2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* Anticipatory & deadline i/o scheduler.
|
|
|
|
*
|
2006-09-04 07:41:16 -06:00
|
|
|
* Copyright (C) 2002 Jens Axboe <axboe@kernel.dk>
|
2005-11-07 01:59:53 -07:00
|
|
|
* Nick Piggin <nickpiggin@yahoo.com.au>
|
2005-04-16 16:20:36 -06:00
|
|
|
*
|
|
|
|
*/
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/fs.h>
|
|
|
|
#include <linux/blkdev.h>
|
|
|
|
#include <linux/elevator.h>
|
|
|
|
#include <linux/bio.h>
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/slab.h>
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/compiler.h>
|
|
|
|
#include <linux/rbtree.h>
|
|
|
|
#include <linux/interrupt.h>
|
|
|
|
|
|
|
|
/*
|
|
|
|
* See Documentation/block/as-iosched.txt
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* max time before a read is submitted.
|
|
|
|
*/
|
|
|
|
#define default_read_expire (HZ / 8)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* ditto for writes, these limits are not hard, even
|
|
|
|
* if the disk is capable of satisfying them.
|
|
|
|
*/
|
|
|
|
#define default_write_expire (HZ / 4)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* read_batch_expire describes how long we will allow a stream of reads to
|
|
|
|
* persist before looking to see whether it is time to switch over to writes.
|
|
|
|
*/
|
|
|
|
#define default_read_batch_expire (HZ / 2)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* write_batch_expire describes how long we want a stream of writes to run for.
|
|
|
|
* This is not a hard limit, but a target we set for the auto-tuning thingy.
|
|
|
|
* See, the problem is: we can send a lot of writes to disk cache / TCQ in
|
|
|
|
* a short amount of time...
|
|
|
|
*/
|
|
|
|
#define default_write_batch_expire (HZ / 8)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* max time we may wait to anticipate a read (default around 6ms)
|
|
|
|
*/
|
|
|
|
#define default_antic_expire ((HZ / 150) ? HZ / 150 : 1)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Keep track of up to 20ms thinktimes. We can go as big as we like here,
|
|
|
|
* however huge values tend to interfere and not decay fast enough. A program
|
|
|
|
* might be in a non-io phase of operation. Waiting on user input for example,
|
|
|
|
* or doing a lengthy computation. A small penalty can be justified there, and
|
|
|
|
* will still catch out those processes that constantly have large thinktimes.
|
|
|
|
*/
|
|
|
|
#define MAX_THINKTIME (HZ/50UL)
|
|
|
|
|
|
|
|
/* Bits in as_io_context.state */
|
|
|
|
enum as_io_states {
|
2005-11-07 01:59:53 -07:00
|
|
|
AS_TASK_RUNNING=0, /* Process has not exited */
|
2005-04-16 16:20:36 -06:00
|
|
|
AS_TASK_IOSTARTED, /* Process has started some IO */
|
|
|
|
AS_TASK_IORUNNING, /* Process has completed some IO */
|
|
|
|
};
|
|
|
|
|
|
|
|
enum anticipation_status {
|
|
|
|
ANTIC_OFF=0, /* Not anticipating (normal operation) */
|
|
|
|
ANTIC_WAIT_REQ, /* The last read has not yet completed */
|
|
|
|
ANTIC_WAIT_NEXT, /* Currently anticipating a request vs
|
|
|
|
last read (which has completed) */
|
|
|
|
ANTIC_FINISHED, /* Anticipating but have found a candidate
|
|
|
|
* or timed out */
|
|
|
|
};
|
|
|
|
|
|
|
|
struct as_data {
|
|
|
|
/*
|
|
|
|
* run time data
|
|
|
|
*/
|
|
|
|
|
|
|
|
struct request_queue *q; /* the "owner" queue */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* requests (as_rq s) are present on both sort_list and fifo_list
|
|
|
|
*/
|
|
|
|
struct rb_root sort_list[2];
|
|
|
|
struct list_head fifo_list[2];
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
struct request *next_rq[2]; /* next in sort order */
|
2009-04-08 03:02:08 -06:00
|
|
|
sector_t last_sector[2]; /* last SYNC & ASYNC sectors */
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
unsigned long exit_prob; /* probability a task will exit while
|
|
|
|
being waited on */
|
2005-11-07 01:59:53 -07:00
|
|
|
unsigned long exit_no_coop; /* probablility an exited task will
|
|
|
|
not be part of a later cooperating
|
|
|
|
request */
|
2005-04-16 16:20:36 -06:00
|
|
|
unsigned long new_ttime_total; /* mean thinktime on new proc */
|
|
|
|
unsigned long new_ttime_mean;
|
|
|
|
u64 new_seek_total; /* mean seek on new proc */
|
|
|
|
sector_t new_seek_mean;
|
|
|
|
|
|
|
|
unsigned long current_batch_expires;
|
|
|
|
unsigned long last_check_fifo[2];
|
|
|
|
int changed_batch; /* 1: waiting for old batch to end */
|
|
|
|
int new_batch; /* 1: waiting on first read complete */
|
2009-04-08 03:02:08 -06:00
|
|
|
int batch_data_dir; /* current batch SYNC / ASYNC */
|
2005-04-16 16:20:36 -06:00
|
|
|
int write_batch_count; /* max # of reqs in a write batch */
|
|
|
|
int current_write_count; /* how many requests left this batch */
|
|
|
|
int write_batch_idled; /* has the write batch gone idle? */
|
|
|
|
|
|
|
|
enum anticipation_status antic_status;
|
|
|
|
unsigned long antic_start; /* jiffies: when it started */
|
|
|
|
struct timer_list antic_timer; /* anticipatory scheduling timer */
|
|
|
|
struct work_struct antic_work; /* Deferred unplugging */
|
|
|
|
struct io_context *io_context; /* Identify the expected process */
|
|
|
|
int ioc_finished; /* IO associated with io_context is finished */
|
|
|
|
int nr_dispatched;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* settings that change how the i/o scheduler behaves
|
|
|
|
*/
|
|
|
|
unsigned long fifo_expire[2];
|
|
|
|
unsigned long batch_expire[2];
|
|
|
|
unsigned long antic_expire;
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* per-request data.
|
|
|
|
*/
|
|
|
|
enum arq_state {
|
|
|
|
AS_RQ_NEW=0, /* New - not referenced and not on any lists */
|
|
|
|
AS_RQ_QUEUED, /* In the request queue. It belongs to the
|
|
|
|
scheduler */
|
|
|
|
AS_RQ_DISPATCHED, /* On the dispatch list. It belongs to the
|
|
|
|
driver now */
|
|
|
|
AS_RQ_PRESCHED, /* Debug poisoning for requests being used */
|
|
|
|
AS_RQ_REMOVED,
|
|
|
|
AS_RQ_MERGED,
|
|
|
|
AS_RQ_POSTSCHED, /* when they shouldn't be */
|
|
|
|
};
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
#define RQ_IOC(rq) ((struct io_context *) (rq)->elevator_private)
|
|
|
|
#define RQ_STATE(rq) ((enum arq_state)(rq)->elevator_private2)
|
|
|
|
#define RQ_SET_STATE(rq, state) ((rq)->elevator_private2 = (void *) state)
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 21:10:01 -06:00
|
|
|
static DEFINE_PER_CPU(unsigned long, ioc_count);
|
2006-03-18 13:05:53 -07:00
|
|
|
static struct completion *ioc_gone;
|
2008-05-29 01:35:22 -06:00
|
|
|
static DEFINE_SPINLOCK(ioc_gone_lock);
|
2006-03-18 13:05:53 -07:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
static void as_move_to_dispatch(struct as_data *ad, struct request *rq);
|
2005-11-11 06:27:09 -07:00
|
|
|
static void as_antic_stop(struct as_data *ad);
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* IO Context helper functions
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* Called to deallocate the as_io_context */
|
|
|
|
static void free_as_io_context(struct as_io_context *aic)
|
|
|
|
{
|
|
|
|
kfree(aic);
|
2006-07-18 21:10:01 -06:00
|
|
|
elv_ioc_count_dec(ioc_count);
|
2008-05-29 01:35:22 -06:00
|
|
|
if (ioc_gone) {
|
|
|
|
/*
|
|
|
|
* AS scheduler is exiting, grab exit lock and check
|
|
|
|
* the pending io context count. If it hits zero,
|
|
|
|
* complete ioc_gone and set it back to NULL.
|
|
|
|
*/
|
|
|
|
spin_lock(&ioc_gone_lock);
|
|
|
|
if (ioc_gone && !elv_ioc_count_read(ioc_count)) {
|
|
|
|
complete(ioc_gone);
|
|
|
|
ioc_gone = NULL;
|
|
|
|
}
|
|
|
|
spin_unlock(&ioc_gone_lock);
|
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2006-03-18 11:21:20 -07:00
|
|
|
static void as_trim(struct io_context *ioc)
|
|
|
|
{
|
2008-02-01 01:44:28 -07:00
|
|
|
spin_lock_irq(&ioc->lock);
|
2006-03-18 13:05:53 -07:00
|
|
|
if (ioc->aic)
|
|
|
|
free_as_io_context(ioc->aic);
|
2006-03-18 11:21:20 -07:00
|
|
|
ioc->aic = NULL;
|
2008-02-01 01:44:28 -07:00
|
|
|
spin_unlock_irq(&ioc->lock);
|
2006-03-18 11:21:20 -07:00
|
|
|
}
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
/* Called when the task exits */
|
|
|
|
static void exit_as_io_context(struct as_io_context *aic)
|
|
|
|
{
|
|
|
|
WARN_ON(!test_bit(AS_TASK_RUNNING, &aic->state));
|
|
|
|
clear_bit(AS_TASK_RUNNING, &aic->state);
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct as_io_context *alloc_as_io_context(void)
|
|
|
|
{
|
|
|
|
struct as_io_context *ret;
|
|
|
|
|
|
|
|
ret = kmalloc(sizeof(*ret), GFP_ATOMIC);
|
|
|
|
if (ret) {
|
|
|
|
ret->dtor = free_as_io_context;
|
|
|
|
ret->exit = exit_as_io_context;
|
|
|
|
ret->state = 1 << AS_TASK_RUNNING;
|
|
|
|
atomic_set(&ret->nr_queued, 0);
|
|
|
|
atomic_set(&ret->nr_dispatched, 0);
|
|
|
|
spin_lock_init(&ret->lock);
|
|
|
|
ret->ttime_total = 0;
|
|
|
|
ret->ttime_samples = 0;
|
|
|
|
ret->ttime_mean = 0;
|
|
|
|
ret->seek_total = 0;
|
|
|
|
ret->seek_samples = 0;
|
|
|
|
ret->seek_mean = 0;
|
2006-07-18 21:10:01 -06:00
|
|
|
elv_ioc_count_inc(ioc_count);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the current task has no AS IO context then create one and initialise it.
