re-enable IOCB_NOWAIT writes to files v5

re-enable IOCB_NOWAIT writes to files v5

[Christoph Hellwig]

Hi all,

commit 66fa3cedf16a ("fs: Add async write file modification handling.")
effectively disabled IOCB_NOWAIT writes as timestamp updates currently
always require blocking, and the modern timestamp resolution means we
always update timestamps.  This leads to a lot of context switches from
applications using io_uring to submit file writes, making it often worse
than using the legacy aio code that is not using IOCB_NOWAIT.

This series allows non-blocking updates for lazytime if the file system
supports it, and adds that support for XFS.

Changes since v4:
 - replace the S_* flags with an enum indicating either access or
   modification time updates to make the logic less fragile and to
   fix a bug in the previous version

Changes since v3:
 - fix was_dirty_time handling in __mark_inode_dirty for the racy flag
   update case
 - refactor inode_update_timestamps to make the lazytime vs blocking
   logical more clear
 - allow non-blocking timestamp updates for fat

Changes since v2:
 - drop patches merged upstream
 - adjust for the inode state accesors
 - keep a check in __writeback_single_inode instead of exercising
   potentially undefined behavior
 - more spelling fixes

Changes since v1:
 - more regular numbering of the S_* flags
 - fix XFS to actually not block
 - don't ignore the generic_update_time return value in
   file_update_time_flags
 - fix the sync_lazytime return value
 - fix an out of data comment in btrfs
 - fix a race that would update i_version before returning -EAGAIN in XFS

Diffstat:
 Documentation/filesystems/locking.rst |    2 
 Documentation/filesystems/vfs.rst     |    6 +
 fs/btrfs/inode.c                      |    8 +-
 fs/fs-writeback.c                     |   33 +++++++---
 fs/gfs2/inode.c                       |    6 +
 fs/inode.c                            |  111 +++++++++++++++++++++-------------
 fs/internal.h                         |    3 
 fs/nfs/inode.c                        |    4 -
 fs/orangefs/inode.c                   |    5 +
 fs/overlayfs/inode.c                  |    2 
 fs/sync.c                             |    4 -
 fs/ubifs/file.c                       |   13 ++-
 fs/xfs/xfs_iops.c                     |   34 +++++++++-
 fs/xfs/xfs_super.c                    |   29 --------
 include/linux/fs.h                    |   27 ++++++--
 include/trace/events/writeback.h      |    6 -
 16 files changed, 182 insertions(+), 111 deletions(-)

[PATCH 01/11] fs: remove inode_update_time

[Christoph Hellwig]

The only external user is gone now, open code it in the two VFS
callers.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
---
 fs/inode.c         | 23 ++++++++---------------
 include/linux/fs.h |  1 -
 2 files changed, 8 insertions(+), 16 deletions(-)

diff --git a/fs/inode.c b/fs/inode.c
index 521383223d8a..07effa0cb999 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -2157,19 +2157,6 @@ int generic_update_time(struct inode *inode, int flags)
 }
 EXPORT_SYMBOL(generic_update_time);
 
-/*
- * This does the actual work of updating an inodes time or version.  Must have
- * had called mnt_want_write() before calling this.
- */
-int inode_update_time(struct inode *inode, int flags)
-{
-	if (inode->i_op->update_time)
-		return inode->i_op->update_time(inode, flags);
-	generic_update_time(inode, flags);
-	return 0;
-}
-EXPORT_SYMBOL(inode_update_time);
-
 /**
  *	atime_needs_update	-	update the access time
  *	@path: the &struct path to update
@@ -2237,7 +2224,10 @@ void touch_atime(const struct path *path)
 	 * We may also fail on filesystems that have the ability to make parts
 	 * of the fs read only, e.g. subvolumes in Btrfs.
 	 */
-	inode_update_time(inode, S_ATIME);
+	if (inode->i_op->update_time)
+		inode->i_op->update_time(inode, S_ATIME);
+	else
+		generic_update_time(inode, S_ATIME);
 	mnt_put_write_access(mnt);
 skip_update:
 	sb_end_write(inode->i_sb);
@@ -2392,7 +2382,10 @@ static int file_update_time_flags(struct file *file, unsigned int flags)
 
 	if (mnt_get_write_access_file(file))
 		return 0;
-	ret = inode_update_time(inode, sync_mode);
+	if (inode->i_op->update_time)
+		ret = inode->i_op->update_time(inode, sync_mode);
+	else
+		ret = generic_update_time(inode, sync_mode);
 	mnt_put_write_access_file(file);
 	return ret;
 }
diff --git a/include/linux/fs.h b/include/linux/fs.h
index f5c9cf28c4dc..ee623c16d835 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -2246,7 +2246,6 @@ enum file_time_flags {
 
 extern bool atime_needs_update(const struct path *, struct inode *);
 extern void touch_atime(const struct path *);
-int inode_update_time(struct inode *inode, int flags);
 
 static inline void file_accessed(struct file *file)
 {
-- 
2.47.3

[PATCH 02/11] fs: allow error returns from generic_update_time

[Christoph Hellwig]

Now that no caller looks at the updated flags, switch generic_update_time
to the same calling convention as the ->update_time method and return 0
or a negative errno.

