On Tuesday 02 November 2004 17:40, Martijn van Oosterhout wrote: > > That's the big question. I thought about deregistering the IP address in > > the stack. A sort of a rollback of what inet_insert_ifa did. > > Print a warning and forget about it. People will notice when they're > not getting any traffic. This is what we currently do. I was just hoping that we could give some better feedback to the user. Consider the folling example: - you have an interface eth1 and assign address 10.10.10.1/16 to it - if you do an 'ip addr add 10.10.10.1/16 dev eth1' again, you get a -EEXIST On our virtual hardware the main reason for not being able to set IP addresses is also, if the IP address is already used by another OS image. So we also get an -EEXIST from the hardware. So I thought we could give this error directly to the user, and not just print a message. The bad thing is, that Linux's EEXIST has a totally different meaning than our hardware's EEXIST. So, I guess the end of the story really is that we can just print an error message, That's one of the problems we have to live with with our current virtualization hardware. But anyway, thanks for the discussion. It helped quite a bit to hear some other's opinions. -- Regards, Thomas. - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html