cvs commit: src/sys/net netisr.c

[Sam Leffler]

On Wednesday 01 October 2003 03:13 pm, Robert Watson wrote:
> (1) Potential lock order interactions.  I believe these are OK given the
>     current model of dropping the driver lock before entering the
>     interface input routines.  However, since the ground doesn't appear to
>     be firm here quite yet...

All drivers I've looked at either drop the lock or mark it recursive so this 
should not be an issue.  If there are any LOR's then they need to be fixed.

I would like to see all instances of recursive lock usage removed but that 
typically requires significant restructuring of the driver.

	Sam