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] [day] [month] [year] [list]
Message-Id: <1466876272-3824-3-git-send-email-manfred@colorfullife.com>
Date:	Sat, 25 Jun 2016 19:37:52 +0200
From:	Manfred Spraul <manfred@...orfullife.com>
To:	"H. Peter Anvin" <hpa@...or.com>,
	Peter Zijlstra <peterz@...radead.org>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Davidlohr Bueso <dave@...olabs.net>
Cc:	LKML <linux-kernel@...r.kernel.org>,
	Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...e.hu>, 1vier1@....de,
	felixh@...ormatik.uni-bremen.de,
	Manfred Spraul <manfred@...orfullife.com>
Subject: [PATCH 2/2] ipc/sem: sem_lock with hysteresis

sysv sem has two lock modes: One with per-semaphore locks, one lock mode
with a single big lock for the whole array.
When switching from the per-semaphore locks to the big lock, all
per-semaphore locks must be scanned for ongoing operations.

The patch adds a hysteresis for switching from the big lock to the per
semaphore locks. This reduces how often the per-semaphore locks must
be scanned.

Passed stress testing with sem-scalebench.

Signed-off-by: Manfred Spraul <manfred@...orfullife.com>
---
 include/linux/sem.h |  2 +-
 ipc/sem.c           | 89 +++++++++++++++++++++++++++++------------------------
 2 files changed, 49 insertions(+), 42 deletions(-)

diff --git a/include/linux/sem.h b/include/linux/sem.h
index d0efd6e..6fb3227 100644
--- a/include/linux/sem.h
+++ b/include/linux/sem.h
@@ -21,7 +21,7 @@ struct sem_array {
 	struct list_head	list_id;	/* undo requests on this array */
 	int			sem_nsems;	/* no. of semaphores in array */
 	int			complex_count;	/* pending complex operations */
-	bool			complex_mode;	/* no parallel simple ops */
+	int			complex_mode;	/* >0: no parallel simple ops */
 };
 
 #ifdef CONFIG_SYSVIPC
diff --git a/ipc/sem.c b/ipc/sem.c
index 538f43a..076b7c9 100644
--- a/ipc/sem.c
+++ b/ipc/sem.c
@@ -161,6 +161,13 @@ static int sysvipc_sem_proc_show(struct seq_file *s, void *it);
 #define SEMOPM_FAST	64  /* ~ 372 bytes on stack */
 
 /*
+ * Switching from the mode suitable for simple ops
+ * to the mode for complex ops is costly. Therefore:
+ * use some hysteresis
+ */
+#define COMPLEX_MODE_ENTER 10
+
+/*
  * Locking:
  * a) global sem_lock() for read/write
  *	sem_undo.id_next,
@@ -269,17 +276,25 @@ static void sem_rcu_free(struct rcu_head *head)
 /*
  * Enter the mode suitable for non-simple operations:
  * Caller must own sem_perm.lock.
+ * Note:
+ * There is no leave complex mode function. Leaving
+ * happens in sem_lock, with some hysteresis.
  */
 static void complexmode_enter(struct sem_array *sma)
 {
 	int i;
 	struct sem *sem;
 
-	if (sma->complex_mode)  {
-		/* We are already in complex_mode. Nothing to do */
+	if (sma->complex_mode > 0)  {
+		/*
+		 * We are already in complex_mode.
+		 * Nothing to do, just increase
+		 * counter until we return to simple mode
+		 */
+		WRITE_ONCE(sma->complex_mode, COMPLEX_MODE_ENTER);
 		return;
 	}
-	WRITE_ONCE(sma->complex_mode, true);
+	WRITE_ONCE(sma->complex_mode, COMPLEX_MODE_ENTER);
 
 	/* We need a full barrier:
 	 * The write to complex_mode must be visible
@@ -294,27 +309,6 @@ static void complexmode_enter(struct sem_array *sma)
 }
 
