>Number:         7801
>Category:       kern
>Synopsis:       1.4 ppp doesn't restart properly(?), eats kernel network buffers.
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people (Kernel Bug People)
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Jun 17 10:50:01 1999
>Last-Modified:
>Originator:     Chris Demetriou
>Organization:
>Release:        1.4
>Environment:
NetBSD brick 1.4 NetBSD 1.4 (BRICK) #0: Wed Jun 16 19:19:08 PDT 1999     cgd@speedy:/a/users/cgd/proj/netbsd/src-1-4-branch/sys/arch/i386/compile/BRICK i386


>Description:
I'm using 1.4's PPP over a dialup modem, connecting to an ISP.  At some
points, it seems to get into a wedged state where data is just queued
in the kernel (and presumably not passed on to the remoted end).
When this happens, netstat -m reports:

136 mbufs in use:
        134 mbufs allocated to data
        2 mbufs allocated to socket names and addresses
30/46 mapped pages in use
112 Kbytes allocated to network (68% in use)
0 requests for memory denied
0 requests for memory delayed
0 calls to protocol drain routines

HUPping pppd causes the network buffers to be drained (presumably
tossed), then the PPP connection restarted, then operation to continue
correctly (until the problem happens again).

This has happened probably 3 or 4 times to my gateway machine since
i've upgraded to 1.4.  (previously, i don't think i ever observed it.)

Unfortunately, I don't have much in the way of a clue about what
causes this problem (since in all but i think one case where it's
happened, i've been off doing something else and my network has
been going merrily along on its own 8-), nor do i have much of a clue
what to do to get more of a clue, since i'm not particularly PPP cluefull.
>How-To-Repeat:
Use 1.4 to do PPP?

I'd very much appreciate guidance from somebody with clue about what I
could do to provide more useful debugging information.
>Fix:


>Audit-Trail:
>Unformatted: