Merge tag 'locking-core-2020-10-12' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull locking updates from Ingo Molnar:
 "These are the locking updates for v5.10:

   - Add deadlock detection for recursive read-locks.

     The rationale is outlined in commit 224ec489d3cd ("lockdep/
     Documention: Recursive read lock detection reasoning")

     The main deadlock pattern we want to detect is:

           TASK A:                 TASK B:

           read_lock(X);
                                   write_lock(X);
           read_lock_2(X);

   - Add "latch sequence counters" (seqcount_latch_t):

     A sequence counter variant where the counter even/odd value is used
     to switch between two copies of protected data. This allows the
     read path, typically NMIs, to safely interrupt the write side
     critical section.

     We utilize this new variant for sched-clock, and to make x86 TSC
     handling safer.

   - Other seqlock cleanups, fixes and enhancements

   - KCSAN updates

   - LKMM updates

   - Misc updates, cleanups and fixes"

* tag 'locking-core-2020-10-12' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (67 commits)
  lockdep: Revert "lockdep: Use raw_cpu_*() for per-cpu variables"
  lockdep: Fix lockdep recursion
  lockdep: Fix usage_traceoverflow
  locking/atomics: Check atomic-arch-fallback.h too
  locking/seqlock: Tweak DEFINE_SEQLOCK() kernel doc
  lockdep: Optimize the memory usage of circular queue
  seqlock: Unbreak lockdep
  seqlock: PREEMPT_RT: Do not starve seqlock_t writers
  seqlock: seqcount_LOCKNAME_t: Introduce PREEMPT_RT support
  seqlock: seqcount_t: Implement all read APIs as statement expressions
  seqlock: Use unique prefix for seqcount_t property accessors
  seqlock: seqcount_LOCKNAME_t: Standardize naming convention
  seqlock: seqcount latch APIs: Only allow seqcount_latch_t
  rbtree_latch: Use seqcount_latch_t
  x86/tsc: Use seqcount_latch_t
  timekeeping: Use seqcount_latch_t
  time/sched_clock: Use seqcount_latch_t
  seqlock: Introduce seqcount_latch_t
  mm/swap: Do not abuse the seqcount_t latching API
  time/sched_clock: Use raw_read_seqcount_latch() during suspend
  ...

1 files changed, 39 insertions, 19 deletions

diff --git a/include/linux/lockdep.h b/include/linux/lockdep.h
index 6a584b3e5c74..f5594879175a 100644
--- a/include/linux/lockdep.h
+++ b/include/linux/lockdep.h
@@ -54,7 +54,11 @@ struct lock_list {
 	struct lock_class		*class;
 	struct lock_class		*links_to;
 	const struct lock_trace		*trace;
-	int				distance;
+	u16				distance;
+	/* bitmap of different dependencies from head to this */
+	u8				dep;
+	/* used by BFS to record whether "prev -> this" only has -(*R)-> */
+	u8				only_xr;
 
 	/*
 	 * The parent field is used to implement breadth-first search, and the
@@ -469,6 +473,20 @@ static inline void print_irqtrace_events(struct task_struct *curr)
 }
 #endif
 
+/* Variable used to make lockdep treat read_lock() as recursive in selftests */
+#ifdef CONFIG_DEBUG_LOCKING_API_SELFTESTS
+extern unsigned int force_read_lock_recursive;
+#else /* CONFIG_DEBUG_LOCKING_API_SELFTESTS */
+#define force_read_lock_recursive 0
+#endif /* CONFIG_DEBUG_LOCKING_API_SELFTESTS */
+
+#ifdef CONFIG_LOCKDEP
+extern bool read_lock_is_recursive(void);
+#else /* CONFIG_LOCKDEP */
+/* If !LOCKDEP, the value is meaningless */
+#define read_lock_is_recursive() 0
+#endif
+
 /*
  * For trivial one-depth nesting of a lock-class, the following
  * global define can be used. (Subsystems with multiple levels
@@ -490,7 +508,14 @@ static inline void print_irqtrace_events(struct task_struct *curr)
 #define spin_release(l, i)			lock_release(l, i)
 
