Re: [PATCH -next v2] locking/osq_lock: annotate a data race in osq_lock

[Paul E. McKenney]

On Sat, May 09, 2020 at 09:01:53AM -0400, Qian Cai wrote:
> 
> 
> > On May 9, 2020, at 12:33 AM, Paul E. McKenney <paulmck@xxxxxxxxxx> wrote:
> > 
> > On Fri, May 08, 2020 at 04:59:05PM -0400, Qian Cai wrote:
> >> 
> >> 
> >>> On Feb 11, 2020, at 8:54 AM, Qian Cai <cai@xxxxxx> wrote:
> >>> 
> >>> prev->next could be accessed concurrently as noticed by KCSAN,
> >>> 
> >>> write (marked) to 0xffff9d3370dbbe40 of 8 bytes by task 3294 on cpu 107:
> >>> osq_lock+0x25f/0x350
> >>> osq_wait_next at kernel/locking/osq_lock.c:79
> >>> (inlined by) osq_lock at kernel/locking/osq_lock.c:185
> >>> rwsem_optimistic_spin
> >>> <snip>
> >>> 
> >>> read to 0xffff9d3370dbbe40 of 8 bytes by task 3398 on cpu 100:
> >>> osq_lock+0x196/0x350
> >>> osq_lock at kernel/locking/osq_lock.c:157
> >>> rwsem_optimistic_spin
> >>> <snip>
> >>> 
> >>> Since the write only stores NULL to prev->next and the read tests if
> >>> prev->next equals to this_cpu_ptr(&osq_node). Even if the value is
> >>> shattered, the code is still working correctly. Thus, mark it as an
> >>> intentional data race using the data_race() macro.
> >>> 
> >>> Signed-off-by: Qian Cai <cai@xxxxxx>
> >> 
> >> Hmm, this patch has been dropped from linux-next from some reasons.
> >> 
> >> Paul, can you pick this up along with KCSAN fixes?
> >> 
> >> https://lore.kernel.org/lkml/1581429255-12542-1-git-send-email-cai@xxxxxx/
> > 
> > I have queued it on -rcu, but it is too late for v5.8 via the -rcu tree,
> > so this is v5.9 at the earliest.  Plus I would need an ack from one of
> > the locking folks.
> 
> Peter, Will, can you give an ACK? This v2 should incorporate all the feedback from Peter,
> 
> https://lore.kernel.org/lkml/20200211124753.GP14914@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/
> 
> V5.9 is fine. All I care about is it is always in linux-next (so the testing bots wonât trigger this over and over again) and to be in mainline at some point in the future.

Ah, and I forgot to ask.  Why "if (data_race(prev->next == node)" instead
of "if (data_race(prev->next) == node"?

							Thanx, Paul

> >>> ---
> >>> 
> >>> v2: insert some code comments.
> >>> 
> >>> kernel/locking/osq_lock.c | 6 +++++-
> >>> 1 file changed, 5 insertions(+), 1 deletion(-)
> >>> 
> >>> diff --git a/kernel/locking/osq_lock.c b/kernel/locking/osq_lock.c
> >>> index 1f7734949ac8..f733bcd99e8a 100644
> >>> --- a/kernel/locking/osq_lock.c
> >>> +++ b/kernel/locking/osq_lock.c
> >>> @@ -154,7 +154,11 @@ bool osq_lock(struct optimistic_spin_queue *lock)
> >>> 	 */
> >>> 
> >>> 	for (;;) {
> >>> -		if (prev->next == node &&
> >>> +		/*
> >>> +		 * cpu_relax() below implies a compiler barrier which would
> >>> +		 * prevent this comparison being optimized away.
> >>> +		 */
> >>> +		if (data_race(prev->next == node) &&
> >>> 		    cmpxchg(&prev->next, node, NULL) == node)
> >>> 			break;
> >>> 
> >>> -- 
> >>> 1.8.3.1
>