[PATCH] Initialize the entire obstack struct [BZ #17919]

[Jim Meyering]

On Tue, Feb 3, 2015 at 8:08 AM, Paul Eggert <eggert@cs.ucla.edu> wrote:
> Does this patch fix any actual bug?  If so, what's the bug?  Is there a
> faster way to fix it, other than zeroing out the entire struct obstack?

I believe that this addresses the MSAN used-uninitialized warning
I explored privately a few months ago. Assuming it's the same one,
initializing the entire struct obstack is not necessary.