> > The transfer/setup problem probably indicates that your pci_map_single
> > code (which you had to write for the SH processor, as I recall) is not
> > behaving correctly.
>
> I want to clarify better the problem that I have with transfer and
> setup:
OK, you're right that data structures allocated on the stack are
not guaranteed to be DMA-able, and where such data is passed
as urb transfer or setup buffers, it should get changed in the source.
Your pci_map_single() must be able to handle a kmalloc'd buffer
though -- if it doesn't, that'd be a bug.
I'll follow up on potential pci_pool problems offline till some more
concrete data shows up -- that hasn't happened on other CPUs.
Though it'd be good to verify you saw a problem with the pre5
code, not the 0323/0324 patches.
- Dave
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
http://lists.sourceforge.net/lists/listinfo/linux-usb-devel
