On Mon, 12 May 2003, Steve Bellovin wrote:

> This sounds a lot like an initialization problem that should be
> fixable.  Is this enough information for someone who understands the
> USB subsystem well enough to hack on it?  (I should note that a disk
> that's inserted at boot time is often not recognized until it's removed
> and reinserted.)

While we are at USB subsystem bugs. There is/was a bug with ohci
showing up with isochronous devices on ugen. Namely whenever
I attached my USB webcam and read some data from it, NetBSD crashed.
Someone posted a workaround, which still does the job on my
1.6.1. (in ohci.c, function ohci_device_isoc_done, comment out
the line "xfer->hcpriv = NULL;").

There was a message on -current with some FreeBSD patches:
http://mail-index.netbsd.org/current-users/2003/04/27/0005.html
including this one for ohci.c:
http://www.clabsms.de/FreeBSD/patch.sys.dev.usb.ohci.c

Is this ohci.c patch worth to look at? Could it be, that it could
solve our problems with isochronous transfers?

Thank you.

-- 
Tom