Searched +hist:5 +hist:accdf82 (Results 1 - 2 of 2) sorted by relevance

/linux-master/fs/
H A Dsuper.cdiff 7366f8b6 Sat Nov 04 08:00:13 MDT 2023 Christian Brauner <brauner@kernel.org> fs: handle freezing from multiple devices

Before [1] freezing a filesystems through the block layer only worked
for the main block device as the owning superblock of additional block
devices could not be found. Any filesystem that made use of multiple
block devices would only be freezable via it's main block device.

For example, consider xfs over device mapper with /dev/dm-0 as main
block device and /dev/dm-1 as external log device. Two freeze requests
before [1]:

(1) dmsetup suspend /dev/dm-0 on the main block device

bdev_freeze(dm-0)
-> dm-0->bd_fsfreeze_count++
-> freeze_super(xfs-sb)

The owning superblock is found and the filesystem gets frozen.
Returns 0.

(2) dmsetup suspend /dev/dm-1 on the log device

bdev_freeze(dm-1)
-> dm-1->bd_fsfreeze_count++

The owning superblock isn't found and only the block device freeze
count is incremented. Returns 0.

Two freeze requests after [1]:

(1') dmsetup suspend /dev/dm-0 on the main block device

bdev_freeze(dm-0)
-> dm-0->bd_fsfreeze_count++
-> freeze_super(xfs-sb)

The owning superblock is found and the filesystem gets frozen.
Returns 0.

(2') dmsetup suspend /dev/dm-1 on the log device

bdev_freeze(dm-0)
-> dm-0->bd_fsfreeze_count++
-> freeze_super(xfs-sb)

The owning superblock is found and the filesystem gets frozen.
Returns -EBUSY.

When (2') is called we initiate a freeze from another block device of
the same superblock. So we increment the bd_fsfreeze_count for that
additional block device. But we now also find the owning superblock for
additional block devices and call freeze_super() again which reports
-EBUSY.

This can be reproduced through xfstests via:

mkfs.xfs -f -m crc=1,reflink=1,rmapbt=1, -i sparse=1 -lsize=1g,logdev=/dev/nvme1n1p4 /dev/nvme1n1p3
mkfs.xfs -f -m crc=1,reflink=1,rmapbt=1, -i sparse=1 -lsize=1g,logdev=/dev/nvme1n1p6 /dev/nvme1n1p5

FSTYP=xfs
export TEST_DEV=/dev/nvme1n1p3
export TEST_DIR=/mnt/test
export TEST_LOGDEV=/dev/nvme1n1p4
export SCRATCH_DEV=/dev/nvme1n1p5
export SCRATCH_MNT=/mnt/scratch
export SCRATCH_LOGDEV=/dev/nvme1n1p6
export USE_EXTERNAL=yes

sudo ./check generic/311

Current semantics allow two concurrent freezers: one initiated from
userspace via FREEZE_HOLDER_USERSPACE and one initiated from the kernel
via FREEZE_HOLDER_KERNEL. If there are multiple concurrent freeze
requests from either FREEZE_HOLDER_USERSPACE or FREEZE_HOLDER_KERNEL
-EBUSY is returned.

We need to preserve these semantics because as they are uapi via
FIFREEZE and FITHAW ioctl()s. IOW, freezes don't nest for FIFREEZE and
FITHAW. Other kernels consumers rely on non-nesting freezes as well.

With freezes initiated from the block layer freezes need to nest if the
same superblock is frozen via multiple devices. So we need to start
counting the number of freeze requests.

If FREEZE_MAY_NEST is passed alongside FREEZE_HOLDER_KERNEL or
FREEZE_HOLDER_USERSPACE we allow the caller to nest freeze calls.

To accommodate the old semantics we split the freeze counter into two
counting kernel initiated and userspace initiated freezes separately. We
can then also stop recording FREEZE_HOLDER_* in struct sb_writers.

We also simplify freezing by making all concurrent freezers share a
single active superblock reference count instead of having separate
references for kernel and userspace. I don't see why we would need two
active reference counts. Neither FREEZE_HOLDER_KERNEL nor
FREEZE_HOLDER_USERSPACE can put the active reference as long as they are
concurrent freezers anwyay. That was already true before we allowed
nesting freezes.

Survives various fstests runs with different options including the
reproducer, online scrub, and online repair, fsfreze, and so on. Also
survives blktests.

Link: https://lore.kernel.org/linux-block/87bkccnwxc.fsf@debian-BULLSEYE-live-builder-AMD64
Link: https://lore.kernel.org/r/20231104-vfs-multi-device-freeze-v2-2-5b5b69626eac@kernel.org
Fixes: 288d8706abfc ("bdev: implement freeze and thaw holder operations") [1] # no backport needed
Tested-by: Chandan Babu R <chandanbabu@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Reported-by: Chandan Babu R <chandanbabu@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff efa5d065 Sat Nov 04 08:00:12 MDT 2023 Christian Brauner <brauner@kernel.org> fs: remove dead check

Above we call super_lock_excl() which waits until the superblock is
SB_BORN and since SB_BORN is never unset once set this check can never
fire. Plus, we also hold an active reference at this point already so
this superblock can't even be shutdown.

Link: https://lore.kernel.org/r/20231104-vfs-multi-device-freeze-v2-1-5b5b69626eac@kernel.org
Tested-by: Chandan Babu R <chandanbabu@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff 49ef8832 Wed Sep 27 07:21:16 MDT 2023 Christian Brauner <brauner@kernel.org> bdev: implement freeze and thaw holder operations

The old method of implementing block device freeze and thaw operations
required us to rely on get_active_super() to walk the list of all
superblocks on the system to find any superblock that might use the
block device. This is wasteful and not very pleasant overall.

Now that we can finally go straight from block device to owning
superblock things become way simpler.

Link: https://lore.kernel.org/r/20231024-vfs-super-freeze-v2-5-599c19f4faac@kernel.org
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff dc3216b1 Mon Aug 28 05:26:24 MDT 2023 Christian Brauner <brauner@kernel.org> super: ensure valid info

For keyed filesystems that recycle superblocks based on s_fs_info or
information contained therein s_fs_info must be kept as long as the
superblock is on the filesystem type super list. This isn't guaranteed
as s_fs_info will be freed latest in sb->kill_sb().

The fix is simply to perform notification and list removal in
kill_anon_super(). Any filesystem needs to free s_fs_info after they
call the kill_*() helpers. If they don't they risk use-after-free right
now so fixing it here is guaranteed that s_fs_info remain valid.

For block backed filesystems notifying in pass sb->kill_sb() in
deactivate_locked_super() remains unproblematic and is required because
multiple other block devices can be shut down after kill_block_super()
has been called from a filesystem's sb->kill_sb() handler. For example,
ext4 and xfs close additional devices. Block based filesystems don't
depend on s_fs_info (btrfs does use s_fs_info but also uses
kill_anon_super() and not kill_block_super().).

Sorry for that braino. Goal should be to unify this behavior during this
cycle obviously. But let's please do a simple bugfix now.

Fixes: 2c18a63b760a ("super: wait until we passed kill super")
Fixes: syzbot+5b64180f8d9e39d3f061@syzkaller.appspotmail.com
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reported-by: syzbot+5b64180f8d9e39d3f061@syzkaller.appspotmail.com
Message-Id: <20230828-vfs-super-fixes-v1-2-b37a4a04a88f@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff dc3216b1 Mon Aug 28 05:26:24 MDT 2023 Christian Brauner <brauner@kernel.org> super: ensure valid info

For keyed filesystems that recycle superblocks based on s_fs_info or
information contained therein s_fs_info must be kept as long as the
superblock is on the filesystem type super list. This isn't guaranteed
as s_fs_info will be freed latest in sb->kill_sb().

The fix is simply to perform notification and list removal in
kill_anon_super(). Any filesystem needs to free s_fs_info after they
call the kill_*() helpers. If they don't they risk use-after-free right
now so fixing it here is guaranteed that s_fs_info remain valid.

For block backed filesystems notifying in pass sb->kill_sb() in
deactivate_locked_super() remains unproblematic and is required because
multiple other block devices can be shut down after kill_block_super()
has been called from a filesystem's sb->kill_sb() handler. For example,
ext4 and xfs close additional devices. Block based filesystems don't
depend on s_fs_info (btrfs does use s_fs_info but also uses
kill_anon_super() and not kill_block_super().).

