On Aug 13, 2009, at 4:34 PM, Joe Perches wrote:

On Thu, 2009-08-13 at 16:24 -0400, Brian Haley wrote:

quoted

Is your arch "um"?  Seems like those are only defined there, I'm  
building
a straight x86 kernel.

Nope.

I did make allyesconfig ; make lib/vsprintf.o lib ctype.o
allnoconfig works though.

quoted

This core dumps when running "test", I'm still trying to track down  
why.

missing return on ip_addr_string

quoted

I think we're thinking too hard about this, I would think we'd always
want to print the shortened IPv6 address in debugging messages with  
%pI6.

True, but you can't tell in sprintf as it's used in seq.

for instance:
net/sunrpc/svcauth_unix.c:		seq_printf(m, "%s %pI6 %s\n", im- 

quoted

m_class, &addr, dom);

This one might be a bad example.  RPC IPv6 support, especially server  
side, isn't written in stone yet.  User space may not even be ready  
for an IPv6 address here; I can check.  If user space happens to be  
flexible here, then it won't matter if this particular instance is  
shorthanded or not.

[ I would think user space in general should be using inet_pton(3)  
everywhere for such interfaces, so the format of these addresses  
wouldn't matter so much.  Probably impossible at this point. ]

quoted

The %pi6 places need to stay since they're an API to userspace.  I  
don't
think we need the extra "c" and "c4" support.

I'm pretty sure it can't change and a new form is needed
so %pi6c should be OK.

I'd rather not use another %p<foo> letter.

I'm not arguing one way or the other, but it would be useful if  
someone could check exactly what the dependencies are right now.  It  
seems like we're speculating a bit.

quoted

One comment on a quick scan of the code:
ip6_addr[8 * 5] is fine here, we won't ever have all eight plus an  
IPv4 address.

I'm fixing it up and will resubmit something working in a little  
while.

cheers, Joe

--
Chuck Lever
chuck[dot]lever[at]oracle[dot]com