From: Jens Axboe <axboe@kernel.dk>
To: linux-block@vger.kernel.org, linux-nvme@lists.infradead.org
Cc: Jens Axboe <axboe@kernel.dk>
Subject: [PATCH 1/7] block: avoid ordered task state change for polled IO
Date: Sat, 17 Nov 2018 14:43:48 -0700 [thread overview]
Message-ID: <20181117214354.822-2-axboe@kernel.dk> (raw)
In-Reply-To: <20181117214354.822-1-axboe@kernel.dk>
For the core poll helper, the task state setting don't need
to imply any atomics, as it's the current task itself that
is being modified and we're not going to sleep.
For IRQ driven, the wakeup path have the necessary barriers
to not need us using the heavy handed version of the task
state setting.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
---
block/blk-mq.c | 4 ++--
fs/block_dev.c | 7 +++++--
fs/iomap.c | 3 ++-
mm/page_io.c | 3 ++-
4 files changed, 11 insertions(+), 6 deletions(-)
diff --git a/block/blk-mq.c b/block/blk-mq.c
index 32b246ed44c0..7fc4abb4cc36 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -3331,12 +3331,12 @@ static bool __blk_mq_poll(struct blk_mq_hw_ctx *hctx, struct request *rq)
ret = q->mq_ops->poll(hctx, rq->tag);
if (ret > 0) {
hctx->poll_success++;
- set_current_state(TASK_RUNNING);
+ __set_current_state(TASK_RUNNING);
return true;
}
if (signal_pending_state(state, current))
- set_current_state(TASK_RUNNING);
+ __set_current_state(TASK_RUNNING);
if (current->state == TASK_RUNNING)
return true;
diff --git a/fs/block_dev.c b/fs/block_dev.c
index 4d79bc80fb41..64ba27b8b754 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -237,9 +237,11 @@ __blkdev_direct_IO_simple(struct kiocb *iocb, struct iov_iter *iter,
qc = submit_bio(&bio);
for (;;) {
- set_current_state(TASK_UNINTERRUPTIBLE);
+ __set_current_state(TASK_UNINTERRUPTIBLE);
+
if (!READ_ONCE(bio.bi_private))
break;
+
if (!(iocb->ki_flags & IOCB_HIPRI) ||
!blk_poll(bdev_get_queue(bdev), qc))
io_schedule();
@@ -415,7 +417,8 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages)
return -EIOCBQUEUED;
for (;;) {
- set_current_state(TASK_UNINTERRUPTIBLE);
+ __set_current_state(TASK_UNINTERRUPTIBLE);
+
if (!READ_ONCE(dio->waiter))
break;
diff --git a/fs/iomap.c b/fs/iomap.c
index b0462b363bad..c5df035ace6f 100644
--- a/fs/iomap.c
+++ b/fs/iomap.c
@@ -1888,7 +1888,8 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter,
return -EIOCBQUEUED;
for (;;) {
- set_current_state(TASK_UNINTERRUPTIBLE);
+ __set_current_state(TASK_UNINTERRUPTIBLE);
+
if (!READ_ONCE(dio->submit.waiter))
break;
diff --git a/mm/page_io.c b/mm/page_io.c
index 57572ff46016..a7271fa481f6 100644
--- a/mm/page_io.c
+++ b/mm/page_io.c
@@ -405,7 +405,8 @@ int swap_readpage(struct page *page, bool synchronous)
bio_get(bio);
qc = submit_bio(bio);
while (synchronous) {
- set_current_state(TASK_UNINTERRUPTIBLE);
+ __set_current_state(TASK_UNINTERRUPTIBLE);
+
if (!READ_ONCE(bio->bi_private))
break;
--
2.17.1
WARNING: multiple messages have this Message-ID (diff)
From: axboe@kernel.dk (Jens Axboe)
Subject: [PATCH 1/7] block: avoid ordered task state change for polled IO
Date: Sat, 17 Nov 2018 14:43:48 -0700 [thread overview]
Message-ID: <20181117214354.822-2-axboe@kernel.dk> (raw)
In-Reply-To: <20181117214354.822-1-axboe@kernel.dk>
For the core poll helper, the task state setting don't need
to imply any atomics, as it's the current task itself that
is being modified and we're not going to sleep.
For IRQ driven, the wakeup path have the necessary barriers
to not need us using the heavy handed version of the task
state setting.
