[PATCH v6 0/5] block: loop: improve loop with AIO

[PATCH v6 0/5] block: loop: improve loop with AIO

[Ming Lei]

Hi Guys,

There are about 3 advantages to use direct I/O and AIO on
read/write loop's backing file:

1) double cache can be avoided, then memory usage gets
decreased a lot

2) not like user space direct I/O, there isn't cost of
pinning pages

3) avoid context switch for obtaining good throughput
- in buffered file read, random I/O throughput is often obtained
only if they are submitted concurrently from lots of tasks; but for
sequential I/O, most of times they can be hit from page cache, so
concurrent submissions often introduce unnecessary context switch
and can't improve throughput much. There was such discussion[1]
to use non-blocking I/O to improve the problem for application.
- with direct I/O and AIO, concurrent submissions can be
avoided and random read throughput can't be affected meantime

So this patchset trys to improve loop via AIO, and about 45% memory
usage can be decreased, see detailed data in commit log of patch4,
also IO throughput isn't affected too.

V6:
        - only patch 4 and patch 5 get updated
        - check lo->lo_offset to decide if direct IO can be supported(4/5)
        - introduce one flag for userspace(losetup) to keep updated
        if using direct I/O to access backing file(4/5)
        - implement patches for util-linux(losetup) so that losetup can
        enable direct I/O feature:(4/5)

           http://kernel.ubuntu.com/git/ming/util-linux.git/log/?h=losetup-dio

        - remove the direct IO control interface from sysfs(4/5)
        - handle partial read in case of direct read (5/5)
        - add more comments for direct IO (5/5)
V5:
        - don't introduce IOCB_DONT_DIRTY_PAGE and bypass dirtying
        for ITER_KVEC and ITER_BVEC direct IO(read), as required by
        Christoph

V4:
        - add detailed commit log for 'use kthread_work'
        - allow userspace(sysfs, losetup) to decide if dio/aio is
        used as suggested by Christoph and Dave Chinner
        - only use dio if the backing block device's min io size
        is 512 as pointed by Dave Chinner & Christoph
V3:
        - based on Al's iov_iter work and Christoph's kiocb changes
        - use kthread_work
        - introduce IOCB_DONT_DIRTY_PAGE flag
        - set QUEUE_FLAG_NOMERGES for loop's request queue
V2:
        - remove 'extra' parameter to aio_kernel_alloc()
        - try to avoid memory allcation inside queue req callback
        - introduce 'use_mq' sysfs file for enabling kernel aio or disabling it
V1:
        - link:
                http://marc.info/?t=140803157700004&r=1&w=2
        - improve failure path in aio_kernel_submit()

 drivers/block/loop.c      | 235 +++++++++++++++++++++++++++++++++++-----------
 drivers/block/loop.h      |  13 +--
 fs/direct-io.c            |   9 +-
 include/uapi/linux/loop.h |   1 +
 4 files changed, 193 insertions(+), 65 deletions(-)

Thanks,
Ming

[PATCH v6 1/5] fs: direct-io: don't dirtying pages for ITER_BVEC/ITER_KVEC direct read

[Ming Lei]

When direct read IO is submitted from kernel, it is often
unnecessary to dirty pages, for example of loop, dirtying pages
have been considered in the upper filesystem(over loop) side
already, and they don't need to be dirtied again.

So this patch doesn't dirtying pages for ITER_BVEC/ITER_KVEC
direct read, and loop should be the 1st case to use ITER_BVEC/ITER_KVEC
for direct read I/O.

The patch is based on previous Dave's patch.

Cc: Dave Kleikamp <dave.kleikamp@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ming Lei <ming.lei@canonical.com>
---
 fs/direct-io.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/fs/direct-io.c b/fs/direct-io.c
index 745d234..07e4b28 100644
--- a/fs/direct-io.c
+++ b/fs/direct-io.c
@@ -120,6 +120,7 @@ struct dio {
 	int page_errors;		/* errno from get_user_pages() */
 	int is_async;			/* is IO async ? */
 	bool defer_completion;		/* defer AIO completion to workqueue? */
+	bool should_dirty;		/* if pages should be dirtied */
 	int io_error;			/* IO error in completion path */
 	unsigned long refcount;		/* direct_io_worker() and bios */
 	struct bio *bio_list;		/* singly linked via bi_private */
@@ -393,7 +394,7 @@ static inline void dio_bio_submit(struct dio *dio, struct dio_submit *sdio)
 	dio->refcount++;
 	spin_unlock_irqrestore(&dio->bio_lock, flags);
 
