Commit 86db1e29 authored by Jens Axboe's avatar Jens Axboe

block: continue ll_rw_blk.c splitup

Adds files for barrier handling, rq execution, io context handling,
mapping data to requests, and queue settings.
Signed-off-by: default avatarJens Axboe <jens.axboe@oracle.com>
parent 8324aa91
......@@ -2,8 +2,9 @@
# Makefile for the kernel block layer
#
obj-$(CONFIG_BLOCK) := elevator.o blk-core.o blk-tag.o blk-sysfs.o ioctl.o \
genhd.o scsi_ioctl.o
obj-$(CONFIG_BLOCK) := elevator.o blk-core.o blk-tag.o blk-sysfs.o \
blk-barrier.o blk-settings.o blk-ioc.o blk-map.o \
blk-exec.o ioctl.o genhd.o scsi_ioctl.o
obj-$(CONFIG_BLK_DEV_BSG) += bsg.o
obj-$(CONFIG_IOSCHED_NOOP) += noop-iosched.o
......
/*
* Functions related to barrier IO handling
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/bio.h>
#include <linux/blkdev.h>
#include "blk.h"
/**
* blk_queue_ordered - does this queue support ordered writes
* @q: the request queue
* @ordered: one of QUEUE_ORDERED_*
* @prepare_flush_fn: rq setup helper for cache flush ordered writes
*
* Description:
* For journalled file systems, doing ordered writes on a commit
* block instead of explicitly doing wait_on_buffer (which is bad
* for performance) can be a big win. Block drivers supporting this
* feature should call this function and indicate so.
*
**/
int blk_queue_ordered(struct request_queue *q, unsigned ordered,
prepare_flush_fn *prepare_flush_fn)
{
if (ordered & (QUEUE_ORDERED_PREFLUSH | QUEUE_ORDERED_POSTFLUSH) &&
prepare_flush_fn == NULL) {
printk(KERN_ERR "blk_queue_ordered: prepare_flush_fn required\n");
return -EINVAL;
}
if (ordered != QUEUE_ORDERED_NONE &&
ordered != QUEUE_ORDERED_DRAIN &&
ordered != QUEUE_ORDERED_DRAIN_FLUSH &&
ordered != QUEUE_ORDERED_DRAIN_FUA &&
ordered != QUEUE_ORDERED_TAG &&
ordered != QUEUE_ORDERED_TAG_FLUSH &&
ordered != QUEUE_ORDERED_TAG_FUA) {
printk(KERN_ERR "blk_queue_ordered: bad value %d\n", ordered);
return -EINVAL;
}
q->ordered = ordered;
q->next_ordered = ordered;
q->prepare_flush_fn = prepare_flush_fn;
return 0;
}
EXPORT_SYMBOL(blk_queue_ordered);
/*
* Cache flushing for ordered writes handling
*/
inline unsigned blk_ordered_cur_seq(struct request_queue *q)
{
if (!q->ordseq)
return 0;
return 1 << ffz(q->ordseq);
}
unsigned blk_ordered_req_seq(struct request *rq)
{
struct request_queue *q = rq->q;
BUG_ON(q->ordseq == 0);
if (rq == &q->pre_flush_rq)
return QUEUE_ORDSEQ_PREFLUSH;
if (rq == &q->bar_rq)
return QUEUE_ORDSEQ_BAR;
if (rq == &q->post_flush_rq)
return QUEUE_ORDSEQ_POSTFLUSH;
/*
* !fs requests don't need to follow barrier ordering. Always
* put them at the front. This fixes the following deadlock.
