cvs commit: src/sys/kern uipc_mbuf.c src/sys/sys mbuf.h

[Dan Nelson]

In the last episode (Feb 24), Mike Silbersack said:
> FWIW, Dan Nelson had a similar double-free problem a while back, I sent 
> him some additional debugging code to see what might be going wrong, but 
> it must not have happened again, he didn't get back to me.
> 
> Dan, any new news?

Nope; that particular panic has not reappeared on my system since you
sent me that extra M_ASSERTVALID call.  Code knows when it's being
watched.  Probably unrelated, but I have not seen a TX underrun error
since Jan 31, and I had been getting a couple every week before that.

> Since he was using if_dc, that leads me to believe that the problem
> is somewhere in the socket layer or in the IFQ.  If Peter can
> reproduce this problem, have him throw a M_ASSERTVALID() after the
> IFQ_DRV_DEQUEUE, that might prevent the double-free earlier.

-- 
	Dan Nelson
	dnelson at allantgroup.com