>Number:         16508
>Category:       kern
>Synopsis:       tlp driver doesn't start correctly on alpha (for me)
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri Apr 26 06:41:00 PDT 2002
>Closed-Date:
>Last-Modified:
>Originator:     Robert Elz
>Release:        NetBSD 1.5ZC sourcves of 2002-04-24 (maybe 25th depends on zone)
>Organization:
	University of Melbourne
>Environment:
	<The following information is extracted from your kernel. Please>
	<append output of "ldd", "ident" where relevant (multiple lines).>
System: NetBSD munnnari.cs.mu.OZ.AU 1.5ZC NetBSD 1.5ZC (MUNNARI) #10: Fri Apr 26 18:18:15 EST 2002 kre@munnnari.cs.mu.OZ.AU:/usr/src/sys/arch/alpha/compile/MUNNARI alpha
Architecture: alpha
Machine: alpha
>Description:
	The tlp driver fails to send/receive packets when the system boots
	using normal boot procedures.  Everything looks fine, but nothing
	gets sent (perhaps received, it is hard to tell)

	Dmesg output for the tlp is ...

tlp0 at pci0 dev 9 function 0: DECchip 21143 Ethernet, pass 4.1
tlp0: interrupting at dec 6600 irq 29
tlp0: DEC , Ethernet address 08:00:2b:87:56:fc
tlp0: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto

	The CPU info ...

NetBSD 1.5ZC (MUNNARI) #10: Fri Apr 26 18:18:15 EST 2002
    kre@munnnari.cs.mu.OZ.AU:/usr/src/sys/arch/alpha/compile/MUNNARI
COMPAQ AlphaServer DS10 466 MHz, s/n AY04111873
8192 byte page size, 1 processor.
total memory = 512 MB
(2824 KB reserved for PROM, 509 MB used by NetBSD)
avail memory = 469 MB
using 3271 buffers containing 26168 KB of memory
mainbus0 (root)
cpu0 at mainbus0: ID 0 (primary), 21264-4
cpu0: VAX FP support, IEEE FP support, Primary Eligible
cpu0: Architecture extensions: 303<PAT,MVI,FIX,BWX>

	I doubt there's much else in the dmesg that will make a
	provide much useful info.

	There are no mii's detected (so I assume none exist).

	The ethernet is connected to some (probably very cheap) 10baseT
	hub at the minute (but that hasn't changed in ages).

	After booting, ifconfig output looks fine, routing table looks
	fine (ie: Nway is doing the right thing, detecting 10baseT etc).

>How-To-Repeat:
	Make a -current kernel on an alpha with 21143 tulip ethernet,
	and have it connected to a 10baseT net (the latter might not be
	a requirement, there's no nearby 100baseTX I can plug in at the
	minute to test that).

	Boot - watch that the network doesn't work at all (just as if the
	interface were not configured)

	This is relatively new, in a 1.5ZB kernel (say a month or so old,
	compiled Mar 12 I believe, would be from sources very close to then)
	there was no problem.   The system was running that kernel until
	this week when I updated it (and the rest of the universe).
>Fix:
	Unknown

Workaround:
	Put the interface in promiscuous mode for a few seconds, allow it
	to receive packets, after that, all is fine.   The easy way to
	accomplish this is to run tcpdump.   I have the following script
	in /etc/rc.d at the minute.   With that there, the system gets
	networking running OK (of course, this wouldn't work if /usr
	was NFS mounted, or something similar).

#!/bin/sh

# PROVIDE: local_stupid
# REQUIRE: mountall
# BEFORE:  cleartmp

local_stupid=YES

. /etc/rc.subr

name="local_stupid"
rcvar=$name
start_cmd="/usr/sbin/tcpdump -c2 -i tlp0 -n ip >/dev/null 2>&1"
stop_cmd=":"

load_rc_config $name
run_rc_command "$1"
>Release-Note:
>Audit-Trail:
>Unformatted:
 	<Please check that the above is correct for the bug being reported,>
 	<and append source date of snapshot, if applicable (one line).>