|
|
|
|
* Then take a ref on the task's io context and return it.
|
|
|
|
*/
|
2006-07-19 15:39:40 -06:00
|
|
|
static struct io_context *as_get_io_context(int node)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2006-07-19 15:39:40 -06:00
|
|
|
struct io_context *ioc = get_io_context(GFP_ATOMIC, node);
|
2005-04-16 16:20:36 -06:00
|
|
|
if (ioc && !ioc->aic) {
|
|
|
|
ioc->aic = alloc_as_io_context();
|
|
|
|
if (!ioc->aic) {
|
|
|
|
put_io_context(ioc);
|
|
|
|
ioc = NULL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return ioc;
|
|
|
|
}
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
static void as_put_io_context(struct request *rq)
|
2005-10-20 08:42:29 -06:00
|
|
|
{
|
|
|
|
struct as_io_context *aic;
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
if (unlikely(!RQ_IOC(rq)))
|
2005-10-20 08:42:29 -06:00
|
|
|
return;
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
aic = RQ_IOC(rq)->aic;
|
2005-10-20 08:42:29 -06:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
if (rq_is_sync(rq) && aic) {
|
2008-02-01 01:44:28 -07:00
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
spin_lock_irqsave(&aic->lock, flags);
|
2005-10-20 08:42:29 -06:00
|
|
|
set_bit(AS_TASK_IORUNNING, &aic->state);
|
|
|
|
aic->last_end_request = jiffies;
|
2008-02-01 01:44:28 -07:00
|
|
|
spin_unlock_irqrestore(&aic->lock, flags);
|
2005-10-20 08:42:29 -06:00
|
|
|
}
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
put_io_context(RQ_IOC(rq));
|
2005-10-20 08:42:29 -06:00
|
|
|
}
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* rb tree support functions
|
|
|
|
*/
|
2006-07-28 01:26:13 -06:00
|
|
|
#define RQ_RB_ROOT(ad, rq) (&(ad)->sort_list[rq_is_sync((rq))])
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
static void as_add_rq_rb(struct as_data *ad, struct request *rq)
|
2005-11-11 06:27:09 -07:00
|
|
|
{
|
2006-07-18 13:06:01 -06:00
|
|
|
struct request *alias;
|
2005-11-11 06:27:09 -07:00
|
|
|
|
2006-07-28 01:26:13 -06:00
|
|
|
while ((unlikely(alias = elv_rb_add(RQ_RB_ROOT(ad, rq), rq)))) {
|
2006-07-18 13:07:29 -06:00
|
|
|
as_move_to_dispatch(ad, alias);
|
2005-11-11 06:27:09 -07:00
|
|
|
as_antic_stop(ad);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
static inline void as_del_rq_rb(struct as_data *ad, struct request *rq)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2006-07-28 01:26:13 -06:00
|
|
|
elv_rb_del(RQ_RB_ROOT(ad, rq), rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* IO Scheduler proper
|
|
|
|
*/
|
|
|
|
|
|
|
|
#define MAXBACK (1024 * 1024) /*
|
|
|
|
* Maximum distance the disk will go backward
|
|
|
|
* for a request.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#define BACK_PENALTY 2
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_choose_req selects the preferred one of two requests of the same data_dir
|
|
|
|
* ignoring time - eg. timeouts, which is the job of as_dispatch_request
|
|
|
|
*/
|
2006-07-18 13:07:29 -06:00
|
|
|
static struct request *
|
|
|
|
as_choose_req(struct as_data *ad, struct request *rq1, struct request *rq2)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
int data_dir;
|
|
|
|
sector_t last, s1, s2, d1, d2;
|
|
|
|
int r1_wrap=0, r2_wrap=0; /* requests are behind the disk head */
|
|
|
|
const sector_t maxback = MAXBACK;
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
if (rq1 == NULL || rq1 == rq2)
|
|
|
|
return rq2;
|
|
|
|
if (rq2 == NULL)
|
|
|
|
return rq1;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
data_dir = rq_is_sync(rq1);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
last = ad->last_sector[data_dir];
|
2009-05-07 07:24:39 -06:00
|
|
|
s1 = blk_rq_pos(rq1);
|
|
|
|
s2 = blk_rq_pos(rq2);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
BUG_ON(data_dir != rq_is_sync(rq2));
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Strict one way elevator _except_ in the case where we allow
|
|
|
|
* short backward seeks which are biased as twice the cost of a
|
|
|
|
* similar forward seek.
|
|
|
|
*/
|
|
|
|
if (s1 >= last)
|
|
|
|
d1 = s1 - last;
|
|
|
|
else if (s1+maxback >= last)
|
|
|
|
d1 = (last - s1)*BACK_PENALTY;
|
|
|
|
else {
|
|
|
|
r1_wrap = 1;
|
|
|
|
d1 = 0; /* shut up, gcc */
|
|
|
|
}
|
|
|
|
|
|
|
|
if (s2 >= last)
|
|
|
|
d2 = s2 - last;
|
|
|
|
else if (s2+maxback >= last)
|
|
|
|
d2 = (last - s2)*BACK_PENALTY;
|
|
|
|
else {
|
|
|
|
r2_wrap = 1;
|
|
|
|
d2 = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Found required data */
|
|
|
|
if (!r1_wrap && r2_wrap)
|
2006-07-18 13:07:29 -06:00
|
|
|
return rq1;
|
2005-04-16 16:20:36 -06:00
|
|
|
else if (!r2_wrap && r1_wrap)
|
2006-07-18 13:07:29 -06:00
|
|
|
return rq2;
|
2005-04-16 16:20:36 -06:00
|
|
|
else if (r1_wrap && r2_wrap) {
|
|
|
|
/* both behind the head */
|
|
|
|
if (s1 <= s2)
|
2006-07-18 13:07:29 -06:00
|
|
|
return rq1;
|
2005-04-16 16:20:36 -06:00
|
|
|
else
|
2006-07-18 13:07:29 -06:00
|
|
|
return rq2;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Both requests in front of the head */
|
|
|
|
if (d1 < d2)
|
2006-07-18 13:07:29 -06:00
|
|
|
return rq1;
|
2005-04-16 16:20:36 -06:00
|
|
|
else if (d2 < d1)
|
2006-07-18 13:07:29 -06:00
|
|
|
return rq2;
|
2005-04-16 16:20:36 -06:00
|
|
|
else {
|
|
|
|
if (s1 >= s2)
|
2006-07-18 13:07:29 -06:00
|
|
|
return rq1;
|
2005-04-16 16:20:36 -06:00
|
|
|
else
|
2006-07-18 13:07:29 -06:00
|
|
|
return rq2;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2006-07-18 13:07:29 -06:00
|
|
|
* as_find_next_rq finds the next request after @prev in elevator order.
|
2005-04-16 16:20:36 -06:00
|
|
|
* this with as_choose_req form the basis for how the scheduler chooses
|
|
|
|
* what request to process next. Anticipation works on top of this.
|
|
|
|
*/
|
2006-07-18 13:07:29 -06:00
|
|
|
static struct request *
|
|
|
|
as_find_next_rq(struct as_data *ad, struct request *last)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct rb_node *rbnext = rb_next(&last->rb_node);
|
|
|
|
struct rb_node *rbprev = rb_prev(&last->rb_node);
|
2006-07-18 13:07:29 -06:00
|
|
|
struct request *next = NULL, *prev = NULL;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:06:01 -06:00
|
|
|
BUG_ON(RB_EMPTY_NODE(&last->rb_node));
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (rbprev)
|
2006-07-18 13:07:29 -06:00
|
|
|
prev = rb_entry_rq(rbprev);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (rbnext)
|
2006-07-18 13:07:29 -06:00
|
|
|
next = rb_entry_rq(rbnext);
|
2005-04-16 16:20:36 -06:00
|
|
|
else {
|
2006-07-28 01:26:13 -06:00
|
|
|
const int data_dir = rq_is_sync(last);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:06:01 -06:00
|
|
|
rbnext = rb_first(&ad->sort_list[data_dir]);
|
|
|
|
if (rbnext && rbnext != &last->rb_node)
|
2006-07-18 13:07:29 -06:00
|
|
|
next = rb_entry_rq(rbnext);
|
2006-07-18 13:06:01 -06:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:06:01 -06:00
|
|
|
return as_choose_req(ad, next, prev);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* anticipatory scheduling functions follow
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_antic_expired tells us when we have anticipated too long.
|
|
|
|
* The funny "absolute difference" math on the elapsed time is to handle
|
|
|
|
* jiffy wraps, and disks which have been idle for 0x80000000 jiffies.
|
|
|
|
*/
|
|
|
|
static int as_antic_expired(struct as_data *ad)
|
|
|
|
{
|
|
|
|
long delta_jif;
|
|
|
|
|
|
|
|
delta_jif = jiffies - ad->antic_start;
|
|
|
|
if (unlikely(delta_jif < 0))
|
|
|
|
delta_jif = -delta_jif;
|
|
|
|
if (delta_jif < ad->antic_expire)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_antic_waitnext starts anticipating that a nice request will soon be
|
|
|
|
* submitted. See also as_antic_waitreq
|
|
|
|
*/
|
|
|
|
static void as_antic_waitnext(struct as_data *ad)
|
|
|
|
{
|
|
|
|
unsigned long timeout;
|
|
|
|
|
|
|
|
BUG_ON(ad->antic_status != ANTIC_OFF
|
|
|
|
&& ad->antic_status != ANTIC_WAIT_REQ);
|
|
|
|
|
|
|
|
timeout = ad->antic_start + ad->antic_expire;
|
|
|
|
|
|
|
|
mod_timer(&ad->antic_timer, timeout);
|
|
|
|
|
|
|
|
ad->antic_status = ANTIC_WAIT_NEXT;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_antic_waitreq starts anticipating. We don't start timing the anticipation
|
|
|
|
* until the request that we're anticipating on has finished. This means we
|
|
|
|
* are timing from when the candidate process wakes up hopefully.
