[PATCH V9 00/11] XFS - Enable per-file/per-directory DAX operations V9

[PATCH V9 00/11] XFS - Enable per-file/per-directory DAX operations V9

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

Changes from V8:
	Rebase to 5.7-rc2
	Change ALWAYS/NEVER bits to be 26/27
	Remove XFS_IDONTCACHE -> lift to I_DONTCACHE
		use mark_inode_dontcache() in XFS
	create xfs_dax_mode enum
	use xfs signature styles
	Change xfs_inode_enabe_dax() -> xfs_inode_should_enable()
		Based on feedback to ext4 series
	Fix locking of DCACHE_DONTCACHE
	Change flag_inode_dontcache() -> mark_inode_dontcache()
	Change xfs_ioctl_setattr_dax_invalidate() -> xfs_ioctl_dax_check_set_cache()
	Documentation cleanups
	Clean up all commit messages


At LSF/MM'19 [1] [2] we discussed applications that overestimate memory
consumption due to their inability to detect whether the kernel will
instantiate page cache for a file, and cases where a global dax enable via a
mount option is too coarse.

The following patch series enables the use of DAX on individual files and/or
directories on xfs, and lays some groundwork to do so in ext4.  It further
enhances the dax mount option to be a tri-state of 'always', 'never', or
'iflag' (default).  Furthermore, it maintians '-o dax' to be equivalent to '-o
dax=always'.

The insight at LSF/MM was to separate the per-mount or per-file "physical"
(FS_XFLAG_DAX) capability switch from an "effective" (S_DAX) attribute for the
file.

At LSF/MM we discussed the difficulties of switching the DAX state of a file
with active mappings / page cache.  It was thought the races could be avoided
by limiting DAX state flips to 0-length files.

However, this turns out to not be true.[3][5] This is because address space
operations (a_ops) may be in use at any time the inode is referenced.

For this reason direct manipulation of the FS_XFLAG_DAX is allowed but the
operation of the file (S_DAX) is not immediately changed.

Details of when and how DAX state can be changed on a file is included in a
documentation patch.

It should be noted that FS_XFLAG_DAX inheritance is not shown in this patch set
as it was maintained from previous work on XFS.  FS_XFLAG_DAX and it's
inheritance will need to be added to other file systems for user control. 


[1] https://lwn.net/Articles/787973/
[2] https://lwn.net/Articles/787233/
[3] https://lkml.org/lkml/2019/10/20/96
[4] https://patchwork.kernel.org/patch/11310511/
[5] https://lore.kernel.org/lkml/20200405061945.GA94792@iweiny-DESK2.sc.intel.com/


To: linux-kernel@vger.kernel.org
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Darrick J. Wong" <darrick.wong@oracle.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: "Theodore Y. Ts'o" <tytso@mit.edu>
Cc: Jan Kara <jack@suse.cz>
Cc: linux-ext4@vger.kernel.org
Cc: linux-xfs@vger.kernel.org
Cc: linux-fsdevel@vger.kernel.org

Changes from V7:
	Add DCACHE_DONTCACHE
	If mount override don't worry about inode caching
	Change mount flags to NEVER/ALWAYS
	Clean up xfs_inode_enable_dax()
	Clarify comments
	Update documentation

Changes from V6:
	Incorporate feedback on patches
	Add ability to change FS_XFLAG_DAX on files at any time.
		Add a don't cache flag to the VFS layer
		Preserve internal XFS IDONTCACHE behavior for bulkstat
		operations

Changes from V5:
	* make dax mount option a tri-state
	* Reject changes to FS_XFLAG_DAX for regular files
		- Allow only on directories
	* Update documentation

Changes from V4:
	* Open code the aops lock rather than add it to the xfs_ilock()
	  subsystem (Darrick's comments were obsoleted by this change)
	* Fix lkp build suggestions and bugs

Changes from V3:
	* Remove global locking...  :-D
	* put back per inode locking and remove pre-mature optimizations
	* Fix issues with Directories having IS_DAX() set
	* Fix kernel crash issues reported by Jeff
	* Add some clean up patches
	* Consolidate diflags to iflags functions
	* Update/add documentation
	* Reorder/rename patches quite a bit

Changes from V2:

	* Move i_dax_sem to be a global percpu_rw_sem rather than per inode
		Internal discussions with Dan determined this would be easier,
		just as performant, and slightly less overhead that having it
		in the SB as suggested by Jan
	* Fix locking order in comments and throughout code
	* Change "mode" to "state" throughout commits
	* Add CONFIG_FS_DAX wrapper to disable inode_[un]lock_state() when not
		configured
	* Add static branch for which is activated by a device which supports
		DAX in XFS
	* Change "lock/unlock" to up/down read/write as appropriate
		Previous names were over simplified
	* Update comments/documentation

	* Remove the xfs specific lock to the vfs (global) layer.
	* Fix i_dax_sem locking order and comments

	* Move 'i_mapped' count from struct inode to struct address_space and
		rename it to mmap_count
	* Add inode_has_mappings() call

	* Fix build issues
	* Clean up syntax spacing and minor issues
	* Update man page text for STATX_ATTR_DAX
	* Add reviewed-by's
	* Rebase to 5.6

	Rename patch:
		from: fs/xfs: Add lock/unlock state to xfs
		to: fs/xfs: Add write DAX lock to xfs layer
	Add patch:
		fs/xfs: Clarify lockdep dependency for xfs_isilocked()
	Drop patch:
		fs/xfs: Fix truncate up


Ira Weiny (11):
  fs/xfs: Remove unnecessary initialization of i_rwsem
  fs: Remove unneeded IS_DAX() check in io_is_direct()
  fs/stat: Define DAX statx attribute
  fs/xfs: Change XFS_MOUNT_DAX to XFS_MOUNT_DAX_ALWAYS
  fs/xfs: Make DAX mount option a tri-state
  fs/xfs: Create function xfs_inode_should_enable_dax()
  fs/xfs: Combine xfs_diflags_to_linux() and xfs_diflags_to_iflags()
  fs: Lift XFS_IDONTCACNE to the VFS layer
  fs: Introduce DCACHE_DONTCACHE
  fs/xfs: Update xfs_ioctl_setattr_dax_invalidate()
  Documentation/dax: Update Usage section

 Documentation/filesystems/dax.txt | 164 +++++++++++++++++++++++++++++-
 drivers/block/loop.c              |   6 +-
 fs/dcache.c                       |   4 +
 fs/inode.c                        |  15 +++
 fs/stat.c                         |   3 +
 fs/xfs/xfs_icache.c               |   8 +-
 fs/xfs/xfs_inode.h                |   3 +-
 fs/xfs/xfs_ioctl.c                | 141 ++++---------------------
 fs/xfs/xfs_iops.c                 |  72 ++++++++-----
 fs/xfs/xfs_mount.h                |   4 +-
 fs/xfs/xfs_super.c                |  55 ++++++++--
 include/linux/dcache.h            |   2 +
 include/linux/fs.h                |  14 +--
 include/uapi/linux/stat.h         |   1 +
 14 files changed, 321 insertions(+), 171 deletions(-)

-- 
2.25.1

[PATCH V9 01/11] fs/xfs: Remove unnecessary initialization of i_rwsem

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

An earlier call of xfs_reinit_inode() from xfs_iget_cache_hit() already
handles initialization of i_rwsem.

Doing so again is unneeded.

Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>

---
Changes from V4:
	Update commit message to make it clear the xfs_iget_cache_hit()
	is actually doing the initialization via xfs_reinit_inode()

New for V4:

NOTE: This was found while ensuring the new i_aops_sem was properly
handled.  It seems like this is a layering violation so I think it is
worth cleaning up so as to not confuse others.
---
 fs/xfs/xfs_icache.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
index 8bf1d15be3f6..17a0b86fe701 100644
--- a/fs/xfs/xfs_icache.c
+++ b/fs/xfs/xfs_icache.c
@@ -423,6 +423,7 @@ xfs_iget_cache_hit(
 		spin_unlock(&ip->i_flags_lock);
 		rcu_read_unlock();
 
+		ASSERT(!rwsem_is_locked(&inode->i_rwsem));
 		error = xfs_reinit_inode(mp, inode);
 		if (error) {
 			bool wake;
@@ -456,9 +457,6 @@ xfs_iget_cache_hit(
 		ip->i_sick = 0;
 		ip->i_checked = 0;
 
-		ASSERT(!rwsem_is_locked(&inode->i_rwsem));
-		init_rwsem(&inode->i_rwsem);
-
 		spin_unlock(&ip->i_flags_lock);
 		spin_unlock(&pag->pag_ici_lock);
 	} else {
-- 
2.25.1

[PATCH V9 02/11] fs: Remove unneeded IS_DAX() check in io_is_direct()

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

Remove the check because DAX now has it's own read/write methods and
file systems which support DAX check IS_DAX() prior to IOCB_DIRECT on
their own.  Therefore, it does not matter if the file state is DAX when
the iocb flags are created.

Also remove io_is_direct() as it is just a simple flag check.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>

---
Changes from v8:
	Rebase to latest Linus tree

Changes from v6:
	remove io_is_direct() as well.
	Remove Reviews since this is quite a bit different.

Changes from v3:
	Reword commit message.
	Reordered to be a 'pre-cleanup' patch
---
 drivers/block/loop.c | 6 +++---
 include/linux/fs.h   | 7 +------
 2 files changed, 4 insertions(+), 9 deletions(-)

diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index da693e6a834e..14372df0f354 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -634,8 +634,8 @@ static int do_req_filebacked(struct loop_device *lo, struct request *rq)
 
 static inline void loop_update_dio(struct loop_device *lo)
 {
-	__loop_update_dio(lo, io_is_direct(lo->lo_backing_file) |
-			lo->use_dio);
+	__loop_update_dio(lo, (lo->lo_backing_file->f_flags & O_DIRECT) |
+				lo->use_dio);
 }
 
 static void loop_reread_partitions(struct loop_device *lo,
@@ -1028,7 +1028,7 @@ 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_write_cache(lo->lo_queue, true, false);
 
-	if (io_is_direct(lo->lo_backing_file) && inode->i_sb->s_bdev) {
+	if ((lo->lo_backing_file->f_flags & O_DIRECT) && inode->i_sb->s_bdev) {
 		/* In case of direct I/O, match underlying block size */
 		unsigned short bsize = bdev_logical_block_size(
 			inode->i_sb->s_bdev);
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 4f6f59b4f22a..a87cc5845a02 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -3394,11 +3394,6 @@ extern void setattr_copy(struct inode *inode, const struct iattr *attr);
 
 extern int file_update_time(struct file *file);
 
-static inline bool io_is_direct(struct file *filp)
-{
-	return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
-}
-
 static inline bool vma_is_dax(const struct vm_area_struct *vma)
 {
 	return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
@@ -3423,7 +3418,7 @@ static inline int iocb_flags(struct file *file)
 	int res = 0;
 	if (file->f_flags & O_APPEND)
 		res |= IOCB_APPEND;
-	if (io_is_direct(file))
+	if (file->f_flags & O_DIRECT)
 		res |= IOCB_DIRECT;
 	if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
 		res |= IOCB_DSYNC;
-- 
2.25.1

[PATCH V9 03/11] fs/stat: Define DAX statx attribute

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

In order for users to determine if a file is currently operating in DAX
state (effective DAX).  Define a statx attribute value and set that
attribute if the effective DAX flag is set.

To go along with this we propose the following addition to the statx man
page:

STATX_ATTR_DAX

	The file is in the DAX (cpu direct access) state.  DAX state
	attempts to minimize software cache effects for both I/O and
	memory mappings of this file.  It requires a file system which
	has been configured to support DAX.

	DAX generally assumes all accesses are via cpu load / store
	instructions which can minimize overhead for small accesses, but
	may adversely affect cpu utilization for large transfers.

	File I/O is done directly to/from user-space buffers and memory
	mapped I/O may be performed with direct memory mappings that
	bypass kernel page cache.

	While the DAX property tends to result in data being transferred
	synchronously, it does not give the same guarantees of O_SYNC
	where data and the necessary metadata are transferred together.

	A DAX file may support being mapped with the MAP_SYNC flag,
	which enables a program to use CPU cache flush instructions to
	persist CPU store operations without an explicit fsync(2).  See
	mmap(2) for more information.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>

---
Changes from V2:
	Update man page text with comments from Darrick, Jan, Dan, and
	Dave.
---
 fs/stat.c                 | 3 +++
 include/uapi/linux/stat.h | 1 +
 2 files changed, 4 insertions(+)

diff --git a/fs/stat.c b/fs/stat.c
index 030008796479..894699c74dde 100644
--- a/fs/stat.c
+++ b/fs/stat.c
@@ -79,6 +79,9 @@ int vfs_getattr_nosec(const struct path *path, struct kstat *stat,
 	if (IS_AUTOMOUNT(inode))
 		stat->attributes |= STATX_ATTR_AUTOMOUNT;
 
+	if (IS_DAX(inode))
+		stat->attributes |= STATX_ATTR_DAX;
+
 	if (inode->i_op->getattr)
 		return inode->i_op->getattr(path, stat, request_mask,
 					    query_flags);
diff --git a/include/uapi/linux/stat.h b/include/uapi/linux/stat.h
index ad80a5c885d5..e5f9d5517f6b 100644
--- a/include/uapi/linux/stat.h
+++ b/include/uapi/linux/stat.h
@@ -169,6 +169,7 @@ struct statx {
 #define STATX_ATTR_ENCRYPTED		0x00000800 /* [I] File requires key to decrypt in fs */
 #define STATX_ATTR_AUTOMOUNT		0x00001000 /* Dir: Automount trigger */
 #define STATX_ATTR_VERITY		0x00100000 /* [I] Verity protected file */
+#define STATX_ATTR_DAX			0x00002000 /* [I] File is DAX */
 
 
 #endif /* _UAPI_LINUX_STAT_H */
-- 
2.25.1

[PATCH V9 04/11] fs/xfs: Change XFS_MOUNT_DAX to XFS_MOUNT_DAX_ALWAYS

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

In prep for the new tri-state mount option which then introduces
XFS_MOUNT_DAX_NEVER.

Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>

---
Changes from v8
	Move bit to 26
---
 fs/xfs/xfs_iops.c  | 2 +-
 fs/xfs/xfs_mount.h | 3 +--
 fs/xfs/xfs_super.c | 8 ++++----
 3 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c
index f7a99b3bbcf7..462f89af479a 100644
--- a/fs/xfs/xfs_iops.c
+++ b/fs/xfs/xfs_iops.c
@@ -1248,7 +1248,7 @@ xfs_inode_supports_dax(
 		return false;
 
 	/* DAX mount option or DAX iflag must be set. */
-	if (!(mp->m_flags & XFS_MOUNT_DAX) &&
+	if (!(mp->m_flags & XFS_MOUNT_DAX_ALWAYS) &&
 	    !(ip->i_d.di_flags2 & XFS_DIFLAG2_DAX))
 		return false;
 
diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
index b2e4598fdf7d..f6123fb0113c 100644
--- a/fs/xfs/xfs_mount.h
+++ b/fs/xfs/xfs_mount.h
@@ -237,8 +237,7 @@ typedef struct xfs_mount {
 #define XFS_MOUNT_FILESTREAMS	(1ULL << 24)	/* enable the filestreams
 						   allocator */
 #define XFS_MOUNT_NOATTR2	(1ULL << 25)	/* disable use of attr2 format */
-
-#define XFS_MOUNT_DAX		(1ULL << 62)	/* TEST ONLY! */
+#define XFS_MOUNT_DAX_ALWAYS	(1ULL << 26)
 
 /*
  * Max and min values for mount-option defined I/O
diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
index 424bb9a2d532..ce169d1c7474 100644
--- a/fs/xfs/xfs_super.c
+++ b/fs/xfs/xfs_super.c
@@ -129,7 +129,7 @@ xfs_fs_show_options(
 		{ XFS_MOUNT_GRPID,		",grpid" },
 		{ XFS_MOUNT_DISCARD,		",discard" },
 		{ XFS_MOUNT_LARGEIO,		",largeio" },
-		{ XFS_MOUNT_DAX,		",dax" },
+		{ XFS_MOUNT_DAX_ALWAYS,		",dax" },
 		{ 0, NULL }
 	};
 	struct xfs_mount	*mp = XFS_M(root->d_sb);
@@ -1261,7 +1261,7 @@ xfs_fc_parse_param(
 		return 0;
 #ifdef CONFIG_FS_DAX
 	case Opt_dax:
-		mp->m_flags |= XFS_MOUNT_DAX;
+		mp->m_flags |= XFS_MOUNT_DAX_ALWAYS;
 		return 0;
 #endif
 	default:
@@ -1454,7 +1454,7 @@ xfs_fc_fill_super(
 	if (XFS_SB_VERSION_NUM(&mp->m_sb) == XFS_SB_VERSION_5)
 		sb->s_flags |= SB_I_VERSION;
 
-	if (mp->m_flags & XFS_MOUNT_DAX) {
+	if (mp->m_flags & XFS_MOUNT_DAX_ALWAYS) {
 		bool rtdev_is_dax = false, datadev_is_dax;
 
 		xfs_warn(mp,
@@ -1468,7 +1468,7 @@ xfs_fc_fill_super(
 		if (!rtdev_is_dax && !datadev_is_dax) {
 			xfs_alert(mp,
 			"DAX unsupported by block device. Turning off DAX.");
-			mp->m_flags &= ~XFS_MOUNT_DAX;
+			mp->m_flags &= ~XFS_MOUNT_DAX_ALWAYS;
 		}
 		if (xfs_sb_version_hasreflink(&mp->m_sb)) {
 			xfs_alert(mp,
-- 
2.25.1

[PATCH V9 05/11] fs/xfs: Make DAX mount option a tri-state

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

As agreed upon[1].  We make the dax mount option a tri-state.  '-o dax'
continues to operate the same.  We add 'always', 'never', and 'inode'
(default).