Signed-off-by: Jens Axboe <axboe at kernel.dk>
---
block/blk-mq.c | 4 ++--
fs/block_dev.c | 7 +++++--
fs/iomap.c | 3 ++-
mm/page_io.c | 3 ++-
4 files changed, 11 insertions(+), 6 deletions(-)
diff --git a/block/blk-mq.c b/block/blk-mq.c
index 32b246ed44c0..7fc4abb4cc36 100644
--- a/block/blk-mq.c
+++ b/block/blk-mq.c
@@ -3331,12 +3331,12 @@ static bool __blk_mq_poll(struct blk_mq_hw_ctx *hctx, struct request *rq)
ret = q->mq_ops->poll(hctx, rq->tag);
if (ret > 0) {
hctx->poll_success++;
- set_current_state(TASK_RUNNING);
+ __set_current_state(TASK_RUNNING);
return true;
}
if (signal_pending_state(state, current))
- set_current_state(TASK_RUNNING);
+ __set_current_state(TASK_RUNNING);
if (current->state == TASK_RUNNING)
return true;
diff --git a/fs/block_dev.c b/fs/block_dev.c
index 4d79bc80fb41..64ba27b8b754 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -237,9 +237,11 @@ __blkdev_direct_IO_simple(struct kiocb *iocb, struct iov_iter *iter,
qc = submit_bio(&bio);
for (;;) {
- set_current_state(TASK_UNINTERRUPTIBLE);
+ __set_current_state(TASK_UNINTERRUPTIBLE);
+
if (!READ_ONCE(bio.bi_private))
break;
+
if (!(iocb->ki_flags & IOCB_HIPRI) ||
!blk_poll(bdev_get_queue(bdev), qc))
io_schedule();
@@ -415,7 +417,8 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages)
return -EIOCBQUEUED;
for (;;) {
- set_current_state(TASK_UNINTERRUPTIBLE);
+ __set_current_state(TASK_UNINTERRUPTIBLE);
+
if (!READ_ONCE(dio->waiter))
break;
diff --git a/fs/iomap.c b/fs/iomap.c
index b0462b363bad..c5df035ace6f 100644
--- a/fs/iomap.c
+++ b/fs/iomap.c
@@ -1888,7 +1888,8 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter,
return -EIOCBQUEUED;
for (;;) {
- set_current_state(TASK_UNINTERRUPTIBLE);
+ __set_current_state(TASK_UNINTERRUPTIBLE);
+
if (!READ_ONCE(dio->submit.waiter))
break;
diff --git a/mm/page_io.c b/mm/page_io.c
index 57572ff46016..a7271fa481f6 100644
--- a/mm/page_io.c
+++ b/mm/page_io.c
@@ -405,7 +405,8 @@ int swap_readpage(struct page *page, bool synchronous)
bio_get(bio);
qc = submit_bio(bio);
while (synchronous) {
- set_current_state(TASK_UNINTERRUPTIBLE);
+ __set_current_state(TASK_UNINTERRUPTIBLE);
+
if (!READ_ONCE(bio->bi_private))
break;
--
2.17.1
next prev parent reply other threads:[~2018-11-17 21:44 UTC|newest]
Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-11-17 21:43 [PATCHSET v4] Various block optimizations Jens Axboe
2018-11-17 21:43 ` Jens Axboe
2018-11-17 21:43 ` Jens Axboe [this message]
2018-11-17 21:43 ` [PATCH 1/7] block: avoid ordered task state change for polled IO Jens Axboe
2018-11-19 7:58 ` Christoph Hellwig
2018-11-19 7:58 ` Christoph Hellwig
2018-11-17 21:43 ` [PATCH 2/7] block: have ->poll_fn() return number of entries polled Jens Axboe
2018-11-17 21:43 ` Jens Axboe
2018-11-19 7:58 ` Christoph Hellwig
2018-11-19 7:58 ` Christoph Hellwig
2018-11-25 10:52 ` Sagi Grimberg
2018-11-25 10:52 ` Sagi Grimberg
2018-11-25 13:41 ` Jens Axboe
2018-11-25 13:41 ` Jens Axboe
2018-11-27 10:10 ` Sagi Grimberg
2018-11-27 10:10 ` Sagi Grimberg
2018-11-27 15:20 ` Jens Axboe
2018-11-27 15:20 ` Jens Axboe
2018-11-17 21:43 ` [PATCH 3/7] nvme-fc: remove unused poll implementation Jens Axboe
2018-11-17 21:43 ` Jens Axboe
2018-11-19 7:59 ` Christoph Hellwig
2018-11-19 7:59 ` Christoph Hellwig
2018-11-19 15:19 ` Jens Axboe
2018-11-19 15:19 ` Jens Axboe
2018-11-19 19:00 ` James Smart
2018-11-19 19:00 ` James Smart
2018-11-17 21:43 ` [PATCH 4/7] blk-mq: when polling for IO, look for any completion Jens Axboe
2018-11-17 21:43 ` Jens Axboe
2018-11-19 8:02 ` Christoph Hellwig
2018-11-19 8:02 ` Christoph Hellwig
2018-11-19 15:20 ` Jens Axboe
2018-11-19 15:20 ` Jens Axboe
2018-11-17 21:43 ` [PATCH 5/7] blk-mq: remove 'tag' parameter from mq_ops->poll() Jens Axboe
2018-11-17 21:43 ` Jens Axboe
2018-11-19 8:02 ` Christoph Hellwig
2018-11-19 8:02 ` Christoph Hellwig
2018-11-17 21:43 ` [PATCH 6/7] block: make blk_poll() take a parameter on whether to spin or not Jens Axboe
2018-11-17 21:43 ` Jens Axboe
2018-11-19 8:05 ` Christoph Hellwig
2018-11-19 8:05 ` Christoph Hellwig
2018-11-19 15:21 ` Jens Axboe
2018-11-19 15:21 ` Jens Axboe
2018-11-20 9:11 ` Sagi Grimberg
2018-11-20 9:11 ` Sagi Grimberg
2018-11-17 21:43 ` [PATCH 7/7] blk-mq: ensure mq_ops ->poll() is entered at least once Jens Axboe
2018-11-17 21:43 ` Jens Axboe
2018-11-19 8:05 ` Christoph Hellwig
2018-11-19 8:05 ` Christoph Hellwig
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20181117214354.822-2-axboe@kernel.dk \
--to=axboe@kernel.dk \
--cc=linux-block@vger.kernel.org \
--cc=linux-nvme@lists.infradead.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.