-	if (dio->is_async && dio->rw == READ)
+	if (dio->is_async && dio->rw == READ && dio->should_dirty)
 		bio_set_pages_dirty(bio);
 
 	if (sdio->submit_io)
@@ -464,13 +465,14 @@ static int dio_bio_complete(struct dio *dio, struct bio *bio)
 	if (!uptodate)
 		dio->io_error = -EIO;
 
-	if (dio->is_async && dio->rw == READ) {
+	if (dio->is_async && dio->rw == READ && dio->should_dirty) {
 		bio_check_pages_dirty(bio);	/* transfers ownership */
 	} else {
 		bio_for_each_segment_all(bvec, bio, i) {
 			struct page *page = bvec->bv_page;
 
-			if (dio->rw == READ && !PageCompound(page))
+			if (dio->rw == READ && !PageCompound(page) &&
+					dio->should_dirty)
 				set_page_dirty_lock(page);
 			page_cache_release(page);
 		}
@@ -1217,6 +1219,7 @@ do_blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
 	spin_lock_init(&dio->bio_lock);
 	dio->refcount = 1;
 
+	dio->should_dirty = (iter->type == ITER_IOVEC);
 	sdio.iter = iter;
 	sdio.final_block_in_request =
 		(offset + iov_iter_count(iter)) >> blkbits;
-- 
1.9.1

[PATCH v6 2/5] block: loop: set QUEUE_FLAG_NOMERGES for request queue of loop

[Ming Lei]

It doesn't make sense to enable merge because the I/O
submitted to backing file is handled page by page.

Signed-off-by: Ming Lei <ming.lei@canonical.com>
---
 drivers/block/loop.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 40580dc..10cc583 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -1595,6 +1595,12 @@ static int loop_add(struct loop_device **l, int i)
 	}
 	lo->lo_queue->queuedata = lo;
 
+	/*
+	 * It doesn't make sense to enable merge because the I/O
+	 * submitted to backing file is handled page by page.
+	 */
+	queue_flag_set_unlocked(QUEUE_FLAG_NOMERGES, lo->lo_queue);
+
 	INIT_LIST_HEAD(&lo->write_cmd_head);
 	INIT_WORK(&lo->write_work, loop_queue_write_work);
 
-- 
1.9.1

[PATCH v6 3/5] block: loop: use kthread_work

[Ming Lei]

The following patch will use dio/aio to submit IO to backing file,
then it needn't to schedule IO concurrently from work, so
use kthread_work for decreasing context switch cost a lot.

For non-AIO case, single thread has been used for long long time,
and it was just converted to work in v4.0, which has caused performance
regression for fedora live booting already. In discussion[1], even
though submitting I/O via work concurrently can improve random read IO
throughput, meantime it might hurt sequential read IO performance, so
better to restore to single thread behaviour.

For the following AIO support, it is better to use multi hw-queue
with per-hwq kthread than current work approach suppose there is so
high performance requirement for loop.

[1] http://marc.info/?t=143082678400002&r=1&w=2

Signed-off-by: Ming Lei <ming.lei@canonical.com>
---
 drivers/block/loop.c | 79 ++++++++++++++++------------------------------------
 drivers/block/loop.h | 10 +++----
 2 files changed, 28 insertions(+), 61 deletions(-)

diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 10cc583..555b895 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -688,6 +688,23 @@ static void loop_config_discard(struct loop_device *lo)
 	queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, q);
 }
 