|
|
|
|
*/
|
|
|
|
static void as_antic_waitreq(struct as_data *ad)
|
|
|
|
{
|
|
|
|
BUG_ON(ad->antic_status == ANTIC_FINISHED);
|
|
|
|
if (ad->antic_status == ANTIC_OFF) {
|
|
|
|
if (!ad->io_context || ad->ioc_finished)
|
|
|
|
as_antic_waitnext(ad);
|
|
|
|
else
|
|
|
|
ad->antic_status = ANTIC_WAIT_REQ;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This is called directly by the functions in this file to stop anticipation.
|
|
|
|
* We kill the timer and schedule a call to the request_fn asap.
|
|
|
|
*/
|
|
|
|
static void as_antic_stop(struct as_data *ad)
|
|
|
|
{
|
|
|
|
int status = ad->antic_status;
|
|
|
|
|
|
|
|
if (status == ANTIC_WAIT_REQ || status == ANTIC_WAIT_NEXT) {
|
|
|
|
if (status == ANTIC_WAIT_NEXT)
|
|
|
|
del_timer(&ad->antic_timer);
|
|
|
|
ad->antic_status = ANTIC_FINISHED;
|
|
|
|
/* see as_work_handler */
|
2008-07-28 05:08:45 -06:00
|
|
|
kblockd_schedule_work(ad->q, &ad->antic_work);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_antic_timeout is the timer function set by as_antic_waitnext.
|
|
|
|
*/
|
|
|
|
static void as_antic_timeout(unsigned long data)
|
|
|
|
{
|
|
|
|
struct request_queue *q = (struct request_queue *)data;
|
|
|
|
struct as_data *ad = q->elevator->elevator_data;
|
|
|
|
unsigned long flags;
|
|
|
|
|
|
|
|
spin_lock_irqsave(q->queue_lock, flags);
|
|
|
|
if (ad->antic_status == ANTIC_WAIT_REQ
|
|
|
|
|| ad->antic_status == ANTIC_WAIT_NEXT) {
|
2008-01-21 12:03:12 -07:00
|
|
|
struct as_io_context *aic;
|
|
|
|
spin_lock(&ad->io_context->lock);
|
|
|
|
aic = ad->io_context->aic;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
ad->antic_status = ANTIC_FINISHED;
|
2008-07-28 05:08:45 -06:00
|
|
|
kblockd_schedule_work(q, &ad->antic_work);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (aic->ttime_samples == 0) {
|
2005-11-07 01:59:53 -07:00
|
|
|
/* process anticipated on has exited or timed out*/
|
2005-04-16 16:20:36 -06:00
|
|
|
ad->exit_prob = (7*ad->exit_prob + 256)/8;
|
|
|
|
}
|
2005-11-07 01:59:53 -07:00
|
|
|
if (!test_bit(AS_TASK_RUNNING, &aic->state)) {
|
|
|
|
/* process not "saved" by a cooperating request */
|
|
|
|
ad->exit_no_coop = (7*ad->exit_no_coop + 256)/8;
|
|
|
|
}
|
2008-01-21 12:03:12 -07:00
|
|
|
spin_unlock(&ad->io_context->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
spin_unlock_irqrestore(q->queue_lock, flags);
|
|
|
|
}
|
|
|
|
|
2005-11-07 01:59:53 -07:00
|
|
|
static void as_update_thinktime(struct as_data *ad, struct as_io_context *aic,
|
|
|
|
unsigned long ttime)
|
|
|
|
{
|
|
|
|
/* fixed point: 1.0 == 1<<8 */
|
|
|
|
if (aic->ttime_samples == 0) {
|
|
|
|
ad->new_ttime_total = (7*ad->new_ttime_total + 256*ttime) / 8;
|
|
|
|
ad->new_ttime_mean = ad->new_ttime_total / 256;
|
|
|
|
|
|
|
|
ad->exit_prob = (7*ad->exit_prob)/8;
|
|
|
|
}
|
|
|
|
aic->ttime_samples = (7*aic->ttime_samples + 256) / 8;
|
|
|
|
aic->ttime_total = (7*aic->ttime_total + 256*ttime) / 8;
|
|
|
|
aic->ttime_mean = (aic->ttime_total + 128) / aic->ttime_samples;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void as_update_seekdist(struct as_data *ad, struct as_io_context *aic,
|
|
|
|
sector_t sdist)
|
|
|
|
{
|
|
|
|
u64 total;
|
|
|
|
|
|
|
|
if (aic->seek_samples == 0) {
|
|
|
|
ad->new_seek_total = (7*ad->new_seek_total + 256*(u64)sdist)/8;
|
|
|
|
ad->new_seek_mean = ad->new_seek_total / 256;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Don't allow the seek distance to get too large from the
|
|
|
|
* odd fragment, pagein, etc
|
|
|
|
*/
|
|
|
|
if (aic->seek_samples <= 60) /* second&third seek */
|
|
|
|
sdist = min(sdist, (aic->seek_mean * 4) + 2*1024*1024);
|
|
|
|
else
|
|
|
|
sdist = min(sdist, (aic->seek_mean * 4) + 2*1024*64);
|
|
|
|
|
|
|
|
aic->seek_samples = (7*aic->seek_samples + 256) / 8;
|
|
|
|
aic->seek_total = (7*aic->seek_total + (u64)256*sdist) / 8;
|
|
|
|
total = aic->seek_total + (aic->seek_samples/2);
|
|
|
|
do_div(total, aic->seek_samples);
|
|
|
|
aic->seek_mean = (sector_t)total;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_update_iohist keeps a decaying histogram of IO thinktimes, and
|
|
|
|
* updates @aic->ttime_mean based on that. It is called when a new
|
|
|
|
* request is queued.
|
|
|
|
*/
|
|
|
|
static void as_update_iohist(struct as_data *ad, struct as_io_context *aic,
|
|
|
|
struct request *rq)
|
|
|
|
{
|
2006-07-28 01:26:13 -06:00
|
|
|
int data_dir = rq_is_sync(rq);
|
2005-11-07 01:59:53 -07:00
|
|
|
unsigned long thinktime = 0;
|
|
|
|
sector_t seek_dist;
|
|
|
|
|
|
|
|
if (aic == NULL)
|
|
|
|
return;
|
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
if (data_dir == BLK_RW_SYNC) {
|
2005-11-07 01:59:53 -07:00
|
|
|
unsigned long in_flight = atomic_read(&aic->nr_queued)
|
|
|
|
+ atomic_read(&aic->nr_dispatched);
|
|
|
|
spin_lock(&aic->lock);
|
|
|
|
if (test_bit(AS_TASK_IORUNNING, &aic->state) ||
|
|
|
|
test_bit(AS_TASK_IOSTARTED, &aic->state)) {
|
|
|
|
/* Calculate read -> read thinktime */
|
|
|
|
if (test_bit(AS_TASK_IORUNNING, &aic->state)
|
|
|
|
&& in_flight == 0) {
|
|
|
|
thinktime = jiffies - aic->last_end_request;
|
|
|
|
thinktime = min(thinktime, MAX_THINKTIME-1);
|
|
|
|
}
|
|
|
|
as_update_thinktime(ad, aic, thinktime);
|
|
|
|
|
|
|
|
/* Calculate read -> read seek distance */
|
2009-05-07 07:24:39 -06:00
|
|
|
if (aic->last_request_pos < blk_rq_pos(rq))
|
|
|
|
seek_dist = blk_rq_pos(rq) -
|
|
|
|
aic->last_request_pos;
|
2005-11-07 01:59:53 -07:00
|
|
|
else
|
2009-05-07 07:24:39 -06:00
|
|
|
seek_dist = aic->last_request_pos -
|
|
|
|
blk_rq_pos(rq);
|
2005-11-07 01:59:53 -07:00
|
|
|
as_update_seekdist(ad, aic, seek_dist);
|
|
|
|
}
|
2009-05-07 07:24:39 -06:00
|
|
|
aic->last_request_pos = blk_rq_pos(rq) + blk_rq_sectors(rq);
|
2005-11-07 01:59:53 -07:00
|
|
|
set_bit(AS_TASK_IOSTARTED, &aic->state);
|
|
|
|
spin_unlock(&aic->lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* as_close_req decides if one request is considered "close" to the
|
|
|
|
* previous one issued.
|
|
|
|
*/
|
2005-11-07 01:59:53 -07:00
|
|
|
static int as_close_req(struct as_data *ad, struct as_io_context *aic,
|
2006-07-18 13:07:29 -06:00
|
|
|
struct request *rq)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2007-05-08 01:26:34 -06:00
|
|
|
unsigned long delay; /* jiffies */
|
2005-04-16 16:20:36 -06:00
|
|
|
sector_t last = ad->last_sector[ad->batch_data_dir];
|
2009-05-07 07:24:39 -06:00
|
|
|
sector_t next = blk_rq_pos(rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
sector_t delta; /* acceptable close offset (in sectors) */
|
2005-11-07 01:59:53 -07:00
|
|
|
sector_t s;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (ad->antic_status == ANTIC_OFF || !ad->ioc_finished)
|
|
|
|
delay = 0;
|
|
|
|
else
|
2007-05-08 01:26:34 -06:00
|
|
|
delay = jiffies - ad->antic_start;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2005-11-07 01:59:53 -07:00
|
|
|
if (delay == 0)
|
|
|
|
delta = 8192;
|
2007-05-08 01:26:34 -06:00
|
|
|
else if (delay <= (20 * HZ / 1000) && delay <= ad->antic_expire)
|
2005-11-07 01:59:53 -07:00
|
|
|
delta = 8192 << delay;
|
2005-04-16 16:20:36 -06:00
|
|
|
else
|
|
|
|
return 1;
|
|
|
|
|
2005-11-07 01:59:53 -07:00
|
|
|
if ((last <= next + (delta>>1)) && (next <= last + delta))
|
|
|
|
return 1;
|
|
|
|
|
|
|
|
if (last < next)
|
|
|
|
s = next - last;
|
|
|
|
else
|
|
|
|
s = last - next;
|
|
|
|
|
|
|
|
if (aic->seek_samples == 0) {
|
|
|
|
/*
|
|
|
|
* Process has just started IO. Use past statistics to
|
|
|
|
* gauge success possibility
|
|
|
|
*/
|
|
|
|
if (ad->new_seek_mean > s) {
|
|
|
|
/* this request is better than what we're expecting */
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
} else {
|
|
|
|
if (aic->seek_mean > s) {
|
|
|
|
/* this request is better than what we're expecting */
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_can_break_anticipation returns true if we have been anticipating this
|
|
|
|
* request.
