2019-10-23 00:25:58 +08:00
|
|
|
// SPDX-License-Identifier: GPL-2.0
|
|
|
|
/*
|
|
|
|
* Basic worker thread pool for io_uring
|
|
|
|
*
|
|
|
|
* Copyright (C) 2019 Jens Axboe
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/errno.h>
|
|
|
|
#include <linux/sched/signal.h>
|
|
|
|
#include <linux/mm.h>
|
|
|
|
#include <linux/sched/mm.h>
|
|
|
|
#include <linux/percpu.h>
|
|
|
|
#include <linux/slab.h>
|
|
|
|
#include <linux/rculist_nulls.h>
|
2020-10-22 23:02:50 +08:00
|
|
|
#include <linux/cpu.h>
|
2021-02-17 05:15:30 +08:00
|
|
|
#include <linux/tracehook.h>
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2020-10-22 23:02:50 +08:00
|
|
|
#include "../kernel/sched/sched.h"
|
2019-10-23 00:25:58 +08:00
|
|
|
#include "io-wq.h"
|
|
|
|
|
|
|
|
#define WORKER_IDLE_TIMEOUT (5 * HZ)
|
|
|
|
|
|
|
|
enum {
|
|
|
|
IO_WORKER_F_UP = 1, /* up and active */
|
|
|
|
IO_WORKER_F_RUNNING = 2, /* account as running */
|
|
|
|
IO_WORKER_F_FREE = 4, /* worker on free list */
|
2020-09-27 02:37:46 +08:00
|
|
|
IO_WORKER_F_FIXED = 8, /* static idle worker */
|
|
|
|
IO_WORKER_F_BOUND = 16, /* is doing bounded work */
|
2019-10-23 00:25:58 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
enum {
|
|
|
|
IO_WQ_BIT_EXIT = 0, /* wq exiting */
|
2020-12-21 01:47:42 +08:00
|
|
|
IO_WQ_BIT_ERROR = 1, /* error on setup */
|
2019-10-23 00:25:58 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
enum {
|
|
|
|
IO_WQE_FLAG_STALLED = 1, /* stalled on hash */
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* One for each thread in a wqe pool
|
|
|
|
*/
|
|
|
|
struct io_worker {
|
|
|
|
refcount_t ref;
|
|
|
|
unsigned flags;
|
|
|
|
struct hlist_nulls_node nulls_node;
|
2019-11-14 04:54:49 +08:00
|
|
|
struct list_head all_list;
|
2019-10-23 00:25:58 +08:00
|
|
|
struct task_struct *task;
|
|
|
|
struct io_wqe *wqe;
|
2019-11-14 00:43:34 +08:00
|
|
|
|
2019-10-23 00:25:58 +08:00
|
|
|
struct io_wq_work *cur_work;
|
2019-11-14 00:43:34 +08:00
|
|
|
spinlock_t lock;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
|
|
|
struct rcu_head rcu;
|
|
|
|
};
|
|
|
|
|
|
|
|
#if BITS_PER_LONG == 64
|
|
|
|
#define IO_WQ_HASH_ORDER 6
|
|
|
|
#else
|
|
|
|
#define IO_WQ_HASH_ORDER 5
|
|
|
|
#endif
|
|
|
|
|
2020-03-24 03:57:22 +08:00
|
|
|
#define IO_WQ_NR_HASH_BUCKETS (1u << IO_WQ_HASH_ORDER)
|
|
|
|
|
2019-11-08 02:41:16 +08:00
|
|
|
struct io_wqe_acct {
|
|
|
|
unsigned nr_workers;
|
|
|
|
unsigned max_workers;
|
|
|
|
atomic_t nr_running;
|
|
|
|
};
|
|
|
|
|
|
|
|
enum {
|
|
|
|
IO_WQ_ACCT_BOUND,
|
|
|
|
IO_WQ_ACCT_UNBOUND,
|
|
|
|
};
|
|
|
|
|
2019-10-23 00:25:58 +08:00
|
|
|
/*
|
|
|
|
* Per-node worker thread pool
|
|
|
|
*/
|
|
|
|
struct io_wqe {
|
|
|
|
struct {
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spinlock_t lock;
|
2019-11-27 02:59:32 +08:00
|
|
|
struct io_wq_work_list work_list;
|
2019-10-23 00:25:58 +08:00
|
|
|
unsigned long hash_map;
|
|
|
|
unsigned flags;
|
|
|
|
} ____cacheline_aligned_in_smp;
|
|
|
|
|
|
|
|
int node;
|
2019-11-08 02:41:16 +08:00
|
|
|
struct io_wqe_acct acct[2];
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2019-11-14 23:00:41 +08:00
|
|
|
struct hlist_nulls_head free_list;
|
2019-11-14 04:54:49 +08:00
|
|
|
struct list_head all_list;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
|
|
|
struct io_wq *wq;
|
2020-03-24 03:57:22 +08:00
|
|
|
struct io_wq_work *hash_tail[IO_WQ_NR_HASH_BUCKETS];
|
2019-10-23 00:25:58 +08:00
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Per io_wq state
|
|
|
|
*/
|
|
|
|
struct io_wq {
|
|
|
|
struct io_wqe **wqes;
|
|
|
|
unsigned long state;
|
|
|
|
|
2020-03-04 21:14:12 +08:00
|
|
|
free_work_fn *free_work;
|
2020-06-09 02:08:20 +08:00
|
|
|
io_wq_work_fn *do_work;
|
2019-11-13 13:31:31 +08:00
|
|
|
|
2019-10-23 00:25:58 +08:00
|
|
|
struct task_struct *manager;
|
2019-11-08 02:41:16 +08:00
|
|
|
struct user_struct *user;
|
2019-10-23 00:25:58 +08:00
|
|
|
refcount_t refs;
|
|
|
|
struct completion done;
|
2020-01-24 06:33:32 +08:00
|
|
|
|
2020-10-22 23:02:50 +08:00
|
|
|
struct hlist_node cpuhp_node;
|
2021-02-17 05:15:30 +08:00
|
|
|
|
|
|
|
pid_t task_pid;
|
2019-10-23 00:25:58 +08:00
|
|
|
};
|
|
|
|
|
2020-10-22 23:02:50 +08:00
|
|
|
static enum cpuhp_state io_wq_online;
|
|
|
|
|
2019-10-23 00:25:58 +08:00
|
|
|
static bool io_worker_get(struct io_worker *worker)
|
|
|
|
{
|
|
|
|
return refcount_inc_not_zero(&worker->ref);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void io_worker_release(struct io_worker *worker)
|
|
|
|
{
|
|
|
|
if (refcount_dec_and_test(&worker->ref))
|
|
|
|
wake_up_process(worker->task);
|
|
|
|
}
|
|
|
|
|
2019-11-08 02:41:16 +08:00
|
|
|
static inline struct io_wqe_acct *io_work_get_acct(struct io_wqe *wqe,
|
|
|
|
struct io_wq_work *work)
|
|
|
|
{
|
|
|
|
if (work->flags & IO_WQ_WORK_UNBOUND)
|
|
|
|
return &wqe->acct[IO_WQ_ACCT_UNBOUND];
|
|
|
|
|
|
|
|
return &wqe->acct[IO_WQ_ACCT_BOUND];
|
|
|
|
}
|
|
|
|
|
2021-02-18 00:00:57 +08:00
|
|
|
static inline struct io_wqe_acct *io_wqe_get_acct(struct io_worker *worker)
|
2019-11-08 02:41:16 +08:00
|
|
|
{
|
2021-02-18 00:00:57 +08:00
|
|
|
struct io_wqe *wqe = worker->wqe;
|
|
|
|
|
2019-11-08 02:41:16 +08:00
|
|
|
if (worker->flags & IO_WORKER_F_BOUND)
|
|
|
|
return &wqe->acct[IO_WQ_ACCT_BOUND];
|
|
|
|
|
|
|
|
return &wqe->acct[IO_WQ_ACCT_UNBOUND];
|
|
|
|
}
|
|
|
|
|
2019-10-23 00:25:58 +08:00
|
|
|
static void io_worker_exit(struct io_worker *worker)
|
|
|
|
{
|
|
|
|
struct io_wqe *wqe = worker->wqe;
|
2021-02-18 00:00:57 +08:00
|
|
|
struct io_wqe_acct *acct = io_wqe_get_acct(worker);
|
2019-10-23 00:25:58 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If we're not at zero, someone else is holding a brief reference
|
|
|
|
* to the worker. Wait for that to go away.