This prepares for adding non-blocking timestamp updates that could return
-EAGAIN.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
---
 fs/gfs2/inode.c    | 3 +--
 fs/inode.c         | 4 ++--
 fs/ubifs/file.c    | 6 ++----
 fs/xfs/xfs_iops.c  | 6 ++----
 include/linux/fs.h | 2 +-
 5 files changed, 8 insertions(+), 13 deletions(-)

diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c
index 36618e353199..e08eb419347c 100644
--- a/fs/gfs2/inode.c
+++ b/fs/gfs2/inode.c
@@ -2257,8 +2257,7 @@ static int gfs2_update_time(struct inode *inode, int flags)
 		if (error)
 			return error;
 	}
-	generic_update_time(inode, flags);
-	return 0;
+	return generic_update_time(inode, flags);
 }
 
 static const struct inode_operations gfs2_file_iops = {
diff --git a/fs/inode.c b/fs/inode.c
index 07effa0cb999..7eb28dd45a5a 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -2141,7 +2141,7 @@ EXPORT_SYMBOL(inode_update_timestamps);
  * or S_VERSION need to be updated we attempt to update all three of them. S_ATIME
  * updates can be handled done independently of the rest.
  *
- * Returns a S_* mask indicating which fields were updated.
+ * Returns a negative error value on error, else 0.
  */
 int generic_update_time(struct inode *inode, int flags)
 {
@@ -2153,7 +2153,7 @@ int generic_update_time(struct inode *inode, int flags)
 	if (updated & S_VERSION)
 		dirty_flags |= I_DIRTY_SYNC;
 	__mark_inode_dirty(inode, dirty_flags);
-	return updated;
+	return 0;
 }
 EXPORT_SYMBOL(generic_update_time);
 
diff --git a/fs/ubifs/file.c b/fs/ubifs/file.c
index c3265b8804f5..ec1bb9f43acc 100644
--- a/fs/ubifs/file.c
+++ b/fs/ubifs/file.c
@@ -1379,10 +1379,8 @@ int ubifs_update_time(struct inode *inode, int flags)
 			.dirtied_ino_d = ALIGN(ui->data_len, 8) };
 	int err, release;
 
-	if (!IS_ENABLED(CONFIG_UBIFS_ATIME_SUPPORT)) {
-		generic_update_time(inode, flags);
-		return 0;
-	}
+	if (!IS_ENABLED(CONFIG_UBIFS_ATIME_SUPPORT))
+		return generic_update_time(inode, flags);
 
 	err = ubifs_budget_space(c, &req);
 	if (err)
diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c
index ad94fbf55014..9dedb54e3cb0 100644
--- a/fs/xfs/xfs_iops.c
+++ b/fs/xfs/xfs_iops.c
@@ -1197,10 +1197,8 @@ xfs_vn_update_time(
 
 	if (inode->i_sb->s_flags & SB_LAZYTIME) {
 		if (!((flags & S_VERSION) &&
-		      inode_maybe_inc_iversion(inode, false))) {
-			generic_update_time(inode, flags);
-			return 0;
-		}
+		      inode_maybe_inc_iversion(inode, false)))
+			return generic_update_time(inode, flags);
 
 		/* Capture the iversion update that just occurred */
 		log_flags |= XFS_ILOG_CORE;
diff --git a/include/linux/fs.h b/include/linux/fs.h
index ee623c16d835..fccb0a38cb74 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -2399,7 +2399,7 @@ extern void ihold(struct inode * inode);
 extern void iput(struct inode *);
 void iput_not_last(struct inode *);
 int inode_update_timestamps(struct inode *inode, int flags);
-int generic_update_time(struct inode *, int);
+int generic_update_time(struct inode *inode, int flags);
 
 /* /sys/fs */
 extern struct kobject *fs_kobj;
-- 
2.47.3

[PATCH 03/11] nfs: split nfs_update_timestamps

[Christoph Hellwig]

The VFS paths update either the atime or ctime and mtime but never mix
between atime and the others.  Split nfs_update_timestamps to match this
to prepare for cleaning up the VFS interfaces.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 fs/nfs/inode.c | 31 +++++++++++++++----------------
 1 file changed, 15 insertions(+), 16 deletions(-)

diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 84049f3cd340..3be8ba7b98c5 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -669,35 +669,31 @@ static void nfs_set_timestamps_to_ts(struct inode *inode, struct iattr *attr)
 	NFS_I(inode)->cache_validity &= ~cache_flags;
 }
 
-static void nfs_update_timestamps(struct inode *inode, unsigned int ia_valid)
+static void nfs_update_atime(struct inode *inode)
 {
-	enum file_time_flags time_flags = 0;
-	unsigned int cache_flags = 0;
+	inode_update_timestamps(inode, S_ATIME);
+	NFS_I(inode)->cache_validity &= ~NFS_INO_INVALID_ATIME;
+}
 