|
|
|
|
*
|
|
|
|
* It also returns true if the process against which we are anticipating
|
|
|
|
* submits a write - that's presumably an fsync, O_SYNC write, etc. We want to
|
|
|
|
* dispatch it ASAP, because we know that application will not be submitting
|
|
|
|
* any new reads.
|
|
|
|
*
|
2005-11-07 01:59:53 -07:00
|
|
|
* If the task which has submitted the request has exited, break anticipation.
|
2005-04-16 16:20:36 -06:00
|
|
|
*
|
|
|
|
* If this task has queued some other IO, do not enter enticipation.
|
|
|
|
*/
|
2006-07-18 13:07:29 -06:00
|
|
|
static int as_can_break_anticipation(struct as_data *ad, struct request *rq)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct io_context *ioc;
|
|
|
|
struct as_io_context *aic;
|
|
|
|
|
|
|
|
ioc = ad->io_context;
|
|
|
|
BUG_ON(!ioc);
|
2008-01-21 12:03:12 -07:00
|
|
|
spin_lock(&ioc->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
if (rq && ioc == RQ_IOC(rq)) {
|
2005-04-16 16:20:36 -06:00
|
|
|
/* request from same process */
|
2008-01-21 12:03:12 -07:00
|
|
|
spin_unlock(&ioc->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ad->ioc_finished && as_antic_expired(ad)) {
|
|
|
|
/*
|
|
|
|
* In this situation status should really be FINISHED,
|
|
|
|
* however the timer hasn't had the chance to run yet.
|
|
|
|
*/
|
2008-01-21 12:03:12 -07:00
|
|
|
spin_unlock(&ioc->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
aic = ioc->aic;
|
2008-01-21 12:03:12 -07:00
|
|
|
if (!aic) {
|
|
|
|
spin_unlock(&ioc->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
return 0;
|
2008-01-21 12:03:12 -07:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (atomic_read(&aic->nr_queued) > 0) {
|
|
|
|
/* process has more requests queued */
|
2008-01-21 12:03:12 -07:00
|
|
|
spin_unlock(&ioc->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (atomic_read(&aic->nr_dispatched) > 0) {
|
|
|
|
/* process has more requests dispatched */
|
2008-01-21 12:03:12 -07:00
|
|
|
spin_unlock(&ioc->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
if (rq && rq_is_sync(rq) && as_close_req(ad, aic, rq)) {
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* Found a close request that is not one of ours.
|
|
|
|
*
|
2005-11-07 01:59:53 -07:00
|
|
|
* This makes close requests from another process update
|
|
|
|
* our IO history. Is generally useful when there are
|
2005-04-16 16:20:36 -06:00
|
|
|
* two or more cooperating processes working in the same
|
|
|
|
* area.
|
|
|
|
*/
|
2005-11-07 01:59:53 -07:00
|
|
|
if (!test_bit(AS_TASK_RUNNING, &aic->state)) {
|
|
|
|
if (aic->ttime_samples == 0)
|
|
|
|
ad->exit_prob = (7*ad->exit_prob + 256)/8;
|
|
|
|
|
|
|
|
ad->exit_no_coop = (7*ad->exit_no_coop)/8;
|
|
|
|
}
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
as_update_iohist(ad, aic, rq);
|
2008-01-21 12:03:12 -07:00
|
|
|
spin_unlock(&ioc->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2005-11-07 01:59:53 -07:00
|
|
|
if (!test_bit(AS_TASK_RUNNING, &aic->state)) {
|
|
|
|
/* process anticipated on has exited */
|
|
|
|
if (aic->ttime_samples == 0)
|
|
|
|
ad->exit_prob = (7*ad->exit_prob + 256)/8;
|
|
|
|
|
2008-01-21 12:03:12 -07:00
|
|
|
if (ad->exit_no_coop > 128) {
|
|
|
|
spin_unlock(&ioc->lock);
|
2005-11-07 01:59:53 -07:00
|
|
|
return 1;
|
2008-01-21 12:03:12 -07:00
|
|
|
}
|
2005-11-07 01:59:53 -07:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
if (aic->ttime_samples == 0) {
|
2008-01-21 12:03:12 -07:00
|
|
|
if (ad->new_ttime_mean > ad->antic_expire) {
|
|
|
|
spin_unlock(&ioc->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
return 1;
|
2008-01-21 12:03:12 -07:00
|
|
|
}
|
|
|
|
if (ad->exit_prob * ad->exit_no_coop > 128*256) {
|
|
|
|
spin_unlock(&ioc->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
return 1;
|
2008-01-21 12:03:12 -07:00
|
|
|
}
|
2005-04-16 16:20:36 -06:00
|
|
|
} else if (aic->ttime_mean > ad->antic_expire) {
|
|
|
|
/* the process thinks too much between requests */
|
2008-01-21 12:03:12 -07:00
|
|
|
spin_unlock(&ioc->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
return 1;
|
|
|
|
}
|
2008-01-21 12:03:12 -07:00
|
|
|
spin_unlock(&ioc->lock);
|
2005-04-16 16:20:36 -06:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2006-07-18 13:07:29 -06:00
|
|
|
* as_can_anticipate indicates whether we should either run rq
|
2005-04-16 16:20:36 -06:00
|
|
|
* or keep anticipating a better request.
|
|
|
|
*/
|
2006-07-18 13:07:29 -06:00
|
|
|
static int as_can_anticipate(struct as_data *ad, struct request *rq)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2008-09-25 03:37:50 -06:00
|
|
|
#if 0 /* disable for now, we need to check tag level as well */
|
2008-09-24 05:03:33 -06:00
|
|
|
/*
|
|
|
|
* SSD device without seek penalty, disable idling
|
|
|
|
*/
|
2008-09-25 03:37:50 -06:00
|
|
|
if (blk_queue_nonrot(ad->q)) axman
|
2008-09-24 05:03:33 -06:00
|
|
|
return 0;
|
2008-09-25 03:37:50 -06:00
|
|
|
#endif
|
2008-09-24 05:03:33 -06:00
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
if (!ad->io_context)
|
|
|
|
/*
|
|
|
|
* Last request submitted was a write
|
|
|
|
*/
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
if (ad->antic_status == ANTIC_FINISHED)
|
|
|
|
/*
|
|
|
|
* Don't restart if we have just finished. Run the next request
|
|
|
|
*/
|
|
|
|
return 0;
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
if (as_can_break_anticipation(ad, rq))
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* This request is a good candidate. Don't keep anticipating,
|
|
|
|
* run it.
|
|
|
|
*/
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* OK from here, we haven't finished, and don't have a decent request!
|
|
|
|
* Status is either ANTIC_OFF so start waiting,
|
|
|
|
* ANTIC_WAIT_REQ so continue waiting for request to finish
|
|
|
|
* or ANTIC_WAIT_NEXT so continue waiting for an acceptable request.
|
|
|
|
*/
|
|
|
|
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2006-07-18 13:07:29 -06:00
|
|
|
* as_update_rq must be called whenever a request (rq) is added to
|
2005-04-16 16:20:36 -06:00
|
|
|
* the sort_list. This function keeps caches up to date, and checks if the
|
|
|
|
* request might be one we are "anticipating"
|
|
|
|
*/
|
2006-07-18 13:07:29 -06:00
|
|
|
static void as_update_rq(struct as_data *ad, struct request *rq)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2006-07-18 13:07:29 -06:00
|
|
|
const int data_dir = rq_is_sync(rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
/* keep the next_rq cache up to date */
|
|
|
|
ad->next_rq[data_dir] = as_choose_req(ad, rq, ad->next_rq[data_dir]);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
/*
|
|
|
|
* have we been anticipating this request?
|
|
|
|
* or does it come from the same process as the one we are anticipating
|
|
|
|
* for?
|
|
|
|
*/
|
|
|
|
if (ad->antic_status == ANTIC_WAIT_REQ
|
|
|
|
|| ad->antic_status == ANTIC_WAIT_NEXT) {
|
2006-07-18 13:07:29 -06:00
|
|
|
if (as_can_break_anticipation(ad, rq))
|
2005-04-16 16:20:36 -06:00
|
|
|
as_antic_stop(ad);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Gathers timings and resizes the write batch automatically
|
|
|
|
*/
|
|
|
|
static void update_write_batch(struct as_data *ad)
|
|
|
|
{
|
2009-04-08 03:02:08 -06:00
|
|
|
unsigned long batch = ad->batch_expire[BLK_RW_ASYNC];
|
2005-04-16 16:20:36 -06:00
|
|
|
long write_time;
|
|
|
|
|
|
|
|
write_time = (jiffies - ad->current_batch_expires) + batch;
|
|
|
|
if (write_time < 0)
|
|
|
|
write_time = 0;
|
|
|
|
|
|
|
|
if (write_time > batch && !ad->write_batch_idled) {
|
|
|
|
if (write_time > batch * 3)
|
|
|
|
ad->write_batch_count /= 2;
|
|
|
|
else
|
|
|
|
ad->write_batch_count--;
|
|
|
|
} else if (write_time < batch && ad->current_write_count == 0) {
|
|
|
|
if (batch > write_time * 3)
|
|
|
|
ad->write_batch_count *= 2;
|
|
|
|
else
|
|
|
|
ad->write_batch_count++;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ad->write_batch_count < 1)
|
|
|
|
ad->write_batch_count = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_completed_request is to be called when a request has completed and
|
|
|
|
* returned something to the requesting process, be it an error or data.
|
|
|
|
*/
|
2007-07-24 01:28:11 -06:00
|
|
|
static void as_completed_request(struct request_queue *q, struct request *rq)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct as_data *ad = q->elevator->elevator_data;
|
|
|
|
|
|
|
|
WARN_ON(!list_empty(&rq->queuelist));
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
if (RQ_STATE(rq) != AS_RQ_REMOVED) {
|
2008-07-25 20:45:38 -06:00
|
|
|
WARN(1, "rq->state %d\n", RQ_STATE(rq));
|
2005-04-16 16:20:36 -06:00
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ad->changed_batch && ad->nr_dispatched == 1) {
|
block: Fix the starving writes bug in the anticipatory IO scheduler
AS scheduler alternates between issuing read and write batches. It does
the batch switch only after all requests from the previous batch are
completed.
