New acceleration architecture

[Thomas Winischhofer]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Keith Packard wrote:
> On Sat, 2005-07-02 at 15:35 +0200, Thomas Winischhofer wrote:
> 
> 
>>exaDestroyPixmap is eventually called AFTER the driver's CloseScreen()
>>is called. In the driver's CloseScreen(), exaDriverFini() is called (as
>>supposed to, I strongly believe), freeing the ExaScr private and thereby
>>destroying the offscreen memory manager.
> 
> 
> That seems like the real problem; exaDestroyPixmap shouldn't be called
> after CloseScreen. Can you figure out where this call comes from and why
> it wasn't done before the screen was closed?

Tried this, but yet without a result. I can only (reliably) reproduce
this with my old 4MB SiS6326 card in the box; if another card is
installed in the very same box, the call to DestroyPixmap is not issued.
Although I find this hard to believe, it seems to relate to the amount
of memory, strangely. This card has only 4MB, and the offscreen area
handled by exa is only 1.5MB. That is the only difference I can think
of. The exa initialisation code is identical for all my supported
hardware, and so is the rest of the software...

The call is coming from one of the CloseScreen-wrappers following the
driver's own CloseScreen(), but I haven't had time to dive into this
jungle.

The 24/32 bpp problem is still entirely unsolved. Was too busy with the
XGI driver over the weekend to look into this either.

Thomas

- --
Thomas Winischhofer
Vienna/Austria
thomas AT winischhofer DOT net          http://www.winischhofer.net/
twini AT xfree86 DOT org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFCyG3AzydIRAktyUcRAjDYAJ9YLTZkc8p55sTUcAGMSj0G6YTFqgCg1RAy
ZmeGTQxRU8D4FO0Ic67HMuI=
=xspQ
-----END PGP SIGNATURE-----