On Sun, 30 Jul 2006, David Brownell wrote: > > It's documented in usb.c:usb_buffer_alloc. Obviously the idea is to map a > > frequently-used DMA buffer just once instead of over and over. > > Erm, no ... buffer allocation and buffer mapping are distinct operations, > that one only affects allocation.
Right. I should have said: "The idea is to allocate a frequently-used DMA buffer in a DMA-consistent memory area and thereby avoid the need to map it over and over." > Note by the way that recent kernels added a new dma operation: > > dma_mapping_error(dma_addr_t) > > returns boolean ... finally, it's at least possible to report mapping > errors, alhough the invalid DMA address is still arch-specific (and usually > all-ones). However hcd_buffer_alloc doesn't use the arch-specific value. It's hard-coded to all-ones. It's also worth pointing out that there is now a dma_supported() routine. We could start using it. > > Are you asking why the method is per-bus? I don't know; the current > > implementation doesn't use this feature. > > At least sl811-hcd could use it, and in general the dma issues need > to be tied to the underlying usb_bus not a usb_device since that's how > the operations work. What could the HCDs do that would be any better than hcd_buffer_alloc? Alan Stern ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys -- and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
