From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Paul E. McKenney" Subject: Re: [PATCH] fix a race condition in cancelable mcs spinlocks Date: Mon, 2 Jun 2014 06:24:54 -0700 Message-ID: <20140602132454.GO22231@linux.vnet.ibm.com> References: <20140601192026.GE16155@laptop.programming.kicks-ass.net> <20140601213003.GG16155@laptop.programming.kicks-ass.net> Reply-To: paulmck@linux.vnet.ibm.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Peter Zijlstra , John David Anglin , Linus Torvalds , jejb@parisc-linux.org, deller@gmx.de, linux-parisc@vger.kernel.org, linux-kernel@vger.kernel.org, chegu_vinod@hp.com, Waiman.Long@hp.com, tglx@linutronix.de, riel@redhat.com, akpm@linux-foundation.org, davidlohr@hp.com, hpa@zytor.com, andi@firstfloor.org, aswin@hp.com, scott.norton@hp.com, Jason Low To: Mikulas Patocka Return-path: In-Reply-To: List-ID: List-Id: linux-parisc.vger.kernel.org On Mon, Jun 02, 2014 at 05:19:39AM -0400, Mikulas Patocka wrote: > > > On Sun, 1 Jun 2014, Peter Zijlstra wrote: > > > On Sun, Jun 01, 2014 at 04:46:26PM -0400, John David Anglin wrote: > > > On 1-Jun-14, at 3:20 PM, Peter Zijlstra wrote: > > > > > > >>If you write to some variable with ACCESS_ONCE and use cmpxchg or xchg > > > >>at > > > >>the same time, you break it. ACCESS_ONCE doesn't take the hashed > > > >>spinlock, > > > >>so, in this case, cmpxchg or xchg isn't really atomic at all. > > > > > > > >And this is really the first place in the kernel that breaks like this? > > > >I've been using xchg() and cmpxchg() without such consideration for > > > >quite a while. > > > > > > I believe Mikulas is correct. Even in a controlled situation where a > > > cmpxchg operation > > > is used to implement pthread_spin_lock() in userspace, we found recently > > > that the lock > > > must be released with a cmpxchg operation and not a simple write on SMP > > > systems. > > > There is a race in the cache operations or instruction ordering that's not > > > present with > > > the ldcw instruction. > > > > Oh, I'm not arguing that. He's quite right that its broken, but this > > form of atomic ops is also quite insane and unusual. Most sane machines > > don't have this problem. > > > > My main concern is how are we going to avoid breaking parisc (and I > > think sparc32, which is similarly retarded) in the future; we should > > invest in machinery to find and detect these things. > > Grep the kernel for "\" and "\" and replace them with > atomic types and atomic access functions. Not so good for pointers, though. Defeats type-checking, for one thing. An example of this is use of xchg() for atomically enqueuing RCU callbacks in kernel/rcu/tree_plugin.h. I still like the idea of PA-RISC's compiler implementing ACCESS_ONCE() as needed to make things work on that architecture. Thanx, Paul