[PATCH stable] lock: handle race in trylock/testlock on Windows

[Yuya Nishihara]

On Wed, 02 Oct 2013 18:11:09 +0400, Nikolaj Sjujskij wrote:
> On Wed, 02 Oct 2013 16:59:28 +0400, Yuya Nishihara <yuya at tcha.org> wrote:
> 
> > # HG changeset patch
> > # User Yuya Nishihara <yuya at tcha.org>
> > # Date 1380716200 -32400
> > #      Wed Oct 02 21:16:40 2013 +0900
> > # Branch stable
> > # Node ID 964b123c4831abac69efc5a089c446222915713f
> > # Parent  4d513f96a565639c0879086b68cf72a1e01e596c
> > lock: handle race in trylock/testlock on Windows
> >
> > readlock may raise IOError if the system does not support symlink.
> >
> > diff --git a/mercurial/lock.py b/mercurial/lock.py
> > --- a/mercurial/lock.py
> > +++ b/mercurial/lock.py
> > @@ -100,7 +100,7 @@ class lock(object):
> >          """
> >          try:
> >              locker = util.readlock(self.f)
> > -        except OSError, why:
> > +        except (OSError, IOError), why:
> >              if why.errno == errno.ENOENT:
> >                  return None
> >              raise
> 
>   Could IOError.errno == errno.ENOENT? Otherwise this change doesn't seem  
> to really change anything.

Yes.  I've confirmed the problem on Windows by inserting sleep(1) in trylock.
And if errno isn't ENOENT, the exception must be raised.

Regards,