+static void loop_unprepare_queue(struct loop_device *lo)
+{
+	flush_kthread_worker(&lo->worker);
+	kthread_stop(lo->worker_task);
+}
+
+static int loop_prepare_queue(struct loop_device *lo)
+{
+	init_kthread_worker(&lo->worker);
+	lo->worker_task = kthread_run(kthread_worker_fn,
+			&lo->worker, "loop%d", lo->lo_number);
+	if (IS_ERR(lo->worker_task))
+		return -ENOMEM;
+	set_user_nice(lo->worker_task, MIN_NICE);
+	return 0;
+}
+
 static int loop_set_fd(struct loop_device *lo, fmode_t mode,
 		       struct block_device *bdev, unsigned int arg)
 {
@@ -745,11 +762,8 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode,
 	size = get_loop_size(lo, file);
 	if ((loff_t)(sector_t)size != size)
 		goto out_putf;
-	error = -ENOMEM;
-	lo->wq = alloc_workqueue("kloopd%d",
-			WQ_MEM_RECLAIM | WQ_HIGHPRI | WQ_UNBOUND, 16,
-			lo->lo_number);
-	if (!lo->wq)
+	error = loop_prepare_queue(lo);
+	if (error)
 		goto out_putf;
 
 	error = 0;
@@ -903,8 +917,7 @@ static int loop_clr_fd(struct loop_device *lo)
 	lo->lo_flags = 0;
 	if (!part_shift)
 		lo->lo_disk->flags |= GENHD_FL_NO_PART_SCAN;
-	destroy_workqueue(lo->wq);
-	lo->wq = NULL;
+	loop_unprepare_queue(lo);
 	mutex_unlock(&lo->lo_ctl_mutex);
 	/*
 	 * Need not hold lo_ctl_mutex to fput backing file.
@@ -1461,23 +1474,7 @@ static int loop_queue_rq(struct blk_mq_hw_ctx *hctx,
 	if (lo->lo_state != Lo_bound)
 		return -EIO;
 
-	if (cmd->rq->cmd_flags & REQ_WRITE) {
-		struct loop_device *lo = cmd->rq->q->queuedata;
-		bool need_sched = true;
-
-		spin_lock_irq(&lo->lo_lock);
-		if (lo->write_started)
-			need_sched = false;
-		else
-			lo->write_started = true;
-		list_add_tail(&cmd->list, &lo->write_cmd_head);
-		spin_unlock_irq(&lo->lo_lock);
-
-		if (need_sched)
-			queue_work(lo->wq, &lo->write_work);
-	} else {
-		queue_work(lo->wq, &cmd->read_work);
-	}
+	queue_kthread_work(&lo->worker, &cmd->work);
 
 	return BLK_MQ_RQ_QUEUE_OK;
 }
@@ -1499,35 +1496,10 @@ static void loop_handle_cmd(struct loop_cmd *cmd)
 	blk_mq_complete_request(cmd->rq);
 }
 
-static void loop_queue_write_work(struct work_struct *work)
-{
-	struct loop_device *lo =
-		container_of(work, struct loop_device, write_work);
-	LIST_HEAD(cmd_list);
-
-	spin_lock_irq(&lo->lo_lock);
- repeat:
-	list_splice_init(&lo->write_cmd_head, &cmd_list);
-	spin_unlock_irq(&lo->lo_lock);
-
-	while (!list_empty(&cmd_list)) {
-		struct loop_cmd *cmd = list_first_entry(&cmd_list,
-				struct loop_cmd, list);
-		list_del_init(&cmd->list);
-		loop_handle_cmd(cmd);
-	}
-
-	spin_lock_irq(&lo->lo_lock);
-	if (!list_empty(&lo->write_cmd_head))
-		goto repeat;
-	lo->write_started = false;
-	spin_unlock_irq(&lo->lo_lock);
-}
-
-static void loop_queue_read_work(struct work_struct *work)
+static void loop_queue_work(struct kthread_work *work)
 {
 	struct loop_cmd *cmd =
-		container_of(work, struct loop_cmd, read_work);
+		container_of(work, struct loop_cmd, work);
 
 	loop_handle_cmd(cmd);
 }
@@ -1539,7 +1511,7 @@ static int loop_init_request(void *data, struct request *rq,
 	struct loop_cmd *cmd = blk_mq_rq_to_pdu(rq);
 
 	cmd->rq = rq;
-	INIT_WORK(&cmd->read_work, loop_queue_read_work);
+	init_kthread_work(&cmd->work, loop_queue_work);
 
 	return 0;
 }
@@ -1601,9 +1573,6 @@ static int loop_add(struct loop_device **l, int i)
 	 */
 	queue_flag_set_unlocked(QUEUE_FLAG_NOMERGES, lo->lo_queue);
 
