From: Goldwyn Rodrigues <rgoldwyn@suse.de>
To: Christoph Hellwig <hch@infradead.org>
Cc: linux-btrfs@vger.kernel.org, darrick.wong@oracle.com,
fdmanana@kernel.org, nborisov@suse.com, dsterba@suse.cz,
jthumshirn@suse.de, linux-fsdevel@vger.kernel.org
Subject: Re: [PATCH 4/8] btrfs: Switch to iomap_dio_rw() for dio
Date: Thu, 2 Jan 2020 12:01:27 -0600 [thread overview]
Message-ID: <20200102180127.65oh2zmclpmy75ix@fiona> (raw)
In-Reply-To: <20191221144226.GA25804@infradead.org>
On 6:42 21/12, Christoph Hellwig wrote:
> So Ilooked into the "unlocked" direct I/O case, and I think the current
> code using dio_sem is really sketchy. What btrfs really needs to do is
> take i_rwsem shared by default for direct writes, and only upgrade to
> the exclusive lock when needed, similar to xfs and the WIP ext4 code.
Sketchy in what sense? I am not trying to second-guess, but I want to
know where it could fail. I would want it to be simpler as well, but if
we can perform direct writes without locking, why should we introduce
locks.
>
> While looking for that I also noticed two other things:
>
> - check_direct_IO looks pretty bogus
> - btrfs_direct_IO really should be split and folded into the two
> callers
Thanks for the cleanups. I will incorporate these.
>
> Untested patches attached. The first should probably go into a prep
> patch, and the second could be folded into this one.
> From bc285e440a50140beb456f11e545a049bdf51ec1 Mon Sep 17 00:00:00 2001
> From: Christoph Hellwig <hch@lst.de>
> Date: Sat, 21 Dec 2019 15:17:26 +0100
> Subject: btrfs: remove direct I/O aligment checks
>
> The direct I/O code itself already checks for the proper sector
> size alignment, so remove the duplicate checks. The remainder of
> check_direct_IO is not ony needed for reads and can be moved to
> file.c and outside of i_rwsem.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
> fs/btrfs/file.c | 34 +++++++++++++++++++++++++++-------
> fs/btrfs/inode.c | 37 -------------------------------------
> 2 files changed, 27 insertions(+), 44 deletions(-)
>
> diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
> index a6d41d7bf362..0522f6d45a98 100644
> --- a/fs/btrfs/file.c
> +++ b/fs/btrfs/file.c
> @@ -3444,21 +3444,41 @@ static int btrfs_file_open(struct inode *inode, struct file *filp)
> return generic_file_open(inode, filp);
> }
>
> -static ssize_t btrfs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
> +/*
> + * If there are duplicate iov_base's in this iovec, fall back to buffered I/O
> + * to avoid checksum errors.
> + */
> +static bool btrfs_direct_read_ok(struct kiocb *iocb, struct iov_iter *iter)
> {
> - ssize_t ret = 0;
> + int seg, i;
>
> - if (iocb->ki_flags & IOCB_DIRECT) {
> + if (!iter_is_iovec(iter))
> + return true;
> +
> + for (seg = 0; seg < iter->nr_segs; seg++) {
> + for (i = seg + 1; i < iter->nr_segs; i++) {
> + if (iter->iov[seg].iov_base == iter->iov[i].iov_base)
> + return false;
> + }
> + }
> +
> + return true;
> +}
> +
> +
> +static ssize_t btrfs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
> +{
> + if ((iocb->ki_flags & IOCB_DIRECT) && btrfs_direct_read_ok(iocb, to)) {
> struct inode *inode = file_inode(iocb->ki_filp);
> + ssize_t ret;
>
> inode_lock_shared(inode);
> ret = btrfs_direct_IO(iocb, to);
> inode_unlock_shared(inode);
> - if (ret < 0)
> - return ret;
> - }
>
> - return generic_file_buffered_read(iocb, to, ret);
> + return ret;
> + }
> + return generic_file_buffered_read(iocb, to, 0);
> }
>
> const struct file_operations btrfs_file_operations = {
> diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
> index 824f318cee5e..18d153a62655 100644
> --- a/fs/btrfs/inode.c
> +++ b/fs/btrfs/inode.c
> @@ -8581,39 +8581,6 @@ static blk_qc_t btrfs_submit_direct(struct bio *dio_bio, struct file *file,
> return BLK_QC_T_NONE;
> }
>
> -static ssize_t check_direct_IO(struct btrfs_fs_info *fs_info,
> - const struct iov_iter *iter, loff_t offset)
> -{
> - int seg;
> - int i;
> - unsigned int blocksize_mask = fs_info->sectorsize - 1;
> - ssize_t retval = -EINVAL;
> -
> - if (offset & blocksize_mask)
> - goto out;
> -
> - if (iov_iter_alignment(iter) & blocksize_mask)
> - goto out;
> -
> - /* If this is a write we don't need to check anymore */
> - if (iov_iter_rw(iter) != READ || !iter_is_iovec(iter))
> - return 0;
> - /*
> - * Check to make sure we don't have duplicate iov_base's in this
> - * iovec, if so return EINVAL, otherwise we'll get csum errors
> - * when reading back.
