MII register mis-read in devn-tulip.so?

I have a general quandary concerning the devn-tulip.so device driver under
RTP (6.1.0). I have a laptop that this environment is currently running
under and the pccard facilities are functioning correctly. When I insert
two different CardBus FastE devices into the system, they are not
auto-enumerated, but do display (via pci and pin) as vid = 0x13d1 (Abocom)
and did = 0xab02 (FE2500 series). One is the SMC SMC8035TX (= to Abocom
FE2500MX, dongle-less), and the other is the LinkSys PCMPC200 (= to Abocom
FE2500 w/ dongle).

The tulip driver will recognize them as “chipset=2500”, but ALWAYS
autodetects media = “MII No Link”. Near as I can tell, because these
devices use an MII PHY after the tulip chipset, it is not correctly
determining link status/media from the MII registers? It bugs me that the
tulip driver does recognize the chipsets but does not seem to pick-up on the
media status from the MII. In this scenario, my media value should be 5
(100bTFD), but it will not take. The command I issue is as follows:

io-net -dtulip verbose=1,media=5 -ptcpip

I have also tried variants as basic as:

io-net -dtulip verbose=1

All attempts end in the same MII No Link message. The end result is I can
bind an IP stack to the en0 device, I can ping it, etc (ie, it instantiates
as a normal device), but it simply thinks there is no PHYS link active.
This is a tad bit distressing to say the least. I noticed that Janet
Geddes’ post from 1/23/01 contained some of the same symptoms as is I have
observed here. Any ideas (Chris McKillop?) or am I re-writing the tulip
driver?

BTW, the verbose option is purely for diagnostic purposes, and I have cycled
through media values 0 - 10, only 5 (100bTFD) and 9 (MII) were accepted
values given my Full Duplex FastE network I am running.

Thanks,

Alex

In article <9j4194$mms$2@inn.qnx.com>, asenkevitch@hotmail.com says…

I have a general quandary concerning the devn-tulip.so device driver under
RTP (6.1.0). I have a laptop that this environment is currently running
under and the pccard facilities are functioning correctly. When I insert
two different CardBus FastE devices into the system, they are not
auto-enumerated, but do display (via pci and pin) as vid = 0x13d1 (Abocom)
and did = 0xab02 (FE2500 series). One is the SMC SMC8035TX (= to Abocom
FE2500MX, dongle-less), and the other is the LinkSys PCMPC200 (= to Abocom
FE2500 w/ dongle).

The tulip driver will recognize them as “chipset=2500”, but ALWAYS
autodetects media = “MII No Link”. Near as I can tell, because these
devices use an MII PHY after the tulip chipset, it is not correctly
determining link status/media from the MII registers? It bugs me that the
tulip driver does recognize the chipsets but does not seem to pick-up on the
media status from the MII. In this scenario, my media value should be 5
(100bTFD), but it will not take. The command I issue is as follows:

Is the cable plugged in at both ends?
Is it the correct type of cable? (Not crossover, probably)
Does the switch/hub at the other end support Auto-negotiation?
Is the cable plugged into a/the correct port on the switch/hub?

io-net -dtulip verbose=1,media=5 -ptcpip

I have also tried variants as basic as:

io-net -dtulip verbose=1

All attempts end in the same MII No Link message. The end result is I can
bind an IP stack to the en0 device, I can ping it, etc (ie, it instantiates
as a normal device), but it simply thinks there is no PHYS link active.
This is a tad bit distressing to say the least. I noticed that Janet
Geddes’ post from 1/23/01 contained some of the same symptoms as is I have
observed here. Any ideas (Chris McKillop?) or am I re-writing the tulip
driver?

BTW, the verbose option is purely for diagnostic purposes, and I have cycled
through media values 0 - 10, only 5 (100bTFD) and 9 (MII) were accepted
values given my Full Duplex FastE network I am running.

Thanks,

Alex






\


Stephen Munnings
Software Developer
Corman Technologies Inc.

No, the switch is a 10/100 auto-sensing switch that, at last look, was
functioning. I was showing link light on the switch, and on the SMC NIC,
the 100, FullD, and Link LED’s were lit. The only LED not indicating
anything (unlit) was the activity LED (unfortunately).

Alex

Stephen Munnings <steve@cormantech.com> wrote in message
news:MPG.15bf5aaa4b50395c98969a@inn.qnx.com

In article <9j4194$mms$> 2@inn.qnx.com> >, > asenkevitch@hotmail.com > says…
I have a general quandary concerning the devn-tulip.so device driver
under
RTP (6.1.0). I have a laptop that this environment is currently running
under and the pccard facilities are functioning correctly. When I insert
two different CardBus FastE devices into the system, they are not
auto-enumerated, but do display (via pci and pin) as vid = 0x13d1
(Abocom)
and did = 0xab02 (FE2500 series). One is the SMC SMC8035TX (= to Abocom
FE2500MX, dongle-less), and the other is the LinkSys PCMPC200 (= to
Abocom
FE2500 w/ dongle).

The tulip driver will recognize them as “chipset=2500”, but ALWAYS
autodetects media = “MII No Link”. Near as I can tell, because these
devices use an MII PHY after the tulip chipset, it is not correctly
determining link status/media from the MII registers? It bugs me that
the
tulip driver does recognize the chipsets but does not seem to pick-up on
the
media status from the MII. In this scenario, my media value should be 5
(100bTFD), but it will not take. The command I issue is as follows:


Is the cable plugged in at both ends?
Is it the correct type of cable? (Not crossover, probably)
Does the switch/hub at the other end support Auto-negotiation?
Is the cable plugged into a/the correct port on the switch/hub?

io-net -dtulip verbose=1,media=5 -ptcpip

I have also tried variants as basic as:

io-net -dtulip verbose=1

All attempts end in the same MII No Link message. The end result is I
can
bind an IP stack to the en0 device, I can ping it, etc (ie, it
instantiates
as a normal device), but it simply thinks there is no PHYS link active.
This is a tad bit distressing to say the least. I noticed that Janet
Geddes’ post from 1/23/01 contained some of the same symptoms as is I
have
observed here. Any ideas (Chris McKillop?) or am I re-writing the tulip
driver?

BTW, the verbose option is purely for diagnostic purposes, and I have
cycled
through media values 0 - 10, only 5 (100bTFD) and 9 (MII) were accepted
values given my Full Duplex FastE network I am running.

Thanks,

Alex









\

Stephen Munnings
Software Developer
Corman Technologies Inc.