On Mon, Dec 10, 2001 at 10:06:45AM -0800, Philip Brown wrote:
>On Mon, Dec 10, 2001 at 09:56:28AM -0800, Daryll Strauss wrote:
>> On Mon, Dec 10, 2001 at 09:50:42AM -0800, Philip Brown wrote:
>> > But I thought that GATT is simply a scatter/gather table, so
>> > you only have to update the GATT when you "allocate and bind" pages.
>> > Then, if you "allocate and bind" the whole range at once, you're done, and
>> > you dont have to do any cache flushing from that point on.
>> > So sounds like you are agreeing with me?? :->
>> The interface was developed with input from XFree86, nVidia, and Xi
>> Graphics. It is also intended to support architectures other than
>> x86. Therefore this is a nice general solution to the problem.
>Fair enough, as far as architecture goes.
>But since I'm only coding a back-end for solaris, which only has AGP on
>x86, do you see a problem with me IMPLEMENTING it this way? :-)
Here is the XFree86 point of view: If you want the result to be
usable with XFree86, then just make sure that it will fit in with
the AGP abstractions provided in XFree86's os-support layer.
XFree86 drivers that need AGP GART functionality, like the i810
driver, use solely those interfaces. You can hide the OS-specific
details in the Solaris part of the XFree86 os-support lyaer.
David Dawes Email: dawes@...
Founder/President, Release Engineer Phone: +1 570 764 0288
The XFree86 Project, Inc http://www.xfree86.org/~dawes