linux/fs/gfs2
Christian Brauner 0ede61d858
file: convert to SLAB_TYPESAFE_BY_RCU
In recent discussions around some performance improvements in the file
handling area we discussed switching the file cache to rely on
SLAB_TYPESAFE_BY_RCU which allows us to get rid of call_rcu() based
freeing for files completely. This is a pretty sensitive change overall
but it might actually be worth doing.

The main downside is the subtlety. The other one is that we should
really wait for Jann's patch to land that enables KASAN to handle
SLAB_TYPESAFE_BY_RCU UAFs. Currently it doesn't but a patch for this
exists.

With SLAB_TYPESAFE_BY_RCU objects may be freed and reused multiple times
which requires a few changes. So it isn't sufficient anymore to just
acquire a reference to the file in question under rcu using
atomic_long_inc_not_zero() since the file might have already been
recycled and someone else might have bumped the reference.

In other words, callers might see reference count bumps from newer
users. For this reason it is necessary to verify that the pointer is the
same before and after the reference count increment. This pattern can be
seen in get_file_rcu() and __files_get_rcu().

In addition, it isn't possible to access or check fields in struct file
without first aqcuiring a reference on it. Not doing that was always
very dodgy and it was only usable for non-pointer data in struct file.
With SLAB_TYPESAFE_BY_RCU it is necessary that callers first acquire a
reference under rcu or they must hold the files_lock of the fdtable.
Failing to do either one of this is a bug.

Thanks to Jann for pointing out that we need to ensure memory ordering
between reallocations and pointer check by ensuring that all subsequent
loads have a dependency on the second load in get_file_rcu() and
providing a fixup that was folded into this patch.

Cc: Jann Horn <jannh@google.com>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
2023-10-19 11:02:48 +02:00
..
acl.c gfs2: convert to ctime accessor functions 2023-07-24 10:29:59 +02:00
acl.h fs: port ->set_acl() to pass mnt_idmap 2023-01-19 09:24:27 +01:00
aops.c gfs2 fixes 2023-09-05 13:00:28 -07:00
aops.h gfs2: support ludicrously large folios in gfs2_trans_add_databufs() 2023-06-19 16:19:30 -07:00
bmap.c gfs2 fixes 2023-09-05 13:00:28 -07:00
bmap.h gfs2: Eliminate gfs2_trim_blocks 2023-04-18 14:40:12 +02:00
dentry.c Reinstate "GFS2: free disk inode which is deleted by remote node -V2" 2023-03-23 19:37:56 +01:00
dir.c gfs2: convert to ctime accessor functions 2023-07-24 10:29:59 +02:00
dir.h gfs2: Delete an unnecessary check before brelse() 2019-09-04 20:22:17 +02:00
export.c Change calling conventions for filldir_t 2022-08-17 17:25:04 -04:00
file.c dlm for 6.6 2023-08-31 15:02:12 -07:00
gfs2.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 398 2019-06-05 17:37:12 +02:00
glock.c file: convert to SLAB_TYPESAFE_BY_RCU 2023-10-19 11:02:48 +02:00
glock.h gfs2: Remove LM_FLAG_PRIORITY flag 2023-09-05 15:58:16 +02:00
glops.c gfs2 fixes 2023-09-05 13:00:28 -07:00
glops.h gfs2: Split up gfs2_meta_sync into inode and rgrp versions 2020-10-29 22:16:46 +01:00
incore.h gfs2: change qd_slot_count to qd_slot_ref 2023-09-05 15:58:18 +02:00
inode.c gfs2 fixes 2023-09-05 13:00:28 -07:00
inode.h fs: port ->permission() to pass mnt_idmap 2023-01-19 09:24:28 +01:00
Kconfig fs: add CONFIG_BUFFER_HEAD 2023-08-02 09:13:09 -06:00
lock_dlm.c gfs2: Remove LM_FLAG_PRIORITY flag 2023-09-05 15:58:16 +02:00
log.c gfs2: Sanitize kthread stopping 2023-09-05 15:58:17 +02:00
log.h fs/gfs2: Use the enum req_op and blk_opf_t types 2022-07-14 12:14:32 -06:00
lops.c gfs2: Use mapping->gfp_mask for metadata inodes 2023-09-05 15:58:15 +02:00
lops.h fs/gfs2: Use the enum req_op and blk_opf_t types 2022-07-14 12:14:32 -06:00
main.c gfs2: Rename "freeze_workqueue" to "gfs2_freeze" 2023-09-05 15:58:17 +02:00
Makefile License cleanup: add SPDX GPL-2.0 license identifier to files with no license 2017-11-02 11:10:55 +01:00
meta_io.c gfs2: replace obvious uses of b_page with b_folio 2023-01-18 17:12:40 -08:00
meta_io.h gfs2: Use container_of() for gfs2_glock(aspace) 2022-05-24 21:29:14 +02:00
ops_fstype.c gfs2: Introduce new quota=quiet mount option 2023-09-05 15:58:17 +02:00
quota.c gfs2 fixes 2023-09-05 13:00:28 -07:00
quota.h gfs2: move privileged user check to gfs2_quota_lock_check 2020-05-08 18:47:58 +02:00
recovery.c gfs2: Rename "gfs_recovery" workqueue to "gfs2_recovery" 2023-09-05 15:58:17 +02:00
recovery.h gfs2: Rename "gfs_recovery" workqueue to "gfs2_recovery" 2023-09-05 15:58:17 +02:00
rgrp.c gfs2: Update rl_unlinked before releasing rgrp lock 2023-06-06 18:35:06 +02:00
rgrp.h Merge part of branch 'for-next.instantiate' into for-next 2022-08-05 18:37:03 +02:00
super.c gfs2 fixes 2023-09-05 13:00:28 -07:00
super.h gfs2: Fix asynchronous thread destruction 2023-09-05 15:58:17 +02:00
sys.c gfs2 fixes 2023-09-05 13:00:28 -07:00
sys.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 398 2019-06-05 17:37:12 +02:00
trace_gfs2.h gfs2: Remove 'first' trace_gfs2_promote argument 2021-10-25 08:42:19 +02:00
trans.c gfs2: Fix freeze consistency check in gfs2_trans_add_meta 2023-08-07 18:40:51 +02:00
trans.h Merge branches 'rgrp-glock-sharing' and 'gfs2-revoke' from https://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2.git 2021-02-23 18:54:22 +01:00
util.c gfs2: Fix withdraw race 2023-09-05 15:58:17 +02:00
util.h gfs2: gfs2_freeze_lock_shared cleanup 2023-07-03 22:30:26 +02:00
xattr.c gfs2: convert to ctime accessor functions 2023-07-24 10:29:59 +02:00
xattr.h treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 398 2019-06-05 17:37:12 +02:00