From cb65537ee1134d3cc55c1fa83952bc8eb1212833 Mon Sep 17 00:00:00 2001 From: David Howells Date: Fri, 10 May 2013 19:50:26 +0100 Subject: Add wait_on_atomic_t() and wake_up_atomic_t() Add wait_on_atomic_t() and wake_up_atomic_t() to indicate became-zero events on atomic_t types. This uses the bit-wake waitqueue table. The key is set to a value outside of the number of bits in a long so that wait_on_bit() won't be woken up accidentally. What I'm using this for is: in a following patch I add a counter to struct fscache_cookie to count the number of outstanding operations that need access to netfs data. The way this works is: (1) When a cookie is allocated, the counter is initialised to 1. (2) When an operation wants to access netfs data, it calls atomic_inc_unless() to increment the counter before it does so. If it was 0, then the counter isn't incremented, the operation isn't permitted to access the netfs data (which might by this point no longer exist) and the operation aborts in some appropriate manner. (3) When an operation finishes with the netfs data, it decrements the counter and if it reaches 0, calls wake_up_atomic_t() on it - the assumption being that it was the last blocker. (4) When a cookie is released, the counter is decremented and the releaser uses wait_on_atomic_t() to wait for the counter to become 0 - which should indicate no one is using the netfs data any longer. The netfs data can then be destroyed. There are some alternatives that I have thought of and that have been suggested by Tejun Heo: (A) Using wait_on_bit() to wait on a bit in the counter. This doesn't work because if that bit happens to be 0 then the wait won't happen - even if the counter is non-zero. (B) Using wait_on_bit() to wait on a flag elsewhere which is cleared when the counter reaches 0. Such a flag would be redundant and would add complexity. (C) Adding a waitqueue to fscache_cookie - this would expand that struct by several words for an event that happens just once in each cookie's lifetime. Further, cookies are generally per-file so there are likely to be a lot of them. (D) Similar to (C), but add a pointer to a waitqueue in the cookie instead of a waitqueue. This would add single word per cookie and so would be less of an expansion - but still an expansion. (E) Adding a static waitqueue to the fscache module. Generally this would be fine, but under certain circumstances many cookies will all get added at the same time (eg. NFS umount, cache withdrawal) thereby presenting scaling issues. Note that the wait may be significant as disk I/O may be in progress. So, I think reusing the wait_on_bit() waitqueue set is reasonable. I don't make much use of the waitqueue I need on a per-cookie basis, but sometimes I have a huge flood of the cookies to deal with. I also don't want to add a whole new set of global waitqueue tables specifically for the dec-to-0 event if I can reuse the bit tables. Signed-off-by: David Howells Tested-By: Milosz Tanski Acked-by: Jeff Layton --- include/linux/wait.h | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) (limited to 'include/linux/wait.h') diff --git a/include/linux/wait.h b/include/linux/wait.h index ac38be2692d8..5bacfc4b336d 100644 --- a/include/linux/wait.h +++ b/include/linux/wait.h @@ -23,6 +23,7 @@ struct __wait_queue { struct wait_bit_key { void *flags; int bit_nr; +#define WAIT_ATOMIC_T_BIT_NR -1 }; struct wait_bit_queue { @@ -60,6 +61,9 @@ struct task_struct; #define __WAIT_BIT_KEY_INITIALIZER(word, bit) \ { .flags = word, .bit_nr = bit, } +#define __WAIT_ATOMIC_T_KEY_INITIALIZER(p) \ + { .flags = p, .bit_nr = WAIT_ATOMIC_T_BIT_NR, } + extern void __init_waitqueue_head(wait_queue_head_t *q, const char *name, struct lock_class_key *); #define init_waitqueue_head(q) \ @@ -146,8 +150,10 @@ void __wake_up_bit(wait_queue_head_t *, void *, int); int __wait_on_bit(wait_queue_head_t *, struct wait_bit_queue *, int (*)(void *), unsigned); int __wait_on_bit_lock(wait_queue_head_t *, struct wait_bit_queue *, int (*)(void *), unsigned); void wake_up_bit(void *, int); +void wake_up_atomic_t(atomic_t *); int out_of_line_wait_on_bit(void *, int, int (*)(void *), unsigned); int out_of_line_wait_on_bit_lock(void *, int, int (*)(void *), unsigned); +int out_of_line_wait_on_atomic_t(atomic_t *, int (*)(atomic_t *), unsigned); wait_queue_head_t *bit_waitqueue(void *, int); #define wake_up(x) __wake_up(x, TASK_NORMAL, 1, NULL) @@ -896,5 +902,23 @@ static inline int wait_on_bit_lock(void *word, int bit, return 0; return out_of_line_wait_on_bit_lock(word, bit, action, mode); } + +/** + * wait_on_atomic_t - Wait for an atomic_t to become 0 + * @val: The atomic value being waited on, a kernel virtual address + * @action: the function used to sleep, which may take special actions + * @mode: the task state to sleep in + * + * Wait for an atomic_t to become 0. We abuse the bit-wait waitqueue table for + * the purpose of getting a waitqueue, but we set the key to a bit number + * outside of the target 'word'. + */ +static inline +int wait_on_atomic_t(atomic_t *val, int (*action)(atomic_t *), unsigned mode) +{ + if (atomic_read(val) == 0) + return 0; + return out_of_line_wait_on_atomic_t(val, action, mode); +} #endif -- cgit v1.2.3 From d79ff142624e1be080ad8d09101f7004d79c36e1 Mon Sep 17 00:00:00 2001 From: Martin Peschke Date: Thu, 22 Aug 2013 17:45:36 +0200 Subject: [SCSI] zfcp: fix lock imbalance by reworking request queue locking This patch adds wait_event_interruptible_lock_irq_timeout(), which is a straight-forward descendant of wait_event_interruptible_timeout() and wait_event_interruptible_lock_irq(). The zfcp driver used to call wait_event_interruptible_timeout() in combination with some intricate and error-prone locking. Using wait_event_interruptible_lock_irq_timeout() as a replacement nicely cleans up that locking. This rework removes a situation that resulted in a locking imbalance in zfcp_qdio_sbal_get(): BUG: workqueue leaked lock or atomic: events/1/0xffffff00/10 last function: zfcp_fc_wka_port_offline+0x0/0xa0 [zfcp] It was introduced by commit c2af7545aaff3495d9bf9a7608c52f0af86fb194 "[SCSI] zfcp: Do not wait for SBALs on stopped queue", which had a new code path related to ZFCP_STATUS_ADAPTER_QDIOUP that took an early exit without a required lock being held. The problem occured when a special, non-SCSI I/O request was being submitted in process context, when the adapter's queues had been torn down. In this case the bug surfaced when the Fibre Channel port connection for a well-known address was closed during a concurrent adapter shut-down procedure, which is a rare constellation. This patch also fixes these warnings from the sparse tool (make C=1): drivers/s390/scsi/zfcp_qdio.c:224:12: warning: context imbalance in 'zfcp_qdio_sbal_check' - wrong count at exit drivers/s390/scsi/zfcp_qdio.c:244:5: warning: context imbalance in 'zfcp_qdio_sbal_get' - unexpected unlock Last but not least, we get rid of that crappy lock-unlock-lock sequence at the beginning of the critical section. It is okay to call zfcp_erp_adapter_reopen() with req_q_lock held. Reported-by: Mikulas Patocka Reported-by: