Skip to contentSkip to conversation

2.6.0-test5 vs. Ethernet cards

DISCUSSION Norman Diamond 3 participants 7 messages Sep 14, 2003 – Sep 17, 2003

lore
Share & tools
lore
#0 lore raw

When trying to do a shutdown or reboot while an Ethernet card is present, 2.6.0-test5 continues to have problems which earlier 2.6.0-test versions had, which can be solved by booting 2.4.19.

Shutdown messages appear on the text console as follows:
[...]
Shutting down PCMCIA unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1 [...]

The only way to shut down at this point is to turn off the power.

An alternative workaround is to manually type "ifconfig eth0 down" before trying to shut down or reboot. This takes a certain amount of activation of human memory at the moment just before shutting down, and I have not been up to the task. Since this step was not necessary under 2.4.19, it is easier to boot 2.4.19 instead.

Command "/etc/rc.d/pcmcia restart" suffers the same as a shutdown or reboot. Again similarly, in 2.4.19 it works.

#1 lore raw
Quoted 14 lines · Norman Diamond

Shutdown messages appear on the text console as follows:
[...]
Shutting down PCMCIA unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1 [...]

Quoted 1 line

The only way to shut down at this point is to turn off the power.

IIRC the problem is your hotplug scripts. Maybe the hotplug folk can tell you the minimum version for 2.6.

Signature · 5 lines
Russell King ([email protected])	http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
  2.6 ARM Linux   - http://www.arm.linux.org.uk/
  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
  2.6 Serial core
#2 lore raw

"Russell King" <[email protected]> replied to me:

Quoted 7 lines
Quoted 4 lines

Shutting down PCMCIA unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
[...]
The only way to shut down at this point is to turn off the power.

IIRC the problem is your hotplug scripts. Maybe the hotplug folk can tell you the minimum version for 2.6.

Then I wonder why the interface came up automatically when the card was inserted. By the way the relevant module is pcnet_cs, which is 16-bit PCMCIA. I didn't guess that the hotplug scripts were used for that. But I'll try to find time to test it with new hotplug scripts some weekend.

(Can't do it now, I'm in the middle of installing SuSE 8.2 on that machine.)

#3 lore raw
Quoted 10 lines · Norman Diamond

"Russell King" <[email protected]> replied to me:

Quoted 7 lines
Quoted 4 lines

Shutting down PCMCIA unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
[...]
The only way to shut down at this point is to turn off the power.

IIRC the problem is your hotplug scripts. Maybe the hotplug folk can tell you the minimum version for 2.6.

Quoted 6 lines

Then I wonder why the interface came up automatically when the card was inserted. By the way the relevant module is pcnet_cs, which is 16-bit PCMCIA. I didn't guess that the hotplug scripts were used for that. But I'll try to find time to test it with new hotplug scripts some weekend.

(Can't do it now, I'm in the middle of installing SuSE 8.2 on that machine.)

I'll "remind" the linux-net mailing list about this outstanding problem. It would be nice if some of the net people looked into this bug which, according to my mailbox, has been hanging around since 16th June...

There comes a point when one guy has to say "I've dealt enough with this problem" and that time is now. linux-net - please investigate this weird usage count bug.

Thanks.

Signature · 5 lines
Russell King ([email protected])	http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
  2.6 ARM Linux   - http://www.arm.linux.org.uk/
  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
  2.6 Serial core
#4 lore raw
In reply to Russell King
Quoted 15 lines · Russell King
Quoted 14 lines · Norman Diamond

Shutdown messages appear on the text console as follows:
[...]
Shutting down PCMCIA unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1 [...]

Quoted 4 lines
Quoted 1 line

The only way to shut down at this point is to turn off the power.

IIRC the problem is your hotplug scripts. Maybe the hotplug folk can tell you the minimum version for 2.6.

The last release version is the best for 2.6, but this doesn't look like a hotplug script issue at all.

thanks,

greg k-h

#5 lore raw
Quoted 16 lines · Greg KH
Quoted 15 lines · Russell King
Quoted 14 lines · Norman Diamond

Shutdown messages appear on the text console as follows:
[...]
Shutting down PCMCIA unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1 [...]

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

The only way to shut down at this point is to turn off the power.

IIRC the problem is your hotplug scripts. Maybe the hotplug folk can tell you the minimum version for 2.6.

The last release version is the best for 2.6, but this doesn't look like a hotplug script issue at all.

Hmm, ok. However, in the past when people have upgraded their hotplug scripts, the problem goes away.

Whatever, it's certainly not a PCMCIA issue either, so I'm at a loss what to do about these reports.

I can only think that the right answer is to bat them all at the netdev list, since it is a network device issue.

Signature · 5 lines
Russell King ([email protected])	http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
  2.6 ARM Linux   - http://www.arm.linux.org.uk/
  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
  2.6 Serial core
#6 lore raw
Quoted 12 lines · Russell King
Quoted 11 lines · Greg KH
Quoted 10 lines · Russell King
Quoted 8 lines · Norman Diamond

Shutdown messages appear on the text console as follows:
[...]
Shutting down PCMCIA unregister_netdevice: waiting for eth0 to become free. Usage count = 1
unregister_netdevice: waiting for eth0 to become free. Usage count = 1
[...]
The only way to shut down at this point is to turn off the power.

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

IIRC the problem is your hotplug scripts. Maybe the hotplug folk can tell you the minimum version for 2.6.

The last release version is the best for 2.6, but this doesn't look like a hotplug script issue at all.

Hmm, ok. However, in the past when people have upgraded their hotplug scripts, the problem goes away.

It will be a while before I have time to try it, sorry. SuSE 8.2 is partially installed on that machine now and one day next weekend will probably finish just that.

Quoted 4 lines

Whatever, it's certainly not a PCMCIA issue either, so I'm at a loss what to do about these reports.
I can only think that the right answer is to bat them all at the netdev list, since it is a network device issue.

Stephen Hemminger sent private e-mail guessing that IPV6 bugs might be the cause. I did select a number of IPV6 options when building 2.6.0-test[1..5]. Sorry I don't remember if my 2.4.19 included any IPV6
support, and I'm not planning to try a regression test to that one. I'll
try to find time some weekend to experiment with 2.6.0-test5 and 2.4.20-SuSE8.2 with and without IPV6.

I don't have access to IPV6 at present but figure I will someday, and if I ever have time to experiment with it then the experimentation time will not be wasted. Most of the world (except for North America) will not be able to survive without IPV6.