When switching to a write batch, if there is an on-going read request,
it waits for its completion and indicates its intention of switching by
setting ad->changed_batch and the new direction but does not update the
batch_expire_time for the new write batch which it does in the case of
no previous pending requests.
On completion of the read request, it sees that we were waiting for the
switch and schedules work for kblockd right away and resets the
ad->changed_data flag.
Now when kblockd enters dispatch_request where it is expected to pick
up a write request, it in turn ends the write batch because the
batch_expire_timer was not updated and shows the expire timestamp for
the previous batch.
This results in the write starvation for all the cases where there is
the intention for switching to a write batch, but there is a previous
in-flight read request and the batch gets reverted to a read_batch
right away.
This also holds true in the reverse case (switching from a write batch
to a read batch with an in-flight write request).
I've checked that this bug exists on 2.6.11, 2.6.18, 2.6.24 and
linux-2.6-block git HEAD. I've tested the fix on x86 platforms with
SCSI drives where the driver asks for the next request while a current
request is in-flight.
This patch is based off linux-2.6-block git HEAD.
Bug reproduction:
A simple scenario which reproduces this bug is:
- dd if=/dev/hda3 of=/dev/null &
- lilo
The lilo takes forever to complete.
This can also be reproduced fairly easily with the earlier dd and
another test
program doing msync().
The example test program below should print out a message after every
iteration
but it simply hangs forever. With this bugfix it makes forward progress.
====
Example test program using msync() (thanks to suleiman AT google DOT
com)
inline uint64_t
rdtsc(void)
{
int64_t tsc;
__asm __volatile("rdtsc" : "=A" (tsc));
return (tsc);
}
int
main(int argc, char **argv)
{
struct stat st;
uint64_t e, s, t;
char *p, q;
long i;
int fd;
if (argc < 2) {
printf("Usage: %s <file>\n", argv[0]);
return (1);
}
if ((fd = open(argv[1], O_RDWR | O_NOATIME)) < 0)
err(1, "open");
if (fstat(fd, &st) < 0)
err(1, "fstat");
p = mmap(NULL, st.st_size, PROT_READ | PROT_WRITE,
MAP_SHARED, fd, 0);
t = 0;
for (i = 0; i < 1000; i++) {
*p = 0;
msync(p, 4096, MS_SYNC);
s = rdtsc();
*p = 0;
__asm __volatile(""::: "memory");
e = rdtsc();
if (argc > 2)
printf("%d: %lld cycles %jd %jd\n",
i, e - s, (intmax_t)s, (intmax_t)e);
t += e - s;
}
printf("average time: %lld cycles\n", t / 1000);
return (0);
}
Cc: <stable@kernel.org>
Acked-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
2008-06-16 10:37:08 -06:00
|
|
|
ad->current_batch_expires = jiffies +
|
|
|
|
ad->batch_expire[ad->batch_data_dir];
|
2008-07-28 05:08:45 -06:00
|
|
|
kblockd_schedule_work(q, &ad->antic_work);
|
2005-04-16 16:20:36 -06:00
|
|
|
ad->changed_batch = 0;
|
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
if (ad->batch_data_dir == BLK_RW_SYNC)
|
2005-04-16 16:20:36 -06:00
|
|
|
ad->new_batch = 1;
|
|
|
|
}
|
|
|
|
WARN_ON(ad->nr_dispatched == 0);
|
|
|
|
ad->nr_dispatched--;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Start counting the batch from when a request of that direction is
|
|
|
|
* actually serviced. This should help devices with big TCQ windows
|
|
|
|
* and writeback caches
|
|
|
|
*/
|
2006-07-28 01:26:13 -06:00
|
|
|
if (ad->new_batch && ad->batch_data_dir == rq_is_sync(rq)) {
|
2005-04-16 16:20:36 -06:00
|
|
|
update_write_batch(ad);
|
|
|
|
ad->current_batch_expires = jiffies +
|
2009-04-08 03:02:08 -06:00
|
|
|
ad->batch_expire[BLK_RW_SYNC];
|
2005-04-16 16:20:36 -06:00
|
|
|
ad->new_batch = 0;
|
|
|
|
}
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
if (ad->io_context == RQ_IOC(rq) && ad->io_context) {
|
2005-04-16 16:20:36 -06:00
|
|
|
ad->antic_start = jiffies;
|
|
|
|
ad->ioc_finished = 1;
|
|
|
|
if (ad->antic_status == ANTIC_WAIT_REQ) {
|
|
|
|
/*
|
|
|
|
* We were waiting on this request, now anticipate
|
|
|
|
* the next one
|
|
|
|
*/
|
|
|
|
as_antic_waitnext(ad);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
as_put_io_context(rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
out:
|
2006-07-18 13:07:29 -06:00
|
|
|
RQ_SET_STATE(rq, AS_RQ_POSTSCHED);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_remove_queued_request removes a request from the pre dispatch queue
|
|
|
|
* without updating refcounts. It is expected the caller will drop the
|
|
|
|
* reference unless it replaces the request at somepart of the elevator
|
|
|
|
* (ie. the dispatch queue)
|
|
|
|
*/
|
2007-07-24 01:28:11 -06:00
|
|
|
static void as_remove_queued_request(struct request_queue *q,
|
|
|
|
struct request *rq)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2006-07-28 01:26:13 -06:00
|
|
|
const int data_dir = rq_is_sync(rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
struct as_data *ad = q->elevator->elevator_data;
|
2006-07-18 13:07:29 -06:00
|
|
|
struct io_context *ioc;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
WARN_ON(RQ_STATE(rq) != AS_RQ_QUEUED);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
ioc = RQ_IOC(rq);
|
|
|
|
if (ioc && ioc->aic) {
|
|
|
|
BUG_ON(!atomic_read(&ioc->aic->nr_queued));
|
|
|
|
atomic_dec(&ioc->aic->nr_queued);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2006-07-18 13:07:29 -06:00
|
|
|
* Update the "next_rq" cache if we are about to remove its
|
2005-04-16 16:20:36 -06:00
|
|
|
* entry
|
|
|
|
*/
|
2006-07-18 13:07:29 -06:00
|
|
|
if (ad->next_rq[data_dir] == rq)
|
|
|
|
ad->next_rq[data_dir] = as_find_next_rq(ad, rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-13 01:12:14 -06:00
|
|
|
rq_fifo_clear(rq);
|
2006-07-18 13:07:29 -06:00
|
|
|
as_del_rq_rb(ad, rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2007-12-05 03:06:50 -07:00
|
|
|
* as_fifo_expired returns 0 if there are no expired requests on the fifo,
|
2005-04-16 16:20:36 -06:00
|
|
|
* 1 otherwise. It is ratelimited so that we only perform the check once per
|
|
|
|
* `fifo_expire' interval. Otherwise a large number of expired requests
|
|
|
|
* would create a hopeless seekstorm.
|
|
|
|
*
|
|
|
|
* See as_antic_expired comment.
|
|
|
|
*/
|
|
|
|
static int as_fifo_expired(struct as_data *ad, int adir)
|
|
|
|
{
|
2006-07-13 01:12:14 -06:00
|
|
|
struct request *rq;
|
2005-04-16 16:20:36 -06:00
|
|
|
long delta_jif;
|
|
|
|
|
|
|
|
delta_jif = jiffies - ad->last_check_fifo[adir];
|
|
|
|
if (unlikely(delta_jif < 0))
|
|
|
|
delta_jif = -delta_jif;
|
|
|
|
if (delta_jif < ad->fifo_expire[adir])
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
ad->last_check_fifo[adir] = jiffies;
|
|
|
|
|
|
|
|
if (list_empty(&ad->fifo_list[adir]))
|
|
|
|
return 0;
|
|
|
|
|
2006-07-13 01:12:14 -06:00
|
|
|
rq = rq_entry_fifo(ad->fifo_list[adir].next);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-13 01:12:14 -06:00
|
|
|
return time_after(jiffies, rq_fifo_time(rq));
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_batch_expired returns true if the current batch has expired. A batch
|
|
|
|
* is a set of reads or a set of writes.
|
|
|
|
*/
|
|
|
|
static inline int as_batch_expired(struct as_data *ad)
|
|
|
|
{
|
|
|
|
if (ad->changed_batch || ad->new_batch)
|
|
|
|
return 0;
|
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
if (ad->batch_data_dir == BLK_RW_SYNC)
|
2005-04-16 16:20:36 -06:00
|
|
|
/* TODO! add a check so a complete fifo gets written? */
|
|
|
|
return time_after(jiffies, ad->current_batch_expires);
|
|
|
|
|
|
|
|
return time_after(jiffies, ad->current_batch_expires)
|
|
|
|
|| ad->current_write_count == 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* move an entry to dispatch queue
|
|
|
|
*/
|
2006-07-18 13:07:29 -06:00
|
|
|
static void as_move_to_dispatch(struct as_data *ad, struct request *rq)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2006-07-28 01:26:13 -06:00
|
|
|
const int data_dir = rq_is_sync(rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:06:01 -06:00
|
|
|
BUG_ON(RB_EMPTY_NODE(&rq->rb_node));
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
as_antic_stop(ad);
|
|
|
|
ad->antic_status = ANTIC_OFF;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This has to be set in order to be correctly updated by
|
2006-07-18 13:07:29 -06:00
|
|
|
* as_find_next_rq
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
2009-05-07 07:24:39 -06:00
|
|
|
ad->last_sector[data_dir] = blk_rq_pos(rq) + blk_rq_sectors(rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
if (data_dir == BLK_RW_SYNC) {
|
2006-07-18 13:07:29 -06:00
|
|
|
struct io_context *ioc = RQ_IOC(rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
/* In case we have to anticipate after this */
|
2006-07-18 13:07:29 -06:00
|
|
|
copy_io_context(&ad->io_context, &ioc);
|
2005-04-16 16:20:36 -06:00
|
|
|
} else {
|
|
|
|
if (ad->io_context) {
|
|
|
|
put_io_context(ad->io_context);
|
|
|
|
ad->io_context = NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (ad->current_write_count != 0)
|
|
|
|
ad->current_write_count--;
|
|
|
|
}
|
|
|
|
ad->ioc_finished = 0;
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
ad->next_rq[data_dir] = as_find_next_rq(ad, rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
/*
|
|
|
|
* take it off the sort and fifo list, add to dispatch queue
|
|
|
|
*/
|
|
|
|
as_remove_queued_request(ad->q, rq);
|
2006-07-18 13:07:29 -06:00
|
|
|
WARN_ON(RQ_STATE(rq) != AS_RQ_QUEUED);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2005-10-20 08:42:29 -06:00
|
|
|
elv_dispatch_sort(ad->q, rq);
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
RQ_SET_STATE(rq, AS_RQ_DISPATCHED);
|
|
|
|
if (RQ_IOC(rq) && RQ_IOC(rq)->aic)
|
|
|
|
atomic_inc(&RQ_IOC(rq)->aic->nr_dispatched);
|
2005-04-16 16:20:36 -06:00
|
|
|
ad->nr_dispatched++;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_dispatch_request selects the best request according to
|
|
|
|
* read/write expire, batch expire, etc, and moves it to the dispatch
|
|
|
|
* queue. Returns 1 if a request was found, 0 otherwise.