*
* http://thread.gmane.org/gmane.linux.kernel/537473
*/
if (!blk_fs_request(rq))
return QUEUE_ORDSEQ_DRAIN;
if ((rq->cmd_flags & REQ_ORDERED_COLOR) ==
(q->orig_bar_rq->cmd_flags & REQ_ORDERED_COLOR))
return QUEUE_ORDSEQ_DRAIN;
else
return QUEUE_ORDSEQ_DONE;
}
void blk_ordered_complete_seq(struct request_queue *q, unsigned seq, int error)
{
struct request *rq;
if (error && !q->orderr)
q->orderr = error;
BUG_ON(q->ordseq & seq);
q->ordseq |= seq;
if (blk_ordered_cur_seq(q) != QUEUE_ORDSEQ_DONE)
return;
/*
* Okay, sequence complete.
*/
q->ordseq = 0;
rq = q->orig_bar_rq;
if (__blk_end_request(rq, q->orderr, blk_rq_bytes(rq)))
BUG();
}
static void pre_flush_end_io(struct request *rq, int error)
{
elv_completed_request(rq->q, rq);
blk_ordered_complete_seq(rq->q, QUEUE_ORDSEQ_PREFLUSH, error);
}
static void bar_end_io(struct request *rq, int error)
{
elv_completed_request(rq->q, rq);
blk_ordered_complete_seq(rq->q, QUEUE_ORDSEQ_BAR, error);
}
static void post_flush_end_io(struct request *rq, int error)
{
elv_completed_request(rq->q, rq);
blk_ordered_complete_seq(rq->q, QUEUE_ORDSEQ_POSTFLUSH, error);
}
static void queue_flush(struct request_queue *q, unsigned which)
{
struct request *rq;
rq_end_io_fn *end_io;
if (which == QUEUE_ORDERED_PREFLUSH) {
rq = &q->pre_flush_rq;
end_io = pre_flush_end_io;
} else {
rq = &q->post_flush_rq;
end_io = post_flush_end_io;
}
rq->cmd_flags = REQ_HARDBARRIER;
rq_init(q, rq);
rq->elevator_private = NULL;
rq->elevator_private2 = NULL;
rq->rq_disk = q->bar_rq.rq_disk;
rq->end_io = end_io;
q->prepare_flush_fn(q, rq);
elv_insert(q, rq, ELEVATOR_INSERT_FRONT);
}
static inline struct request *start_ordered(struct request_queue *q,
struct request *rq)
{
q->orderr = 0;
q->ordered = q->next_ordered;
q->ordseq |= QUEUE_ORDSEQ_STARTED;
/*
* Prep proxy barrier request.
*/
blkdev_dequeue_request(rq);
q->orig_bar_rq = rq;
rq = &q->bar_rq;
rq->cmd_flags = 0;
rq_init(q, rq);
if (bio_data_dir(q->orig_bar_rq->bio) == WRITE)
rq->cmd_flags |= REQ_RW;
if (q->ordered & QUEUE_ORDERED_FUA)
rq->cmd_flags |= REQ_FUA;
rq->elevator_private = NULL;
rq->elevator_private2 = NULL;
init_request_from_bio(rq, q->orig_bar_rq->bio);
rq->end_io = bar_end_io;
/*
* Queue ordered sequence. As we stack them at the head, we
* need to queue in reverse order. Note that we rely on that
* no fs request uses ELEVATOR_INSERT_FRONT and thus no fs
* request gets inbetween ordered sequence. If this request is
* an empty barrier, we don't need to do a postflush ever since
* there will be no data written between the pre and post flush.
* Hence a single flush will suffice.
*/
if ((q->ordered & QUEUE_ORDERED_POSTFLUSH) && !blk_empty_barrier(rq))
queue_flush(q, QUEUE_ORDERED_POSTFLUSH);
else
q->ordseq |= QUEUE_ORDSEQ_POSTFLUSH;
elv_insert(q, rq, ELEVATOR_INSERT_FRONT);
if (q->ordered & QUEUE_ORDERED_PREFLUSH) {
queue_flush(q, QUEUE_ORDERED_PREFLUSH);
rq = &q->pre_flush_rq;
} else
q->ordseq |= QUEUE_ORDSEQ_PREFLUSH;
if ((q->ordered & QUEUE_ORDERED_TAG) || q->in_flight == 0)
q->ordseq |= QUEUE_ORDSEQ_DRAIN;
else
rq = NULL;
return rq;
}
int blk_do_ordered(struct request_queue *q, struct request **rqp)
{
struct request *rq = *rqp;
const int is_barrier = blk_fs_request(rq) && blk_barrier_rq(rq);
if (!q->ordseq) {
if (!is_barrier)
return 1;
if (q->next_ordered != QUEUE_ORDERED_NONE) {
*rqp = start_ordered(q, rq);
return 1;
} else {
/*
* This can happen when the queue switches to
* ORDERED_NONE while this request is on it.