|
|
|
|
*/
|
|
|
|
set_current_state(TASK_INTERRUPTIBLE);
|
|
|
|
if (!refcount_dec_and_test(&worker->ref))
|
|
|
|
schedule();
|
|
|
|
__set_current_state(TASK_RUNNING);
|
|
|
|
|
|
|
|
preempt_disable();
|
|
|
|
current->flags &= ~PF_IO_WORKER;
|
|
|
|
if (worker->flags & IO_WORKER_F_RUNNING)
|
2019-11-08 02:41:16 +08:00
|
|
|
atomic_dec(&acct->nr_running);
|
|
|
|
if (!(worker->flags & IO_WORKER_F_BOUND))
|
|
|
|
atomic_dec(&wqe->wq->user->processes);
|
2019-10-23 00:25:58 +08:00
|
|
|
worker->flags = 0;
|
|
|
|
preempt_enable();
|
|
|
|
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_lock_irq(&wqe->lock);
|
2019-10-23 00:25:58 +08:00
|
|
|
hlist_nulls_del_rcu(&worker->nulls_node);
|
2019-11-14 04:54:49 +08:00
|
|
|
list_del_rcu(&worker->all_list);
|
2019-11-08 02:41:16 +08:00
|
|
|
acct->nr_workers--;
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_unlock_irq(&wqe->lock);
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2019-11-02 15:55:01 +08:00
|
|
|
kfree_rcu(worker, rcu);
|
2020-09-26 21:26:55 +08:00
|
|
|
if (refcount_dec_and_test(&wqe->wq->refs))
|
|
|
|
complete(&wqe->wq->done);
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2019-11-08 02:41:16 +08:00
|
|
|
static inline bool io_wqe_run_queue(struct io_wqe *wqe)
|
|
|
|
__must_hold(wqe->lock)
|
|
|
|
{
|
2019-11-27 02:59:32 +08:00
|
|
|
if (!wq_list_empty(&wqe->work_list) &&
|
|
|
|
!(wqe->flags & IO_WQE_FLAG_STALLED))
|
2019-11-08 02:41:16 +08:00
|
|
|
return true;
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Check head of free list for an available worker. If one isn't available,
|
|
|
|
* caller must wake up the wq manager to create one.
|
|
|
|
*/
|
|
|
|
static bool io_wqe_activate_free_worker(struct io_wqe *wqe)
|
|
|
|
__must_hold(RCU)
|
|
|
|
{
|
|
|
|
struct hlist_nulls_node *n;
|
|
|
|
struct io_worker *worker;
|
|
|
|
|
2019-11-14 23:00:41 +08:00
|
|
|
n = rcu_dereference(hlist_nulls_first_rcu(&wqe->free_list));
|
2019-11-08 02:41:16 +08:00
|
|
|
if (is_a_nulls(n))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
worker = hlist_nulls_entry(n, struct io_worker, nulls_node);
|
|
|
|
if (io_worker_get(worker)) {
|
2019-12-08 12:03:59 +08:00
|
|
|
wake_up_process(worker->task);
|
2019-11-08 02:41:16 +08:00
|
|
|
io_worker_release(worker);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We need a worker. If we find a free one, we're good. If not, and we're
|
|
|
|
* below the max number of workers, wake up the manager to create one.
|
|
|
|
*/
|
|
|
|
static void io_wqe_wake_worker(struct io_wqe *wqe, struct io_wqe_acct *acct)
|
|
|
|
{
|
|
|
|
bool ret;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Most likely an attempt to queue unbounded work on an io_wq that
|
|
|
|
* wasn't setup with any unbounded workers.
|
|
|
|
*/
|
|
|
|
WARN_ON_ONCE(!acct->max_workers);
|
|
|
|
|
|
|
|
rcu_read_lock();
|
|
|
|
ret = io_wqe_activate_free_worker(wqe);
|
|
|
|
rcu_read_unlock();
|
|
|
|
|
|
|
|
if (!ret && acct->nr_workers < acct->max_workers)
|
|
|
|
wake_up_process(wqe->wq->manager);
|
|
|
|
}
|
|
|
|
|
2021-02-18 00:00:57 +08:00
|
|
|
static void io_wqe_inc_running(struct io_worker *worker)
|
2019-11-08 02:41:16 +08:00
|
|
|
{
|
2021-02-18 00:00:57 +08:00
|
|
|
struct io_wqe_acct *acct = io_wqe_get_acct(worker);
|
2019-11-08 02:41:16 +08:00
|
|
|
|
|
|
|
atomic_inc(&acct->nr_running);
|
|
|
|
}
|
|
|
|
|
2021-02-18 00:00:57 +08:00
|
|
|
static void io_wqe_dec_running(struct io_worker *worker)
|
2019-11-08 02:41:16 +08:00
|
|
|
__must_hold(wqe->lock)
|
|
|
|
{
|
2021-02-18 00:00:57 +08:00
|
|
|
struct io_wqe_acct *acct = io_wqe_get_acct(worker);
|
|
|
|
struct io_wqe *wqe = worker->wqe;
|
2019-11-08 02:41:16 +08:00
|
|
|
|
|
|
|
if (atomic_dec_and_test(&acct->nr_running) && io_wqe_run_queue(wqe))
|
|
|
|
io_wqe_wake_worker(wqe, acct);
|
|
|
|
}
|
|
|
|
|
2021-02-18 00:00:57 +08:00
|
|
|
static void io_worker_start(struct io_worker *worker)
|
2019-10-23 00:25:58 +08:00
|
|
|
{
|
|
|
|
worker->flags |= (IO_WORKER_F_UP | IO_WORKER_F_RUNNING);
|
2021-02-18 00:00:57 +08:00
|
|
|
io_wqe_inc_running(worker);
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Worker will start processing some work. Move it to the busy list, if
|
|
|
|
* it's currently on the freelist
|
|
|
|
*/
|
|
|
|
static void __io_worker_busy(struct io_wqe *wqe, struct io_worker *worker,
|
|
|
|
struct io_wq_work *work)
|
|
|
|
__must_hold(wqe->lock)
|
|
|
|
{
|
2019-11-08 02:41:16 +08:00
|
|
|
bool worker_bound, work_bound;
|
|
|
|
|
2019-10-23 00:25:58 +08:00
|
|
|
if (worker->flags & IO_WORKER_F_FREE) {
|
|
|
|
worker->flags &= ~IO_WORKER_F_FREE;
|
|
|
|
hlist_nulls_del_init_rcu(&worker->nulls_node);
|
|
|
|
}
|
2019-11-08 02:41:16 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If worker is moving from bound to unbound (or vice versa), then
|
|
|
|
* ensure we update the running accounting.
