DragonFlyBSD

(I haven't had a chance to try Sephe's latest commit yet, I'll try
it out tonight after I get the next batch of HAMMER work committed).

-Matt

:Hi,
:
:I'm running -HEAD as of today and nfe0 doesn't seem to work 100%.
:
:I can ping to another computer on the same network (directly connected
:via a 100 MBit/s switch), but I can't ping the DragonFly box from the
:other machine. Also, dhclient doesn't work on the DragonFly box!
:
:I've also tried setting hw.nfe0.imtimer=-125 but that doesn't help
:either.
:
:Any hints?
:
:Regards,
:
: Michael

Try backing out a rev or two. Well, try backing out one rev of
    if_nfe.c (back down to 1.30) and see if that works ok.

-Matt
                    Matthew Dillon 
                    <dillon@backplane.com>

Nope. Same problem occurs. Btw, is it required to rebuilt the whole
kernel, or is it enough to just "make && make install" in an already
compiled "sys/compile/KERNCONFIG" directory?

Regards,

Michael

nfe works here.

thomas

$ pciconf -lv | grep -A4 '^nfe0'
nfe0@pci0:8:0:    class=0x068000 card=0x82391043 chip=0x037310de rev=0xa3 hdr=0x00
    vendor   = 'Nvidia Corp'
    device   = 'MCP55 Ethernet'
    class    = bridge
    subclass = PCI-unknown

Try
if_nfe.c rev 1.24
if_nfereg.h rev 1.6
if_nfevar.h rev 1.3

Give me the output of 'ifconfig nfe0' and the dmesg, if possible.

Best Regards,
sephe

No luck either! I can ping out, but not in, as before.

To rule out other possible problems, it works fine when booting from a
FreeBSD 7.0 cdrom.

Here is pciconf, ifconfig and dmesg output:

$ pciconf -lv | grep -A4 '^nfe0'
nfe0@pci0:10:0: class=0x020000 card=0x82b31043 chip=0x054c10de rev=0xa2
hdr=0x00
vendor = 'Nvidia Corp'
device = 'MCP67 Ethernet'
class = network
subclass = ethernet

$ ifconfig nfe0
nfe0: flags=8802<BROADCAST,SIMPLEX,MULTICAST> mtu 1500
options=8<VLAN_MTU>
ether 4f:ad:6d:c6:1f:00
media: Ethernet autoselect
status: active

$ dmesg | grep nfe0
nfe0: <NVIDIA MCP67 Gigabit Ethernet> port 0xc880-0xc887 mem
0xdfefe000-0xdfefe00f,0xdfefe400-0xdfefe4ff,0xdfefc000-0xdfefcfff irq 10
at device 10.0 on pci0
miibus1: <MII bus> on nfe0
nfe0: MAC address: 4f:ad:6d:c6:1f:00

Hehe, I see. I wanted to change that code for a long time. MCP61 and
above has different MAC address order. Linux also added some code to
adjust TX_POLL, but it does not affect MCP61 and above. Please test
the attached patch.

Best Regards,
sephe

Hehe, I know what you mean :)

After the patch:

ether 00:1f:c6:6d:ad:4f

And it works like a charm! Thank you very much!

Regards,

Michael

fix committed