Re: new usb2 stuff

On Tue, Apr 15, 2003 at 08:28:42AM -0400, Bryan Liesner wrote:
> 
> Unless I compile in ohci along with ehci, I get the following error:
> gravy kernel: usb0: wrong number of companions (2 != 0)

EHCI controllers can't work alone.
You need to have either UHCI or OHCI controller compiled in, depending
on what combination your controller chip has.

> Using both ohci and ehci results in probing the attached devices
> twice, then removing the ohci device in favor of the ehci device.
> So, in the dmesg, I see my usb hard drive twice, assigning umass0,
> then umass1. After umass1 is assigned, umass0 is then detached.
> 
> kernel: umass0: Buslink USB 2.0 Hard Drive, rev 2.00/11.00, addr 2
> kernel: umass1: Buslink USB 2.0 Hard Drive, rev 2.00/11.00, addr 2
> kernel: umass0: at uhub2 port 4 (addr 2) disconnected
> kernel: umass0: detached
> 
> Is this the intended behavior?

No, but known:
[88]cicely8# head ehci.c
/* $NetBSD: ehci.c,v 1.46 2003/03/09 19:51:13 augustss Exp $    */
/* $FreeBSD$ */

/*
 * TODO
 *  hold off explorations by companion controllers until ehci has started.
 */

/*
 * Copyright (c) 2001 The NetBSD Foundation, Inc.

The problem is that we have to probe the companion controllers first,
but until the EHCI controller is enabled the 1.1 fallback causes them
to be probed by the companion controllers first.
The EHCI controller than takes over.

The EHCI manpage is on the way.

> I've been using a partitioned off usb hard drive as a tape device,
> just using the raw devices as tapes. (Much cheaper than a tape drive).
> My nightly dump caused a panic, although shorter transfers seem ok.
> 
> I found this in messages:
> kernel: umass1: Invalid CSW: sig 0x00010001 should be 0x53425355

Is this with USB 2.0 or USB 1.x?
Do you use a 2.0 capable cable?
Can you show us usbdevs -v and dmesg output?

> Sorry I don't have more info at this time, I've been flooded at work
> and haven't had the time to reproduce and create a core dump yet.
> 
> Finally, I see no difference in speed between the 1.0 drivers and the
> 2.0 drivers.

I did testings with an ASUS SCB-1608-D drive and was able to see
Datarates up to 3.5M while readings CDs.
I don't know if this was the drive limit, or the limit of my test host.
It was also possible to write CDRs faster than with USB1.1.

-- 
B.Walter                   BWCT                http://www.bwct.de
ticso_at_bwct.de                                  info_at_bwct.de