Re: wm(4) issue/question

[Chris Ross <cross+netbsd%distal.com@localhost>]

On Mar 2, 2010, at 03:13, SAITOH Masanobu wrote:
> Hello, Chris.

  Hello...

> >  I have an i386 router with a fiber Intel GigE card ("Intel i82543GC
> >  1000BASE-X Ethernet, rev. 2").  This is connected to a SX GBIC on a
> >  switch, and I'm running numerous 802.1q VLANs across it.  This is  
> > all
> >  working fine.
>
> I have a card which has the same chip.
>
> >  I am trying to replace the 10/100 switches (with Gigabit GBIC ports)
> >  with full gigabit switches.  I have a Dell 2724 with SX SFP's in it,
> >  and I tried connecting the i386 with wm(4) to one of them last  
> > night,
> >  after trunking that new switch into the existing fabric.
> >
> >  I was getting 80% or more packet loss between my desktop and the
>
> Strange.

  I also didn't expect it.  :-)

> What NetBSD version are you running? Could you give me the full dmesg
> and ifconfig wm0 of your system?

  I am running a recent (late Jan, early Feb 2010) netbsd-5 image.   
dmesg and ifconfig output attached.

> I'll try to reproduce the bug.
>
> PR#40981 is a bug that device timeouts occur, but I suspect that  
> it's a
> different bug.
>
>         http://www.netbsd.org/cgi-bin/query-pr-single.pl?number=40981

  It may be related, but the circumstances certainly seem different.   
And I'm seeing significant network communication issues, beyond only  
the messages, so that I think makes it a different issue.

  Thanks...

                        - Chris

Attachment: dmesg.boot
Description: Binary data

wm0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
        
capabilities=2bf00<IP4CSUM_Rx,IP4CSUM_Tx,TCP4CSUM_Rx,TCP4CSUM_Tx,UDP4CSUM_Rx,UDP4CSUM_Tx,TCP6CSUM_Tx,UDP6CSUM_Tx>
        enabled=0
        address: 00:03:47:0d:eb:86
        media: Ethernet autoselect (1000baseSX full-duplex,flowcontrol,rxpause)
        status: active
        inet6 fe80::203:47ff:fe0d:eb86%wm0 prefixlen 64 scopeid 0x2