On Sat, 4 Nov 2000, Stefan Seefeld wrote:

> "Jon M. Taylor" wrote:
> > 
> > On Sat, 4 Nov 2000, Stefan Seefeld wrote:
> > 
> >         Your mail client is wrapping your posts at 84 columns....
> > 
> > > In fact, I think video memory management should be at the very core of GGI, 
>together
> > > with drawing primitives. Every advanced program will require that.
> > 
> >         _Resources_ are at the core of GGI.  Not every target has video
> > memory to manage.  We _should_ write an extension for this, but to do it
> > properly requires a great deal of OS-level code and probably direct
> > integration with the VM subsystems of the OS itself.  Seeing as how
> > Linux's method of "managing" the AGP GART pagetable mappings is to expose
> > a /dev/agpgart and use ioctl()s to establish the mappings <vomit>, or that
> > MTRRs are set/read using write()/read() on /proc/mtrr <gag!>, I think
> > we'll be waiting for this until 2.5/2.6 kernels at the earliest.
> 
> What do the kernel developers think about that ? 

        Right now, they are focused on getting 2.4 out the door.  The
rawio stuff (kiobufs, kiovecs) that went into 2.3/2.4 is a step in the
right direction - it allows userspace to have much more fine-grained
control over the low-level VM mappings (pagetables, physical-to-virtual
page mappings, page cacheing, etc) of memory mapped files.  But I have
not seen any mention of such a comprehensive "page coloring" or "page
tagging" system as I proposed.  Lots of driver subsystems could make
their memory-management routines much more generic if such a system was
present....

> Doesn't Alan work on some video
> architecture ? 

        Alan has worked on the video4linux API in the past - I do not know
if he continues to do so.

> Are they totally satisfied with the current state ? 

        Linus appears to be happy to let the DRI people deal with this
sort of thing for now.  Their system is a _little_ more powerful than
/dev/agpgart (they implement some generic reader/writer locking and
synchronized DMA flushes), but not much.

> Or do they
> assume that there is no life after X ? ;)

        It seems that way.

Jon

---
'Cloning and the reprogramming of DNA is the first serious step in 
becoming one with God.'
        - Scientist G. Richard Seed

Reply via email to