Sorry for that braino. Goal should be to unify this behavior during this
cycle obviously. But let's please do a simple bugfix now.

Fixes: 2c18a63b760a ("super: wait until we passed kill super")
Fixes: syzbot+5b64180f8d9e39d3f061@syzkaller.appspotmail.com
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reported-by: syzbot+5b64180f8d9e39d3f061@syzkaller.appspotmail.com
Message-Id: <20230828-vfs-super-fixes-v1-2-b37a4a04a88f@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff 5e874914 Fri Aug 18 08:00:50 MDT 2023 Christian Brauner <brauner@kernel.org> super: wait for nascent superblocks

Recent patches experiment with making it possible to allocate a new
superblock before opening the relevant block device. Naturally this has
intricate side-effects that we get to learn about while developing this.

Superblock allocators such as sget{_fc}() return with s_umount of the
new superblock held and lock ordering currently requires that block
level locks such as bdev_lock and open_mutex rank above s_umount.

Before aca740cecbe5 ("fs: open block device after superblock creation")
ordering was guaranteed to be correct as block devices were opened prior
to superblock allocation and thus s_umount wasn't held. But now s_umount
must be dropped before opening block devices to avoid locking
violations.

This has consequences. The main one being that iterators over
@super_blocks and @fs_supers that grab a temporary reference to the
superblock can now also grab s_umount before the caller has managed to
open block devices and called fill_super(). So whereas before such
iterators or concurrent mounts would have simply slept on s_umount until
SB_BORN was set or the superblock was discard due to initalization
failure they can now needlessly spin through sget{_fc}().

If the caller is sleeping on bdev_lock or open_mutex one caller waiting
on SB_BORN will always spin somewhere and potentially this can go on for
quite a while.

It should be possible to drop s_umount while allowing iterators to wait
on a nascent superblock to either be born or discarded. This patch
implements a wait_var_event() mechanism allowing iterators to sleep
until they are woken when the superblock is born or discarded.

This also allows us to avoid relooping through @fs_supers and
@super_blocks if a superblock isn't yet born or dying.

Link: aca740cecbe5 ("fs: open block device after superblock creation")
Reviewed-by: Jan Kara <jack@suse.cz>
Message-Id: <20230818-vfs-super-fixes-v3-v3-3-9f0b1876e46b@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff 22ed7ecd Wed Aug 02 05:57:06 MDT 2023 Christian Brauner <brauner@kernel.org> fs: add FSCONFIG_CMD_CREATE_EXCL

Summary
=======

This introduces FSCONFIG_CMD_CREATE_EXCL which will allows userspace to
implement something like mount -t ext4 --exclusive /dev/sda /B which
fails if a superblock for the requested filesystem does already exist:

Before this patch
-----------------

$ sudo ./move-mount -f xfs -o source=/dev/sda4 /A
Requesting filesystem type xfs
Mount options requested: source=/dev/sda4
Attaching mount at /A
Moving single attached mount
Setting key(source) with val(/dev/sda4)

$ sudo ./move-mount -f xfs -o source=/dev/sda4 /B
Requesting filesystem type xfs
Mount options requested: source=/dev/sda4
Attaching mount at /B
Moving single attached mount
Setting key(source) with val(/dev/sda4)

After this patch with --exclusive as a switch for FSCONFIG_CMD_CREATE_EXCL
--------------------------------------------------------------------------

$ sudo ./move-mount -f xfs --exclusive -o source=/dev/sda4 /A
Requesting filesystem type xfs
Request exclusive superblock creation
Mount options requested: source=/dev/sda4
Attaching mount at /A
Moving single attached mount
Setting key(source) with val(/dev/sda4)

$ sudo ./move-mount -f xfs --exclusive -o source=/dev/sda4 /B
Requesting filesystem type xfs
Request exclusive superblock creation
Mount options requested: source=/dev/sda4
Attaching mount at /B
Moving single attached mount
Setting key(source) with val(/dev/sda4)
Device or resource busy | move-mount.c: 300: do_fsconfig: i xfs: reusing existing filesystem not allowed

Details
=======

As mentioned on the list (cf. [1]-[3]) mount requests like
mount -t ext4 /dev/sda /A are ambigous for userspace. Either a new
superblock has been created and mounted or an existing superblock has
been reused and a bind-mount has been created.

This becomes clear in the following example where two processes create
the same mount for the same block device:

P1 P2
fd_fs = fsopen("ext4"); fd_fs = fsopen("ext4");
fsconfig(fd_fs, FSCONFIG_SET_STRING, "source", "/dev/sda"); fsconfig(fd_fs, FSCONFIG_SET_STRING, "source", "/dev/sda");
fsconfig(fd_fs, FSCONFIG_SET_STRING, "dax", "always"); fsconfig(fd_fs, FSCONFIG_SET_STRING, "resuid", "1000");

// wins and creates superblock
fsconfig(fd_fs, FSCONFIG_CMD_CREATE, ...)
// finds compatible superblock of P1
// spins until P1 sets SB_BORN and grabs a reference
fsconfig(fd_fs, FSCONFIG_CMD_CREATE, ...)

fd_mnt1 = fsmount(fd_fs); fd_mnt2 = fsmount(fd_fs);
move_mount(fd_mnt1, "/A") move_mount(fd_mnt2, "/B")

Not just does P2 get a bind-mount but the mount options that P2
requestes are silently ignored. The VFS itself doesn't, can't and
shouldn't enforce filesystem specific mount option compatibility. It
only enforces incompatibility for read-only <-> read-write transitions:

mount -t ext4 /dev/sda /A
mount -t ext4 -o ro /dev/sda /B

The read-only request will fail with EBUSY as the VFS can't just
silently transition a superblock from read-write to read-only or vica
versa without risking security issues.

To userspace this silent superblock reuse can become a security issue in
because there is currently no straightforward way for userspace to know
that they did indeed manage to create a new superblock and didn't just
reuse an existing one.

This adds a new FSCONFIG_CMD_CREATE_EXCL command to fsconfig() that
returns EBUSY if an existing superblock would be reused. Userspace that
needs to be sure that it did create a new superblock with the requested
mount options can request superblock creation using this command. If the
command succeeds they can be sure that they did create a new superblock
with the requested mount options.

This requires the new mount api. With the old mount api it would be
necessary to plumb this through every legacy filesystem's
file_system_type->mount() method. If they want this feature they are
most welcome to switch to the new mount api.

Following is an analysis of the effect of FSCONFIG_CMD_CREATE_EXCL on
each high-level superblock creation helper:

(1) get_tree_nodev()

Always allocate new superblock. Hence, FSCONFIG_CMD_CREATE and
FSCONFIG_CMD_CREATE_EXCL are equivalent.

The binderfs or overlayfs filesystems are examples.

(4) get_tree_keyed()

Finds an existing superblock based on sb->s_fs_info. Hence,
FSCONFIG_CMD_CREATE would reuse an existing superblock whereas
FSCONFIG_CMD_CREATE_EXCL would reject it with EBUSY.

The mqueue or nfsd filesystems are examples.

(2) get_tree_bdev()

This effectively works like get_tree_keyed().

The ext4 or xfs filesystems are examples.

(3) get_tree_single()

Only one superblock of this filesystem type can ever exist.
Hence, FSCONFIG_CMD_CREATE would reuse an existing superblock
whereas FSCONFIG_CMD_CREATE_EXCL would reject it with EBUSY.

The securityfs or configfs filesystems are examples.

Note that some single-instance filesystems never destroy the
superblock once it has been created during the first mount. For
example, if securityfs has been mounted at least onces then the
created superblock will never be destroyed again as long as there is
still an LSM making use it. Consequently, even if securityfs is
unmounted and the superblock seemingly destroyed it really isn't
which means that FSCONFIG_CMD_CREATE_EXCL will continue rejecting
reusing an existing superblock.

This is acceptable thugh since special purpose filesystems such as
this shouldn't have a need to use FSCONFIG_CMD_CREATE_EXCL anyway
and if they do it's probably to make sure that mount options aren't
ignored.

Following is an analysis of the effect of FSCONFIG_CMD_CREATE_EXCL on
filesystems that make use of the low-level sget_fc() helper directly.
They're all effectively variants on get_tree_keyed(), get_tree_bdev(),
or get_tree_nodev():

(5) mtd_get_sb()

Similar logic to get_tree_keyed().

