On Sat, Mar 10, 2007 at 02:12:04AM -0500, Dmitry Torokhov wrote:
> On Saturday 10 March 2007 01:55, Greg KH wrote:
> > On Fri, Mar 09, 2007 at 10:54:43PM -0800, Greg KH wrote:
> > > On Sat, Mar 10, 2007 at 01:37:34AM -0500, Dmitry Torokhov wrote:
> > > > Greg,
> > > >
> > > > Please consider applying the patch below. It switches struct device_type
> > > > to using attribute groups which os more flexible. I am using it in my
> > > > input class_device -> device conversion (which is 99% done btw).
> > >
> > > Argh, I never sent you my version of that, did I? Very sorry about
> > > that, I was working on fixing up the device namespace issue first, which
> > > isn't done yet :(
> > >
> > > Anyway, my patch that did that is below, feel free to use it or not if
> > > you want.
> > >
> > > > I looked through -mm and the latest git and there does not seem to be
> > > > any users of struct device_type yet...
> > >
> > > Yes, the input patch below uses it and I have a block-device patch from
> > > Kay in my tree that Andrew doesn't pull from (as it's usually really
> > > messed up and I know to hide this kind of breakage from him...)
> >
> > Oops, that patch didn't use it, this follow-on patch from Kay uses them.
>
> Ok, so input portion in your tree does not use type->attrs so we don't
> have a conflict here. Unless my patch messes up Kay's blockdev patch
> badly I'd like you to accept it. Input uses 3 attribute groups and I
> don't want to open-code their creation/removal.
I'll take your patch and see if it messes up Kay's. If it does, I'm
sure he will fix it up for me later :)