tulip initialization?

hi,

I've got a NIC that uses the tulip driver but i have a strange problem.

With a 2.0.39 kernel in the same box, the box is booted, the tulip
module loaded and I am able to ping boxes etc.

However, with the same box with a 2.6.12 kernel, the box is booted, the
module loaded but I am not able to get out of the box. The routes etc
are correctly setup and the power/activity LEDs are lit at the back of
the NIC, although not the 100Mbit LED.

To 'enable' the box to ping other machines, I have to remove and plug
the network cable back into the box, at which point the 100MBit LED is
lit -- is there a change to the tulip driver requiring additional
params at module initialization time?

The network cable is fine as I did not have this problem with the same
h/w but with a 2.0.x kernel in the same box.

Any thoughts? TIA
-r

In <(E-Mail Removed) .com>, whatdoineed2do
wrote:

> hi,
>
> I've got a NIC that uses the tulip driver but i have a strange problem.
>
> With a 2.0.39 kernel in the same box, the box is booted, the tulip module
> loaded and I am able to ping boxes etc.
>
> However, with the same box with a 2.6.12 kernel, the box is booted, the
> module loaded but I am not able to get out of the box. The routes etc are
> correctly setup and the power/activity LEDs are lit at the back of the
> NIC, although not the 100Mbit LED.
>
> To 'enable' the box to ping other machines, I have to remove and plug the
> network cable back into the box, at which point the 100MBit LED is lit --
> is there a change to the tulip driver requiring additional params at
> module initialization time?
>
> The network cable is fine as I did not have this problem with the same h/w
> but with a 2.0.x kernel in the same box.
>
> Any thoughts? TIA
> -r

Does your card use a real DEC chipset (21x4x)? You might try using the
de4x5 driver instead. There have been problems with the tulip driver on
DEC chipsets for a long time (since the 2.2. days at least) and they
haven't really been fixed since de4x5 also works with those chipsets.

--
Not to have been a dupe, that will have been my best possesion, my best
deed, to have been a dupe, wishing I wasn't, thinking I wasn't, knowing
I was, not being a dupe of not being a dupe.
--Samuel Beckett, The Unnamable

i've just tried using the de4x5 driver but that caused other problems
(ifconfig reports TX had over 8000errors in the 20mins i had the link
up and the link was slow because of this) so that isnt viable for me --
but it did resolve the problem that i mentioned before.

the other thing i had noticed was that when the machine is booting to
the boatloader (ie before any OS is loaded) with the network cable
plugged in, the NIC leds are all lit including the 100Mbit LED (ie
showing that the NIC has sensed that it is connected to a 100Mbit lan).
However, as soon as the tulip driver is loaded the 100Mbit LED turns
off.

This to me suggests that the driver forces the card into the wrong
mode. whilst booting, i've seen the tulip report that "non MII
100baseTx-FDX" so i pressume is this is what the NIC is reporting.
with that said, do you know how i could force the NIC to operate in
this mode?

> with that said, do you know how i could force the NIC to operate in
> this mode?
Will miitool do this for you? Other possibilities: try the old tulip
driver, or change the hardware at the other end (add a hub or change hub
manufacturer)

On 15 Aug 2005, (E-Mail Removed) <(E-Mail Removed)> wrote:
> hi,
>
> I've got a NIC that uses the tulip driver but i have a strange problem.
>
> With a 2.0.39 kernel in the same box, the box is booted, the tulip
> module loaded and I am able to ping boxes etc.
>
> However, with the same box with a 2.6.12 kernel, the box is booted, the
> module loaded but I am not able to get out of the box. The routes etc
> are correctly setup and the power/activity LEDs are lit at the back of
> the NIC, although not the 100Mbit LED.
>
> To 'enable' the box to ping other machines, I have to remove and plug
> the network cable back into the box, at which point the 100MBit LED is
> lit -- is there a change to the tulip driver requiring additional
> params at module initialization time?
>
> The network cable is fine as I did not have this problem with the same
> h/w but with a 2.0.x kernel in the same box.

I had a similar problem with what I think was a Netgear 310TX. The older
kernel 2.2 tulip module worked fine. With later 2.2 kernels it broke
(would not work until pinged from network).

The tulip in kernel 2.4 appeared to work, but would die overnight (TX
errors unless I took the interface down and up).

My solution was to run Mandrake 7.0 on that box with most recent 2.2
kernel, but using its "old_tulip" module, until that PC died. The nic
lived on in an NT box at work until that became obsolete, and the hard
drive is in my current Linux Celeron 300 pppoe/firewall/router.

i found a soln. the driver would report:

tulip0: Transceiver selection forced to 100baseTx-FDX.
tulip0: EEPROM default media type Autosense.
tulip0: Index #0 - Media 10baseT (#0) described by a 21140 non-MII
(0) bloc
k.
tulip0: Index #1 - Media 100baseTx (#3) described by a 21140
non-MII (0) bl
ock.
tulip0: Index #2 - Media 10baseT-FDX (#4) described by a 21140
non-MII (0)
block.
tulip0: Index #3 - Media 100baseTx-FDX (#5) described by a 21140
non-MII (0
) block

and i knew that the NIC could do 100base full duplex so using the above
info (the '5' is the idx into the tulip driver's media selection
table), i loaded the driver as:

modprobe tulip options=5

and when the network came up, everything was fine. an update to
modprobe.conf, reboot and that solved the problem so it does look like
something was forcing the NIC into a mode it didnt quite like. for
completeness:

# /etc/modprobe.conf
alias tulip eth0
options tulip options=5