reproduceable kernel panic when trying to use tap0 interface (sparc64)

[Robert Watson]

On Mon, 29 May 2006, Michael Ortmann wrote:

> im using 6_stable on sparc64 and get a 100% reproduceable kernel panic. it 
> crashes when i try to usr/create the tap0 interface. (i discovered it when i 
> tried to run openvpn). so i guess it may be the tap driver on sparc64. i can 
> provice kernel core and offer my help. i wrote to the sparc64 mailinglist 
> before but now i guess its better off here.

This sounds like a kernel code alignment bug, which is likely easy to fix. 
However...

> == how to reproduce the kernel panic ==
>
> # cat /dev/zero >/dev/tap0
>
> tap0: Ethernet address: 00:bd:00:02:10:00
> panic: trap: memory address not aligned
> cpuid = 0
> KDB: enter: panic
> [thread 449 tid 100044]
> Stopped at kdb_enter+0x3c: ta %xcc, 1

Whoops, the gdb stack trace below looks corrupted and/or wrong.  Could you 
instead provide the output of the "trace" command in DDB?  DDB traces can be 
more reliable under some circumstances, and more resistant to mistakes such as 
matching the wrong kernel to the wrong core, gdb bugs, and so on.

Robert N M Watson