[1] https://lore.kernel.org/lkml/20200405061945.GA94792@iweiny-DESK2.sc.intel.com/

Signed-off-by: Ira Weiny <ira.weiny@intel.com>

---
Changes from v8:
	Move NEVER bit to 27
	Use xfs signature style
	use xfs_dax_mode enum

Changes from v7:
	Change to XFS_MOUNT_DAX_NEVER

Changes from v6:
	Use 2 flag bits rather than a field.
	change iflag to inode

Changes from v5:
	New Patch
---
 fs/xfs/xfs_mount.h |  1 +
 fs/xfs/xfs_super.c | 51 ++++++++++++++++++++++++++++++++++++++++++----
 2 files changed, 48 insertions(+), 4 deletions(-)

diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
index f6123fb0113c..37bfb50db809 100644
--- a/fs/xfs/xfs_mount.h
+++ b/fs/xfs/xfs_mount.h
@@ -238,6 +238,7 @@ typedef struct xfs_mount {
 						   allocator */
 #define XFS_MOUNT_NOATTR2	(1ULL << 25)	/* disable use of attr2 format */
 #define XFS_MOUNT_DAX_ALWAYS	(1ULL << 26)
+#define XFS_MOUNT_DAX_NEVER	(1ULL << 27)
 
 /*
  * Max and min values for mount-option defined I/O
diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
index ce169d1c7474..0d0f74786799 100644
--- a/fs/xfs/xfs_super.c
+++ b/fs/xfs/xfs_super.c
@@ -47,6 +47,39 @@ static struct kset *xfs_kset;		/* top-level xfs sysfs dir */
 static struct xfs_kobj xfs_dbg_kobj;	/* global debug sysfs attrs */
 #endif
 
+enum xfs_dax_mode {
+	XFS_DAX_INODE = 0,
+	XFS_DAX_ALWAYS = 1,
+	XFS_DAX_NEVER = 2,
+};
+
+static void
+xfs_mount_set_dax_mode(
+	struct xfs_mount *mp,
+	enum xfs_dax_mode mode)
+{
+	switch (mode) {
+		case XFS_DAX_INODE:
+			mp->m_flags &= ~(XFS_MOUNT_DAX_ALWAYS | XFS_MOUNT_DAX_NEVER);
+			break;
+		case XFS_DAX_ALWAYS:
+			mp->m_flags |= XFS_MOUNT_DAX_ALWAYS;
+			mp->m_flags &= ~XFS_MOUNT_DAX_NEVER;
+			break;
+		case XFS_DAX_NEVER:
+			mp->m_flags |= XFS_MOUNT_DAX_NEVER;
+			mp->m_flags &= ~XFS_MOUNT_DAX_ALWAYS;
+			break;
+	}
+}
+
+static const struct constant_table dax_param_enums[] = {
+	{"inode",	XFS_DAX_INODE },
+	{"always",	XFS_DAX_ALWAYS },
+	{"never",	XFS_DAX_NEVER },
+	{}
+};
+
 /*
  * Table driven mount option parser.
  */
@@ -59,7 +92,7 @@ enum {
 	Opt_filestreams, Opt_quota, Opt_noquota, Opt_usrquota, Opt_grpquota,
 	Opt_prjquota, Opt_uquota, Opt_gquota, Opt_pquota,
 	Opt_uqnoenforce, Opt_gqnoenforce, Opt_pqnoenforce, Opt_qnoenforce,
-	Opt_discard, Opt_nodiscard, Opt_dax,
+	Opt_discard, Opt_nodiscard, Opt_dax, Opt_dax_enum,
 };
 
 static const struct fs_parameter_spec xfs_fs_parameters[] = {
@@ -103,6 +136,7 @@ static const struct fs_parameter_spec xfs_fs_parameters[] = {
 	fsparam_flag("discard",		Opt_discard),
 	fsparam_flag("nodiscard",	Opt_nodiscard),
 	fsparam_flag("dax",		Opt_dax),
+	fsparam_enum("dax",		Opt_dax_enum, dax_param_enums),
 	{}
 };
 
@@ -129,7 +163,6 @@ xfs_fs_show_options(
 		{ XFS_MOUNT_GRPID,		",grpid" },
 		{ XFS_MOUNT_DISCARD,		",discard" },
 		{ XFS_MOUNT_LARGEIO,		",largeio" },
-		{ XFS_MOUNT_DAX_ALWAYS,		",dax" },
 		{ 0, NULL }
 	};
 	struct xfs_mount	*mp = XFS_M(root->d_sb);
@@ -185,6 +218,13 @@ xfs_fs_show_options(
 	if (!(mp->m_qflags & XFS_ALL_QUOTA_ACCT))
 		seq_puts(m, ",noquota");
 
+	if (mp->m_flags & XFS_MOUNT_DAX_ALWAYS)
+		seq_puts(m, ",dax=always");
+	else if (mp->m_flags & XFS_MOUNT_DAX_NEVER)
+		seq_puts(m, ",dax=never");
+	else
+		seq_puts(m, ",dax=inode");
+
 	return 0;
 }
 
@@ -1261,7 +1301,10 @@ xfs_fc_parse_param(
 		return 0;
 #ifdef CONFIG_FS_DAX
 	case Opt_dax:
-		mp->m_flags |= XFS_MOUNT_DAX_ALWAYS;
+		xfs_mount_set_dax_mode(mp, XFS_DAX_ALWAYS);
+		return 0;
+	case Opt_dax_enum:
+		xfs_mount_set_dax_mode(mp, result.uint_32);
 		return 0;
 #endif
 	default:
@@ -1468,7 +1511,7 @@ xfs_fc_fill_super(
 		if (!rtdev_is_dax && !datadev_is_dax) {
 			xfs_alert(mp,
 			"DAX unsupported by block device. Turning off DAX.");
-			mp->m_flags &= ~XFS_MOUNT_DAX_ALWAYS;
+			xfs_mount_set_dax_mode(mp, XFS_DAX_NEVER);
 		}
 		if (xfs_sb_version_hasreflink(&mp->m_sb)) {
 			xfs_alert(mp,
-- 
2.25.1

[PATCH V9 06/11] fs/xfs: Create function xfs_inode_should_enable_dax()

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

xfs_inode_supports_dax() should reflect if the inode can support DAX not
that it is enabled for DAX.

Change the use of xfs_inode_supports_dax() to reflect only if the inode
and underlying storage support dax.

Add a new function xfs_inode_should_enable_dax() which reflects if the
inode should be enabled for DAX.

Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>

---
Changes from V8
	Change to 'should enable' (feedback given by Jan in ext4 series)
		Darrick I've preserved your Reviewed-by for now LMK if
		that is an issue.

Changes from v7:
	Move S_ISREG check first
	use IS_ENABLED(CONFIG_FS_DAX) rather than duplicated function

Changes from v6:
	Change enable checks to be sequential logic.
	Update for 2 bit tri-state option.
	Make 'static' consistent.
	Don't set S_DAX if !CONFIG_FS_DAX

Changes from v5:
	Update to reflect the new tri-state mount option

Changes from v3:
	Update functions and names to be more clear
	Update commit message
	Merge with
		'fs/xfs: Clean up DAX support check'
		don't allow IS_DAX() on a directory
		use STATIC macro for static
		make xfs_inode_supports_dax() static
---
 fs/xfs/xfs_iops.c | 28 ++++++++++++++++++++++------
 1 file changed, 22 insertions(+), 6 deletions(-)

diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c
index 462f89af479a..1814f10e43d3 100644
--- a/fs/xfs/xfs_iops.c
+++ b/fs/xfs/xfs_iops.c
@@ -1243,13 +1243,12 @@ xfs_inode_supports_dax(
 {
 	struct xfs_mount	*mp = ip->i_mount;
 
-	/* Only supported on non-reflinked files. */
-	if (!S_ISREG(VFS_I(ip)->i_mode) || xfs_is_reflink_inode(ip))
+	/* Only supported on regular files. */
+	if (!S_ISREG(VFS_I(ip)->i_mode))
 		return false;
 
-	/* DAX mount option or DAX iflag must be set. */
-	if (!(mp->m_flags & XFS_MOUNT_DAX_ALWAYS) &&
-	    !(ip->i_d.di_flags2 & XFS_DIFLAG2_DAX))
+	/* Only supported on non-reflinked files. */
+	if (xfs_is_reflink_inode(ip))
 		return false;
 
 	/* Block size must match page size */
@@ -1260,6 +1259,23 @@ xfs_inode_supports_dax(
 	return xfs_inode_buftarg(ip)->bt_daxdev != NULL;
 }
 
+static bool
+xfs_inode_should_enable_dax(
+	struct xfs_inode *ip)
+{
+	if (!IS_ENABLED(CONFIG_FS_DAX))
+		return false;
+	if (ip->i_mount->m_flags & XFS_MOUNT_DAX_NEVER)
+		return false;
+	if (!xfs_inode_supports_dax(ip))
+		return false;
+	if (ip->i_mount->m_flags & XFS_MOUNT_DAX_ALWAYS)
+		return true;
+	if (ip->i_d.di_flags2 & XFS_DIFLAG2_DAX)
+		return true;
+	return false;
+}
+
 STATIC void
 xfs_diflags_to_iflags(
 	struct inode		*inode,
@@ -1278,7 +1294,7 @@ xfs_diflags_to_iflags(
 		inode->i_flags |= S_SYNC;
 	if (flags & XFS_DIFLAG_NOATIME)
 		inode->i_flags |= S_NOATIME;
-	if (xfs_inode_supports_dax(ip))
+	if (xfs_inode_should_enable_dax(ip))
 		inode->i_flags |= S_DAX;
 }
 
-- 
2.25.1

[PATCH V9 07/11] fs/xfs: Combine xfs_diflags_to_linux() and xfs_diflags_to_iflags()

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

The functionality in xfs_diflags_to_linux() and xfs_diflags_to_iflags() are
nearly identical.  The only difference is that *_to_linux() is called after
inode setup and disallows changing the DAX flag.

Combining them can be done with a flag which indicates if this is the initial
setup to allow the DAX flag to be properly set only at init time.

So remove xfs_diflags_to_linux() and call the modified xfs_diflags_to_iflags()
directly.

While we are here simplify xfs_diflags_to_iflags() to take struct xfs_inode and
use xfs_ip2xflags() to ensure future diflags are included correctly.

Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>

---
Changes from V7:
	Clarify with a comment the reason for leaving S_DAX out of the
	mask

Changes from V6:
	Move unrelated hunk to previous patch.
	Change logic for better code generation.

Changes from V5:
	The functions are no longer identical so we can only combine
	them rather than deleting one completely.  This is reflected in
	the new init parameter.
---
 fs/xfs/xfs_inode.h |  1 +
 fs/xfs/xfs_ioctl.c | 33 +--------------------------------
 fs/xfs/xfs_iops.c  | 46 +++++++++++++++++++++++++++-------------------
 3 files changed, 29 insertions(+), 51 deletions(-)

diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index c6a63f6764a6..83073c883fbf 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -467,6 +467,7 @@ int	xfs_break_layouts(struct inode *inode, uint *iolock,
 /* from xfs_iops.c */
 extern void xfs_setup_inode(struct xfs_inode *ip);
 extern void xfs_setup_iops(struct xfs_inode *ip);
+extern void xfs_diflags_to_iflags(struct xfs_inode *ip, bool init);
 
 /*
  * When setting up a newly allocated inode, we need to call
diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
index 309958186d33..104495ac187c 100644
--- a/fs/xfs/xfs_ioctl.c
+++ b/fs/xfs/xfs_ioctl.c
@@ -1201,37 +1201,6 @@ xfs_flags2diflags2(
 	return di_flags2;
 }
 
-STATIC void
-xfs_diflags_to_linux(
-	struct xfs_inode	*ip)
-{
-	struct inode		*inode = VFS_I(ip);
-	unsigned int		xflags = xfs_ip2xflags(ip);
-
-	if (xflags & FS_XFLAG_IMMUTABLE)
-		inode->i_flags |= S_IMMUTABLE;
-	else
-		inode->i_flags &= ~S_IMMUTABLE;
-	if (xflags & FS_XFLAG_APPEND)
-		inode->i_flags |= S_APPEND;
-	else
-		inode->i_flags &= ~S_APPEND;
-	if (xflags & FS_XFLAG_SYNC)
-		inode->i_flags |= S_SYNC;
-	else
-		inode->i_flags &= ~S_SYNC;
-	if (xflags & FS_XFLAG_NOATIME)
-		inode->i_flags |= S_NOATIME;
-	else
-		inode->i_flags &= ~S_NOATIME;
-#if 0	/* disabled until the flag switching races are sorted out */
-	if (xflags & FS_XFLAG_DAX)
-		inode->i_flags |= S_DAX;
-	else
-		inode->i_flags &= ~S_DAX;
-#endif
-}
-
 static int
 xfs_ioctl_setattr_xflags(
 	struct xfs_trans	*tp,
@@ -1269,7 +1238,7 @@ xfs_ioctl_setattr_xflags(
 	ip->i_d.di_flags = xfs_flags2diflags(ip, fa->fsx_xflags);
 	ip->i_d.di_flags2 = di_flags2;
 
-	xfs_diflags_to_linux(ip);
+	xfs_diflags_to_iflags(ip, false);
 	xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG);
 	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
 	XFS_STATS_INC(mp, xs_ig_attrchg);
diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c
index 1814f10e43d3..b70b735fe4a4 100644
--- a/fs/xfs/xfs_iops.c
+++ b/fs/xfs/xfs_iops.c
@@ -1276,26 +1276,34 @@ xfs_inode_should_enable_dax(
 	return false;
 }
 
-STATIC void
+void
 xfs_diflags_to_iflags(
-	struct inode		*inode,
-	struct xfs_inode	*ip)
+	struct xfs_inode	*ip,
+	bool init)
 {
-	uint16_t		flags = ip->i_d.di_flags;
-
-	inode->i_flags &= ~(S_IMMUTABLE | S_APPEND | S_SYNC |
-			    S_NOATIME | S_DAX);
-
-	if (flags & XFS_DIFLAG_IMMUTABLE)
-		inode->i_flags |= S_IMMUTABLE;
-	if (flags & XFS_DIFLAG_APPEND)
-		inode->i_flags |= S_APPEND;
-	if (flags & XFS_DIFLAG_SYNC)
-		inode->i_flags |= S_SYNC;
-	if (flags & XFS_DIFLAG_NOATIME)
-		inode->i_flags |= S_NOATIME;
-	if (xfs_inode_should_enable_dax(ip))
-		inode->i_flags |= S_DAX;
+	struct inode            *inode = VFS_I(ip);
+	unsigned int            xflags = xfs_ip2xflags(ip);
+	unsigned int            flags = 0;
+
+	ASSERT(!(IS_DAX(inode) && init));
+
+	if (xflags & FS_XFLAG_IMMUTABLE)
+		flags |= S_IMMUTABLE;
+	if (xflags & FS_XFLAG_APPEND)
+		flags |= S_APPEND;
+	if (xflags & FS_XFLAG_SYNC)
+		flags |= S_SYNC;
+	if (xflags & FS_XFLAG_NOATIME)
+		flags |= S_NOATIME;
+	if (init && xfs_inode_should_enable_dax(ip))
+		flags |= S_DAX;
+
+	/*
+	 * S_DAX can only be set during inode initialization and is never set by
+	 * the VFS, so we cannot mask off S_DAX in i_flags.
+	 */
+	inode->i_flags &= ~(S_IMMUTABLE | S_APPEND | S_SYNC | S_NOATIME);
+	inode->i_flags |= flags;
 }
 
 /*
@@ -1321,7 +1329,7 @@ xfs_setup_inode(
 	inode_fake_hash(inode);
 
 	i_size_write(inode, ip->i_d.di_size);
-	xfs_diflags_to_iflags(inode, ip);
+	xfs_diflags_to_iflags(ip, true);
 
 	if (S_ISDIR(inode->i_mode)) {
 		/*
-- 
2.25.1

[PATCH V9 08/11] fs: Lift XFS_IDONTCACNE to the VFS layer

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

DAX effective mode (S_DAX) changes requires inode eviction.

XFS has an advisory flag (XFS_IDONTCACHE) to prevent caching of the
inode if no other additional references are taken.  We lift this flag to
the VFS layer and change the behavior slightly by allowing the flag to
remain even if multiple references are taken.

This will expedite the eviction of inodes to change S_DAX.

Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>

---
Changes from V8:
	Remove XFS_IDONTCACHE
---
 fs/xfs/xfs_icache.c | 4 ++--
 fs/xfs/xfs_inode.h  | 2 +-
 fs/xfs/xfs_super.c  | 2 +-
 include/linux/fs.h  | 6 +++++-
 4 files changed, 9 insertions(+), 5 deletions(-)

diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
index 17a0b86fe701..de76f7f60695 100644
--- a/fs/xfs/xfs_icache.c
+++ b/fs/xfs/xfs_icache.c
@@ -477,7 +477,7 @@ xfs_iget_cache_hit(
 		xfs_ilock(ip, lock_flags);
 
 	if (!(flags & XFS_IGET_INCORE))
-		xfs_iflags_clear(ip, XFS_ISTALE | XFS_IDONTCACHE);
+		xfs_iflags_clear(ip, XFS_ISTALE);
 	XFS_STATS_INC(mp, xs_ig_found);
 
 	return 0;
@@ -559,7 +559,7 @@ xfs_iget_cache_miss(
 	 */
 	iflags = XFS_INEW;
 	if (flags & XFS_IGET_DONTCACHE)
-		iflags |= XFS_IDONTCACHE;
+		VFS_I(ip)->i_state |= I_DONTCACHE;
 	ip->i_udquot = NULL;
 	ip->i_gdquot = NULL;
 	ip->i_pdquot = NULL;
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index 83073c883fbf..52b8ee21a0b1 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -218,7 +218,7 @@ static inline bool xfs_inode_has_cow_data(struct xfs_inode *ip)
 #define XFS_IFLOCK		(1 << __XFS_IFLOCK_BIT)
 #define __XFS_IPINNED_BIT	8	 /* wakeup key for zero pin count */
 #define XFS_IPINNED		(1 << __XFS_IPINNED_BIT)
-#define XFS_IDONTCACHE		(1 << 9) /* don't cache the inode long term */
+/* Was XFS_IDONTCACHE 9 */
 #define XFS_IEOFBLOCKS		(1 << 10)/* has the preallocblocks tag set */
 /*
  * If this unlinked inode is in the middle of recovery, don't let drop_inode
diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
index 0d0f74786799..2e165e226e15 100644
--- a/fs/xfs/xfs_super.c
+++ b/fs/xfs/xfs_super.c
@@ -742,7 +742,7 @@ xfs_fs_drop_inode(
 		return 0;
 	}
 
-	return generic_drop_inode(inode) || (ip->i_flags & XFS_IDONTCACHE);
+	return generic_drop_inode(inode);
 }
 
 static void
diff --git a/include/linux/fs.h b/include/linux/fs.h
index a87cc5845a02..44bd45af760f 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -2156,6 +2156,8 @@ static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
  *
  * I_CREATING		New object's inode in the middle of setting up.
  *
+ * I_DONTCACHE		Evict inode as soon as it is not used anymore.
+ *
  * Q: What is the difference between I_WILL_FREE and I_FREEING?
  */
 #define I_DIRTY_SYNC		(1 << 0)
@@ -2178,6 +2180,7 @@ static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
 #define I_WB_SWITCH		(1 << 13)
 #define I_OVL_INUSE		(1 << 14)
 #define I_CREATING		(1 << 15)
+#define I_DONTCACHE		(1 << 16)
 
 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
@@ -3049,7 +3052,8 @@ extern int inode_needs_sync(struct inode *inode);
 extern int generic_delete_inode(struct inode *inode);
 static inline int generic_drop_inode(struct inode *inode)
 {
-	return !inode->i_nlink || inode_unhashed(inode);
+	return !inode->i_nlink || inode_unhashed(inode) ||
+		(inode->i_state & I_DONTCACHE);
 }
 
 extern struct inode *ilookup5_nowait(struct super_block *sb,
-- 
2.25.1

[PATCH V9 09/11] fs: Introduce DCACHE_DONTCACHE

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

DCACHE_DONTCACHE indicates a dentry should not be cached on final
dput().

Also add a helper function to mark DCACHE_DONTCACHE on all dentries
pointing to a specific inode when that inode is being set I_DONTCACHE.

This facilitates dropping dentry references to inodes sooner which
require eviction to swap S_DAX mode.

Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>

---
Changes from V8:
	Update commit message
	Use mark_inode_dontcache in XFS
	Fix locking...  can't use rcu here.
	Change name to mark_inode_dontcache
---
 fs/dcache.c            |  4 ++++
 fs/inode.c             | 15 +++++++++++++++
 fs/xfs/xfs_icache.c    |  2 +-
 include/linux/dcache.h |  2 ++
 include/linux/fs.h     |  1 +
 5 files changed, 23 insertions(+), 1 deletion(-)

diff --git a/fs/dcache.c b/fs/dcache.c
index b280e07e162b..0030fabab2c4 100644
--- a/fs/dcache.c
+++ b/fs/dcache.c
@@ -647,6 +647,10 @@ static inline bool retain_dentry(struct dentry *dentry)
 		if (dentry->d_op->d_delete(dentry))
 			return false;
 	}
+
+	if (unlikely(dentry->d_flags & DCACHE_DONTCACHE))
+		return false;
+
 	/* retain; LRU fodder */
 	dentry->d_lockref.count--;
 	if (unlikely(!(dentry->d_flags & DCACHE_LRU_LIST)))
diff --git a/fs/inode.c b/fs/inode.c
index 93d9252a00ab..da7f3c4926cd 100644
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -1526,6 +1526,21 @@ int generic_delete_inode(struct inode *inode)
 }
 EXPORT_SYMBOL(generic_delete_inode);
 
+void mark_inode_dontcache(struct inode *inode)
+{
+	struct dentry *de;
+
+	spin_lock(&inode->i_lock);
+	hlist_for_each_entry(de, &inode->i_dentry, d_u.d_alias) {
+		spin_lock(&de->d_lock);
+		de->d_flags |= DCACHE_DONTCACHE;
+		spin_unlock(&de->d_lock);
+	}
+	spin_unlock(&inode->i_lock);
+	inode->i_state |= I_DONTCACHE;
+}
+EXPORT_SYMBOL(mark_inode_dontcache);
+
 /*
  * Called when we're dropping the last reference
  * to an inode.
diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
index de76f7f60695..3c8f44477804 100644
--- a/fs/xfs/xfs_icache.c
+++ b/fs/xfs/xfs_icache.c
@@ -559,7 +559,7 @@ xfs_iget_cache_miss(
 	 */
 	iflags = XFS_INEW;
 	if (flags & XFS_IGET_DONTCACHE)
-		VFS_I(ip)->i_state |= I_DONTCACHE;
+		mark_inode_dontcache(VFS_I(ip));
 	ip->i_udquot = NULL;
 	ip->i_gdquot = NULL;
 	ip->i_pdquot = NULL;
diff --git a/include/linux/dcache.h b/include/linux/dcache.h
index c1488cc84fd9..56b1482d9223 100644
--- a/include/linux/dcache.h
+++ b/include/linux/dcache.h
@@ -177,6 +177,8 @@ struct dentry_operations {
 
 #define DCACHE_REFERENCED		0x00000040 /* Recently used, don't discard. */
 
+#define DCACHE_DONTCACHE		0x00000080 /* don't cache on final dput() */
+
 #define DCACHE_CANT_MOUNT		0x00000100
 #define DCACHE_GENOCIDE			0x00000200
 #define DCACHE_SHRINK_LIST		0x00000400
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 44bd45af760f..064168ec2e0b 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -3055,6 +3055,7 @@ static inline int generic_drop_inode(struct inode *inode)
 	return !inode->i_nlink || inode_unhashed(inode) ||
 		(inode->i_state & I_DONTCACHE);
 }
+extern void mark_inode_dontcache(struct inode *inode);
 
 extern struct inode *ilookup5_nowait(struct super_block *sb,
 		unsigned long hashval, int (*test)(struct inode *, void *),
-- 
2.25.1

[PATCH V9 10/11] fs/xfs: Update xfs_ioctl_setattr_dax_invalidate()

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

Because of the separation of FS_XFLAG_DAX from S_DAX and the delayed
setting of S_DAX, data invalidation no longer needs to happen when
FS_XFLAG_DAX is changed.

