Subject: Re: kern/33935
To: None <kern-bug-people@netbsd.org, gnats-admin@netbsd.org,>
From: George Michaelson <ggm@apnic.net>
List: netbsd-bugs
Date: 02/23/2007 00:20:02
The following reply was made to PR kern/33935; it has been noted by GNATS.
From: George Michaelson <ggm@apnic.net>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/33935
Date: Fri, 23 Feb 2007 10:18:29 +1000
Based on this mail, I believe the PR should close. It appears to be
resolved in current/4.x. -The stall is probably something to be tracked
in another PR.
(I sent the h/w to Matthias)
-G
Begin forwarded message:
Date: Wed, 21 Feb 2007 21:23:26 +0100
From: Matthias Drochner <M.Drochner@fz-juelich.de>
To: George Michaelson <ggm@apnic.net>
Subject: Re: does your work on that iKey device mean I can close this
pr?
ggm@apnic.net said:
> Hi. I wondered if your work with that iKey meant I could close this
> pr:
> http://www.netbsd.org/cgi-bin/query-pr-single.pl?number=33935
I believe I've found what caused these crashes and why I couldn't
reproduce them: The "poll" function checks the endpoint type in the
endpoint descriptor, but the ikey does use the default "control
endpoint" which no descriptor exists for.
An unrelated change (addition of read-ahead) changed the order
of checks so that the code in question is not reached. So -current
(and 4.x) are OK. From code inspection, 3.x is still affected.
Since the read-ahead stuff is too complex to be pulled up to 3.x;
we'll need a simpler patch.
(btw, the "stall after a while" with my USB2 hub seems to be a subtle
timing problem rather than a toggle bit bug. It disappears as soon
as I add some debug printouts.)
best regards
Matthias