(6) afs_get_tree()

Similar logic to get_tree_keyed().

(7) ceph_get_tree()

Similar logic to get_tree_keyed().

Already explicitly allows forcing the allocation of a new superblock
via CEPH_OPT_NOSHARE. This turns it into get_tree_nodev().

(8) fuse_get_tree_submount()

Similar logic to get_tree_nodev().

(9) fuse_get_tree()

Forces reuse of existing FUSE superblock.

Forces reuse of existing superblock if passed in file refers to an
existing FUSE connection.
If FSCONFIG_CMD_CREATE_EXCL is specified together with an fd
referring to an existing FUSE connections this would cause the
superblock reusal to fail. If reusing is the intent then
FSCONFIG_CMD_CREATE_EXCL shouldn't be specified.

(10) fuse_get_tree()
-> get_tree_nodev()

Same logic as in get_tree_nodev().

(11) fuse_get_tree()
-> get_tree_bdev()

Same logic as in get_tree_bdev().

(12) virtio_fs_get_tree()

Same logic as get_tree_keyed().

(13) gfs2_meta_get_tree()

Forces reuse of existing gfs2 superblock.

Mounting gfs2meta enforces that a gf2s superblock must already
exist. If not, it will error out. Consequently, mounting gfs2meta
with FSCONFIG_CMD_CREATE_EXCL would always fail. If reusing is the
intent then FSCONFIG_CMD_CREATE_EXCL shouldn't be specified.

(14) kernfs_get_tree()

Similar logic to get_tree_keyed().

(15) nfs_get_tree_common()

Similar logic to get_tree_keyed().

Already explicitly allows forcing the allocation of a new superblock
via NFS_MOUNT_UNSHARED. This effectively turns it into
get_tree_nodev().

Link: [1] https://lore.kernel.org/linux-block/20230704-fasching-wertarbeit-7c6ffb01c83d@brauner
Link: [2] https://lore.kernel.org/linux-block/20230705-pumpwerk-vielversprechend-a4b1fd947b65@brauner
Link: [3] https://lore.kernel.org/linux-fsdevel/20230725-einnahmen-warnschilder-17779aec0a97@brauner
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Aleksa Sarai <cyphar@cyphar.com>
Message-Id: <20230802-vfs-super-exclusive-v2-4-95dc4e41b870@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff 6a320739 Sun Aug 06 17:26:25 MDT 2023 Christoph Hellwig <hch@lst.de> fs, block: remove bdev->bd_super

bdev->bd_super is unused now, remove it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Christian Brauner <brauner@kernel.org>
Message-Id: <20230807112625.652089-5-hch@lst.de>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff ec64036e Tue Feb 07 23:21:06 MST 2023 Eric Biggers <ebiggers@google.com> fs/super.c: stop calling fscrypt_destroy_keyring() from __put_super()

Now that the key associated with the "test_dummy_operation" mount option
is added on-demand when it's needed, rather than immediately when the
filesystem is mounted, fscrypt_destroy_keyring() no longer needs to be
called from __put_super() to avoid a memory leak on mount failure.

Remove this call, which was causing confusion because it appeared to be
a sleep-in-atomic bug (though it wasn't, for a somewhat-subtle reason).

Signed-off-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20230208062107.199831-5-ebiggers@kernel.org
diff e33c267a Tue May 31 21:22:24 MDT 2022 Roman Gushchin <roman.gushchin@linux.dev> mm: shrinkers: provide shrinkers with names

Currently shrinkers are anonymous objects. For debugging purposes they
can be identified by count/scan function names, but it's not always
useful: e.g. for superblock's shrinkers it's nice to have at least an
idea of to which superblock the shrinker belongs.

This commit adds names to shrinkers. register_shrinker() and
prealloc_shrinker() functions are extended to take a format and arguments
to master a name.

In some cases it's not possible to determine a good name at the time when
a shrinker is allocated. For such cases shrinker_debugfs_rename() is
provided.

The expected format is:
<subsystem>-<shrinker_type>[:<instance>]-<id>
For some shrinkers an instance can be encoded as (MAJOR:MINOR) pair.

After this change the shrinker debugfs directory looks like:
$ cd /sys/kernel/debug/shrinker/
$ ls
dquota-cache-16 sb-devpts-28 sb-proc-47 sb-tmpfs-42
mm-shadow-18 sb-devtmpfs-5 sb-proc-48 sb-tmpfs-43
mm-zspool:zram0-34 sb-hugetlbfs-17 sb-pstore-31 sb-tmpfs-44
rcu-kfree-0 sb-hugetlbfs-33 sb-rootfs-2 sb-tmpfs-49
sb-aio-20 sb-iomem-12 sb-securityfs-6 sb-tracefs-13
sb-anon_inodefs-15 sb-mqueue-21 sb-selinuxfs-22 sb-xfs:vda1-36
sb-bdev-3 sb-nsfs-4 sb-sockfs-8 sb-zsmalloc-19
sb-bpf-32 sb-pipefs-14 sb-sysfs-26 thp-deferred_split-10
sb-btrfs:vda2-24 sb-proc-25 sb-tmpfs-1 thp-zero-9
sb-cgroup2-30 sb-proc-39 sb-tmpfs-27 xfs-buf:vda1-37
sb-configfs-23 sb-proc-41 sb-tmpfs-29 xfs-inodegc:vda1-38
sb-dax-11 sb-proc-45 sb-tmpfs-35
sb-debugfs-7 sb-proc-46 sb-tmpfs-40

[roman.gushchin@linux.dev: fix build warnings]
Link: https://lkml.kernel.org/r/Yr+ZTnLb9lJk6fJO@castle
Reported-by: kernel test robot <lkp@intel.com>
Link: https://lkml.kernel.org/r/20220601032227.4076670-4-roman.gushchin@linux.dev
Signed-off-by: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Cc: Dave Chinner <dchinner@redhat.com>
Cc: Hillf Danton <hdanton@sina.com>
Cc: Kent Overstreet <kent.overstreet@gmail.com>
Cc: Muchun Song <songmuchun@bytedance.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
/linux-master/include/linux/
H A Dfs.hdiff fe3944fb Fri Feb 02 01:39:23 MST 2024 Bart Van Assche <bvanassche@acm.org> fs: Move enum rw_hint into a new header file

Move enum rw_hint into a new header file to prepare for using this data
type in the block layer. Add the attribute __packed to reduce the space
occupied by instances of this data type from four bytes to one byte.
Change the data type of i_write_hint from u8 into enum rw_hint.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Acked-by: Chao Yu <chao@kernel.org> # for the F2FS part
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Jan Kara <jack@suse.cz>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20240202203926.2478590-5-bvanassche@acm.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff d3b1a9a7 Fri Feb 02 01:33:04 MST 2024 JonasZhou <JonasZhou@zhaoxin.com> fs/address_space: move i_mmap_rwsem to mitigate a false sharing with i_mmap.

In the struct address_space, there is a 32-byte gap between i_mmap
and i_mmap_rwsem. Due to the alignment of struct address_space
variables to 8 bytes, in certain situations, i_mmap and i_mmap_rwsem
may end up in the same CACHE line.

While running Unixbench/execl, we observe high false sharing issues
when accessing i_mmap against i_mmap_rwsem. We move i_mmap_rwsem
after i_private_list, ensuring a 64-byte gap between i_mmap and
i_mmap_rwsem.

For Intel Silver machines (2 sockets) using kernel v6.8 rc-2, the score
of Unixbench/execl improves by ~3.94%, and the score of Unixbench/shell
improves by ~3.26%.

