Huh? It shouldn't be. Though maybe the 2.4 code has issues that are now gone in 2.6 kernels.
Yes, I talked about the current usb-2.4 bk tree.
OK, that wasn't quite clear. In 2.4, only the EHCI code is really expecting the "hcd.c" support to behave. In general we want the individual HCDs to avoid replicating code for common tasks ... but on 2.4 it's not entirely practical.
pdev is still used at the end of hcd_submit_urb() without some #ifdef PCI around. For somebody that is not deep in the USB code (like me, laughter) it's not really obvious what to do instead at this place.
Hey, it's not clear what do do in any case ... even in 2.6, Linux doesn't have very good support for the necessary things!
That without PCI one has to write his own probe() function should however be doable.
Full support for USB without PCI should be very doable. But most of the patches to enable that haven't been submitted to the mainstream kernels ... either for 2.4 or (sadly) 2.6 yet. Most of that hardware seems to use OHCI, though there are exceptions like the SL-811 support and some Philips chips.
A quick check on hcd.c in usb-2.6 bk revealed that there is no #ifdef PCI anywhere. And I see calls to pci_unmap_single(). Not sure how this would work out on non-PCI systems. The culprit is usb_hcd_giveback_urb().
There's a PCI conditional in the Makefile... but you're right about the giveback() path, it should dma_unmap_single(). Those lines of code never got updated when those DMA calls got added.
- Dave
------------------------------------------------------- This SF.net email is sponsored by: SF.net Giveback Program. Does SourceForge.net help you be more productive? Does it help you create better code? SHARE THE LOVE, and help us help YOU! Click Here: http://sourceforge.net/donate/ _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