|
|
|
|
*/
|
2007-07-24 01:28:11 -06:00
|
|
|
static int as_dispatch_request(struct request_queue *q, int force)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2005-10-20 08:42:29 -06:00
|
|
|
struct as_data *ad = q->elevator->elevator_data;
|
2009-04-08 03:02:08 -06:00
|
|
|
const int reads = !list_empty(&ad->fifo_list[BLK_RW_SYNC]);
|
|
|
|
const int writes = !list_empty(&ad->fifo_list[BLK_RW_ASYNC]);
|
2006-07-18 13:07:29 -06:00
|
|
|
struct request *rq;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2005-10-20 08:42:29 -06:00
|
|
|
if (unlikely(force)) {
|
|
|
|
/*
|
|
|
|
* Forced dispatch, accounting is useless. Reset
|
|
|
|
* accounting states and dump fifo_lists. Note that
|
2009-04-08 03:02:08 -06:00
|
|
|
* batch_data_dir is reset to BLK_RW_SYNC to avoid
|
2005-10-20 08:42:29 -06:00
|
|
|
* screwing write batch accounting as write batch
|
|
|
|
* accounting occurs on W->R transition.
|
|
|
|
*/
|
|
|
|
int dispatched = 0;
|
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
ad->batch_data_dir = BLK_RW_SYNC;
|
2005-10-20 08:42:29 -06:00
|
|
|
ad->changed_batch = 0;
|
|
|
|
ad->new_batch = 0;
|
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
while (ad->next_rq[BLK_RW_SYNC]) {
|
|
|
|
as_move_to_dispatch(ad, ad->next_rq[BLK_RW_SYNC]);
|
2005-10-20 08:42:29 -06:00
|
|
|
dispatched++;
|
|
|
|
}
|
2009-04-08 03:02:08 -06:00
|
|
|
ad->last_check_fifo[BLK_RW_SYNC] = jiffies;
|
2005-10-20 08:42:29 -06:00
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
while (ad->next_rq[BLK_RW_ASYNC]) {
|
|
|
|
as_move_to_dispatch(ad, ad->next_rq[BLK_RW_ASYNC]);
|
2005-10-20 08:42:29 -06:00
|
|
|
dispatched++;
|
|
|
|
}
|
2009-04-08 03:02:08 -06:00
|
|
|
ad->last_check_fifo[BLK_RW_ASYNC] = jiffies;
|
2005-10-20 08:42:29 -06:00
|
|
|
|
|
|
|
return dispatched;
|
|
|
|
}
|
|
|
|
|
2005-04-16 16:20:36 -06:00
|
|
|
/* Signal that the write batch was uncontended, so we can't time it */
|
2009-04-08 03:02:08 -06:00
|
|
|
if (ad->batch_data_dir == BLK_RW_ASYNC && !reads) {
|
2005-04-16 16:20:36 -06:00
|
|
|
if (ad->current_write_count == 0 || !writes)
|
|
|
|
ad->write_batch_idled = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!(reads || writes)
|
|
|
|
|| ad->antic_status == ANTIC_WAIT_REQ
|
|
|
|
|| ad->antic_status == ANTIC_WAIT_NEXT
|
|
|
|
|| ad->changed_batch)
|
|
|
|
return 0;
|
|
|
|
|
2005-11-07 01:59:53 -07:00
|
|
|
if (!(reads && writes && as_batch_expired(ad))) {
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* batch is still running or no reads or no writes
|
|
|
|
*/
|
2006-07-18 13:07:29 -06:00
|
|
|
rq = ad->next_rq[ad->batch_data_dir];
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
if (ad->batch_data_dir == BLK_RW_SYNC && ad->antic_expire) {
|
|
|
|
if (as_fifo_expired(ad, BLK_RW_SYNC))
|
2005-04-16 16:20:36 -06:00
|
|
|
goto fifo_expired;
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
if (as_can_anticipate(ad, rq)) {
|
2005-04-16 16:20:36 -06:00
|
|
|
as_antic_waitreq(ad);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
if (rq) {
|
2005-04-16 16:20:36 -06:00
|
|
|
/* we have a "next request" */
|
|
|
|
if (reads && !writes)
|
|
|
|
ad->current_batch_expires =
|
2009-04-08 03:02:08 -06:00
|
|
|
jiffies + ad->batch_expire[BLK_RW_SYNC];
|
2005-04-16 16:20:36 -06:00
|
|
|
goto dispatch_request;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* at this point we are not running a batch. select the appropriate
|
|
|
|
* data direction (read / write)
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (reads) {
|
2009-04-08 03:02:08 -06:00
|
|
|
BUG_ON(RB_EMPTY_ROOT(&ad->sort_list[BLK_RW_SYNC]));
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
if (writes && ad->batch_data_dir == BLK_RW_SYNC)
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* Last batch was a read, switch to writes
|
|
|
|
*/
|
|
|
|
goto dispatch_writes;
|
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
if (ad->batch_data_dir == BLK_RW_ASYNC) {
|
2005-04-16 16:20:36 -06:00
|
|
|
WARN_ON(ad->new_batch);
|
|
|
|
ad->changed_batch = 1;
|
|
|
|
}
|
2009-04-08 03:02:08 -06:00
|
|
|
ad->batch_data_dir = BLK_RW_SYNC;
|
|
|
|
rq = rq_entry_fifo(ad->fifo_list[BLK_RW_SYNC].next);
|
2005-04-16 16:20:36 -06:00
|
|
|
ad->last_check_fifo[ad->batch_data_dir] = jiffies;
|
|
|
|
goto dispatch_request;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* the last batch was a read
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (writes) {
|
|
|
|
dispatch_writes:
|
2009-04-08 03:02:08 -06:00
|
|
|
BUG_ON(RB_EMPTY_ROOT(&ad->sort_list[BLK_RW_ASYNC]));
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
if (ad->batch_data_dir == BLK_RW_SYNC) {
|
2005-04-16 16:20:36 -06:00
|
|
|
ad->changed_batch = 1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* new_batch might be 1 when the queue runs out of
|
|
|
|
* reads. A subsequent submission of a write might
|
|
|
|
* cause a change of batch before the read is finished.
|
|
|
|
*/
|
|
|
|
ad->new_batch = 0;
|
|
|
|
}
|
2009-04-08 03:02:08 -06:00
|
|
|
ad->batch_data_dir = BLK_RW_ASYNC;
|
2005-04-16 16:20:36 -06:00
|
|
|
ad->current_write_count = ad->write_batch_count;
|
|
|
|
ad->write_batch_idled = 0;
|
2009-04-08 03:02:08 -06:00
|
|
|
rq = rq_entry_fifo(ad->fifo_list[BLK_RW_ASYNC].next);
|
|
|
|
ad->last_check_fifo[BLK_RW_ASYNC] = jiffies;
|
2005-04-16 16:20:36 -06:00
|
|
|
goto dispatch_request;
|
|
|
|
}
|
|
|
|
|
|
|
|
BUG();
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
dispatch_request:
|
|
|
|
/*
|
|
|
|
* If a request has expired, service it.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (as_fifo_expired(ad, ad->batch_data_dir)) {
|
|
|
|
fifo_expired:
|
2006-07-18 13:07:29 -06:00
|
|
|
rq = rq_entry_fifo(ad->fifo_list[ad->batch_data_dir].next);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
if (ad->changed_batch) {
|
|
|
|
WARN_ON(ad->new_batch);
|
|
|
|
|
|
|
|
if (ad->nr_dispatched)
|
|
|
|
return 0;
|
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
if (ad->batch_data_dir == BLK_RW_ASYNC)
|
2005-04-16 16:20:36 -06:00
|
|
|
ad->current_batch_expires = jiffies +
|
2009-04-08 03:02:08 -06:00
|
|
|
ad->batch_expire[BLK_RW_ASYNC];
|
2005-04-16 16:20:36 -06:00
|
|
|
else
|
|
|
|
ad->new_batch = 1;
|
|
|
|
|
|
|
|
ad->changed_batch = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2006-07-18 13:07:29 -06:00
|
|
|
* rq is the selected appropriate request.