|
|
|
|
*/
|
2019-11-19 14:22:16 +08:00
|
|
|
worker_bound = (worker->flags & IO_WORKER_F_BOUND) != 0;
|
|
|
|
work_bound = (work->flags & IO_WQ_WORK_UNBOUND) == 0;
|
|
|
|
if (worker_bound != work_bound) {
|
2021-02-18 00:00:57 +08:00
|
|
|
io_wqe_dec_running(worker);
|
2019-11-08 02:41:16 +08:00
|
|
|
if (work_bound) {
|
|
|
|
worker->flags |= IO_WORKER_F_BOUND;
|
|
|
|
wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers--;
|
|
|
|
wqe->acct[IO_WQ_ACCT_BOUND].nr_workers++;
|
|
|
|
atomic_dec(&wqe->wq->user->processes);
|
|
|
|
} else {
|
|
|
|
worker->flags &= ~IO_WORKER_F_BOUND;
|
|
|
|
wqe->acct[IO_WQ_ACCT_UNBOUND].nr_workers++;
|
|
|
|
wqe->acct[IO_WQ_ACCT_BOUND].nr_workers--;
|
|
|
|
atomic_inc(&wqe->wq->user->processes);
|
|
|
|
}
|
2021-02-18 00:00:57 +08:00
|
|
|
io_wqe_inc_running(worker);
|
2019-11-08 02:41:16 +08:00
|
|
|
}
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* No work, worker going to sleep. Move to freelist, and unuse mm if we
|
|
|
|
* have one attached. Dropping the mm may potentially sleep, so we drop
|
|
|
|
* the lock in that case and return success. Since the caller has to
|
|
|
|
* retry the loop in that case (we changed task state), we don't regrab
|
|
|
|
* the lock if we return success.
|
|
|
|
*/
|
|
|
|
static bool __io_worker_idle(struct io_wqe *wqe, struct io_worker *worker)
|
|
|
|
__must_hold(wqe->lock)
|
|
|
|
{
|
|
|
|
if (!(worker->flags & IO_WORKER_F_FREE)) {
|
|
|
|
worker->flags |= IO_WORKER_F_FREE;
|
2019-11-14 23:00:41 +08:00
|
|
|
hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list);
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
return false;
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2020-03-14 05:31:05 +08:00
|
|
|
static inline unsigned int io_get_work_hash(struct io_wq_work *work)
|
|
|
|
{
|
|
|
|
return work->flags >> IO_WQ_HASH_SHIFT;
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct io_wq_work *io_get_next_work(struct io_wqe *wqe)
|
2019-10-23 00:25:58 +08:00
|
|
|
__must_hold(wqe->lock)
|
|
|
|
{
|
2019-11-27 02:59:32 +08:00
|
|
|
struct io_wq_work_node *node, *prev;
|
2020-03-24 03:57:22 +08:00
|
|
|
struct io_wq_work *work, *tail;
|
2020-03-14 05:31:05 +08:00
|
|
|
unsigned int hash;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2019-11-27 02:59:32 +08:00
|
|
|
wq_list_for_each(node, prev, &wqe->work_list) {
|
|
|
|
work = container_of(node, struct io_wq_work, list);
|
|
|
|
|
2019-10-23 00:25:58 +08:00
|
|
|
/* not hashed, can run anytime */
|
2020-03-14 05:31:04 +08:00
|
|
|
if (!io_wq_is_hashed(work)) {
|
2020-03-24 03:57:22 +08:00
|
|
|
wq_list_del(&wqe->work_list, node, prev);
|
2019-10-23 00:25:58 +08:00
|
|
|
return work;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* hashed, can run if not already running */
|
2020-03-14 05:31:05 +08:00
|
|
|
hash = io_get_work_hash(work);
|
|
|
|
if (!(wqe->hash_map & BIT(hash))) {
|
|
|
|
wqe->hash_map |= BIT(hash);
|
2020-03-24 03:57:22 +08:00
|
|
|
/* all items with this hash lie in [work, tail] */
|
|
|
|
tail = wqe->hash_tail[hash];
|
|
|
|
wqe->hash_tail[hash] = NULL;
|
|
|
|
wq_list_cut(&wqe->work_list, &tail->list, prev);
|
2019-10-23 00:25:58 +08:00
|
|
|
return work;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
static void io_flush_signals(void)
|
2020-01-28 07:34:48 +08:00
|
|
|
{
|
2021-02-17 05:15:30 +08:00
|
|
|
if (unlikely(test_tsk_thread_flag(current, TIF_NOTIFY_SIGNAL))) {
|
|
|
|
if (current->task_works)
|
|
|
|
task_work_run();
|
|
|
|
clear_tsk_thread_flag(current, TIF_NOTIFY_SIGNAL);
|
2020-01-28 07:34:48 +08:00
|
|
|
}
|
2020-03-04 21:14:09 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
static void io_assign_current_work(struct io_worker *worker,
|
|
|
|
struct io_wq_work *work)
|
|
|
|
{
|
2020-03-14 05:31:03 +08:00
|
|
|
if (work) {
|
2021-02-17 05:15:30 +08:00
|
|
|
io_flush_signals();
|
2020-03-14 05:31:03 +08:00
|
|
|
cond_resched();
|
|
|
|
}
|
2020-03-04 21:14:09 +08:00
|
|
|
|
|
|
|
spin_lock_irq(&worker->lock);
|
|
|
|
worker->cur_work = work;
|
|
|
|
spin_unlock_irq(&worker->lock);
|
|
|
|
}
|
|
|
|
|
2020-03-14 05:31:05 +08:00
|
|
|
static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work);
|
|
|
|
|
2019-10-23 00:25:58 +08:00
|
|
|
static void io_worker_handle_work(struct io_worker *worker)
|
|
|
|
__releases(wqe->lock)
|
|
|
|
{
|
|
|
|
struct io_wqe *wqe = worker->wqe;
|
|
|
|
struct io_wq *wq = wqe->wq;
|
|
|
|
|
|
|
|
do {
|
2020-03-24 03:57:22 +08:00
|
|
|
struct io_wq_work *work;
|
2020-03-04 21:14:11 +08:00
|
|
|
get_next:
|
2019-10-23 00:25:58 +08:00
|
|
|
/*
|
|
|
|
* If we got some work, mark us as busy. If we didn't, but
|
|
|
|
* the list isn't empty, it means we stalled on hashed work.
|
|
|
|
* Mark us stalled so we don't keep looking for work when we
|
|
|
|
* can't make progress, any work completion or insertion will
|
|
|
|
* clear the stalled flag.