Change xfs_ioctl_setattr_dax_invalidate() to be
xfs_ioctl_dax_check_set_cache() and alter the code to reflect the new
functionality.

Furthermore, we no longer need the locking so we remove the join_flags
logic.

Signed-off-by: Ira Weiny <ira.weiny@intel.com>

---
Changes from V8:
	Change name of function to xfs_ioctl_dax_check_set_cache()
	Update commit message
	Fix bit manipulations

Changes from V7:
	Use new flag_inode_dontcache()
	Skip don't cache if mount over ride is active.

Changes from v6:
	Fix completely broken implementation and update commit message.
	Use the new VFS layer I_DONTCACHE to facilitate inode eviction
	and S_DAX changing on drop_caches

Changes from v5:
	New patch
---
 fs/xfs/xfs_ioctl.c | 108 +++++++++------------------------------------
 1 file changed, 20 insertions(+), 88 deletions(-)

diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
index 104495ac187c..b87b571a6748 100644
--- a/fs/xfs/xfs_ioctl.c
+++ b/fs/xfs/xfs_ioctl.c
@@ -1245,64 +1245,26 @@ xfs_ioctl_setattr_xflags(
 	return 0;
 }
 
-/*
- * If we are changing DAX flags, we have to ensure the file is clean and any
- * cached objects in the address space are invalidated and removed. This
- * requires us to lock out other IO and page faults similar to a truncate
- * operation. The locks need to be held until the transaction has been committed
- * so that the cache invalidation is atomic with respect to the DAX flag
- * manipulation.
- */
-static int
-xfs_ioctl_setattr_dax_invalidate(
+static void
+xfs_ioctl_dax_check_set_cache(
 	struct xfs_inode	*ip,
-	struct fsxattr		*fa,
-	int			*join_flags)
+	struct fsxattr		*fa)
 {
-	struct inode		*inode = VFS_I(ip);
-	struct super_block	*sb = inode->i_sb;
-	int			error;
-
-	*join_flags = 0;
-
-	/*
-	 * It is only valid to set the DAX flag on regular files and
-	 * directories on filesystems where the block size is equal to the page
-	 * size. On directories it serves as an inherited hint so we don't
-	 * have to check the device for dax support or flush pagecache.
-	 */
-	if (fa->fsx_xflags & FS_XFLAG_DAX) {
-		struct xfs_buftarg	*target = xfs_inode_buftarg(ip);
-
-		if (!bdev_dax_supported(target->bt_bdev, sb->s_blocksize))
-			return -EINVAL;
-	}
-
-	/* If the DAX state is not changing, we have nothing to do here. */
-	if ((fa->fsx_xflags & FS_XFLAG_DAX) && IS_DAX(inode))
-		return 0;
-	if (!(fa->fsx_xflags & FS_XFLAG_DAX) && !IS_DAX(inode))
-		return 0;
+	struct xfs_mount	*mp = ip->i_mount;
+	struct inode            *inode = VFS_I(ip);
 
 	if (S_ISDIR(inode->i_mode))
-		return 0;
-
-	/* lock, flush and invalidate mapping in preparation for flag change */
-	xfs_ilock(ip, XFS_MMAPLOCK_EXCL | XFS_IOLOCK_EXCL);
-	error = filemap_write_and_wait(inode->i_mapping);
-	if (error)
-		goto out_unlock;
-	error = invalidate_inode_pages2(inode->i_mapping);
-	if (error)
-		goto out_unlock;
-
-	*join_flags = XFS_MMAPLOCK_EXCL | XFS_IOLOCK_EXCL;
-	return 0;
+		return;
 
-out_unlock:
-	xfs_iunlock(ip, XFS_MMAPLOCK_EXCL | XFS_IOLOCK_EXCL);
-	return error;
+	if ((mp->m_flags & XFS_MOUNT_DAX_ALWAYS) ||
+	    (mp->m_flags & XFS_MOUNT_DAX_NEVER))
+		return;
 
+	if (((fa->fsx_xflags & FS_XFLAG_DAX) &&
+	    !(ip->i_d.di_flags2 & XFS_DIFLAG2_DAX)) ||
+	    (!(fa->fsx_xflags & FS_XFLAG_DAX) &&
+	     (ip->i_d.di_flags2 & XFS_DIFLAG2_DAX)))
+		mark_inode_dontcache(inode);
 }
 
 /*
@@ -1310,17 +1272,10 @@ xfs_ioctl_setattr_dax_invalidate(
  * have permission to do so. On success, return a clean transaction and the
  * inode locked exclusively ready for further operation specific checks. On
  * failure, return an error without modifying or locking the inode.
- *
- * The inode might already be IO locked on call. If this is the case, it is
- * indicated in @join_flags and we take full responsibility for ensuring they
- * are unlocked from now on. Hence if we have an error here, we still have to
- * unlock them. Otherwise, once they are joined to the transaction, they will
- * be unlocked on commit/cancel.
  */
 static struct xfs_trans *
 xfs_ioctl_setattr_get_trans(
-	struct xfs_inode	*ip,
-	int			join_flags)
+	struct xfs_inode	*ip)
 {
 	struct xfs_mount	*mp = ip->i_mount;
 	struct xfs_trans	*tp;
@@ -1337,8 +1292,7 @@ xfs_ioctl_setattr_get_trans(
 		goto out_unlock;
 
 	xfs_ilock(ip, XFS_ILOCK_EXCL);
-	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL | join_flags);
-	join_flags = 0;
+	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
 
 	/*
 	 * CAP_FOWNER overrides the following restrictions:
@@ -1359,8 +1313,6 @@ xfs_ioctl_setattr_get_trans(
 out_cancel:
 	xfs_trans_cancel(tp);
 out_unlock:
-	if (join_flags)
-		xfs_iunlock(ip, join_flags);
 	return ERR_PTR(error);
 }
 
@@ -1486,7 +1438,6 @@ xfs_ioctl_setattr(
 	struct xfs_dquot	*pdqp = NULL;
 	struct xfs_dquot	*olddquot = NULL;
 	int			code;
-	int			join_flags = 0;
 
 	trace_xfs_ioctl_setattr(ip);
 
@@ -1510,18 +1461,9 @@ xfs_ioctl_setattr(
 			return code;
 	}
 
-	/*
-	 * Changing DAX config may require inode locking for mapping
-	 * invalidation. These need to be held all the way to transaction commit
-	 * or cancel time, so need to be passed through to
-	 * xfs_ioctl_setattr_get_trans() so it can apply them to the join call
-	 * appropriately.
-	 */
-	code = xfs_ioctl_setattr_dax_invalidate(ip, fa, &join_flags);
-	if (code)
-		goto error_free_dquots;
+	xfs_ioctl_dax_check_set_cache(ip, fa);
 
-	tp = xfs_ioctl_setattr_get_trans(ip, join_flags);
+	tp = xfs_ioctl_setattr_get_trans(ip);
 	if (IS_ERR(tp)) {
 		code = PTR_ERR(tp);
 		goto error_free_dquots;
@@ -1651,7 +1593,6 @@ xfs_ioc_setxflags(
 	struct fsxattr		fa;
 	struct fsxattr		old_fa;
 	unsigned int		flags;
-	int			join_flags = 0;
 	int			error;
 
 	if (copy_from_user(&flags, arg, sizeof(flags)))
@@ -1668,18 +1609,9 @@ xfs_ioc_setxflags(
 	if (error)
 		return error;
 
-	/*
-	 * Changing DAX config may require inode locking for mapping
-	 * invalidation. These need to be held all the way to transaction commit
-	 * or cancel time, so need to be passed through to
-	 * xfs_ioctl_setattr_get_trans() so it can apply them to the join call
-	 * appropriately.
-	 */
-	error = xfs_ioctl_setattr_dax_invalidate(ip, &fa, &join_flags);
-	if (error)
-		goto out_drop_write;
+	xfs_ioctl_dax_check_set_cache(ip, &fa);
 
-	tp = xfs_ioctl_setattr_get_trans(ip, join_flags);
+	tp = xfs_ioctl_setattr_get_trans(ip);
 	if (IS_ERR(tp)) {
 		error = PTR_ERR(tp);
 		goto out_drop_write;
-- 
2.25.1

[PATCH V9 11/11] Documentation/dax: Update Usage section

[ira.weiny]

From: Ira Weiny <ira.weiny@intel.com>

Update the Usage section to reflect the new individual dax selection
functionality.

Signed-off-by: Ira Weiny <ira.weiny@intel.com>

---
Changes from V8:
	Updates from Darrick

Changes from V7:
	Cleanups/clarifications from Darrick and Dan

Changes from V6:
	Update to allow setting FS_XFLAG_DAX any time.
	Update with list of behaviors from Darrick
	https://lore.kernel.org/lkml/20200409165927.GD6741@magnolia/

Changes from V5:
	Update to reflect the agreed upon semantics
	https://lore.kernel.org/lkml/20200405061945.GA94792@iweiny-DESK2.sc.intel.com/
---
 Documentation/filesystems/dax.txt | 164 +++++++++++++++++++++++++++++-
 1 file changed, 161 insertions(+), 3 deletions(-)

diff --git a/Documentation/filesystems/dax.txt b/Documentation/filesystems/dax.txt
index 679729442fd2..8f4ab08be715 100644
--- a/Documentation/filesystems/dax.txt
+++ b/Documentation/filesystems/dax.txt
@@ -17,11 +17,169 @@ For file mappings, the storage device is mapped directly into userspace.
 Usage
 -----
 
-If you have a block device which supports DAX, you can make a filesystem
+If you have a block device which supports DAX, you can make a file system
 on it as usual.  The DAX code currently only supports files with a block
 size equal to your kernel's PAGE_SIZE, so you may need to specify a block
-size when creating the filesystem.  When mounting it, use the "-o dax"
-option on the command line or add 'dax' to the options in /etc/fstab.
+size when creating the file system.
+
+Currently 3 filesystems support DAX: ext2, ext4 and xfs.  Enabling DAX on them
+is different.
+
+Enabling DAX on ext4 and ext2
+-----------------------------
+
+When mounting the filesystem, use the "-o dax" option on the command line or
+add 'dax' to the options in /etc/fstab.  This works to enable DAX on all files
+within the filesystem.  It is equivalent to the '-o dax=always' behavior below.
+
+
+Enabling DAX on xfs
+-------------------
+
+Summary
+-------
+
+ 1. There exists an in-kernel file access mode flag S_DAX that corresponds to
+    the statx flag STATX_ATTR_DAX.  See the manpage for statx(2) for details
+    about this access mode.
+
+ 2. There exists an advisory file inode flag FS_XFLAG_DAX that is
+    inherited from the parent directory FS_XFLAG_DAX inode flag at file
+    creation time.  This advisory flag can be set or cleared at any
+    time, but doing so does not immediately affect the S_DAX state.
+
+    Unless overridden by mount options (see (3)), if FS_XFLAG_DAX is set
+    and the fs is on pmem then it will enable S_DAX at inode load time;
+    if FS_XFLAG_DAX is not set, it will not enable S_DAX.
+
+ 3. There exists a dax= mount option.
+
+    "-o dax=never"  means "never set S_DAX, ignore FS_XFLAG_DAX."
+
+    "-o dax=always" means "always set S_DAX (at least on pmem),
+                    and ignore FS_XFLAG_DAX."
+
+    "-o dax"        is an alias for "dax=always".
+
+    "-o dax=inode"  means "follow FS_XFLAG_DAX" and is the default.
+
+ 4. There exists an advisory directory inode flag FS_XFLAG_DAX that can
+    be set or cleared at any time.  The flag state is inherited by any files or
+    subdirectories when they are created within that directory.
+
+ 5. Programs that require a specific file access mode (DAX or not DAX)
+    can do one of the following:
+
+    (a) Create files in directories that the FS_XFLAG_DAX flag set as
+        needed; or
+
+    (b) Have the administrator set an override via mount option; or
+
+    (c) Set or clear the file's FS_XFLAG_DAX flag as needed.  Programs
+        must then cause the kernel to evict the inode from memory.  This
+        can be done by:
+
+        i>  Closing the file and re-opening the file and using statx to
+            see if the fs has changed the S_DAX flag; and
+
+        ii> If the file still does not have the desired S_DAX access
+            mode, either unmount and remount the filesystem, or close
+            the file and use drop_caches.
+
+ 6. It is expected that users who want to squeeze every last bit of performance
+    out of the particular rough and tumble bits of their storage will also be
+    exposed to the difficulties of what happens when the operating system can't
+    totally virtualize those hardware capabilities.  DAX is such a feature.
+
+
+Details
+-------
+
+There are 2 per-file dax flags.  One is a physical inode setting (FS_XFLAG_DAX)
+and the other a currently enabled state (S_DAX).
+
+FS_XFLAG_DAX is maintained, on disk, on individual inodes.  It is preserved
+within the file system.  This 'physical' config setting can be set using an
+ioctl and/or an application such as "xfs_io -c 'chattr [-+]x'".  Files and
+directories automatically inherit FS_XFLAG_DAX from their parent directory
+_when_ _created_.  Therefore, setting FS_XFLAG_DAX at directory creation time
+can be used to set a default behavior for an entire sub-tree.  (Doing so on the
+root directory acts to set a default for the entire file system.)
+
+To clarify inheritance here are 3 examples:
+
+Example A:
+
+mkdir -p a/b/c
+xfs_io 'chattr +x' a
+mkdir a/b/c/d
+mkdir a/e
+
+	dax: a,e
+	no dax: b,c,d
+
+Example B:
+
+mkdir a
+xfs_io 'chattr +x' a
+mkdir -p a/b/c/d
+
+	dax: a,b,c,d
+	no dax:
+
+Example C:
+
+mkdir -p a/b/c
+xfs_io 'chattr +x' c
+mkdir a/b/c/d
+
+	dax: c,d
+	no dax: a,b
+
+
+The current enabled state (S_DAX) is set when a file inode is _loaded_ based on
+the underlying media support, the value of FS_XFLAG_DAX, and the file systems
+dax mount option setting.  See below.
+
+statx can be used to query S_DAX.  NOTE that a directory will never have S_DAX
+set and therefore statx will never indicate that S_DAX is set on directories.
+
+NOTE: Setting the FS_XFLAG_DAX (specifically or through inheritance) occurs
+even if the underlying media does not support dax and/or the file system is
+overridden with a mount option.
+
+
+Overriding FS_XFLAG_DAX (dax= mount option)
+-------------------------------------------
+
+There exists a dax mount option.  Using the mount option does not change the
+physical configured state of individual files but overrides the S_DAX operating
+state when inodes are loaded.
+
+Given underlying media support, the dax mount option is a tri-state option
+(never, always, inode) with the following meanings:
+
+   "-o dax=never" means "never set S_DAX, ignore FS_XFLAG_DAX"
+   "-o dax=always" means "always set S_DAX, ignore FS_XFLAG_DAX"
+        "-o dax" by itself means "dax=always" to remain compatible with older
+	         kernels
+   "-o dax=inode" means "follow FS_XFLAG_DAX"
+
+The default state is 'inode'.  Given underlying media support, the following
+algorithm is used to determine the effective mode of the file S_DAX on a
+capable device.
+
+	S_DAX = FS_XFLAG_DAX;
+
+	if (dax_mount == "always")
+		S_DAX = true;
+	else if (dax_mount == "off"
+		S_DAX = false;
+
+To reiterate: Setting, and inheritance, continues to affect FS_XFLAG_DAX even
+while the file system is mounted with a dax override.  However, in-core inode
+state (S_DAX) will continue to be overridden until the filesystem is remounted
+with dax=inode and the inode is evicted."
 
 
 Implementation Tips for Block Driver Writers
-- 
2.25.1

Re: [PATCH V9 02/11] fs: Remove unneeded IS_DAX() check in io_is_direct()

[Darrick J. Wong]

On Tue, Apr 21, 2020 at 12:17:44PM -0700, ira.weiny@intel.com wrote:
> From: Ira Weiny <ira.weiny@intel.com>
> 
> Remove the check because DAX now has it's own read/write methods and
> file systems which support DAX check IS_DAX() prior to IOCB_DIRECT on
> their own.  Therefore, it does not matter if the file state is DAX when
> the iocb flags are created.
> 
> Also remove io_is_direct() as it is just a simple flag check.
> 
> Reviewed-by: Christoph Hellwig <hch@lst.de>
> Signed-off-by: Ira Weiny <ira.weiny@intel.com>

Looks fine to me,
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>

--D

> 
> ---
> Changes from v8:
> 	Rebase to latest Linus tree
> 
> Changes from v6:
> 	remove io_is_direct() as well.
> 	Remove Reviews since this is quite a bit different.
> 
> Changes from v3:
> 	Reword commit message.
> 	Reordered to be a 'pre-cleanup' patch
> ---
>  drivers/block/loop.c | 6 +++---
>  include/linux/fs.h   | 7 +------
>  2 files changed, 4 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/block/loop.c b/drivers/block/loop.c
> index da693e6a834e..14372df0f354 100644
> --- a/drivers/block/loop.c
> +++ b/drivers/block/loop.c
> @@ -634,8 +634,8 @@ static int do_req_filebacked(struct loop_device *lo, struct request *rq)
>  
>  static inline void loop_update_dio(struct loop_device *lo)
>  {
> -	__loop_update_dio(lo, io_is_direct(lo->lo_backing_file) |
> -			lo->use_dio);
> +	__loop_update_dio(lo, (lo->lo_backing_file->f_flags & O_DIRECT) |
> +				lo->use_dio);
>  }
>  
>  static void loop_reread_partitions(struct loop_device *lo,
> @@ -1028,7 +1028,7 @@ 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_write_cache(lo->lo_queue, true, false);
>  
> -	if (io_is_direct(lo->lo_backing_file) && inode->i_sb->s_bdev) {
> +	if ((lo->lo_backing_file->f_flags & O_DIRECT) && inode->i_sb->s_bdev) {
>  		/* In case of direct I/O, match underlying block size */
>  		unsigned short bsize = bdev_logical_block_size(
>  			inode->i_sb->s_bdev);
> diff --git a/include/linux/fs.h b/include/linux/fs.h
> index 4f6f59b4f22a..a87cc5845a02 100644
> --- a/include/linux/fs.h
> +++ b/include/linux/fs.h
> @@ -3394,11 +3394,6 @@ extern void setattr_copy(struct inode *inode, const struct iattr *attr);
>  
>  extern int file_update_time(struct file *file);
>  
> -static inline bool io_is_direct(struct file *filp)
> -{
> -	return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
> -}
> -
>  static inline bool vma_is_dax(const struct vm_area_struct *vma)
>  {
>  	return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
> @@ -3423,7 +3418,7 @@ static inline int iocb_flags(struct file *file)
>  	int res = 0;
>  	if (file->f_flags & O_APPEND)
>  		res |= IOCB_APPEND;
> -	if (io_is_direct(file))
> +	if (file->f_flags & O_DIRECT)
>  		res |= IOCB_DIRECT;
>  	if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
>  		res |= IOCB_DSYNC;
> -- 
> 2.25.1
> 

Re: [PATCH V9 05/11] fs/xfs: Make DAX mount option a tri-state

[Darrick J. Wong]

