dc0 problems (Was: Re: Heads up: NEWCARD is coming)

[Tomaž Borštnar <tomaz.borstnar@xxxxxxxx>]

der.dragonflybsd.org> <20040217112047.GA1007@xxxxxxxxxxxxxxxxx>
In-Reply-To: <20040217112047.GA1007@xxxxxxxxxxxxxxxxx>
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer-Encoding: 8bit
Lines: 112
Message-ID: <40322b86$0$185$415eb37d@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>
NNTP-Posting-Host: 217.72.82.13
X-Trace: 1077029767 crater_reader.dragonflybsd.org 185 217.72.82.13
Xref: crater_reader.dragonflybsd.org dragonfly.kernel:3853

Joerg Sonnenberger pravi:

>>>>Ifconfig showed the xl device with no media. But simple ifconfig xl0 up 
>>>>solved the problem :) And it correctly discovered my 100Mbps switch :)
>>
>>If the card is plugged in while booting it is correctly activated - up and 
>>configured for media.
> 
> 
> Otherwise not?
It is configured properly. It is just not up (flag UP in ifconfig output) without assigning some IP address or manually forcing it 
to up.


>>But upon invocation (inserting the card or being present while booting) it 
>>locks my sistem so hard that I can not even get into DDB. I have no problem 
>>manually jumping into DDB on this machine otherwise. It does not affect 
>>stability of machine if dc card is not present. Last line before freezing is
>>dc0: <Intel 21143 10/100BaseTX> port 0x1000-0x107f mem 
>>0x88002000-0x880023ff irq 10 at device 0.0 on cardbus0
> 
> 
> It seems to fail while reading the MAC address. Can you give the output
> with hw.cardbus.cis_debug=1 while attaching the card?

Interesting. I restarted the machine which came up fine. Inserted the card and things looked OK:
(having logs because of SSH login via xl card!)

Feb 17 14:38:15  /kernel: Product version: 5.0
Feb 17 14:38:15  /kernel: Product name: Silicom Ltd. | CardBus Ethernet | CBF100
  | 1.00981020 |
Feb 17 14:38:15  /kernel: Manufacturer ID: da018100
Feb 17 14:38:15  /kernel: Functions: Network Adaptor, Multi-Functioned
Feb 17 14:38:15  /kernel: Function Extension: 0102
Feb 17 14:38:15  /kernel: Function Extension: 0280969800
Feb 17 14:38:15  /kernel: Function Extension: 0200e1f505
Feb 17 14:38:15  /kernel: Function Extension: 0301
Feb 17 14:38:15  /kernel: Function Extension: 0501
Feb 17 14:38:15  /kernel: cardbus1: Opening BAR: type=IO, bar=10, len=0080
Feb 17 14:38:15  /kernel: cardbus1: Opening BAR: type=MEM, bar=14, len=0400
Feb 17 14:38:15  /kernel: CIS reading done
Feb 17 14:38:15  /kernel: cardbus1: Non-prefetchable memory at 88002400-880027ff
Feb 17 14:38:15  /kernel: cardbus1: Non-prefetchable memory rid=14 at 88002400-8
80027ff (400)
Feb 17 14:38:15  /kernel: cardbus1: IO port at 1080-10ff
Feb 17 14:38:15  /kernel: cardbus1: IO port rid=10 at 1080-10ff
Feb 17 14:38:15  /kernel: dc0: <Intel 21143 10/100BaseTX> port 0x1080-0x10ff mem
  0x88002400-0x880027ff irq 10 at device 0.0 on cardbus1
Feb 17 14:38:15  /kernel: dc0: Ethernet address: 00:e0:ed:01:bc:03
Feb 17 14:38:15  /kernel: miibus1: <MII bus> on dc0
Feb 17 14:38:15  /kernel: acphy0: <AC101 10/100 media interface> on miibus1
Feb 17 14:38:15  /kernel: acphy0:  10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
# ifconfig dc0
dc0: flags=8802<BROADCAST,SIMPLEX,MULTICAST> mtu 1500
         ether 00:e0:ed:01:bc:03
         media: Ethernet autoselect (none)
         status: no carrier


Until I plugged in utp cable (still working!) and tried using dhclient at which point it locked solidly.

Ok, after reboot...

# ifconfig dc0 inet 192.168.168.168 netmask 255.255.255.0
#
#
# ifconfig dc0
dc0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
         inet 192.168.168.168 netmask 0xffffff00 broadcast 192.168.168.255
         inet6 fe80::2e0:edff:fe01:bc03%dc0 prefixlen 64 scopeid 0x7
         ether 00:e0:ed:01:bc:03
         media: Ethernet autoselect (none)
         status: no carrier

Ok, machine still working... Now try to fix settings for the cable...
And it is up! Replacing IP with LAN IP and it works too. Weird, weird, weird. ifconfig dc0 delete to get rid of IP settings and 
using dhclient dc0. Works too! Finally I physically took out the card and put it back...
machine was so locked that even DDB could not help!

Machine ALWAYS locks up if the card is present while booting.
Machine ALWAYS locks up if I insert the card after the booting is done and issue dhclient dc0 right after it is recognized by the 
system. But seems like auto-detection is the culprit... I rebooted machine, inserted card and was happily recognized. I waited few 
moments before connecting it to the switch and machine was responding all the time until the card recogized the switch - which it 
did successfully:

rs# port show port-status et.2.18
Flags: M - Mirroring enabled  S - SmartTRUNK port
                                            Negot- Link      Admin
Port    Port Type        Duplex  Speed     iation State     State   Flags
et.2.18 100Mbit Ethernet  Full   100Mbits  Auto   Up         Up

If I unplug the cable:
(messages from switch now)
17-Feb-2004 16:48:02 WARNING Port     et.2.18 Down

and put it back...
17-Feb-2004 16:48:18 INFO    Port     et.2.18 Up

rs# port show port-status et.2.18
Flags: M - Mirroring enabled  S - SmartTRUNK port
                                            Negot- Link      Admin
Port    Port Type        Duplex  Speed     iation State     State   Flags
et.2.18 100Mbit Ethernet  Full   100Mbits  Auto   Up         Up

So something on the machine is still alive, but my console is frozen.

Maybe xl device helped dc not to freeze as things were looking much better while xl was also present in the system?

Toma?