On Thu, 29 Apr 2010, Jeremy Ramer wrote: > I have traced this hang as far as the function map_urb_for_dma() > called in usb_hcd_submit_urb(). Since it seems DMA-related I have > made the following patch to my driver that seems to have fixed the > hang. Are there any concerns with allocating the DMA coherent memory > in the probe? Is it better to do this allocation on every transfer? If you're going to use coherent memory, you might as well allocate it just once. The real question is whether it's better to use coherent memory vs. ordinary streaming mappings. For the x86 architecture there's not much difference. The memory allocated by usb_alloc_buffer() uses up an entire page, but for only two buffers that won't matter. So you can continue doing this if it solves your problem. On the other hand, I would be interested to learn what's going wrong, if you can track down the cause of your hangs into the DMA mapping code. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
