diff --git a/include/linux/workqueue.h b/include/linux/workqueue.h index d85c85a57064f4462fafdfbb9fc5d6339cbeb35a..35eee10263c27de32953b1fb9b4288408df834b1 100644 --- a/include/linux/workqueue.h +++ b/include/linux/workqueue.h @@ -30,16 +30,16 @@ void delayed_work_timer_fn(struct timer_list *t); enum { WORK_STRUCT_PENDING_BIT = 0, /* work item is pending execution */ - WORK_STRUCT_INACTIVE_BIT= 1, /* work item is inactive */ - WORK_STRUCT_PWQ_BIT = 2, /* data points to pwq */ - WORK_STRUCT_LINKED_BIT = 3, /* next work is linked to this one */ + WORK_STRUCT_INACTIVE_BIT, /* work item is inactive */ + WORK_STRUCT_PWQ_BIT, /* data points to pwq */ + WORK_STRUCT_LINKED_BIT, /* next work is linked to this one */ #ifdef CONFIG_DEBUG_OBJECTS_WORK - WORK_STRUCT_STATIC_BIT = 4, /* static initializer (debugobjects) */ - WORK_STRUCT_COLOR_SHIFT = 5, /* color for workqueue flushing */ -#else - WORK_STRUCT_COLOR_SHIFT = 4, /* color for workqueue flushing */ + WORK_STRUCT_STATIC_BIT, /* static initializer (debugobjects) */ #endif + WORK_STRUCT_FLAG_BITS, + /* color for workqueue flushing */ + WORK_STRUCT_COLOR_SHIFT = WORK_STRUCT_FLAG_BITS, WORK_STRUCT_COLOR_BITS = 4, WORK_STRUCT_PENDING = 1 << WORK_STRUCT_PENDING_BIT, @@ -58,25 +58,37 @@ enum { WORK_CPU_UNBOUND = NR_CPUS, /* - * Reserve 8 bits off of pwq pointer w/ debugobjects turned off. - * This makes pwqs aligned to 256 bytes and allows 16 workqueue - * flush colors. + * When WORK_STRUCT_PWQ is set, reserve 8 bits off of pwq pointer w/ + * debugobjects turned off. This makes pwqs aligned to 256 bytes (512 + * bytes w/ DEBUG_OBJECTS_WORK) and allows 16 workqueue flush colors. + * + * MSB + * [ pwq pointer ] [ flush color ] [ STRUCT flags ] + * 4 bits 4 or 5 bits */ - WORK_STRUCT_FLAG_BITS = WORK_STRUCT_COLOR_SHIFT + - WORK_STRUCT_COLOR_BITS, + WORK_STRUCT_PWQ_SHIFT = WORK_STRUCT_COLOR_SHIFT + WORK_STRUCT_COLOR_BITS, - /* data contains off-queue information when !WORK_STRUCT_PWQ */ - WORK_OFFQ_FLAG_BASE = WORK_STRUCT_COLOR_SHIFT, + /* + * data contains off-queue information when !WORK_STRUCT_PWQ. + * + * MSB + * [ pool ID ] [ disable depth ] [ OFFQ flags ] [ STRUCT flags ] + * 16 bits 1 bit 4 or 5 bits + */ + WORK_OFFQ_FLAG_SHIFT = WORK_STRUCT_FLAG_BITS, + WORK_OFFQ_CANCELING_BIT = WORK_OFFQ_FLAG_SHIFT, + WORK_OFFQ_FLAG_END, + WORK_OFFQ_FLAG_BITS = WORK_OFFQ_FLAG_END - WORK_OFFQ_FLAG_SHIFT, - __WORK_OFFQ_CANCELING = WORK_OFFQ_FLAG_BASE, + WORK_OFFQ_DISABLE_SHIFT = WORK_OFFQ_FLAG_SHIFT + WORK_OFFQ_FLAG_BITS, + WORK_OFFQ_DISABLE_BITS = 16, /* - * When a work item is off queue, its high bits point to the last - * pool it was on. Cap at 31 bits and use the highest number to - * indicate that no pool is associated. + * When a work item is off queue, the high bits encode off-queue flags + * and the last pool it was on. Cap pool ID to 31 bits and use the + * highest number to indicate that no pool is associated. */ - WORK_OFFQ_FLAG_BITS = 1, - WORK_OFFQ_POOL_SHIFT = WORK_OFFQ_FLAG_BASE + WORK_OFFQ_FLAG_BITS, + WORK_OFFQ_POOL_SHIFT = WORK_OFFQ_DISABLE_SHIFT + WORK_OFFQ_DISABLE_BITS, WORK_OFFQ_LEFT = BITS_PER_LONG - WORK_OFFQ_POOL_SHIFT, WORK_OFFQ_POOL_BITS = WORK_OFFQ_LEFT <= 31 ? WORK_OFFQ_LEFT : 31, @@ -89,12 +101,12 @@ enum { }; /* Convenience constants - of type 'unsigned long', not 'enum'! */ -#define WORK_OFFQ_CANCELING (1ul << __WORK_OFFQ_CANCELING) +#define