On Oct 29, 2011, at 1:38 AM, Greg KH wrote:
On Fri, Oct 28, 2011 at 11:48:12PM +0200, Hans J. Koch wrote:
quoted
On Fri, Oct 28, 2011 at 10:50:29AM -0500, Kumar Gala wrote:
quoted
For some devices, the default behavior of pgprot_noncached() is not
appropriate for all of its mappable regions. This provides a means =
for
quoted
quoted
the kernel side of the UIO driver to override the flags without =
having
quoted
quoted
to implement its own full mmap callback.
=20
Thanks for also providing an example driver showing the use of this.
You should also post this driver in a mainline-ready version, I'm a =
bit
quoted
uncomfortable with adding a new function pointer without having any =
users.
=20
I'm more than "uncomfortable", I'll refuse to take any such patch =
unless
there is a in-kernel user, otherwise it makes no sense to add the
pointer at all.
=20
thanks,
=20
greg k-h
I'm in agreement with this view. I wanted to post this to make sure the =
direction we took was ok so when the upstream driver is posted this =
patch / change isn't a concern.
- k=