|
|
|
|
*/
|
2020-03-14 05:31:05 +08:00
|
|
|
work = io_get_next_work(wqe);
|
2019-10-23 00:25:58 +08:00
|
|
|
if (work)
|
|
|
|
__io_worker_busy(wqe, worker, work);
|
2019-11-27 02:59:32 +08:00
|
|
|
else if (!wq_list_empty(&wqe->work_list))
|
2019-10-23 00:25:58 +08:00
|
|
|
wqe->flags |= IO_WQE_FLAG_STALLED;
|
|
|
|
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_unlock_irq(&wqe->lock);
|
2019-10-23 00:25:58 +08:00
|
|
|
if (!work)
|
|
|
|
break;
|
2020-03-04 21:14:10 +08:00
|
|
|
io_assign_current_work(worker, work);
|
2019-11-14 00:43:34 +08:00
|
|
|
|
2020-03-04 21:14:09 +08:00
|
|
|
/* handle a whole dependent link */
|
|
|
|
do {
|
2021-02-04 21:52:08 +08:00
|
|
|
struct io_wq_work *next_hashed, *linked;
|
2020-07-25 19:42:00 +08:00
|
|
|
unsigned int hash = io_get_work_hash(work);
|
2020-03-04 21:14:09 +08:00
|
|
|
|
2020-03-24 03:57:22 +08:00
|
|
|
next_hashed = wq_next_work(work);
|
2021-02-04 21:52:08 +08:00
|
|
|
wq->do_work(work);
|
|
|
|
io_assign_current_work(worker, NULL);
|
2020-03-04 21:14:09 +08:00
|
|
|
|
2021-02-04 21:52:08 +08:00
|
|
|
linked = wq->free_work(work);
|
2020-03-24 03:57:22 +08:00
|
|
|
work = next_hashed;
|
|
|
|
if (!work && linked && !io_wq_is_hashed(linked)) {
|
|
|
|
work = linked;
|
|
|
|
linked = NULL;
|
|
|
|
}
|
|
|
|
io_assign_current_work(worker, work);
|
|
|
|
if (linked)
|
|
|
|
io_wqe_enqueue(wqe, linked);
|
|
|
|
|
|
|
|
if (hash != -1U && !next_hashed) {
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_lock_irq(&wqe->lock);
|
2020-03-04 21:14:09 +08:00
|
|
|
wqe->hash_map &= ~BIT_ULL(hash);
|
|
|
|
wqe->flags &= ~IO_WQE_FLAG_STALLED;
|
2020-03-04 21:14:11 +08:00
|
|
|
/* skip unnecessary unlock-lock wqe->lock */
|
|
|
|
if (!work)
|
|
|
|
goto get_next;
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_unlock_irq(&wqe->lock);
|
2019-11-13 13:31:31 +08:00
|
|
|
}
|
2020-03-04 21:14:10 +08:00
|
|
|
} while (work);
|
2019-11-13 13:31:31 +08:00
|
|
|
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_lock_irq(&wqe->lock);
|
2019-10-23 00:25:58 +08:00
|
|
|
} while (1);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int io_wqe_worker(void *data)
|
|
|
|
{
|
|
|
|
struct io_worker *worker = data;
|
|
|
|
struct io_wqe *wqe = worker->wqe;
|
|
|
|
struct io_wq *wq = wqe->wq;
|
|
|
|
|
2021-02-18 00:00:57 +08:00
|
|
|
io_worker_start(worker);
|
2019-10-23 00:25:58 +08:00
|
|
|
|
|
|
|
while (!test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
|
2019-12-08 12:03:59 +08:00
|
|
|
set_current_state(TASK_INTERRUPTIBLE);
|
2019-12-08 12:06:46 +08:00
|
|
|
loop:
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_lock_irq(&wqe->lock);
|
2019-10-23 00:25:58 +08:00
|
|
|
if (io_wqe_run_queue(wqe)) {
|
|
|
|
__set_current_state(TASK_RUNNING);
|
|
|
|
io_worker_handle_work(worker);
|
2019-12-08 12:06:46 +08:00
|
|
|
goto loop;
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
/* drops the lock on success, retry */
|
|
|
|
if (__io_worker_idle(wqe, worker)) {
|
|
|
|
__release(&wqe->lock);
|
2019-12-08 12:06:46 +08:00
|
|
|
goto loop;
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_unlock_irq(&wqe->lock);
|
2021-02-17 05:15:30 +08:00
|
|
|
io_flush_signals();
|
2019-10-23 00:25:58 +08:00
|
|
|
if (schedule_timeout(WORKER_IDLE_TIMEOUT))
|
|
|
|
continue;
|
2021-02-17 05:15:30 +08:00
|
|
|
if (fatal_signal_pending(current))
|
|
|
|
break;
|
2019-10-23 00:25:58 +08:00
|
|
|
/* timed out, exit unless we're the fixed worker */
|
|
|
|
if (test_bit(IO_WQ_BIT_EXIT, &wq->state) ||
|
|
|
|
!(worker->flags & IO_WORKER_F_FIXED))
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_lock_irq(&wqe->lock);
|
2019-11-27 02:59:32 +08:00
|
|
|
if (!wq_list_empty(&wqe->work_list))
|
2019-10-23 00:25:58 +08:00
|
|
|
io_worker_handle_work(worker);
|
|
|
|
else
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_unlock_irq(&wqe->lock);
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
io_worker_exit(worker);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Called when a worker is scheduled in. Mark us as currently running.
|
|
|
|
*/
|
|
|
|
void io_wq_worker_running(struct task_struct *tsk)
|
|
|
|
{
|
2021-02-17 05:15:30 +08:00
|
|
|
struct io_worker *worker = tsk->pf_io_worker;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
if (!worker)
|
|
|
|
return;
|
2019-10-23 00:25:58 +08:00
|
|
|
if (!(worker->flags & IO_WORKER_F_UP))
|
|
|
|
return;
|
|
|
|
if (worker->flags & IO_WORKER_F_RUNNING)
|
|
|
|
return;
|
|
|
|
worker->flags |= IO_WORKER_F_RUNNING;
|
2021-02-18 00:00:57 +08:00
|
|
|
io_wqe_inc_running(worker);
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Called when worker is going to sleep. If there are no workers currently
|
|
|
|
* running and we have work pending, wake up a free one or have the manager
|
|
|
|
* set one up.
|
|
|
|
*/
|
|
|
|
void io_wq_worker_sleeping(struct task_struct *tsk)
|
|
|
|
{
|
2021-02-17 05:15:30 +08:00
|
|
|
struct io_worker *worker = tsk->pf_io_worker;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
if (!worker)
|
|
|
|
return;
|
2019-10-23 00:25:58 +08:00
|
|
|
if (!(worker->flags & IO_WORKER_F_UP))
|
|
|
|
return;
|
|
|
|
if (!(worker->flags & IO_WORKER_F_RUNNING))
|
|
|
|
return;
|
|
|
|
|
|
|
|
worker->flags &= ~IO_WORKER_F_RUNNING;
|
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
raw_spin_lock_irq(&worker->wqe->lock);
|
2021-02-18 00:00:57 +08:00
|
|
|
io_wqe_dec_running(worker);
|
2021-02-17 05:15:30 +08:00
|
|
|
raw_spin_unlock_irq(&worker->wqe->lock);
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
static int task_thread(void *data, int index)
|
2019-10-23 00:25:58 +08:00
|
|
|
{
|
2021-02-17 05:15:30 +08:00
|
|
|
struct io_worker *worker = data;
|
|
|
|
struct io_wqe *wqe = worker->wqe;
|
2020-09-26 21:26:55 +08:00
|
|
|
struct io_wqe_acct *acct = &wqe->acct[index];
|
2021-02-17 05:15:30 +08:00
|
|
|
struct io_wq *wq = wqe->wq;
|
|
|
|
char buf[TASK_COMM_LEN];
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
