[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20180521153337.GF3803@linux.vnet.ibm.com>
Date: Mon, 21 May 2018 08:33:37 -0700
From: "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
To: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: Roman Pen <roman.penyaev@...fitbricks.com>,
linux-block <linux-block@...r.kernel.org>,
linux-rdma <linux-rdma@...r.kernel.org>,
Jens Axboe <axboe@...nel.dk>,
Christoph Hellwig <hch@...radead.org>,
Sagi Grimberg <sagi@...mberg.me>,
Bart Van Assche <bart.vanassche@...disk.com>,
Or Gerlitz <ogerlitz@...lanox.com>,
Doug Ledford <dledford@...hat.com>,
swapnil.ingle@...fitbricks.com, danil.kipnis@...fitbricks.com,
Jinpu Wang <jinpu.wang@...fitbricks.com>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2 01/26] rculist: introduce list_next_or_null_rr_rcu()
On Mon, May 21, 2018 at 08:16:59AM -0700, Linus Torvalds wrote:
> On Mon, May 21, 2018 at 6:51 AM Roman Penyaev <
> roman.penyaev@...fitbricks.com> wrote:
>
> > No, I continue from the pointer, which I assigned on the previous IO
> > in order to send IO fairly and keep load balanced.
>
> Right. And that's exactly what has both me and Paul nervous. You're no
> longer in the RCU domain. You're using a pointer where the lifetime has
> nothing to do with RCU any more.
>
> Can it be done? Sure. But you need *other* locking for it (that you haven't
> explained), and it's fragile as hell.
He looks to actually have it right, but I would want to see a big comment
on the read side noting the leak of the pointer and documenting why it
is OK.
Thanx, Paul
> It's probably best to not use RCU for it at all, but depend on that "other
> locking" that you have to have anyway, to keep the pointer valid over the
> non-RCU region.
>
> Linus
>
Powered by blists - more mailing lists