On Mon, Apr 05, 2021 at 06:23:49PM +0000, Al Viro wrote: > On Mon, Apr 05, 2021 at 07:08:01PM +0200, Christian Brauner wrote: > > > Ah dentry count of -127 looks... odd. > > dead + 1... > > void lockref_mark_dead(struct lockref *lockref) > { > assert_spin_locked(&lockref->lock); > lockref->count = -128; > } > > IOW, a leaked (uncounted) reference to dentry, that got dget() called on > it after dentry had been freed. > > IOW, current->fs->pwd.dentry happens to point to an already freed > struct dentry here. Joy... > > Could you slap > > spin_lock(¤t->fs->lock); > WARN_ON(d_count(current->fs->pwd.dentry) < 0); > spin_unlock(¤t->fs->lock); > > before and after calls of io_issue_sqe() and see if it triggers? We > definitely > are seeing buggered dentry refcounting here.
Check if this helps, please. diff --git a/fs/namei.c b/fs/namei.c index 216f16e74351..82344f1139ff 100644 --- a/fs/namei.c +++ b/fs/namei.c @@ -2289,6 +2289,9 @@ static const char *path_init(struct nameidata *nd, unsigned flags) int error; const char *s = nd->name->name; + nd->path.mnt = NULL; + nd->path.dentry = NULL; + /* LOOKUP_CACHED requires RCU, ask caller to retry */ if ((flags & (LOOKUP_RCU | LOOKUP_CACHED)) == LOOKUP_CACHED) return ERR_PTR(-EAGAIN); @@ -2322,8 +2325,6 @@ static const char *path_init(struct nameidata *nd, unsigned flags) } nd->root.mnt = NULL; - nd->path.mnt = NULL; - nd->path.dentry = NULL; /* Absolute pathname -- fetch the root (LOOKUP_IN_ROOT uses nd->dfd). */ if (*s == '/' && !(flags & LOOKUP_IN_ROOT)) {