-	if (ia_valid & ATTR_MTIME) {
-		time_flags |= S_MTIME | S_CTIME;
-		cache_flags |= NFS_INO_INVALID_CTIME | NFS_INO_INVALID_MTIME;
-	}
-	if (ia_valid & ATTR_ATIME) {
-		time_flags |= S_ATIME;
-		cache_flags |= NFS_INO_INVALID_ATIME;
-	}
-	inode_update_timestamps(inode, time_flags);
-	NFS_I(inode)->cache_validity &= ~cache_flags;
+static void nfs_update_mtime(struct inode *inode)
+{
+	inode_update_timestamps(inode, S_MTIME | S_CTIME);
+	NFS_I(inode)->cache_validity &=
+		~(NFS_INO_INVALID_CTIME | NFS_INO_INVALID_MTIME);
 }
 
 void nfs_update_delegated_atime(struct inode *inode)
 {
 	spin_lock(&inode->i_lock);
 	if (nfs_have_delegated_atime(inode))
-		nfs_update_timestamps(inode, ATTR_ATIME);
+		nfs_update_atime(inode);
 	spin_unlock(&inode->i_lock);
 }
 
 void nfs_update_delegated_mtime_locked(struct inode *inode)
 {
 	if (nfs_have_delegated_mtime(inode))
-		nfs_update_timestamps(inode, ATTR_MTIME);
+		nfs_update_mtime(inode);
 }
 
 void nfs_update_delegated_mtime(struct inode *inode)
@@ -747,7 +743,10 @@ nfs_setattr(struct mnt_idmap *idmap, struct dentry *dentry,
 						ATTR_ATIME|ATTR_ATIME_SET);
 			}
 		} else {
-			nfs_update_timestamps(inode, attr->ia_valid);
+			if (attr->ia_valid & ATTR_MTIME)
+				nfs_update_mtime(inode);
+			if (attr->ia_valid & ATTR_ATIME)
+				nfs_update_atime(inode);
 			attr->ia_valid &= ~(ATTR_MTIME|ATTR_ATIME);
 		}
 		spin_unlock(&inode->i_lock);
-- 
2.47.3

[PATCH 04/11] fat: cleanup the flags for fat_truncate_time

[Christoph Hellwig]

Fat only has a single on-disk timestamp covering ctime and mtime.  Add
fat-specific flags that indicate which timestamp fat_truncate_time should
update to make this more clear.  This allows removing no-op
fat_truncate_time calls with the S_CTIME flag and prepares for removing
the S_* flags.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 fs/fat/dir.c         |  2 +-
 fs/fat/fat.h         |  8 ++++----
 fs/fat/file.c        | 14 ++++++--------
 fs/fat/inode.c       |  2 +-
 fs/fat/misc.c        | 39 ++++++++++++++++-----------------------
 fs/fat/namei_msdos.c | 13 +++++--------
 fs/fat/namei_vfat.c  |  9 ++++-----
 7 files changed, 37 insertions(+), 50 deletions(-)

diff --git a/fs/fat/dir.c b/fs/fat/dir.c
index 92b091783966..3d03bff40944 100644
--- a/fs/fat/dir.c
+++ b/fs/fat/dir.c
@@ -1080,7 +1080,7 @@ int fat_remove_entries(struct inode *dir, struct fat_slot_info *sinfo)
 		}
 	}
 
-	fat_truncate_time(dir, NULL, S_ATIME|S_MTIME);
+	fat_truncate_time(dir, NULL, FAT_UPDATE_ATIME | FAT_UPDATE_CMTIME);
 	if (IS_DIRSYNC(dir))
 		(void)fat_sync_inode(dir);
 	else
diff --git a/fs/fat/fat.h b/fs/fat/fat.h
index d3e426de5f01..767b566b1cab 100644
--- a/fs/fat/fat.h
+++ b/fs/fat/fat.h
@@ -468,10 +468,10 @@ extern void fat_time_unix2fat(struct msdos_sb_info *sbi, struct timespec64 *ts,
 			      __le16 *time, __le16 *date, u8 *time_cs);
 extern struct timespec64 fat_truncate_atime(const struct msdos_sb_info *sbi,
 					    const struct timespec64 *ts);
-extern struct timespec64 fat_truncate_mtime(const struct msdos_sb_info *sbi,
-					    const struct timespec64 *ts);
-extern int fat_truncate_time(struct inode *inode, struct timespec64 *now,
-			     int flags);
+#define FAT_UPDATE_ATIME	(1u << 0)
+#define FAT_UPDATE_CMTIME	(1u << 1)
+void fat_truncate_time(struct inode *inode, struct timespec64 *now,
+		unsigned int flags);
 extern int fat_update_time(struct inode *inode, int flags);
 extern int fat_sync_bhs(struct buffer_head **bhs, int nr_bhs);
 
diff --git a/fs/fat/file.c b/fs/fat/file.c
index 4fc49a614fb8..f9bc93411aa2 100644
--- a/fs/fat/file.c
+++ b/fs/fat/file.c
@@ -224,7 +224,7 @@ static int fat_cont_expand(struct inode *inode, loff_t size)
 	if (err)
 		goto out;
 
-	fat_truncate_time(inode, NULL, S_CTIME|S_MTIME);
+	fat_truncate_time(inode, NULL, FAT_UPDATE_CMTIME);
 	mark_inode_dirty(inode);
 	if (IS_SYNC(inode)) {
 		int err2;
@@ -327,7 +327,7 @@ static int fat_free(struct inode *inode, int skip)
 		MSDOS_I(inode)->i_logstart = 0;
 	}
 	MSDOS_I(inode)->i_attrs |= ATTR_ARCH;
-	fat_truncate_time(inode, NULL, S_CTIME|S_MTIME);
+	fat_truncate_time(inode, NULL, FAT_UPDATE_CMTIME);
 	if (wait) {
 		err = fat_sync_inode(inode);
 		if (err) {
@@ -553,15 +553,13 @@ int fat_setattr(struct mnt_idmap *idmap, struct dentry *dentry,
 	}
 
