From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
To: Lance Roy <ldr709@gmail.com>
Cc: bobby.prani@gmail.com, linux-kernel@vger.kernel.org,
mingo@kernel.org, jiangshanlai@gmail.com, dipankar@in.ibm.com,
akpm@linux-foundation.org, mathieu.desnoyers@efficios.com,
josh@joshtriplett.org, tglx@linutronix.de, peterz@infradead.org,
rostedt@goodmis.org, dhowells@redhat.com, edumazet@google.com,
dvhart@linux.intel.com, fweisbec@gmail.com, oleg@redhat.com
Subject: Re: [PATCH] srcu: Implement more-efficient reader counts
Date: Tue, 24 Jan 2017 09:07:19 -0800 [thread overview]
Message-ID: <20170124170719.GQ28085@linux.vnet.ibm.com> (raw)
In-Reply-To: <20170123192645.69507b2f@gmail.com>
On Mon, Jan 23, 2017 at 07:26:45PM -0800, Lance Roy wrote:
> > Yeah, we did have this same conversation awhile back, didn't we?
> >
> > Back then, did I think to ask if this could be minimized or even prevented
> > by adding memory barriers appropriately? ;-)
> >
> > Thanx, Paul
>
> Yes, it can be fixed by adding a memory barrier after incrementing ->completed
> inside srcu_flip(). The upper limit on NR_CPUS turns out to be more complicated
> than this, as it needs to deal with highly nested read side critical sections
> mixed with the critical section loops, but only the one memory barrier should
> be necessary.
Something like this, then?
Thanx, Paul
------------------------------------------------------------------------
commit 35be9e413dde662fc9661352e595105ac4b0b167
Author: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Date: Tue Jan 24 08:51:34 2017 -0800
srcu: Reduce probability of SRCU ->unlock_count[] counter overflow
Because there are no memory barriers between the srcu_flip() ->completed
increment and the summation of the read-side ->unlock_count[] counters,
both the compiler and the CPU can reorder the summation with the
->completed increment. If the updater is preempted long enough during
this process, the read-side counters could overflow, resulting in a
too-short grace period.
This commit therefore adds a memory barrier just after the ->completed
increment, ensuring that if the summation misses an increment of
->unlock_count[] from __srcu_read_unlock(), the next __srcu_read_lock()
will see the new value of ->completed, thus bounding the number of
->unlock_count[] increments that can be missed to NR_CPUS. The actual
overflow computation is more complex due to the possibility of nesting
of __srcu_read_lock().
Reported-by: Lance Roy <ldr709@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
diff --git a/kernel/rcu/srcu.c b/kernel/rcu/srcu.c
index d3378ceb9762..aefe3ab20a6a 100644
--- a/kernel/rcu/srcu.c
+++ b/kernel/rcu/srcu.c
@@ -337,7 +337,16 @@ static bool try_check_zero(struct srcu_struct *sp, int idx, int trycount)
*/
static void srcu_flip(struct srcu_struct *sp)
{
- sp->completed++;
+ WRITE_ONCE(sp->completed, sp->completed + 1);
+
+ /*
+ * Ensure that if the updater misses an __srcu_read_unlock()
+ * increment, that task's next __srcu_read_lock() will see the
+ * above counter update. Note that both this memory barrier
+ * and the one in srcu_readers_active_idx_check() provide the
+ * guarantee for __srcu_read_lock().