sprintf(buf, "iou-wrk-%d", wq->task_pid);
|
|
|
|
set_task_comm(current, buf);
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
current->pf_io_worker = worker;
|
|
|
|
worker->task = current;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
set_cpus_allowed_ptr(current, cpumask_of_node(wqe->node));
|
|
|
|
current->flags |= PF_NO_SETAFFINITY;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_lock_irq(&wqe->lock);
|
2019-11-14 23:00:41 +08:00
|
|
|
hlist_nulls_add_head_rcu(&worker->nulls_node, &wqe->free_list);
|
2019-11-14 04:54:49 +08:00
|
|
|
list_add_tail_rcu(&worker->all_list, &wqe->all_list);
|
2019-10-23 00:25:58 +08:00
|
|
|
worker->flags |= IO_WORKER_F_FREE;
|
2019-11-08 02:41:16 +08:00
|
|
|
if (index == IO_WQ_ACCT_BOUND)
|
|
|
|
worker->flags |= IO_WORKER_F_BOUND;
|
|
|
|
if (!acct->nr_workers && (worker->flags & IO_WORKER_F_BOUND))
|
2019-10-23 00:25:58 +08:00
|
|
|
worker->flags |= IO_WORKER_F_FIXED;
|
2019-11-08 02:41:16 +08:00
|
|
|
acct->nr_workers++;
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_unlock_irq(&wqe->lock);
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2019-11-08 02:41:16 +08:00
|
|
|
if (index == IO_WQ_ACCT_UNBOUND)
|
|
|
|
atomic_inc(&wq->user->processes);
|
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
io_wqe_worker(data);
|
|
|
|
do_exit(0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int task_thread_bound(void *data)
|
|
|
|
{
|
|
|
|
return task_thread(data, IO_WQ_ACCT_BOUND);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int task_thread_unbound(void *data)
|
|
|
|
{
|
|
|
|
return task_thread(data, IO_WQ_ACCT_UNBOUND);
|
|
|
|
}
|
|
|
|
|
|
|
|
static pid_t fork_thread(int (*fn)(void *), void *arg)
|
|
|
|
{
|
|
|
|
unsigned long flags = CLONE_FS|CLONE_FILES|CLONE_SIGHAND|CLONE_THREAD|
|
|
|
|
CLONE_IO|SIGCHLD;
|
|
|
|
struct kernel_clone_args args = {
|
|
|
|
.flags = ((lower_32_bits(flags) | CLONE_VM |
|
|
|
|
CLONE_UNTRACED) & ~CSIGNAL),
|
|
|
|
.exit_signal = (lower_32_bits(flags) & CSIGNAL),
|
|
|
|
.stack = (unsigned long)fn,
|
|
|
|
.stack_size = (unsigned long)arg,
|
|
|
|
};
|
|
|
|
|
|
|
|
return kernel_clone(&args);
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool create_io_worker(struct io_wq *wq, struct io_wqe *wqe, int index)
|
|
|
|
{
|
|
|
|
struct io_worker *worker;
|
|
|
|
pid_t pid;
|
|
|
|
|
|
|
|
worker = kzalloc_node(sizeof(*worker), GFP_KERNEL, wqe->node);
|
|
|
|
if (!worker)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
refcount_set(&worker->ref, 1);
|
|
|
|
worker->nulls_node.pprev = NULL;
|
|
|
|
worker->wqe = wqe;
|
|
|
|
spin_lock_init(&worker->lock);
|
|
|
|
|
|
|
|
if (index == IO_WQ_ACCT_BOUND)
|
|
|
|
pid = fork_thread(task_thread_bound, worker);
|
|
|
|
else
|
|
|
|
pid = fork_thread(task_thread_unbound, worker);
|
|
|
|
if (pid < 0) {
|
|
|
|
kfree(worker);
|
|
|
|
return false;
|
|
|
|
}
|
2020-09-26 21:26:55 +08:00
|
|
|
refcount_inc(&wq->refs);
|
2019-11-19 23:37:07 +08:00
|
|
|
return true;
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2019-11-08 02:41:16 +08:00
|
|
|
static inline bool io_wqe_need_worker(struct io_wqe *wqe, int index)
|
2019-10-23 00:25:58 +08:00
|
|
|
__must_hold(wqe->lock)
|
|
|
|
{
|
2019-11-08 02:41:16 +08:00
|
|
|
struct io_wqe_acct *acct = &wqe->acct[index];
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2019-11-08 02:41:16 +08:00
|
|
|
/* if we have available workers or no work, no need */
|
2019-11-14 23:00:41 +08:00
|
|
|
if (!hlist_nulls_empty(&wqe->free_list) || !io_wqe_run_queue(wqe))
|
2019-11-08 02:41:16 +08:00
|
|
|
return false;
|
|
|
|
return acct->nr_workers < acct->max_workers;
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2020-09-26 21:26:55 +08:00
|
|
|
/*
|
|
|
|
* Iterate the passed in list and call the specific function for each
|
|
|
|
* worker that isn't exiting
|
|
|
|
*/
|
|
|
|
static bool io_wq_for_each_worker(struct io_wqe *wqe,
|
|
|
|
bool (*func)(struct io_worker *, void *),
|
|
|
|
void *data)
|
|
|
|
{
|
|
|
|
struct io_worker *worker;
|
|
|
|
bool ret = false;
|
|
|
|
|
|
|
|
list_for_each_entry_rcu(worker, &wqe->all_list, all_list) {
|
|
|
|
if (io_worker_get(worker)) {
|
|
|
|
/* no task if node is/was offline */
|
|
|
|
if (worker->task)
|
|
|
|
ret = func(worker, data);
|
|
|
|
io_worker_release(worker);
|
|
|
|
if (ret)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool io_wq_worker_wake(struct io_worker *worker, void *data)
|
|
|
|
{
|
|
|
|
wake_up_process(worker->task);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2019-10-23 00:25:58 +08:00
|
|
|
/*
|
|
|
|
* Manager thread. Tasked with creating new workers, if we need them.
|
|
|
|
*/
|
|
|
|
static int io_wq_manager(void *data)
|
|
|
|
{
|
|
|
|
struct io_wq *wq = data;
|
2021-02-17 05:15:30 +08:00
|
|
|
char buf[TASK_COMM_LEN];
|
2019-11-27 02:10:20 +08:00
|
|
|
int node;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
sprintf(buf, "iou-mgr-%d", wq->task_pid);
|
|
|
|
set_task_comm(current, buf);
|
|
|
|
current->flags |= PF_IO_WORKER;
|
|
|
|
wq->manager = current;
|
|
|
|
|
2019-11-19 23:37:07 +08:00
|
|
|
complete(&wq->done);
|
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
while (!test_bit(IO_WQ_BIT_EXIT, &wq->state)) {
|
2019-11-27 02:10:20 +08:00
|
|
|
for_each_node(node) {
|
|
|
|
struct io_wqe *wqe = wq->wqes[node];
|
2019-11-08 02:41:16 +08:00
|
|
|
bool fork_worker[2] = { false, false };
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2020-02-11 21:30:06 +08:00
|
|
|
if (!node_online(node))
|
|
|
|
continue;
|
|
|
|
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_lock_irq(&wqe->lock);
|
2019-11-08 02:41:16 +08:00
|
|
|
if (io_wqe_need_worker(wqe, IO_WQ_ACCT_BOUND))
|
|
|
|
fork_worker[IO_WQ_ACCT_BOUND] = true;
|
|
|
|
if (io_wqe_need_worker(wqe, IO_WQ_ACCT_UNBOUND))
|
|
|
|
fork_worker[IO_WQ_ACCT_UNBOUND] = true;
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_unlock_irq(&wqe->lock);
|
2019-11-08 02:41:16 +08:00
|
|
|
