lockdep: add support for queued rwlock
Subsystems (2)
v5 cover lockdep: add support for queued rwlock
v4->v5:
- Add patch 2 to update the locking selftest code to handle recursive read_lock correctly. Patch 1 has no change.
v3->v4:
- Document the new read state and move the conditional compilation code to lockdep.h.
v2->v3:
- Add a new read mode (3) for rwlock (used in lock_acquire_shared_cond_recursive()) to avoid conflict with other use cases of lock_acquire_shared_recursive().
v1->v2:
- Use less conditional & make it easier to read
With the merging of qrwlock into 3.16, it was found that the btrfs filesystem hanged readily. A fix was devised and merged into rc2 and the use of recursive read_lock call was part of the problem.
This patch series addes code to the lockdep subsystem to catch this kind of recursive read_lock calls in kernel code. It also updates the locking selftest to handle recursive read_lock correctly so that it won't complain about test failures whether or not queue rwlock is configured.
Waiman Long (2):
lockdep: restrict the use of recursive read_lock with qrwlock
locking-selftest: Support queue rwlock
include/linux/lockdep.h | 12 ++++++++++++ kernel/locking/lockdep.c | 6 ++++++ lib/locking-selftest.c | 16 +++++++++++++--- 3 files changed, 31 insertions(+), 3 deletions(-)
PATCH v5 1/2 lockdep: restrict the use of recursive read_lock with qrwlock
Unlike the original unfair rwlock implementation, queued rwlock will grant lock according to the chronological sequence of the lock requests except when the lock requester is in the interrupt context. Consequently, recursive read_lock calls will now hang the process if there is a write_lock call somewhere in between the read_lock calls.
This patch updates the lockdep implementation to look for recursive read_lock calls when queued rwlock is being used. A new read state (3) is used to mark those read_lock call that cannot be recursively called except in the interrupt context. The new read state does exhaust the 2 bits available in held_lock:read bit field. The addition of any new read state in the future may require a redesign of how all those bits are squeezed together in the held_lock structure.
--- include/linux/lockdep.h | 12 ++++++++++++ kernel/locking/lockdep.c | 6 ++++++ 2 files changed, 18 insertions(+), 0 deletions(-)
include/linux/lockdep.h +12 -0
@@ -478,16 +478,28 @@ static inline void print_irqtrace_events(struct task_struct *curr)* on the per lock-class debug mode: */ +/* +* Read states in the 2-bit held_lock:read field: +* 0: Exclusive lock +* 1: Shareable lock, cannot be recursively called +* 2: Shareable lock, can be recursively called +* 3: Shareable lock, cannot be recursively called except in interrupt context +*/ #define lock_acquire_exclusive (l, s, t, n, i) lock_acquire(l, s, t, 0, 1, n, i)#define lock_acquire_shared (l, s, t, n, i) lock_acquire(l, s, t, 1, 1, n, i)#define lock_acquire_shared_recursive (l, s, t, n, i) lock_acquire(l, s, t, 2, 1, n, i) +#define lock_acquire_shared_irecursive (l, s, t, n, i) lock_acquire(l, s, t, 3, 1, n, i)#define spin_acquire (l, s, t, i) lock_acquire_exclusive(l, s, t, NULL, i)#define spin_acquire_nest (l, s, t, n, i) lock_acquire_exclusive(l, s, t, n, i)#define spin_release (l, n, i) lock_release(l, n, i)#define rwlock_acquire (l, s, t, i) lock_acquire_exclusive(l, s, t, NULL, i) +#ifdef CONFIG_QUEUE_RWLOCK +#define rwlock_acquire_read (l, s, t, i) lock_acquire_shared_irecursive(l, s, t, NULL, i) +#else #define rwlock_acquire_read (l, s, t, i) lock_acquire_shared_recursive(l, s, t, NULL, i) +#endif #define rwlock_release (l, n, i) lock_release(l, n, i)#define seqcount_acquire (l, s, t, i) lock_acquire_exclusive(l, s, t, NULL, i)
kernel/locking/lockdep.c +6 -0
@@ -1774,6 +1774,12 @@ check_deadlock(struct task_struct *curr, struct held_lock *next,return 2 ;/* +* Recursive read-lock allowed only in interrupt context +*/ +if ((read ==3 ) && prev->read &&in_interrupt ()) +return 2 ; + +/* * We're holding the nest_lock, which serializes this lock's * nesting behaviour. */
Signature · 1 line
1.7.1
tip-bot for Waiman Long
merge bot[tip:locking/core] locking/lockdep: Restrict the use of recursive read_lock() with qrwlock
locking/lockdep: Restrict the use of recursive read_lock() with qrwlock
Commit-ID: e0645a111cb44e01adc6bfff34f683323863f4d2 Gitweb: http://git.kernel.org/tip/e0645a111cb44e01adc6bfff34f683323863f4d2 Author: Waiman Long <[email protected]> AuthorDate: Thu, 26 Jun 2014 13:39:10 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Thu, 17 Jul 2014 12:32:52 +0200
locking/lockdep: Restrict the use of recursive read_lock() with qrwlock
Unlike the original unfair rwlock implementation, queued rwlock will grant lock according to the chronological sequence of the lock requests except when the lock requester is in the interrupt context. Consequently, recursive read_lock calls will now hang the process if there is a write_lock call somewhere in between the read_lock calls.
