I've just committed a change to the USB stack (rev 1.130 of
sys/dev/usb_subr.c) that seems to fix the problem people had seen with
Apple iPods and a couple of other devices where the scsibus would
attach but no sd device was ever seen (the iPod was NAKing the INQUIRY
command). The change is to not send a CLEAR_FEATURE(ENDPOINT_HALT)
message when each pipe is opened; the iPod apparantly can't deal with
that command when it is first attached, and it's being invoked in a
kind of belt-and-suspenders manner to set up a device which should be
set up anyway.

However, this changes does alter the sequence of packets we send to
pretty much every USB device, so it has the potential to expose some
devices that are buggy in the opposite direction. If any of your USB
devices suddenly stop working properly, please file a PR.

        - Nathan