 	/*
-	 * setattr_copy can't truncate these appropriately, so we'll
-	 * copy them ourselves
+	 * setattr_copy can't truncate these appropriately, so we'll copy them
+	 * ourselves.  See fat_truncate_time for the c/mtime logic on fat.
 	 */
 	if (attr->ia_valid & ATTR_ATIME)
-		fat_truncate_time(inode, &attr->ia_atime, S_ATIME);
-	if (attr->ia_valid & ATTR_CTIME)
-		fat_truncate_time(inode, &attr->ia_ctime, S_CTIME);
+		fat_truncate_time(inode, &attr->ia_atime, FAT_UPDATE_ATIME);
 	if (attr->ia_valid & ATTR_MTIME)
-		fat_truncate_time(inode, &attr->ia_mtime, S_MTIME);
+		fat_truncate_time(inode, &attr->ia_mtime, FAT_UPDATE_CMTIME);
 	attr->ia_valid &= ~(ATTR_ATIME|ATTR_CTIME|ATTR_MTIME);
 
 	setattr_copy(idmap, inode, attr);
diff --git a/fs/fat/inode.c b/fs/fat/inode.c
index 0b6009cd1844..59fa90617b5b 100644
--- a/fs/fat/inode.c
+++ b/fs/fat/inode.c
@@ -246,7 +246,7 @@ static int fat_write_end(const struct kiocb *iocb,
 	if (err < len)
 		fat_write_failed(mapping, pos + len);
 	if (!(err < 0) && !(MSDOS_I(inode)->i_attrs & ATTR_ARCH)) {
-		fat_truncate_time(inode, NULL, S_CTIME|S_MTIME);
+		fat_truncate_time(inode, NULL, FAT_UPDATE_CMTIME);
 		MSDOS_I(inode)->i_attrs |= ATTR_ARCH;
 		mark_inode_dirty(inode);
 	}
diff --git a/fs/fat/misc.c b/fs/fat/misc.c
index 950da09f0961..f4a1fa58bf05 100644
--- a/fs/fat/misc.c
+++ b/fs/fat/misc.c
@@ -299,43 +299,36 @@ struct timespec64 fat_truncate_atime(const struct msdos_sb_info *sbi,
 }
 
 /*
- * truncate mtime to 2 second granularity
- */
-struct timespec64 fat_truncate_mtime(const struct msdos_sb_info *sbi,
-				     const struct timespec64 *ts)
-{
-	return fat_timespec64_trunc_2secs(*ts);
-}
-
-/*
- * truncate the various times with appropriate granularity:
- *   all times in root node are always 0
+ * Update the in-inode atime and/or mtime after truncating the timestamp to the
+ * granularity.  All timestamps in root inode are always 0.
+ *
+ * ctime and mtime share the same on-disk field, and should be identical in
+ * memory.  All mtime updates will be applied to ctime, but ctime updates are
+ * ignored.
  */
-int fat_truncate_time(struct inode *inode, struct timespec64 *now, int flags)
+void fat_truncate_time(struct inode *inode, struct timespec64 *now,
+		unsigned int flags)
 {
 	struct msdos_sb_info *sbi = MSDOS_SB(inode->i_sb);
 	struct timespec64 ts;
 
 	if (inode->i_ino == MSDOS_ROOT_INO)
-		return 0;
+		return;
 
 	if (now == NULL) {
 		now = &ts;
 		ts = current_time(inode);
 	}
 
-	if (flags & S_ATIME)
+	if (flags & FAT_UPDATE_ATIME)
 		inode_set_atime_to_ts(inode, fat_truncate_atime(sbi, now));
-	/*
-	 * ctime and mtime share the same on-disk field, and should be
-	 * identical in memory. all mtime updates will be applied to ctime,
-	 * but ctime updates are ignored.
-	 */
-	if (flags & S_MTIME)
-		inode_set_mtime_to_ts(inode,
-				      inode_set_ctime_to_ts(inode, fat_truncate_mtime(sbi, now)));
+	if (flags & FAT_UPDATE_CMTIME) {
+		/* truncate mtime to 2 second granularity */
+		struct timespec64 mtime = fat_timespec64_trunc_2secs(*now);
 
-	return 0;
+		inode_set_mtime_to_ts(inode, mtime);
+		inode_set_ctime_to_ts(inode, mtime);
+	}
 }
 EXPORT_SYMBOL_GPL(fat_truncate_time);
 
diff --git a/fs/fat/namei_msdos.c b/fs/fat/namei_msdos.c
index 0b920ee40a7f..ba0152ed0810 100644
--- a/fs/fat/namei_msdos.c
+++ b/fs/fat/namei_msdos.c
@@ -251,7 +251,7 @@ static int msdos_add_entry(struct inode *dir, const unsigned char *name,
 	if (err)
 		return err;
 
