Hello,
The following patch is against current gadget-2.4/drivers/usb/gadget/net2280.c
It applies to 2.6 as well.
and addresses some issues with maxpacket isochronous endpoints.
- DMA "scatter/gather" (a.k.a. "ring", USE_DMA_CHAINING) mode cannot be used for IN or OUT endpoints with non-0-mod-4 MPS
Yeah, I want to look at that again ... though I seem to recall some NetChip doc saying the issue was only for IN endpoints.
(Hardware couldn't add the end-of-packet bit to FIFO words in that case, without a software-assisted "validate", but it had no such issue with OUT packets from the host.)
- write_fifo: must also explicitly "validate" max length isochronous packets if maxpacket is not a multiple of 4 bytes.
Shouldn't that happen with interrupt transfers too? Not bulk, which only allows certain sizes; but a three byte interrupt maxpacket is legal.
- dma_done: print some info if being verbose
That is, a per-request trace on the return path. I'll likely add that as an "#if 0" like the per-request trace on submit, or maybe add a USB_TRACE flag like "goku_udc" uses.
Thanks, I'll have a look but won't merge it until after I've run it through some new test software.
I've not really used ISO on the net2280 yet -- is it behaving for you? Including high bandwidth modes?
- Dave
------------------------------------------------------- This SF.net email is sponsored by OSDN's Audience Survey. Help shape OSDN's sites and tell us what you think. Take this five minute survey and you could win a $250 Gift Certificate. http://www.wrgsurveys.com/2003/osdntech03.php?site=8 _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
