linux-kernel - [PATCH v2 4/5] locking/pvstat: Add stat counter to track _Q_SLOW

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Tue, 31 May 2016 12:53:50 -0400
From:	Waiman Long <Waiman.Long@....com>
To:	Peter Zijlstra <peterz@...radead.org>,
	Ingo Molnar <mingo@...hat.com>
Cc:	linux-kernel@...r.kernel.org,
	Pan Xinhui <xinhui@...ux.vnet.ibm.com>,
	Boqun Feng <boqun.feng@...il.com>,
	Scott J Norton <scott.norton@....com>,
	Douglas Hatch <doug.hatch@....com>,
	Waiman Long <Waiman.Long@....com>
Subject: [PATCH v2 4/5] locking/pvstat: Add stat counter to track _Q_SLOW_VAL race

This patch adds a new stat counter to track if the _Q_SLOW_VAL race
between lock owner running pv_kick_node() and queue head running
pv_wait_head_or_lock() is happening.

Signed-off-by: Waiman Long <Waiman.Long@....com>
---
 kernel/locking/qspinlock_paravirt.h |    4 +++-
 kernel/locking/qspinlock_stat.h     |    5 ++++-
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/kernel/locking/qspinlock_paravirt.h b/kernel/locking/qspinlock_paravirt.h
index 54e03b9..3d57296 100644
--- a/kernel/locking/qspinlock_paravirt.h
+++ b/kernel/locking/qspinlock_paravirt.h
@@ -468,8 +468,10 @@ pv_wait_head_or_lock(struct qspinlock *lock, struct mcs_spinlock *node)
 					WRITE_ONCE(l->locked, _Q_LOCKED_VAL);
 					clear_pending(lock);
 					goto gotlock;
+				} else {
+					/* old == _Q_SLOW_VAL. */
+					qstat_inc(qstat_pv_slow_race, true);
 				}
-				/* old == _Q_SLOW_VAL. */
 			}
 		}
 		clear_pending(lock);	/* Enable lock stealing */
diff --git a/kernel/locking/qspinlock_stat.h b/kernel/locking/qspinlock_stat.h
index 3cd07a9..a40262e 100644
--- a/kernel/locking/qspinlock_stat.h
+++ b/kernel/locking/qspinlock_stat.h
@@ -24,6 +24,7 @@
  *   pv_latency_wake	- average latency (ns) from vCPU kick to wakeup
  *   pv_lock_slowpath	- # of locking operations via the slowpath
  *   pv_lock_stealing	- # of lock stealing operations
+ *   pv_slow_race	- # of _Q_SLOW_VAL races between lock owner & qhead
  *   pv_spurious_wakeup	- # of spurious wakeups in non-head vCPUs
  *   pv_wait_again	- # of wait's after a queue head vCPU kick
  *   pv_wait_early	- # of early vCPU wait's
@@ -48,6 +49,7 @@ enum qlock_stats {
 	qstat_pv_latency_wake,
 	qstat_pv_lock_slowpath,
 	qstat_pv_lock_stealing,
+	qstat_pv_slow_race,
 	qstat_pv_spurious_wakeup,
 	qstat_pv_wait_again,
 	qstat_pv_wait_early,
@@ -69,11 +71,12 @@ static const char * const qstat_names[qstat_num + 1] = {
 	[qstat_pv_hash_hops]	   = "pv_hash_hops",
 	[qstat_pv_kick_unlock]     = "pv_kick_unlock",
 	[qstat_pv_kick_wake]       = "pv_kick_wake",
-	[qstat_pv_spurious_wakeup] = "pv_spurious_wakeup",
 	[qstat_pv_latency_kick]	   = "pv_latency_kick",
 	[qstat_pv_latency_wake]    = "pv_latency_wake",
 	[qstat_pv_lock_slowpath]   = "pv_lock_slowpath",
 	[qstat_pv_lock_stealing]   = "pv_lock_stealing",
+	[qstat_pv_slow_race]	   = "pv_slow_race",
+	[qstat_pv_spurious_wakeup] = "pv_spurious_wakeup",
 	[qstat_pv_wait_again]      = "pv_wait_again",
 	[qstat_pv_wait_early]      = "pv_wait_early",
 	[qstat_pv_wait_head]       = "pv_wait_head",
-- 
1.7.1