Wouldn't it be better if you didn't do the bogus locking in the attach
routine?  It would be better to not lock at all and not initialize the
ISR until last.  That's what I did with if_rl.  Let's not work around
the bogus locking, but insteal eliminate it.  There's no way that the
driver can interrupt itself until an interrupt happens.  If there's no
way to interrupt, no locking is needed :-)

Also, most of the driver locking that's in the tree is likely bogus
anyway, since the network stack still uses giant to a large extent.

Warner

To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Reply via email to