if (fork_worker[IO_WQ_ACCT_BOUND])
|
|
|
|
create_io_worker(wq, wqe, IO_WQ_ACCT_BOUND);
|
|
|
|
if (fork_worker[IO_WQ_ACCT_UNBOUND])
|
|
|
|
create_io_worker(wq, wqe, IO_WQ_ACCT_UNBOUND);
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
set_current_state(TASK_INTERRUPTIBLE);
|
|
|
|
schedule_timeout(HZ);
|
2021-02-17 05:15:30 +08:00
|
|
|
if (fatal_signal_pending(current))
|
|
|
|
set_bit(IO_WQ_BIT_EXIT, &wq->state);
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2020-09-26 21:26:55 +08:00
|
|
|
if (refcount_dec_and_test(&wq->refs)) {
|
2019-11-19 23:37:07 +08:00
|
|
|
complete(&wq->done);
|
2021-02-17 05:15:30 +08:00
|
|
|
do_exit(0);
|
2020-09-26 21:26:55 +08:00
|
|
|
}
|
|
|
|
/* if ERROR is set and we get here, we have workers to wake */
|
|
|
|
if (test_bit(IO_WQ_BIT_ERROR, &wq->state)) {
|
|
|
|
rcu_read_lock();
|
|
|
|
for_each_node(node)
|
|
|
|
io_wq_for_each_worker(wq->wqes[node], io_wq_worker_wake, NULL);
|
|
|
|
rcu_read_unlock();
|
|
|
|
}
|
2021-02-17 05:15:30 +08:00
|
|
|
do_exit(0);
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2019-11-08 02:41:16 +08:00
|
|
|
static bool io_wq_can_queue(struct io_wqe *wqe, struct io_wqe_acct *acct,
|
|
|
|
struct io_wq_work *work)
|
|
|
|
{
|
|
|
|
bool free_worker;
|
|
|
|
|
|
|
|
if (!(work->flags & IO_WQ_WORK_UNBOUND))
|
|
|
|
return true;
|
|
|
|
if (atomic_read(&acct->nr_running))
|
|
|
|
return true;
|
|
|
|
|
|
|
|
rcu_read_lock();
|
2019-11-14 23:00:41 +08:00
|
|
|
free_worker = !hlist_nulls_empty(&wqe->free_list);
|
2019-11-08 02:41:16 +08:00
|
|
|
rcu_read_unlock();
|
|
|
|
if (free_worker)
|
|
|
|
return true;
|
|
|
|
|
|
|
|
if (atomic_read(&wqe->wq->user->processes) >= acct->max_workers &&
|
|
|
|
!(capable(CAP_SYS_RESOURCE) || capable(CAP_SYS_ADMIN)))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2020-03-04 21:14:12 +08:00
|
|
|
static void io_run_cancel(struct io_wq_work *work, struct io_wqe *wqe)
|
2020-03-02 00:18:19 +08:00
|
|
|
{
|
2020-03-04 21:14:12 +08:00
|
|
|
struct io_wq *wq = wqe->wq;
|
|
|
|
|
2020-03-02 00:18:19 +08:00
|
|
|
do {
|
|
|
|
work->flags |= IO_WQ_WORK_CANCEL;
|
2021-02-04 21:52:08 +08:00
|
|
|
wq->do_work(work);
|
|
|
|
work = wq->free_work(work);
|
2020-03-02 00:18:19 +08:00
|
|
|
} while (work);
|
|
|
|
}
|
|
|
|
|
2020-03-24 03:57:22 +08:00
|
|
|
static void io_wqe_insert_work(struct io_wqe *wqe, struct io_wq_work *work)
|
|
|
|
{
|
|
|
|
unsigned int hash;
|
|
|
|
struct io_wq_work *tail;
|
|
|
|
|
|
|
|
if (!io_wq_is_hashed(work)) {
|
|
|
|
append:
|
|
|
|
wq_list_add_tail(&work->list, &wqe->work_list);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
hash = io_get_work_hash(work);
|
|
|
|
tail = wqe->hash_tail[hash];
|
|
|
|
wqe->hash_tail[hash] = work;
|
|
|
|
if (!tail)
|
|
|
|
goto append;
|
|
|
|
|
|
|
|
wq_list_add_after(&work->list, &tail->list, &wqe->work_list);
|
|
|
|
}
|
|
|
|
|
2019-10-23 00:25:58 +08:00
|
|
|
static void io_wqe_enqueue(struct io_wqe *wqe, struct io_wq_work *work)
|
|
|
|
{
|
2019-11-08 02:41:16 +08:00
|
|
|
struct io_wqe_acct *acct = io_work_get_acct(wqe, work);
|
2019-12-17 23:46:33 +08:00
|
|
|
int work_flags;
|
2019-10-23 00:25:58 +08:00
|
|
|
unsigned long flags;
|
|
|
|
|
2019-11-08 02:41:16 +08:00
|
|
|
/*
|
|
|
|
* Do early check to see if we need a new unbound worker, and if we do,
|
|
|
|
* if we're allowed to do so. This isn't 100% accurate as there's a
|
|
|
|
* gap between this check and incrementing the value, but that's OK.
|
|
|
|
* It's close enough to not be an issue, fork() has the same delay.
|
|
|
|
*/
|
|
|
|
if (unlikely(!io_wq_can_queue(wqe, acct, work))) {
|
2020-03-04 21:14:12 +08:00
|
|
|
io_run_cancel(work, wqe);
|
2019-11-08 02:41:16 +08:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2019-12-17 23:46:33 +08:00
|
|
|
work_flags = work->flags;
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_lock_irqsave(&wqe->lock, flags);
|
2020-03-24 03:57:22 +08:00
|
|
|
io_wqe_insert_work(wqe, work);
|
2019-10-23 00:25:58 +08:00
|
|
|
wqe->flags &= ~IO_WQE_FLAG_STALLED;
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_unlock_irqrestore(&wqe->lock, flags);
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2019-12-17 23:46:33 +08:00
|
|
|
if ((work_flags & IO_WQ_WORK_CONCURRENT) ||
|
|
|
|
!atomic_read(&acct->nr_running))
|
2019-11-08 02:41:16 +08:00
|
|
|
io_wqe_wake_worker(wqe, acct);
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
void io_wq_enqueue(struct io_wq *wq, struct io_wq_work *work)
|
|
|
|
{
|
|
|
|
struct io_wqe *wqe = wq->wqes[numa_node_id()];
|
|
|
|
|
|
|
|
io_wqe_enqueue(wqe, work);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2020-03-14 05:31:04 +08:00
|
|
|
* Work items that hash to the same value will not be done in parallel.
|
|
|
|
* Used to limit concurrent writes, generally hashed by inode.
|
2019-10-23 00:25:58 +08:00
|
|
|
*/
|
2020-03-14 05:31:04 +08:00
|
|
|
void io_wq_hash_work(struct io_wq_work *work, void *val)
|
2019-10-23 00:25:58 +08:00
|
|
|
{
|
2020-03-14 05:31:04 +08:00
|
|
|
unsigned int bit;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
|
|
|
bit = hash_ptr(val, IO_WQ_HASH_ORDER);
|
|
|
|
work->flags |= (IO_WQ_WORK_HASHED | (bit << IO_WQ_HASH_SHIFT));
|
|
|
|
}
|
|
|
|
|
2019-10-29 11:49:21 +08:00
|
|
|
struct io_cb_cancel_data {
|
2020-03-07 06:15:39 +08:00
|
|
|
work_cancel_fn *fn;
|
|
|
|
void *data;
|
2020-06-15 15:24:03 +08:00
|
|
|
int nr_running;
|
|
|
|
int nr_pending;
|
|
|
|
bool cancel_all;
|
2019-10-29 11:49:21 +08:00
|
|
|
};
|
|
|
|
|
2020-03-07 06:15:39 +08:00
|
|
|
static bool io_wq_worker_cancel(struct io_worker *worker, void *data)
|
2019-10-29 11:49:21 +08:00
|
|
|
{
|
2020-03-07 06:15:39 +08:00
|
|
|
struct io_cb_cancel_data *match = data;
|
2019-11-06 04:51:51 +08:00
|
|
|
unsigned long flags;
|
2019-10-29 11:49:21 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Hold the lock to avoid ->cur_work going out of scope, caller
|
2019-11-14 00:43:34 +08:00
|
|
|
* may dereference the passed in work.
