[PATCHES][RFC] the meat of tree-in-dcache series
Al Viro
viro at zeniv.linux.org.uk
Sat Sep 20 17:41:56 AEST 2025
[my apologies for the size of patchbomb and the size of Cc; this series
really affects a lot of places, if lightly]
Some filesystems use a kinda-sorta controlled dentry refcount leak to pin
dentries of created objects in dcache (and undo it when removing those).
Reference is grabbed and not released, but it's not actually _stored_
anywhere. That works, but it's hard to follow and verify; among other
things, we have no way to tell _which_ of the increments is intended
to be an unpaired one. Worse, on removal we need to decide whether
the reference had already been dropped, which can be non-trivial if
that removal is on umount and we need to figure out if this dentry is
pinned due to e.g. unlink() not done. Usually that is handled by using
kill_litter_super() as ->kill_sb(), but there are open-coded special
cases of the same (consider e.g. /proc/self).
Things get simpler if we introduce a new dentry flag (DCACHE_PERSISTENT)
marking those "leaked" dentries. Having it set claims responsibility
for +1 in refcount.
The end result this series is aiming for:
* get these unbalanced dget() and dput() replaced with new primitives that
would, in addition to adjusting refcount, set and clear persistency flag.
* instead of having kill_litter_super() mess with removing the remaining
"leaked" references (e.g. for all tmpfs files that hadn't been removed
prior to umount), have the regular shrink_dcache_for_umount() strip
DCACHE_PERSISTENT of all dentries, dropping the corresponding
reference if it had been set. After that kill_litter_super() becomes
an equivalent of kill_anon_super().
Doing that in a single step is not feasible - it would affect too many places
in too many filesystems. It has to be split into a series.
This work has really started early in 2024; quite a few preliminary pieces
have already gone into mainline. This chunk is finally getting to the
meat of that stuff - infrastructure and most of the conversions to it.
Some pieces are still sitting in the local branches, but the bulk of
that stuff is here.
The branch is -rc5-based; it lives in
git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs.git #work.persistency
individual patches in followups.
Please, help with review and testing; it does appear to survive the local beating,
but extra eyes on it would be very welcome.
First two commits add a couple of useful helpers, the next three add the
infrastructure and the rest consists of per-filesystem conversions.
Shortlog:
new helper: simple_remove_by_name()
new helper: simple_done_creating()
introduce a flag for explicitly marking persistently pinned dentries
primitives for maintaining persisitency
convert simple_{link,unlink,rmdir,rename,fill_super}() to new primitives
convert ramfs and tmpfs
procfs: make /self and /thread_self dentries persistent
configfs, securityfs: kill_litter_super() not needed
convert xenfs
convert smackfs
convert hugetlbfs
convert mqueue
convert bpf
convert dlmfs
convert fuse_ctl
convert pstore
convert tracefs
convert debugfs
debugfs: remove duplicate checks in callers of start_creating()
convert efivarfs
convert spufs
convert ibmasmfs
ibmasmfs: get rid of ibmasmfs_dir_ops
convert devpts
binderfs: use simple_start_creating()
binderfs_binder_ctl_create(): kill a bogus check
convert binderfs
autofs_{rmdir,unlink}: dentry->d_fsdata->dentry == dentry there
convert autofs
convert binfmt_misc
convert selinuxfs
functionfs: switch to simple_remove_by_name()
convert functionfs
gadgetfs: switch to simple_remove_by_name()
convert gadgetfs
hypfs: don't pin dentries twice
hypfs: switch hypfs_create_str() to returning int
hypfs: swich hypfs_create_u64() to returning int
convert hypfs
Diffstat:
arch/powerpc/platforms/cell/spufs/inode.c | 15 ++---
arch/s390/hypfs/hypfs.h | 6 +-
arch/s390/hypfs/hypfs_diag_fs.c | 60 +++++++-------------
arch/s390/hypfs/hypfs_vm_fs.c | 21 +++----
arch/s390/hypfs/inode.c | 82 +++++++++------------------
drivers/android/binderfs.c | 82 +++++++--------------------
drivers/base/devtmpfs.c | 2 +-
drivers/misc/ibmasm/ibmasmfs.c | 24 ++++----
drivers/usb/gadget/function/f_fs.c | 54 ++++++++----------
drivers/usb/gadget/legacy/inode.c | 49 ++++++++--------
drivers/xen/xenfs/super.c | 2 +-
fs/autofs/inode.c | 2 +-
fs/autofs/root.c | 11 ++--
fs/binfmt_misc.c | 69 +++++++++++------------
fs/configfs/mount.c | 2 +-
fs/dcache.c | 93 ++++++++++++++++++++++++-------
fs/debugfs/inode.c | 32 +++--------
fs/devpts/inode.c | 57 +++++++------------
fs/efivarfs/inode.c | 7 +--
fs/efivarfs/super.c | 5 +-
fs/fuse/control.c | 19 ++++---
fs/hugetlbfs/inode.c | 12 ++--
fs/libfs.c | 31 +++++++++--
fs/ocfs2/dlmfs/dlmfs.c | 8 +--
fs/proc/base.c | 6 +-
fs/proc/internal.h | 1 +
fs/proc/root.c | 14 ++---
fs/proc/self.c | 10 +---
fs/proc/thread_self.c | 11 +---
fs/pstore/inode.c | 7 ++-
fs/ramfs/inode.c | 8 +--
fs/tracefs/event_inode.c | 4 +-
fs/tracefs/inode.c | 13 ++---
include/linux/dcache.h | 3 +
include/linux/fs.h | 3 +
include/linux/proc_fs.h | 2 -
init/do_mounts.c | 2 +-
ipc/mqueue.c | 12 +---
kernel/bpf/inode.c | 15 ++---
mm/shmem.c | 23 ++------
security/inode.c | 2 +-
security/selinux/selinuxfs.c | 52 +++++++++--------
security/smack/smackfs.c | 2 +-
43 files changed, 412 insertions(+), 523 deletions(-)
More information about the Linuxppc-dev
mailing list