Re: [RFC 1/2] PCI-Express Non-Transparent Bridge Support

On Sun, Jul 15, 2012 at 04:50:41PM -0700, Jon Mason wrote:

On Sat, Jul 14, 2012 at 10:04:11AM -0700, Greg KH wrote:

quoted

On Fri, Jul 13, 2012 at 02:44:59PM -0700, Jon Mason wrote:

quoted

The NTB device driver is needed to configure these memory windows, doorbell, and
scratch-pad registers as well as use them in such a way as they can be turned
into a viable communication channel to the remote system.  ntb_hw.[ch]
determines the usage model (NTB to NTB or NTB to Root Port) and abstracts away
the underlying hardware to provide access and a common interface to the doorbell
registers, scratch pads, and memory windows.  These hardware interfaces are
exported so that other, non-mainlined kernel drivers can access these.

Why would you have non-mainlined drivers?

Can you submit the drivers at the same time so we see how you are using
these new interfaces?

There are none at this time.  In the near future, the transport will
be modified to use IOAT instead of the CPU copy to improve throughput
performance, and it may be beneficial to have that separate.  If you
wish for me to remove the hooks until it is necessary for that, then I
can.

Yes, please do so, we don't add apis for things that are not in-kernel
as they almost always need to change once we actually get a user of
them, as you know.

thanks,

greg k-h