Re: Call for testers - dc ethernet driver patch

On Wed, 7 Jan 2004 11:32:23 -0800 (PST), Don Lewis <truckman_at_FreeBSD.org> 
wrote:

> The dc driver has been broken ever since it was converted to use busdma.
> The symptoms are that it reports spurious output errors and collisions,
> even when running in full-duplex mode.  I believe the patch below fixes
> the problem, but I do not have the appropriate hardware to test it.  I
> have had one gotten one report so far from someone who claims this patch
> fixed the problems he was seeing.  I'd appreciate further testing of
> this patch, especially with as many of the chipsets supported by the
> driver as possible so that I can commit this patch.  Please mention the
> type of hardware you are using (machine architecture and card probe
> message) in your reports. This is somewhat time critical because I've
> had users express interest in getting this into 5.2.

Thanks, it's improvement.. I have LinkSys v2.0 and 5.0, I usually get like 
two to three 'TX underrun.....' times when I transfer the over 1gig 
between two machines and now I only get one. I have tested it like three 
times and just one at each time. The collisions is better.

Before:
==================
Name    Mtu Network       Address              Ipkts Ierrs    Opkts Oerrs  
Coll
dc0    1500 <Link#1>      00:04:5a:79:36:96   192317     0   148854     0 
21803

I usually get the collisions by the small things such as only surf, ssh or 
whatever.
==================

After:
==================
Name    Mtu Network       Address              Ipkts Ierrs    Opkts Oerrs  
Coll
dc0    1500 <Link#1>      00:04:5a:79:36:96   390765     0  1392718     
0     0

Now, it's better.
==================

On the other machine that has LinkSys v2.0:
==================
Name    Mtu Network       Address              Ipkts Ierrs    Opkts Oerrs  
Coll
dc0    1500 <Link#1>      00:a0:cc:34:c2:a6  1085003     0   181102     
1     0

I think, an error might be from LOR (very easy to reproduce and your patch 
doesn't cause this, because before/after patch I get LOR).. I am planning 
to search in the mailing to make sure if there's no same LOR then I will 
report. It's transfer around over 1gig by NFS.
==================

Linksys v2.0 (LNE100):
==================
dc0_at_pci0:11:0:  class=0x020000 card=0xc00111ad chip=0xc11511ad rev=0x25 
hdr=0x00    vendor   = 'Lite-On Communications Inc'
     device   = 'LC82C115 PNIC II 10/100 PCI MAC/PHY'
     class    = network
     subclass = ethernet

dc0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
         inet 192.168.1.9 netmask 0xffffff00 broadcast 192.168.1.255
         inet6 fe80::2a0:ccff:fe34:c2a6%dc0 prefixlen 64 scopeid 0x1
         inet 192.168.1.10 netmask 0xffffffff broadcast 192.168.1.10
         inet 192.168.1.11 netmask 0xffffffff broadcast 192.168.1.11
         ether 00:a0:cc:34:c2:a6
         media: Ethernet autoselect (100baseTX <full-duplex>)
         status: active
==================

Linksys: v5.0 (LNE100):
==================
dc0_at_pci0:12:0:  class=0x020000 card=0x05741317 chip=0x09851317 rev=0x11 
hdr=0x00    vendor   = 'ADMtek Inc'
     device   = 'AM983 Comet Fast Ethernet Adapter'
     class    = network
     subclass = ethernet

dc0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
         inet 192.168.1.7 netmask 0xffffff00 broadcast 192.168.1.255
         inet6 fe80::204:5aff:fe79:3696%dc0 prefixlen 64 scopeid 0x1
         ether 00:04:5a:79:36:96
         media: Ethernet autoselect (100baseTX <full-duplex>)
         status: active
==================

Cheers,
Mezz


-- 
bsdforums.org 's moderator, mezz.