This patch updates the lockdep implementation to look for recursive read_lock calls when queued rwlock is being used. A new read state (3) is used to mark those read_lock call that cannot be recursively called except in the interrupt context. The new read state does exhaust the 2 bits available in held_lock:read bit field. The addition of any new read state in the future may require a redesign of how all those bits are squeezed together in the held_lock structure.
--- include/linux/lockdep.h | 10 +++++++++- kernel/locking/lockdep.c | 6 ++++++ 2 files changed, 15 insertions(+), 1 deletion(-)
include/linux/lockdep.h +9 -1
@@ -478,16 +478,24 @@ static inline void print_irqtrace_events(struct task_struct *curr)* on the per lock-class debug mode: */ +/* +* Read states in the 2-bit held_lock:read field: +* 0: Exclusive lock +* 1: Shareable lock, cannot be recursively called +* 2: Shareable lock, can be recursively called +* 3: Shareable lock, cannot be recursively called except in interrupt context +*/ #define lock_acquire_exclusive (l, s, t, n, i) lock_acquire(l, s, t, 0, 1, n, i)#define lock_acquire_shared (l, s, t, n, i) lock_acquire(l, s, t, 1, 1, n, i)#define lock_acquire_shared_recursive (l, s, t, n, i) lock_acquire(l, s, t, 2, 1, n, i) +#define lock_acquire_shared_irecursive (l, s, t, n, i) lock_acquire(l, s, t, 3, 1, n, i)#define spin_acquire (l, s, t, i) lock_acquire_exclusive(l, s, t, NULL, i)#define spin_acquire_nest (l, s, t, n, i) lock_acquire_exclusive(l, s, t, n, i)#define spin_release (l, n, i) lock_release(l, n, 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) lock_acquire_shared_irecursive(l, s, t, NULL, i)#define rwlock_release (l, n, i) lock_release(l, n, i)#define seqcount_acquire (l, s, t, i) lock_acquire_exclusive(l, s, t, NULL, i)
kernel/locking/lockdep.c +6 -0
@@ -1776,6 +1776,12 @@ check_deadlock(struct task_struct *curr, struct held_lock *next,return 2 ;/* +* Recursive read-lock allowed only in interrupt context +*/ +if ((read ==3 ) && prev->read &&in_interrupt ()) +return 2 ; + +/* * We're holding the nest_lock, which serializes this lock's * nesting behaviour. */
PATCH v5 2/2 locking-selftest: Support queue rwlock
In reply to Waiman LongThe queue rwlock does not support the use of recursive read-lock in the process context. With changes in the lockdep code to check and disallow recursive read-lock when queue rwlock is configured, it is also necessary for the locking selftest to be updated to change the process context recursive read locking results from SUCCESS to FAILURE for queue rwlock.
--- lib/locking-selftest.c | 16 +++++++++++++--- 1 files changed, 13 insertions(+), 3 deletions(-)
lib/locking-selftest.c +13 -3
@@ -940,6 +940,16 @@ GENERATE_PERMUTATIONS_3_EVENTS(irq_read_recursion_soft) init_rwsem(&rwsem_##x); \ } while (0) +/* +* If queue rwlock is used, recursive read-lock is not allowed in the +* process context. It is allowed in the interrupt context. +*/ +#ifdef CONFIG_QUEUE_RWLOCK +#define RRLOCK_RESULT FAILURE +#else +#define RRLOCK_RESULT SUCCESS +#endif +static void reset_locks (void ) {local_irq_disable ();
@@ -1069,7 +1079,7 @@ static inline void print_testname(const char *testname)print_testname (desc); \ dotest(name##_spin, FAILURE, LOCKTYPE_SPIN); \ dotest(name##_wlock, FAILURE, LOCKTYPE_RWLOCK); \ - dotest(name##_rlock, SUCCESS, LOCKTYPE_RWLOCK); \ + dotest(name##_rlock, RRLOCK_RESULT, LOCKTYPE_RWLOCK); \ dotest(name##_mutex, FAILURE, LOCKTYPE_MUTEX); \ dotest(name##_wsem, FAILURE, LOCKTYPE_RWSEM); \ dotest(name##_rsem, FAILURE, LOCKTYPE_RWSEM); \