-	fat_truncate_time(dir, ts, S_CTIME|S_MTIME);
+	fat_truncate_time(dir, ts, FAT_UPDATE_CMTIME);
 	if (IS_DIRSYNC(dir))
 		(void)fat_sync_inode(dir);
 	else
@@ -295,7 +295,7 @@ static int msdos_create(struct mnt_idmap *idmap, struct inode *dir,
 		err = PTR_ERR(inode);
 		goto out;
 	}
-	fat_truncate_time(inode, &ts, S_ATIME|S_CTIME|S_MTIME);
+	fat_truncate_time(inode, &ts, FAT_UPDATE_ATIME | FAT_UPDATE_CMTIME);
 	/* timestamp is already written, so mark_inode_dirty() is unneeded. */
 
 	d_instantiate(dentry, inode);
@@ -328,7 +328,6 @@ static int msdos_rmdir(struct inode *dir, struct dentry *dentry)
 	drop_nlink(dir);
 
 	clear_nlink(inode);
-	fat_truncate_time(inode, NULL, S_CTIME);
 	fat_detach(inode);
 out:
 	mutex_unlock(&MSDOS_SB(sb)->s_lock);
@@ -382,7 +381,7 @@ static struct dentry *msdos_mkdir(struct mnt_idmap *idmap, struct inode *dir,
 		goto out;
 	}
 	set_nlink(inode, 2);
-	fat_truncate_time(inode, &ts, S_ATIME|S_CTIME|S_MTIME);
+	fat_truncate_time(inode, &ts, FAT_UPDATE_ATIME | FAT_UPDATE_CMTIME);
 	/* timestamp is already written, so mark_inode_dirty() is unneeded. */
 
 	d_instantiate(dentry, inode);
@@ -415,7 +414,6 @@ static int msdos_unlink(struct inode *dir, struct dentry *dentry)
 	if (err)
 		goto out;
 	clear_nlink(inode);
-	fat_truncate_time(inode, NULL, S_CTIME);
 	fat_detach(inode);
 out:
 	mutex_unlock(&MSDOS_SB(sb)->s_lock);
@@ -480,7 +478,7 @@ static int do_msdos_rename(struct inode *old_dir, unsigned char *old_name,
 				mark_inode_dirty(old_inode);
 
 			inode_inc_iversion(old_dir);
-			fat_truncate_time(old_dir, NULL, S_CTIME|S_MTIME);
+			fat_truncate_time(old_dir, NULL, FAT_UPDATE_CMTIME);
 			if (IS_DIRSYNC(old_dir))
 				(void)fat_sync_inode(old_dir);
 			else
@@ -540,7 +538,7 @@ static int do_msdos_rename(struct inode *old_dir, unsigned char *old_name,
 	if (err)
 		goto error_dotdot;
 	inode_inc_iversion(old_dir);
-	fat_truncate_time(old_dir, &ts, S_CTIME|S_MTIME);
+	fat_truncate_time(old_dir, &ts, FAT_UPDATE_CMTIME);
 	if (IS_DIRSYNC(old_dir))
 		(void)fat_sync_inode(old_dir);
 	else
@@ -550,7 +548,6 @@ static int do_msdos_rename(struct inode *old_dir, unsigned char *old_name,
 		drop_nlink(new_inode);
 		if (is_dir)
 			drop_nlink(new_inode);
-		fat_truncate_time(new_inode, &ts, S_CTIME);
 	}
 out:
 	brelse(sinfo.bh);
diff --git a/fs/fat/namei_vfat.c b/fs/fat/namei_vfat.c
index 5dbc4cbb8fce..e46f34cade1a 100644
--- a/fs/fat/namei_vfat.c
+++ b/fs/fat/namei_vfat.c
@@ -676,7 +676,7 @@ static int vfat_add_entry(struct inode *dir, const struct qstr *qname,
 		goto cleanup;
 
 	/* update timestamp */
-	fat_truncate_time(dir, ts, S_CTIME|S_MTIME);
+	fat_truncate_time(dir, ts, FAT_UPDATE_CMTIME);
 	if (IS_DIRSYNC(dir))
 		(void)fat_sync_inode(dir);
 	else
@@ -806,7 +806,7 @@ static int vfat_rmdir(struct inode *dir, struct dentry *dentry)
 	drop_nlink(dir);
 
 	clear_nlink(inode);
-	fat_truncate_time(inode, NULL, S_ATIME|S_MTIME);
+	fat_truncate_time(inode, NULL, FAT_UPDATE_ATIME | FAT_UPDATE_CMTIME);
 	fat_detach(inode);
 	vfat_d_version_set(dentry, inode_query_iversion(dir));
 out:
@@ -832,7 +832,7 @@ static int vfat_unlink(struct inode *dir, struct dentry *dentry)
 	if (err)
 		goto out;
 	clear_nlink(inode);
-	fat_truncate_time(inode, NULL, S_ATIME|S_MTIME);
+	fat_truncate_time(inode, NULL, FAT_UPDATE_ATIME | FAT_UPDATE_CMTIME);
 	fat_detach(inode);
 	vfat_d_version_set(dentry, inode_query_iversion(dir));
 out:
@@ -918,7 +918,7 @@ static int vfat_update_dotdot_de(struct inode *dir, struct inode *inode,
 static void vfat_update_dir_metadata(struct inode *dir, struct timespec64 *ts)
 {
 	inode_inc_iversion(dir);
-	fat_truncate_time(dir, ts, S_CTIME | S_MTIME);
+	fat_truncate_time(dir, ts, FAT_UPDATE_CMTIME);
 	if (IS_DIRSYNC(dir))
 		(void)fat_sync_inode(dir);
 	else
@@ -996,7 +996,6 @@ static int vfat_rename(struct inode *old_dir, struct dentry *old_dentry,
 		drop_nlink(new_inode);
 		if (is_dir)
 			drop_nlink(new_inode);
-		fat_truncate_time(new_inode, &ts, S_CTIME);
 	}
 out:
 	brelse(sinfo.bh);
-- 
2.47.3