*/
blkdev_dequeue_request(rq);
if (__blk_end_request(rq, -EOPNOTSUPP,
blk_rq_bytes(rq)))
BUG();
*rqp = NULL;
return 0;
}
}
/*
* Ordered sequence in progress
*/
/* Special requests are not subject to ordering rules. */
if (!blk_fs_request(rq) &&
rq != &q->pre_flush_rq && rq != &q->post_flush_rq)
return 1;
if (q->ordered & QUEUE_ORDERED_TAG) {
/* Ordered by tag. Blocking the next barrier is enough. */
if (is_barrier && rq != &q->bar_rq)
*rqp = NULL;
} else {
/* Ordered by draining. Wait for turn. */
WARN_ON(blk_ordered_req_seq(rq) < blk_ordered_cur_seq(q));
if (blk_ordered_req_seq(rq) > blk_ordered_cur_seq(q))
*rqp = NULL;
}
return 1;
}
static void bio_end_empty_barrier(struct bio *bio, int err)
{
if (err)
clear_bit(BIO_UPTODATE, &bio->bi_flags);
complete(bio->bi_private);
}
/**
* blkdev_issue_flush - queue a flush
* @bdev: blockdev to issue flush for
* @error_sector: error sector
*
* Description:
* Issue a flush for the block device in question. Caller can supply
* room for storing the error offset in case of a flush error, if they
* wish to. Caller must run wait_for_completion() on its own.
*/
int blkdev_issue_flush(struct block_device *bdev, sector_t *error_sector)
{
DECLARE_COMPLETION_ONSTACK(wait);
struct request_queue *q;
struct bio *bio;
int ret;
if (bdev->bd_disk == NULL)
return -ENXIO;
q = bdev_get_queue(bdev);
if (!q)
return -ENXIO;
bio = bio_alloc(GFP_KERNEL, 0);
if (!bio)
return -ENOMEM;
bio->bi_end_io = bio_end_empty_barrier;
bio->bi_private = &wait;
bio->bi_bdev = bdev;
submit_bio(1 << BIO_RW_BARRIER, bio);
wait_for_completion(&wait);
/*
* The driver must store the error location in ->bi_sector, if
* it supports it. For non-stacked drivers, this should be copied
* from rq->sector.
*/
if (error_sector)
*error_sector = bio->bi_sector;
ret = 0;
if (!bio_flagged(bio, BIO_UPTODATE))
ret = -EIO;
bio_put(bio);
return ret;
}
EXPORT_SYMBOL(blkdev_issue_flush);
This diff is collapsed.
/*
* Functions related to setting various queue properties from drivers
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/bio.h>
#include <linux/blkdev.h>
#include "blk.h"
/*
* for max sense size
*/
#include <scsi/scsi_cmnd.h>
/**
* blk_end_sync_rq - executes a completion event on a request
* @rq: request to complete
* @error: end io status of the request
*/
void blk_end_sync_rq(struct request *rq, int error)
{
struct completion *waiting = rq->end_io_data;
rq->end_io_data = NULL;
__blk_put_request(rq->q, rq);
/*
* complete last, if this is a stack request the process (and thus
* the rq pointer) could be invalid right after this complete()
*/
complete(waiting);
}
EXPORT_SYMBOL(blk_end_sync_rq);
/**
* blk_execute_rq_nowait - insert a request into queue for execution
* @q: queue to insert the request in
* @bd_disk: matching gendisk
* @rq: request to insert
* @at_head: insert request at head or tail of queue
* @done: I/O completion handler
*
* Description:
* Insert a fully prepared request at the back of the io scheduler queue
* for execution. Don't wait for completion.
