Re: [usb] Test Suites for USB Host

[Mike Kentley <mkentley@highdesertdesigncenter.com>]

The testbench requriements really should follow from a verification plan
document based upon the USB device module spec and the USB 2.0
specification. That's how you make sure you got all the necessary
features in the BFM's.  Here's a short list of some favorite things to
test.

1) Should be able to inject errors into any field of outgoing
packet(i.e. pid/crc/data/sof)
2) Should be able to vary host's outgoing clock to stress device's clock
recovery circuit.
3) Should be able to send any type of packet at any time
    a)  need to be able to simulate the case where the host did not
correctly receive the ACK to setup packet, I should be able to send back
to back setup packets
    b) should be able to follow the same enumeration sequence that a PC
based host controller would follow.
    c) need to be able to send "incorrect" handshakes to packets to see
how the device handles them.
4) You can probably leave out support for split transactions, as they
are invisible to the USB device (unless you plan to build a hub).

Something to think about: Do you need a "sideband" channel between the
USB host and device to maintain synchronization during the tests?
 
 
begin:vcard 
n:Kentley;Mike
tel;pager:<a href="mailto:5419483810@page.nextel.com">Page Me!</a>
tel;cell:541-948-3810
tel;fax:541-318-6148
tel;work:541-948-3810
x-mozilla-html:TRUE
url:http://www.highdesertdesigncenter.com
org:High Desert Design Center, Inc.
adr:;;;Bend;OR;;USA
version:2.1
email;internet:mkentley@ieee.org
title:IC Design Consultant
fn:Mike Kentley
end:vcard