On Mon, Feb 02, 2004, Sancho Dauskardt <[EMAIL PROTECTED]> wrote:
> At 22:38 01.02.04 -0500, Alan Stern wrote:
> >Johannes (or anyone else who can help):
> >
> >I could use some advice on tracking this problem down.  You can follow the
> >entire discussion in the thread going forward from
> >
> >http://marc.theaimsgroup.com/?l=linux-usb-devel&m=107550399505642&w=2
> >
> >Basically the problem goes like this.  The device (a digital camera) sends
> >a babble reply to a bulk-in transfer, and after that the host controller
> >doesn't seem to do anything.  In particular, it doesn't raise any
> >interrupt requests, even after uhci_set_next_interrupt().  Note that the
> >controller is still running (the Run-Stop bit in the USBCMD register is
> >set) and the schedule is not corrupted at all.  And interrupt routing is
> >okay, because USB transfers worked correctly prior to the babble
> >occurrence.
> 
> 
> >The email thread shows that detailed debugging that justifies these
> >conclusion.  Furthermore, several people have reported similar problems.
> >The real questions are: Is this actually a fault in the controller
> >hardware?  And if it is, how can we work around it?
> 
> It's not a bug, it's a feature: VIA UHCI controllers turn themselves off 
> after bable. HC_RESET is the only way to get them going again.
> I've got a USB card-reader which babbles on card-removal.
> changes along the line of
>         if( status & TD_CTRL_BABBLE  ){
>                 reset_hc();
>                 ...
>         }
> hacked into uhci_hcd.c::hacked into uhci_map_status() result in enumeration 
> after babble, but keep the entire usb system running.

That's liberal use of the term "feature", but this VIA problem is known.

I think even usb-uhci.c has a workaround for it (I don't know how well
it works).

I had been waiting for a better solution since reenumerating doesn't
seem like an acceptable workaround to me. No device should cause this
sort of problem.

But I guess it's better than leaving the HC completely useless.

JE



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to