Next-in-Thread Next Message

Question DLink DFE 660 - Bad performance 

Forum: PCMCIA Network Adapter Issues
Date: 2000, Aug 02
From: hvx

Hi!

Im using a DLink DFE-660 card with is an 100MBit card, but I get only about 200k/s. I tried 2 cards, a new and an old one, neither one of them was faster than that. Apart from that, the card seems to work just fine.

Here is some output from differnt commands, maybe somebody can give me a little advise how to get more speed out of this card. Thank you, M.Melms

Im using linux 2.2.16 with cardmgr 3.1.19.

bobby:~ # ifconfig eth0
eth0      Link encap:Ethernet  HWaddr 00:50:BA:74:55:50  
          inet addr:192.168.100.80  Bcast:192.168.100.255  Mask:255.255.255.0
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:1362 errors:0 dropped:0 overruns:0 frame:0
          TX packets:2281 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:100 
          Interrupt:10 Base address:0x200 

When I put in the card, /var/log/messages sais:

Aug  2 21:53:35 bobby kernel: cs: cb_alloc(bus 35): vendor 0x1011, device 0x0019
Aug  2 21:53:35 bobby cardmgr[78]: initializing socket 1
Aug  2 21:53:35 bobby cardmgr[78]: socket 1: Linksys EtherFast 10/100
Aug  2 21:53:36 bobby cardmgr[78]: executing: 'modprobe cb_enabler'
Aug  2 21:53:36 bobby cardmgr[78]: executing: 'modprobe tulip_cb'
Aug  2 21:53:36 bobby kernel: cs: cb_config(bus 35)
Aug  2 21:53:36 bobby kernel:   fn 0 bar 1: io 0x200-0x27f
Aug  2 21:53:36 bobby kernel:   fn 0 bar 2: mem 0x600c0000-0x600c03ff
Aug  2 21:53:36 bobby kernel:   fn 0 rom: mem 0x60080000-0x600bffff
Aug  2 21:53:36 bobby kernel:   irq 10
Aug  2 21:53:36 bobby kernel: cs: cb_enable(bus 35)
Aug  2 21:53:36 bobby kernel:   bridge io map 0 (flags 0x21): 0x200-0x27f
Aug  2 21:53:36 bobby kernel:   bridge mem map 0 (flags 0x1): 0x60080000-0x600c0
fff
Aug  2 21:53:36 bobby kernel: tulip_attach(device 23:00.0)
Aug  2 21:53:36 bobby kernel: tulip.c:v0.91g-ppc 7/16/99 becker@cesdis.gsfc.nasa
.gov (modified by danilo@cs.uni-magdeburg.de for XIRCOM CBE, fixed by Doug Ledfo rd) Aug 2 21:53:36 bobby kernel: eth0: Digital DS21143 Tulip rev 65 at 0x200, 00:50 :BA:74:55:50, IRQ 10.
Aug  2 21:53:36 bobby kernel: eth0:  EEPROM default media type Autosense.
Aug  2 21:53:36 bobby kernel: eth0:  Index #0 - Media MII (#11) described by a 2
1142 MII PHY (3) block. Aug 2 21:53:36 bobby kernel: eth0: MII transceiver #0 config 3000 status 7809 advertising 01e1. Aug 2 21:53:36 bobby cardmgr[78]: executing: './network start eth0'

Next-in-Thread Next Message

Messages Inline: 0 1

None First guess: problem negotiating correct media type

Re: Question DLink DFE 660 - Bad performance
Date: 2000, Aug 02
From: David Hinds <dhinds@pcmcia.sourceforge.org>

Maybe the card is confused about whether this is a full or half duplex
link.  What sort of device are you connecting to?  Hub or switch?  Do
you know if it supports autonegotiation?

Try adding to /etc/pcmcia/config.opts a line like

   module "tulip_cb" opts "options=N"

where N is a number matching your media type, taken from
http://www.scyld.com/network/tulip.html.

-- Dave

Feedback thanks, but it didnt work

Re: None First guess: problem negotiating correct media type (David Hinds)
Date: 2000, Aug 02
From: hvx

> Maybe the card is confused about whether this is a full or half duplex
> link.  

Thats what I though, too, since the LED saying Half/Full Duplex always goes on and off when there is traffic.

> What sort of device are you connecting to? Hub or switch?

Its a micronet switch.

> Do
> you know if it supports autonegotiation?

I have no documentation, so I dont - any way to figure it out ?

> Try adding to /etc/pcmcia/config.opts a line like
> 
>      module "tulip_cb" opts "options=N"

If I do so, the card wont work at all. Even if I put "11" in there, which is the mode the card is usually in, it wont work. The logs dont say much besides

eth0: Using user-specified media MII

and I got a

kernel: eth0: Tx hung, 29 vs. 11.

a few times. Anything else I can try ? thanks, M.Melms

None Re: Feedback: thanks, but it didnt work

Re: Feedback thanks, but it didnt work
Date: 2000, Aug 02
From: David Hinds <dhinds@pcmcia.sourceforge.org>

> kernel: eth0: Tx hung, 29 vs. 11.

Do you ever get this message in normal operation (with no module
options)?  Can you show more of your PCMCIA startup messages (the PCIC
probe, etc)?

Is the 200 kb/sec in both directions?  Does "ifconfig" show lots of
errors of any sort?

Are you using a good (i.e., Category 5) 100baseT cable?

-- Dave

Feedback Re: Feedback: thanks, but it didnt work

Re: None Re: Feedback: thanks, but it didnt work (David Hinds)
Date: 2000, Aug 02
From: hvx

> Do you ever get this message in normal operation (with no 
> module
> options)? 

no, never.

>  Can you show more of your PCMCIA startup messages (the PCIC
> probe, etc)?

like when I start cardmgr ? This would be:

Aug  2 23:19:51 bobby cardmgr[1741]: starting, version is 3.1.19
Aug  2 23:19:51 bobby cardmgr[1741]: watching 2 sockets
Aug  2 23:19:51 bobby kernel: cs: IO port probe 0x1000-0x17ff: clean.
Aug  2 23:19:51 bobby kernel: cs: IO port probe 0x0100-0x04ff: clean.
Aug  2 23:19:51 bobby kernel: cs: IO port probe 0x0a00-0x0aff: clean.

> Is the 200 kb/sec in both directions?

no, its worse ;-) I logged into our server via FTP. When I put a file I get 200K/s When I get a file I get 5 k/s

I noticed that the longer the download takes, the slower the LEDs blink - speed goes down.

>  Does "ifconfig" show lots of
> errors of any sort?
>

 TX packets:888 errors:14 dropped:0 overruns:0 carrier:14
                  collisions:0 txqueuelen:100 

Thats all, after a few MBs traffic.

> Are you using a good (i.e., Category 5) 100baseT cable?

Very good ones. Thanks, M.Melms

DLink DFE 660 - Bad performance


Members Subscribe No Admin Mode Show Frames Help for HyperNews at pcmcia-cs.sourceforge.net 1.10
[ Edit This Forum ]