 /*
- * Try to leave the mode that disallows simple operations:
- * Caller must own sem_perm.lock.
- */
-static void complexmode_tryleave(struct sem_array *sma)
-{
-	if (sma->complex_count)  {
-		/* Complex ops are sleeping.
-		 * We must stay in complex mode
-		 */
-		return;
-	}
-	/*
-	 * Immediately after setting complex_mode to false,
-	 * a simple op can start. Thus: all memory writes
-	 * performed by the current operation must be visible
-	 * before we set complex_mode to false.
-	 */
-	smp_store_release(&sma->complex_mode, false);
-}
-
-/*
  * If the request contains only one semaphore operation, and there are
  * no complex transactions pending, lock only the semaphore involved.
  * Otherwise, lock the entire semaphore array, since we either have
@@ -372,27 +366,42 @@ static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
 	ipc_lock_object(&sma->sem_perm);
 
 	if (sma->complex_count == 0) {
-		/* False alarm:
-		 * There is no complex operation, thus we can switch
-		 * back to the fast path.
-		 */
-		spin_lock(&sem->lock);
-		ipc_unlock_object(&sma->sem_perm);
-		return sops->sem_num;
-	} else {
-		/* Not a false alarm, thus complete the sequence for a
-		 * full lock.
+		/*
+		 * Check if fast path is possible:
+		 * There is no complex operation, check hysteresis
+		 * If 0, switch back to the fast path.
 		 */
-		complexmode_enter(sma);
-		return -1;
+		if (sma->complex_mode > 0) {
+			/* Note:
+			 * Immediately after setting complex_mode to 0,
+			 * a simple op could start.
+			 * The data it would access was written by the
+			 * previous owner of sem->sem_perm.lock, i.e
+			 * a release and an acquire memory barrier ago.
+			 * No need for another barrier.
+			 */
+			WRITE_ONCE(sma->complex_mode, sma->complex_mode-1);
+		}
+		if (sma->complex_mode == 0) {
+			spin_lock(&sem->lock);
+			ipc_unlock_object(&sma->sem_perm);
+			return sops->sem_num;
+		}
 	}
+	/*
+	 * Not a false alarm, full lock is required.
+	 * Since we are already in complex_mode (either because of waiting
+	 * complex ops or due to hysteresis), there is not need for a
+	 * complexmode_enter().
+	 */
+	WARN_ON(sma->complex_mode == 0);
+	return -1;
 }
 
 static inline void sem_unlock(struct sem_array *sma, int locknum)
 {
 	if (locknum == -1) {
 		unmerge_queues(sma);
-		complexmode_tryleave(sma);
 		ipc_unlock_object(&sma->sem_perm);
 	} else {
 		struct sem *sem = sma->sem_base + locknum;
@@ -544,7 +553,7 @@ static int newary(struct ipc_namespace *ns, struct ipc_params *params)
 	}
 
 	sma->complex_count = 0;
-	sma->complex_mode = true; /* dropped by sem_unlock below */
+	WRITE_ONCE(sma->complex_mode, COMPLEX_MODE_ENTER);
 	INIT_LIST_HEAD(&sma->pending_alter);
 	INIT_LIST_HEAD(&sma->pending_const);
 	INIT_LIST_HEAD(&sma->list_id);
@@ -2201,7 +2210,7 @@ static int sysvipc_sem_proc_show(struct seq_file *s, void *it)
 	 * The proc interface isn't aware of sem_lock(), it calls
 	 * ipc_lock_object() directly (in sysvipc_find_ipc).
 	 * In order to stay compatible with sem_lock(), we must
-	 * enter / leave complex_mode.
+	 * enter complex_mode.
 	 */
 	complexmode_enter(sma);
 
@@ -2220,8 +2229,6 @@ static int sysvipc_sem_proc_show(struct seq_file *s, void *it)
 		   sem_otime,
 		   sma->sem_ctime);
 
-	complexmode_tryleave(sma);
-
 	return 0;
 }
 #endif
-- 
2.5.5

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