Re: [PATCH] nfs: fix dead code of ipv6_addr_scope

[Alexander Aring <alex.aring@gmail.com>]

Hi Hannes,

On Sun, Dec 22, 2013 at 03:30:54AM +0100, Hannes Frederic Sowa wrote:
> On Sat, Dec 21, 2013 at 02:32:54PM +0100, Alexander Aring wrote:
> > Hi Hannes,
> > 
> > On Sat, Dec 21, 2013 at 01:44:40PM +0100, Hannes Frederic Sowa wrote:
> > > On Sat, Dec 21, 2013 at 05:39:04AM +0100, Alexander Aring wrote:
> > > > The correct way to check on IPV6_ADDR_SCOPE_LINKLOCAL is to check with
> > > > the ipv6_addr_src_scope function.
> > > > 
> > > > Currently this can't be work, because ipv6_addr_scope returns a int with
> > > > a mask of IPV6_ADDR_SCOPE_MASK (0x00f0U) and IPV6_ADDR_SCOPE_LINKLOCAL
> > > > is 0x02. So the condition is always false.
> > > > 
> > > > Signed-off-by: Alexander Aring <alex.aring@gmail.com>
> > > > ---
> > > > I think ipv6_addr_src_scope should be correct, can somebody from netdev ml
> > > > confirm this please?
> > > > I stumple over that and I did not compile and test it. Maybe this is something
> > > > for stable?
> > > > 
> > > >  fs/nfs/nfs4filelayoutdev.c | 2 +-
> > > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > > 
> > > > diff --git a/fs/nfs/nfs4filelayoutdev.c b/fs/nfs/nfs4filelayoutdev.c
> > > > index c7c295e5..efac602 100644
> > > > --- a/fs/nfs/nfs4filelayoutdev.c
> > > > +++ b/fs/nfs/nfs4filelayoutdev.c
> > > > @@ -95,7 +95,7 @@ same_sockaddr(struct sockaddr *addr1, struct sockaddr *addr2)
> > > >  		b6 = (struct sockaddr_in6 *)addr2;
> > > >  
> > > >  		/* LINKLOCAL addresses must have matching scope_id */
> > > > -		if (ipv6_addr_scope(&a6->sin6_addr) ==
> > > > +		if (ipv6_addr_src_scope(&a6->sin6_addr) ==
> > > >  		    IPV6_ADDR_SCOPE_LINKLOCAL &&
> > > >  		    a6->sin6_scope_id != b6->sin6_scope_id)
> > > >  			return false;
> > > 
> > > Good catch!
> > > 
> > thanks.
> > 
> > I am still unsure if sctp is correct or not, I think it isn't correct.
> > Because we compare and don't check if any bit is set.
> > 
> > We don't use IPV6_ADDR_SCOPE_TYPE here. We use IPV6_ADDR_TYPE. But we can't
> > compare it.
> 
> Actually, this is fine, too. ipv6_addr_scope does mask the addr_type with
> IPV6_ADDR_SCOPE_MASK (which is 0x00f0U). If you look at addrconf_core.c you
> see that the 4 bits stand by itself each time.
> 
> Actually it seems ipv6_addr_src_scope is better suitable for multicast scope
> handling and ipv6_addr_scope with IFA_{HOST,LINK,SITE} is fine for
> non-multicast. In this case there is no difference.
> 
ah thanks, now I understand it!

so an alternative would be:

if (ipv6_addr_scope(&a6->sin6_addr) & IPV6_ADDR_LINKLOCAL &&
	a6->sin6_scope_id != b6->sin6_scope_id)
	...

maybe this is a little bit faster instead of ipv6_addr_src_scope.
Should I resend a v2 with the faster solution?

> Maybe an int ipv6_cmp_sockaddr(struct in6_addr *a1, int scope1,
> 			  struct in6_addr *a2, int scope2)
> or
> int ipv6_cmp_sockaddr(struct sockaddr_in6 *s1,
> 		      struct sockaddr_in6 *s2)
> 

I don't understand why we need such a function here. We only check if
"a6" is linklocal and has a different sin6_scope_id than "b6" sin6_scope_id
and we don't compare "a6" and "b6" here (then "b6" should be a
linklocal, too). I think it's too abstract for me what exactly "compare"
means in this case. :-)

- Alex