On Tue, Apr 21, 2020 at 12:17:47PM -0700, ira.weiny@intel.com wrote:
> From: Ira Weiny <ira.weiny@intel.com>
> 
> As agreed upon[1].  We make the dax mount option a tri-state.  '-o dax'
> continues to operate the same.  We add 'always', 'never', and 'inode'
> (default).
> 
> [1] https://lore.kernel.org/lkml/20200405061945.GA94792@iweiny-DESK2.sc.intel.com/
> 
> Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> 
> ---
> Changes from v8:
> 	Move NEVER bit to 27
> 	Use xfs signature style
> 	use xfs_dax_mode enum
> 
> Changes from v7:
> 	Change to XFS_MOUNT_DAX_NEVER
> 
> Changes from v6:
> 	Use 2 flag bits rather than a field.
> 	change iflag to inode
> 
> Changes from v5:
> 	New Patch
> ---
>  fs/xfs/xfs_mount.h |  1 +
>  fs/xfs/xfs_super.c | 51 ++++++++++++++++++++++++++++++++++++++++++----
>  2 files changed, 48 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
> index f6123fb0113c..37bfb50db809 100644
> --- a/fs/xfs/xfs_mount.h
> +++ b/fs/xfs/xfs_mount.h
> @@ -238,6 +238,7 @@ typedef struct xfs_mount {
>  						   allocator */
>  #define XFS_MOUNT_NOATTR2	(1ULL << 25)	/* disable use of attr2 format */
>  #define XFS_MOUNT_DAX_ALWAYS	(1ULL << 26)
> +#define XFS_MOUNT_DAX_NEVER	(1ULL << 27)
>  
>  /*
>   * Max and min values for mount-option defined I/O
> diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
> index ce169d1c7474..0d0f74786799 100644
> --- a/fs/xfs/xfs_super.c
> +++ b/fs/xfs/xfs_super.c
> @@ -47,6 +47,39 @@ static struct kset *xfs_kset;		/* top-level xfs sysfs dir */
>  static struct xfs_kobj xfs_dbg_kobj;	/* global debug sysfs attrs */
>  #endif
>  
> +enum xfs_dax_mode {
> +	XFS_DAX_INODE = 0,
> +	XFS_DAX_ALWAYS = 1,
> +	XFS_DAX_NEVER = 2,
> +};
> +
> +static void
> +xfs_mount_set_dax_mode(
> +	struct xfs_mount *mp,
> +	enum xfs_dax_mode mode)

Indent between the type name and the variable name, please.

static void
xfs_mount_set_dax_mode(
	struct xfs_mount	*mp,
	enum xfs_dax_mode	mode)

> +{
> +	switch (mode) {
> +		case XFS_DAX_INODE:
> +			mp->m_flags &= ~(XFS_MOUNT_DAX_ALWAYS | XFS_MOUNT_DAX_NEVER);
> +			break;
> +		case XFS_DAX_ALWAYS:
> +			mp->m_flags |= XFS_MOUNT_DAX_ALWAYS;
> +			mp->m_flags &= ~XFS_MOUNT_DAX_NEVER;
> +			break;
> +		case XFS_DAX_NEVER:
> +			mp->m_flags |= XFS_MOUNT_DAX_NEVER;
> +			mp->m_flags &= ~XFS_MOUNT_DAX_ALWAYS;
> +			break;

You can skip one level of indentation, e.g.

	switch (mode) {
	case XFS_DAX_INODE:
		mp->m_flags &= ~(XFS_MOUNT_DAX_ALWAYS | XFS_MOUNT_DAX_NEVER);
		break;

> +	}
> +}
> +
> +static const struct constant_table dax_param_enums[] = {
> +	{"inode",	XFS_DAX_INODE },
> +	{"always",	XFS_DAX_ALWAYS },
> +	{"never",	XFS_DAX_NEVER },
> +	{}
> +};
> +
>  /*
>   * Table driven mount option parser.
>   */
> @@ -59,7 +92,7 @@ enum {
>  	Opt_filestreams, Opt_quota, Opt_noquota, Opt_usrquota, Opt_grpquota,
>  	Opt_prjquota, Opt_uquota, Opt_gquota, Opt_pquota,
>  	Opt_uqnoenforce, Opt_gqnoenforce, Opt_pqnoenforce, Opt_qnoenforce,
> -	Opt_discard, Opt_nodiscard, Opt_dax,
> +	Opt_discard, Opt_nodiscard, Opt_dax, Opt_dax_enum,
>  };
>  
>  static const struct fs_parameter_spec xfs_fs_parameters[] = {
> @@ -103,6 +136,7 @@ static const struct fs_parameter_spec xfs_fs_parameters[] = {
>  	fsparam_flag("discard",		Opt_discard),
>  	fsparam_flag("nodiscard",	Opt_nodiscard),
>  	fsparam_flag("dax",		Opt_dax),
> +	fsparam_enum("dax",		Opt_dax_enum, dax_param_enums),
>  	{}
>  };
>  
> @@ -129,7 +163,6 @@ xfs_fs_show_options(
>  		{ XFS_MOUNT_GRPID,		",grpid" },
>  		{ XFS_MOUNT_DISCARD,		",discard" },
>  		{ XFS_MOUNT_LARGEIO,		",largeio" },
> -		{ XFS_MOUNT_DAX_ALWAYS,		",dax" },
>  		{ 0, NULL }
>  	};
>  	struct xfs_mount	*mp = XFS_M(root->d_sb);
> @@ -185,6 +218,13 @@ xfs_fs_show_options(
>  	if (!(mp->m_qflags & XFS_ALL_QUOTA_ACCT))
>  		seq_puts(m, ",noquota");
>  
> +	if (mp->m_flags & XFS_MOUNT_DAX_ALWAYS)
> +		seq_puts(m, ",dax=always");
> +	else if (mp->m_flags & XFS_MOUNT_DAX_NEVER)
> +		seq_puts(m, ",dax=never");
> +	else
> +		seq_puts(m, ",dax=inode");

No need to report dax=inode, since that's the default.

--D

> +
>  	return 0;
>  }
>  
> @@ -1261,7 +1301,10 @@ xfs_fc_parse_param(
>  		return 0;
>  #ifdef CONFIG_FS_DAX
>  	case Opt_dax:
> -		mp->m_flags |= XFS_MOUNT_DAX_ALWAYS;
> +		xfs_mount_set_dax_mode(mp, XFS_DAX_ALWAYS);
> +		return 0;
> +	case Opt_dax_enum:
> +		xfs_mount_set_dax_mode(mp, result.uint_32);
>  		return 0;
>  #endif
>  	default:
> @@ -1468,7 +1511,7 @@ xfs_fc_fill_super(
>  		if (!rtdev_is_dax && !datadev_is_dax) {
>  			xfs_alert(mp,
>  			"DAX unsupported by block device. Turning off DAX.");
> -			mp->m_flags &= ~XFS_MOUNT_DAX_ALWAYS;
> +			xfs_mount_set_dax_mode(mp, XFS_DAX_NEVER);
>  		}
>  		if (xfs_sb_version_hasreflink(&mp->m_sb)) {
>  			xfs_alert(mp,
> -- 
> 2.25.1
> 

Re: [PATCH V9 08/11] fs: Lift XFS_IDONTCACNE to the VFS layer

[Darrick J. Wong]

> Subject: [PATCH V9 08/11] fs: Lift XFS_IDONTCACNE to the VFS layer

This still has a misspelling in the subject line.

On Tue, Apr 21, 2020 at 12:17:50PM -0700, ira.weiny@intel.com wrote:
> From: Ira Weiny <ira.weiny@intel.com>
> 
> DAX effective mode (S_DAX) changes requires inode eviction.
> 
> XFS has an advisory flag (XFS_IDONTCACHE) to prevent caching of the
> inode if no other additional references are taken.  We lift this flag to
> the VFS layer and change the behavior slightly by allowing the flag to
> remain even if multiple references are taken.
> 
> This will expedite the eviction of inodes to change S_DAX.
> 
> Cc: Al Viro <viro@zeniv.linux.org.uk>
> Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> 
> ---
> Changes from V8:
> 	Remove XFS_IDONTCACHE
> ---
>  fs/xfs/xfs_icache.c | 4 ++--
>  fs/xfs/xfs_inode.h  | 2 +-
>  fs/xfs/xfs_super.c  | 2 +-
>  include/linux/fs.h  | 6 +++++-
>  4 files changed, 9 insertions(+), 5 deletions(-)
> 
> diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
> index 17a0b86fe701..de76f7f60695 100644
> --- a/fs/xfs/xfs_icache.c
> +++ b/fs/xfs/xfs_icache.c
> @@ -477,7 +477,7 @@ xfs_iget_cache_hit(
>  		xfs_ilock(ip, lock_flags);
>  
>  	if (!(flags & XFS_IGET_INCORE))
> -		xfs_iflags_clear(ip, XFS_ISTALE | XFS_IDONTCACHE);
> +		xfs_iflags_clear(ip, XFS_ISTALE);
>  	XFS_STATS_INC(mp, xs_ig_found);
>  
>  	return 0;
> @@ -559,7 +559,7 @@ xfs_iget_cache_miss(
>  	 */
>  	iflags = XFS_INEW;
>  	if (flags & XFS_IGET_DONTCACHE)
> -		iflags |= XFS_IDONTCACHE;
> +		VFS_I(ip)->i_state |= I_DONTCACHE;
>  	ip->i_udquot = NULL;
>  	ip->i_gdquot = NULL;
>  	ip->i_pdquot = NULL;
> diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
> index 83073c883fbf..52b8ee21a0b1 100644
> --- a/fs/xfs/xfs_inode.h
> +++ b/fs/xfs/xfs_inode.h
> @@ -218,7 +218,7 @@ static inline bool xfs_inode_has_cow_data(struct xfs_inode *ip)
>  #define XFS_IFLOCK		(1 << __XFS_IFLOCK_BIT)
>  #define __XFS_IPINNED_BIT	8	 /* wakeup key for zero pin count */
>  #define XFS_IPINNED		(1 << __XFS_IPINNED_BIT)
> -#define XFS_IDONTCACHE		(1 << 9) /* don't cache the inode long term */
> +/* Was XFS_IDONTCACHE 9 */
>  #define XFS_IEOFBLOCKS		(1 << 10)/* has the preallocblocks tag set */

These are incore state flags, you can change XFS_IEOFBLOCKS to (1 << 9).

--D

>  /*
>   * If this unlinked inode is in the middle of recovery, don't let drop_inode
> diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
> index 0d0f74786799..2e165e226e15 100644
> --- a/fs/xfs/xfs_super.c
> +++ b/fs/xfs/xfs_super.c
> @@ -742,7 +742,7 @@ xfs_fs_drop_inode(
>  		return 0;
>  	}
>  
> -	return generic_drop_inode(inode) || (ip->i_flags & XFS_IDONTCACHE);
> +	return generic_drop_inode(inode);
>  }
>  
>  static void
> diff --git a/include/linux/fs.h b/include/linux/fs.h
> index a87cc5845a02..44bd45af760f 100644
> --- a/include/linux/fs.h
> +++ b/include/linux/fs.h
> @@ -2156,6 +2156,8 @@ static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
>   *
>   * I_CREATING		New object's inode in the middle of setting up.
>   *
> + * I_DONTCACHE		Evict inode as soon as it is not used anymore.
> + *
>   * Q: What is the difference between I_WILL_FREE and I_FREEING?
>   */
>  #define I_DIRTY_SYNC		(1 << 0)
> @@ -2178,6 +2180,7 @@ static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
>  #define I_WB_SWITCH		(1 << 13)
>  #define I_OVL_INUSE		(1 << 14)
>  #define I_CREATING		(1 << 15)
> +#define I_DONTCACHE		(1 << 16)
>  
>  #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
>  #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
> @@ -3049,7 +3052,8 @@ extern int inode_needs_sync(struct inode *inode);
>  extern int generic_delete_inode(struct inode *inode);
>  static inline int generic_drop_inode(struct inode *inode)
>  {
> -	return !inode->i_nlink || inode_unhashed(inode);
> +	return !inode->i_nlink || inode_unhashed(inode) ||
> +		(inode->i_state & I_DONTCACHE);
>  }
>  
>  extern struct inode *ilookup5_nowait(struct super_block *sb,
> -- 
> 2.25.1
> 

Re: [PATCH V9 09/11] fs: Introduce DCACHE_DONTCACHE

[Darrick J. Wong]

On Tue, Apr 21, 2020 at 12:17:51PM -0700, ira.weiny@intel.com wrote:
> From: Ira Weiny <ira.weiny@intel.com>
> 
> DCACHE_DONTCACHE indicates a dentry should not be cached on final
> dput().
> 
> Also add a helper function to mark DCACHE_DONTCACHE on all dentries
> pointing to a specific inode when that inode is being set I_DONTCACHE.
> 
> This facilitates dropping dentry references to inodes sooner which
> require eviction to swap S_DAX mode.
> 
> Cc: Al Viro <viro@zeniv.linux.org.uk>
> Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> 
> ---
> Changes from V8:
> 	Update commit message
> 	Use mark_inode_dontcache in XFS
> 	Fix locking...  can't use rcu here.
> 	Change name to mark_inode_dontcache
> ---
>  fs/dcache.c            |  4 ++++
>  fs/inode.c             | 15 +++++++++++++++
>  fs/xfs/xfs_icache.c    |  2 +-
>  include/linux/dcache.h |  2 ++
>  include/linux/fs.h     |  1 +
>  5 files changed, 23 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/dcache.c b/fs/dcache.c
> index b280e07e162b..0030fabab2c4 100644
> --- a/fs/dcache.c
> +++ b/fs/dcache.c
> @@ -647,6 +647,10 @@ static inline bool retain_dentry(struct dentry *dentry)
>  		if (dentry->d_op->d_delete(dentry))
>  			return false;
>  	}
> +
> +	if (unlikely(dentry->d_flags & DCACHE_DONTCACHE))
> +		return false;
> +
>  	/* retain; LRU fodder */
>  	dentry->d_lockref.count--;
>  	if (unlikely(!(dentry->d_flags & DCACHE_LRU_LIST)))
> diff --git a/fs/inode.c b/fs/inode.c
> index 93d9252a00ab..da7f3c4926cd 100644
> --- a/fs/inode.c
> +++ b/fs/inode.c
> @@ -1526,6 +1526,21 @@ int generic_delete_inode(struct inode *inode)
>  }
>  EXPORT_SYMBOL(generic_delete_inode);
>  
> +void mark_inode_dontcache(struct inode *inode)
> +{
> +	struct dentry *de;
> +
> +	spin_lock(&inode->i_lock);
> +	hlist_for_each_entry(de, &inode->i_dentry, d_u.d_alias) {
> +		spin_lock(&de->d_lock);
> +		de->d_flags |= DCACHE_DONTCACHE;
> +		spin_unlock(&de->d_lock);
> +	}
> +	spin_unlock(&inode->i_lock);
> +	inode->i_state |= I_DONTCACHE;
> +}
> +EXPORT_SYMBOL(mark_inode_dontcache);
> +
>  /*
>   * Called when we're dropping the last reference
>   * to an inode.
> diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
> index de76f7f60695..3c8f44477804 100644
> --- a/fs/xfs/xfs_icache.c
> +++ b/fs/xfs/xfs_icache.c
> @@ -559,7 +559,7 @@ xfs_iget_cache_miss(
>  	 */
>  	iflags = XFS_INEW;
>  	if (flags & XFS_IGET_DONTCACHE)
> -		VFS_I(ip)->i_state |= I_DONTCACHE;
> +		mark_inode_dontcache(VFS_I(ip));
>  	ip->i_udquot = NULL;
>  	ip->i_gdquot = NULL;
>  	ip->i_pdquot = NULL;
> diff --git a/include/linux/dcache.h b/include/linux/dcache.h
> index c1488cc84fd9..56b1482d9223 100644
> --- a/include/linux/dcache.h
> +++ b/include/linux/dcache.h
> @@ -177,6 +177,8 @@ struct dentry_operations {
>  
>  #define DCACHE_REFERENCED		0x00000040 /* Recently used, don't discard. */
>  
> +#define DCACHE_DONTCACHE		0x00000080 /* don't cache on final dput() */

"Purge from memory on final dput()"?

--D

> +
>  #define DCACHE_CANT_MOUNT		0x00000100
>  #define DCACHE_GENOCIDE			0x00000200
>  #define DCACHE_SHRINK_LIST		0x00000400
> diff --git a/include/linux/fs.h b/include/linux/fs.h
> index 44bd45af760f..064168ec2e0b 100644
> --- a/include/linux/fs.h
> +++ b/include/linux/fs.h
> @@ -3055,6 +3055,7 @@ static inline int generic_drop_inode(struct inode *inode)
>  	return !inode->i_nlink || inode_unhashed(inode) ||
>  		(inode->i_state & I_DONTCACHE);
>  }
> +extern void mark_inode_dontcache(struct inode *inode);
>  
>  extern struct inode *ilookup5_nowait(struct super_block *sb,
>  		unsigned long hashval, int (*test)(struct inode *, void *),
> -- 
> 2.25.1
> 

Re: [PATCH V9 10/11] fs/xfs: Update xfs_ioctl_setattr_dax_invalidate()

[Darrick J. Wong]

On Tue, Apr 21, 2020 at 12:17:52PM -0700, ira.weiny@intel.com wrote:
> From: Ira Weiny <ira.weiny@intel.com>
> 
> Because of the separation of FS_XFLAG_DAX from S_DAX and the delayed
> setting of S_DAX, data invalidation no longer needs to happen when
> FS_XFLAG_DAX is changed.
> 
> Change xfs_ioctl_setattr_dax_invalidate() to be
> xfs_ioctl_dax_check_set_cache() and alter the code to reflect the new
> functionality.
> 
> Furthermore, we no longer need the locking so we remove the join_flags
> logic.
> 
> Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> 
> ---
> Changes from V8:
> 	Change name of function to xfs_ioctl_dax_check_set_cache()
> 	Update commit message
> 	Fix bit manipulations
> 
> Changes from V7:
> 	Use new flag_inode_dontcache()
> 	Skip don't cache if mount over ride is active.
> 
> Changes from v6:
> 	Fix completely broken implementation and update commit message.
> 	Use the new VFS layer I_DONTCACHE to facilitate inode eviction
> 	and S_DAX changing on drop_caches
> 
> Changes from v5:
> 	New patch
> ---
>  fs/xfs/xfs_ioctl.c | 108 +++++++++------------------------------------
>  1 file changed, 20 insertions(+), 88 deletions(-)
> 
> diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
> index 104495ac187c..b87b571a6748 100644
> --- a/fs/xfs/xfs_ioctl.c
> +++ b/fs/xfs/xfs_ioctl.c
> @@ -1245,64 +1245,26 @@ xfs_ioctl_setattr_xflags(
>  	return 0;
>  }
>  
> -/*
> - * If we are changing DAX flags, we have to ensure the file is clean and any
> - * cached objects in the address space are invalidated and removed. This
> - * requires us to lock out other IO and page faults similar to a truncate
> - * operation. The locks need to be held until the transaction has been committed
> - * so that the cache invalidation is atomic with respect to the DAX flag
> - * manipulation.
> - */
> -static int
> -xfs_ioctl_setattr_dax_invalidate(
> +static void
> +xfs_ioctl_dax_check_set_cache(

That's a ... strange name.  Set cache on what?

Oh, this is the function that sets I_DONTCACHE if an FS_XFLAG_DAX change
could have an immediate effect on S_DAX (assuming no other users).  What
do you think of the following?

	/*
	 * If a change to FS_XFLAG_DAX will result in an change to S_DAX
	 * the next time the incore inode is initialized, set the VFS
	 * I_DONTCACHE flag to try to hurry that along.
	 */
	static void
	xfs_ioctl_try_change_vfs_dax(...)

