Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 03 Dec 2004 11:40:08 -0800
From:      Julian Elischer <julian@elischer.org>
To:        "M. Warner Losh" <imp@bsdimp.com>
Cc:        freebsd-usb@freebsd.org
Subject:   Re: Question about pipes and ugen
Message-ID:  <41B0C118.8070001@elischer.org>
In-Reply-To: <20041201.204502.51859780.imp@bsdimp.com>
References:  <20041201.160833.35218540.imp@bsdimp.com> <41AE5B3C.8010003@elischer.org> <20041201.204502.51859780.imp@bsdimp.com>

next in thread | previous in thread | raw e-mail | index | archive | help


M. Warner Losh wrote:

>In message: <41AE5B3C.8010003@elischer.org>
>            Julian Elischer <julian@elischer.org> writes:
>: >Right now, when the usb stack opens a pipe, it resets the pipe using
>: >the ClearFeature(Stall) message.  This appears to work great for the
>: >vast majority of usb devices that use things like ugen.  However,
>: >there are some that do not like this at all.  Fidora Core 1 Linux, for
>: >example, never seems to send these reset messages.  Consequently,
>: >there are some devices that work well on Linux, but fail utterly on
>: >FreeBSD.
>: >
>: 
>: On the other hand linux's usblib sends a ClearFeature(halt) as well, 
>: which we don't even have
>: in our version yet..
>: 
>: I'm slowly working on this..
>
>Actually, linux's stack does no such thing in Fidora Core 1.  I've
>verified this with a usb sniffer.
>

I didn't say stack.. I said libusb (excepty I spelled it usblib)

it recovers on several machines where FreeBSD looses communications with 
the target after an error.

>
>: >>From my reading of the standard, it appears that FreeBSD is doing
>: >standards conforming things.  Can anybody confirm this?  Even if it is
>: >standards conforming, is there a desire for FreeBSD to not do this?
>: >
>: 
>: I believe that this would have to be a "quirk" and enterred into the 
>: "quirk" database so that
>: such devices don't try do that..  (The quirks database is something we 
>: did in SCSI
>: and has been duplicated in USB ,where badly behaved devices are 
>: identified and
>: special behaviour can be requested for them.)
>
>I'm not sure that the pipe open commands have access to the requisite
>information...  or if things would even work since the toggle would be
>messed up if we don't send this command...
>
>Warner
>  
>



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?41B0C118.8070001>