Baseline:
-------------------------------------------------------------
162 546 748 11374 21 0xffff92e266af90c0
-------------------------------------------------------------
46.89% 44.65% 0.00% 0.00% 0x0 1 1 0xffffffff86d5fb96 460 258 271 1069 32 [k] __handle_mm_fault [kernel.vmlinux] memory.c:2940 0 1
4.21% 4.41% 0.00% 0.00% 0x4 1 1 0xffffffff86d0ed54 473 311 288 95 28 [k] filemap_read [kernel.vmlinux] atomic.h:23 0 1
0.00% 0.00% 0.04% 4.76% 0x8 1 1 0xffffffff86d4bcf1 0 0 0 5 4 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:204 0 1
6.41% 6.02% 0.00% 0.00% 0x8 1 1 0xffffffff86d4ba85 411 271 339 210 32 [k] vma_interval_tree_insert [kernel.vmlinux] interval_tree.c:23 0 1
0.00% 0.00% 0.47% 95.24% 0x10 1 1 0xffffffff86d4bd34 0 0 0 74 32 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:339 0 1
0.37% 0.13% 0.00% 0.00% 0x10 1 1 0xffffffff86d4bb4f 328 212 380 7 5 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
5.13% 5.08% 0.00% 0.00% 0x10 1 1 0xffffffff86d4bb4b 416 255 357 197 32 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
1.10% 0.53% 0.00% 0.00% 0x28 1 1 0xffffffff86e06eb8 395 228 351 24 14 [k] do_dentry_open [kernel.vmlinux] open.c:966 0 1
1.10% 2.14% 57.07% 0.00% 0x38 1 1 0xffffffff878c9225 1364 792 462 7003 32 [k] down_write [kernel.vmlinux] atomic64_64.h:109 0 1
0.00% 0.00% 0.01% 0.00% 0x38 1 1 0xffffffff878c8e75 0 0 252 3 2 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:109 0 1
0.00% 0.13% 0.00% 0.00% 0x38 1 1 0xffffffff878c8e23 0 596 63 2 2 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:15 0 1
2.38% 2.94% 6.53% 0.00% 0x38 1 1 0xffffffff878c8ccb 1150 818 570 1197 32 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:109 0 1
30.59% 32.22% 0.00% 0.00% 0x38 1 1 0xffffffff878c8cb4 423 251 380 648 32 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:15 0 1
1.83% 1.74% 35.88% 0.00% 0x38 1 1 0xffffffff86b4f833 1217 1112 565 4586 32 [k] up_write [kernel.vmlinux] atomic64_64.h:91 0 1

with this change:
-------------------------------------------------------------
360 12 300 57 35 0xffff982cdae76400
-------------------------------------------------------------
50.00% 59.67% 0.00% 0.00% 0x0 1 1 0xffffffff8215fb86 352 200 191 558 32 [k] __handle_mm_fault [kernel.vmlinux] memory.c:2940 0 1
8.33% 5.00% 0.00% 0.00% 0x4 1 1 0xffffffff8210ed44 370 284 263 42 24 [k] filemap_read [kernel.vmlinux] atomic.h:23 0 1
0.00% 0.00% 5.26% 2.86% 0x8 1 1 0xffffffff8214bce1 0 0 0 4 4 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:204 0 1
33.33% 14.33% 0.00% 0.00% 0x8 1 1 0xffffffff8214ba75 344 186 219 140 32 [k] vma_interval_tree_insert [kernel.vmlinux] interval_tree.c:23 0 1
0.00% 0.00% 94.74% 97.14% 0x10 1 1 0xffffffff8214bd24 0 0 0 88 29 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:339 0 1
8.33% 20.00% 0.00% 0.00% 0x10 1 1 0xffffffff8214bb3b 296 209 226 167 31 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
0.00% 0.67% 0.00% 0.00% 0x28 1 1 0xffffffff82206f45 0 140 334 4 3 [k] do_dentry_open [kernel.vmlinux] open.c:966 0 1
0.00% 0.33% 0.00% 0.00% 0x38 1 1 0xffffffff8250a6c4 0 286 126 5 5 [k] errseq_sample [kernel.vmlinux] errseq.c:125 0

Signed-off-by: JonasZhou <JonasZhou@zhaoxin.com>
Link: https://lore.kernel.org/r/20240202083304.10995-1-JonasZhou-oc@zhaoxin.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff d3b1a9a7 Fri Feb 02 01:33:04 MST 2024 JonasZhou <JonasZhou@zhaoxin.com> fs/address_space: move i_mmap_rwsem to mitigate a false sharing with i_mmap.

In the struct address_space, there is a 32-byte gap between i_mmap
and i_mmap_rwsem. Due to the alignment of struct address_space
variables to 8 bytes, in certain situations, i_mmap and i_mmap_rwsem
may end up in the same CACHE line.

While running Unixbench/execl, we observe high false sharing issues
when accessing i_mmap against i_mmap_rwsem. We move i_mmap_rwsem
after i_private_list, ensuring a 64-byte gap between i_mmap and
i_mmap_rwsem.

For Intel Silver machines (2 sockets) using kernel v6.8 rc-2, the score
of Unixbench/execl improves by ~3.94%, and the score of Unixbench/shell
improves by ~3.26%.

Baseline:
-------------------------------------------------------------
162 546 748 11374 21 0xffff92e266af90c0
-------------------------------------------------------------
46.89% 44.65% 0.00% 0.00% 0x0 1 1 0xffffffff86d5fb96 460 258 271 1069 32 [k] __handle_mm_fault [kernel.vmlinux] memory.c:2940 0 1
4.21% 4.41% 0.00% 0.00% 0x4 1 1 0xffffffff86d0ed54 473 311 288 95 28 [k] filemap_read [kernel.vmlinux] atomic.h:23 0 1
0.00% 0.00% 0.04% 4.76% 0x8 1 1 0xffffffff86d4bcf1 0 0 0 5 4 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:204 0 1
6.41% 6.02% 0.00% 0.00% 0x8 1 1 0xffffffff86d4ba85 411 271 339 210 32 [k] vma_interval_tree_insert [kernel.vmlinux] interval_tree.c:23 0 1
0.00% 0.00% 0.47% 95.24% 0x10 1 1 0xffffffff86d4bd34 0 0 0 74 32 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:339 0 1
0.37% 0.13% 0.00% 0.00% 0x10 1 1 0xffffffff86d4bb4f 328 212 380 7 5 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
5.13% 5.08% 0.00% 0.00% 0x10 1 1 0xffffffff86d4bb4b 416 255 357 197 32 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
1.10% 0.53% 0.00% 0.00% 0x28 1 1 0xffffffff86e06eb8 395 228 351 24 14 [k] do_dentry_open [kernel.vmlinux] open.c:966 0 1
1.10% 2.14% 57.07% 0.00% 0x38 1 1 0xffffffff878c9225 1364 792 462 7003 32 [k] down_write [kernel.vmlinux] atomic64_64.h:109 0 1
0.00% 0.00% 0.01% 0.00% 0x38 1 1 0xffffffff878c8e75 0 0 252 3 2 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:109 0 1
0.00% 0.13% 0.00% 0.00% 0x38 1 1 0xffffffff878c8e23 0 596 63 2 2 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:15 0 1
2.38% 2.94% 6.53% 0.00% 0x38 1 1 0xffffffff878c8ccb 1150 818 570 1197 32 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:109 0 1
30.59% 32.22% 0.00% 0.00% 0x38 1 1 0xffffffff878c8cb4 423 251 380 648 32 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:15 0 1
1.83% 1.74% 35.88% 0.00% 0x38 1 1 0xffffffff86b4f833 1217 1112 565 4586 32 [k] up_write [kernel.vmlinux] atomic64_64.h:91 0 1

with this change:
-------------------------------------------------------------
360 12 300 57 35 0xffff982cdae76400
-------------------------------------------------------------
50.00% 59.67% 0.00% 0.00% 0x0 1 1 0xffffffff8215fb86 352 200 191 558 32 [k] __handle_mm_fault [kernel.vmlinux] memory.c:2940 0 1
8.33% 5.00% 0.00% 0.00% 0x4 1 1 0xffffffff8210ed44 370 284 263 42 24 [k] filemap_read [kernel.vmlinux] atomic.h:23 0 1
0.00% 0.00% 5.26% 2.86% 0x8 1 1 0xffffffff8214bce1 0 0 0 4 4 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:204 0 1
33.33% 14.33% 0.00% 0.00% 0x8 1 1 0xffffffff8214ba75 344 186 219 140 32 [k] vma_interval_tree_insert [kernel.vmlinux] interval_tree.c:23 0 1
0.00% 0.00% 94.74% 97.14% 0x10 1 1 0xffffffff8214bd24 0 0 0 88 29 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:339 0 1
8.33% 20.00% 0.00% 0.00% 0x10 1 1 0xffffffff8214bb3b 296 209 226 167 31 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
0.00% 0.67% 0.00% 0.00% 0x28 1 1 0xffffffff82206f45 0 140 334 4 3 [k] do_dentry_open [kernel.vmlinux] open.c:966 0 1
0.00% 0.33% 0.00% 0.00% 0x38 1 1 0xffffffff8250a6c4 0 286 126 5 5 [k] errseq_sample [kernel.vmlinux] errseq.c:125 0

Signed-off-by: JonasZhou <JonasZhou@zhaoxin.com>
Link: https://lore.kernel.org/r/20240202083304.10995-1-JonasZhou-oc@zhaoxin.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff d3b1a9a7 Fri Feb 02 01:33:04 MST 2024 JonasZhou <JonasZhou@zhaoxin.com> fs/address_space: move i_mmap_rwsem to mitigate a false sharing with i_mmap.