--D

>  	struct xfs_inode	*ip,
> -	struct fsxattr		*fa,
> -	int			*join_flags)
> +	struct fsxattr		*fa)
>  {
> -	struct inode		*inode = VFS_I(ip);
> -	struct super_block	*sb = inode->i_sb;
> -	int			error;
> -
> -	*join_flags = 0;
> -
> -	/*
> -	 * It is only valid to set the DAX flag on regular files and
> -	 * directories on filesystems where the block size is equal to the page
> -	 * size. On directories it serves as an inherited hint so we don't
> -	 * have to check the device for dax support or flush pagecache.
> -	 */
> -	if (fa->fsx_xflags & FS_XFLAG_DAX) {
> -		struct xfs_buftarg	*target = xfs_inode_buftarg(ip);
> -
> -		if (!bdev_dax_supported(target->bt_bdev, sb->s_blocksize))
> -			return -EINVAL;
> -	}
> -
> -	/* If the DAX state is not changing, we have nothing to do here. */
> -	if ((fa->fsx_xflags & FS_XFLAG_DAX) && IS_DAX(inode))
> -		return 0;
> -	if (!(fa->fsx_xflags & FS_XFLAG_DAX) && !IS_DAX(inode))
> -		return 0;
> +	struct xfs_mount	*mp = ip->i_mount;
> +	struct inode            *inode = VFS_I(ip);
>  
>  	if (S_ISDIR(inode->i_mode))
> -		return 0;
> -
> -	/* lock, flush and invalidate mapping in preparation for flag change */
> -	xfs_ilock(ip, XFS_MMAPLOCK_EXCL | XFS_IOLOCK_EXCL);
> -	error = filemap_write_and_wait(inode->i_mapping);
> -	if (error)
> -		goto out_unlock;
> -	error = invalidate_inode_pages2(inode->i_mapping);
> -	if (error)
> -		goto out_unlock;
> -
> -	*join_flags = XFS_MMAPLOCK_EXCL | XFS_IOLOCK_EXCL;
> -	return 0;
> +		return;
>  
> -out_unlock:
> -	xfs_iunlock(ip, XFS_MMAPLOCK_EXCL | XFS_IOLOCK_EXCL);
> -	return error;
> +	if ((mp->m_flags & XFS_MOUNT_DAX_ALWAYS) ||
> +	    (mp->m_flags & XFS_MOUNT_DAX_NEVER))
> +		return;
>  
> +	if (((fa->fsx_xflags & FS_XFLAG_DAX) &&
> +	    !(ip->i_d.di_flags2 & XFS_DIFLAG2_DAX)) ||
> +	    (!(fa->fsx_xflags & FS_XFLAG_DAX) &&
> +	     (ip->i_d.di_flags2 & XFS_DIFLAG2_DAX)))
> +		mark_inode_dontcache(inode);
>  }
>  
>  /*
> @@ -1310,17 +1272,10 @@ xfs_ioctl_setattr_dax_invalidate(
>   * have permission to do so. On success, return a clean transaction and the
>   * inode locked exclusively ready for further operation specific checks. On
>   * failure, return an error without modifying or locking the inode.
> - *
> - * The inode might already be IO locked on call. If this is the case, it is
> - * indicated in @join_flags and we take full responsibility for ensuring they
> - * are unlocked from now on. Hence if we have an error here, we still have to
> - * unlock them. Otherwise, once they are joined to the transaction, they will
> - * be unlocked on commit/cancel.
>   */
>  static struct xfs_trans *
>  xfs_ioctl_setattr_get_trans(
> -	struct xfs_inode	*ip,
> -	int			join_flags)
> +	struct xfs_inode	*ip)
>  {
>  	struct xfs_mount	*mp = ip->i_mount;
>  	struct xfs_trans	*tp;
> @@ -1337,8 +1292,7 @@ xfs_ioctl_setattr_get_trans(
>  		goto out_unlock;
>  
>  	xfs_ilock(ip, XFS_ILOCK_EXCL);
> -	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL | join_flags);
> -	join_flags = 0;
> +	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
>  
>  	/*
>  	 * CAP_FOWNER overrides the following restrictions:
> @@ -1359,8 +1313,6 @@ xfs_ioctl_setattr_get_trans(
>  out_cancel:
>  	xfs_trans_cancel(tp);
>  out_unlock:
> -	if (join_flags)
> -		xfs_iunlock(ip, join_flags);
>  	return ERR_PTR(error);
>  }
>  
> @@ -1486,7 +1438,6 @@ xfs_ioctl_setattr(
>  	struct xfs_dquot	*pdqp = NULL;
>  	struct xfs_dquot	*olddquot = NULL;
>  	int			code;
> -	int			join_flags = 0;
>  
>  	trace_xfs_ioctl_setattr(ip);
>  
> @@ -1510,18 +1461,9 @@ xfs_ioctl_setattr(
>  			return code;
>  	}
>  
> -	/*
> -	 * Changing DAX config may require inode locking for mapping
> -	 * invalidation. These need to be held all the way to transaction commit
> -	 * or cancel time, so need to be passed through to
> -	 * xfs_ioctl_setattr_get_trans() so it can apply them to the join call
> -	 * appropriately.
> -	 */
> -	code = xfs_ioctl_setattr_dax_invalidate(ip, fa, &join_flags);
> -	if (code)
> -		goto error_free_dquots;
> +	xfs_ioctl_dax_check_set_cache(ip, fa);
>  
> -	tp = xfs_ioctl_setattr_get_trans(ip, join_flags);
> +	tp = xfs_ioctl_setattr_get_trans(ip);
>  	if (IS_ERR(tp)) {
>  		code = PTR_ERR(tp);
>  		goto error_free_dquots;
> @@ -1651,7 +1593,6 @@ xfs_ioc_setxflags(
>  	struct fsxattr		fa;
>  	struct fsxattr		old_fa;
>  	unsigned int		flags;
> -	int			join_flags = 0;
>  	int			error;
>  
>  	if (copy_from_user(&flags, arg, sizeof(flags)))
> @@ -1668,18 +1609,9 @@ xfs_ioc_setxflags(
>  	if (error)
>  		return error;
>  
> -	/*
> -	 * Changing DAX config may require inode locking for mapping
> -	 * invalidation. These need to be held all the way to transaction commit
> -	 * or cancel time, so need to be passed through to
> -	 * xfs_ioctl_setattr_get_trans() so it can apply them to the join call
> -	 * appropriately.
> -	 */
> -	error = xfs_ioctl_setattr_dax_invalidate(ip, &fa, &join_flags);
> -	if (error)
> -		goto out_drop_write;
> +	xfs_ioctl_dax_check_set_cache(ip, &fa);
>  
> -	tp = xfs_ioctl_setattr_get_trans(ip, join_flags);
> +	tp = xfs_ioctl_setattr_get_trans(ip);
>  	if (IS_ERR(tp)) {
>  		error = PTR_ERR(tp);
>  		goto out_drop_write;
> -- 
> 2.25.1
> 

Re: [PATCH V9 11/11] Documentation/dax: Update Usage section

[Darrick J. Wong]

On Tue, Apr 21, 2020 at 12:17:53PM -0700, ira.weiny@intel.com wrote:
> From: Ira Weiny <ira.weiny@intel.com>
> 
> Update the Usage section to reflect the new individual dax selection
> functionality.
> 
> Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> 
> ---
> Changes from V8:
> 	Updates from Darrick
> 
> Changes from V7:
> 	Cleanups/clarifications from Darrick and Dan
> 
> Changes from V6:
> 	Update to allow setting FS_XFLAG_DAX any time.
> 	Update with list of behaviors from Darrick
> 	https://lore.kernel.org/lkml/20200409165927.GD6741@magnolia/
> 
> Changes from V5:
> 	Update to reflect the agreed upon semantics
> 	https://lore.kernel.org/lkml/20200405061945.GA94792@iweiny-DESK2.sc.intel.com/
> ---
>  Documentation/filesystems/dax.txt | 164 +++++++++++++++++++++++++++++-
>  1 file changed, 161 insertions(+), 3 deletions(-)
> 
> diff --git a/Documentation/filesystems/dax.txt b/Documentation/filesystems/dax.txt
> index 679729442fd2..8f4ab08be715 100644
> --- a/Documentation/filesystems/dax.txt
> +++ b/Documentation/filesystems/dax.txt
> @@ -17,11 +17,169 @@ For file mappings, the storage device is mapped directly into userspace.
>  Usage
>  -----
>  
> -If you have a block device which supports DAX, you can make a filesystem
> +If you have a block device which supports DAX, you can make a file system
>  on it as usual.  The DAX code currently only supports files with a block
>  size equal to your kernel's PAGE_SIZE, so you may need to specify a block
> -size when creating the filesystem.  When mounting it, use the "-o dax"
> -option on the command line or add 'dax' to the options in /etc/fstab.
> +size when creating the file system.
> +
> +Currently 3 filesystems support DAX: ext2, ext4 and xfs.  Enabling DAX on them
> +is different.
> +
> +Enabling DAX on ext4 and ext2
> +-----------------------------
> +
> +When mounting the filesystem, use the "-o dax" option on the command line or
> +add 'dax' to the options in /etc/fstab.  This works to enable DAX on all files
> +within the filesystem.  It is equivalent to the '-o dax=always' behavior below.
> +
> +
> +Enabling DAX on xfs
> +-------------------
> +
> +Summary
> +-------
> +
> + 1. There exists an in-kernel file access mode flag S_DAX that corresponds to
> +    the statx flag STATX_ATTR_DAX.  See the manpage for statx(2) for details
> +    about this access mode.
> +
> + 2. There exists an advisory file inode flag FS_XFLAG_DAX that is
> +    inherited from the parent directory FS_XFLAG_DAX inode flag at file
> +    creation time.  This advisory flag can be set or cleared at any
> +    time, but doing so does not immediately affect the S_DAX state.
> +
> +    Unless overridden by mount options (see (3)), if FS_XFLAG_DAX is set
> +    and the fs is on pmem then it will enable S_DAX at inode load time;
> +    if FS_XFLAG_DAX is not set, it will not enable S_DAX.
> +
> + 3. There exists a dax= mount option.
> +
> +    "-o dax=never"  means "never set S_DAX, ignore FS_XFLAG_DAX."
> +
> +    "-o dax=always" means "always set S_DAX (at least on pmem),
> +                    and ignore FS_XFLAG_DAX."
> +
> +    "-o dax"        is an alias for "dax=always".
> +
> +    "-o dax=inode"  means "follow FS_XFLAG_DAX" and is the default.
> +
> + 4. There exists an advisory directory inode flag FS_XFLAG_DAX that can
> +    be set or cleared at any time.  The flag state is inherited by any files or
> +    subdirectories when they are created within that directory.
> +
> + 5. Programs that require a specific file access mode (DAX or not DAX)
> +    can do one of the following:
> +
> +    (a) Create files in directories that the FS_XFLAG_DAX flag set as
> +        needed; or
> +
> +    (b) Have the administrator set an override via mount option; or
> +
> +    (c) Set or clear the file's FS_XFLAG_DAX flag as needed.  Programs
> +        must then cause the kernel to evict the inode from memory.  This
> +        can be done by:
> +
> +        i>  Closing the file and re-opening the file and using statx to
> +            see if the fs has changed the S_DAX flag; and
> +
> +        ii> If the file still does not have the desired S_DAX access
> +            mode, either unmount and remount the filesystem, or close
> +            the file and use drop_caches.
> +
> + 6. It is expected that users who want to squeeze every last bit of performance
> +    out of the particular rough and tumble bits of their storage will also be
> +    exposed to the difficulties of what happens when the operating system can't
> +    totally virtualize those hardware capabilities.  DAX is such a feature.
> +
> +
> +Details
> +-------
> +
> +There are 2 per-file dax flags.  One is a physical inode setting (FS_XFLAG_DAX)
> +and the other a currently enabled state (S_DAX).
> +
> +FS_XFLAG_DAX is maintained, on disk, on individual inodes.  It is preserved
> +within the file system.  This 'physical' config setting can be set using an
> +ioctl and/or an application such as "xfs_io -c 'chattr [-+]x'".  Files and
> +directories automatically inherit FS_XFLAG_DAX from their parent directory
> +_when_ _created_.  Therefore, setting FS_XFLAG_DAX at directory creation time
> +can be used to set a default behavior for an entire sub-tree.  (Doing so on the
> +root directory acts to set a default for the entire file system.)
> +
> +To clarify inheritance here are 3 examples:
> +
> +Example A:
> +
> +mkdir -p a/b/c
> +xfs_io 'chattr +x' a
> +mkdir a/b/c/d
> +mkdir a/e
> +
> +	dax: a,e
> +	no dax: b,c,d
> +
> +Example B:
> +
> +mkdir a
> +xfs_io 'chattr +x' a
> +mkdir -p a/b/c/d
> +
> +	dax: a,b,c,d
> +	no dax:
> +
> +Example C:
> +
> +mkdir -p a/b/c
> +xfs_io 'chattr +x' c
> +mkdir a/b/c/d
> +
> +	dax: c,d
> +	no dax: a,b
> +
> +
> +The current enabled state (S_DAX) is set when a file inode is _loaded_ based on
> +the underlying media support, the value of FS_XFLAG_DAX, and the file systems
> +dax mount option setting.  See below.
> +
> +statx can be used to query S_DAX.  NOTE that a directory will never have S_DAX
> +set and therefore statx will never indicate that S_DAX is set on directories.
> +
> +NOTE: Setting the FS_XFLAG_DAX (specifically or through inheritance) occurs
> +even if the underlying media does not support dax and/or the file system is
> +overridden with a mount option.
> +
> +
> +Overriding FS_XFLAG_DAX (dax= mount option)
> +-------------------------------------------
> +
> +There exists a dax mount option.  Using the mount option does not change the
> +physical configured state of individual files but overrides the S_DAX operating
> +state when inodes are loaded.
> +
> +Given underlying media support, the dax mount option is a tri-state option
> +(never, always, inode) with the following meanings:
> +
> +   "-o dax=never" means "never set S_DAX, ignore FS_XFLAG_DAX"
> +   "-o dax=always" means "always set S_DAX, ignore FS_XFLAG_DAX"
> +        "-o dax" by itself means "dax=always" to remain compatible with older
> +	         kernels
> +   "-o dax=inode" means "follow FS_XFLAG_DAX"
> +
> +The default state is 'inode'.  Given underlying media support, the following
> +algorithm is used to determine the effective mode of the file S_DAX on a
> +capable device.
> +
> +	S_DAX = FS_XFLAG_DAX;
> +
> +	if (dax_mount == "always")
> +		S_DAX = true;
> +	else if (dax_mount == "off"

Missing trailing parenthesis here.

> +		S_DAX = false;
> +
> +To reiterate: Setting, and inheritance, continues to affect FS_XFLAG_DAX even
> +while the file system is mounted with a dax override.  However, in-core inode
> +state (S_DAX) will continue to be overridden until the filesystem is remounted
> +with dax=inode and the inode is evicted."

Trailing double-quote here.

Otherwise, this looks good to me.

--D

>  
>  
>  Implementation Tips for Block Driver Writers
> -- 
> 2.25.1
> 

Re: [PATCH V9 05/11] fs/xfs: Make DAX mount option a tri-state

[Ira Weiny]

On Tue, Apr 21, 2020 at 01:16:41PM -0700, Darrick J. Wong wrote:
> On Tue, Apr 21, 2020 at 12:17:47PM -0700, ira.weiny@intel.com wrote:
> > From: Ira Weiny <ira.weiny@intel.com>
> > 
> > As agreed upon[1].  We make the dax mount option a tri-state.  '-o dax'
> > continues to operate the same.  We add 'always', 'never', and 'inode'
> > (default).
> > 
> > [1] https://lore.kernel.org/lkml/20200405061945.GA94792@iweiny-DESK2.sc.intel.com/
> > 
> > Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> > 
> > ---
> > Changes from v8:
> > 	Move NEVER bit to 27
> > 	Use xfs signature style
> > 	use xfs_dax_mode enum
> > 
> > Changes from v7:
> > 	Change to XFS_MOUNT_DAX_NEVER
> > 
> > Changes from v6:
> > 	Use 2 flag bits rather than a field.
> > 	change iflag to inode
> > 
> > Changes from v5:
> > 	New Patch
> > ---
> >  fs/xfs/xfs_mount.h |  1 +
> >  fs/xfs/xfs_super.c | 51 ++++++++++++++++++++++++++++++++++++++++++----
> >  2 files changed, 48 insertions(+), 4 deletions(-)
> > 
> > diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
> > index f6123fb0113c..37bfb50db809 100644
> > --- a/fs/xfs/xfs_mount.h
> > +++ b/fs/xfs/xfs_mount.h
> > @@ -238,6 +238,7 @@ typedef struct xfs_mount {
> >  						   allocator */
> >  #define XFS_MOUNT_NOATTR2	(1ULL << 25)	/* disable use of attr2 format */
> >  #define XFS_MOUNT_DAX_ALWAYS	(1ULL << 26)
> > +#define XFS_MOUNT_DAX_NEVER	(1ULL << 27)
> >  
> >  /*
> >   * Max and min values for mount-option defined I/O
> > diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
> > index ce169d1c7474..0d0f74786799 100644
> > --- a/fs/xfs/xfs_super.c
> > +++ b/fs/xfs/xfs_super.c
> > @@ -47,6 +47,39 @@ static struct kset *xfs_kset;		/* top-level xfs sysfs dir */
> >  static struct xfs_kobj xfs_dbg_kobj;	/* global debug sysfs attrs */
> >  #endif
> >  
> > +enum xfs_dax_mode {
> > +	XFS_DAX_INODE = 0,
> > +	XFS_DAX_ALWAYS = 1,
> > +	XFS_DAX_NEVER = 2,
> > +};
> > +
> > +static void
> > +xfs_mount_set_dax_mode(
> > +	struct xfs_mount *mp,
> > +	enum xfs_dax_mode mode)
> 
> Indent between the type name and the variable name, please.
> 
> static void
> xfs_mount_set_dax_mode(
> 	struct xfs_mount	*mp,
> 	enum xfs_dax_mode	mode)

Done.

> 
> > +{
> > +	switch (mode) {
> > +		case XFS_DAX_INODE:
> > +			mp->m_flags &= ~(XFS_MOUNT_DAX_ALWAYS | XFS_MOUNT_DAX_NEVER);
> > +			break;
> > +		case XFS_DAX_ALWAYS:
> > +			mp->m_flags |= XFS_MOUNT_DAX_ALWAYS;
> > +			mp->m_flags &= ~XFS_MOUNT_DAX_NEVER;
> > +			break;
> > +		case XFS_DAX_NEVER:
> > +			mp->m_flags |= XFS_MOUNT_DAX_NEVER;
> > +			mp->m_flags &= ~XFS_MOUNT_DAX_ALWAYS;
> > +			break;
> 
> You can skip one level of indentation, e.g.
> 
> 	switch (mode) {
> 	case XFS_DAX_INODE:
> 		mp->m_flags &= ~(XFS_MOUNT_DAX_ALWAYS | XFS_MOUNT_DAX_NEVER);
> 		break;

done.

> 
> > +	}
> > +}
> > +
> > +static const struct constant_table dax_param_enums[] = {
> > +	{"inode",	XFS_DAX_INODE },
> > +	{"always",	XFS_DAX_ALWAYS },
> > +	{"never",	XFS_DAX_NEVER },
> > +	{}
> > +};
> > +
> >  /*
> >   * Table driven mount option parser.
> >   */
> > @@ -59,7 +92,7 @@ enum {
> >  	Opt_filestreams, Opt_quota, Opt_noquota, Opt_usrquota, Opt_grpquota,
> >  	Opt_prjquota, Opt_uquota, Opt_gquota, Opt_pquota,
> >  	Opt_uqnoenforce, Opt_gqnoenforce, Opt_pqnoenforce, Opt_qnoenforce,
> > -	Opt_discard, Opt_nodiscard, Opt_dax,
> > +	Opt_discard, Opt_nodiscard, Opt_dax, Opt_dax_enum,
> >  };
> >  
> >  static const struct fs_parameter_spec xfs_fs_parameters[] = {
> > @@ -103,6 +136,7 @@ static const struct fs_parameter_spec xfs_fs_parameters[] = {
> >  	fsparam_flag("discard",		Opt_discard),
> >  	fsparam_flag("nodiscard",	Opt_nodiscard),
> >  	fsparam_flag("dax",		Opt_dax),
> > +	fsparam_enum("dax",		Opt_dax_enum, dax_param_enums),
> >  	{}
> >  };
> >  
> > @@ -129,7 +163,6 @@ xfs_fs_show_options(
> >  		{ XFS_MOUNT_GRPID,		",grpid" },
> >  		{ XFS_MOUNT_DISCARD,		",discard" },
> >  		{ XFS_MOUNT_LARGEIO,		",largeio" },
> > -		{ XFS_MOUNT_DAX_ALWAYS,		",dax" },
> >  		{ 0, NULL }
> >  	};
> >  	struct xfs_mount	*mp = XFS_M(root->d_sb);
> > @@ -185,6 +218,13 @@ xfs_fs_show_options(
> >  	if (!(mp->m_qflags & XFS_ALL_QUOTA_ACCT))
> >  		seq_puts(m, ",noquota");
> >  
> > +	if (mp->m_flags & XFS_MOUNT_DAX_ALWAYS)
> > +		seq_puts(m, ",dax=always");
> > +	else if (mp->m_flags & XFS_MOUNT_DAX_NEVER)
> > +		seq_puts(m, ",dax=never");
> > +	else
> > +		seq_puts(m, ",dax=inode");
> 
> No need to report dax=inode, since that's the default.

done.

thanks,
Ira

> 
> --D
> 
> > +
> >  	return 0;
> >  }
> >  
> > @@ -1261,7 +1301,10 @@ xfs_fc_parse_param(
> >  		return 0;
> >  #ifdef CONFIG_FS_DAX
> >  	case Opt_dax:
> > -		mp->m_flags |= XFS_MOUNT_DAX_ALWAYS;
> > +		xfs_mount_set_dax_mode(mp, XFS_DAX_ALWAYS);
> > +		return 0;
> > +	case Opt_dax_enum:
> > +		xfs_mount_set_dax_mode(mp, result.uint_32);
> >  		return 0;
> >  #endif
> >  	default:
> > @@ -1468,7 +1511,7 @@ xfs_fc_fill_super(
> >  		if (!rtdev_is_dax && !datadev_is_dax) {
> >  			xfs_alert(mp,
> >  			"DAX unsupported by block device. Turning off DAX.");
> > -			mp->m_flags &= ~XFS_MOUNT_DAX_ALWAYS;
> > +			xfs_mount_set_dax_mode(mp, XFS_DAX_NEVER);
> >  		}
> >  		if (xfs_sb_version_hasreflink(&mp->m_sb)) {
> >  			xfs_alert(mp,
> > -- 
> > 2.25.1
> > 

Re: [PATCH V9 08/11] fs: Lift XFS_IDONTCACNE to the VFS layer

[Ira Weiny]

On Tue, Apr 21, 2020 at 01:23:14PM -0700, Darrick J. Wong wrote:
> > Subject: [PATCH V9 08/11] fs: Lift XFS_IDONTCACNE to the VFS layer
> 
> This still has a misspelling in the subject line.

