linux-kernel - [RFC PATCH 10/12] locking/rwsem: Mark rwsem.wait

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]

Message-Id: <1541709268-3766-11-git-send-email-longman@redhat.com>
Date:   Thu,  8 Nov 2018 15:34:26 -0500
From:   Waiman Long <longman@...hat.com>
To:     Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        Will Deacon <will.deacon@....com>,
        Thomas Gleixner <tglx@...utronix.de>
Cc:     linux-kernel@...r.kernel.org, kasan-dev@...glegroups.com,
        linux-mm@...ck.org, Petr Mladek <pmladek@...e.com>,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
        Andrey Ryabinin <aryabinin@...tuozzo.com>,
        Tejun Heo <tj@...nel.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Waiman Long <longman@...hat.com>
Subject: [RFC PATCH 10/12] locking/rwsem: Mark rwsem.wait_lock as a terminal lock

The wait_lock in a rwsem is always acquired with IRQ disabled. For the
rwsem-xadd.c implementation, no other lock will be called while holding
the wait_lock. So it satisfies the condition of being a terminal lock.
By marking it as terminal,  the lockdep overhead will be reduced.

Signed-off-by: Waiman Long <longman@...hat.com>
---
 include/linux/rwsem.h       | 11 ++++++++++-
 kernel/locking/rwsem-xadd.c |  1 +
 2 files changed, 11 insertions(+), 1 deletion(-)

diff --git a/include/linux/rwsem.h b/include/linux/rwsem.h
index 67dbb57..a2a2385 100644
--- a/include/linux/rwsem.h
+++ b/include/linux/rwsem.h
@@ -77,16 +77,25 @@ static inline int rwsem_is_locked(struct rw_semaphore *sem)
 # define __RWSEM_DEP_MAP_INIT(lockname)
 #endif
 
+/*
+ * The wait_lock is marked as a terminal lock to reduce lockdep overhead
+ * when the rwsem-xadd.c is used. This is implied when
+ * CONFIG_RWSEM_SPIN_ON_OWNER is true. The rwsem-spinlock.c implementation
+ * allows calling wake_up_process() while holding the wait_lock. So it
+ * can't be marked as terminal in this case.
+ */
 #ifdef CONFIG_RWSEM_SPIN_ON_OWNER
 #define __RWSEM_OPT_INIT(lockname) , .osq = OSQ_LOCK_UNLOCKED, .owner = NULL
+#define __RWSEM_WAIT_LOCK_INIT(x)	__RAW_TERMINAL_SPIN_LOCK_UNLOCKED(x)
 #else
 #define __RWSEM_OPT_INIT(lockname)
+#define __RWSEM_WAIT_LOCK_INIT(x)	__RAW_SPIN_LOCK_UNLOCKED(x)
 #endif
 
 #define __RWSEM_INITIALIZER(name)				\
 	{ __RWSEM_INIT_COUNT(name),				\
 	  .wait_list = LIST_HEAD_INIT((name).wait_list),	\
-	  .wait_lock = __RAW_SPIN_LOCK_UNLOCKED(name.wait_lock)	\
+	  .wait_lock = __RWSEM_WAIT_LOCK_INIT(name.wait_lock)	\
 	  __RWSEM_OPT_INIT(name)				\
 	  __RWSEM_DEP_MAP_INIT(name) }
 
diff --git a/kernel/locking/rwsem-xadd.c b/kernel/locking/rwsem-xadd.c
index 09b1800..3dbe593 100644
--- a/kernel/locking/rwsem-xadd.c
+++ b/kernel/locking/rwsem-xadd.c
@@ -85,6 +85,7 @@ void __init_rwsem(struct rw_semaphore *sem, const char *name,
 #endif
 	atomic_long_set(&sem->count, RWSEM_UNLOCKED_VALUE);
 	raw_spin_lock_init(&sem->wait_lock);
+	lockdep_set_terminal_class(&sem->wait_lock);
 	INIT_LIST_HEAD(&sem->wait_list);
 #ifdef CONFIG_RWSEM_SPIN_ON_OWNER
 	sem->owner = NULL;
-- 
1.8.3.1