From mboxrd@z Thu Jan 1 00:00:00 1970 From: Kent Overstreet Subject: Re: [PATCH 22/25] Generic dynamic per cpu refcounting Date: Thu, 29 Nov 2012 12:22:31 -0800 Message-ID: <20121129202231.GJ15094@google.com> References: <1354121029-1376-1-git-send-email-koverstreet@google.com> <1354121029-1376-23-git-send-email-koverstreet@google.com> <20121129185720.GE15094@google.com> <20121129185953.GW16230@one.firstfloor.org> <20121129191214.GG15094@google.com> <20121129192003.GX16230@one.firstfloor.org> <20121129192925.GH15094@google.com> <20121129193452.GI19042@kvack.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Andi Kleen , linux-kernel@vger.kernel.org, linux-aio@kvack.org, linux-fsdevel@vger.kernel.org, zab@redhat.com, jmoyer@redhat.com, axboe@kernel.dk, viro@zeniv.linux.org.uk To: Benjamin LaHaise Return-path: Received: from mail-pa0-f46.google.com ([209.85.220.46]:53054 "EHLO mail-pa0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752545Ab2K2UWf (ORCPT ); Thu, 29 Nov 2012 15:22:35 -0500 Received: by mail-pa0-f46.google.com with SMTP id bh2so7882603pad.19 for ; Thu, 29 Nov 2012 12:22:35 -0800 (PST) Content-Disposition: inline In-Reply-To: <20121129193452.GI19042@kvack.org> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: On Thu, Nov 29, 2012 at 02:34:52PM -0500, Benjamin LaHaise wrote: > On Thu, Nov 29, 2012 at 11:29:25AM -0800, Kent Overstreet wrote: > > There's some kind of symmetry going on here, and if I'd been awake more > > in college I could probably say exactly why it works, but it does. > > I think the catch is that using only a 32 bit counter is something the > user could arbitrarily control the sum of all parts. I think a 64 bit > counter may be required to ensure no overflow occurs. Otherwise, an > overflow could result in a premature free when there are still 2^32 > objects active thanks to a malicious user (possible on systems with lots > of memory these days -- remote, but possible). That's no different from regular atomic_t - but you're right, we should be using size_t for anything userspace can manipulate. Not gonna worry about it in this patch though because the refcount was an atomic_t before and userspace can only do one get per thread.