Messages in this thread | | | From | (Alan Cox) | Subject | Re: Linux Graphics Architecture | Date | Mon, 8 Feb 1999 03:14:17 +0000 (GMT) |
| |
> >> There are several other serious problems, including the implicit > >> assumption that graphics hardware will enforce security. > > > > You don't need hardware contexts. > > Read that again. Here: > "must be context-switchable" > "must be able to be performed preemptively" > "in the middle of a command" > > Without that, you can crash the card when two apps send interleaved > commands to the server. Do you intend to disable context switching > and prohibit SMP hardware?
Of course not. The requirement is at the software interface level, not the silicon on the card level.
So you can do several things. Eg you can use the mmu to block access by another context pending I/O completion.
> Consider the common Matrox Millennium: > > 1. write 0x00000000 (commonly a black pixel) to video memory > 2. tell the card to DMA that into the kernel > 3. have fun as root
Ok then the matrox is a case where it can't do direct 2D render. It needs helper code in the Xserver. It isnt the only one - try writing random bit patterns to a Voodoo card (this btw is why the glide device driver isnt worth a lot..)
Alan
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |