Date: Tue, 19 Dec 2017 12:22:32 -0800
From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
To: Rao Shoaib <rao.shoaib@oracle.com>
Cc: Matthew Wilcox <willy@infradead.org>, linux-kernel@vger.kernel.org,
        brouer@redhat.com, linux-mm@kvack.org
Subject: Re: [PATCH] kfree_rcu() should use the new kfree_bulk() interface
 for freeing rcu structures
Reply-To: paulmck@linux.vnet.ibm.com
References: <rao.shoaib@oracle.com>
 <1513705948-31072-1-git-send-email-rao.shoaib@oracle.com>
 <20171219193039.GB6515@bombadil.infradead.org>
 <24c9f1c0-58d4-5d27-8795-d211693455dd@oracle.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <24c9f1c0-58d4-5d27-8795-d211693455dd@oracle.com>
User-Agent: Mutt/1.5.21 (2010-09-15)
Message-Id: <20171219202232.GE7829@linux.vnet.ibm.com>
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1362
Lines: 33

On Tue, Dec 19, 2017 at 11:56:30AM -0800, Rao Shoaib wrote:
> On 12/19/2017 11:30 AM, Matthew Wilcox wrote:
> >On Tue, Dec 19, 2017 at 09:52:27AM -0800, rao.shoaib@oracle.com wrote:

[ . . . ]

> >I've been doing a lot of thinking about this because I really want a
> >way to kfree_rcu() an object without embedding a struct rcu_head in it.
> >But I see no way to do that today; even if we have an external memory
> >allocation to point to the object to be freed, we have to keep track of
> >the grace periods.
> I am not sure I understand. If you had external memory you can
> easily do that.
> I am exactly doing that, the only reason the RCU structure is needed
> is to get the pointer to the object being freed.

This can be done as long as you are willing to either:

1.	Occasionally have kfree_rcu() wait for a grace period.

2.	Occasionally have kfree_rcu() allocate memory.

3.	Keep the rcu_head, but use it only when you would otherwise
	have to accept the above two penalties.  (The point of this
	is that tracking lists of memory waiting for a grace period
	using dense arrays improves cache locality.)

There might be others, and if you come up with one, please don't keep it
a secret.  The C++ standards committee insisted on an interface using
option #2 above.  (There is also an option to use their equivalent of
an rcu_head.)

							Thanx, Paul