Impact of changes made to umass.c at r232358

[Hans Petter Selasky]

On Friday 06 April 2012 18:19:10 Terence Telkamp wrote:
> One thing I forgot to mention is that I have a machine in the kernel
> debugger after the panic, so I can capture more information if any would
> be helpful.
> 
> -----Original Message-----
> From: owner-freebsd-usb at freebsd.org [mailto:owner-freebsd-usb at freebsd.org]
> On Behalf Of Terence Telkamp Sent: Friday, April 06, 2012 11:17 AM
> To: freebsd-usb at freebsd.org
> Subject: Impact of changes made to umass.c at r232358
> 
> I am seeing a kernel panic in FreeBSD 8.1, which is reproduced after
> physically attaching and detaching a USB device several times.  The kernel
> debugger shows that the panic happens in camisr where the cam_sim and its
> associated mutex are clearly destroyed.  sim->refcount is 0, sim->softc is
> 1 (UMASS_GONE), and the sim->mtx is destroyed (mtx_lock = 6).
> 
> This looks very similar to FreeBSD PR kern/153514, which is unfortunately
> unresolved.
> 
> http://www.freebsd.org/cgi/query-pr.cgi?pr=153514
> 
> 
> Is it possible that the changes made to umass.c at r232358 might fix this
> issue?
> 

Hi,

I need to see the backtrace. The revision you mention fixes a detach panic 
issue.

Current issues are happening inside CAM layer, and that is outside of USB.

If you search this list for usbconfig and umass, you will find some e-mails 
describing how you can stress test umass attach/detach.

--HPS