> - */
> - for (seg = 0; seg < iter->nr_segs; seg++) {
> - for (i = seg + 1; i < iter->nr_segs; i++) {
> - if (iter->iov[seg].iov_base == iter->iov[i].iov_base)
> - goto out;
> - }
> - }
> - retval = 0;
> -out:
> - return retval;
> -}
> -
> static const struct iomap_ops btrfs_dio_iomap_ops = {
> .iomap_begin = btrfs_dio_iomap_begin,
> .iomap_end = btrfs_dio_iomap_end,
> @@ -8635,7 +8602,6 @@ ssize_t btrfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
> {
> struct file *file = iocb->ki_filp;
> struct inode *inode = file->f_mapping->host;
> - struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
> struct extent_changeset *data_reserved = NULL;
> loff_t offset = iocb->ki_pos;
> size_t count = 0;
> @@ -8644,9 +8610,6 @@ ssize_t btrfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
>
> lockdep_assert_held(&inode->i_rwsem);
>
> - if (check_direct_IO(fs_info, iter, offset))
> - return 0;
> -
> count = iov_iter_count(iter);
> if (iov_iter_rw(iter) == WRITE) {
> /*
> --
> 2.24.0
>
> From 7194fa1986a48af46d2b01457865066cdbd14e35 Mon Sep 17 00:00:00 2001
> From: Christoph Hellwig <hch@lst.de>
> Date: Sat, 21 Dec 2019 15:23:41 +0100
> Subject: btrfs: split btrfs_direct_IO
>
> The read and write versions don't have anything in common except
> for the call to iomap_dio_rw. So split this function, and merge
> each half into its only caller.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
> fs/btrfs/ctree.h | 4 ++-
> fs/btrfs/file.c | 44 +++++++++++++++++++++++++----
> fs/btrfs/inode.c | 72 ++++--------------------------------------------
> 3 files changed, 48 insertions(+), 72 deletions(-)
>
> diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
> index 8faa069b0a73..fccbbfebdf88 100644
> --- a/fs/btrfs/ctree.h
> +++ b/fs/btrfs/ctree.h
> @@ -28,6 +28,7 @@
> #include <linux/dynamic_debug.h>
> #include <linux/refcount.h>
> #include <linux/crc32c.h>
> +#include <linux/iomap.h>
> #include "extent-io-tree.h"
> #include "extent_io.h"
> #include "extent_map.h"
> @@ -2904,7 +2905,8 @@ int btrfs_writepage_cow_fixup(struct page *page, u64 start, u64 end);
> void btrfs_writepage_endio_finish_ordered(struct page *page, u64 start,
> u64 end, int uptodate);
> extern const struct dentry_operations btrfs_dentry_operations;
> -ssize_t btrfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
> +const struct iomap_ops btrfs_dio_iomap_ops;
> +const struct iomap_dio_ops btrfs_dio_ops;
>
> /* ioctl.c */
> long btrfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
> diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
> index 0522f6d45a98..ed0b2e015d8d 100644
> --- a/fs/btrfs/file.c
> +++ b/fs/btrfs/file.c
> @@ -1822,17 +1822,50 @@ static noinline ssize_t btrfs_buffered_write(struct kiocb *iocb,
> return num_written ? num_written : ret;
> }
>
> -static ssize_t __btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from)
> +static ssize_t btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from)
> {
> struct file *file = iocb->ki_filp;
> struct inode *inode = file_inode(file);
> - loff_t pos;
> + size_t count = iov_iter_count(from);
> + struct extent_changeset *data_reserved = NULL;
> + loff_t pos = iocb->ki_pos;
> ssize_t written;
> ssize_t written_buffered;
> loff_t endbyte;
> + bool relock = false;
> int err;
>
> - written = btrfs_direct_IO(iocb, from);
> + /*
> + * If the write DIO is beyond the EOF, we need update the isize, but
> + * it is protected by i_mutex. So we can not unlock the i_mutex in
> + * this case.