-	INIT_LIST_HEAD(&lo->write_cmd_head);
-	INIT_WORK(&lo->write_work, loop_queue_write_work);
-
 	disk = lo->lo_disk = alloc_disk(1 << part_shift);
 	if (!disk)
 		goto out_free_queue;
diff --git a/drivers/block/loop.h b/drivers/block/loop.h
index 25e8997..b6c7d21 100644
--- a/drivers/block/loop.h
+++ b/drivers/block/loop.h
@@ -14,7 +14,7 @@
 #include <linux/blk-mq.h>
 #include <linux/spinlock.h>
 #include <linux/mutex.h>
-#include <linux/workqueue.h>
+#include <linux/kthread.h>
 #include <uapi/linux/loop.h>
 
 /* Possible states of device */
@@ -54,12 +54,10 @@ struct loop_device {
 	gfp_t		old_gfp_mask;
 
 	spinlock_t		lo_lock;
-	struct workqueue_struct *wq;
-	struct list_head	write_cmd_head;
-	struct work_struct	write_work;
-	bool			write_started;
 	int			lo_state;
 	struct mutex		lo_ctl_mutex;
+	struct kthread_worker	worker;
+	struct task_struct	*worker_task;
 
 	struct request_queue	*lo_queue;
 	struct blk_mq_tag_set	tag_set;
@@ -67,7 +65,7 @@ struct loop_device {
 };
 
 struct loop_cmd {
-	struct work_struct read_work;
+	struct kthread_work work;
 	struct request *rq;
 	struct list_head list;
 };
-- 
1.9.1

[PATCH v6 4/5] block: loop: prepare for supporing direct IO

[Ming Lei]

This patches provides one interface for enabling direct IO
from user space:

	- userspace(such as losetup) can pass 'file' which is
	opened/fcntl as O_DIRECT

Also __loop_update_dio() is introduced to check if direct I/O
can be used on current loop setting.

The last big change is to introduce LO_FLAGS_DIRECT_IO flag
for userspace to know if direct IO is used to access backing
file.

Cc: linux-api@vger.kernel.org
Signed-off-by: Ming Lei <ming.lei@canonical.com>
---
 drivers/block/loop.c      | 56 +++++++++++++++++++++++++++++++++++++++++++++++
 drivers/block/loop.h      |  1 +
 include/uapi/linux/loop.h |  1 +
 3 files changed, 58 insertions(+)

diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 555b895..599d668 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -421,6 +421,51 @@ struct switch_request {
 	struct completion wait;
 };
 
+static void __loop_update_dio(struct loop_device *lo, bool dio)
+{
+	struct file *file = lo->lo_backing_file;
+	struct inode *inode = file->f_mapping->host;
+	bool use_dio;
+
+	/*
+	 * loop block's logical block size is 512, now
+	 * we support direct I/O only if the backing
+	 * block devices' minimize I/O size is 512 and
+	 * the offset is aligned with 512.
+	 */
+	if (dio) {
+		if (inode->i_sb->s_bdev &&
+			bdev_io_min(inode->i_sb->s_bdev) == 512 &&
+			!(lo->lo_offset & 511))
+			use_dio = true;
+		else
+			use_dio = false;
+	} else {
+		use_dio = false;
+	}
+
+	if (lo->use_dio == use_dio)
+		return;
+
+	/* flush dirty pages before changing direct IO */
+	vfs_fsync(file, 0);
+
+	/*
+	 * The flag of LO_FLAGS_DIRECT_IO is handled similarly with
+	 * LO_FLAGS_READ_ONLY, both are set from kernel, and losetup
+	 * will get updated by ioctl(LOOP_GET_STATUS)
+	 */
+	blk_mq_freeze_queue(lo->lo_queue);
+	lo->use_dio = use_dio;
+	lo->lo_flags |= use_dio ? LO_FLAGS_DIRECT_IO : 0;
+	blk_mq_unfreeze_queue(lo->lo_queue);
+}
+
+static inline void loop_update_dio(struct loop_device *lo)
+{
+	__loop_update_dio(lo, io_is_direct(lo->lo_backing_file));
+}
+
 /*
  * Do the actual switch; called from the BIO completion routine
  */
@@ -441,6 +486,7 @@ static void do_loop_switch(struct loop_device *lo, struct switch_request *p)
 		mapping->host->i_bdev->bd_block_size : PAGE_SIZE;
 	lo->old_gfp_mask = mapping_gfp_mask(mapping);
 	mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));
+	loop_update_dio(lo);
 }
 
