From mboxrd@z Thu Jan  1 00:00:00 1970
From: Bob Peterson <rpeterso@redhat.com>
Date: Thu, 31 Jan 2019 10:04:18 -0500 (EST)
Subject: [Cluster-devel] [PATCH 2/2] gfs2: Fix lru_count going negative
In-Reply-To: <1247235469.68837763.1548945397399.JavaMail.zimbra@redhat.com>
References: <20190131105543.15421-1-ross.lagerwall@citrix.com>
	<20190131105543.15421-3-ross.lagerwall@citrix.com>
	<1247235469.68837763.1548945397399.JavaMail.zimbra@redhat.com>
Message-ID: <359471440.68850881.1548947058822.JavaMail.zimbra@redhat.com>
List-Id: <cluster-devel.redhat.com>
To: cluster-devel.redhat.com
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit

----- Original Message -----
> > +
> > +	if (!test_bit(GLF_LRU, &gl->gl_flags)) {
> > +		set_bit(GLF_LRU, &gl->gl_flags);
> >  		atomic_inc(&lru_count);
> > +	}
> 
> The above may be simplified to something like:
> +	if (!test_and_set_bit(GLF_LRU, &gl->gl_flags))
>  		atomic_inc(&lru_count);

Scratch that. Andreas says test_and_set_bit() and similar are much more
expensive cpu-wise, and we're already protected from races by the
lru_lock, so I guess the original is better after all.

Bob Peterson