> + */
> + if (pos + count <= inode->i_size) {
> + inode_unlock(inode);
> + relock = true;
> + } else {
> + if (iocb->ki_flags & IOCB_NOWAIT)
> + return -EAGAIN;
> + }
> +
> + err = btrfs_delalloc_reserve_space(inode, &data_reserved, pos, count);
> + if (err) {
> + if (relock)
> + inode_lock(inode);
> + return err;
> + }
> +
> + down_read(&BTRFS_I(inode)->dio_sem);
> + written = iomap_dio_rw(iocb, from, &btrfs_dio_iomap_ops, &btrfs_dio_ops,
> + is_sync_kiocb(iocb));
> + up_read(&BTRFS_I(inode)->dio_sem);
> + if (written >= 0 && (size_t)written < count)
> + btrfs_delalloc_release_space(inode, data_reserved,
> + pos, count - (size_t)written, true);
> + btrfs_delalloc_release_extents(BTRFS_I(inode), count);
> + if (relock)
> + inode_lock(inode);
> + extent_changeset_free(data_reserved);
>
> if (written < 0 || !iov_iter_count(from))
> return written;
> @@ -1975,7 +2008,7 @@ static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
> atomic_inc(&BTRFS_I(inode)->sync_writers);
>
> if (iocb->ki_flags & IOCB_DIRECT) {
> - num_written = __btrfs_direct_write(iocb, from);
> + num_written = btrfs_direct_write(iocb, from);
> } else {
> num_written = btrfs_buffered_write(iocb, from);
> if (num_written > 0)
> @@ -3473,7 +3506,8 @@ static ssize_t btrfs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
> ssize_t ret;
>
> inode_lock_shared(inode);
> - ret = btrfs_direct_IO(iocb, to);
> + ret = iomap_dio_rw(iocb, to, &btrfs_dio_iomap_ops,
> + &btrfs_dio_ops, is_sync_kiocb(iocb));
> inode_unlock_shared(inode);
>
> return ret;
> diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
> index 18d153a62655..7b747270ec40 100644
> --- a/fs/btrfs/inode.c
> +++ b/fs/btrfs/inode.c
> @@ -29,7 +29,6 @@
> #include <linux/iversion.h>
> #include <linux/swap.h>
> #include <linux/sched/mm.h>
> -#include <linux/iomap.h>
> #include <asm/unaligned.h>
> #include "misc.h"
> #include "ctree.h"
> @@ -7856,6 +7855,11 @@ static int btrfs_dio_iomap_end(struct inode *inode, loff_t pos, loff_t length,
> return 0;
> }
>
> +const struct iomap_ops btrfs_dio_iomap_ops = {
> + .iomap_begin = btrfs_dio_iomap_begin,
> + .iomap_end = btrfs_dio_iomap_end,
> +};
> +
> static inline blk_status_t submit_dio_repair_bio(struct inode *inode,
> struct bio *bio,
> int mirror_num)
> @@ -8581,74 +8585,10 @@ static blk_qc_t btrfs_submit_direct(struct bio *dio_bio, struct file *file,
> return BLK_QC_T_NONE;
> }
>
> -static const struct iomap_ops btrfs_dio_iomap_ops = {
> - .iomap_begin = btrfs_dio_iomap_begin,
> - .iomap_end = btrfs_dio_iomap_end,
> -};
> -
> -static const struct iomap_dio_ops btrfs_dops = {
> +const struct iomap_dio_ops btrfs_dio_ops = {
> .submit_io = btrfs_submit_direct,
> };
>
> -
> -/*
> - * btrfs_direct_IO - perform direct I/O
> - * inode->i_rwsem must be locked before calling this function, shared or exclusive.
> - * @iocb - kernel iocb
> - * @iter - iter to/from data is copied
> - */
> -
> -ssize_t btrfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
> -{
> - struct file *file = iocb->ki_filp;
> - struct inode *inode = file->f_mapping->host;
> - struct extent_changeset *data_reserved = NULL;
> - loff_t offset = iocb->ki_pos;
> - size_t count = 0;
> - bool relock = false;
> - ssize_t ret;
> -
> - lockdep_assert_held(&inode->i_rwsem);
> -
> - count = iov_iter_count(iter);
> - if (iov_iter_rw(iter) == WRITE) {
> - /*
> - * If the write DIO is beyond the EOF, we need update
> - * the isize, but it is protected by i_mutex. So we can
> - * not unlock the i_mutex at this case.