 /*
@@ -627,11 +673,19 @@ static ssize_t loop_attr_partscan_show(struct loop_device *lo, char *buf)
 	return sprintf(buf, "%s\n", partscan ? "1" : "0");
 }
 
+static ssize_t loop_attr_dio_show(struct loop_device *lo, char *buf)
+{
+	int dio = (lo->lo_flags & LO_FLAGS_DIRECT_IO);
+
+	return sprintf(buf, "%s\n", dio ? "1" : "0");
+}
+
 LOOP_ATTR_RO(backing_file);
 LOOP_ATTR_RO(offset);
 LOOP_ATTR_RO(sizelimit);
 LOOP_ATTR_RO(autoclear);
 LOOP_ATTR_RO(partscan);
+LOOP_ATTR_RO(dio);
 
 static struct attribute *loop_attrs[] = {
 	&loop_attr_backing_file.attr,
@@ -639,6 +693,7 @@ static struct attribute *loop_attrs[] = {
 	&loop_attr_sizelimit.attr,
 	&loop_attr_autoclear.attr,
 	&loop_attr_partscan.attr,
+	&loop_attr_dio.attr,
 	NULL,
 };
 
@@ -783,6 +838,7 @@ static int loop_set_fd(struct loop_device *lo, fmode_t mode,
 	if (!(lo_flags & LO_FLAGS_READ_ONLY) && file->f_op->fsync)
 		blk_queue_flush(lo->lo_queue, REQ_FLUSH);
 
+	loop_update_dio(lo);
 	set_capacity(lo->lo_disk, size);
 	bd_set_size(bdev, size << 9);
 	loop_sysfs_init(lo);
diff --git a/drivers/block/loop.h b/drivers/block/loop.h
index b6c7d21..d1de221 100644
--- a/drivers/block/loop.h
+++ b/drivers/block/loop.h
@@ -58,6 +58,7 @@ struct loop_device {
 	struct mutex		lo_ctl_mutex;
 	struct kthread_worker	worker;
 	struct task_struct	*worker_task;
+	bool			use_dio;
 
 	struct request_queue	*lo_queue;
 	struct blk_mq_tag_set	tag_set;
diff --git a/include/uapi/linux/loop.h b/include/uapi/linux/loop.h
index e0cecd2..949851c 100644
--- a/include/uapi/linux/loop.h
+++ b/include/uapi/linux/loop.h
@@ -21,6 +21,7 @@ enum {
 	LO_FLAGS_READ_ONLY	= 1,
 	LO_FLAGS_AUTOCLEAR	= 4,
 	LO_FLAGS_PARTSCAN	= 8,
+	LO_FLAGS_DIRECT_IO	= 16,
 };
 
 #include <asm/posix_types.h>	/* for __kernel_old_dev_t */
-- 
1.9.1

[PATCH v6 5/5] block: loop: support DIO & AIO

[Ming Lei]

There are about 3 advantages to use direct I/O and AIO on
read/write loop's backing file:

1) double cache can be avoided, then memory usage gets
decreased a lot

2) not like user space direct I/O, there isn't cost of
pinning pages

3) avoid context switch for obtaining good throughput
- in buffered file read, random I/O top throughput is often obtained
only if they are submitted concurrently from lots of tasks; but for
sequential I/O, most of times they can be hit from page cache, so
concurrent submissions often introduce unnecessary context switch
and can't improve throughput much. There was such discussion[1]
to use non-blocking I/O to improve the problem for application.
- with direct I/O and AIO, concurrent submissions can be
avoided and random read throughput can't be affected meantime

Follows my fio test result:

1. 16 jobs fio test inside ext4 file system over loop block
1) How to run
	- linux kernel: 4.1.0-rc2-next-20150506 with the patchset
	- the loop block is over one image on HDD.
	- linux psync, 16 jobs, size 400M, ext4 over loop block
	- test result: IOPS from fio output

2) Throughput result:
        -------------------------------------------------------------
        test cases          |randread   |read   |randwrite  |write  |
        -------------------------------------------------------------
        base                |240        |8705   |3763       |20914
        -------------------------------------------------------------
        base+loop aio       |242        |9258   |4577       |21451
        -------------------------------------------------------------

3) context switch
        - context switch decreased by ~16% with loop aio for randread,
	and decreased by ~33% for read

4) memory usage
	- After these four tests with loop aio: ~10% memory becomes used
	- After these four tests without loop aio: more than 55% memory
	becomes used

2. single job fio test inside ext4 file system over loop block(for Maxim Patlasov)
1) How to run
	- linux kernel: 4.1.0-rc2-next-20150506 with the patchset
	- the loop block is over one image on HDD.
	- linux psync, 1 job, size 4000M, ext4 over loop block
	- test result: IOPS from fio output

2) Throughput result:
        -------------------------------------------------------------
        test cases          |randread   |read   |randwrite  |write  |
        -------------------------------------------------------------
        base                |109        |21180  |4192       |22782
        -------------------------------------------------------------
        base+loop aio       |114        |21018  |5404       |22670
        -------------------------------------------------------------

3) context switch
        - context switch decreased by ~10% with loop aio for randread,
	and decreased by ~50% for read

4) memory usage
	- After these four tests with loop aio: ~10% memory becomes used
	- After these four tests without loop aio: more than 55% memory
	becomes used

Both 'context switch' and 'memory usage' data are got from sar.

[1] https://lwn.net/Articles/612483/
[2] sar graph when running fio over loop without the patchset
http://kernel.ubuntu.com/~ming/block/loop-aio/v3/lo-nonaio.pdf

[3] sar graph when running fio over loop with the patchset
http://kernel.ubuntu.com/~ming/block/loop-aio/v3/lo-aio.pdf

[4] sar graph when running fio over loop without the patchset
http://kernel.ubuntu.com/~ming/block/loop-aio/v3/lo-nonaio-1job.pdf

[5] sar graph when running fio over loop with the patchset
http://kernel.ubuntu.com/~ming/block/loop-aio/v3/lo-aio-1job.pdf

Signed-off-by: Ming Lei <ming.lei@canonical.com>
---
 drivers/block/loop.c | 98 ++++++++++++++++++++++++++++++++++++++++++++++++++--
 drivers/block/loop.h |  2 ++
 2 files changed, 97 insertions(+), 3 deletions(-)

diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 599d668..b0ff9c6 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -389,6 +389,90 @@ static int lo_req_flush(struct loop_device *lo, struct request *rq)
 	return ret;
 }
 