@@ -1830,14 +1840,14 @@ void locking_selftest(void)printk (" --------------------------------------------------------------------------\n" );print_testname ("recursive read-lock" );printk (" |" ); -dotest (rlock_AA1,SUCCESS ,LOCKTYPE_RWLOCK ); +dotest (rlock_AA1,RRLOCK_RESULT ,LOCKTYPE_RWLOCK );printk (" |" );dotest (rsem_AA1,FAILURE ,LOCKTYPE_RWSEM );printk ("\n" );print_testname ("recursive read-lock #2" );printk (" |" ); -dotest (rlock_AA1B,SUCCESS ,LOCKTYPE_RWLOCK ); +dotest (rlock_AA1B,RRLOCK_RESULT ,LOCKTYPE_RWLOCK );printk (" |" );dotest (rsem_AA1B,FAILURE ,LOCKTYPE_RWSEM );printk ("\n" );
Signature · 1 line
1.7.1
tip-bot for Waiman Long
merge bot[tip:locking/core] locking/selftest: Support queued rwlock
locking/selftest: Support queued rwlock
Commit-ID: 586fefe5bbdc931fb0725b850f7002f6d71a1aa3 Gitweb: http://git.kernel.org/tip/586fefe5bbdc931fb0725b850f7002f6d71a1aa3 Author: Waiman Long <[email protected]> AuthorDate: Thu, 26 Jun 2014 13:39:11 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Thu, 17 Jul 2014 12:32:53 +0200
locking/selftest: Support queued rwlock
The queued rwlock does not support the use of recursive read-lock in the process context. With changes in the lockdep code to check and disallow recursive read-lock when queued rwlock is configured, it is also necessary for the locking selftest to be updated to change the process context recursive read locking results from SUCCESS to FAILURE for queued rwlock.
--- lib/locking-selftest.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
lib/locking-selftest.c +3 -3
@@ -1069,7 +1069,7 @@ static inline void print_testname(const char *testname)print_testname (desc); \ dotest(name##_spin, FAILURE, LOCKTYPE_SPIN); \ dotest(name##_wlock, FAILURE, LOCKTYPE_RWLOCK); \ - dotest(name##_rlock, SUCCESS, LOCKTYPE_RWLOCK); \ + dotest(name##_rlock, FAILURE, LOCKTYPE_RWLOCK); \ dotest(name##_mutex, FAILURE, LOCKTYPE_MUTEX); \ dotest(name##_wsem, FAILURE, LOCKTYPE_RWSEM); \ dotest(name##_rsem, FAILURE, LOCKTYPE_RWSEM); \
@@ -1830,14 +1830,14 @@ void locking_selftest(void)printk (" --------------------------------------------------------------------------\n" );print_testname ("recursive read-lock" );printk (" |" ); -dotest (rlock_AA1,SUCCESS ,LOCKTYPE_RWLOCK ); +dotest (rlock_AA1,FAILURE ,LOCKTYPE_RWLOCK );printk (" |" );dotest (rsem_AA1,FAILURE ,LOCKTYPE_RWSEM );printk ("\n" );print_testname ("recursive read-lock #2" );printk (" |" ); -dotest (rlock_AA1B,SUCCESS ,LOCKTYPE_RWLOCK ); +dotest (rlock_AA1B,FAILURE ,LOCKTYPE_RWLOCK );printk (" |" );dotest (rsem_AA1B,FAILURE ,LOCKTYPE_RWSEM );printk ("\n" );
Peter Zijlstra
maintainerQuoted 3 lines · Waiman Long
v4->v5:
- Add patch 2 to update the locking selftest code to handle recursiveread_lockcorrectly. Patch 1 has no change.
I removed all CONFIG_QUEUE_RWLOCK dependencies and made lockdep unconditionally assume the stronger constraints.
Since we want all code 'clean' for the strongest possible implementation, everybody should run with those semantics, it doesn't make sense to have that configurable.
Eg. someone on (say ARM, which doesn't -- yet -- have QUEUE_RWLOCK) could unwittingly introduce faulty code.
Quoted 7 lines · Peter Zijlstra
Quoted 3 lines · Waiman Long
v4->v5:
- Add patch 2 to update the locking selftest code to handle recursiveread_lockcorrectly. Patch 1 has no change.I removed all
CONFIG_QUEUE_RWLOCKdependencies and made lockdep unconditionally assume the stronger constraints.
Quoted 6 lines
Since we want all code 'clean' for the strongest possible implementation, everybody should run with those semantics, it doesn't make sense to have that configurable.
Eg. someone on (say ARM, which doesn't -- yet -- have
QUEUE_RWLOCK) could unwittingly introduce faulty code.
I think this is a better way to go. I didn't take this way in my patch for fear that I may be pushing too much.
-Longman