletnieks wrote:
> > > > > next-20151222 fails to build for me:
> > > > >   CC      drivers/media/usb/uvc/uvc_driver.o
> > > > > 
> > > > > drivers/media/usb/uvc/uvc_driver.c: In function 'uvc_probe':
> > > > > drivers/media/usb/uvc/uvc_driver.c:1941:32: error: 'struct uvc_device'
> > > > > has no member named 'mdev'
> > > > > 
> > > > >   if (media_device_register(&dev->mdev) < 0)
> > > > >                                 ^
> > > > > 
> > > > > scripts/Makefile.build:258: recipe for target
> > > > > 'drivers/media/usb/uvc/uvc_driver.o' failed
> > > > > 
> > > > > 'git blame' points at that line being added in:
> > > > > 
> > > > > commit 1590ad7b52714fddc958189103c95541b49b1dae
> > > > > Author: Javier Martinez Canillas <javier@osg.samsung.com>
> > > > > Date:   Fri Dec 11 20:57:08 2015 -0200
> > > > > 
> > > > >     [media] media-device: split media initialization and registration
> > > > > 
> > > > > Not sure what went wrong here.
> > > > 
> > > > It was my forgetting to test with !CONFIG_MEDIA_CONTROLLER...
> > > > 
> > > > Anyways, I've already posted a fix for this:
> > > > 
> > > > https://lkml.org/lkml/2015/12/21/224
> > > 
> > > Thank you for the fix.
> > > 
> > > I know this is an unpopular request, but can't we make this MC rework
> > > series bisectable ? We're introducing bugs, which is unavoidable given
> > > the scope of the change, and I'm really worried about how difficult we'll
> > > make it to debug them if we keep piling even compilation fixes on top.
> > > 
> > > I can spend a day this week rebasing the patches myself if that could
> > > help.
> > 
> > Laurent,
> > 
> > The problem is that those patches got merged already at media_tree,
> > at the media-controller topic branch.
> > 
> > Any rebase there will break the git copies from all developers that are
> > based on it. It will also break the trees at linuxtv.org, since the
> > developer trees share objects with media_tree.git, in order to save
> > space on the servers.
> 
> But that branch hasn't been merged to master, so it doesn't have to be the one 
> we send upstream, does it ? I'm willing to spend time working on the patches 
> if it can help.

My plan is to send the topic branch on a separate pull request and only
merge it back at master after -rc1.

Yet, as far as I'm aware, the is only one patch that caused compilation
breakages (and only when !CONFIG_MEDIA_CONTROLLER). It should be pretty
easy to fold this patch with the two fixup patches. My plan is to do it
just before sending a pull request. So, no need for you to do the rebase.

Yet, as you said you have some spare time, it would be great if you
could test the MC with the drivers whose I don't have hardware access,
in order to check that there's no regression when using the existing
ioctl:
	- vsp1
	- xilinux
	- staging/davinci_vpfe

Also, with regards to staging/omap4iss, I tried to test it on a
Pandaboard, but the driver seems to be broken/incomplete, as it
doesn't have the required pdata to allow testing it, nor it has
been converted to DT. This seems to be a known issue:
	https://www.mail-archive.com/linux-media@vger.kernel.org/msg89247.html

so, I guess we should add a "depends on BROKEN" for staging/omap4iss
until someone converts it to DT and adds the needed bindings.

Regards,
Mauro



> 
> > What we could try to do is to fold them just before sending the pull
> > request upstream, as we're using tags for pull requests.
> > 
> > I'll do that during the merge window, if someone reminds me about
> > what patches should be fold. I guess there are only two or three
> > patches to be fold, as the only compilation breakages I'm aware are
> > the ones related to Javier's patch series that broke media_device
> > init from the media devnode creation.
>