> - */
> - if (offset + count <= inode->i_size) {
> - inode_unlock(inode);
> - relock = true;
> - } else if (iocb->ki_flags & IOCB_NOWAIT) {
> - ret = -EAGAIN;
> - goto out;
> - }
> - ret = btrfs_delalloc_reserve_space(inode, &data_reserved,
> - offset, count);
> - if (ret)
> - goto out;
> -
> - down_read(&BTRFS_I(inode)->dio_sem);
> - }
> -
> - ret = iomap_dio_rw(iocb, iter, &btrfs_dio_iomap_ops, &btrfs_dops,
> - is_sync_kiocb(iocb));
> -
> - if (iov_iter_rw(iter) == WRITE) {
> - up_read(&BTRFS_I(inode)->dio_sem);
> - if (ret >= 0 && (size_t)ret < count)
> - btrfs_delalloc_release_space(inode, data_reserved,
> - offset, count - (size_t)ret, true);
> - btrfs_delalloc_release_extents(BTRFS_I(inode), count);
> - }
> -out:
> - if (relock)
> - inode_lock(inode);
> - extent_changeset_free(data_reserved);
> - return ret;
> -}
> -
> #define BTRFS_FIEMAP_FLAGS (FIEMAP_FLAG_SYNC)
>
> static int btrfs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
> --
> 2.24.0
>
--
Goldwyn
next prev parent reply other threads:[~2020-01-02 18:01 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-12-13 19:57 [PATCH 0/8 v6] btrfs direct-io using iomap Goldwyn Rodrigues
2019-12-13 19:57 ` [PATCH 1/8] fs: Export generic_file_buffered_read() Goldwyn Rodrigues
2019-12-13 19:57 ` [PATCH 2/8] iomap: add a filesystem hook for direct I/O bio submission Goldwyn Rodrigues
2019-12-14 0:31 ` Darrick J. Wong
2019-12-18 2:02 ` Darrick J. Wong
2019-12-13 19:57 ` [PATCH 3/8] iomap: Move lockdep_assert_held() to iomap_dio_rw() calls Goldwyn Rodrigues
2019-12-14 0:32 ` Darrick J. Wong
2019-12-18 2:04 ` Darrick J. Wong
2019-12-21 13:41 ` Christoph Hellwig
2019-12-21 13:42 ` Christoph Hellwig
2019-12-21 18:02 ` Darrick J. Wong
2019-12-13 19:57 ` [PATCH 4/8] btrfs: Switch to iomap_dio_rw() for dio Goldwyn Rodrigues
2019-12-21 14:42 ` Christoph Hellwig
2020-01-02 18:01 ` Goldwyn Rodrigues [this message]
2020-01-07 17:23 ` Christoph Hellwig
2020-01-07 11:59 ` Goldwyn Rodrigues
2020-01-07 17:21 ` Christoph Hellwig
2019-12-13 19:57 ` [PATCH 5/8] fs: Remove dio_end_io() Goldwyn Rodrigues
2019-12-13 19:57 ` [PATCH 6/8] btrfs: Wait for extent bits to release page Goldwyn Rodrigues
2019-12-13 19:57 ` [PATCH 7/8] btrfs: Use ->iomap_end() instead of btrfs_dio_data Goldwyn Rodrigues
2019-12-13 19:57 ` [PATCH 8/8] btrfs: remove BTRFS_INODE_READDIO_NEED_LOCK Goldwyn Rodrigues
2019-12-16 0:01 ` [PATCH 0/8 v6] btrfs direct-io using iomap Nikolay Borisov
2019-12-16 12:41 ` Goldwyn Rodrigues
-- strict thread matches above, loose matches on Subject: below --
2019-12-10 23:01 [PATCH 0/8 v4] " Goldwyn Rodrigues
2019-12-10 23:01 ` [PATCH 4/8] btrfs: Switch to iomap_dio_rw() for dio Goldwyn Rodrigues
2019-12-11 8:58 ` Filipe Manana
2019-12-11 10:43 ` Nikolay Borisov
2019-12-05 15:56 [PATCH 0/8 v3] btrfs direct-io using iomap Goldwyn Rodrigues
2019-12-05 15:56 ` [PATCH 4/8] btrfs: Switch to iomap_dio_rw() for dio Goldwyn Rodrigues
2019-12-05 17:18 ` Johannes Thumshirn
2019-12-05 17:19 ` Christoph Hellwig
2019-12-05 17:32 ` Johannes Thumshirn
2019-12-05 17:33 ` Christoph Hellwig
2019-12-05 17:36 ` Johannes Thumshirn
2019-12-05 17:37 ` Christoph Hellwig
2019-12-05 17:37 ` Christoph Hellwig
2019-12-05 17:40 ` Johannes Thumshirn
2019-12-05 17:44 ` Goldwyn Rodrigues
2019-12-05 22:59 ` Nikolay Borisov
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=20200102180127.65oh2zmclpmy75ix@fiona \
--to=rgoldwyn@suse.de \
--cc=darrick.wong@oracle.com \
--cc=dsterba@suse.cz \
--cc=fdmanana@kernel.org \
--cc=hch@infradead.org \
--cc=jthumshirn@suse.de \
--cc=linux-btrfs@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=nborisov@suse.com \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).