These patches are against the newest roccat file versions (after
applying everything I released the last days).
They fix some inattentive (though not critical) failures from previous
patches and change hid-roccat-koneplus and hid-roccat-pyra to use the
new sysfs_create_bin_group()
koneplus should be finished now (until I find errors)
Stefan
On Mon, Nov 15, 2010 at 09:32:41PM +0100, Stefan Achatz wrote:
> These patches are against the newest roccat file versions (after
> applying everything I released the last days).
> They fix some inattentive (though not critical) failures from previous
> patches and change hid-roccat-koneplus and hid-roccat-pyra to use the
> new sysfs_create_bin_group()
> koneplus should be finished now (until I find errors)
Wait, they still don't resolve the main problem of this api is still
racy and I can't accept it.
thanks,
greg k-h
On Mon, 15 Nov 2010, Greg KH wrote:
> > These patches are against the newest roccat file versions (after
> > applying everything I released the last days).
> > They fix some inattentive (though not critical) failures from previous
> > patches and change hid-roccat-koneplus and hid-roccat-pyra to use the
> > new sysfs_create_bin_group()
> > koneplus should be finished now (until I find errors)
>
> Wait, they still don't resolve the main problem of this api is still
> racy and I can't accept it.
Yeah, I have already pointed Stefan to
http://lkml.org/lkml/2010/11/13/174
which he might have missed.
Thanks,
--
Jiri Kosina
SUSE Labs, Novell Inc.
> On Mon, 15 Nov 2010, Greg KH wrote:
>
> > > These patches are against the newest roccat file versions (after
> > > applying everything I released the last days).
> > > They fix some inattentive (though not critical) failures from previous
> > > patches and change hid-roccat-koneplus and hid-roccat-pyra to use the
> > > new sysfs_create_bin_group()
> > > koneplus should be finished now (until I find errors)
> >
> > Wait, they still don't resolve the main problem of this api is still
> > racy and I can't accept it.
>
> Yeah, I have already pointed Stefan to
>
> http://lkml.org/lkml/2010/11/13/174
>
> which he might have missed.
Hello again,
I don't see why these patches can't be applied. The bin attribute group functions standing for themself seem to be useful. The kone driver is already in the stable kernel and has less code with these patches. Though this does not change its functionality its a code cleanup. Thats also true for pyra and koneplus. So for me these are clear improvements until I make progress in the race problem. I will do it, but being a newbe I don't know where to start (feel free to point me to something), so it might take quite some time and sure needs some questions asked.
Have a nice day
Stefan
On Tue, 16 Nov 2010, Stefan Achatz wrote:
> > > > These patches are against the newest roccat file versions (after
> > > > applying everything I released the last days).
> > > > They fix some inattentive (though not critical) failures from previous
> > > > patches and change hid-roccat-koneplus and hid-roccat-pyra to use the
> > > > new sysfs_create_bin_group()
> > > > koneplus should be finished now (until I find errors)
> > >
> > > Wait, they still don't resolve the main problem of this api is still
> > > racy and I can't accept it.
> >
> > Yeah, I have already pointed Stefan to
> >
> > http://lkml.org/lkml/2010/11/13/174
> >
> > which he might have missed.
>
> Hello again,
>
> I don't see why these patches can't be applied. The bin attribute group
> functions standing for themself seem to be useful. The kone driver is
> already in the stable kernel and has less code with these patches.
> Though this does not change its functionality its a code cleanup. Thats
> also true for pyra and koneplus. So for me these are clear improvements
> until I make progress in the race problem. I will do it, but being a
> newbe I don't know where to start (feel free to point me to something),
> so it might take quite some time and sure needs some questions asked.
Hi Stefan,
the way the driver does things is racy and needs fixing. Introducing
interface into sysfs/driver core, which we now know is racy, is not
something we'd like to do.
Thanks,
--
Jiri Kosina
SUSE Labs, Novell Inc.