 #define rwlock_acquire(l, s, t, i)		lock_acquire_exclusive(l, s, t, NULL, i)
-#define rwlock_acquire_read(l, s, t, i)		lock_acquire_shared_recursive(l, s, t, NULL, i)
+#define rwlock_acquire_read(l, s, t, i)					\
+do {									\
+	if (read_lock_is_recursive())					\
+		lock_acquire_shared_recursive(l, s, t, NULL, i);	\
+	else								\
+		lock_acquire_shared(l, s, t, NULL, i);			\
+} while (0)
+
 #define rwlock_release(l, i)			lock_release(l, i)
 
 #define seqcount_acquire(l, s, t, i)		lock_acquire_exclusive(l, s, t, NULL, i)
@@ -512,19 +537,19 @@ static inline void print_irqtrace_events(struct task_struct *curr)
 #define lock_map_release(l)			lock_release(l, _THIS_IP_)
 
 #ifdef CONFIG_PROVE_LOCKING
-# define might_lock(lock) 						\
+# define might_lock(lock)						\
 do {									\
 	typecheck(struct lockdep_map *, &(lock)->dep_map);		\
 	lock_acquire(&(lock)->dep_map, 0, 0, 0, 1, NULL, _THIS_IP_);	\
 	lock_release(&(lock)->dep_map, _THIS_IP_);			\
 } while (0)
-# define might_lock_read(lock) 						\
+# define might_lock_read(lock)						\
 do {									\
 	typecheck(struct lockdep_map *, &(lock)->dep_map);		\
 	lock_acquire(&(lock)->dep_map, 0, 0, 1, 1, NULL, _THIS_IP_);	\
 	lock_release(&(lock)->dep_map, _THIS_IP_);			\
 } while (0)
-# define might_lock_nested(lock, subclass) 				\
+# define might_lock_nested(lock, subclass)				\
 do {									\
 	typecheck(struct lockdep_map *, &(lock)->dep_map);		\
 	lock_acquire(&(lock)->dep_map, subclass, 0, 1, 1, NULL,		\
@@ -534,44 +559,39 @@ do {									\
 
 DECLARE_PER_CPU(int, hardirqs_enabled);
 DECLARE_PER_CPU(int, hardirq_context);
+DECLARE_PER_CPU(unsigned int, lockdep_recursion);
 
-/*
- * The below lockdep_assert_*() macros use raw_cpu_read() to access the above
- * per-cpu variables. This is required because this_cpu_read() will potentially
- * call into preempt/irq-disable and that obviously isn't right. This is also
- * correct because when IRQs are enabled, it doesn't matter if we accidentally
- * read the value from our previous CPU.
- */
+#define __lockdep_enabled	(debug_locks && !this_cpu_read(lockdep_recursion))
 
 #define lockdep_assert_irqs_enabled()					\
 do {									\
-	WARN_ON_ONCE(debug_locks && !raw_cpu_read(hardirqs_enabled));	\
+	WARN_ON_ONCE(__lockdep_enabled && !this_cpu_read(hardirqs_enabled)); \
 } while (0)
 
 #define lockdep_assert_irqs_disabled()					\
 do {									\
-	WARN_ON_ONCE(debug_locks && raw_cpu_read(hardirqs_enabled));	\
+	WARN_ON_ONCE(__lockdep_enabled && this_cpu_read(hardirqs_enabled)); \
 } while (0)
 
 #define lockdep_assert_in_irq()						\
 do {									\
-	WARN_ON_ONCE(debug_locks && !raw_cpu_read(hardirq_context));	\
+	WARN_ON_ONCE(__lockdep_enabled && !this_cpu_read(hardirq_context)); \
 } while (0)
 
 #define lockdep_assert_preemption_enabled()				\
 do {									\
 	WARN_ON_ONCE(IS_ENABLED(CONFIG_PREEMPT_COUNT)	&&		\
-		     debug_locks			&&		\
+		     __lockdep_enabled			&&		\
 		     (preempt_count() != 0		||		\
-		      !raw_cpu_read(hardirqs_enabled)));		\
+		      !this_cpu_read(hardirqs_enabled)));		\
 } while (0)
 
 #define lockdep_assert_preemption_disabled()				\
 do {									\
 	WARN_ON_ONCE(IS_ENABLED(CONFIG_PREEMPT_COUNT)	&&		\
-		     debug_locks			&&		\
+		     __lockdep_enabled			&&		\
 		     (preempt_count() == 0		&&		\
-		      raw_cpu_read(hardirqs_enabled)));			\
+		      this_cpu_read(hardirqs_enabled)));		\
 } while (0)
 
 #else