Re: re-enable IOCB_NOWAIT writes to files v5

[Christoph Hellwig]

Sorry, sending this got interrupted by the broken train wifi.
Please look at the immediately following full series instead.

re-enable IOCB_NOWAIT writes to files v6

[Christoph Hellwig]

Hi all,

commit 66fa3cedf16a ("fs: Add async write file modification handling.")
effectively disabled IOCB_NOWAIT writes as timestamp updates currently
always require blocking, and the modern timestamp resolution means we
always update timestamps.  This leads to a lot of context switches from
applications using io_uring to submit file writes, making it often worse
than using the legacy aio code that is not using IOCB_NOWAIT.

This series allows non-blocking updates for lazytime if the file system
supports it, and adds that support for XFS.

Changes since v5:
 - sample ctime before calling inode_set_ctime_current
 - fix a mild bisection hazard in fat

Changes since v4:
 - replace the S_* flags with an enum indicating either access or
   modification time updates to make the logic less fragile and to
   fix a bug in the previous version

Changes since v3:
 - fix was_dirty_time handling in __mark_inode_dirty for the racy flag
   update case
 - refactor inode_update_timestamps to make the lazytime vs blocking
   logical more clear
 - allow non-blocking timestamp updates for fat

Changes since v2:
 - drop patches merged upstream
 - adjust for the inode state accesors
 - keep a check in __writeback_single_inode instead of exercising
   potentially undefined behavior
 - more spelling fixes

Changes since v1:
 - more regular numbering of the S_* flags
 - fix XFS to actually not block
 - don't ignore the generic_update_time return value in
   file_update_time_flags
 - fix the sync_lazytime return value
 - fix an out of data comment in btrfs
 - fix a race that would update i_version before returning -EAGAIN in XFS

Diffstat:
 Documentation/filesystems/locking.rst |    2 
 Documentation/filesystems/vfs.rst     |    6 +
 fs/btrfs/inode.c                      |    8 +-
 fs/fs-writeback.c                     |   33 +++++++---
 fs/gfs2/inode.c                       |    6 +
 fs/inode.c                            |  111 +++++++++++++++++++++-------------
 fs/internal.h                         |    3 
 fs/nfs/inode.c                        |    4 -
 fs/orangefs/inode.c                   |    5 +
 fs/overlayfs/inode.c                  |    2 
 fs/sync.c                             |    4 -
 fs/ubifs/file.c                       |   13 ++-
 fs/xfs/xfs_iops.c                     |   34 +++++++++-
 fs/xfs/xfs_super.c                    |   29 --------
 include/linux/fs.h                    |   27 ++++++--
 include/trace/events/writeback.h      |    6 -
 16 files changed, 182 insertions(+), 111 deletions(-)

[PATCH 01/11] fs: remove inode_update_time

[Christoph Hellwig]

The only external user is gone now, open code it in the two VFS
callers.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
---
 fs/inode.c         | 23 ++++++++---------------
 include/linux/fs.h |  1 -
 2 files changed, 8 insertions(+), 16 deletions(-)

diff --git a/fs/inode.c b/fs/inode.c
index 521383223d8a..07effa0cb999 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -2157,19 +2157,6 @@ int generic_update_time(struct inode *inode, int flags)
 }
 EXPORT_SYMBOL(generic_update_time);
 
-/*
- * This does the actual work of updating an inodes time or version.  Must have
- * had called mnt_want_write() before calling this.
- */
-int inode_update_time(struct inode *inode, int flags)
-{
-	if (inode->i_op->update_time)
-		return inode->i_op->update_time(inode, flags);
-	generic_update_time(inode, flags);
-	return 0;
-}
-EXPORT_SYMBOL(inode_update_time);
-
 /**
  *	atime_needs_update	-	update the access time
  *	@path: the &struct path to update
@@ -2237,7 +2224,10 @@ void touch_atime(const struct path *path)
 	 * We may also fail on filesystems that have the ability to make parts
 	 * of the fs read only, e.g. subvolumes in Btrfs.
 	 */
-	inode_update_time(inode, S_ATIME);
+	if (inode->i_op->update_time)
+		inode->i_op->update_time(inode, S_ATIME);
+	else
+		generic_update_time(inode, S_ATIME);
 	mnt_put_write_access(mnt);
 skip_update:
 	sb_end_write(inode->i_sb);
@@ -2392,7 +2382,10 @@ static int file_update_time_flags(struct file *file, unsigned int flags)
 
