Now that I think about it, probably not -- usb seems like it is a victim,
not the culprit.

Interrupt handling failing somewhere, perhaps? USB also has to deal with
interrupts, and -- another data point - I just managed to get my CATS box to
crash again, this time running the regression suite for the Kaffe JVM...
this time while running the ThreadInterrupt test... and if I ignore it and
go on, I get lots of SEGVs every time any sort of thread-handling code is
invoked that uses interrupts. I can't type "t" or even "show reg" though --
the machine locked up tight after I typed that at the debugger prompt.

This was with kaffe-1.0b3.tar.gz, for reference. I still don't know whether
interrupts or any one thing in particular is causing this, so this is just
my uninformed speculation.

-Edwin

-----Original Message-----
From: Richard Earnshaw [mailto:rearnsha@arm.com]
Sent: Monday, December 14, 1998 7:56 AM
To: Todd Whitesel
Cc: port-arm32@netbsd.org; richard.earnshaw@arm.com
Subject: Re: arm32 kernel crashes 



toddpw@best.com said:
> Aye Carumba!! USB in -current was broken for a few days this week. See
> http://mail-index.netbsd.org/mlist/current-users/1998/12/12/0001.html
> and get a fresh SUP. Uninitialized struct pointers, very bad. 


Hmm, I supped a kernel on Thursday last week, and it locks the machine up 
solid when a PPP connects (mouse and keyboard completely dead), but this 
is on a RISC PC, which shouldn't have any USB code enabled.  Any chance 
these are related?

R.