Dave Airlie wrote:
> On 5/7/07, Thomas Hellström <[EMAIL PROTECTED]> wrote:
>> Dave Airlie wrote:
>> >> I've started a cleanup branch,
>> >>
>> >> I've just checked in the mm_ioctl split out into separate parts, I'll
>> >> try and get fence and buffer done as well..
>> >>
>> >> This will break compatiblity but to be honest, anyone who has
>> deployed
>> >> this beyond embedded system work (i.e. TG and me), deserves what they
>> >> get for integrating unreleased code :-)
>> >
>> > Of course quite how to fix up the buffer object ioctl chain scheme I'm
>> > not sure, I'm not 100% sure this is really a win in most situations??
>> > have we any numbers?
>> >
>> > Dave.
>> Dave,
>> It's good that you've found time to do this.
>>
>> The IOCTL chaining is presently only used to submit a list of buffers
>> for validation. Any other use can go away, and if we can find a better
>> way of submitting a buffer list (be it a linked list or array) for a
>> validatebufferlist we can and should probably use that.
>
> Okay I've cleaned up this stuff as best I can, it builds I haven't
> tested it as my crash box is hooked up doing some other testing at the
> moment... but I kept the list for the fence/validate operations...
>
> I've also removed the typedefs from most of the code and I don't see a
> major impact on readability so I may proceed to do this in more
> places..
>
Cool!
I'll try it out as soon as there is time.
BTW what rules should we use for scalar types in drm?
I assume "unsigned", "int", "unsigned long" etc are OK when no specific
size is needed in kernel code?
"u32" and similar can be used in linux-only kernel code. In shared code
we use "uint32_t" and similar?
"__u32" is used in linux-only user-space interface code. What about
shared interface code? "unsigned"? "uint32_t"?
>>
>
> This interface sounds okay for tiling, I'd need to play around with it
> to be sure it would work on radeon, which I probably won't get to
> before I go on holidays... but I was looking at doing this for my own
> project as it might save me 2-3MBs of memory I'm needlessly mapping
> into the GTT at the moment to fill in the tile gaps..
>
Yes, I have an agpgart patch around somewhere that allows tiled
insertions and removals, but it still needs some form of TTM support.
I'll be needing something similar towards the summer as well, mainly to
be able to tile private back-buffers with a limited number of tiling
("fence") registers. With the buffer object vm we can reuse tiling
registers, Not from GPU-validated buffers but at least from mapped
buffers. We can kill the user-space mappings and steal the tiling
register. As soon as a nopfn() is triggered we give it (or another one)
back, and remap the page(s).
Let's keep in touch about this.
> Dave.
/Thomas
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
--
_______________________________________________
Dri-devel mailing list
Dri-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dri-devel