+ */
+ smp_mb(); /* D */ /* Pairs with C. */
}
/*
next prev parent reply other threads:[~2017-01-24 17:07 UTC|newest]
Thread overview: 43+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-01-14 9:19 [PATCH tip/core/rcu 0/3] SRCU updates for 4.11 Paul E. McKenney
2017-01-14 9:19 ` [PATCH tip/core/rcu 1/3] srcu: More efficient reader counts Paul E. McKenney
2017-01-14 9:31 ` Ingo Molnar
2017-01-14 19:48 ` Paul E. McKenney
2017-01-14 9:20 ` [PATCH tip/core/rcu 2/3] srcu: Force full grace-period ordering Paul E. McKenney
2017-01-14 9:35 ` Ingo Molnar
2017-01-14 19:54 ` Paul E. McKenney
2017-01-14 21:41 ` Paul E. McKenney
2017-01-15 7:11 ` Ingo Molnar
2017-01-15 7:40 ` Paul E. McKenney
2017-01-15 7:57 ` Ingo Molnar
2017-01-15 9:24 ` Paul E. McKenney
2017-01-15 9:40 ` Ingo Molnar
2017-01-15 19:45 ` Paul E. McKenney
2017-01-16 6:56 ` Ingo Molnar
2017-01-23 8:12 ` Michael Ellerman
2017-01-24 2:45 ` Paul E. McKenney
2017-01-15 6:54 ` Ingo Molnar
2017-01-14 9:20 ` [PATCH tip/core/rcu 3/3] rcutorture: Add CBMC-based formal verification for SRCU Paul E. McKenney
2017-01-15 22:41 ` [PATCH tip/core/rcu v2 0/3] SRCU updates for 4.11 Paul E. McKenney
2017-01-15 22:42 ` [PATCH v2 tip/core/rcu 1/3] srcu: Implement more-efficient reader counts Paul E. McKenney
2017-01-23 20:17 ` Lance Roy
2017-01-23 20:17 ` [PATCH] SRCU: More efficient " Lance Roy
2017-01-23 20:35 ` Paul E. McKenney
2017-01-23 21:33 ` Lance Roy
2017-01-23 21:35 ` [PATCH] srcu: Implement more-efficient " Lance Roy
2017-01-24 0:42 ` Paul E. McKenney
2017-01-24 0:53 ` Lance Roy
2017-01-24 1:57 ` Paul E. McKenney
2017-01-24 3:26 ` Lance Roy
2017-01-24 17:07 ` Paul E. McKenney [this message]
2017-01-15 22:42 ` [PATCH v2 tip/core/rcu 2/3] srcu: Force full grace-period ordering Paul E. McKenney
2017-01-23 8:38 ` Lance Roy
2017-01-23 19:12 ` Paul E. McKenney
2017-01-23 20:06 ` Lance Roy
2017-01-15 22:42 ` [PATCH v2 tip/core/rcu 3/3] rcutorture: Add CBMC-based formal verification for SRCU Paul E. McKenney
2017-01-24 22:00 ` [PATCH v3 tip/core/rcu 0/4] SRCU updates for 4.11 Paul E. McKenney
2017-01-24 22:00 ` [PATCH v3 tip/core/rcu 1/4] srcu: Implement more-efficient reader counts Paul E. McKenney
2017-01-25 18:17 ` Lance Roy
2017-01-25 21:03 ` Paul E. McKenney
2017-01-24 22:00 ` [PATCH v3 tip/core/rcu 2/4] srcu: Force full grace-period ordering Paul E. McKenney
2017-01-24 22:00 ` [PATCH v3 tip/core/rcu 3/4] rcutorture: Add CBMC-based formal verification for SRCU Paul E. McKenney
2017-01-24 22:00 ` [PATCH v3 tip/core/rcu 4/4] srcu: Reduce probability of SRCU ->unlock_count[] counter overflow Paul E. McKenney
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170124170719.GQ28085@linux.vnet.ibm.com \
--to=paulmck@linux.vnet.ibm.com \
--cc=akpm@linux-foundation.org \
--cc=bobby.prani@gmail.com \
--cc=dhowells@redhat.com \
--cc=dipankar@in.ibm.com \
--cc=dvhart@linux.intel.com \
--cc=edumazet@google.com \
--cc=fweisbec@gmail.com \
--cc=jiangshanlai@gmail.com \
--cc=josh@joshtriplett.org \
--cc=ldr709@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mathieu.desnoyers@efficios.com \
--cc=mingo@kernel.org \
--cc=oleg@redhat.com \
--cc=peterz@infradead.org \
--cc=rostedt@goodmis.org \
--cc=tglx@linutronix.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).