[klibc] problems with ipconfig command (static config without interface configures "lo")

[H. Peter Anvin]

Dirk von Suchodoletz wrote:

> 
>>static unsigned int default_caps = CAP_DHCP | CAP_BOOTP | CAP_RARP;
>>
>>
>>this should be enough to get the interface up:
>>ip=<client-ip>:<server-ip>:<gw-ip>:<netmask>
>>
>>not 'ip=<client-ip>:<server-ip>:<gw-ip>:<netmask>:::none' like it is
>>now. Even 'ip=<client-ip>' should be enough, but parse_device() would
> 
> 
> That would made sense to use just 'ip=<client-ip>' because the ipconfig
> tries to make sense of the IP and set netmask and broadcast accordingly 
> ... (I saw that in earlier versions that for an IP 132.230.4.123 the tool 
> used class B netmask even provided with a class C via dhcp - bug fixed 
> versions ago ... But this mechanism makes sense for the problem stated 
> above ...)
> 

I don't think so.  In fact, a default netmask of 255.255.255.0 would 
make more sense, since pretty much ALL networks > class C are subnetted, 
and have been for a decade or more.

The classful Internet must die.  One of the things I liked about the 
Linux IP stack back in 1993 was that it was one of the first ones which 
was 100% classless already back then.

	-hpa