 	if (mnt_get_write_access_file(file))
 		return 0;
-	ret = inode_update_time(inode, sync_mode);
+	if (inode->i_op->update_time)
+		ret = inode->i_op->update_time(inode, sync_mode);
+	else
+		ret = generic_update_time(inode, sync_mode);
 	mnt_put_write_access_file(file);
 	return ret;
 }
diff --git a/include/linux/fs.h b/include/linux/fs.h
index f5c9cf28c4dc..ee623c16d835 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -2246,7 +2246,6 @@ enum file_time_flags {
 
 extern bool atime_needs_update(const struct path *, struct inode *);
 extern void touch_atime(const struct path *);
-int inode_update_time(struct inode *inode, int flags);
 
 static inline void file_accessed(struct file *file)
 {
-- 
2.47.3

re-enable IOCB_NOWAIT writes to files v5

[Christoph Hellwig]

Hi all,

commit 66fa3cedf16a ("fs: Add async write file modification handling.")
effectively disabled IOCB_NOWAIT writes as timestamp updates currently
always require blocking, and the modern timestamp resolution means we
always update timestamps.  This leads to a lot of context switches from
applications using io_uring to submit file writes, making it often worse
than using the legacy aio code that is not using IOCB_NOWAIT.

This series allows non-blocking updates for lazytime if the file system
supports it, and adds that support for XFS.

Changes since v4:
 - replace the S_* flags with an enum indicating either access or
   modification time updates to make the logic less fragile and to
   fix a bug in the previous version

Changes since v3:
 - fix was_dirty_time handling in __mark_inode_dirty for the racy flag
   update case
 - refactor inode_update_timestamps to make the lazytime vs blocking
   logical more clear
 - allow non-blocking timestamp updates for fat

Changes since v2:
 - drop patches merged upstream
 - adjust for the inode state accesors
 - keep a check in __writeback_single_inode instead of exercising
   potentially undefined behavior
 - more spelling fixes

Changes since v1:
 - more regular numbering of the S_* flags
 - fix XFS to actually not block
 - don't ignore the generic_update_time return value in
   file_update_time_flags
 - fix the sync_lazytime return value
 - fix an out of data comment in btrfs
 - fix a race that would update i_version before returning -EAGAIN in XFS

Diffstat:
 Documentation/filesystems/locking.rst |    2 
 Documentation/filesystems/vfs.rst     |    6 +
 fs/btrfs/inode.c                      |    8 +-
 fs/fs-writeback.c                     |   33 +++++++---
 fs/gfs2/inode.c                       |    6 +
 fs/inode.c                            |  111 +++++++++++++++++++++-------------
 fs/internal.h                         |    3 
 fs/nfs/inode.c                        |    4 -
 fs/orangefs/inode.c                   |    5 +
 fs/overlayfs/inode.c                  |    2 
 fs/sync.c                             |    4 -
 fs/ubifs/file.c                       |   13 ++-
 fs/xfs/xfs_iops.c                     |   34 +++++++++-
 fs/xfs/xfs_super.c                    |   29 --------
 include/linux/fs.h                    |   27 ++++++--
 include/trace/events/writeback.h      |    6 -
 16 files changed, 182 insertions(+), 111 deletions(-)

[PATCH 01/11] fs: remove inode_update_time

[Christoph Hellwig]

The only external user is gone now, open code it in the two VFS
callers.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
---
 fs/inode.c         | 23 ++++++++---------------
 include/linux/fs.h |  1 -
 2 files changed, 8 insertions(+), 16 deletions(-)

diff --git a/fs/inode.c b/fs/inode.c
index 521383223d8a..07effa0cb999 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -2157,19 +2157,6 @@ int generic_update_time(struct inode *inode, int flags)
 }
 EXPORT_SYMBOL(generic_update_time);
 
-/*
- * This does the actual work of updating an inodes time or version.  Must have
- * had called mnt_want_write() before calling this.
- */
-int inode_update_time(struct inode *inode, int flags)
-{
-	if (inode->i_op->update_time)
-		return inode->i_op->update_time(inode, flags);
-	generic_update_time(inode, flags);
-	return 0;
-}
-EXPORT_SYMBOL(inode_update_time);
-
 /**
  *	atime_needs_update	-	update the access time
  *	@path: the &struct path to update
@@ -2237,7 +2224,10 @@ void touch_atime(const struct path *path)
 	 * We may also fail on filesystems that have the ability to make parts
 	 * of the fs read only, e.g. subvolumes in Btrfs.
 	 */
-	inode_update_time(inode, S_ATIME);
+	if (inode->i_op->update_time)
+		inode->i_op->update_time(inode, S_ATIME);
+	else
+		generic_update_time(inode, S_ATIME);
 	mnt_put_write_access(mnt);
 skip_update:
 	sb_end_write(inode->i_sb);
@@ -2392,7 +2382,10 @@ static int file_update_time_flags(struct file *file, unsigned int flags)
 
 	if (mnt_get_write_access_file(file))
 		return 0;
-	ret = inode_update_time(inode, sync_mode);
+	if (inode->i_op->update_time)
+		ret = inode->i_op->update_time(inode, sync_mode);
+	else
+		ret = generic_update_time(inode, sync_mode);
 	mnt_put_write_access_file(file);
 	return ret;
 }
diff --git a/include/linux/fs.h b/include/linux/fs.h
index f5c9cf28c4dc..ee623c16d835 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -2246,7 +2246,6 @@ enum file_time_flags {
 
 extern bool atime_needs_update(const struct path *, struct inode *);
 extern void touch_atime(const struct path *);
-int inode_update_time(struct inode *inode, int flags);
 
 static inline void file_accessed(struct file *file)
 {
-- 
2.47.3

re-enable IOCB_NOWAIT writes to files v4

[Christoph Hellwig]

