On Thu, Jul 9, 2020 at 12:57 AM Gerd Hoffmann <kra...@redhat.com> wrote:
> > Starting at line 1746 is the first CBW, it's for an Inquiry command.
> >
> > Starting at line 1759 is the response, notice at line 1761 the MSD debug
> > says "Data in 64/36", which is strange.
>
> Not really. First is the packet size, second is the (remaining) data
> size. Inquiry data is 36 bytes, and dwc2 uses 64 byte instead of 36
> byte transfers.
>
> > Then the MSD defers the packet, even though the full 36 bytes has
> > already been received.
>
> Yes, and this is the problem. The condition checks whenever there is
> room left in the usb packet. But we should also check whenever there
> is actually more data pending, so how about this:
>
> if (p->actual_length < p->iov.size && s->mode == USB_MSDM_DATAIN) {
> DPRINTF("Deferring packet %p [wait data-in]\n", p);
>
> take care,
> Gerd
>
>
Hi Gerd,
Hey, that works! But we still need to keep the rest of 7ad3d51ebb8a,
or else s->data_len eventually goes negative and we hit the assertion
on line 447 "Assertion `le32_to_cpu(s->csw.residue) == 0' failed.
But with that, hcd-dwc2 seems to work fine, and also the error
messages that Sai saw with hcd-xhci are gone.
Do you want to submit the patch for this?
Thanks,
Paul