WORK_OFFQ_CANCELING (1ul << WORK_OFFQ_CANCELING_BIT) +#define WORK_OFFQ_FLAG_MASK (((1ul << WORK_OFFQ_FLAG_BITS) - 1) << WORK_OFFQ_FLAG_SHIFT) +#define WORK_OFFQ_DISABLE_MASK (((1ul << WORK_OFFQ_DISABLE_BITS) - 1) << WORK_OFFQ_DISABLE_SHIFT) #define WORK_OFFQ_POOL_NONE ((1ul << WORK_OFFQ_POOL_BITS) - 1) #define WORK_STRUCT_NO_POOL (WORK_OFFQ_POOL_NONE << WORK_OFFQ_POOL_SHIFT) - -#define WORK_STRUCT_FLAG_MASK ((1ul << WORK_STRUCT_FLAG_BITS) - 1) -#define WORK_STRUCT_WQ_DATA_MASK (~WORK_STRUCT_FLAG_MASK) +#define WORK_STRUCT_PWQ_MASK (~((1ul << WORK_STRUCT_PWQ_SHIFT) - 1)) struct work_struct { atomic_long_t data; @@ -533,6 +545,14 @@ extern bool flush_delayed_work(struct delayed_work *dwork); extern bool cancel_delayed_work(struct delayed_work *dwork); extern bool cancel_delayed_work_sync(struct delayed_work *dwork); +extern bool disable_work(struct work_struct *work); +extern bool disable_work_sync(struct work_struct *work); +extern bool enable_work(struct work_struct *work); + +extern bool disable_delayed_work(struct delayed_work *dwork); +extern bool disable_delayed_work_sync(struct delayed_work *dwork); +extern bool enable_delayed_work(struct delayed_work *dwork); + extern bool flush_rcu_work(struct rcu_work *rwork); extern void workqueue_set_max_active(struct workqueue_struct *wq, diff --git a/kernel/workqueue.c b/kernel/workqueue.c index 5aa75b8233486485bff6eb2bc24a849b604636b3..dc9cd3b1bb5fd1ce22667873ddae73587b23ebed 100644 --- a/kernel/workqueue.c +++ b/kernel/workqueue.c @@ -111,6 +111,11 @@ enum { WQ_NAME_LEN = 24, }; +enum work_cancel_flags { + WORK_CANCEL_DELAYED = 1 << 0, /* canceling a delayed_work */ + WORK_CANCEL_DISABLE = 1 << 1, /* canceling to disable */ +}; + /* * Structure fields follow one of the following exclusion rules. * @@ -219,7 +224,7 @@ enum pool_workqueue_stats { }; /* - * The per-pool workqueue. While queued, the lower WORK_STRUCT_FLAG_BITS + * The per-pool workqueue. While queued, bits below WORK_PWQ_SHIFT * of work_struct->data are used for flags and the remaining high bits * point to the pwq; thus, pwqs need to be aligned at two's power of the * number of flag bits. @@ -265,7 +270,7 @@ struct pool_workqueue { */ struct kthread_work release_work; struct rcu_head rcu; -} __aligned(1 << WORK_STRUCT_FLAG_BITS); +} __aligned(1 << WORK_STRUCT_PWQ_SHIFT); /* * Structure used to wait for workqueue flush. @@ -341,6 +346,12 @@ struct wq_pod_type { static struct wq_pod_type wq_pod_types[WQ_AFFN_NR_TYPES]; static enum wq_affn_scope wq_affn_dfl = WQ_AFFN_CACHE; +struct work_offq_data { + u32 pool_id; + u32 disable; + u32 flags; +}; + static const char *wq_affn_names[WQ_AFFN_NR_TYPES] = { [WQ_AFFN_DFL] = "default", [WQ_AFFN_CPU] = "cpu", @@ -413,6 +424,12 @@ static struct workqueue_attrs *unbound_std_wq_attrs[NR_STD_WORKER_POOLS]; /* I: attributes used when instantiating ordered pools on demand */ static struct workqueue_attrs *ordered_wq_attrs[NR_STD_WORKER_POOLS]; +/* + * Used to synchronize multiple cancel_sync attempts on the same work item. See + * work_grab_pending() and __cancel_work_sync(). + */ +static DECLARE_WAIT_QUEUE_HEAD(wq_cancel_waitq); + /* * I: kthread_worker to release pwq's. pwq release needs to be bounced to a * process context while holding a pool lock. Bounce to a dedicated kthread @@ -647,10 +664,9 @@ static int work_next_color(int