[c-nsp] connection speed at physical layer

Martin T m4rtntns at gmail.com
Thu Jul 21 15:48:05 EDT 2011


Hi John,
I determine it with:

ethtool -s eth0 advertise 0x02

For example "0x04" would be 100BASE-TX/Half or "0x20" would be
1000BASE-T/Full. "ethtool eth0" gives:

root at martin-ThinkPad-T60:/home/martin# ethtool eth0
Settings for eth0:
	Supported ports: [ TP ]
	Supported link modes:   10baseT/Half 10baseT/Full
	                        100baseT/Half 100baseT/Full
	                        1000baseT/Full
	Supports auto-negotiation: Yes
	Advertised link modes:  10baseT/Full
	Advertised pause frame use: No
	Advertised auto-negotiation: Yes
	Speed: 10Mb/s
	Duplex: Full
	Port: Twisted Pair
	PHYAD: 1
	Transceiver: internal
	Auto-negotiation: on
	MDI-X: off
	Supports Wake-on: pumbag
	Wake-on: g
	Current message level: 0x00000001 (1)
	Link detected: yes
root at martin-ThinkPad-T60:/home/martin#

As you can see, after executing "ethtool -s eth0 advertise 0x02", the
only link mode I advertise, is "10baseT/Full". It's same as for
example on Cisco 4500 platform many line cards with GI ports support
"speed auto 100" or "speed auto 10" which means that they advertise
themselves as 100BASE-TX or 10BASE-T ports.


The reason I asked this question at first place, was that I heard a
story from a friend of mine, that if you connect for example FE port
and GI port like this:

Cisco2950[Fa0/1] <-> [Gi0/11]Cisco2960

..using "speed auto" and "duplex auto" on both Fa0/1 and Gi0/11 port
the link will of course come up at 100BASE-TX/Full mode, but actually
GI interface will send electrical signals 10 times faster than FE port
sends towards GI port and this fills inbound buffers of FE port
faster. However, now it's sorted out, that it was a myth :)

regards,
martin


2011/7/21 John Gill <johgill at cisco.com>:
> Hi Martin,
> I am not sure how you determine you only advertise 10/full, but speed
> detection is not hard, but rather duplex is where it's more necessary.
>
> The interface on the switch is able to operate in 10 or 100 at the phy, in
> this case it knows it can understand the 10Mb/s signaling coming from the
> T60, so it steps down and that is the only way link is established.  I have
> not heard of two phy's operating at different speeds on the same link.
>
> John Gill
> cisco
>
>
> On 7/21/11 7:42 AM, Martin T wrote:
>>
>> Martin,
>> I'm running GNU/Linux(Debian). I set this manually for testing
>> purposes. In other words eth0 interface is configured to allow
>> autonegotiation, but advertises only 10BASE-T/Full.
>>
>> So it's clear, that if GE interface is configured to 10BASE-T mode, it
>> will send traffic on physical layer at max 10Mbps. Or for example if
>> FE interface is set to 10BASE-T mode, it will not send electrical
>> signals at rate of 100Mbps, but instead 10 times slower?
>>
>>
>> regards,
>> martin
>>
>>
>> 2011/7/21 Martin Barry<marty at supine.com>:
>>>
>>> $quoted_author = "Martin T" ;
>>>>
>>>> This might be a bit stupid question, but I have following topology:
>>>>
>>>> IBM_T60[eth0]<->  [Fa0/2]WS-C2950C-24
>>>>
>>>> eth0 port is an "Intel Corporation 82573L" GE port and Fa0/2 port in
>>>> WS-C2950C-24 switch is of course FE port. I have enabled
>>>> autonegotiation on eth0 port, but it advertises only 10BASE-T/Full
>>>> mode, so switch port, which is configured to "speed auto" and "duplex
>>>> auto", is up with following settings:
>>>>
>>>> Fa0/2     ->  T60             connected    1          a-full   a-10
>>>> 10/100BaseTX
>>>
>>> What operating system are you running on the T60? You need to figure out
>>> why
>>> it's not offering the full range of speeds during auto-negotiation.
>>>
>>>
>>>> Am I correct, that actual signalling on physical layer is 10BASE-T? Or
>>>> is eth0 port actually sending at 1Gbps(1000BASE-T signalling) rate and
>>>> Fa0/2(100BASE-TX signalling) is sending at 100Mbps rate on physical
>>>> layer?
>>>
>>> Your first instinct is correct, both sides with be operating at 10/full
>>> rather than the 100/full which you would obviously prefer.
>>>
>>> cheers
>>> Marty
>>> _______________________________________________
>>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>>>
>>
>> _______________________________________________
>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>>
>



More information about the cisco-nsp mailing list