At home here I have a notebook and desktop PC and I can connect between them via a hub and everything works but when I try a crossover cable instead I get one way transmission, from the desktop to the notebook.
Thinking it could be a cable fault I connected the cable the other way found (so the pairs in use were the opposite way round) but still the notebook could see packets coming from the desktop and not the other way round.
Does anyone have any tips?
The desktop is running kernel 2.4.29 and the following is from 'dmesg' for when the crossover cable is in use:
natsemi.c:v1.07 1/9/2001 Written by Donald Becker becker@scyld.com http://www.scyld.com/network/natsemi.html (unofficial 2.4.x kernel port, version 1.07+LK1.0.14, Nov 27, 2001 Jeff Garzik, Tjeerd Mulder) AMD756: dev 100b:0020, router pirq : 2 get irq : 11 PCI: Found IRQ 11 for device 00:0a.0 eth0: NatSemi DP8381[56] at 0xf0929000, 00:a0:cc:7b:33:e6, IRQ 11. eth0: link up. eth0: Setting full-duplex based on negotiated link capability. eth0: Autonegotiation advertising 0x5e1 partner 0x00.
Reverting back to the hub, I get, again from the desktop:
eth0: link up. eth0: Link wake-up event 0x40020b eth0: Autonegotiation advertising 0x5e1 partner 0x21. eth0: Setting half-duplex based on negotiated link capability.
On Thu, Sep 12, 2002 at 11:53:38PM +0100, Steve Fosdick wrote:
At home here I have a notebook and desktop PC and I can connect between them via a hub and everything works but when I try a crossover cable instead I get one way transmission, from the desktop to the notebook.
Thinking it could be a cable fault I connected the cable the other way found (so the pairs in use were the opposite way round) but still the notebook could see packets coming from the desktop and not the other way round.
Does anyone have any tips?
yeah crossover cables are not actually supposed to work, they do by virtue of having the correct wires in the correct place but they are not supposed to work. I would give you the exact reasons if i had my Ethernet book at home but its at work where i have been "fixing" ethernet things there.
also did you try tcpdump on both ends to just to make sure things couldn't be seen? and are you sure that one of the cards isn't autonegotiating 10 base and the other 100 when you are using the crossover cable?
HTH Adam
Adam,
Thanks for such a quick response.
On Fri, 13 Sep 2002 00:04:50 +0100 Adam Bower abower@thebowery.co.uk wrote:
On Thu, Sep 12, 2002 at 11:53:38PM +0100, Steve Fosdick wrote:
At home here I have a notebook and desktop PC and I can connect between them via a hub and everything works but when I try a crossover cable instead I get one way transmission, from the desktop to the notebook.
Thinking it could be a cable fault I connected the cable the other way found (so the pairs in use were the opposite way round) but still the notebook could see packets coming from the desktop and not the other way round.
Does anyone have any tips?
yeah crossover cables are not actually supposed to work, they do by virtue of having the correct wires in the correct place but they are not supposed to work. I would give you the exact reasons if i had my Ethernet book at home but its at work where i have been "fixing" ethernet things there.
Interesting. I'd be interested to haer why.
also did you try tcpdump on both ends to just to make sure things couldn't be seen?
No. They way I diagnosed one way transmission was just to look at the packet counters reported at each end by 'ifconfig' assuming that these work at ethernet level rather than IP level.
When I pinged the notebook from the desktop the TX packet counter went up on the desktop and the RX packet counter went up on the notebook. When I pinged the desktop from the notebook the TX packet counter went up on the notebook but the RX counter on the desktop stayed at zero.
and are you sure that one of the cards isn't autonegotiating 10 base and the other 100 when you are using the crossover cable?
That's a possibility I suppose. Perhaps one of those companion diag programs that sometimes come with the network drivers could decode the status value so I can see if this is the case. If it is then I guess the answer may be to lock down one or both ends rather than rely on auto negotiation.
Steve.
On Fri, 13 Sep 2002 00:25:52 +0100 Steve Fosdick fozzy@pelvoux.demon.co.uk wrote:
On Fri, 13 Sep 2002 00:04:50 +0100 Adam Bower abower@thebowery.co.uk wrote:
also did you try tcpdump on both ends to just to make sure things couldn't be seen?
No. They way I diagnosed one way transmission was just to look at the packet counters reported at each end by 'ifconfig' assuming that these work at ethernet level rather than IP level.
When I pinged the notebook from the desktop the TX packet counter went up on the desktop and the RX packet counter went up on the notebook. When I pinged the desktop from the notebook the TX packet counter went up on the notebook but the RX counter on the desktop stayed at zero.
I have also now tried tcpdump as a cross check. It shows arp requests being sent from the desktop to the notebook but nothing coming back.
and are you sure that one of the cards isn't autonegotiating 10 base and the other 100 when you are using the crossover cable?
That's a possibility I suppose. Perhaps one of those companion diag programs that sometimes come with the network drivers could decode the status value so I can see if this is the case. If it is then I guess the answer may be to lock down one or both ends rather than rely on auto negotiation.
I have looked at the desktop end with mii-diag and it reports 100-BaseTx-FD. I have also tried 100-Base-TX-HD and both 10-BaseT options from the desktop end and none of them makes a difference. I'll look at the notebook end tomorrow.
Steve. z
-
Adam Bower -
Steve Fosdick