kernel_optimize_test/Documentation/usb/dma.txt

In Linux 2.5 kernels (and later), USB device drivers have additional control
over how DMA may be used to perform I/O operations.  The APIs are detailed
in the kernel usb programming guide (kerneldoc, from the source code).


API OVERVIEW

The big picture is that USB drivers can continue to ignore most DMA issues,
though they still must provide DMA-ready buffers (see DMA-mapping.txt).
That's how they've worked through the 2.4 (and earlier) kernels.

OR:  they can now be DMA-aware.

- New calls enable DMA-aware drivers, letting them allocate dma buffers and
  manage dma mappings for existing dma-ready buffers (see below).

- URBs have an additional "transfer_dma" field, as well as a transfer_flags
  bit saying if it's valid.  (Control requests also have "setup_dma" and a
  corresponding transfer_flags bit.)

- "usbcore" will map those DMA addresses, if a DMA-aware driver didn't do
  it first and set URB_NO_TRANSFER_DMA_MAP or URB_NO_SETUP_DMA_MAP.  HCDs
  don't manage dma mappings for URBs.

- There's a new "generic DMA API", parts of which are usable by USB device
  drivers.  Never use dma_set_mask() on any USB interface or device; that
  would potentially break all devices sharing that bus.


ELIMINATING COPIES

It's good to avoid making CPUs copy data needlessly.  The costs can add up,
and effects like cache-trashing can impose subtle penalties.

- When you're allocating a buffer for DMA purposes anyway, use the buffer
  primitives.  Think of them as kmalloc and kfree that give you the right
  kind of addresses to store in urb->transfer_buffer and urb->transfer_dma,
  while guaranteeing that no hidden copies through DMA "bounce" buffers will
  slow things down.  You'd also set URB_NO_TRANSFER_DMA_MAP in
  urb->transfer_flags:

	void *usb_buffer_alloc (struct usb_device *dev, size_t size,
		int mem_flags, dma_addr_t *dma);

	void usb_buffer_free (struct usb_device *dev, size_t size,
		void *addr, dma_addr_t dma);

  For control transfers you can use the buffer primitives or not for each
  of the transfer buffer and setup buffer independently.  Set the flag bits
  URB_NO_TRANSFER_DMA_MAP and URB_NO_SETUP_DMA_MAP to indicate which
  buffers you have prepared.  For non-control transfers URB_NO_SETUP_DMA_MAP
  is ignored.

  The memory buffer returned is "dma-coherent"; sometimes you might need to
  force a consistent memory access ordering by using memory barriers.  It's
  not using a streaming DMA mapping, so it's good for small transfers on
  systems where the I/O would otherwise tie up an IOMMU mapping.  (See
  Documentation/DMA-mapping.txt for definitions of "coherent" and "streaming"
  DMA mappings.)

  Asking for 1/Nth of a page (as well as asking for N pages) is reasonably
  space-efficient.

- Devices on some EHCI controllers could handle DMA to/from high memory.
  Driver probe() routines can notice this using a generic DMA call, then
  tell higher level code (network, scsi, etc) about it like this:

	if (dma_supported (&intf->dev, 0xffffffffffffffffULL))
		net->features |= NETIF_F_HIGHDMA;

  That can eliminate dma bounce buffering of requests that originate (or
  terminate) in high memory, in cases where the buffers aren't allocated
  with usb_buffer_alloc() but instead are dma-mapped.


WORKING WITH EXISTING BUFFERS

Existing buffers aren't usable for DMA without first being mapped into the
DMA address space of the device.

- When you're using scatterlists, you can map everything at once.  On some
  systems, this kicks in an IOMMU and turns the scatterlists into single
  DMA transactions:

	int usb_buffer_map_sg (struct usb_device *dev, unsigned pipe,
		struct scatterlist *sg, int nents);

	void usb_buffer_dmasync_sg (struct usb_device *dev, unsigned pipe,
		struct scatterlist *sg, int n_hw_ents);

	void usb_buffer_unmap_sg (struct usb_device *dev, unsigned pipe,
		struct scatterlist *sg, int n_hw_ents);

  It's probably easier to use the new usb_sg_*() calls, which do the DMA
  mapping and apply other tweaks to make scatterlist i/o be fast.

- Some drivers may prefer to work with the model that they're mapping large
  buffers, synchronizing their safe re-use.  (If there's no re-use, then let
  usbcore do the map/unmap.)  Large periodic transfers make good examples
  here, since it's cheaper to just synchronize the buffer than to unmap it
  each time an urb completes and then re-map it on during resubmission.

  These calls all work with initialized urbs:  urb->dev, urb->pipe,
  urb->transfer_buffer, and urb->transfer_buffer_length must all be
  valid when these calls are used (urb->setup_packet must be valid too
  if urb is a control request):

	struct urb *usb_buffer_map (struct urb *urb);

	void usb_buffer_dmasync (struct urb *urb);

	void usb_buffer_unmap (struct urb *urb);

  The calls manage urb->transfer_dma for you, and set URB_NO_TRANSFER_DMA_MAP
  so that usbcore won't map or unmap the buffer.  The same goes for
  urb->setup_dma and URB_NO_SETUP_DMA_MAP for control requests.