Avoid race when updating nr_unused count of unused dentries.
A message that is not in the archive (sent off-list or not yet received): 20040528143740.31687.patches@notabene
PATCH 1/1 Avoid race when updating nr_unused count of unused dentries.
d_count == 1 is no guarantee that dentry is on the dentry_unused list, even if it has just been incremented inside dcache_lock, as dput can decrement at any time.
This test from Greg Banks is much safer, and is more transparently correct.
From: Greg Banks <[email protected]>
Signed-off-by Neil Brown <[email protected]>
----------- Diffstat output ------------
./fs/dcache.c | 7 ++++--- 1 files changed, 4 insertions(+), 3 deletions(-)
fs/dcache.c renamed from fs/dcache.c~current~ +4 -3
@@ -259,7 +259,7 @@ int d_invalidate(struct dentry * dentry)static inline struct dentry *__dget_locked (struct dentry *dentry) {atomic_inc (&dentry->d_count); -if (atomic_read (&dentry->d_count) ==1 ) { +if (!list_empty (&dentry->d_lru)) { dentry_stat.nr_unused--;list_del_init (&dentry->d_lru); }
@@ -935,8 +935,9 @@ struct dentry *d_splice_alias(struct ino* lookup is going on. * * dentry_unused list is not updated even if lookup finds the required dentry -* in there. It is updated in places such as prune_dcache, shrink_dcache_sb and -* select_parent. This laziness saves lookup from dcache_lock acquisition. +* in there. It is updated in places such as prune_dcache, shrink_dcache_sb, +* select_parent and __dget_locked. This laziness saves lookup from dcache_lock +* acquisition. * * d_lookup() is protected against the concurrent renames in some unrelated * directory using the seqlockt_t rename_lock.