color) * contain the pointer to the queued pwq. Once execution starts, the flag * is cleared and the high bits contain OFFQ flags and pool ID. * - * set_work_pwq(), set_work_pool_and_clear_pending(), mark_work_canceling() - * and clear_work_data() can be used to set the pwq, pool or clear - * work->data. These functions should only be called while the work is - * owned - ie. while the PENDING bit is set. + * set_work_pwq(), set_work_pool_and_clear_pending() and mark_work_canceling() + * can be used to set the pwq, pool or clear work->data. These functions should + * only be called while the work is owned - ie. while the PENDING bit is set. * * get_work_pool() and get_work_pwq() can be used to obtain the pool or pwq * corresponding to a work. Pool is available once the work has been @@ -662,29 +678,28 @@ static int work_next_color(int color) * but stay off timer and worklist for arbitrarily long and nobody should * try to steal the PENDING bit. */ -static inline void set_work_data(struct work_struct *work, unsigned long data, - unsigned long flags) +static inline void set_work_data(struct work_struct *work, unsigned long data) { WARN_ON_ONCE(!work_pending(work)); - atomic_long_set(&work->data, data | flags | work_static(work)); + atomic_long_set(&work->data, data | work_static(work)); } static void set_work_pwq(struct work_struct *work, struct pool_workqueue *pwq, - unsigned long extra_flags) + unsigned long flags) { - set_work_data(work, (unsigned long)pwq, - WORK_STRUCT_PENDING | WORK_STRUCT_PWQ | extra_flags); + set_work_data(work, (unsigned long)pwq | WORK_STRUCT_PENDING | + WORK_STRUCT_PWQ | flags); } static void set_work_pool_and_keep_pending(struct work_struct *work, - int pool_id) + int pool_id, unsigned long flags) { - set_work_data(work, (unsigned long)pool_id << WORK_OFFQ_POOL_SHIFT, - WORK_STRUCT_PENDING); + set_work_data(work, ((unsigned long)pool_id << WORK_OFFQ_POOL_SHIFT) | + WORK_STRUCT_PENDING | flags); } static void set_work_pool_and_clear_pending(struct work_struct *work, - int pool_id) + int pool_id, unsigned long flags) { /* * The following wmb is paired with the implied mb in @@ -693,7 +708,8 @@ static void set_work_pool_and_clear_pending(struct work_struct *work, * owner. */ smp_wmb(); - set_work_data(work, (unsigned long)pool_id << WORK_OFFQ_POOL_SHIFT, 0); + set_work_data(work, ((unsigned long)pool_id << WORK_OFFQ_POOL_SHIFT) | + flags); /* * The following mb guarantees that previous clear of a PENDING bit * will not be reordered with any speculative LOADS or STORES from @@ -725,15 +741,9 @@ static void set_work_pool_and_clear_pending(struct work_struct *work, smp_mb(); } -static void clear_work_data(struct work_struct *work) -{ - smp_wmb(); /* see set_work_pool_and_clear_pending() */ - set_work_data(work, WORK_STRUCT_NO_POOL, 0); -} - static inline struct pool_workqueue *work_struct_pwq(unsigned long data) { - return (struct pool_workqueue *)(data & WORK_STRUCT_WQ_DATA_MASK); + return (struct pool_workqueue *)(data & WORK_STRUCT_PWQ_MASK); } static struct pool_workqueue *get_work_pwq(struct work_struct *work) @@ -778,29 +788,26 @@ static struct worker_pool *get_work_pool(struct work_struct *work) return idr_find(&worker_pool_idr, pool_id); } -/** - * get_work_pool_id - return the worker pool ID a given work is associated with - * @work: the work item of interest - * - * Return: The worker_pool ID @work was last associated with. - * %WORK_OFFQ_POOL_NONE if none. - */ -static int get_work_pool_id(struct work_struct *work) +static unsigned long shift_and_mask(unsigned long v, u32 shift, u32 bits) { - unsigned long data = atomic_long_read(&work->data); + return (v >> shift) & ((1U << bits) - 1); +} - if (data & WORK_STRUCT_PWQ) - return work_struct_pwq(data)->pool->id; +static void work_offqd_unpack(struct work_offq_data *offqd, unsigned long data) +{ + WARN_ON_ONCE(data & WORK_STRUCT_PWQ); - return data >> WORK_OFFQ_POOL_SHIFT; + offqd->pool_id = shift_and_mask(data, WORK_OFFQ_POOL_SHIFT, + WORK_OFFQ_POOL_BITS); + offqd->disable = shift_and_mask(data, WORK_OFFQ_DISABLE_SHIFT, + WORK_OFFQ_DISABLE_BITS); + offqd->flags = data & WORK_OFFQ_FLAG_MASK; } -static void mark_work_canceling(struct work_struct *work) +static unsigned long work_offqd_pack_flags(struct work_offq_data *offqd) { - unsigned long pool_id = get_work_pool_id(work); - - pool_id <<= WORK_OFFQ_POOL_SHIFT; - set_work_data(work, pool_id | WORK_OFFQ_CANCELING, WORK_STRUCT_PENDING); + return ((unsigned long)offqd->disable << WORK_OFFQ_DISABLE_SHIFT) | + ((unsigned long)offqd->flags); } static bool work_is_canceling(struct work_struct *work) @@ -1521,7 +1528,7 @@ static void pwq_dec_nr_in_flight(struct pool_workqueue *pwq, unsigned long work_ /** * try_to_grab_pending - steal work item from worklist and disable irq * @work: work item to steal - * @is_dwork: @work is a delayed_work + * @cflags: %WORK_CANCEL_ flags * @flags: place to store irq state * * Try to grab PENDING bit of @work. This function can handle @work in any @@ -1548,7 +1555,7 @@ static void pwq_dec_nr_in_flight(struct pool_workqueue *pwq, unsigned long work_ * * This function is safe to call from any context including IRQ handler. */ -static int try_to_grab_pending(struct work_struct *work, bool is_dwork, +static int try_to_grab_pending(struct work_struct *work, u32 cflags, unsigned long *flags) { struct worker_pool *pool; @@ -1557,7 +1564,7 @@ static int try_to_grab_pending(struct work_struct *work, bool is_dwork, local_irq_save(*flags); /* try to steal the timer if it exists */ - if (is_dwork) { + if (cflags & WORK_CANCEL_DELAYED) { struct delayed_work *dwork = to_delayed_work(work); /* @@ -1613,7 +1620,7 @@ static int try_to_grab_pending(struct work_struct *work, bool is_dwork, pwq_dec_nr_in_flight(pwq, *work_data_bits(work)); /* work->data points to pwq iff queued, point to pool */ - set_work_pool_and_keep_pending(work, pool->id); + set_work_pool_and_keep_pending(work, pool->id, 0); raw_spin_unlock(&pool->lock); rcu_read_unlock(); @@ -1629,6 +1636,75 @@ static int try_to_grab_pending(struct work_struct *work, bool is_dwork, return -EAGAIN; } +struct cwt_wait { + wait_queue_entry_t wait; + struct work_struct *work; +}; + +static int cwt_wakefn(wait_queue_entry_t *wait, unsigned mode, int sync, void *key) +{ + struct cwt_wait *cwait = container_of(wait, struct cwt_wait, wait); + + if (cwait->work != key) + return 0; + return autoremove_wake_function(wait, mode, sync, key); +} + +/** + * work_grab_pending - steal work item from worklist and disable irq + * @work: work item to steal + * @cflags: %WORK_CANCEL_ flags + * @irq_flags: place to store IRQ state + * + * Grab PENDING bit of @work. @work can be in any stable state - idle, on timer + * or on worklist. + * + * Must be called in process context. IRQ is disabled on return with IRQ state + * stored in *@irq_flags. The caller is responsible for re-enabling it using + * local_irq_restore(). + * + * Returns %true if @work was pending. %false if idle. + */ +static bool work_grab_pending(struct work_struct *work, u32 cflags, + unsigned long *irq_flags) +{ + struct cwt_wait cwait; + int ret; + + might_sleep(); +repeat: + ret = try_to_grab_pending(work, cflags, irq_flags); + if (likely(ret >= 0)) + return ret; + if (ret != -ENOENT) + goto repeat; + + /* + * Someone is already canceling. Wait for it to finish. flush_work() + * doesn't work for PREEMPT_NONE because we may get woken up between + * @work's completion and the other canceling task resuming and clearing + * CANCELING - flush_work() will return false immediately as @work is no + * longer busy, try_to_grab_pending() will return -ENOENT as @work is + * still being canceled and the other canceling task won't be able to + * clear CANCELING as we're hogging the CPU. + * + * Let's wait for completion using a waitqueue. As this may lead to the + * thundering herd problem, use a custom wake function which matches + * @work along with exclusive wait and wakeup. + */ + init_wait(&cwait.wait); + cwait.wait.func = cwt_wakefn; + cwait.work = work; + + prepare_to_wait_exclusive(&wq_cancel_waitq, &cwait.wait, + TASK_UNINTERRUPTIBLE); + if (work_is_canceling(work)) + schedule(); + finish_wait(&wq_cancel_waitq, &cwait.wait); + + goto repeat; +} + /** * insert_work - insert a work into a pool * @pwq: pwq @work belongs to @@ -1809,6 +1885,21 @@ static void __queue_work(int cpu, struct workqueue_struct *wq, rcu_read_unlock(); } +static bool clear_pending_if_disabled(struct work_struct *work) +{ + unsigned long data = *work_data_bits(work); + struct work_offq_data offqd; + + if (likely((data & WORK_STRUCT_PWQ) || + !(data & WORK_OFFQ_DISABLE_MASK))) + return false; + + work_offqd_unpack(&offqd, data); + set_work_pool_and_clear_pending(work, offqd.pool_id, + work_offqd_pack_flags(&offqd)); + return true; +} + /** * queue_work_on - queue work on specific cpu * @cpu: CPU number to execute work on @@ -1831,7 +1922,8 @@ bool queue_work_on(int cpu, struct workqueue_struct *wq, local_irq_save(flags); - if (!test_and_set_bit(WORK_STRUCT_PENDING_BIT, work_data_bits(work))) { + if (!test_and_set_bit(WORK_STRUCT_PENDING_BIT, work_data_bits(work)) && + !clear_pending_if_disabled(work)) { __queue_work(cpu, wq, work); ret = true; } @@ -1909,7 +2001,8 @@ bool queue_work_node(int node, struct workqueue_struct *wq, local_irq_save(flags); - if (!test_and_set_bit(WORK_STRUCT_PENDING_BIT, work_data_bits(work))) { + if (!test_and_set_bit(WORK_STRUCT_PENDING_BIT, work_data_bits(work)) && + !clear_pending_if_disabled(work)) { int cpu = select_numa_node_cpu(node); __queue_work(cpu, wq, work); @@ -1983,7 +2076,8 @@ bool queue_delayed_work_on(int cpu, struct workqueue_struct *wq, /* read the comment in __queue_work() */ local_irq_save(flags); - if (!test_and_set_bit(WORK_STRUCT_PENDING_BIT, work_data_bits(work))) { + if (!test_and_set_bit(WORK_STRUCT_PENDING_BIT, work_data_bits(work)) && + !clear_pending_if_disabled(work)) { __queue_delayed_work(cpu, wq, dwork, delay); ret = true; } @@ -2018,7 +2112,8 @@ bool mod_delayed_work_on(int cpu, struct