In the struct address_space, there is a 32-byte gap between i_mmap
and i_mmap_rwsem. Due to the alignment of struct address_space
variables to 8 bytes, in certain situations, i_mmap and i_mmap_rwsem
may end up in the same CACHE line.

While running Unixbench/execl, we observe high false sharing issues
when accessing i_mmap against i_mmap_rwsem. We move i_mmap_rwsem
after i_private_list, ensuring a 64-byte gap between i_mmap and
i_mmap_rwsem.

For Intel Silver machines (2 sockets) using kernel v6.8 rc-2, the score
of Unixbench/execl improves by ~3.94%, and the score of Unixbench/shell
improves by ~3.26%.

Baseline:
-------------------------------------------------------------
162 546 748 11374 21 0xffff92e266af90c0
-------------------------------------------------------------
46.89% 44.65% 0.00% 0.00% 0x0 1 1 0xffffffff86d5fb96 460 258 271 1069 32 [k] __handle_mm_fault [kernel.vmlinux] memory.c:2940 0 1
4.21% 4.41% 0.00% 0.00% 0x4 1 1 0xffffffff86d0ed54 473 311 288 95 28 [k] filemap_read [kernel.vmlinux] atomic.h:23 0 1
0.00% 0.00% 0.04% 4.76% 0x8 1 1 0xffffffff86d4bcf1 0 0 0 5 4 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:204 0 1
6.41% 6.02% 0.00% 0.00% 0x8 1 1 0xffffffff86d4ba85 411 271 339 210 32 [k] vma_interval_tree_insert [kernel.vmlinux] interval_tree.c:23 0 1
0.00% 0.00% 0.47% 95.24% 0x10 1 1 0xffffffff86d4bd34 0 0 0 74 32 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:339 0 1
0.37% 0.13% 0.00% 0.00% 0x10 1 1 0xffffffff86d4bb4f 328 212 380 7 5 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
5.13% 5.08% 0.00% 0.00% 0x10 1 1 0xffffffff86d4bb4b 416 255 357 197 32 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
1.10% 0.53% 0.00% 0.00% 0x28 1 1 0xffffffff86e06eb8 395 228 351 24 14 [k] do_dentry_open [kernel.vmlinux] open.c:966 0 1
1.10% 2.14% 57.07% 0.00% 0x38 1 1 0xffffffff878c9225 1364 792 462 7003 32 [k] down_write [kernel.vmlinux] atomic64_64.h:109 0 1
0.00% 0.00% 0.01% 0.00% 0x38 1 1 0xffffffff878c8e75 0 0 252 3 2 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:109 0 1
0.00% 0.13% 0.00% 0.00% 0x38 1 1 0xffffffff878c8e23 0 596 63 2 2 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:15 0 1
2.38% 2.94% 6.53% 0.00% 0x38 1 1 0xffffffff878c8ccb 1150 818 570 1197 32 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:109 0 1
30.59% 32.22% 0.00% 0.00% 0x38 1 1 0xffffffff878c8cb4 423 251 380 648 32 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:15 0 1
1.83% 1.74% 35.88% 0.00% 0x38 1 1 0xffffffff86b4f833 1217 1112 565 4586 32 [k] up_write [kernel.vmlinux] atomic64_64.h:91 0 1

with this change:
-------------------------------------------------------------
360 12 300 57 35 0xffff982cdae76400
-------------------------------------------------------------
50.00% 59.67% 0.00% 0.00% 0x0 1 1 0xffffffff8215fb86 352 200 191 558 32 [k] __handle_mm_fault [kernel.vmlinux] memory.c:2940 0 1
8.33% 5.00% 0.00% 0.00% 0x4 1 1 0xffffffff8210ed44 370 284 263 42 24 [k] filemap_read [kernel.vmlinux] atomic.h:23 0 1
0.00% 0.00% 5.26% 2.86% 0x8 1 1 0xffffffff8214bce1 0 0 0 4 4 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:204 0 1
33.33% 14.33% 0.00% 0.00% 0x8 1 1 0xffffffff8214ba75 344 186 219 140 32 [k] vma_interval_tree_insert [kernel.vmlinux] interval_tree.c:23 0 1
0.00% 0.00% 94.74% 97.14% 0x10 1 1 0xffffffff8214bd24 0 0 0 88 29 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:339 0 1
8.33% 20.00% 0.00% 0.00% 0x10 1 1 0xffffffff8214bb3b 296 209 226 167 31 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
0.00% 0.67% 0.00% 0.00% 0x28 1 1 0xffffffff82206f45 0 140 334 4 3 [k] do_dentry_open [kernel.vmlinux] open.c:966 0 1
0.00% 0.33% 0.00% 0.00% 0x38 1 1 0xffffffff8250a6c4 0 286 126 5 5 [k] errseq_sample [kernel.vmlinux] errseq.c:125 0

Signed-off-by: JonasZhou <JonasZhou@zhaoxin.com>
Link: https://lore.kernel.org/r/20240202083304.10995-1-JonasZhou-oc@zhaoxin.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff d3b1a9a7 Fri Feb 02 01:33:04 MST 2024 JonasZhou <JonasZhou@zhaoxin.com> fs/address_space: move i_mmap_rwsem to mitigate a false sharing with i_mmap.

In the struct address_space, there is a 32-byte gap between i_mmap
and i_mmap_rwsem. Due to the alignment of struct address_space
variables to 8 bytes, in certain situations, i_mmap and i_mmap_rwsem
may end up in the same CACHE line.

While running Unixbench/execl, we observe high false sharing issues
when accessing i_mmap against i_mmap_rwsem. We move i_mmap_rwsem
after i_private_list, ensuring a 64-byte gap between i_mmap and
i_mmap_rwsem.

For Intel Silver machines (2 sockets) using kernel v6.8 rc-2, the score
of Unixbench/execl improves by ~3.94%, and the score of Unixbench/shell
improves by ~3.26%.

Baseline:
-------------------------------------------------------------
162 546 748 11374 21 0xffff92e266af90c0
-------------------------------------------------------------
46.89% 44.65% 0.00% 0.00% 0x0 1 1 0xffffffff86d5fb96 460 258 271 1069 32 [k] __handle_mm_fault [kernel.vmlinux] memory.c:2940 0 1
4.21% 4.41% 0.00% 0.00% 0x4 1 1 0xffffffff86d0ed54 473 311 288 95 28 [k] filemap_read [kernel.vmlinux] atomic.h:23 0 1
0.00% 0.00% 0.04% 4.76% 0x8 1 1 0xffffffff86d4bcf1 0 0 0 5 4 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:204 0 1
6.41% 6.02% 0.00% 0.00% 0x8 1 1 0xffffffff86d4ba85 411 271 339 210 32 [k] vma_interval_tree_insert [kernel.vmlinux] interval_tree.c:23 0 1
0.00% 0.00% 0.47% 95.24% 0x10 1 1 0xffffffff86d4bd34 0 0 0 74 32 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:339 0 1
0.37% 0.13% 0.00% 0.00% 0x10 1 1 0xffffffff86d4bb4f 328 212 380 7 5 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
5.13% 5.08% 0.00% 0.00% 0x10 1 1 0xffffffff86d4bb4b 416 255 357 197 32 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
1.10% 0.53% 0.00% 0.00% 0x28 1 1 0xffffffff86e06eb8 395 228 351 24 14 [k] do_dentry_open [kernel.vmlinux] open.c:966 0 1
1.10% 2.14% 57.07% 0.00% 0x38 1 1 0xffffffff878c9225 1364 792 462 7003 32 [k] down_write [kernel.vmlinux] atomic64_64.h:109 0 1
0.00% 0.00% 0.01% 0.00% 0x38 1 1 0xffffffff878c8e75 0 0 252 3 2 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:109 0 1
0.00% 0.13% 0.00% 0.00% 0x38 1 1 0xffffffff878c8e23 0 596 63 2 2 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:15 0 1
2.38% 2.94% 6.53% 0.00% 0x38 1 1 0xffffffff878c8ccb 1150 818 570 1197 32 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:109 0 1
30.59% 32.22% 0.00% 0.00% 0x38 1 1 0xffffffff878c8cb4 423 251 380 648 32 [k] rwsem_down_write_slowpath [kernel.vmlinux] atomic64_64.h:15 0 1
1.83% 1.74% 35.88% 0.00% 0x38 1 1 0xffffffff86b4f833 1217 1112 565 4586 32 [k] up_write [kernel.vmlinux] atomic64_64.h:91 0 1