Hi all,

commit 66fa3cedf16a ("fs: Add async write file modification handling.")
effectively disabled IOCB_NOWAIT writes as timestamp updates currently
always require blocking, and the modern timestamp resolution means we
always update timestamps.  This leads to a lot of context switches from
applications using io_uring to submit file writes, making it often worse
than using the legacy aio code that is not using IOCB_NOWAIT.

This series allows non-blocking updates for lazytime if the file system
supports it, and adds that support for XFS.

Changes since v3:
 - fix was_dirty_time handling in __mark_inode_dirty for the racy flag
   update case
 - refactor inode_update_timestamps to make the lazytime vs blocking
   logical more clear
 - allow non-blocking timestamp updates for fat

Changes since v2:
 - drop patches merged upstream
 - adjust for the inode state accesors
 - keep a check in __writeback_single_inode instead of exercising
   potentially undefined behavior
 - more spelling fixes

Changes since v1:
 - more regular numbering of the S_* flags
 - fix XFS to actually not block
 - don't ignore the generic_update_time return value in
   file_update_time_flags
 - fix the sync_lazytime return value
 - fix an out of data comment in btrfs
 - fix a race that would update i_version before returning -EAGAIN in XFS

Diffstat:
 Documentation/filesystems/locking.rst |    2 
 Documentation/filesystems/vfs.rst     |    6 +
 fs/btrfs/inode.c                      |    8 +-
 fs/fs-writeback.c                     |   33 +++++++---
 fs/gfs2/inode.c                       |    6 +
 fs/inode.c                            |  111 +++++++++++++++++++++-------------
 fs/internal.h                         |    3 
 fs/nfs/inode.c                        |    4 -
 fs/orangefs/inode.c                   |    5 +
 fs/overlayfs/inode.c                  |    2 
 fs/sync.c                             |    4 -
 fs/ubifs/file.c                       |   13 ++-
 fs/xfs/xfs_iops.c                     |   34 +++++++++-
 fs/xfs/xfs_super.c                    |   29 --------
 include/linux/fs.h                    |   27 ++++++--
 include/trace/events/writeback.h      |    6 -
 16 files changed, 182 insertions(+), 111 deletions(-)

[PATCH 01/11] fs: remove inode_update_time

[Christoph Hellwig]

The only external user is gone now, open code it in the two VFS
callers.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
---
 fs/inode.c         | 23 ++++++++---------------
 include/linux/fs.h |  1 -
 2 files changed, 8 insertions(+), 16 deletions(-)

diff --git a/fs/inode.c b/fs/inode.c
index 521383223d8a..07effa0cb999 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -2157,19 +2157,6 @@ int generic_update_time(struct inode *inode, int flags)
 }
 EXPORT_SYMBOL(generic_update_time);
 
-/*
- * This does the actual work of updating an inodes time or version.  Must have
- * had called mnt_want_write() before calling this.
- */
-int inode_update_time(struct inode *inode, int flags)
-{
-	if (inode->i_op->update_time)
-		return inode->i_op->update_time(inode, flags);
-	generic_update_time(inode, flags);
-	return 0;
-}
-EXPORT_SYMBOL(inode_update_time);
-
 /**
  *	atime_needs_update	-	update the access time
  *	@path: the &struct path to update
@@ -2237,7 +2224,10 @@ void touch_atime(const struct path *path)
 	 * We may also fail on filesystems that have the ability to make parts
 	 * of the fs read only, e.g. subvolumes in Btrfs.
 	 */
-	inode_update_time(inode, S_ATIME);
+	if (inode->i_op->update_time)
+		inode->i_op->update_time(inode, S_ATIME);
+	else
+		generic_update_time(inode, S_ATIME);
 	mnt_put_write_access(mnt);
 skip_update:
 	sb_end_write(inode->i_sb);
@@ -2392,7 +2382,10 @@ static int file_update_time_flags(struct file *file, unsigned int flags)
 
 	if (mnt_get_write_access_file(file))
 		return 0;
-	ret = inode_update_time(inode, sync_mode);
+	if (inode->i_op->update_time)
+		ret = inode->i_op->update_time(inode, sync_mode);
+	else
+		ret = generic_update_time(inode, sync_mode);
 	mnt_put_write_access_file(file);
 	return ret;
 }
diff --git a/include/linux/fs.h b/include/linux/fs.h
index f5c9cf28c4dc..ee623c16d835 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -2246,7 +2246,6 @@ enum file_time_flags {
 
 extern bool atime_needs_update(const struct path *, struct inode *);
 extern void touch_atime(const struct path *);
-int inode_update_time(struct inode *inode, int flags);
 
 static inline void file_accessed(struct file *file)
 {
-- 
2.47.3