workqueue_struct *wq, int ret; do { - ret = try_to_grab_pending(&dwork->work, true, &flags); + ret = try_to_grab_pending(&dwork->work, WORK_CANCEL_DELAYED, + &flags); } while (unlikely(ret == -EAGAIN)); if (likely(ret >= 0)) { @@ -2055,7 +2150,12 @@ bool queue_rcu_work(struct workqueue_struct *wq, struct rcu_work *rwork) { struct work_struct *work = &rwork->work; - if (!test_and_set_bit(WORK_STRUCT_PENDING_BIT, work_data_bits(work))) { + /* + * rcu_work can't be canceled or disabled. Warn if the user reached + * inside @rwork and disabled the inner work. + */ + if (!test_and_set_bit(WORK_STRUCT_PENDING_BIT, work_data_bits(work)) && + !WARN_ON_ONCE(clear_pending_if_disabled(work))) { rwork->wq = wq; call_rcu_hurry(&rwork->rcu, rcu_work_rcufn); return true; @@ -2599,7 +2699,7 @@ __acquires(&pool->lock) * PENDING and queued state changes happen together while IRQ is * disabled. */ - set_work_pool_and_clear_pending(work, pool->id); + set_work_pool_and_clear_pending(work, pool->id, 0); pwq->stats[PWQ_STAT_STARTED]++; raw_spin_unlock_irq(&pool->lock); @@ -3452,61 +3552,37 @@ bool flush_work(struct work_struct *work) } EXPORT_SYMBOL_GPL(flush_work); -struct cwt_wait { - wait_queue_entry_t wait; - struct work_struct *work; -}; - -static int cwt_wakefn(wait_queue_entry_t *wait, unsigned mode, int sync, void *key) +static void work_offqd_disable(struct work_offq_data *offqd) { - struct cwt_wait *cwait = container_of(wait, struct cwt_wait, wait); + const unsigned long max = (1lu << WORK_OFFQ_DISABLE_BITS) - 1; - if (cwait->work != key) - return 0; - return autoremove_wake_function(wait, mode, sync, key); + if (likely(offqd->disable < max)) + offqd->disable++; + else + WARN_ONCE(true, "workqueue: work disable count overflowed\n"); +} + +static void work_offqd_enable(struct work_offq_data *offqd) +{ + if (likely(offqd->disable > 0)) + offqd->disable--; + else + WARN_ONCE(true, "workqueue: work disable count underflowed\n"); } -static bool __cancel_work_timer(struct work_struct *work, bool is_dwork) +static bool __cancel_work_sync(struct work_struct *work, u32 cflags) { - static DECLARE_WAIT_QUEUE_HEAD(cancel_waitq); + struct work_offq_data offqd; unsigned long flags; - int ret; + bool ret; - do { - ret = try_to_grab_pending(work, is_dwork, &flags); - /* - * If someone else is already canceling, wait for it to - * finish. flush_work() doesn't work for PREEMPT_NONE - * because we may get scheduled between @work's completion - * and the other canceling task resuming and clearing - * CANCELING - flush_work() will return false immediately - * as @work is no longer busy, try_to_grab_pending() will - * return -ENOENT as @work is still being canceled and the - * other canceling task won't be able to clear CANCELING as - * we're hogging the CPU. - * - * Let's wait for completion using a waitqueue. As this - * may lead to the thundering herd problem, use a custom - * wake function which matches @work along with exclusive - * wait and wakeup. - */ - if (unlikely(ret == -ENOENT)) { - struct cwt_wait cwait; - - init_wait(&cwait.wait); - cwait.wait.func = cwt_wakefn; - cwait.work = work; - - prepare_to_wait_exclusive(&cancel_waitq, &cwait.wait, - TASK_UNINTERRUPTIBLE); - if (work_is_canceling(work)) - schedule(); - finish_wait(&cancel_waitq, &cwait.wait); - } - } while (unlikely(ret < 0)); + /* claim @work and tell