with this change:
-------------------------------------------------------------
360 12 300 57 35 0xffff982cdae76400
-------------------------------------------------------------
50.00% 59.67% 0.00% 0.00% 0x0 1 1 0xffffffff8215fb86 352 200 191 558 32 [k] __handle_mm_fault [kernel.vmlinux] memory.c:2940 0 1
8.33% 5.00% 0.00% 0.00% 0x4 1 1 0xffffffff8210ed44 370 284 263 42 24 [k] filemap_read [kernel.vmlinux] atomic.h:23 0 1
0.00% 0.00% 5.26% 2.86% 0x8 1 1 0xffffffff8214bce1 0 0 0 4 4 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:204 0 1
33.33% 14.33% 0.00% 0.00% 0x8 1 1 0xffffffff8214ba75 344 186 219 140 32 [k] vma_interval_tree_insert [kernel.vmlinux] interval_tree.c:23 0 1
0.00% 0.00% 94.74% 97.14% 0x10 1 1 0xffffffff8214bd24 0 0 0 88 29 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:339 0 1
8.33% 20.00% 0.00% 0.00% 0x10 1 1 0xffffffff8214bb3b 296 209 226 167 31 [k] vma_interval_tree_remove [kernel.vmlinux] rbtree_augmented.h:338 0 1
0.00% 0.67% 0.00% 0.00% 0x28 1 1 0xffffffff82206f45 0 140 334 4 3 [k] do_dentry_open [kernel.vmlinux] open.c:966 0 1
0.00% 0.33% 0.00% 0.00% 0x38 1 1 0xffffffff8250a6c4 0 286 126 5 5 [k] errseq_sample [kernel.vmlinux] errseq.c:125 0

Signed-off-by: JonasZhou <JonasZhou@zhaoxin.com>
Link: https://lore.kernel.org/r/20240202083304.10995-1-JonasZhou-oc@zhaoxin.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff 7366f8b6 Sat Nov 04 08:00:13 MDT 2023 Christian Brauner <brauner@kernel.org> fs: handle freezing from multiple devices

Before [1] freezing a filesystems through the block layer only worked
for the main block device as the owning superblock of additional block
devices could not be found. Any filesystem that made use of multiple
block devices would only be freezable via it's main block device.

For example, consider xfs over device mapper with /dev/dm-0 as main
block device and /dev/dm-1 as external log device. Two freeze requests
before [1]:

(1) dmsetup suspend /dev/dm-0 on the main block device

bdev_freeze(dm-0)
-> dm-0->bd_fsfreeze_count++
-> freeze_super(xfs-sb)

The owning superblock is found and the filesystem gets frozen.
Returns 0.

(2) dmsetup suspend /dev/dm-1 on the log device

bdev_freeze(dm-1)
-> dm-1->bd_fsfreeze_count++

The owning superblock isn't found and only the block device freeze
count is incremented. Returns 0.

Two freeze requests after [1]:

(1') dmsetup suspend /dev/dm-0 on the main block device

bdev_freeze(dm-0)
-> dm-0->bd_fsfreeze_count++
-> freeze_super(xfs-sb)

The owning superblock is found and the filesystem gets frozen.
Returns 0.

(2') dmsetup suspend /dev/dm-1 on the log device

bdev_freeze(dm-0)
-> dm-0->bd_fsfreeze_count++
-> freeze_super(xfs-sb)

The owning superblock is found and the filesystem gets frozen.
Returns -EBUSY.

When (2') is called we initiate a freeze from another block device of
the same superblock. So we increment the bd_fsfreeze_count for that
additional block device. But we now also find the owning superblock for
additional block devices and call freeze_super() again which reports
-EBUSY.

This can be reproduced through xfstests via:

mkfs.xfs -f -m crc=1,reflink=1,rmapbt=1, -i sparse=1 -lsize=1g,logdev=/dev/nvme1n1p4 /dev/nvme1n1p3
mkfs.xfs -f -m crc=1,reflink=1,rmapbt=1, -i sparse=1 -lsize=1g,logdev=/dev/nvme1n1p6 /dev/nvme1n1p5

FSTYP=xfs
export TEST_DEV=/dev/nvme1n1p3
export TEST_DIR=/mnt/test
export TEST_LOGDEV=/dev/nvme1n1p4
export SCRATCH_DEV=/dev/nvme1n1p5
export SCRATCH_MNT=/mnt/scratch
export SCRATCH_LOGDEV=/dev/nvme1n1p6
export USE_EXTERNAL=yes

sudo ./check generic/311

Current semantics allow two concurrent freezers: one initiated from
userspace via FREEZE_HOLDER_USERSPACE and one initiated from the kernel
via FREEZE_HOLDER_KERNEL. If there are multiple concurrent freeze
requests from either FREEZE_HOLDER_USERSPACE or FREEZE_HOLDER_KERNEL
-EBUSY is returned.

We need to preserve these semantics because as they are uapi via
FIFREEZE and FITHAW ioctl()s. IOW, freezes don't nest for FIFREEZE and
FITHAW. Other kernels consumers rely on non-nesting freezes as well.

With freezes initiated from the block layer freezes need to nest if the
same superblock is frozen via multiple devices. So we need to start
counting the number of freeze requests.

If FREEZE_MAY_NEST is passed alongside FREEZE_HOLDER_KERNEL or
FREEZE_HOLDER_USERSPACE we allow the caller to nest freeze calls.

To accommodate the old semantics we split the freeze counter into two
counting kernel initiated and userspace initiated freezes separately. We
can then also stop recording FREEZE_HOLDER_* in struct sb_writers.

We also simplify freezing by making all concurrent freezers share a
single active superblock reference count instead of having separate
references for kernel and userspace. I don't see why we would need two
active reference counts. Neither FREEZE_HOLDER_KERNEL nor
FREEZE_HOLDER_USERSPACE can put the active reference as long as they are
concurrent freezers anwyay. That was already true before we allowed
nesting freezes.

Survives various fstests runs with different options including the
reproducer, online scrub, and online repair, fsfreze, and so on. Also
survives blktests.

Link: https://lore.kernel.org/linux-block/87bkccnwxc.fsf@debian-BULLSEYE-live-builder-AMD64
Link: https://lore.kernel.org/r/20231104-vfs-multi-device-freeze-v2-2-5b5b69626eac@kernel.org
Fixes: 288d8706abfc ("bdev: implement freeze and thaw holder operations") [1] # no backport needed
Tested-by: Chandan Babu R <chandanbabu@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Reported-by: Chandan Babu R <chandanbabu@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff 5aa8fd9c Mon Sep 11 18:25:50 MDT 2023 Jeff Layton <jlayton@kernel.org> fs: add a new SB_I_NOUMASK flag

SB_POSIXACL must be set when a filesystem supports POSIX ACLs, but NFSv4
also sets this flag to prevent the VFS from applying the umask on
newly-created files. NFSv4 doesn't support POSIX ACLs however, which
causes confusion when other subsystems try to test for them.

Add a new SB_I_NOUMASK flag that allows filesystems to opt-in to umask
stripping without advertising support for POSIX ACLs. Set the new flag
on NFSv4 instead of SB_POSIXACL.

Also, move mode_strip_umask to namei.h and convert init_mknod and
init_mkdir to use it.

Signed-off-by: Jeff Layton <jlayton@kernel.org>
Message-Id: <20230911-acl-fix-v3-1-b25315333f6c@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff 03adc61e Thu Oct 12 15:55:18 MDT 2023 Dan Clash <daclash@linux.microsoft.com> audit,io_uring: io_uring openat triggers audit reference count underflow

An io_uring openat operation can update an audit reference count
from multiple threads resulting in the call trace below.