Heiko Carstens Signed-off-by: Martin Peschke Cc: stable@vger.kernel.org #2.6.35+ Signed-off-by: Steffen Maier Signed-off-by: James Bottomley --- drivers/s390/scsi/zfcp_qdio.c | 8 ++---- include/linux/wait.h | 57 +++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 59 insertions(+), 6 deletions(-) (limited to 'include/linux/wait.h') diff --git a/drivers/s390/scsi/zfcp_qdio.c b/drivers/s390/scsi/zfcp_qdio.c index 665e3cfaaf85..de0598eaacd2 100644 --- a/drivers/s390/scsi/zfcp_qdio.c +++ b/drivers/s390/scsi/zfcp_qdio.c @@ -224,11 +224,9 @@ int zfcp_qdio_sbals_from_sg(struct zfcp_qdio *qdio, struct zfcp_qdio_req *q_req, static int zfcp_qdio_sbal_check(struct zfcp_qdio *qdio) { - spin_lock_irq(&qdio->req_q_lock); if (atomic_read(&qdio->req_q_free) || !(atomic_read(&qdio->adapter->status) & ZFCP_STATUS_ADAPTER_QDIOUP)) return 1; - spin_unlock_irq(&qdio->req_q_lock); return 0; } @@ -246,9 +244,8 @@ int zfcp_qdio_sbal_get(struct zfcp_qdio *qdio) { long ret; - spin_unlock_irq(&qdio->req_q_lock); - ret = wait_event_interruptible_timeout(qdio->req_q_wq, - zfcp_qdio_sbal_check(qdio), 5 * HZ); + ret = wait_event_interruptible_lock_irq_timeout(qdio->req_q_wq, + zfcp_qdio_sbal_check(qdio), qdio->req_q_lock, 5 * HZ); if (!(atomic_read(&qdio->adapter->status) & ZFCP_STATUS_ADAPTER_QDIOUP)) return -EIO; @@ -262,7 +259,6 @@ int zfcp_qdio_sbal_get(struct zfcp_qdio *qdio) zfcp_erp_adapter_reopen(qdio->adapter, 0, "qdsbg_1"); } - spin_lock_irq(&qdio->req_q_lock); return -EIO; } diff --git a/include/linux/wait.h b/include/linux/wait.h index f487a4750b7f..a67fc1635592 100644 --- a/include/linux/wait.h +++ b/include/linux/wait.h @@ -811,6 +811,63 @@ do { \ __ret; \ }) +#define __wait_event_interruptible_lock_irq_timeout(wq, condition, \ + lock, ret) \ +do { \ + DEFINE_WAIT(__wait); \ + \ + for (;;) { \ + prepare_to_wait(&wq, &__wait, TASK_INTERRUPTIBLE); \ + if (condition) \ + break; \ + if (signal_pending(current)) { \ + ret = -ERESTARTSYS; \ + break; \ + } \ + spin_unlock_irq(&lock); \ + ret = schedule_timeout(ret); \ + spin_lock_irq(&lock); \ + if (!ret) \ + break; \ + } \ + finish_wait(&wq, &__wait); \ +} while (0) + +/** + * wait_event_interruptible_lock_irq_timeout - sleep until a condition gets true or a timeout elapses. + * The condition is checked under the lock. This is expected + * to be called with the lock taken. + * @wq: the waitqueue to wait on + * @condition: a C expression for the event to wait for + * @lock: a locked spinlock_t, which will be released before schedule() + * and reacquired afterwards. + * @timeout: timeout, in jiffies + * + * The process is put to sleep (TASK_INTERRUPTIBLE) until the + * @condition evaluates to true or signal is received. The @condition is + * checked each time the waitqueue @wq is woken up. + * + * wake_up() has to be called after changing any variable that could + * change the result of the wait condition. + * + * This is supposed to be called while holding the lock. The lock is + * dropped before going to sleep and is reacquired afterwards. + * + * The function returns 0 if the @timeout elapsed, -ERESTARTSYS if it + * was interrupted by a signal, and the remaining jiffies otherwise + * if the condition evaluated to true before the timeout elapsed. + */ +#define wait_event_interruptible_lock_irq_timeout(wq, condition, lock, \ + timeout) \ +({ \ + int __ret = timeout; \ + \ + if (!(condition)) \ + __wait_event_interruptible_lock_irq_timeout( \ + wq, condition, lock, __ret); \ + __ret; \ +}) + /* * These are the old interfaces to sleep waiting for an event. -- cgit v1.2.3