*/
void blk_execute_rq_nowait(struct request_queue *q, struct gendisk *bd_disk,
struct request *rq, int at_head,
rq_end_io_fn *done)
{
int where = at_head ? ELEVATOR_INSERT_FRONT : ELEVATOR_INSERT_BACK;
rq->rq_disk = bd_disk;
rq->cmd_flags |= REQ_NOMERGE;
rq->end_io = done;
WARN_ON(irqs_disabled());
spin_lock_irq(q->queue_lock);
__elv_add_request(q, rq, where, 1);
__generic_unplug_device(q);
spin_unlock_irq(q->queue_lock);
}
EXPORT_SYMBOL_GPL(blk_execute_rq_nowait);
/**
* blk_execute_rq - insert a request into queue for execution
* @q: queue to insert the request in
* @bd_disk: matching gendisk
* @rq: request to insert
* @at_head: insert request at head or tail of queue
*
* Description:
* Insert a fully prepared request at the back of the io scheduler queue
* for execution and wait for completion.
*/
int blk_execute_rq(struct request_queue *q, struct gendisk *bd_disk,
struct request *rq, int at_head)
{
DECLARE_COMPLETION_ONSTACK(wait);
char sense[SCSI_SENSE_BUFFERSIZE];
int err = 0;
/*
* we need an extra reference to the request, so we can look at
* it after io completion
*/
rq->ref_count++;
if (!rq->sense) {
memset(sense, 0, sizeof(sense));
rq->sense = sense;
rq->sense_len = 0;
}
rq->end_io_data = &wait;
blk_execute_rq_nowait(q, bd_disk, rq, at_head, blk_end_sync_rq);
wait_for_completion(&wait);
if (rq->errors)
err = -EIO;
return err;
}
EXPORT_SYMBOL(blk_execute_rq);
/*
* Functions related to io context handling
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/bio.h>
#include <linux/blkdev.h>
#include <linux/bootmem.h> /* for max_pfn/max_low_pfn */
#include "blk.h"
/*
* For io context allocations
*/
static struct kmem_cache *iocontext_cachep;
static void cfq_dtor(struct io_context *ioc)
{
struct cfq_io_context *cic[1];
int r;
/*
* We don't have a specific key to lookup with, so use the gang
* lookup to just retrieve the first item stored. The cfq exit
* function will iterate the full tree, so any member will do.
*/
r = radix_tree_gang_lookup(&ioc->radix_root, (void **) cic, 0, 1);
if (r > 0)
cic[0]->dtor(ioc);
}
/*
* IO Context helper functions. put_io_context() returns 1 if there are no
* more users of this io context, 0 otherwise.