A call to io_uring_submit() with a single openat op with a flag of
IOSQE_ASYNC results in the following reference count updates.

These first part of the system call performs two increments that do not race.

do_syscall_64()
__do_sys_io_uring_enter()
io_submit_sqes()
io_openat_prep()
__io_openat_prep()
getname()
getname_flags() /* update 1 (increment) */
__audit_getname() /* update 2 (increment) */

The openat op is queued to an io_uring worker thread which starts the
opportunity for a race. The system call exit performs one decrement.

do_syscall_64()
syscall_exit_to_user_mode()
syscall_exit_to_user_mode_prepare()
__audit_syscall_exit()
audit_reset_context()
putname() /* update 3 (decrement) */

The io_uring worker thread performs one increment and two decrements.
These updates can race with the system call decrement.

io_wqe_worker()
io_worker_handle_work()
io_wq_submit_work()
io_issue_sqe()
io_openat()
io_openat2()
do_filp_open()
path_openat()
__audit_inode() /* update 4 (increment) */
putname() /* update 5 (decrement) */
__audit_uring_exit()
audit_reset_context()
putname() /* update 6 (decrement) */

The fix is to change the refcnt member of struct audit_names
from int to atomic_t.

kernel BUG at fs/namei.c:262!
Call Trace:
...
? putname+0x68/0x70
audit_reset_context.part.0.constprop.0+0xe1/0x300
__audit_uring_exit+0xda/0x1c0
io_issue_sqe+0x1f3/0x450
? lock_timer_base+0x3b/0xd0
io_wq_submit_work+0x8d/0x2b0
? __try_to_del_timer_sync+0x67/0xa0
io_worker_handle_work+0x17c/0x2b0
io_wqe_worker+0x10a/0x350

Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/lkml/MW2PR2101MB1033FFF044A258F84AEAA584F1C9A@MW2PR2101MB1033.namprd21.prod.outlook.com/
Fixes: 5bd2182d58e9 ("audit,io_uring,io-wq: add some basic audit support to io_uring")
Signed-off-by: Dan Clash <daclash@linux.microsoft.com>
Link: https://lore.kernel.org/r/20231012215518.GA4048@linuxonhyperv3.guj3yctzbm1etfxqx2vob5hsef.xx.internal.cloudapp.net
Reviewed-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff 03adc61e Thu Oct 12 15:55:18 MDT 2023 Dan Clash <daclash@linux.microsoft.com> audit,io_uring: io_uring openat triggers audit reference count underflow

An io_uring openat operation can update an audit reference count
from multiple threads resulting in the call trace below.

A call to io_uring_submit() with a single openat op with a flag of
IOSQE_ASYNC results in the following reference count updates.

These first part of the system call performs two increments that do not race.

do_syscall_64()
__do_sys_io_uring_enter()
io_submit_sqes()
io_openat_prep()
__io_openat_prep()
getname()
getname_flags() /* update 1 (increment) */
__audit_getname() /* update 2 (increment) */

The openat op is queued to an io_uring worker thread which starts the
opportunity for a race. The system call exit performs one decrement.

do_syscall_64()
syscall_exit_to_user_mode()
syscall_exit_to_user_mode_prepare()
__audit_syscall_exit()
audit_reset_context()
putname() /* update 3 (decrement) */

The io_uring worker thread performs one increment and two decrements.
These updates can race with the system call decrement.

io_wqe_worker()
io_worker_handle_work()
io_wq_submit_work()
io_issue_sqe()
io_openat()
io_openat2()
do_filp_open()
path_openat()
__audit_inode() /* update 4 (increment) */
putname() /* update 5 (decrement) */
__audit_uring_exit()
audit_reset_context()
putname() /* update 6 (decrement) */

The fix is to change the refcnt member of struct audit_names
from int to atomic_t.

kernel BUG at fs/namei.c:262!
Call Trace:
...
? putname+0x68/0x70
audit_reset_context.part.0.constprop.0+0xe1/0x300
__audit_uring_exit+0xda/0x1c0
io_issue_sqe+0x1f3/0x450
? lock_timer_base+0x3b/0xd0
io_wq_submit_work+0x8d/0x2b0
? __try_to_del_timer_sync+0x67/0xa0
io_worker_handle_work+0x17c/0x2b0
io_wqe_worker+0x10a/0x350

Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/lkml/MW2PR2101MB1033FFF044A258F84AEAA584F1C9A@MW2PR2101MB1033.namprd21.prod.outlook.com/
Fixes: 5bd2182d58e9 ("audit,io_uring,io-wq: add some basic audit support to io_uring")
Signed-off-by: Dan Clash <daclash@linux.microsoft.com>
Link: https://lore.kernel.org/r/20231012215518.GA4048@linuxonhyperv3.guj3yctzbm1etfxqx2vob5hsef.xx.internal.cloudapp.net
Reviewed-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Christian Brauner <brauner@kernel.org>
diff aee79d4e Sun Jul 16 08:56:54 MDT 2023 Zhu, Lipeng <lipeng.zhu@intel.com> fs/address_space: add alignment padding for i_map and i_mmap_rwsem to mitigate a false sharing.

When running UnixBench/Shell Scripts, we observed high false sharing for
accessing i_mmap against i_mmap_rwsem.

UnixBench/Shell Scripts are typical load/execute command test scenarios,
which concurrently launch->execute->exit a lot of shell commands. A lot
of processes invoke vma_interval_tree_remove which touch "i_mmap", the
call stack:

----vma_interval_tree_remove
|----unlink_file_vma
| free_pgtables
| |----exit_mmap
| | mmput
| | |----begin_new_exec
| | | load_elf_binary
| | | bprm_execve

Meanwhile, there are a lot of processes touch 'i_mmap_rwsem' to acquire
the semaphore in order to access 'i_mmap'. In existing 'address_space'
layout, 'i_mmap' and 'i_mmap_rwsem' are in the same cacheline.

The patch places the i_mmap and i_mmap_rwsem in separate cache lines to
avoid this false sharing problem.

With this patch, based on kernel v6.4.0, on Intel Sapphire Rapids
112C/224T platform, the score improves by ~5.3%. And perf c2c tool shows
the false sharing is resolved as expected, the symbol
vma_interval_tree_remove disappeared in cache line 0 after this change.

Baseline:
=================================================
Shared Cache Line Distribution Pareto
=================================================
-------------------------------------------------------------
0 3729 5791 0 0 0xff19b3818445c740
-------------------------------------------------------------
3.27% 3.02% 0.00% 0.00% 0x18 0 1 0xffffffffa194403b 604 483 389 692 203 [k] vma_interval_tree_insert [kernel.kallsyms] vma_interval_tree_insert+75 0 1
4.13% 3.63% 0.00% 0.00% 0x20 0 1 0xffffffffa19440a2 553 413 415 962 215 [k] vma_interval_tree_remove [kernel.kallsyms] vma_interval_tree_remove+18 0 1
2.04% 1.35% 0.00% 0.00% 0x28 0 1 0xffffffffa219a1d6 1210 855 460 1229 222 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+678 0 1
0.62% 1.85% 0.00% 0.00% 0x28 0 1 0xffffffffa219a1bf 762 329 577 527 198 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+655 0 1
0.48% 0.31% 0.00% 0.00% 0x28 0 1 0xffffffffa219a58c 1677 1476 733 1544 224 [k] down_write [kernel.kallsyms] down_write+28 0 1
0.05% 0.07% 0.00% 0.00% 0x28 0 1 0xffffffffa219a21d 1040 819 689 33 27 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+749 0 1
0.00% 0.05% 0.00% 0.00% 0x28 0 1 0xffffffffa17707db 0 1005 786 1373 223 [k] up_write [kernel.kallsyms] up_write+27 0 1
0.00% 0.02% 0.00% 0.00% 0x28 0 1 0xffffffffa219a064 0 233 778 32 30 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+308 0 1
33.82% 34.10% 0.00% 0.00% 0x30 0 1 0xffffffffa1770945 779 495 534 6011 224 [k] rwsem_spin_on_owner [kernel.kallsyms] rwsem_spin_on_owner+53 0 1
17.06% 15.28% 0.00% 0.00% 0x30 0 1 0xffffffffa1770915 593 438 468 2715 224 [k] rwsem_spin_on_owner [kernel.kallsyms] rwsem_spin_on_owner+5 0 1
3.54% 3.52% 0.00% 0.00% 0x30 0 1 0xffffffffa2199f84 881 601 583 1421 223 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+84 0 1

