akr / fsij.org wrote:
> I tried to workaround this issue at r45047.
> However I don't have an environment to reproduce the problem.

I don't know how to reproduce the problem, either.  netlink sockets
should not block like that.

r45047 looks correct.  A few minor comments:
* backwards for loop for list is confusing to me,
  any reason for not reversing list declaration?
* why xmalloc + MEMZERO?  xcalloc is shorter and generates smaller code