Linus,
Its a huge improvement over 2.4, things are broken out into different pieces
much in the same style as 2.5.  However I won't lie in that the probe code
in 2.5 is much cleaner right now since its broken into seperate modules.
Part of the work I've been doing though will enable multiple agp devices by
scaning the bus for agp devices.  That bus scanning code when beefed up will
make the driver initialization trivial and clean.  That bits not quite there
yet.

You will find though that a large portion of this code cleans up alot of
nasties that have been there since the beginning.  Things flow a hell of
alot better and are easier to follow.  Alot of code has been consolidated
and broken into multiple functions.  The code in agp_vm.c is probably the
most interesting from that standpoint, and will give you a basic flavor of
the degree of the changes.  Right now the code does need a little bit of
final organization, and the arch specific bits need to be extracted into a
seperate file from agp_vm.c.

One thing I did find though is that I'm doing some vm stuff that noone else
has done in a module before.  I had to duplicate core kernel functionality
to make this code work as a module on current kernels.  Most of it will be
ripped out before a patch is submitted and the proper symbols will be
exported, so don't be too worried about it.

At any rate, if you have the time give it a gander...

-Jeff

-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]On Behalf Of Linus
Torvalds
Sent: Friday, January 10, 2003 10:46 PM
To: Jeff Hartmann
Cc: Dieter N�tzel; [EMAIL PROTECTED]
Subject: RE: [Dri-devel] [ANNOUNCE] AGPGART Version 2.0



On Fri, 10 Jan 2003, Jeff Hartmann wrote:
>
> I have been in contact with Dave Jones, and I'm going to work with him to
> get a patch into 2.5 with all this functionality, however this is not
based
> on 2.5 stuff.  This is significantly different functionality as to what
has
> been provided in 2.5.  This is bleeding edge new agp 3.0 and agpgart vm
> stuff.

I hope it's at least as well organized as the 2.5.x stuff is these days..
The old AGP code was just horribly badly done from a "everything depended
on everything else" standpoint, and the current 2.5.x code is hugely
improved in this respect. I'd hate to go backwards..

Functionality is often not even as important as maintainability, and we
_finally_ reached a point where the 2.5.x AGP code is at least
maintainable.

                Linus



-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_______________________________________________
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel



-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_______________________________________________
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel

Reply via email to