With this change:
-------------------------------------------------------------
0 556 838 0 0 0xff2780d7965d2780
-------------------------------------------------------------
0.18% 0.60% 0.00% 0.00% 0x8 0 1 0xffffffffafff27b8 503 453 569 14 13 [k] do_dentry_open [kernel.kallsyms] do_dentry_open+456 0 1
0.54% 0.12% 0.00% 0.00% 0x8 0 1 0xffffffffaffc51ac 510 199 428 15 12 [k] hugepage_vma_check [kernel.kallsyms] hugepage_vma_check+252 0 1
1.80% 2.15% 0.00% 0.00% 0x18 0 1 0xffffffffb079a1d6 1778 799 343 215 136 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+678 0 1
0.54% 1.31% 0.00% 0.00% 0x18 0 1 0xffffffffb079a1bf 547 296 528 91 71 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+655 0 1
0.72% 0.72% 0.00% 0.00% 0x18 0 1 0xffffffffb079a58c 1479 1534 676 288 163 [k] down_write [kernel.kallsyms] down_write+28 0 1
0.00% 0.12% 0.00% 0.00% 0x18 0 1 0xffffffffafd707db 0 2381 744 282 158 [k] up_write [kernel.kallsyms] up_write+27 0 1
0.00% 0.12% 0.00% 0.00% 0x18 0 1 0xffffffffb079a064 0 239 518 6 6 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+308 0 1
46.58% 47.02% 0.00% 0.00% 0x20 0 1 0xffffffffafd70945 704 403 499 1137 219 [k] rwsem_spin_on_owner [kernel.kallsyms] rwsem_spin_on_owner+53 0 1
23.92% 25.78% 0.00% 0.00% 0x20 0 1 0xffffffffafd70915 558 413 500 542 185 [k] rwsem_spin_on_owner [kernel.kallsyms] rwsem_spin_on_owner+5 0 1

v1->v2: change padding to exchange fields.

Link: https://lkml.kernel.org/r/20230716145653.20122-1-lipeng.zhu@intel.com
Signed-off-by: Lipeng Zhu <lipeng.zhu@intel.com>
Reviewed-by: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Yu Ma <yu.ma@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
diff aee79d4e Sun Jul 16 08:56:54 MDT 2023 Zhu, Lipeng <lipeng.zhu@intel.com> fs/address_space: add alignment padding for i_map and i_mmap_rwsem to mitigate a false sharing.

When running UnixBench/Shell Scripts, we observed high false sharing for
accessing i_mmap against i_mmap_rwsem.

UnixBench/Shell Scripts are typical load/execute command test scenarios,
which concurrently launch->execute->exit a lot of shell commands. A lot
of processes invoke vma_interval_tree_remove which touch "i_mmap", the
call stack:

----vma_interval_tree_remove
|----unlink_file_vma
| free_pgtables
| |----exit_mmap
| | mmput
| | |----begin_new_exec
| | | load_elf_binary
| | | bprm_execve

Meanwhile, there are a lot of processes touch 'i_mmap_rwsem' to acquire
the semaphore in order to access 'i_mmap'. In existing 'address_space'
layout, 'i_mmap' and 'i_mmap_rwsem' are in the same cacheline.

The patch places the i_mmap and i_mmap_rwsem in separate cache lines to
avoid this false sharing problem.

With this patch, based on kernel v6.4.0, on Intel Sapphire Rapids
112C/224T platform, the score improves by ~5.3%. And perf c2c tool shows
the false sharing is resolved as expected, the symbol
vma_interval_tree_remove disappeared in cache line 0 after this change.

Baseline:
=================================================
Shared Cache Line Distribution Pareto
=================================================
-------------------------------------------------------------
0 3729 5791 0 0 0xff19b3818445c740
-------------------------------------------------------------
3.27% 3.02% 0.00% 0.00% 0x18 0 1 0xffffffffa194403b 604 483 389 692 203 [k] vma_interval_tree_insert [kernel.kallsyms] vma_interval_tree_insert+75 0 1
4.13% 3.63% 0.00% 0.00% 0x20 0 1 0xffffffffa19440a2 553 413 415 962 215 [k] vma_interval_tree_remove [kernel.kallsyms] vma_interval_tree_remove+18 0 1
2.04% 1.35% 0.00% 0.00% 0x28 0 1 0xffffffffa219a1d6 1210 855 460 1229 222 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+678 0 1
0.62% 1.85% 0.00% 0.00% 0x28 0 1 0xffffffffa219a1bf 762 329 577 527 198 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+655 0 1
0.48% 0.31% 0.00% 0.00% 0x28 0 1 0xffffffffa219a58c 1677 1476 733 1544 224 [k] down_write [kernel.kallsyms] down_write+28 0 1
0.05% 0.07% 0.00% 0.00% 0x28 0 1 0xffffffffa219a21d 1040 819 689 33 27 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+749 0 1
0.00% 0.05% 0.00% 0.00% 0x28 0 1 0xffffffffa17707db 0 1005 786 1373 223 [k] up_write [kernel.kallsyms] up_write+27 0 1
0.00% 0.02% 0.00% 0.00% 0x28 0 1 0xffffffffa219a064 0 233 778 32 30 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+308 0 1
33.82% 34.10% 0.00% 0.00% 0x30 0 1 0xffffffffa1770945 779 495 534 6011 224 [k] rwsem_spin_on_owner [kernel.kallsyms] rwsem_spin_on_owner+53 0 1
17.06% 15.28% 0.00% 0.00% 0x30 0 1 0xffffffffa1770915 593 438 468 2715 224 [k] rwsem_spin_on_owner [kernel.kallsyms] rwsem_spin_on_owner+5 0 1
3.54% 3.52% 0.00% 0.00% 0x30 0 1 0xffffffffa2199f84 881 601 583 1421 223 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+84 0 1

With this change:
-------------------------------------------------------------
0 556 838 0 0 0xff2780d7965d2780
-------------------------------------------------------------
0.18% 0.60% 0.00% 0.00% 0x8 0 1 0xffffffffafff27b8 503 453 569 14 13 [k] do_dentry_open [kernel.kallsyms] do_dentry_open+456 0 1
0.54% 0.12% 0.00% 0.00% 0x8 0 1 0xffffffffaffc51ac 510 199 428 15 12 [k] hugepage_vma_check [kernel.kallsyms] hugepage_vma_check+252 0 1
1.80% 2.15% 0.00% 0.00% 0x18 0 1 0xffffffffb079a1d6 1778 799 343 215 136 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+678 0 1
0.54% 1.31% 0.00% 0.00% 0x18 0 1 0xffffffffb079a1bf 547 296 528 91 71 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+655 0 1
0.72% 0.72% 0.00% 0.00% 0x18 0 1 0xffffffffb079a58c 1479 1534 676 288 163 [k] down_write [kernel.kallsyms] down_write+28 0 1
0.00% 0.12% 0.00% 0.00% 0x18 0 1 0xffffffffafd707db 0 2381 744 282 158 [k] up_write [kernel.kallsyms] up_write+27 0 1
0.00% 0.12% 0.00% 0.00% 0x18 0 1 0xffffffffb079a064 0 239 518 6 6 [k] rwsem_down_write_slowpath [kernel.kallsyms] rwsem_down_write_slowpath+308 0 1
46.58% 47.02% 0.00% 0.00% 0x20 0 1 0xffffffffafd70945 704 403 499 1137 219 [k] rwsem_spin_on_owner [kernel.kallsyms] rwsem_spin_on_owner+53 0 1
23.92% 25.78% 0.00% 0.00% 0x20 0 1 0xffffffffafd70915 558 413 500 542 185 [k] rwsem_spin_on_owner [kernel.kallsyms] rwsem_spin_on_owner+5 0 1

v1->v2: change padding to exchange fields.

Link: https://lkml.kernel.org/r/20230716145653.20122-1-lipeng.zhu@intel.com
Signed-off-by: Lipeng Zhu <lipeng.zhu@intel.com>
Reviewed-by: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Yu Ma <yu.ma@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

Completed in 809 milliseconds