+static inline void handle_partial_read(struct loop_cmd *cmd, long bytes)
+{
+	if (bytes < 0 || (cmd->rq->cmd_flags & REQ_WRITE))
+		return;
+
+	if (unlikely(bytes < blk_rq_bytes(cmd->rq))) {
+		struct bio *bio = cmd->rq->bio;
+
+		bio_advance(bio, bytes);
+		zero_fill_bio(bio);
+	}
+}
+
+static void lo_rw_aio_complete(struct kiocb *iocb, long ret, long ret2)
+{
+	struct loop_cmd *cmd = container_of(iocb, struct loop_cmd, iocb);
+	struct request *rq = cmd->rq;
+
+	handle_partial_read(cmd, ret);
+
+	if (ret > 0)
+		ret = 0;
+	else if (ret < 0)
+		ret = -EIO;
+
+	rq->errors = ret;
+	blk_mq_complete_request(rq);
+}
+
+static int lo_rw_aio(struct loop_device *lo, struct loop_cmd *cmd,
+		     loff_t pos, bool rw)
+{
+	struct iov_iter iter;
+	struct bio_vec *bvec;
+	struct bio *bio = cmd->rq->bio;
+	struct file *file = lo->lo_backing_file;
+	int ret;
+
+	/* nomerge for loop request queue */
+	WARN_ON(cmd->rq->bio != cmd->rq->biotail);
+
+	bvec = __bvec_iter_bvec(bio->bi_io_vec, bio->bi_iter);
+	iov_iter_bvec(&iter, ITER_BVEC | rw, bvec,
+		      bio_segments(bio), blk_rq_bytes(cmd->rq));
+
+	cmd->iocb.ki_pos = pos;
+	cmd->iocb.ki_filp = file;
+	cmd->iocb.ki_complete = lo_rw_aio_complete;
+	cmd->iocb.ki_flags = IOCB_DIRECT;
+
+	if (rw == WRITE)
+		ret = file->f_op->write_iter(&cmd->iocb, &iter);
+	else
+		ret = file->f_op->read_iter(&cmd->iocb, &iter);
+
+	if (ret != -EIOCBQUEUED)
+		cmd->iocb.ki_complete(&cmd->iocb, ret, 0);
+	return 0;
+}
+
+
+static inline int lo_rw_simple(struct loop_device *lo,
+		struct request *rq, loff_t pos, bool rw)
+{
+	struct loop_cmd *cmd = blk_mq_rq_to_pdu(rq);
+
+	if (cmd->use_aio)
+		return lo_rw_aio(lo, cmd, pos, rw);
+
+	/*
+	 * lo_write_simple and lo_read_simple should have been covered
+	 * by io submit style function like lo_rw_aio(), one blocker
+	 * is that lo_read_simple() need to call flush_dcache_page after
+	 * the page is written from kernel, and it isn't easy to handle
+	 * this in io submit style function which submits all segments
+	 * of the req at one time. And direct read IO doesn't need to
+	 * run flush_dcache_page().
+	 */
+	if (rw == WRITE)
+		return lo_write_simple(lo, rq, pos);
+	else
+		return lo_read_simple(lo, rq, pos);
+}
+
 static int do_req_filebacked(struct loop_device *lo, struct request *rq)
 {
 	loff_t pos;
@@ -404,13 +488,13 @@ static int do_req_filebacked(struct loop_device *lo, struct request *rq)
 		else if (lo->transfer)
 			ret = lo_write_transfer(lo, rq, pos);
 		else
-			ret = lo_write_simple(lo, rq, pos);
+			ret = lo_rw_simple(lo, rq, pos, WRITE);
 
 	} else {
 		if (lo->transfer)
 			ret = lo_read_transfer(lo, rq, pos);
 		else
-			ret = lo_read_simple(lo, rq, pos);
+			ret = lo_rw_simple(lo, rq, pos, READ);
 	}
 
 	return ret;
@@ -1530,6 +1614,12 @@ static int loop_queue_rq(struct blk_mq_hw_ctx *hctx,
 	if (lo->lo_state != Lo_bound)
 		return -EIO;
 
+	if (lo->use_dio && !lo->transfer &&
+			!(cmd->rq->cmd_flags & (REQ_FLUSH | REQ_DISCARD)))
+		cmd->use_aio = true;
+	else
+		cmd->use_aio = false;
+
 	queue_kthread_work(&lo->worker, &cmd->work);
 
 	return BLK_MQ_RQ_QUEUE_OK;
@@ -1549,7 +1639,9 @@ static void loop_handle_cmd(struct loop_cmd *cmd)
  failed:
 	if (ret)
 		cmd->rq->errors = -EIO;
-	blk_mq_complete_request(cmd->rq);
+	/* complete non-aio request */
+	if (!cmd->use_aio || ret)
+		blk_mq_complete_request(cmd->rq);
 }
 
 static void loop_queue_work(struct kthread_work *work)
diff --git a/drivers/block/loop.h b/drivers/block/loop.h
index d1de221..fb2237c 100644
--- a/drivers/block/loop.h
+++ b/drivers/block/loop.h
@@ -69,6 +69,8 @@ struct loop_cmd {
 	struct kthread_work work;
 	struct request *rq;
 	struct list_head list;
+	bool use_aio;           /* use AIO interface to handle I/O */
+	struct kiocb iocb;
 };
 
 /* Support for loadable transfer modules */
-- 
1.9.1