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
2006-07-18 13:07:29 -06:00
|
|
|
as_move_to_dispatch(ad, rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2006-07-18 13:07:29 -06:00
|
|
|
* add rq to rbtree and fifo
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
2007-07-24 01:28:11 -06:00
|
|
|
static void as_add_request(struct request_queue *q, struct request *rq)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2005-10-20 08:42:29 -06:00
|
|
|
struct as_data *ad = q->elevator->elevator_data;
|
2005-04-16 16:20:36 -06:00
|
|
|
int data_dir;
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
RQ_SET_STATE(rq, AS_RQ_NEW);
|
2005-10-20 08:42:29 -06:00
|
|
|
|
2006-07-28 01:26:13 -06:00
|
|
|
data_dir = rq_is_sync(rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-19 15:39:40 -06:00
|
|
|
rq->elevator_private = as_get_io_context(q->node);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
if (RQ_IOC(rq)) {
|
|
|
|
as_update_iohist(ad, RQ_IOC(rq)->aic, rq);
|
|
|
|
atomic_inc(&RQ_IOC(rq)->aic->nr_queued);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
as_add_rq_rb(ad, rq);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2005-11-11 06:27:09 -07:00
|
|
|
/*
|
2007-12-05 03:06:50 -07:00
|
|
|
* set expire time and add to fifo list
|
2005-11-11 06:27:09 -07:00
|
|
|
*/
|
2006-07-13 01:12:14 -06:00
|
|
|
rq_set_fifo_time(rq, jiffies + ad->fifo_expire[data_dir]);
|
|
|
|
list_add_tail(&rq->queuelist, &ad->fifo_list[data_dir]);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
as_update_rq(ad, rq); /* keep state machine up to date */
|
|
|
|
RQ_SET_STATE(rq, AS_RQ_QUEUED);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2007-07-24 01:28:11 -06:00
|
|
|
static void as_activate_request(struct request_queue *q, struct request *rq)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2006-07-18 13:07:29 -06:00
|
|
|
WARN_ON(RQ_STATE(rq) != AS_RQ_DISPATCHED);
|
|
|
|
RQ_SET_STATE(rq, AS_RQ_REMOVED);
|
|
|
|
if (RQ_IOC(rq) && RQ_IOC(rq)->aic)
|
|
|
|
atomic_dec(&RQ_IOC(rq)->aic->nr_dispatched);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2007-07-24 01:28:11 -06:00
|
|
|
static void as_deactivate_request(struct request_queue *q, struct request *rq)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2006-07-18 13:07:29 -06:00
|
|
|
WARN_ON(RQ_STATE(rq) != AS_RQ_REMOVED);
|
|
|
|
RQ_SET_STATE(rq, AS_RQ_DISPATCHED);
|
|
|
|
if (RQ_IOC(rq) && RQ_IOC(rq)->aic)
|
|
|
|
atomic_inc(&RQ_IOC(rq)->aic->nr_dispatched);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as_queue_empty tells us if there are requests left in the device. It may
|
|
|
|
* not be the case that a driver can get the next request even if the queue
|
|
|
|
* is not empty - it is used in the block layer to check for plugging and
|
|
|
|
* merging opportunities
|
|
|
|
*/
|
2007-07-24 01:28:11 -06:00
|
|
|
static int as_queue_empty(struct request_queue *q)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct as_data *ad = q->elevator->elevator_data;
|
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
return list_empty(&ad->fifo_list[BLK_RW_ASYNC])
|
|
|
|
&& list_empty(&ad->fifo_list[BLK_RW_SYNC]);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
2007-07-24 01:28:11 -06:00
|
|
|
as_merge(struct request_queue *q, struct request **req, struct bio *bio)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct as_data *ad = q->elevator->elevator_data;
|
|
|
|
sector_t rb_key = bio->bi_sector + bio_sectors(bio);
|
|
|
|
struct request *__rq;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* check for front merge
|
|
|
|
*/
|
2006-07-18 13:06:01 -06:00
|
|
|
__rq = elv_rb_find(&ad->sort_list[bio_data_dir(bio)], rb_key);
|
2006-07-28 01:23:08 -06:00
|
|
|
if (__rq && elv_rq_merge_ok(__rq, bio)) {
|
|
|
|
*req = __rq;
|
|
|
|
return ELEVATOR_FRONT_MERGE;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
return ELEVATOR_NO_MERGE;
|
|
|
|
}
|
|
|
|
|
2007-07-24 01:28:11 -06:00
|
|
|
static void as_merged_request(struct request_queue *q, struct request *req,
|
|
|
|
int type)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct as_data *ad = q->elevator->elevator_data;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* if the merge was a front merge, we need to reposition request
|
|
|
|
*/
|
2006-07-18 13:06:01 -06:00
|
|
|
if (type == ELEVATOR_FRONT_MERGE) {
|
2006-07-18 13:07:29 -06:00
|
|
|
as_del_rq_rb(ad, req);
|
|
|
|
as_add_rq_rb(ad, req);
|
2005-04-16 16:20:36 -06:00
|
|
|
/*
|
|
|
|
* Note! At this stage of this and the next function, our next
|
|
|
|
* request may not be optimal - eg the request may have "grown"
|
|
|
|
* behind the disk head. We currently don't bother adjusting.
|
|
|
|
*/
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2007-07-24 01:28:11 -06:00
|
|
|
static void as_merged_requests(struct request_queue *q, struct request *req,
|
2005-11-07 01:59:53 -07:00
|
|
|
struct request *next)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
/*
|
2006-07-18 13:07:29 -06:00
|
|
|
* if next expires before rq, assign its expire time to arq
|
|
|
|
* and move into next position (next will be deleted) in fifo
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
2006-07-13 01:12:14 -06:00
|
|
|
if (!list_empty(&req->queuelist) && !list_empty(&next->queuelist)) {
|
|
|
|
if (time_before(rq_fifo_time(next), rq_fifo_time(req))) {
|
|
|
|
list_move(&req->queuelist, &next->queuelist);
|
|
|
|
rq_set_fifo_time(req, rq_fifo_time(next));
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* kill knowledge of next, this one is a goner
|
|
|
|
*/
|
|
|
|
as_remove_queued_request(q, next);
|
2006-07-18 13:07:29 -06:00
|
|
|
as_put_io_context(next);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2006-07-18 13:07:29 -06:00
|
|
|
RQ_SET_STATE(next, AS_RQ_MERGED);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This is executed in a "deferred" process context, by kblockd. It calls the
|
|
|
|
* driver's request_fn so the driver can submit that request.
|
|
|
|
*
|
|
|
|
* IMPORTANT! This guy will reenter the elevator, so set up all queue global
|
|
|
|
* state before calling, and don't rely on any state over calls.
|
|
|
|
*
|
|
|
|
* FIXME! dispatch queue is not a queue at all!
|
|
|
|
*/
|
2006-11-22 07:55:48 -07:00
|
|
|
static void as_work_handler(struct work_struct *work)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2006-11-22 07:55:48 -07:00
|
|
|
struct as_data *ad = container_of(work, struct as_data, antic_work);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2009-04-22 20:05:17 -06:00
|
|
|
blk_run_queue(ad->q);
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
2007-07-24 01:28:11 -06:00
|
|
|
static int as_may_queue(struct request_queue *q, int rw)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
int ret = ELV_MQUEUE_MAY;
|
|
|
|
struct as_data *ad = q->elevator->elevator_data;
|
|
|
|
struct io_context *ioc;
|
|
|
|
if (ad->antic_status == ANTIC_WAIT_REQ ||
|
|
|
|
ad->antic_status == ANTIC_WAIT_NEXT) {
|
2006-07-19 15:39:40 -06:00
|
|
|
ioc = as_get_io_context(q->node);
|
2005-04-16 16:20:36 -06:00
|
|
|
if (ad->io_context == ioc)
|
|
|
|
ret = ELV_MQUEUE_MUST;
|
|
|
|
put_io_context(ioc);
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2008-10-31 03:05:07 -06:00
|
|
|
static void as_exit_queue(struct elevator_queue *e)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct as_data *ad = e->elevator_data;
|
|
|
|
|
|
|
|
del_timer_sync(&ad->antic_timer);
|
2008-12-03 04:41:39 -07:00
|
|
|
cancel_work_sync(&ad->antic_work);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
BUG_ON(!list_empty(&ad->fifo_list[BLK_RW_SYNC]));
|
|
|
|
BUG_ON(!list_empty(&ad->fifo_list[BLK_RW_ASYNC]));
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
put_io_context(ad->io_context);
|
|
|
|
kfree(ad);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2006-07-18 13:07:29 -06:00
|
|
|
* initialize elevator private data (as_data).
