git: d53633bfcf24 - main - rename(2): do not allow to rename root vnode of the mounted filesystem
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Sat, 06 Jun 2026 03:29:10 UTC
The branch main has been updated by kib:
URL: https://cgit.FreeBSD.org/src/commit/?id=d53633bfcf24a3eb3711e24b597aa8301f92b958
commit d53633bfcf24a3eb3711e24b597aa8301f92b958
Author: Konstantin Belousov <kib@FreeBSD.org>
AuthorDate: 2026-06-04 19:56:36 +0000
Commit: Konstantin Belousov <kib@FreeBSD.org>
CommitDate: 2026-06-06 03:28:28 +0000
rename(2): do not allow to rename root vnode of the mounted filesystem
Check for tdvp being vp_crossmp. This cannot happen for the normal
rename cases, but could if the target path specified by the syscall
points to the nullfs mount over the regular file. In this case namei()
cannot step over crossmp, and keep it in ni_dvp.
Since crossmp VOP_GETWRITEMOUNT() returns NULL mp, we retry the locking
dance since the belief is that NULL return is transient.
PR: 295826
Reviewed by: markj
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D57453
---
sys/kern/vfs_lookup.c | 2 +-
sys/kern/vfs_syscalls.c | 9 +++++++++
sys/sys/vnode.h | 1 +
3 files changed, 11 insertions(+), 1 deletion(-)
diff --git a/sys/kern/vfs_lookup.c b/sys/kern/vfs_lookup.c
index c1363c0104d3..5d66070ca142 100644
--- a/sys/kern/vfs_lookup.c
+++ b/sys/kern/vfs_lookup.c
@@ -113,7 +113,7 @@ SDT_PROBE_DEFINE4(vfs, namei, lookup, return, "int", "struct vnode *", "bool",
uma_zone_t namei_zone;
/* Placeholder vnode for mp traversal. */
-static struct vnode *vp_crossmp;
+struct vnode *vp_crossmp;
static int
crossmp_vop_islocked(struct vop_islocked_args *ap)
diff --git a/sys/kern/vfs_syscalls.c b/sys/kern/vfs_syscalls.c
index 71d37e08c65b..7ca7ccd582e1 100644
--- a/sys/kern/vfs_syscalls.c
+++ b/sys/kern/vfs_syscalls.c
@@ -3823,6 +3823,15 @@ again:
}
tdvp = tond.ni_dvp;
tvp = tond.ni_vp;
+ if (tdvp == vp_crossmp) {
+ /*
+ * Rename of the root vnode of the mounted
+ * filesystem. It is possible to get there with the
+ * nullfs mount over the regular file.
+ */
+ error = EBUSY;
+ goto out;
+ }
if (tvp != NULL && (flags & AT_RENAME_NOREPLACE) != 0) {
/*
* Often filesystems need to relock the vnodes in
diff --git a/sys/sys/vnode.h b/sys/sys/vnode.h
index 41b5e21fb879..99e90aa4187e 100644
--- a/sys/sys/vnode.h
+++ b/sys/sys/vnode.h
@@ -462,6 +462,7 @@ extern struct mount *rootdevmp; /* "/dev" mount */
extern u_long desiredvnodes; /* number of vnodes desired */
extern struct uma_zone *namei_zone;
extern struct vattr va_null; /* predefined null vattr structure */
+extern struct vnode *vp_crossmp;
extern u_int vn_lock_pair_pause_max;