linux/fs/io-wq.h
Xiaoguang Wang 7cdaf587de io_uring: avoid whole io_wq_work copy for requests completed inline
If requests can be submitted and completed inline, we don't need to
initialize whole io_wq_work in io_init_req(), which is an expensive
operation, add a new 'REQ_F_WORK_INITIALIZED' to determine whether
io_wq_work is initialized and add a helper io_req_init_async(), users
must call io_req_init_async() for the first time touching any members
of io_wq_work.

I use /dev/nullb0 to evaluate performance improvement in my physical
machine:
  modprobe null_blk nr_devices=1 completion_nsec=0
  sudo taskset -c 60 fio  -name=fiotest -filename=/dev/nullb0 -iodepth=128
  -thread -rw=read -ioengine=io_uring -direct=1 -bs=4k -size=100G -numjobs=1
  -time_based -runtime=120

before this patch:
Run status group 0 (all jobs):
   READ: bw=724MiB/s (759MB/s), 724MiB/s-724MiB/s (759MB/s-759MB/s),
   io=84.8GiB (91.1GB), run=120001-120001msec

With this patch:
Run status group 0 (all jobs):
   READ: bw=761MiB/s (798MB/s), 761MiB/s-761MiB/s (798MB/s-798MB/s),
   io=89.2GiB (95.8GB), run=120001-120001msec

About 5% improvement.

Signed-off-by: Xiaoguang Wang <xiaoguang.wang@linux.alibaba.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2020-06-10 17:58:46 -06:00

154 lines
3.6 KiB
C

#ifndef INTERNAL_IO_WQ_H
#define INTERNAL_IO_WQ_H
struct io_wq;
enum {
IO_WQ_WORK_CANCEL = 1,
IO_WQ_WORK_HASHED = 4,
IO_WQ_WORK_UNBOUND = 32,
IO_WQ_WORK_NO_CANCEL = 256,
IO_WQ_WORK_CONCURRENT = 512,
IO_WQ_HASH_SHIFT = 24, /* upper 8 bits are used for hash key */
};
enum io_wq_cancel {
IO_WQ_CANCEL_OK, /* cancelled before started */
IO_WQ_CANCEL_RUNNING, /* found, running, and attempted cancelled */
IO_WQ_CANCEL_NOTFOUND, /* work not found */
};
struct io_wq_work_node {
struct io_wq_work_node *next;
};
struct io_wq_work_list {
struct io_wq_work_node *first;
struct io_wq_work_node *last;
};
static inline void wq_list_add_after(struct io_wq_work_node *node,
struct io_wq_work_node *pos,
struct io_wq_work_list *list)
{
struct io_wq_work_node *next = pos->next;
pos->next = node;
node->next = next;
if (!next)
list->last = node;
}
static inline void wq_list_add_tail(struct io_wq_work_node *node,
struct io_wq_work_list *list)
{
if (!list->first) {
list->last = node;
WRITE_ONCE(list->first, node);
} else {
list->last->next = node;
list->last = node;
}
}
static inline void wq_list_cut(struct io_wq_work_list *list,
struct io_wq_work_node *last,
struct io_wq_work_node *prev)
{
/* first in the list, if prev==NULL */
if (!prev)
WRITE_ONCE(list->first, last->next);
else
prev->next = last->next;
if (last == list->last)
list->last = prev;
last->next = NULL;
}
static inline void wq_list_del(struct io_wq_work_list *list,
struct io_wq_work_node *node,
struct io_wq_work_node *prev)
{
wq_list_cut(list, node, prev);
}
#define wq_list_for_each(pos, prv, head) \
for (pos = (head)->first, prv = NULL; pos; prv = pos, pos = (pos)->next)
#define wq_list_empty(list) (READ_ONCE((list)->first) == NULL)
#define INIT_WQ_LIST(list) do { \
(list)->first = NULL; \
(list)->last = NULL; \
} while (0)
struct io_wq_work {
struct io_wq_work_node list;
struct files_struct *files;
struct mm_struct *mm;
const struct cred *creds;
struct fs_struct *fs;
unsigned flags;
pid_t task_pid;
};
static inline struct io_wq_work *wq_next_work(struct io_wq_work *work)
{
if (!work->list.next)
return NULL;
return container_of(work->list.next, struct io_wq_work, list);
}
typedef void (free_work_fn)(struct io_wq_work *);
typedef void (io_wq_work_fn)(struct io_wq_work **);
struct io_wq_data {
struct user_struct *user;
io_wq_work_fn *do_work;
free_work_fn *free_work;
};
struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data);
bool io_wq_get(struct io_wq *wq, struct io_wq_data *data);
void io_wq_destroy(struct io_wq *wq);
void io_wq_enqueue(struct io_wq *wq, struct io_wq_work *work);
void io_wq_hash_work(struct io_wq_work *work, void *val);
static inline bool io_wq_is_hashed(struct io_wq_work *work)
{
return work->flags & IO_WQ_WORK_HASHED;
}
void io_wq_cancel_all(struct io_wq *wq);
enum io_wq_cancel io_wq_cancel_work(struct io_wq *wq, struct io_wq_work *cwork);
enum io_wq_cancel io_wq_cancel_pid(struct io_wq *wq, pid_t pid);
typedef bool (work_cancel_fn)(struct io_wq_work *, void *);
enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel,
void *data);
struct task_struct *io_wq_get_task(struct io_wq *wq);
#if defined(CONFIG_IO_WQ)
extern void io_wq_worker_sleeping(struct task_struct *);
extern void io_wq_worker_running(struct task_struct *);
#else
static inline void io_wq_worker_sleeping(struct task_struct *tsk)
{
}
static inline void io_wq_worker_running(struct task_struct *tsk)
{
}
#endif
static inline bool io_wq_current_is_worker(void)
{
return in_task() && (current->flags & PF_IO_WORKER);
}
#endif