Re: via PCI DMA blitblt added.

[Thomas Hellstrom]

Alan Cox wrote: On Sul, 2005-09-25 at 15:06 +0200, Thomas Hellstrom wrote: *VIA docs are rumored to require the (src_addr%4 == dest_addr%4) for all lines, and this is what is implemented as a sanity check. However, apparently there is a bug in the chip that also requires (system_addr&15 == 0) for all lines. VIA has been contacted about this, but have failed to provide any answers. If these alignments cannot be met, a user-space bounce buffer needs to be used, and this will be implemented for Xv. Really EXA needs to allow the driver to specify this and allocate padded and aligned buffer blocks when neccessary. That would avoid the rather undesirable bounce buffer second copy. I agree fully. For Xv the bounce-buffer copy is quite cheap (system to system memcpy), rarely used and the size of the buffer is seldom altered. For EXA the situation will be very different. For OpenGL, the application will primarily be texture downloads from frame-buffer to system and I'm not sure that Mesa takes into account padding / alignment either? /Thomas   ------------------------------------------------------- SF.Net email is sponsored by: Tame your development challenges with Apache's Geronimo App Server. Download it for free - -and be entered to win a 42" plasma tv or your very own Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php -- _______________________________________________ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel