From: Nitesh Shetty <nj.shetty@samsung.com> To: linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org, viro@zeniv.linux.org.uk Cc: linux-block@vger.kernel.org, axboe@fb.com, linux-nvme@lists.infradead.org, nj.shetty@samsung.com, joshi.k@samsung.com Subject: [PATCH] blk: optimization for classic polling Date: Sun, 30 May 2083 09:51:06 +0530 [thread overview] Message-ID: <3578876466-3733-1-git-send-email-nj.shetty@samsung.com> (raw) In-Reply-To: CGME20180208143655epcas1p444bf34705526b7839a5a135f82761aad@epcas1p4.samsung.com This removes the dependency on interrupts to wake up task. Set task state as TASK_RUNNING, if need_resched() returns true, while polling for IO completion. Earlier, polling task used to sleep, relying on interrupt to wake it up. This made some IO take very long when interrupt-coalescing is enabled in NVMe. Reference: http://lists.infradead.org/pipermail/linux-nvme/2018-February/015435.html Signed-off-by: Nitesh Shetty <nj.shetty@samsung.com> --- fs/block_dev.c | 16 ++++++++++++---- fs/direct-io.c | 8 ++++++-- fs/iomap.c | 10 +++++++--- 3 files changed, 25 insertions(+), 9 deletions(-) diff --git a/fs/block_dev.c b/fs/block_dev.c index 4a181fc..a87d8b7 100644 --- a/fs/block_dev.c +++ b/fs/block_dev.c @@ -236,9 +236,13 @@ __blkdev_direct_IO_simple(struct kiocb *iocb, struct iov_iter *iter, 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)) + if (!(iocb->ki_flags & IOCB_HIPRI)) io_schedule(); + else if (!blk_poll(bdev_get_queue(bdev), qc)) { + if (need_resched()) + set_current_state(TASK_RUNNING); + io_schedule(); + } } __set_current_state(TASK_RUNNING); @@ -401,9 +405,13 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages) if (!READ_ONCE(dio->waiter)) break; - if (!(iocb->ki_flags & IOCB_HIPRI) || - !blk_poll(bdev_get_queue(bdev), qc)) + if (!(iocb->ki_flags & IOCB_HIPRI)) io_schedule(); + else if (!blk_poll(bdev_get_queue(bdev), qc)) { + if (need_resched()) + set_current_state(TASK_RUNNING); + io_schedule(); + } } __set_current_state(TASK_RUNNING); diff --git a/fs/direct-io.c b/fs/direct-io.c index a0ca9e4..c815ac9 100644 --- a/fs/direct-io.c +++ b/fs/direct-io.c @@ -518,9 +518,13 @@ static struct bio *dio_await_one(struct dio *dio) __set_current_state(TASK_UNINTERRUPTIBLE); dio->waiter = current; spin_unlock_irqrestore(&dio->bio_lock, flags); - if (!(dio->iocb->ki_flags & IOCB_HIPRI) || - !blk_poll(dio->bio_disk->queue, dio->bio_cookie)) + if (!(dio->iocb->ki_flags & IOCB_HIPRI)) io_schedule(); + else if (!blk_poll(dio->bio_disk->queue, dio->bio_cookie)) { + if (need_resched()) + __set_current_state(TASK_RUNNING); + io_schedule(); + } /* wake up sets us TASK_RUNNING */ spin_lock_irqsave(&dio->bio_lock, flags); dio->waiter = NULL; diff --git a/fs/iomap.c b/fs/iomap.c index afd1635..b51569d 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -1072,10 +1072,14 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter, break; if (!(iocb->ki_flags & IOCB_HIPRI) || - !dio->submit.last_queue || - !blk_poll(dio->submit.last_queue, - dio->submit.cookie)) + !dio->submit.last_queue) io_schedule(); + else if (!blk_poll(dio->submit.last_queue, + dio->submit.cookie)) { + if (need_resched()) + set_current_state(TASK_RUNNING); + io_schedule(); + } } __set_current_state(TASK_RUNNING); } -- 2.7.4
WARNING: multiple messages have this Message-ID (diff)
From: nj.shetty@samsung.com (Nitesh Shetty) Subject: [PATCH] blk: optimization for classic polling Date: Thu, 08 Feb 2018 14:37:20 -0000 [thread overview] Message-ID: <3578876466-3733-1-git-send-email-nj.shetty@samsung.com> (raw) In-Reply-To: CGME20180208143655epcas1p444bf34705526b7839a5a135f82761aad@epcas1p4.samsung.com This removes the dependency on interrupts to wake up task. Set task state as TASK_RUNNING, if need_resched() returns true, while polling for IO completion. Earlier, polling task used to sleep, relying on interrupt to wake it up. This made some IO take very long when interrupt-coalescing is enabled in NVMe. Reference: http://lists.infradead.org/pipermail/linux-nvme/2018-February/015435.html Signed-off-by: Nitesh Shetty <nj.shetty at samsung.com> --- fs/block_dev.c | 16 ++++++++++++---- fs/direct-io.c | 8 ++++++-- fs/iomap.c | 10 +++++++--- 3 files changed, 25 insertions(+), 9 deletions(-) diff --git a/fs/block_dev.c b/fs/block_dev.c index 4a181fc..a87d8b7 100644 --- a/fs/block_dev.c +++ b/fs/block_dev.c @@ -236,9 +236,13 @@ __blkdev_direct_IO_simple(struct kiocb *iocb, struct iov_iter *iter, 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)) + if (!