[PATCH 0/3] RFC: addition to DMA API

[Alan Stern]

On Fri, 2 Sep 2011, Ming Lei wrote:

> >> I am not sure if there are other devices except for EHCI(maybe have uhci/ohci).
> >
> > Yes: UHCI, OHCI, EHCI, and XHCI all poll memory constantly.
> 
> In fact, the flush may be not required for ohci and xhci case, since there is
> already one mmio register writing at the end of .enqueue path in ohci/xhci
> driver.(just a glance at the code of ohci/xhci, please correct if I am wrong)

I don't know about xhci, but you're right about ohci.  However, there's 
no guarantee that the mmio write will always remain there; somebody 
might change the code so that the write takes place only when it is 
needed instead of every time.

> For UHCI, looks like it has not been used on ARM, so maybe can ignore it.
> UHCI is to support a slow usb 1.1 transfer, so I am wondering if the flush
> can produce a obvious performance boost.

Believe me, even with USB-1.1 a 20-ms delay will be noticeable.

> So looks like the flush only makes sense on EHCI.

Assuming ARM is the only architecture that needs it.

> If the above is not wrong, is it really needed to introduce a general DMA API
> only for EHCI?

Maybe not.  I'm hoping that people will identify the underlying cause
for these delayed write-backs and fix it.  Then no changes at all would
be needed in the USB stack.

Alan Stern