Subject: Re: kern/25659
To: John Refling <>
From: Manuel Bouyer <>
List: netbsd-bugs
Date: 07/08/2005 16:11:45
On Fri, Jul 08, 2005 at 12:06:21AM -0700, John Refling wrote:
>    Well, after a long time to test this, that patch didn't
> do anything to make the superdisk work.  I find it strange
> that most of my usb -> ide stuff does not work under Netbsd.
> These are the superdisks, the HP CD writers, etc.  Not uncommon
> items!

Well, I have only one USB->IDE adapter (I can't remember which brand, but
it also has a firefire connector) and it worked without problems.

>    For fun, I thought I'd replace the CD writer in the HP
> external USB CDR with an IDE hard drive!  That wasn't recognised
> under windows or NetBSD.  I just thought that any IDE device
> would be identified at the end of the USB bus.  Oh Well!

The USB/IDE bridge has to know about the command set of the device plugged
in. ATA and ATAPI are very different, and I'm not surprised that some
of these chips implements only one of them.

>    Note that the original problem, the PCMCIA IDE controller
> really does now work with the initial one line delay() patch that
> you gave me.  I suppose that should get included in the next
> release, if no objections.

It's already commited, and pulled up to netbsd-2 and netbsd-3

>    Are you also the PCMCIA/SCSI guru?  I have an adaptec aha-
> 1450 and -1460 PCMCIA SCSI controller.  I have a simple program
> to just write sequential data to the entire raw partition.
> Works (worked?) fine on the other adaptec aic- controllers (ISA,
> PCI) but on the PCMCIA bus, brings the system load to between 1.3
> and 1.6 and top shows 99% interrupt.  Where should I post that?

Does data get transfered ? It's possible that the system is looping
on interrupts, but it's also possible that it has to transfers all data
via PIO though the slow PCMCIA bus, which would explain the high
CPU load in interrupt context.

Manuel Bouyer <>
     NetBSD: 26 ans d'experience feront toujours la difference