|
2019-10-29 11:49:21 +08:00
|
|
|
*/
|
2019-11-14 00:43:34 +08:00
|
|
|
spin_lock_irqsave(&worker->lock, flags);
|
2019-10-29 11:49:21 +08:00
|
|
|
if (worker->cur_work &&
|
2020-03-07 06:15:39 +08:00
|
|
|
match->fn(worker->cur_work, match->data)) {
|
2021-02-17 05:15:30 +08:00
|
|
|
set_notify_signal(worker->task);
|
2020-06-15 15:24:03 +08:00
|
|
|
match->nr_running++;
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
2019-11-14 00:43:34 +08:00
|
|
|
spin_unlock_irqrestore(&worker->lock, flags);
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2020-06-15 15:24:03 +08:00
|
|
|
return match->nr_running && !match->cancel_all;
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2020-08-24 01:33:10 +08:00
|
|
|
static inline void io_wqe_remove_pending(struct io_wqe *wqe,
|
|
|
|
struct io_wq_work *work,
|
|
|
|
struct io_wq_work_node *prev)
|
|
|
|
{
|
|
|
|
unsigned int hash = io_get_work_hash(work);
|
|
|
|
struct io_wq_work *prev_work = NULL;
|
|
|
|
|
|
|
|
if (io_wq_is_hashed(work) && work == wqe->hash_tail[hash]) {
|
|
|
|
if (prev)
|
|
|
|
prev_work = container_of(prev, struct io_wq_work, list);
|
|
|
|
if (prev_work && io_get_work_hash(prev_work) == hash)
|
|
|
|
wqe->hash_tail[hash] = prev_work;
|
|
|
|
else
|
|
|
|
wqe->hash_tail[hash] = NULL;
|
|
|
|
}
|
|
|
|
wq_list_del(&wqe->work_list, &work->list, prev);
|
|
|
|
}
|
|
|
|
|
2020-06-15 15:24:03 +08:00
|
|
|
static void io_wqe_cancel_pending_work(struct io_wqe *wqe,
|
2020-06-15 15:24:02 +08:00
|
|
|
struct io_cb_cancel_data *match)
|
2019-10-23 00:25:58 +08:00
|
|
|
{
|
2019-11-27 02:59:32 +08:00
|
|
|
struct io_wq_work_node *node, *prev;
|
2019-10-23 00:25:58 +08:00
|
|
|
struct io_wq_work *work;
|
2019-11-06 04:51:51 +08:00
|
|
|
unsigned long flags;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2020-06-15 15:24:03 +08:00
|
|
|
retry:
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_lock_irqsave(&wqe->lock, flags);
|
2019-11-27 02:59:32 +08:00
|
|
|
wq_list_for_each(node, prev, &wqe->work_list) {
|
|
|
|
work = container_of(node, struct io_wq_work, list);
|
2020-06-15 15:24:03 +08:00
|
|
|
if (!match->fn(work, match->data))
|
|
|
|
continue;
|
2020-08-24 01:33:10 +08:00
|
|
|
io_wqe_remove_pending(wqe, work, prev);
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_unlock_irqrestore(&wqe->lock, flags);
|
2020-06-15 15:24:03 +08:00
|
|
|
io_run_cancel(work, wqe);
|
|
|
|
match->nr_pending++;
|
|
|
|
if (!match->cancel_all)
|
|
|
|
return;
|
|
|
|
|
|
|
|
/* not safe to continue after unlock */
|
|
|
|
goto retry;
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_unlock_irqrestore(&wqe->lock, flags);
|
2020-06-15 15:24:02 +08:00
|
|
|
}
|
|
|
|
|
2020-06-15 15:24:03 +08:00
|
|
|
static void io_wqe_cancel_running_work(struct io_wqe *wqe,
|
2020-06-15 15:24:02 +08:00
|
|
|
struct io_cb_cancel_data *match)
|
|
|
|
{
|
2019-10-23 00:25:58 +08:00
|
|
|
rcu_read_lock();
|
2020-06-15 15:24:03 +08:00
|
|
|
io_wq_for_each_worker(wqe, io_wq_worker_cancel, match);
|
2019-10-23 00:25:58 +08:00
|
|
|
rcu_read_unlock();
|
|
|
|
}
|
|
|
|
|
2020-03-07 06:15:39 +08:00
|
|
|
enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel,
|
2020-06-15 15:24:03 +08:00
|
|
|
void *data, bool cancel_all)
|
2019-10-23 00:25:58 +08:00
|
|
|
{
|
2020-03-07 06:15:39 +08:00
|
|
|
struct io_cb_cancel_data match = {
|
2020-06-15 15:24:03 +08:00
|
|
|
.fn = cancel,
|
|
|
|
.data = data,
|
|
|
|
.cancel_all = cancel_all,
|
2020-02-09 10:13:32 +08:00
|
|
|
};
|
2019-11-27 02:10:20 +08:00
|
|
|
int node;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2020-06-15 15:24:02 +08:00
|
|
|
/*
|
|
|
|
* First check pending list, if we're lucky we can just remove it
|
|
|
|
* from there. CANCEL_OK means that the work is returned as-new,
|
|
|
|
* no completion will be posted for it.
|
|
|
|
*/
|
2019-11-27 02:10:20 +08:00
|
|
|
for_each_node(node) {
|
|
|
|
struct io_wqe *wqe = wq->wqes[node];
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2020-06-15 15:24:03 +08:00
|
|
|
io_wqe_cancel_pending_work(wqe, &match);
|
|
|
|
if (match.nr_pending && !match.cancel_all)
|
2020-06-15 15:24:02 +08:00
|
|
|
return IO_WQ_CANCEL_OK;
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2020-06-15 15:24:02 +08:00
|
|
|
/*
|
|
|
|
* Now check if a free (going busy) or busy worker has the work
|
|
|
|
* currently running. If we find it there, we'll return CANCEL_RUNNING
|
|
|
|
* as an indication that we attempt to signal cancellation. The
|
|
|
|
* completion will run normally in this case.