Sorry...  My reliance on spell check is thwarted by macros again!  ;-)

> 
> On Tue, Apr 21, 2020 at 12:17:50PM -0700, ira.weiny@intel.com wrote:
> > From: Ira Weiny <ira.weiny@intel.com>
> > 
> > DAX effective mode (S_DAX) changes requires inode eviction.
> > 
> > XFS has an advisory flag (XFS_IDONTCACHE) to prevent caching of the
> > inode if no other additional references are taken.  We lift this flag to
> > the VFS layer and change the behavior slightly by allowing the flag to
> > remain even if multiple references are taken.
> > 
> > This will expedite the eviction of inodes to change S_DAX.
> > 
> > Cc: Al Viro <viro@zeniv.linux.org.uk>
> > Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> > 
> > ---
> > Changes from V8:
> > 	Remove XFS_IDONTCACHE
> > ---
> >  fs/xfs/xfs_icache.c | 4 ++--
> >  fs/xfs/xfs_inode.h  | 2 +-
> >  fs/xfs/xfs_super.c  | 2 +-
> >  include/linux/fs.h  | 6 +++++-
> >  4 files changed, 9 insertions(+), 5 deletions(-)
> > 
> > diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
> > index 17a0b86fe701..de76f7f60695 100644
> > --- a/fs/xfs/xfs_icache.c
> > +++ b/fs/xfs/xfs_icache.c
> > @@ -477,7 +477,7 @@ xfs_iget_cache_hit(
> >  		xfs_ilock(ip, lock_flags);
> >  
> >  	if (!(flags & XFS_IGET_INCORE))
> > -		xfs_iflags_clear(ip, XFS_ISTALE | XFS_IDONTCACHE);
> > +		xfs_iflags_clear(ip, XFS_ISTALE);
> >  	XFS_STATS_INC(mp, xs_ig_found);
> >  
> >  	return 0;
> > @@ -559,7 +559,7 @@ xfs_iget_cache_miss(
> >  	 */
> >  	iflags = XFS_INEW;
> >  	if (flags & XFS_IGET_DONTCACHE)
> > -		iflags |= XFS_IDONTCACHE;
> > +		VFS_I(ip)->i_state |= I_DONTCACHE;
> >  	ip->i_udquot = NULL;
> >  	ip->i_gdquot = NULL;
> >  	ip->i_pdquot = NULL;
> > diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
> > index 83073c883fbf..52b8ee21a0b1 100644
> > --- a/fs/xfs/xfs_inode.h
> > +++ b/fs/xfs/xfs_inode.h
> > @@ -218,7 +218,7 @@ static inline bool xfs_inode_has_cow_data(struct xfs_inode *ip)
> >  #define XFS_IFLOCK		(1 << __XFS_IFLOCK_BIT)
> >  #define __XFS_IPINNED_BIT	8	 /* wakeup key for zero pin count */
> >  #define XFS_IPINNED		(1 << __XFS_IPINNED_BIT)
> > -#define XFS_IDONTCACHE		(1 << 9) /* don't cache the inode long term */
> > +/* Was XFS_IDONTCACHE 9 */
> >  #define XFS_IEOFBLOCKS		(1 << 10)/* has the preallocblocks tag set */
> 
> These are incore state flags, you can change XFS_IEOFBLOCKS to (1 << 9).

Sounds good changed.

Thanks,
Ira

> 
> --D
> 
> >  /*
> >   * If this unlinked inode is in the middle of recovery, don't let drop_inode
> > diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
> > index 0d0f74786799..2e165e226e15 100644
> > --- a/fs/xfs/xfs_super.c
> > +++ b/fs/xfs/xfs_super.c
> > @@ -742,7 +742,7 @@ xfs_fs_drop_inode(
> >  		return 0;
> >  	}
> >  
> > -	return generic_drop_inode(inode) || (ip->i_flags & XFS_IDONTCACHE);
> > +	return generic_drop_inode(inode);
> >  }
> >  
> >  static void
> > diff --git a/include/linux/fs.h b/include/linux/fs.h
> > index a87cc5845a02..44bd45af760f 100644
> > --- a/include/linux/fs.h
> > +++ b/include/linux/fs.h
> > @@ -2156,6 +2156,8 @@ static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
> >   *
> >   * I_CREATING		New object's inode in the middle of setting up.
> >   *
> > + * I_DONTCACHE		Evict inode as soon as it is not used anymore.
> > + *
> >   * Q: What is the difference between I_WILL_FREE and I_FREEING?
> >   */
> >  #define I_DIRTY_SYNC		(1 << 0)
> > @@ -2178,6 +2180,7 @@ static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
> >  #define I_WB_SWITCH		(1 << 13)
> >  #define I_OVL_INUSE		(1 << 14)
> >  #define I_CREATING		(1 << 15)
> > +#define I_DONTCACHE		(1 << 16)
> >  
> >  #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
> >  #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
> > @@ -3049,7 +3052,8 @@ extern int inode_needs_sync(struct inode *inode);
> >  extern int generic_delete_inode(struct inode *inode);
> >  static inline int generic_drop_inode(struct inode *inode)
> >  {
> > -	return !inode->i_nlink || inode_unhashed(inode);
> > +	return !inode->i_nlink || inode_unhashed(inode) ||
> > +		(inode->i_state & I_DONTCACHE);
> >  }
> >  
> >  extern struct inode *ilookup5_nowait(struct super_block *sb,
> > -- 
> > 2.25.1
> > 

Re: [PATCH V9 09/11] fs: Introduce DCACHE_DONTCACHE

[Ira Weiny]

On Tue, Apr 21, 2020 at 01:25:19PM -0700, Darrick J. Wong wrote:
> On Tue, Apr 21, 2020 at 12:17:51PM -0700, ira.weiny@intel.com wrote:
> > From: Ira Weiny <ira.weiny@intel.com>
> > 
> > DCACHE_DONTCACHE indicates a dentry should not be cached on final
> > dput().
> > 
> > Also add a helper function to mark DCACHE_DONTCACHE on all dentries
> > pointing to a specific inode when that inode is being set I_DONTCACHE.
> > 
> > This facilitates dropping dentry references to inodes sooner which
> > require eviction to swap S_DAX mode.
> > 
> > Cc: Al Viro <viro@zeniv.linux.org.uk>
> > Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> > 
> > ---

[snip]

> > diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
> > index de76f7f60695..3c8f44477804 100644
> > --- a/fs/xfs/xfs_icache.c
> > +++ b/fs/xfs/xfs_icache.c
> > @@ -559,7 +559,7 @@ xfs_iget_cache_miss(
> >  	 */
> >  	iflags = XFS_INEW;
> >  	if (flags & XFS_IGET_DONTCACHE)
> > -		VFS_I(ip)->i_state |= I_DONTCACHE;
> > +		mark_inode_dontcache(VFS_I(ip));
> >  	ip->i_udquot = NULL;
> >  	ip->i_gdquot = NULL;
> >  	ip->i_pdquot = NULL;
> > diff --git a/include/linux/dcache.h b/include/linux/dcache.h
> > index c1488cc84fd9..56b1482d9223 100644
> > --- a/include/linux/dcache.h
> > +++ b/include/linux/dcache.h
> > @@ -177,6 +177,8 @@ struct dentry_operations {
> >  
> >  #define DCACHE_REFERENCED		0x00000040 /* Recently used, don't discard. */
> >  
> > +#define DCACHE_DONTCACHE		0x00000080 /* don't cache on final dput() */
> 
> "Purge from memory on final dput()"?

Sounds good to me,
Ira

> 
> --D
> 

Re: [PATCH V9 10/11] fs/xfs: Update xfs_ioctl_setattr_dax_invalidate()

[Dave Chinner]

On Tue, Apr 21, 2020 at 01:31:40PM -0700, Darrick J. Wong wrote:
> On Tue, Apr 21, 2020 at 12:17:52PM -0700, ira.weiny@intel.com wrote:
> > From: Ira Weiny <ira.weiny@intel.com>
> > 
> > Because of the separation of FS_XFLAG_DAX from S_DAX and the delayed
> > setting of S_DAX, data invalidation no longer needs to happen when
> > FS_XFLAG_DAX is changed.
> > 
> > Change xfs_ioctl_setattr_dax_invalidate() to be
> > xfs_ioctl_dax_check_set_cache() and alter the code to reflect the new
> > functionality.
> > 
> > Furthermore, we no longer need the locking so we remove the join_flags
> > logic.
> > 
> > Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> > 
> > ---
> > Changes from V8:
> > 	Change name of function to xfs_ioctl_dax_check_set_cache()
> > 	Update commit message
> > 	Fix bit manipulations
> > 
> > Changes from V7:
> > 	Use new flag_inode_dontcache()
> > 	Skip don't cache if mount over ride is active.
> > 
> > Changes from v6:
> > 	Fix completely broken implementation and update commit message.
> > 	Use the new VFS layer I_DONTCACHE to facilitate inode eviction
> > 	and S_DAX changing on drop_caches
> > 
> > Changes from v5:
> > 	New patch
> > ---
> >  fs/xfs/xfs_ioctl.c | 108 +++++++++------------------------------------
> >  1 file changed, 20 insertions(+), 88 deletions(-)
> > 
> > diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
> > index 104495ac187c..b87b571a6748 100644
> > --- a/fs/xfs/xfs_ioctl.c
> > +++ b/fs/xfs/xfs_ioctl.c
> > @@ -1245,64 +1245,26 @@ xfs_ioctl_setattr_xflags(
> >  	return 0;
> >  }
> >  
> > -/*
> > - * If we are changing DAX flags, we have to ensure the file is clean and any
> > - * cached objects in the address space are invalidated and removed. This
> > - * requires us to lock out other IO and page faults similar to a truncate
> > - * operation. The locks need to be held until the transaction has been committed
> > - * so that the cache invalidation is atomic with respect to the DAX flag
> > - * manipulation.
> > - */
> > -static int
> > -xfs_ioctl_setattr_dax_invalidate(
> > +static void
> > +xfs_ioctl_dax_check_set_cache(
> 
> That's a ... strange name.  Set cache on what?
> 
> Oh, this is the function that sets I_DONTCACHE if an FS_XFLAG_DAX change
> could have an immediate effect on S_DAX (assuming no other users).  What
> do you think of the following?
> 
> 	/*
> 	 * If a change to FS_XFLAG_DAX will result in an change to S_DAX
> 	 * the next time the incore inode is initialized, set the VFS
> 	 * I_DONTCACHE flag to try to hurry that along.
> 	 */
> 	static void
> 	xfs_ioctl_try_change_vfs_dax(...)

That doesn't seem any better. This is a preparation function now, in
that it can't fail and doesn't change the outcome of the operation
being performed. So, IMO, calling it something like
xfs_ioctl_setattr_prepare_dax() would be a better name for it.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH V9 10/11] fs/xfs: Update xfs_ioctl_setattr_dax_invalidate()

[Ira Weiny]

On Wed, Apr 22, 2020 at 07:30:49AM +1000, Dave Chinner wrote:
> On Tue, Apr 21, 2020 at 01:31:40PM -0700, Darrick J. Wong wrote:
> > On Tue, Apr 21, 2020 at 12:17:52PM -0700, ira.weiny@intel.com wrote:
> > > From: Ira Weiny <ira.weiny@intel.com>
> > > 
> > > Because of the separation of FS_XFLAG_DAX from S_DAX and the delayed
> > > setting of S_DAX, data invalidation no longer needs to happen when
> > > FS_XFLAG_DAX is changed.
> > > 
> > > Change xfs_ioctl_setattr_dax_invalidate() to be
> > > xfs_ioctl_dax_check_set_cache() and alter the code to reflect the new
> > > functionality.
> > > 
> > > Furthermore, we no longer need the locking so we remove the join_flags
> > > logic.
> > > 
> > > Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> > > 
> > > ---
> > > Changes from V8:
> > > 	Change name of function to xfs_ioctl_dax_check_set_cache()
> > > 	Update commit message
> > > 	Fix bit manipulations
> > > 
> > > Changes from V7:
> > > 	Use new flag_inode_dontcache()
> > > 	Skip don't cache if mount over ride is active.
> > > 
> > > Changes from v6:
> > > 	Fix completely broken implementation and update commit message.
> > > 	Use the new VFS layer I_DONTCACHE to facilitate inode eviction
> > > 	and S_DAX changing on drop_caches
> > > 
> > > Changes from v5:
> > > 	New patch
> > > ---
> > >  fs/xfs/xfs_ioctl.c | 108 +++++++++------------------------------------
> > >  1 file changed, 20 insertions(+), 88 deletions(-)
> > > 
> > > diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
> > > index 104495ac187c..b87b571a6748 100644
> > > --- a/fs/xfs/xfs_ioctl.c
> > > +++ b/fs/xfs/xfs_ioctl.c
> > > @@ -1245,64 +1245,26 @@ xfs_ioctl_setattr_xflags(
> > >  	return 0;
> > >  }
> > >  
> > > -/*
> > > - * If we are changing DAX flags, we have to ensure the file is clean and any
> > > - * cached objects in the address space are invalidated and removed. This
> > > - * requires us to lock out other IO and page faults similar to a truncate
> > > - * operation. The locks need to be held until the transaction has been committed
> > > - * so that the cache invalidation is atomic with respect to the DAX flag
> > > - * manipulation.
> > > - */
> > > -static int
> > > -xfs_ioctl_setattr_dax_invalidate(
> > > +static void
> > > +xfs_ioctl_dax_check_set_cache(
> > 
> > That's a ... strange name.  Set cache on what?
> > 
> > Oh, this is the function that sets I_DONTCACHE if an FS_XFLAG_DAX change
> > could have an immediate effect on S_DAX (assuming no other users).  What
> > do you think of the following?
> > 
> > 	/*
> > 	 * If a change to FS_XFLAG_DAX will result in an change to S_DAX
> > 	 * the next time the incore inode is initialized, set the VFS
> > 	 * I_DONTCACHE flag to try to hurry that along.
> > 	 */
> > 	static void
> > 	xfs_ioctl_try_change_vfs_dax(...)
> 
> That doesn't seem any better. This is a preparation function now, in
> that it can't fail and doesn't change the outcome of the operation
> being performed. So, IMO, calling it something like
> xfs_ioctl_setattr_prepare_dax() would be a better name for it.

But it does potentially (after a check) set I_DONTCACHE.

What about?

xfs_ioctl_dax_check_set_dontcache()?

Ira

> 
> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david@fromorbit.com

Re: [PATCH V9 11/11] Documentation/dax: Update Usage section

[Ira Weiny]

On Tue, Apr 21, 2020 at 01:34:23PM -0700, Darrick J. Wong wrote:
> On Tue, Apr 21, 2020 at 12:17:53PM -0700, ira.weiny@intel.com wrote:
> > From: Ira Weiny <ira.weiny@intel.com>
> > 
> > Update the Usage section to reflect the new individual dax selection
> > functionality.
> > 

[snip]

> > +
> > +The default state is 'inode'.  Given underlying media support, the following
> > +algorithm is used to determine the effective mode of the file S_DAX on a
> > +capable device.
> > +
> > +	S_DAX = FS_XFLAG_DAX;
> > +
> > +	if (dax_mount == "always")
> > +		S_DAX = true;
> > +	else if (dax_mount == "off"
> 
> Missing trailing parenthesis here.

Done.

> 
> > +		S_DAX = false;
> > +
> > +To reiterate: Setting, and inheritance, continues to affect FS_XFLAG_DAX even
> > +while the file system is mounted with a dax override.  However, in-core inode
> > +state (S_DAX) will continue to be overridden until the filesystem is remounted
> > +with dax=inode and the inode is evicted."
> 
> Trailing double-quote here.

Fixed done.

> 
> Otherwise, this looks good to me.

thanks,
Ira

Re: [PATCH V9 10/11] fs/xfs: Update xfs_ioctl_setattr_dax_invalidate()

[Dave Chinner]

On Tue, Apr 21, 2020 at 02:43:29PM -0700, Ira Weiny wrote:
> On Wed, Apr 22, 2020 at 07:30:49AM +1000, Dave Chinner wrote:
> > On Tue, Apr 21, 2020 at 01:31:40PM -0700, Darrick J. Wong wrote:
> > > On Tue, Apr 21, 2020 at 12:17:52PM -0700, ira.weiny@intel.com wrote:
> > > > From: Ira Weiny <ira.weiny@intel.com>
> > > > 
> > > > Because of the separation of FS_XFLAG_DAX from S_DAX and the delayed
> > > > setting of S_DAX, data invalidation no longer needs to happen when
> > > > FS_XFLAG_DAX is changed.
> > > > 
> > > > Change xfs_ioctl_setattr_dax_invalidate() to be
> > > > xfs_ioctl_dax_check_set_cache() and alter the code to reflect the new
> > > > functionality.
> > > > 
> > > > Furthermore, we no longer need the locking so we remove the join_flags
> > > > logic.
> > > > 
> > > > Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> > > > 
> > > > ---
> > > > Changes from V8:
> > > > 	Change name of function to xfs_ioctl_dax_check_set_cache()
> > > > 	Update commit message
> > > > 	Fix bit manipulations
> > > > 
> > > > Changes from V7:
> > > > 	Use new flag_inode_dontcache()
> > > > 	Skip don't cache if mount over ride is active.
> > > > 
> > > > Changes from v6:
> > > > 	Fix completely broken implementation and update commit message.
> > > > 	Use the new VFS layer I_DONTCACHE to facilitate inode eviction
> > > > 	and S_DAX changing on drop_caches
> > > > 
> > > > Changes from v5:
> > > > 	New patch
> > > > ---
> > > >  fs/xfs/xfs_ioctl.c | 108 +++++++++------------------------------------
> > > >  1 file changed, 20 insertions(+), 88 deletions(-)
> > > > 
> > > > diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
> > > > index 104495ac187c..b87b571a6748 100644
> > > > --- a/fs/xfs/xfs_ioctl.c
> > > > +++ b/fs/xfs/xfs_ioctl.c
> > > > @@ -1245,64 +1245,26 @@ xfs_ioctl_setattr_xflags(
> > > >  	return 0;
> > > >  }
> > > >  
> > > > -/*
> > > > - * If we are changing DAX flags, we have to ensure the file is clean and any
> > > > - * cached objects in the address space are invalidated and removed. This
> > > > - * requires us to lock out other IO and page faults similar to a truncate
> > > > - * operation. The locks need to be held until the transaction has been committed
> > > > - * so that the cache invalidation is atomic with respect to the DAX flag
> > > > - * manipulation.
> > > > - */
> > > > -static int
> > > > -xfs_ioctl_setattr_dax_invalidate(
> > > > +static void
> > > > +xfs_ioctl_dax_check_set_cache(
> > > 
> > > That's a ... strange name.  Set cache on what?
> > > 
> > > Oh, this is the function that sets I_DONTCACHE if an FS_XFLAG_DAX change
> > > could have an immediate effect on S_DAX (assuming no other users).  What
> > > do you think of the following?
> > > 
> > > 	/*
> > > 	 * If a change to FS_XFLAG_DAX will result in an change to S_DAX
> > > 	 * the next time the incore inode is initialized, set the VFS
> > > 	 * I_DONTCACHE flag to try to hurry that along.
> > > 	 */
> > > 	static void
> > > 	xfs_ioctl_try_change_vfs_dax(...)
> > 
> > That doesn't seem any better. This is a preparation function now, in
> > that it can't fail and doesn't change the outcome of the operation
> > being performed. So, IMO, calling it something like
> > xfs_ioctl_setattr_prepare_dax() would be a better name for it.
> 
> But it does potentially (after a check) set I_DONTCACHE.