(iocb->ki_flags & IOCB_HIPRI)) io_schedule(); + else if (!blk_poll(bdev_get_queue(bdev), qc)) { + if (need_resched()) + set_current_state(TASK_RUNNING); + io_schedule(); + } } __set_current_state(TASK_RUNNING); @@ -401,9 +405,13 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages) if (!READ_ONCE(dio->waiter)) break; - if (!(iocb->ki_flags & IOCB_HIPRI) || - !blk_poll(bdev_get_queue(bdev), qc)) + if (!(iocb->ki_flags & IOCB_HIPRI)) io_schedule(); + else if (!blk_poll(bdev_get_queue(bdev), qc)) { + if (need_resched()) + set_current_state(TASK_RUNNING); + io_schedule(); + } } __set_current_state(TASK_RUNNING); diff --git a/fs/direct-io.c b/fs/direct-io.c index a0ca9e4..c815ac9 100644 --- a/fs/direct-io.c +++ b/fs/direct-io.c @@ -518,9 +518,13 @@ static struct bio *dio_await_one(struct dio *dio) __set_current_state(TASK_UNINTERRUPTIBLE); dio->waiter = current; spin_unlock_irqrestore(&dio->bio_lock, flags); - if (!(dio->iocb->ki_flags & IOCB_HIPRI) || - !blk_poll(dio->bio_disk->queue, dio->bio_cookie)) + if (!(dio->iocb->ki_flags & IOCB_HIPRI)) io_schedule(); + else if (!blk_poll(dio->bio_disk->queue, dio->bio_cookie)) { + if (need_resched()) + __set_current_state(TASK_RUNNING); + io_schedule(); + } /* wake up sets us TASK_RUNNING */ spin_lock_irqsave(&dio->bio_lock, flags); dio->waiter = NULL; diff --git a/fs/iomap.c b/fs/iomap.c index afd1635..b51569d 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -1072,10 +1072,14 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter, break; if (!(iocb->ki_flags & IOCB_HIPRI) || - !dio->submit.last_queue || - !blk_poll(dio->submit.last_queue, - dio->submit.cookie)) + !dio->submit.last_queue) io_schedule(); + else if (!blk_poll(dio->submit.last_queue, + dio->submit.cookie)) { + if (need_resched()) + set_current_state(TASK_RUNNING); + io_schedule(); + } } __set_current_state(TASK_RUNNING); } -- 2.7.4
next parent reply other threads:[~2018-02-08 14:36 UTC|newest] Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top [not found] <CGME20180208143655epcas1p444bf34705526b7839a5a135f82761aad@epcas1p4.samsung.com> 2018-02-08 14:37 ` Nitesh Shetty [this message] 2083-05-30 4:21 ` [PATCH] blk: optimization for classic polling Nitesh Shetty 2018-02-08 15:27 ` Keith Busch 2018-02-08 15:27 ` Keith Busch 2018-02-08 15:27 ` Keith Busch 2018-02-08 16:03 ` Sagi Grimberg 2018-02-08 16:03 ` Sagi Grimberg 2018-02-12 15:50 ` Bart Van Assche 2018-02-12 15:50 ` Bart Van Assche [not found] ` <CGME20180213154752epcas1p35e3c13d150d2108a06342781ec5e64d7@epcas1p3.samsung.com> 2018-02-13 15:48 ` [PATCH v3] " Nitesh Shetty 2018-02-13 15:48 ` Nitesh Shetty 2018-02-13 16:12 ` Jens Axboe 2018-02-13 16:12 ` Jens Axboe [not found] ` <CGME20180213082701epcas1p4d92a86f127a0918fb9d62120bd121cb6@epcas1p4.samsung.com> 2018-02-13 18:56 ` [PATCH v2 RESENT] " Nitesh Shetty 2018-02-13 18:56 ` Nitesh Shetty 2018-02-13 14:42 ` Jens Axboe 2018-02-13 14:42 ` Jens Axboe 2018-02-20 13:21 ` [PATCH] " Peter Zijlstra 2018-02-20 13:21 ` Peter Zijlstra 2018-02-20 16:27 ` Keith Busch 2018-02-20 16:27 ` Keith Busch 2018-02-20 22:37 ` Jens Axboe 2018-02-20 22:37 ` Jens Axboe 2018-02-21 8:32 ` Peter Zijlstra 2018-02-21 8:32 ` Peter Zijlstra 2018-10-10 18:52 ` Florian Fainelli 2018-10-10 18:52 ` Florian Fainelli
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=3578876466-3733-1-git-send-email-nj.shetty@samsung.com \ --to=nj.shetty@samsung.com \ --cc=axboe@fb.com \ --cc=joshi.k@samsung.com \ --cc=linux-block@vger.kernel.org \ --cc=linux-fsdevel@vger.kernel.org \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-nvme@lists.infradead.org \ --cc=viro@zeniv.linux.org.uk \ /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: linkBe 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.