|
2005-04-16 16:20:36 -06:00
|
|
|
*/
|
2007-07-24 01:28:11 -06:00
|
|
|
static void *as_init_queue(struct request_queue *q)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
|
|
|
struct as_data *ad;
|
|
|
|
|
2007-07-17 05:03:29 -06:00
|
|
|
ad = kmalloc_node(sizeof(*ad), GFP_KERNEL | __GFP_ZERO, q->node);
|
2005-04-16 16:20:36 -06:00
|
|
|
if (!ad)
|
2006-06-08 00:49:06 -06:00
|
|
|
return NULL;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
ad->q = q; /* Identify what queue the data belongs to */
|
|
|
|
|
|
|
|
/* anticipatory scheduling helpers */
|
|
|
|
ad->antic_timer.function = as_antic_timeout;
|
|
|
|
ad->antic_timer.data = (unsigned long)q;
|
|
|
|
init_timer(&ad->antic_timer);
|
2006-11-22 07:55:48 -07:00
|
|
|
INIT_WORK(&ad->antic_work, as_work_handler);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
INIT_LIST_HEAD(&ad->fifo_list[BLK_RW_SYNC]);
|
|
|
|
INIT_LIST_HEAD(&ad->fifo_list[BLK_RW_ASYNC]);
|
|
|
|
ad->sort_list[BLK_RW_SYNC] = RB_ROOT;
|
|
|
|
ad->sort_list[BLK_RW_ASYNC] = RB_ROOT;
|
|
|
|
ad->fifo_expire[BLK_RW_SYNC] = default_read_expire;
|
|
|
|
ad->fifo_expire[BLK_RW_ASYNC] = default_write_expire;
|
2005-04-16 16:20:36 -06:00
|
|
|
ad->antic_expire = default_antic_expire;
|
2009-04-08 03:02:08 -06:00
|
|
|
ad->batch_expire[BLK_RW_SYNC] = default_read_batch_expire;
|
|
|
|
ad->batch_expire[BLK_RW_ASYNC] = default_write_batch_expire;
|
2005-04-16 16:20:36 -06:00
|
|
|
|
2009-04-08 03:02:08 -06:00
|
|
|
ad->current_batch_expires = jiffies + ad->batch_expire[BLK_RW_SYNC];
|
|
|
|
ad->write_batch_count = ad->batch_expire[BLK_RW_ASYNC] / 10;
|
2005-04-16 16:20:36 -06:00
|
|
|
if (ad->write_batch_count < 2)
|
|
|
|
ad->write_batch_count = 2;
|
|
|
|
|
2006-06-08 00:49:06 -06:00
|
|
|
return ad;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* sysfs parts below
|
|
|
|
*/
|
|
|
|
|
|
|
|
static ssize_t
|
|
|
|
as_var_show(unsigned int var, char *page)
|
|
|
|
{
|
|
|
|
return sprintf(page, "%d\n", var);
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t
|
|
|
|
as_var_store(unsigned long *var, const char *page, size_t count)
|
|
|
|
{
|
|
|
|
char *p = (char *) page;
|
|
|
|
|
2005-07-27 12:43:37 -06:00
|
|
|
*var = simple_strtoul(p, &p, 10);
|
2005-04-16 16:20:36 -06:00
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
2008-10-31 03:05:07 -06:00
|
|
|
static ssize_t est_time_show(struct elevator_queue *e, char *page)
|
2005-04-16 16:20:36 -06:00
|
|
|
{
|
2006-03-18 16:35:43 -07:00
|
|
|
struct as_data *ad = e->elevator_data;
|
2005-04-16 16:20:36 -06:00
|
|
|
int pos = 0;
|
|
|
|
|
2005-11-07 01:59:53 -07:00
|
|
|
pos += sprintf(page+pos, "%lu %% exit probability\n",
|
|
|
|
100*ad->exit_prob/256);
|
|
|
|
pos += sprintf(page+pos, "%lu %% probability of exiting without a "
|
|
|
|
"cooperating process submitting IO\n",
|
|
|
|
100*ad->exit_no_coop/256);
|
2005-04-16 16:20:36 -06:00
|
|
|
pos += sprintf(page+pos, "%lu ms new thinktime\n", ad->new_ttime_mean);
|
2005-11-07 01:59:53 -07:00
|
|
|
pos += sprintf(page+pos, "%llu sectors new seek distance\n",
|
|
|
|
(unsigned long long)ad->new_seek_mean);
|
2005-04-16 16:20:36 -06:00
|
|
|
|
|
|
|
return pos;
|
|
|
|
}
|
|
|
|
|
|
|
|
#define SHOW_FUNCTION(__FUNC, __VAR) \
|
2008-10-31 03:05:07 -06:00
|
|
|
static ssize_t __FUNC(struct elevator_queue *e, char *page) \
|
2005-04-16 16:20:36 -06:00
|
|
|
{ \
|
2006-03-18 16:35:43 -07:00
|
|
|
struct as_data *ad = e->elevator_data; \
|
2005-04-16 16:20:36 -06:00
|
|
|
return as_var_show(jiffies_to_msecs((__VAR)), (page)); \
|
|
|
|
}
|
2009-04-08 03:02:08 -06:00
|
|
|
SHOW_FUNCTION(as_read_expire_show, ad->fifo_expire[BLK_RW_SYNC]);
|
|
|
|
SHOW_FUNCTION(as_write_expire_show, ad->fifo_expire[BLK_RW_ASYNC]);
|
2006-03-18 20:27:18 -07:00
|
|
|
SHOW_FUNCTION(as_antic_expire_show, ad->antic_expire);
|
2009-04-08 03:02:08 -06:00
|
|
|
SHOW_FUNCTION(as_read_batch_expire_show, ad->batch_expire[BLK_RW_SYNC]);
|
|
|
|
SHOW_FUNCTION(as_write_batch_expire_show, ad->batch_expire[BLK_RW_ASYNC]);
|
2005-04-16 16:20:36 -06:00
|
|
|
#undef SHOW_FUNCTION
|
|
|
|
|
|
|
|
#define STORE_FUNCTION(__FUNC, __PTR, MIN, MAX) \
|
2008-10-31 03:05:07 -06:00
|
|
|
static ssize_t __FUNC(struct elevator_queue *e, const char *page, size_t count) \
|
2005-04-16 16:20:36 -06:00
|
|
|
{ \
|
2006-03-18 16:35:43 -07:00
|
|
|
struct as_data *ad = e->elevator_data; \
|
|
|
|
int ret = as_var_store(__PTR, (page), count); \
|
2005-04-16 16:20:36 -06:00
|
|
|
if (*(__PTR) < (MIN)) \
|
|
|
|
*(__PTR) = (MIN); \
|
|
|
|
else if (*(__PTR) > (MAX)) \
|
|
|
|
*(__PTR) = (MAX); \
|
|
|
|
*(__PTR) = msecs_to_jiffies(*(__PTR)); \
|
|
|
|
return ret; \
|
|
|
|
}
|
2009-04-08 03:02:08 -06:00
|
|
|
STORE_FUNCTION(as_read_expire_store, &ad->fifo_expire[BLK_RW_SYNC], 0, INT_MAX);
|
|
|
|
STORE_FUNCTION(as_write_expire_store,
|
|
|
|
&ad->fifo_expire[BLK_RW_ASYNC], 0, INT_MAX);
|
2006-03-18 20:27:18 -07:00
|
|
|
STORE_FUNCTION(as_antic_expire_store, &ad->antic_expire, 0, INT_MAX);
|
|
|
|
STORE_FUNCTION(as_read_batch_expire_store,
|
2009-04-08 03:02:08 -06:00
|
|
|
&ad->batch_expire[BLK_RW_SYNC], 0, INT_MAX);
|
2006-03-18 20:27:18 -07:00
|
|
|
STORE_FUNCTION(as_write_batch_expire_store,
|
2009-04-08 03:02:08 -06:00
|
|
|
&ad->batch_expire[BLK_RW_ASYNC], 0, INT_MAX);
|
2005-04-16 16:20:36 -06:00
|
|
|
#undef STORE_FUNCTION
|
|
|
|
|
2006-03-18 20:27:18 -07:00
|
|
|
#define AS_ATTR(name) \
|
|
|
|
__ATTR(name, S_IRUGO|S_IWUSR, as_##name##_show, as_##name##_store)
|
|
|
|
|
|
|
|
static struct elv_fs_entry as_attrs[] = {
|
|
|
|
__ATTR_RO(est_time),
|
|
|
|
AS_ATTR(read_expire),
|
|
|
|
AS_ATTR(write_expire),
|
|
|
|
AS_ATTR(antic_expire),
|
|
|
|
AS_ATTR(read_batch_expire),
|
|
|
|
AS_ATTR(write_batch_expire),
|
|
|
|
__ATTR_NULL
|
2005-04-16 16:20:36 -06:00
|
|
|
};
|
|
|
|
|
|
|
|
static struct elevator_type iosched_as = {
|
|
|
|
.ops = {
|
|
|
|
.elevator_merge_fn = as_merge,
|
|
|
|
.elevator_merged_fn = as_merged_request,
|
|
|
|
.elevator_merge_req_fn = as_merged_requests,
|
2005-10-20 08:42:29 -06:00
|
|
|
.elevator_dispatch_fn = as_dispatch_request,
|
|
|
|
.elevator_add_req_fn = as_add_request,
|
|
|
|
.elevator_activate_req_fn = as_activate_request,
|
2005-04-16 16:20:36 -06:00
|
|
|
.elevator_deactivate_req_fn = as_deactivate_request,
|
|
|
|
.elevator_queue_empty_fn = as_queue_empty,
|
|
|
|
.elevator_completed_req_fn = as_completed_request,
|
2006-07-18 13:06:01 -06:00
|
|
|
.elevator_former_req_fn = elv_rb_former_request,
|
|
|
|
.elevator_latter_req_fn = elv_rb_latter_request,
|
2005-04-16 16:20:36 -06:00
|
|
|
.elevator_may_queue_fn = as_may_queue,
|
|
|
|
.elevator_init_fn = as_init_queue,
|
|
|
|
.elevator_exit_fn = as_exit_queue,
|
2006-03-18 11:21:20 -07:00
|
|
|
.trim = as_trim,
|
2005-04-16 16:20:36 -06:00
|
|
|
},
|
|
|
|
|
2006-03-18 16:35:43 -07:00
|
|
|
.elevator_attrs = as_attrs,
|
2005-04-16 16:20:36 -06:00
|
|
|
.elevator_name = "anticipatory",
|
|
|
|
.elevator_owner = THIS_MODULE,
|
|
|
|
};
|
|
|
|
|
|
|
|
static int __init as_init(void)
|
|
|
|
{
|
2007-12-12 10:51:56 -07:00
|
|
|
elv_register(&iosched_as);
|
|
|
|
|
|
|
|
return 0;
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
static void __exit as_exit(void)
|
|
|
|
{
|
2006-10-01 00:28:10 -06:00
|
|
|
DECLARE_COMPLETION_ONSTACK(all_gone);
|
2005-04-16 16:20:36 -06:00
|
|
|
elv_unregister(&iosched_as);
|
2006-03-18 13:05:53 -07:00
|
|
|
ioc_gone = &all_gone;
|
[PATCH 1/2] iosched: fix typo and barrier()
On rmmod path, cfq/as waits to make sure all io-contexts was
freed. However, it's using complete(), not wait_for_completion().
I think barrier() is not enough in here. To avoid the following case,
this patch replaces barrier() with smb_wmb().
cpu0 visibility cpu1
[ioc_gnone=NULL,ioc_count=1]
ioc_gnone = &all_gone NULL,ioc_count=1
atomic_read(&ioc_count) NULL,ioc_count=1
wait_for_completion() NULL,ioc_count=0 atomic_sub_and_test()
NULL,ioc_count=0 if ( && ioc_gone)
[ioc_gone==NULL,
so doesn't call complete()]
&all_gone,ioc_count=0
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Jens Axboe <axboe@suse.de>
2006-04-18 01:44:06 -06:00
|
|
|
/* ioc_gone's update must be visible before reading ioc_count */
|
|
|
|
smp_wmb();
|
2006-07-18 21:10:01 -06:00
|
|
|
if (elv_ioc_count_read(ioc_count))
|
2008-05-29 01:35:22 -06:00
|
|
|
wait_for_completion(&all_gone);
|
2006-03-18 13:05:53 -07:00
|
|
|
synchronize_rcu();
|
2005-04-16 16:20:36 -06:00
|
|
|
}
|
|
|
|
|
|
|
|
module_init(as_init);
|
|
|
|
module_exit(as_exit);
|
|
|
|
|
|
|
|
MODULE_AUTHOR("Nick Piggin");
|
|
|
|
MODULE_LICENSE("GPL");
|
|
|
|
MODULE_DESCRIPTION("anticipatory IO scheduler");
|