That is an implementation detail - it doesn't change the outcome of
the function, the current behaviour of the inode, or the result of
the ioctl....

> What about?
> 
> xfs_ioctl_dax_check_set_dontcache()?

Then we have to rename it again the moment we change it's
functionality. i.e. we have exactly the same problem we have now
where the function name describes the implementation, not the
operational reason the function is being called...

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH V9 10/11] fs/xfs: Update xfs_ioctl_setattr_dax_invalidate()

[Ira Weiny]

On Wed, Apr 22, 2020 at 09:13:43AM +1000, Dave Chinner wrote:
> On Tue, Apr 21, 2020 at 02:43:29PM -0700, Ira Weiny wrote:
> > On Wed, Apr 22, 2020 at 07:30:49AM +1000, Dave Chinner wrote:
> > > On Tue, Apr 21, 2020 at 01:31:40PM -0700, Darrick J. Wong wrote:
> > > > On Tue, Apr 21, 2020 at 12:17:52PM -0700, ira.weiny@intel.com wrote:
> > > > > From: Ira Weiny <ira.weiny@intel.com>

[snip]

> > > > 
> > > > That's a ... strange name.  Set cache on what?
> > > > 
> > > > Oh, this is the function that sets I_DONTCACHE if an FS_XFLAG_DAX change
> > > > could have an immediate effect on S_DAX (assuming no other users).  What
> > > > do you think of the following?
> > > > 
> > > > 	/*
> > > > 	 * If a change to FS_XFLAG_DAX will result in an change to S_DAX
> > > > 	 * the next time the incore inode is initialized, set the VFS
> > > > 	 * I_DONTCACHE flag to try to hurry that along.
> > > > 	 */
> > > > 	static void
> > > > 	xfs_ioctl_try_change_vfs_dax(...)
> > > 
> > > That doesn't seem any better. This is a preparation function now, in
> > > that it can't fail and doesn't change the outcome of the operation
> > > being performed. So, IMO, calling it something like
> > > xfs_ioctl_setattr_prepare_dax() would be a better name for it.
> > 
> > But it does potentially (after a check) set I_DONTCACHE.
> 
> That is an implementation detail - it doesn't change the outcome of
> the function, the current behaviour of the inode, or the result of
> the ioctl....

I'm confused.  This function does potentially flag the inode to not be cached.
How is that not changing the behavior of the inode?

> 
> > What about?
> > 
> > xfs_ioctl_dax_check_set_dontcache()?
> 
> Then we have to rename it again the moment we change it's
> functionality. i.e. we have exactly the same problem we have now
> where the function name describes the implementation, not the
> operational reason the function is being called...

Ok, I see what you are driving at.  You want this function to potentially do
other things and it should 'prepare' the inode for 'dax stuff'.  Is that about
it?

I'm ok with xfs_ioctl_setattr_prepare_dax() then.

Ira

> 
> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david@fromorbit.com

Re: [PATCH V9 09/11] fs: Introduce DCACHE_DONTCACHE

[Al Viro]

On Tue, Apr 21, 2020 at 01:25:19PM -0700, Darrick J. Wong wrote:

> > DCACHE_DONTCACHE indicates a dentry should not be cached on final
> > dput().
> > 
> > Also add a helper function to mark DCACHE_DONTCACHE on all dentries
> > pointing to a specific inode when that inode is being set I_DONTCACHE.
> > 
> > This facilitates dropping dentry references to inodes sooner which
> > require eviction to swap S_DAX mode.

Explain, please.  Questions:

1) does that ever happen to directories?
2) how much trouble do we get if such inode is *NOT* evicted for, say, several
days?

Re: [PATCH V9 09/11] fs: Introduce DCACHE_DONTCACHE

[Ira Weiny]

On Wed, Apr 22, 2020 at 03:34:07AM +0100, Al Viro wrote:
> On Tue, Apr 21, 2020 at 01:25:19PM -0700, Darrick J. Wong wrote:
> 
> > > DCACHE_DONTCACHE indicates a dentry should not be cached on final
> > > dput().
> > > 
> > > Also add a helper function to mark DCACHE_DONTCACHE on all dentries
> > > pointing to a specific inode when that inode is being set I_DONTCACHE.
> > > 
> > > This facilitates dropping dentry references to inodes sooner which
> > > require eviction to swap S_DAX mode.
> 
> Explain, please.  Questions:
> 
> 1) does that ever happen to directories?

Directories never get S_DAX set.  So the eviction only needs to happen on
inodes.  But that can't happen without dentries also dropping their references.

> 2) how much trouble do we get if such inode is *NOT* evicted for, say, several
> days?

No trouble at all.  Users understand that changing the FS_XFLAG_DAX setting
does _not_ immediately result in S_DAX changing.

It is intended that applications requiring a change of mode would flip the
FS_XFLAG_DAX close the file and wait for the eviction (or force it through a
drop cache if they have permission).

Ira

Re: [PATCH V9 02/11] fs: Remove unneeded IS_DAX() check in io_is_direct()

[Jan Kara]

On Tue 21-04-20 12:17:44, ira.weiny@intel.com wrote:
> From: Ira Weiny <ira.weiny@intel.com>
> 
> Remove the check because DAX now has it's own read/write methods and
> file systems which support DAX check IS_DAX() prior to IOCB_DIRECT on
> their own.  Therefore, it does not matter if the file state is DAX when
> the iocb flags are created.
> 
> Also remove io_is_direct() as it is just a simple flag check.
> 
> Reviewed-by: Christoph Hellwig <hch@lst.de>
> Signed-off-by: Ira Weiny <ira.weiny@intel.com>

Looks good to me. You can add:

Reviewed-by: Jan Kara <jack@suse.cz>

								Honza

> 
> ---
> Changes from v8:
> 	Rebase to latest Linus tree
> 
> Changes from v6:
> 	remove io_is_direct() as well.
> 	Remove Reviews since this is quite a bit different.
> 
> Changes from v3:
> 	Reword commit message.
> 	Reordered to be a 'pre-cleanup' patch
> ---
>  drivers/block/loop.c | 6 +++---
>  include/linux/fs.h   | 7 +------
>  2 files changed, 4 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/block/loop.c b/drivers/block/loop.c
> index da693e6a834e..14372df0f354 100644
> --- a/drivers/block/loop.c
> +++ b/drivers/block/loop.c
> @@ -634,8 +634,8 @@ static int do_req_filebacked(struct loop_device *lo, struct request *rq)
>  
>  static inline void loop_update_dio(struct loop_device *lo)
>  {
> -	__loop_update_dio(lo, io_is_direct(lo->lo_backing_file) |
> -			lo->use_dio);
> +	__loop_update_dio(lo, (lo->lo_backing_file->f_flags & O_DIRECT) |
> +				lo->use_dio);
>  }
>  
>  static void loop_reread_partitions(struct loop_device *lo,
> @@ -1028,7 +1028,7 @@ 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_write_cache(lo->lo_queue, true, false);
>  
> -	if (io_is_direct(lo->lo_backing_file) && inode->i_sb->s_bdev) {
> +	if ((lo->lo_backing_file->f_flags & O_DIRECT) && inode->i_sb->s_bdev) {
>  		/* In case of direct I/O, match underlying block size */
>  		unsigned short bsize = bdev_logical_block_size(
>  			inode->i_sb->s_bdev);
> diff --git a/include/linux/fs.h b/include/linux/fs.h
> index 4f6f59b4f22a..a87cc5845a02 100644
> --- a/include/linux/fs.h
> +++ b/include/linux/fs.h
> @@ -3394,11 +3394,6 @@ extern void setattr_copy(struct inode *inode, const struct iattr *attr);
>  
>  extern int file_update_time(struct file *file);
>  
> -static inline bool io_is_direct(struct file *filp)
> -{
> -	return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
> -}
> -
>  static inline bool vma_is_dax(const struct vm_area_struct *vma)
>  {
>  	return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
> @@ -3423,7 +3418,7 @@ static inline int iocb_flags(struct file *file)
>  	int res = 0;
>  	if (file->f_flags & O_APPEND)
>  		res |= IOCB_APPEND;
> -	if (io_is_direct(file))
> +	if (file->f_flags & O_DIRECT)
>  		res |= IOCB_DIRECT;
>  	if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
>  		res |= IOCB_DSYNC;
> -- 
> 2.25.1
> 
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Re: [PATCH V9 08/11] fs: Lift XFS_IDONTCACNE to the VFS layer

[Jan Kara]

On Tue 21-04-20 12:17:50, ira.weiny@intel.com wrote:
> From: Ira Weiny <ira.weiny@intel.com>
> 
> DAX effective mode (S_DAX) changes requires inode eviction.
> 
> XFS has an advisory flag (XFS_IDONTCACHE) to prevent caching of the
> inode if no other additional references are taken.  We lift this flag to
> the VFS layer and change the behavior slightly by allowing the flag to
> remain even if multiple references are taken.
> 
> This will expedite the eviction of inodes to change S_DAX.
> 
> Cc: Al Viro <viro@zeniv.linux.org.uk>
> Signed-off-by: Ira Weiny <ira.weiny@intel.com>

Besides the two nits Darrick had I didn't find any other problem so feel
free to add:

Reviewed-by: Jan Kara <jack@suse.cz>

to the fixed up version.

								Honza


> 
> ---
> Changes from V8:
> 	Remove XFS_IDONTCACHE
> ---
>  fs/xfs/xfs_icache.c | 4 ++--
>  fs/xfs/xfs_inode.h  | 2 +-
>  fs/xfs/xfs_super.c  | 2 +-
>  include/linux/fs.h  | 6 +++++-
>  4 files changed, 9 insertions(+), 5 deletions(-)
> 
> diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
> index 17a0b86fe701..de76f7f60695 100644
> --- a/fs/xfs/xfs_icache.c
> +++ b/fs/xfs/xfs_icache.c
> @@ -477,7 +477,7 @@ xfs_iget_cache_hit(
>  		xfs_ilock(ip, lock_flags);
>  
>  	if (!(flags & XFS_IGET_INCORE))
> -		xfs_iflags_clear(ip, XFS_ISTALE | XFS_IDONTCACHE);
> +		xfs_iflags_clear(ip, XFS_ISTALE);
>  	XFS_STATS_INC(mp, xs_ig_found);
>  
>  	return 0;
> @@ -559,7 +559,7 @@ xfs_iget_cache_miss(
>  	 */
>  	iflags = XFS_INEW;
>  	if (flags & XFS_IGET_DONTCACHE)
> -		iflags |= XFS_IDONTCACHE;
> +		VFS_I(ip)->i_state |= I_DONTCACHE;
>  	ip->i_udquot = NULL;
>  	ip->i_gdquot = NULL;
>  	ip->i_pdquot = NULL;
> diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
> index 83073c883fbf..52b8ee21a0b1 100644
> --- a/fs/xfs/xfs_inode.h
> +++ b/fs/xfs/xfs_inode.h
> @@ -218,7 +218,7 @@ static inline bool xfs_inode_has_cow_data(struct xfs_inode *ip)
>  #define XFS_IFLOCK		(1 << __XFS_IFLOCK_BIT)
>  #define __XFS_IPINNED_BIT	8	 /* wakeup key for zero pin count */
>  #define XFS_IPINNED		(1 << __XFS_IPINNED_BIT)
> -#define XFS_IDONTCACHE		(1 << 9) /* don't cache the inode long term */
> +/* Was XFS_IDONTCACHE 9 */
>  #define XFS_IEOFBLOCKS		(1 << 10)/* has the preallocblocks tag set */
>  /*
>   * If this unlinked inode is in the middle of recovery, don't let drop_inode
> diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
> index 0d0f74786799..2e165e226e15 100644
> --- a/fs/xfs/xfs_super.c
> +++ b/fs/xfs/xfs_super.c
> @@ -742,7 +742,7 @@ xfs_fs_drop_inode(
>  		return 0;
>  	}
>  
> -	return generic_drop_inode(inode) || (ip->i_flags & XFS_IDONTCACHE);
> +	return generic_drop_inode(inode);
>  }
>  
>  static void
> diff --git a/include/linux/fs.h b/include/linux/fs.h
> index a87cc5845a02..44bd45af760f 100644
> --- a/include/linux/fs.h
> +++ b/include/linux/fs.h
> @@ -2156,6 +2156,8 @@ static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
>   *
>   * I_CREATING		New object's inode in the middle of setting up.
>   *
> + * I_DONTCACHE		Evict inode as soon as it is not used anymore.
> + *
>   * Q: What is the difference between I_WILL_FREE and I_FREEING?
>   */
>  #define I_DIRTY_SYNC		(1 << 0)
> @@ -2178,6 +2180,7 @@ static inline void kiocb_clone(struct kiocb *kiocb, struct kiocb *kiocb_src,
>  #define I_WB_SWITCH		(1 << 13)
>  #define I_OVL_INUSE		(1 << 14)
>  #define I_CREATING		(1 << 15)
> +#define I_DONTCACHE		(1 << 16)
>  
>  #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
>  #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
> @@ -3049,7 +3052,8 @@ extern int inode_needs_sync(struct inode *inode);
>  extern int generic_delete_inode(struct inode *inode);
>  static inline int generic_drop_inode(struct inode *inode)
>  {
> -	return !inode->i_nlink || inode_unhashed(inode);
> +	return !inode->i_nlink || inode_unhashed(inode) ||
> +		(inode->i_state & I_DONTCACHE);
>  }
>  
>  extern struct inode *ilookup5_nowait(struct super_block *sb,
> -- 
> 2.25.1
> 
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Re: [PATCH V9 09/11] fs: Introduce DCACHE_DONTCACHE

[Jan Kara]

On Tue 21-04-20 12:17:51, ira.weiny@intel.com wrote:
> From: Ira Weiny <ira.weiny@intel.com>
> 
> DCACHE_DONTCACHE indicates a dentry should not be cached on final
> dput().
> 
> Also add a helper function to mark DCACHE_DONTCACHE on all dentries
> pointing to a specific inode when that inode is being set I_DONTCACHE.
> 
> This facilitates dropping dentry references to inodes sooner which
> require eviction to swap S_DAX mode.
> 
> Cc: Al Viro <viro@zeniv.linux.org.uk>
> Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> 
> ---
> Changes from V8:
> 	Update commit message
> 	Use mark_inode_dontcache in XFS
> 	Fix locking...  can't use rcu here.
> 	Change name to mark_inode_dontcache
> ---
>  fs/dcache.c            |  4 ++++
>  fs/inode.c             | 15 +++++++++++++++
>  fs/xfs/xfs_icache.c    |  2 +-
>  include/linux/dcache.h |  2 ++
>  include/linux/fs.h     |  1 +
>  5 files changed, 23 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/dcache.c b/fs/dcache.c
> index b280e07e162b..0030fabab2c4 100644
> --- a/fs/dcache.c
> +++ b/fs/dcache.c
> @@ -647,6 +647,10 @@ static inline bool retain_dentry(struct dentry *dentry)
>  		if (dentry->d_op->d_delete(dentry))
>  			return false;
>  	}
> +
> +	if (unlikely(dentry->d_flags & DCACHE_DONTCACHE))
> +		return false;
> +
>  	/* retain; LRU fodder */
>  	dentry->d_lockref.count--;
>  	if (unlikely(!(dentry->d_flags & DCACHE_LRU_LIST)))
> diff --git a/fs/inode.c b/fs/inode.c
> index 93d9252a00ab..da7f3c4926cd 100644
> --- a/fs/inode.c
> +++ b/fs/inode.c
> @@ -1526,6 +1526,21 @@ int generic_delete_inode(struct inode *inode)
>  }
>  EXPORT_SYMBOL(generic_delete_inode);
>  
> +void mark_inode_dontcache(struct inode *inode)
> +{
> +	struct dentry *de;
> +
> +	spin_lock(&inode->i_lock);
> +	hlist_for_each_entry(de, &inode->i_dentry, d_u.d_alias) {
> +		spin_lock(&de->d_lock);
> +		de->d_flags |= DCACHE_DONTCACHE;
> +		spin_unlock(&de->d_lock);
> +	}
> +	spin_unlock(&inode->i_lock);
> +	inode->i_state |= I_DONTCACHE;

Modification of i_state should happen under i_lock.

> +}
> +EXPORT_SYMBOL(mark_inode_dontcache);
> +
>  /*
>   * Called when we're dropping the last reference
>   * to an inode.
> diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
> index de76f7f60695..3c8f44477804 100644
> --- a/fs/xfs/xfs_icache.c
> +++ b/fs/xfs/xfs_icache.c
> @@ -559,7 +559,7 @@ xfs_iget_cache_miss(
>  	 */
>  	iflags = XFS_INEW;
>  	if (flags & XFS_IGET_DONTCACHE)
> -		VFS_I(ip)->i_state |= I_DONTCACHE;
> +		mark_inode_dontcache(VFS_I(ip));

And I know here modification of i_state didn't happen under i_lock but
that's a special case because we are just instantiating the inode so it was
not a real issue.

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Re: [PATCH V9 11/11] Documentation/dax: Update Usage section

[Jan Kara]

On Tue 21-04-20 12:17:53, ira.weiny@intel.com wrote:
> From: Ira Weiny <ira.weiny@intel.com>
> 
> Update the Usage section to reflect the new individual dax selection
> functionality.
> 
> Signed-off-by: Ira Weiny <ira.weiny@intel.com>

Looks good to me. You can add:

