Re: [PATCH] for ng_ubt2 stalled transfers

From: Lars Engels <lme_at_FreeBSD.org>
Date: Fri, 23 Jan 2009 22:33:23 +0100
On Fri, Jan 23, 2009 at 10:29:52PM +0100, Lars Engels wrote:
> On Fri, Jan 23, 2009 at 11:13:59AM -0800, Maksim Yevmenkin wrote:
> > On Fri, Jan 23, 2009 at 11:12 AM, Maksim Yevmenkin
> > <maksim.yevmenkin_at_gmail.com> wrote:
> > > Lars,
> > >
> > >> Maksim has made a lot of changes to the Bluetooth driver. Maybe he can look
> > >> into it?
> > >>
> > >> This looks like a NULL-pointer access to me.
> > >>
> > >> Maksim:
> > >>
> > >> In the ubt_detach routine, make sure that:
> > >>
> > >> 0) No further commands are accepted from the Netgraph stack.
> > >> 1) You drain the task queue!
> > >> 2) You drain all USB transfers: "usb2_transfer_drain()" called unlocked.
> > >> 3) Set the "sc_node" to NULL.
> > >
> > > please try attached patch for ng_ubt2 that hopefully addresses some of
> > > the issues with stalled transfers. stalled transfers do not happen on
> > > my machine, so i can not test it. i briefly kicked the tires and made
> > > sure other things seems to work fine for me.
> > 
> > sorry, forgot to attach the patch :)
> > 
> > thanks,
> > max
> 
> Thanks for the quick patch, but that did not help:
> 
> ubt0: <Broadcom Corp Foxconn Bluetooth 2.0 plus EDR, class 224/1, rev 2.00/1.00, addr 2> on usbus2
> ubt0:ubt_bulk_read_callback:981: bulk-in transfer failed: USB_ERR_STALLED
> ubt0:ubt_intr_read_callback:834: interrupt transfer failed: USB_ERR_STALLED
> 
> 
> I also have another usb bluetooth stick which shows this on insertion:
> 
> Jan 23 22:13:00 maggie kernel: usb2_alloc_device:1401: set address 2 failed (ignored)
> Jan 23 22:13:00 maggie kernel: usb2_alloc_device:1436: getting device descriptor at addr 2 failed!
> Jan 23 22:13:01 maggie kernel: ugen0.2: <Cambridge Silicon Radio> at usbus0
> Jan 23 22:13:01 maggie kernel: ubt1: <Cambridge Silicon Radio Bluetooth USB dongle, class 224/1, rev 1.10/3.73, addr 2> on usbus0
>

I forgot to mention that both bt devices worked without problems with
the old stack, so there is no hardware issue. :)

Received on Fri Jan 23 2009 - 20:33:24 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:40 UTC