*/
int put_io_context(struct io_context *ioc)
{
if (ioc == NULL)
return 1;
BUG_ON(atomic_read(&ioc->refcount) == 0);
if (atomic_dec_and_test(&ioc->refcount)) {
rcu_read_lock();
if (ioc->aic && ioc->aic->dtor)
ioc->aic->dtor(ioc->aic);
rcu_read_unlock();
cfq_dtor(ioc);
kmem_cache_free(iocontext_cachep, ioc);
return 1;
}
return 0;
}
EXPORT_SYMBOL(put_io_context);
static void cfq_exit(struct io_context *ioc)
{
struct cfq_io_context *cic[1];
int r;
rcu_read_lock();
/*
* See comment for cfq_dtor()
*/
r = radix_tree_gang_lookup(&ioc->radix_root, (void **) cic, 0, 1);
rcu_read_unlock();
if (r > 0)
cic[0]->exit(ioc);
}
/* Called by the exitting task */
void exit_io_context(void)
{
struct io_context *ioc;
task_lock(current);
ioc = current->io_context;
current->io_context = NULL;
task_unlock(current);
if (atomic_dec_and_test(&ioc->nr_tasks)) {
if (ioc->aic && ioc->aic->exit)
ioc->aic->exit(ioc->aic);
cfq_exit(ioc);
put_io_context(ioc);
}
}
struct io_context *alloc_io_context(gfp_t gfp_flags, int node)
{
struct io_context *ret;
ret = kmem_cache_alloc_node(iocontext_cachep, gfp_flags, node);
if (ret) {
atomic_set(&ret->refcount, 1);
atomic_set(&ret->nr_tasks, 1);
spin_lock_init(&ret->lock);
ret->ioprio_changed = 0;
ret->ioprio = 0;
ret->last_waited = jiffies; /* doesn't matter... */
ret->nr_batch_requests = 0; /* because this is 0 */
ret->aic = NULL;
INIT_RADIX_TREE(&ret->radix_root, GFP_ATOMIC | __GFP_HIGH);
ret->ioc_data = NULL;
}
return ret;
}
/*
* If the current task has no IO context then create one and initialise it.
* Otherwise, return its existing IO context.
*
* This returned IO context doesn't have a specifically elevated refcount,
* but since the current task itself holds a reference, the context can be
* used in general code, so long as it stays within `current` context.
*/
struct io_context *current_io_context(gfp_t gfp_flags, int node)
{
struct task_struct *tsk = current;
struct io_context *ret;
ret = tsk->io_context;
if (likely(ret))
return ret;
ret = alloc_io_context(gfp_flags, node);
if (ret) {
/* make sure set_task_ioprio() sees the settings above */
smp_wmb();
tsk->io_context = ret;
}
return ret;
}
/*
* If the current task has no IO context then create one and initialise it.
* If it does have a context, take a ref on it.
*
* This is always called in the context of the task which submitted the I/O.
*/
struct io_context *get_io_context(gfp_t gfp_flags, int node)
{
struct io_context *ret = NULL;
/*
* Check for unlikely race with exiting task. ioc ref count is
* zero when ioc is being detached.
*/
do {
ret = current_io_context(gfp_flags, node);
if (unlikely(!ret))
break;
} while (!atomic_inc_not_zero(&ret->refcount));
return ret;
}
EXPORT_SYMBOL(get_io_context);
void copy_io_context(struct io_context **pdst, struct io_context **psrc)
{
struct io_context *src = *psrc;
struct io_context *dst = *pdst;
if (src) {
BUG_ON(atomic_read(&src->refcount) == 0);
atomic_inc(&src->refcount);
put_io_context(dst);
*pdst = src;
}
}
EXPORT_SYMBOL(copy_io_context);
void swap_io_context(struct io_context **ioc1, struct io_context **ioc2)
{
struct io_context *temp;
temp = *ioc1;
*ioc1 = *ioc2;
*ioc2 = temp;
}
EXPORT_SYMBOL(swap_io_context);
int __init blk_ioc_init(void)
{
iocontext_cachep = kmem_cache_create("blkdev_ioc",
sizeof(struct io_context), 0, SLAB_PANIC, NULL);
return 0;
}
subsys_initcall(blk_ioc_init);
/*
* Functions related to mapping data to requests
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/bio.h>
#include <linux/blkdev.h>
#include "blk.h"
int blk_rq_append_bio(struct request_queue *q, struct request *rq,
struct bio *bio)
{
if (!rq->bio)
blk_rq_bio_prep(q, rq, bio);
else if (!ll_back_merge_fn(q, rq, bio))
return -EINVAL;
else {
rq->biotail->bi_next = bio;
rq->biotail = bio;
rq->data_len += bio->bi_size;
}
return 0;
}
EXPORT_SYMBOL(blk_rq_append_bio);