Reviewed-by: Jan Kara <jack@suse.cz>

								Honza

> 
> ---
> Changes from V8:
> 	Updates from Darrick
> 
> Changes from V7:
> 	Cleanups/clarifications from Darrick and Dan
> 
> Changes from V6:
> 	Update to allow setting FS_XFLAG_DAX any time.
> 	Update with list of behaviors from Darrick
> 	https://lore.kernel.org/lkml/20200409165927.GD6741@magnolia/
> 
> Changes from V5:
> 	Update to reflect the agreed upon semantics
> 	https://lore.kernel.org/lkml/20200405061945.GA94792@iweiny-DESK2.sc.intel.com/
> ---
>  Documentation/filesystems/dax.txt | 164 +++++++++++++++++++++++++++++-
>  1 file changed, 161 insertions(+), 3 deletions(-)
> 
> diff --git a/Documentation/filesystems/dax.txt b/Documentation/filesystems/dax.txt
> index 679729442fd2..8f4ab08be715 100644
> --- a/Documentation/filesystems/dax.txt
> +++ b/Documentation/filesystems/dax.txt
> @@ -17,11 +17,169 @@ For file mappings, the storage device is mapped directly into userspace.
>  Usage
>  -----
>  
> -If you have a block device which supports DAX, you can make a filesystem
> +If you have a block device which supports DAX, you can make a file system
>  on it as usual.  The DAX code currently only supports files with a block
>  size equal to your kernel's PAGE_SIZE, so you may need to specify a block
> -size when creating the filesystem.  When mounting it, use the "-o dax"
> -option on the command line or add 'dax' to the options in /etc/fstab.
> +size when creating the file system.
> +
> +Currently 3 filesystems support DAX: ext2, ext4 and xfs.  Enabling DAX on them
> +is different.
> +
> +Enabling DAX on ext4 and ext2
> +-----------------------------
> +
> +When mounting the filesystem, use the "-o dax" option on the command line or
> +add 'dax' to the options in /etc/fstab.  This works to enable DAX on all files
> +within the filesystem.  It is equivalent to the '-o dax=always' behavior below.
> +
> +
> +Enabling DAX on xfs
> +-------------------
> +
> +Summary
> +-------
> +
> + 1. There exists an in-kernel file access mode flag S_DAX that corresponds to
> +    the statx flag STATX_ATTR_DAX.  See the manpage for statx(2) for details
> +    about this access mode.
> +
> + 2. There exists an advisory file inode flag FS_XFLAG_DAX that is
> +    inherited from the parent directory FS_XFLAG_DAX inode flag at file
> +    creation time.  This advisory flag can be set or cleared at any
> +    time, but doing so does not immediately affect the S_DAX state.
> +
> +    Unless overridden by mount options (see (3)), if FS_XFLAG_DAX is set
> +    and the fs is on pmem then it will enable S_DAX at inode load time;
> +    if FS_XFLAG_DAX is not set, it will not enable S_DAX.
> +
> + 3. There exists a dax= mount option.
> +
> +    "-o dax=never"  means "never set S_DAX, ignore FS_XFLAG_DAX."
> +
> +    "-o dax=always" means "always set S_DAX (at least on pmem),
> +                    and ignore FS_XFLAG_DAX."
> +
> +    "-o dax"        is an alias for "dax=always".
> +
> +    "-o dax=inode"  means "follow FS_XFLAG_DAX" and is the default.
> +
> + 4. There exists an advisory directory inode flag FS_XFLAG_DAX that can
> +    be set or cleared at any time.  The flag state is inherited by any files or
> +    subdirectories when they are created within that directory.
> +
> + 5. Programs that require a specific file access mode (DAX or not DAX)
> +    can do one of the following:
> +
> +    (a) Create files in directories that the FS_XFLAG_DAX flag set as
> +        needed; or
> +
> +    (b) Have the administrator set an override via mount option; or
> +
> +    (c) Set or clear the file's FS_XFLAG_DAX flag as needed.  Programs
> +        must then cause the kernel to evict the inode from memory.  This
> +        can be done by:
> +
> +        i>  Closing the file and re-opening the file and using statx to
> +            see if the fs has changed the S_DAX flag; and
> +
> +        ii> If the file still does not have the desired S_DAX access
> +            mode, either unmount and remount the filesystem, or close
> +            the file and use drop_caches.
> +
> + 6. It is expected that users who want to squeeze every last bit of performance
> +    out of the particular rough and tumble bits of their storage will also be
> +    exposed to the difficulties of what happens when the operating system can't
> +    totally virtualize those hardware capabilities.  DAX is such a feature.
> +
> +
> +Details
> +-------
> +
> +There are 2 per-file dax flags.  One is a physical inode setting (FS_XFLAG_DAX)
> +and the other a currently enabled state (S_DAX).
> +
> +FS_XFLAG_DAX is maintained, on disk, on individual inodes.  It is preserved
> +within the file system.  This 'physical' config setting can be set using an
> +ioctl and/or an application such as "xfs_io -c 'chattr [-+]x'".  Files and
> +directories automatically inherit FS_XFLAG_DAX from their parent directory
> +_when_ _created_.  Therefore, setting FS_XFLAG_DAX at directory creation time
> +can be used to set a default behavior for an entire sub-tree.  (Doing so on the
> +root directory acts to set a default for the entire file system.)
> +
> +To clarify inheritance here are 3 examples:
> +
> +Example A:
> +
> +mkdir -p a/b/c
> +xfs_io 'chattr +x' a
> +mkdir a/b/c/d
> +mkdir a/e
> +
> +	dax: a,e
> +	no dax: b,c,d
> +
> +Example B:
> +
> +mkdir a
> +xfs_io 'chattr +x' a
> +mkdir -p a/b/c/d
> +
> +	dax: a,b,c,d
> +	no dax:
> +
> +Example C:
> +
> +mkdir -p a/b/c
> +xfs_io 'chattr +x' c
> +mkdir a/b/c/d
> +
> +	dax: c,d
> +	no dax: a,b
> +
> +
> +The current enabled state (S_DAX) is set when a file inode is _loaded_ based on
> +the underlying media support, the value of FS_XFLAG_DAX, and the file systems
> +dax mount option setting.  See below.
> +
> +statx can be used to query S_DAX.  NOTE that a directory will never have S_DAX
> +set and therefore statx will never indicate that S_DAX is set on directories.
> +
> +NOTE: Setting the FS_XFLAG_DAX (specifically or through inheritance) occurs
> +even if the underlying media does not support dax and/or the file system is
> +overridden with a mount option.
> +
> +
> +Overriding FS_XFLAG_DAX (dax= mount option)
> +-------------------------------------------
> +
> +There exists a dax mount option.  Using the mount option does not change the
> +physical configured state of individual files but overrides the S_DAX operating
> +state when inodes are loaded.
> +
> +Given underlying media support, the dax mount option is a tri-state option
> +(never, always, inode) with the following meanings:
> +
> +   "-o dax=never" means "never set S_DAX, ignore FS_XFLAG_DAX"
> +   "-o dax=always" means "always set S_DAX, ignore FS_XFLAG_DAX"
> +        "-o dax" by itself means "dax=always" to remain compatible with older
> +	         kernels
> +   "-o dax=inode" means "follow FS_XFLAG_DAX"
> +
> +The default state is 'inode'.  Given underlying media support, the following
> +algorithm is used to determine the effective mode of the file S_DAX on a
> +capable device.
> +
> +	S_DAX = FS_XFLAG_DAX;
> +
> +	if (dax_mount == "always")
> +		S_DAX = true;
> +	else if (dax_mount == "off"
> +		S_DAX = false;
> +
> +To reiterate: Setting, and inheritance, continues to affect FS_XFLAG_DAX even
> +while the file system is mounted with a dax override.  However, in-core inode
> +state (S_DAX) will continue to be overridden until the filesystem is remounted
> +with dax=inode and the inode is evicted."
>  
>  
>  Implementation Tips for Block Driver Writers
> -- 
> 2.25.1
> 
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Re: [PATCH V9 09/11] fs: Introduce DCACHE_DONTCACHE

[Ira Weiny]

On Wed, Apr 22, 2020 at 10:46:47AM +0200, Jan Kara wrote:
> On Tue 21-04-20 12:17:51, ira.weiny@intel.com wrote:
> > From: Ira Weiny <ira.weiny@intel.com>
> > 
> > DCACHE_DONTCACHE indicates a dentry should not be cached on final
> > dput().
> > 
> > Also add a helper function to mark DCACHE_DONTCACHE on all dentries
> > pointing to a specific inode when that inode is being set I_DONTCACHE.
> > 
> > This facilitates dropping dentry references to inodes sooner which
> > require eviction to swap S_DAX mode.
> > 
> > Cc: Al Viro <viro@zeniv.linux.org.uk>
> > Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> > 
> > ---
> > Changes from V8:
> > 	Update commit message
> > 	Use mark_inode_dontcache in XFS
> > 	Fix locking...  can't use rcu here.
> > 	Change name to mark_inode_dontcache
> > ---
> >  fs/dcache.c            |  4 ++++
> >  fs/inode.c             | 15 +++++++++++++++
> >  fs/xfs/xfs_icache.c    |  2 +-
> >  include/linux/dcache.h |  2 ++
> >  include/linux/fs.h     |  1 +
> >  5 files changed, 23 insertions(+), 1 deletion(-)
> > 
> > diff --git a/fs/dcache.c b/fs/dcache.c
> > index b280e07e162b..0030fabab2c4 100644
> > --- a/fs/dcache.c
> > +++ b/fs/dcache.c
> > @@ -647,6 +647,10 @@ static inline bool retain_dentry(struct dentry *dentry)
> >  		if (dentry->d_op->d_delete(dentry))
> >  			return false;
> >  	}
> > +
> > +	if (unlikely(dentry->d_flags & DCACHE_DONTCACHE))
> > +		return false;
> > +
> >  	/* retain; LRU fodder */
> >  	dentry->d_lockref.count--;
> >  	if (unlikely(!(dentry->d_flags & DCACHE_LRU_LIST)))
> > diff --git a/fs/inode.c b/fs/inode.c
> > index 93d9252a00ab..da7f3c4926cd 100644
> > --- a/fs/inode.c
> > +++ b/fs/inode.c
> > @@ -1526,6 +1526,21 @@ int generic_delete_inode(struct inode *inode)
> >  }
> >  EXPORT_SYMBOL(generic_delete_inode);
> >  
> > +void mark_inode_dontcache(struct inode *inode)
> > +{
> > +	struct dentry *de;
> > +
> > +	spin_lock(&inode->i_lock);
> > +	hlist_for_each_entry(de, &inode->i_dentry, d_u.d_alias) {
> > +		spin_lock(&de->d_lock);
> > +		de->d_flags |= DCACHE_DONTCACHE;
> > +		spin_unlock(&de->d_lock);
> > +	}
> > +	spin_unlock(&inode->i_lock);
> > +	inode->i_state |= I_DONTCACHE;
> 
> Modification of i_state should happen under i_lock.

Done.

> 
> > +}
> > +EXPORT_SYMBOL(mark_inode_dontcache);
> > +
> >  /*
> >   * Called when we're dropping the last reference
> >   * to an inode.
> > diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
> > index de76f7f60695..3c8f44477804 100644
> > --- a/fs/xfs/xfs_icache.c
> > +++ b/fs/xfs/xfs_icache.c
> > @@ -559,7 +559,7 @@ xfs_iget_cache_miss(
> >  	 */
> >  	iflags = XFS_INEW;
> >  	if (flags & XFS_IGET_DONTCACHE)
> > -		VFS_I(ip)->i_state |= I_DONTCACHE;
> > +		mark_inode_dontcache(VFS_I(ip));
> 
> And I know here modification of i_state didn't happen under i_lock but
> that's a special case because we are just instantiating the inode so it was
> not a real issue.

Thanks!
Ira

> 
> 								Honza
> -- 
> Jan Kara <jack@suse.com>
> SUSE Labs, CR

Re: [PATCH V9 03/11] fs/stat: Define DAX statx attribute

[Darrick J. Wong]

On Tue, Apr 21, 2020 at 12:17:45PM -0700, ira.weiny@intel.com wrote:
> From: Ira Weiny <ira.weiny@intel.com>
> 
> In order for users to determine if a file is currently operating in DAX
> state (effective DAX).  Define a statx attribute value and set that
> attribute if the effective DAX flag is set.
> 
> To go along with this we propose the following addition to the statx man
> page:
> 
> STATX_ATTR_DAX
> 
> 	The file is in the DAX (cpu direct access) state.  DAX state
> 	attempts to minimize software cache effects for both I/O and
> 	memory mappings of this file.  It requires a file system which
> 	has been configured to support DAX.
> 
> 	DAX generally assumes all accesses are via cpu load / store
> 	instructions which can minimize overhead for small accesses, but
> 	may adversely affect cpu utilization for large transfers.
> 
> 	File I/O is done directly to/from user-space buffers and memory
> 	mapped I/O may be performed with direct memory mappings that
> 	bypass kernel page cache.
> 
> 	While the DAX property tends to result in data being transferred
> 	synchronously, it does not give the same guarantees of O_SYNC
> 	where data and the necessary metadata are transferred together.
> 
> 	A DAX file may support being mapped with the MAP_SYNC flag,
> 	which enables a program to use CPU cache flush instructions to
> 	persist CPU store operations without an explicit fsync(2).  See
> 	mmap(2) for more information.

One thing I hadn't noticed before -- this is a change to userspace API,
so please cc this series to linux-api@vger.kernel.org when you send V10.

Also, I've started to think about commit order sequencing for actually
landing this series.  Usually I try to put vfs and documentation things
before xfs stuff, which means I came up with:

vfs       xfs          I_DONTCACHE
2 3 11    1 4 5 6 7    8 9 10

Note that I separated the DONTCACHE part because it touches VFS
internals, which implies a higher standard of review (aka Al) and I do
not wish to hold up the 2-3-11-1-4-5-6-7 patches if the dontcache part
becomes contentious.

What do you think of that ordering?

(Heck, maybe I'll just put patch 1 in the queue for 5.8 right now...)

--D

> Reviewed-by: Dave Chinner <dchinner@redhat.com>
> Reviewed-by: Jan Kara <jack@suse.cz>
> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
> Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> 
> ---
> Changes from V2:
> 	Update man page text with comments from Darrick, Jan, Dan, and
> 	Dave.
> ---
>  fs/stat.c                 | 3 +++
>  include/uapi/linux/stat.h | 1 +
>  2 files changed, 4 insertions(+)
> 
> diff --git a/fs/stat.c b/fs/stat.c
> index 030008796479..894699c74dde 100644
> --- a/fs/stat.c
> +++ b/fs/stat.c
> @@ -79,6 +79,9 @@ int vfs_getattr_nosec(const struct path *path, struct kstat *stat,
>  	if (IS_AUTOMOUNT(inode))
>  		stat->attributes |= STATX_ATTR_AUTOMOUNT;
>  
> +	if (IS_DAX(inode))
> +		stat->attributes |= STATX_ATTR_DAX;
> +
>  	if (inode->i_op->getattr)
>  		return inode->i_op->getattr(path, stat, request_mask,
>  					    query_flags);
> diff --git a/include/uapi/linux/stat.h b/include/uapi/linux/stat.h
> index ad80a5c885d5..e5f9d5517f6b 100644
> --- a/include/uapi/linux/stat.h
> +++ b/include/uapi/linux/stat.h
> @@ -169,6 +169,7 @@ struct statx {
>  #define STATX_ATTR_ENCRYPTED		0x00000800 /* [I] File requires key to decrypt in fs */
>  #define STATX_ATTR_AUTOMOUNT		0x00001000 /* Dir: Automount trigger */
>  #define STATX_ATTR_VERITY		0x00100000 /* [I] Verity protected file */
> +#define STATX_ATTR_DAX			0x00002000 /* [I] File is DAX */
>  
>  
>  #endif /* _UAPI_LINUX_STAT_H */
> -- 
> 2.25.1
> 

Re: [PATCH V9 03/11] fs/stat: Define DAX statx attribute

[Ira Weiny]

On Wed, Apr 22, 2020 at 09:29:51AM -0700, Darrick J. Wong wrote:
> On Tue, Apr 21, 2020 at 12:17:45PM -0700, ira.weiny@intel.com wrote:
> > From: Ira Weiny <ira.weiny@intel.com>
> > 
> > In order for users to determine if a file is currently operating in DAX
> > state (effective DAX).  Define a statx attribute value and set that
> > attribute if the effective DAX flag is set.
> > 
> > To go along with this we propose the following addition to the statx man
> > page:
> > 
> > STATX_ATTR_DAX
> > 
> > 	The file is in the DAX (cpu direct access) state.  DAX state
> > 	attempts to minimize software cache effects for both I/O and
> > 	memory mappings of this file.  It requires a file system which
> > 	has been configured to support DAX.
> > 
> > 	DAX generally assumes all accesses are via cpu load / store
> > 	instructions which can minimize overhead for small accesses, but
> > 	may adversely affect cpu utilization for large transfers.
> > 
> > 	File I/O is done directly to/from user-space buffers and memory
> > 	mapped I/O may be performed with direct memory mappings that
> > 	bypass kernel page cache.
> > 
> > 	While the DAX property tends to result in data being transferred
> > 	synchronously, it does not give the same guarantees of O_SYNC
> > 	where data and the necessary metadata are transferred together.
> > 
> > 	A DAX file may support being mapped with the MAP_SYNC flag,
> > 	which enables a program to use CPU cache flush instructions to
> > 	persist CPU store operations without an explicit fsync(2).  See
> > 	mmap(2) for more information.
> 
> One thing I hadn't noticed before -- this is a change to userspace API,
> so please cc this series to linux-api@vger.kernel.org when you send V10.

Right!  Glad you caught me on this because I was just preparing to send V10.

Is there someone I could directly mail who needs to look at this?  I guess I
thought we had the important FS people involved for this type of API change.
:-/

> 
> Also, I've started to think about commit order sequencing for actually
> landing this series.  Usually I try to put vfs and documentation things
> before xfs stuff, which means I came up with:
> 
> vfs       xfs          I_DONTCACHE
> 2 3 11    1 4 5 6 7    8 9 10
> 
> Note that I separated the DONTCACHE part because it touches VFS
> internals, which implies a higher standard of review (aka Al) and I do
> not wish to hold up the 2-3-11-1-4-5-6-7 patches if the dontcache part
> becomes contentious.
> 
> What do you think of that ordering?

I think 1 stands on it's own separate from this series...  so I would keep it
first.  Moving Documentation up is easy.

I've changed to this order...

prelim   vfs       xfs        I_DONTCACHE
1        2 3 11    4 5 6 7    8 9 10

Which is pretty much the same now that I look at it!  ;-)

> 
> (Heck, maybe I'll just put patch 1 in the queue for 5.8 right now...)

IMHO, I think 1 and 2 can go.

While patch 2 is in the VFS layer it is very much a DAX thing.  Jan and
Christoph approved it.  I think even Dave approved the version before I
removed io_is_direct() but I don't recall now.

Dan and I also discussed it internally when I first found the issue.  So I'm
very confident in it!  :-D

Unfortunately, 3 and 10 are the critical pieces to the feature.  So we could
move 3 out later after 8 and 9 are approved.  But I don't think it buys us
much to have the tri-state go in without the rest.

Ira

> 
> --D
> 
> > Reviewed-by: Dave Chinner <dchinner@redhat.com>
> > Reviewed-by: Jan Kara <jack@suse.cz>
> > Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
> > Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> > 
> > ---
> > Changes from V2:
> > 	Update man page text with comments from Darrick, Jan, Dan, and
> > 	Dave.
> > ---
> >  fs/stat.c                 | 3 +++
> >  include/uapi/linux/stat.h | 1 +
> >  2 files changed, 4 insertions(+)
> > 
> > diff --git a/fs/stat.c b/fs/stat.c
> > index 030008796479..894699c74dde 100644
> > --- a/fs/stat.c
> > +++ b/fs/stat.c
> > @@ -79,6 +79,9 @@ int vfs_getattr_nosec(const struct path *path, struct kstat *stat,
> >  	if (IS_AUTOMOUNT(inode))
> >  		stat->attributes |= STATX_ATTR_AUTOMOUNT;
> >  
> > +	if (IS_DAX(inode))
> > +		stat->attributes |= STATX_ATTR_DAX;
> > +
> >  	if (inode->i_op->getattr)
> >  		return inode->i_op->getattr(path, stat, request_mask,
> >  					    query_flags);
> > diff --git a/include/uapi/linux/stat.h b/include/uapi/linux/stat.h
> > index ad80a5c885d5..e5f9d5517f6b 100644
> > --- a/include/uapi/linux/stat.h
> > +++ b/include/uapi/linux/stat.h
> > @@ -169,6 +169,7 @@ struct statx {
> >  #define STATX_ATTR_ENCRYPTED		0x00000800 /* [I] File requires key to decrypt in fs */
> >  #define STATX_ATTR_AUTOMOUNT		0x00001000 /* Dir: Automount trigger */
> >  #define STATX_ATTR_VERITY		0x00100000 /* [I] Verity protected file */
> > +#define STATX_ATTR_DAX			0x00002000 /* [I] File is DAX */
> >  
> >  
> >  #endif /* _UAPI_LINUX_STAT_H */
> > -- 
> > 2.25.1
> > 

Re: [PATCH V9 03/11] fs/stat: Define DAX statx attribute

[Jan Kara]

On Wed 22-04-20 11:51:21, Ira Weiny wrote:
> On Wed, Apr 22, 2020 at 09:29:51AM -0700, Darrick J. Wong wrote:
> > On Tue, Apr 21, 2020 at 12:17:45PM -0700, ira.weiny@intel.com wrote:
> > > From: Ira Weiny <ira.weiny@intel.com>
> > > 
> > > In order for users to determine if a file is currently operating in DAX
> > > state (effective DAX).  Define a statx attribute value and set that
> > > attribute if the effective DAX flag is set.
> > > 
> > > To go along with this we propose the following addition to the statx man
> > > page:
> > > 
> > > STATX_ATTR_DAX
> > > 
> > > 	The file is in the DAX (cpu direct access) state.  DAX state
> > > 	attempts to minimize software cache effects for both I/O and
> > > 	memory mappings of this file.  It requires a file system which
> > > 	has been configured to support DAX.
> > > 
> > > 	DAX generally assumes all accesses are via cpu load / store
> > > 	instructions which can minimize overhead for small accesses, but
> > > 	may adversely affect cpu utilization for large transfers.
> > > 
> > > 	File I/O is done directly to/from user-space buffers and memory
> > > 	mapped I/O may be performed with direct memory mappings that
> > > 	bypass kernel page cache.
> > > 
> > > 	While the DAX property tends to result in data being transferred
> > > 	synchronously, it does not give the same guarantees of O_SYNC
> > > 	where data and the necessary metadata are transferred together.
> > > 
> > > 	A DAX file may support being mapped with the MAP_SYNC flag,
> > > 	which enables a program to use CPU cache flush instructions to
> > > 	persist CPU store operations without an explicit fsync(2).  See
> > > 	mmap(2) for more information.
> > 
> > One thing I hadn't noticed before -- this is a change to userspace API,
> > so please cc this series to linux-api@vger.kernel.org when you send V10.
> 
> Right!  Glad you caught me on this because I was just preparing to send V10.
> 
> Is there someone I could directly mail who needs to look at this?  I guess I
> thought we had the important FS people involved for this type of API change.
> :-/

I believe we have all the important people here. But linux-api is a general
fallback list where people reviewing API changes linger. So when changing
user facing API, it is good to CC this list.

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR