Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 04 Mar 2004 22:44:40 +1000
From:      Stephen McKay <smckay@internode.on.net>
To:        freebsd-hardware@freebsd.org
Cc:        Stephen McKay <smckay@internode.on.net>
Subject:   D-Link DGE-530T NIC
Message-ID:  <200403041244.i24CieBN013941@dungeon.home>

next in thread | raw e-mail | index | archive | help
Hi!

I've hacked the sk driver to support D-Link's DGE-530T gigabit NIC.  Nothing
fancy; basically I cloned the 3Com 3C940 case and put in the right PCI IDs
for the D-Link.  After checking the linux driver, that looks like it was the
correct approach.

The problem I now have is that my DGE-530T appears to have a completely blank
VPD (Vital Product Data) EEPROM.  Every byte reads as 0xff.  Does anybody
know if this is normal?  This causes the driver to output:

skc0: bad VPD resource id: expected 82 got ff
skc0: (null)

The driver works just fine after this, and the VPD is only used to print
the product name (which is stored there along with other stuff).  The
3Com 3C940 which is onboard my Asus P4P800 motherboard has a valid VPD
and the product name is printed for it, so at least some clones do it
properly.

If a blank VPD is a somewhat normal case, I'll put in code to ignore
it if it reads as 0xff.  I've already put in code to not print "(null)"
when reading the VPD fails.

After I've got this sorted, I can commit the changes.  Well, after I've
rebuilt my -current box, which had an undignified death recently (so this
was all done on 4.9).

By the way, it's too early to tell if this card is the bottleneck or
not when I did a quick performance test.  A P4 2.8GHz with 3C940 sending
to an Athlon 1.2GHz with DGE-530T got about 60MB/s using ttcp.  Only
20MB/s when disks were involved (using ftp).

Stephen.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200403041244.i24CieBN013941>