> So now I want to implement mmap-interface. Basically the code in bttv is a > good example of how to do it except for the actual capture part(so different > from voodoo w/ all risc code, etc...). Note: bttv 0.8.x handles this completely different compared to the previous versions. > So now if I get an interrupt signaling of a frame captured to framebuffer > what should I do next? > > 2) have maybe a kernel thread that takes care of that an have my interrupt > handler just wake up the thread? Sounds good. > 3) Could I just map parts of the framebuffer memory with mmap to users to > skip this extra copying? Setting up, say, 10 buffers in framebuffer and use > the onchip 2d-routines to copy/transform data into these. Rather less time > consuming then doing it in main memory. Hmm, I'd try how fast the framebuffer memory is first. Be prepeared that reading from there is much slower than reading from main memory. Given this it may not be worth trying to save that extra copy. Copying it to the (fast) main memory might turn out to be faster overall because you read the data once only from the fb. It's also more flexible as you are not limited to the fb memory for your capture buffers. > There's no way, it seems, to have voodoo-chip transfer data to main > memory :-( That would be the perfect way... Gerd -- Get back there in front of the computer NOW. Christmas can wait. -- Linus "the Grinch" Torvalds, 24 Dec 2000 on linux-kernel