|
|
|
|
*/
|
|
|
|
for_each_node(node) {
|
|
|
|
struct io_wqe *wqe = wq->wqes[node];
|
|
|
|
|
2020-06-15 15:24:03 +08:00
|
|
|
io_wqe_cancel_running_work(wqe, &match);
|
|
|
|
if (match.nr_running && !match.cancel_all)
|
2020-06-15 15:24:02 +08:00
|
|
|
return IO_WQ_CANCEL_RUNNING;
|
|
|
|
}
|
|
|
|
|
2020-06-15 15:24:03 +08:00
|
|
|
if (match.nr_running)
|
|
|
|
return IO_WQ_CANCEL_RUNNING;
|
|
|
|
if (match.nr_pending)
|
|
|
|
return IO_WQ_CANCEL_OK;
|
2020-06-15 15:24:02 +08:00
|
|
|
return IO_WQ_CANCEL_NOTFOUND;
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2019-11-25 23:49:20 +08:00
|
|
|
struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
|
2019-10-23 00:25:58 +08:00
|
|
|
{
|
2019-11-27 02:10:20 +08:00
|
|
|
int ret = -ENOMEM, node;
|
2019-10-23 00:25:58 +08:00
|
|
|
struct io_wq *wq;
|
|
|
|
|
2020-06-09 02:08:20 +08:00
|
|
|
if (WARN_ON_ONCE(!data->free_work || !data->do_work))
|
2020-03-04 21:14:12 +08:00
|
|
|
return ERR_PTR(-EINVAL);
|
|
|
|
|
2019-11-27 00:39:45 +08:00
|
|
|
wq = kzalloc(sizeof(*wq), GFP_KERNEL);
|
2019-10-23 00:25:58 +08:00
|
|
|
if (!wq)
|
|
|
|
return ERR_PTR(-ENOMEM);
|
|
|
|
|
2019-11-27 02:10:20 +08:00
|
|
|
wq->wqes = kcalloc(nr_node_ids, sizeof(struct io_wqe *), GFP_KERNEL);
|
2020-10-22 23:02:50 +08:00
|
|
|
if (!wq->wqes)
|
|
|
|
goto err_wq;
|
|
|
|
|
|
|
|
ret = cpuhp_state_add_instance_nocalls(io_wq_online, &wq->cpuhp_node);
|
|
|
|
if (ret)
|
|
|
|
goto err_wqes;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2020-03-04 21:14:12 +08:00
|
|
|
wq->free_work = data->free_work;
|
2020-06-09 02:08:20 +08:00
|
|
|
wq->do_work = data->do_work;
|
2019-11-13 13:31:31 +08:00
|
|
|
|
2019-11-08 02:41:16 +08:00
|
|
|
/* caller must already hold a reference to this */
|
2019-11-25 23:49:20 +08:00
|
|
|
wq->user = data->user;
|
2019-11-08 02:41:16 +08:00
|
|
|
|
2020-10-22 23:02:50 +08:00
|
|
|
ret = -ENOMEM;
|
2019-11-27 02:10:20 +08:00
|
|
|
for_each_node(node) {
|
2019-10-23 00:25:58 +08:00
|
|
|
struct io_wqe *wqe;
|
2020-02-11 21:30:06 +08:00
|
|
|
int alloc_node = node;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2020-02-11 21:30:06 +08:00
|
|
|
if (!node_online(alloc_node))
|
|
|
|
alloc_node = NUMA_NO_NODE;
|
|
|
|
wqe = kzalloc_node(sizeof(struct io_wqe), GFP_KERNEL, alloc_node);
|
2019-10-23 00:25:58 +08:00
|
|
|
if (!wqe)
|
2019-11-27 02:10:20 +08:00
|
|
|
goto err;
|
|
|
|
wq->wqes[node] = wqe;
|
2020-02-11 21:30:06 +08:00
|
|
|
wqe->node = alloc_node;
|
2019-11-08 02:41:16 +08:00
|
|
|
wqe->acct[IO_WQ_ACCT_BOUND].max_workers = bounded;
|
|
|
|
atomic_set(&wqe->acct[IO_WQ_ACCT_BOUND].nr_running, 0);
|
2019-11-25 23:49:20 +08:00
|
|
|
if (wq->user) {
|
2019-11-08 02:41:16 +08:00
|
|
|
wqe->acct[IO_WQ_ACCT_UNBOUND].max_workers =
|
|
|
|
task_rlimit(current, RLIMIT_NPROC);
|
|
|
|
}
|
|
|
|
atomic_set(&wqe->acct[IO_WQ_ACCT_UNBOUND].nr_running, 0);
|
2019-10-23 00:25:58 +08:00
|
|
|
wqe->wq = wq;
|
2020-09-01 16:41:46 +08:00
|
|
|
raw_spin_lock_init(&wqe->lock);
|
2019-11-27 02:59:32 +08:00
|
|
|
INIT_WQ_LIST(&wqe->work_list);
|
2019-11-14 23:00:41 +08:00
|
|
|
INIT_HLIST_NULLS_HEAD(&wqe->free_list, 0);
|
2019-11-14 04:54:49 +08:00
|
|
|
INIT_LIST_HEAD(&wqe->all_list);
|
2019-10-23 00:25:58 +08:00
|
|
|
}
|
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
wq->task_pid = current->pid;
|
2019-10-23 00:25:58 +08:00
|
|
|
init_completion(&wq->done);
|
2021-02-17 05:15:30 +08:00
|
|
|
refcount_set(&wq->refs, 1);
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
current->flags |= PF_IO_WORKER;
|
|
|
|
ret = fork_thread(io_wq_manager, wq);
|
|
|
|
current->flags &= ~PF_IO_WORKER;
|
|
|
|
if (ret >= 0) {
|
2019-11-19 23:37:07 +08:00
|
|
|
wait_for_completion(&wq->done);
|
|
|
|
reinit_completion(&wq->done);
|
2019-10-23 00:25:58 +08:00
|
|
|
return wq;
|
|
|
|
}
|
|
|
|
|
2021-02-17 05:15:30 +08:00
|
|
|
if (refcount_dec_and_test(&wq->refs))
|
|
|
|
complete(&wq->done);
|
2019-11-19 23:37:07 +08:00
|
|
|
err:
|
2020-10-22 23:02:50 +08:00
|
|
|
cpuhp_state_remove_instance_nocalls(io_wq_online, &wq->cpuhp_node);
|
2019-11-27 02:10:20 +08:00
|
|
|
for_each_node(node)
|
|
|
|
kfree(wq->wqes[node]);
|
2020-10-22 23:02:50 +08:00
|
|
|
err_wqes:
|
2019-11-19 23:37:07 +08:00
|
|
|
kfree(wq->wqes);
|
2020-10-22 23:02:50 +08:00
|
|
|
err_wq:
|
2019-11-19 23:37:07 +08:00
|
|
|
kfree(wq);
|
2019-10-23 00:25:58 +08:00
|
|
|
return ERR_PTR(ret);
|
|
|
|
}
|
|
|
|
|
2021-02-17 06:42:24 +08:00
|
|
|
void io_wq_destroy(struct io_wq *wq)
|
2019-10-23 00:25:58 +08:00
|
|
|
{
|
2019-11-27 02:10:20 +08:00
|
|
|
int node;
|
2019-10-23 00:25:58 +08:00
|
|
|
|
2020-10-22 23:02:50 +08:00
|
|
|
cpuhp_state_remove_instance_nocalls(io_wq_online, &wq->cpuhp_node);
|
|
|
|
|
2019-11-19 23:37:07 +08:00
|
|
|
set_bit(IO_WQ_BIT_EXIT, &wq->state);
|
|
|
|
if (wq->manager)
|
2021-02-17 05:15:30 +08:00
|
|
|
wake_up_process(wq->manager);
|
2019-10-23 00:25:58 +08:00
|
|
|
|
|
|
|
rcu_read_lock();
|
2019-11-27 02:10:20 +08:00
|
|
|
for_each_node(node)
|
|
|
|
io_wq_for_each_worker(wq->wqes[node], io_wq_worker_wake, NULL);
|
2019-10-23 00:25:58 +08:00
|
|
|
rcu_read_unlock();
|
|
|
|
|
|
|
|
wait_for_completion(&wq->done);
|
|
|
|
|
2019-11-27 02:10:20 +08:00
|
|
|
for_each_node(node)
|
|
|
|
kfree(wq->wqes[node]);
|
2019-10-23 00:25:58 +08:00
|
|
|
kfree(wq->wqes);
|
|
|
|
kfree(wq);
|
|
|
|
}
|
2020-01-24 06:33:32 +08:00
|
|
|
|
2020-10-22 23:02:50 +08:00
|
|
|
static bool io_wq_worker_affinity(struct io_worker *worker, void *data)
|
|
|
|
{
|
|
|
|
struct task_struct *task = worker->task;
|
|
|
|
struct rq_flags rf;
|
|
|
|
struct rq *rq;
|
|
|
|
|
|
|
|
rq = task_rq_lock(task, &rf);
|
|
|
|
do_set_cpus_allowed(task, cpumask_of_node(worker->wqe->node));
|
|
|
|
task->flags |= PF_NO_SETAFFINITY;
|
|
|
|
task_rq_unlock(rq, task, &rf);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int io_wq_cpu_online(unsigned int cpu, struct hlist_node *node)
|
|
|
|
{
|
|
|
|
struct io_wq *wq = hlist_entry_safe(node, struct io_wq, cpuhp_node);
|
|
|
|
int i;
|
|
|
|
|
|
|
|
rcu_read_lock();
|
|
|
|
for_each_node(i)
|
|
|
|
io_wq_for_each_worker(wq->wqes[i], io_wq_worker_affinity, NULL);
|
|
|
|
rcu_read_unlock();
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static __init int io_wq_init(void)
|
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN, "io-wq/online",
|
|
|
|
io_wq_cpu_online, NULL);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
io_wq_online = ret;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
subsys_initcall(io_wq_init);
|