From: YOSHIFUJI Hideaki / 吉藤英明 [off-list ref]
   Date: Mon, 31 Mar 2003 10:34:51 +0900 (JST)

   In article [ref] (at Mon, 31 Mar 2003 03:35:24 +0900 (JST)), YOSHIFUJI Hideaki / 吉藤英明 [off-list ref] says:
   
   > In article [ref] (at Sun, 30 Mar 2003 18:36:56 +0200), Simone Piunno [off-list ref] says:
   > 
   > >  - locking inside ipv6_add_addr() is simpler and more linear but
   > >    semantically wrong because you're unable to tell the user why his 
   > >    "ip addr add" failed.  E.g. you answer ENOBUFS instead of EEXIST.
   > 
   > We don't want to create duplicate address in any case.
   > ipv6_add_addr() IS right place.
   > And, we can return error code by using IS_ERR() etc.
   > I'll fix this.
   
   Here's the revised patch.

Applied to both 2.4.x and 2.5.x.

BTW, 2.4.x patch failed in two spots, one was author comment
which I easily fixed, second was in privacy code which I did not
apply yet to 2.4.x (I fixed this too, don't worry).

I do not want to put privacy code into 2.4.x until crypto is there.
I plan to put crypto lib into 2.4.22-pre1.