[PATCH 2 of 4 STABLE] lock: avoid unintentional lock acquisition at failure of readlock

[Yuya Nishihara]

On Sat, 29 Apr 2017 17:38:08 +0900, FUJIWARA Katsunori wrote:
> # HG changeset patch
> # User FUJIWARA Katsunori <foozy at lares.dti.ne.jp>
> # Date 1493454422 -32400
> #      Sat Apr 29 17:27:02 2017 +0900
> # Branch stable
> # Node ID d1bf658c4ea6d1f87328884bc00f2b91d884a6d9
> # Parent  7cb8ebcca7d3a542031c41e09dc3ca0e2db36dcc
> lock: avoid unintentional lock acquisition at failure of readlock

> diff --git a/mercurial/lock.py b/mercurial/lock.py
> --- a/mercurial/lock.py
> +++ b/mercurial/lock.py
> @@ -148,6 +148,12 @@ class lock(object):
>                      raise error.LockUnavailable(why.errno, why.strerror,
>                                                  why.filename, self.desc)
>  
> +        if not self.held:
> +            # use 0 to mean "busy for frequent lock/unlock by many processes"
> +            locker = "%s:%d" % (lock._host, 0)
> +            raise error.LockHeld(errno.EAGAIN,
> +                                 self.vfs.join(self.f), self.desc, locker)

I hesitate to use pid=0 as a null value. It's sometimes dangerous though it
would be okay here.