misc/55823: em(4) interfaces (Intel Gigabit) are mute/silent/dumb on reboot.

pak at cns.utoronto.ca pak at cns.utoronto.ca
Wed Aug 20 15:20:14 PDT 2003


>Number:         55823
>Category:       misc
>Synopsis:       em(4) interfaces (Intel Gigabit) are mute/silent/dumb on reboot.
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Wed Aug 20 15:20:11 PDT 2003
>Closed-Date:
>Last-Modified:
>Originator:     pak
>Release:        FreeBSD 4.8-RELEASE i386
>Organization:
Computing and Network Services, U Toronto
>Environment:
System: FreeBSD rodent.utcs 4.8-RELEASE FreeBSD 4.8-RELEASE #0: Mon Apr 7 11:38:19 EDT 2003 root@:/usr/src/sys-48/compile/RODENT i386

	  ---

	[ from "pciconf -l -v" on three(3) DELL GX 260's
	  each with an on-board Intel Gigabit NIC ... ]

	em0 at pci2:12:0:  class=0x020000 card=0x002e1028 chip=0x100e8086 rev=0x02 hdr=0x00
	    vendor   = 'Intel Corporation'
	    device   = '82544XT PRO/1000 MT Gigabit Ethernet Controller'
	    class    = network
	    subclass = ethernet

	  ---

	[ from "pciconf -l -v" on a DELL Dimension XPS R450
	  with an Intel PRO/1000 MT Dual Port adapter ... ]

	em0 at pci0:14:0:  class=0x020000 card=0x10118086 chip=0x10108086 rev=0x01 hdr=0x00
	    vendor   = 'Intel Corporation'
	    device   = '82546EB Gigabit Ethernet Controller (copper)'
	    class    = network
	    subclass = ethernet
	em1 at pci0:14:1:  class=0x020000 card=0x10118086 chip=0x10108086 rev=0x01 hdr=0x00
	    vendor   = 'Intel Corporation'
	    device   = '82546EB Gigabit Ethernet Controller (copper)'
	    class    = network
	    subclass = ethernet

	  ---

>Description:

	On reboot, an em(4) interface will not transmit until its
	twisted-pair cable is physically disconnected and then reconnected.

	While the em(4) interface is mute, a tcpdump will indicate that
	traffic is being received from the NIC (so it's not deaf) and
	traffic is being sent the NIC.  However, listening to the wire
	on a neighbouring system shows that em(4) NIC isn't actually
	transmitting anything.  Only after the em(4) NIC's cable is
	disconnected and reconnected does the expected traffic actually
	appear on the wire.

>How-To-Repeat:

	Using 2 machines, a system with an em(4) interface (box1) and
	a neighbouring system (box2) which can see all traffic coming
	from box1's em(4) interface (eg. via "tcpdump").
	Ideally, both systems are connected to the same hub or switch.

	- reboot box1.

	- run "tcpdump -n -l -e -t -i em0" on box1 to see that traffic
	  meant for transmission is being sent to the NIC (ARPs, etc.)
	  and that traffic is also being received by the NIC.

	- watch on box2 for any traffic coming from box1.
	    [ at this point, as far as box2 can tell, box1 should be silent ]

	- if no traffic from box1 is visible to box2, unplug and then
	  reconnect the twisted-pair cable attached to box1's Gigabit NIC.

	- watch on box2 for any traffic coming from box1.
	    [ box1 should be transmitting traffic that can now be seen by box2 ]

>Fix:

	


>Release-Note:
>Audit-Trail:
>Unformatted:


More information about the freebsd-bugs mailing list