other tasks trying to grab @work to back off */ + ret = work_grab_pending(work, cflags, &flags); - /* tell other tasks trying to grab @work to back off */ - mark_work_canceling(work); + work_offqd_unpack(&offqd, *work_data_bits(work)); + offqd.flags |= WORK_OFFQ_CANCELING; + set_work_pool_and_keep_pending(work, offqd.pool_id, + work_offqd_pack_flags(&offqd)); local_irq_restore(flags); /* @@ -3516,16 +3592,22 @@ static bool __cancel_work_timer(struct work_struct *work, bool is_dwork) if (wq_online) __flush_work(work, true); - clear_work_data(work); + work_offqd_unpack(&offqd, *work_data_bits(work)); + + if (cflags & WORK_CANCEL_DISABLE) + work_offqd_disable(&offqd); /* - * Paired with prepare_to_wait() above so that either - * waitqueue_active() is visible here or !work_is_canceling() is - * visible there. + * smp_mb() at the end of set_work_pool_and_clear_pending() is paired + * with prepare_to_wait() above so that either waitqueue_active() is + * visible here or !work_is_canceling() is visible there. */ - smp_mb(); - if (waitqueue_active(&cancel_waitq)) - __wake_up(&cancel_waitq, TASK_NORMAL, 1, work); + offqd.flags &= ~WORK_OFFQ_CANCELING; + set_work_pool_and_clear_pending(work, WORK_OFFQ_POOL_NONE, + work_offqd_pack_flags(&offqd)); + + if (waitqueue_active(&wq_cancel_waitq)) + __wake_up(&wq_cancel_waitq, TASK_NORMAL, 1, work); return ret; } @@ -3550,7 +3632,7 @@ static bool __cancel_work_timer(struct work_struct *work, bool is_dwork) */ bool cancel_work_sync(struct work_struct *work) { - return __cancel_work_timer(work, false); + return __cancel_work_sync(work, 0); } EXPORT_SYMBOL_GPL(cancel_work_sync); @@ -3596,19 +3678,30 @@ bool flush_rcu_work(struct rcu_work *rwork) } EXPORT_SYMBOL(flush_rcu_work); -static bool __cancel_work(struct work_struct *work, bool is_dwork) +static bool __cancel_work(struct work_struct *work, u32 cflags) { + struct work_offq_data offqd; unsigned long flags; int ret; - do { - ret = try_to_grab_pending(work, is_dwork, &flags); - } while (unlikely(ret == -EAGAIN)); + if (cflags & WORK_CANCEL_DISABLE) { + ret = work_grab_pending(work, cflags, &flags); + } else { + do { + ret = try_to_grab_pending(work, cflags, &flags); + } while (unlikely(ret == -EAGAIN)); - if (unlikely(ret < 0)) - return false; + if (unlikely(ret < 0)) + return false; + } + + work_offqd_unpack(&offqd, *work_data_bits(work)); + + if (cflags & WORK_CANCEL_DISABLE) + work_offqd_disable(&offqd); - set_work_pool_and_clear_pending(work, get_work_pool_id(work)); + set_work_pool_and_clear_pending(work, offqd.pool_id, + work_offqd_pack_flags(&offqd)); local_irq_restore(flags); return ret; } @@ -3618,7 +3711,7 @@ static bool __cancel_work(struct work_struct *work, bool is_dwork) */ bool cancel_work(struct work_struct *work) { - return __cancel_work(work, false); + return __cancel_work(work, 0); } EXPORT_SYMBOL(cancel_work); @@ -3640,7 +3733,7 @@ EXPORT_SYMBOL(cancel_work); */ bool cancel_delayed_work(struct delayed_work *dwork) { - return __cancel_work(&dwork->work, true); + return __cancel_work(&dwork->work, WORK_CANCEL_DELAYED); } EXPORT_SYMBOL(cancel_delayed_work); @@ -3655,10 +3748,109 @@ EXPORT_SYMBOL(cancel_delayed_work); */ bool cancel_delayed_work_sync(struct delayed_work *dwork) { - return __cancel_work_timer(&dwork->work, true); + return __cancel_work_sync(&dwork->work, WORK_CANCEL_DELAYED); } EXPORT_SYMBOL(cancel_delayed_work_sync); +/** + * disable_work - Disable and cancel a work item + * @work: work item to disable + * + * Disable @work by incrementing its disable count and cancel it if currently + * pending. As long as the disable count is non-zero, any attempt to queue @work + * will fail and return %false. The maximum supported disable depth is 2 to the + * power of %WORK_OFFQ_DISABLE_BITS, currently 65536. + * + * Must be called from a sleepable context. Returns %true if @work was pending, + * %false otherwise. + */ +bool disable_work(struct work_struct *work) +{ + return __cancel_work(work, WORK_CANCEL_DISABLE); +} +EXPORT_SYMBOL_GPL(disable_work); + +/** + * disable_work_sync - Disable, cancel and drain a work item + * @work: work item to disable + * + * Similar to disable_work() but also wait for @work to finish if currently + * executing. + * + * Must be called from a sleepable context. Returns %true if @work was pending, + * %false otherwise. + */ +bool disable_work_sync(struct work_struct *work) +{ + return __cancel_work_sync(work, WORK_CANCEL_DISABLE); +} +EXPORT_SYMBOL_GPL(disable_work_sync); + +/** + * enable_work - Enable a work item + * @work: work item to enable + * + * Undo disable_work[_sync]() by decrementing @work's disable count. @work can + * only be queued if its disable count is 0. + * + * Must be called from a sleepable context. Returns %true if the disable count + * reached 0. Otherwise, %false. + */ +bool enable_work(struct work_struct *work) +{ + struct work_offq_data offqd; + unsigned long irq_flags; + + work_grab_pending(work, 0, &irq_flags); + + work_offqd_unpack(&offqd, *work_data_bits(work)); + work_offqd_enable(&offqd); + set_work_pool_and_clear_pending(work, offqd.pool_id, + work_offqd_pack_flags(&offqd)); + local_irq_restore(irq_flags); + + return !offqd.disable; +} +EXPORT_SYMBOL_GPL(enable_work); + +/** + * disable_delayed_work - Disable and cancel a delayed work item + * @dwork: delayed work item to disable + * + * disable_work() for delayed work items. + */ +bool disable_delayed_work(struct delayed_work *dwork) +{ + return __cancel_work(&dwork->work, + WORK_CANCEL_DELAYED | WORK_CANCEL_DISABLE); +} +EXPORT_SYMBOL_GPL(disable_delayed_work); + +/** + * disable_delayed_work_sync - Disable, cancel and drain a delayed work item + * @dwork: delayed work item to disable + * + * disable_work_sync() for delayed work items. + */ +bool disable_delayed_work_sync(struct delayed_work *dwork) +{ + return __cancel_work_sync(&dwork->work, + WORK_CANCEL_DELAYED | WORK_CANCEL_DISABLE); +} +EXPORT_SYMBOL_GPL(disable_delayed_work_sync); + +/** + * enable_delayed_work - Enable a delayed work item + * @dwork: delayed work item to enable + * + * enable_work() for delayed work items. + */ +bool enable_delayed_work(struct delayed_work *dwork) +{ + return enable_work(&dwork->work); +} +EXPORT_SYMBOL_GPL(enable_delayed_work); + /** * schedule_on_each_cpu - execute a function synchronously on each online CPU * @func: the function to call @@ -4212,7 +4404,7 @@ static void pwq_adjust_max_active(struct pool_workqueue *pwq) static void init_pwq(struct pool_workqueue *pwq, struct workqueue_struct *wq, struct worker_pool *pool) { - BUG_ON((unsigned long)pwq & WORK_STRUCT_FLAG_MASK); + BUG_ON((unsigned long)pwq & ~WORK_STRUCT_PWQ_MASK); memset(pwq, 0, sizeof(*pwq));