fhandle: reject detached mounts in capable_wrt_mount()
Subsystems (5)
PATCH 1/1 fhandle: reject detached mounts in capable_wrt_mount()
The recent fhandle RCU fix moved the mount namespace capability check into capable_wrt_mount(), so a non-NULL mnt_namespace survives the ns_capable() dereference. The helper still assumes the later READ_ONCE(mount->mnt_ns) must be non-NULL because may_decode_fh() checked is_mounted() first.
That assumption is not stable. A detached mount from open_tree(..., OPEN_TREE_CLONE) can be dissolved on fput while open_by_handle_at() is between those checks, and umount_tree() can clear mount->mnt_ns. If the helper observes NULL, it dereferences mnt_ns->user_ns and panics.
Return false when the RCU read observes a detached mount. This keeps the relaxed permission path conservative: a mount no longer attached to a namespace cannot authorize open_by_handle_at() access.
---
Bug found and triaged by David Lee from Trail of Bits.
Trail of Bits has a minimal PoC that triggers this crash on a custom kernel build, which can be shared further if needed.
fs/fhandle.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
fs/fhandle.c +1 -1
@@ -295,7 +295,7 @@ static bool capable_wrt_mount(struct mount *mount)*/ guard (rcu)(); mnt_ns =READ_ONCE (mount->mnt_ns); -return ns_capable (mnt_ns->user_ns,CAP_SYS_ADMIN ); +return mnt_ns &&ns_capable (mnt_ns->user_ns,CAP_SYS_ADMIN ); } static inline int may_decode_fh(struct handle_to_path_ctx *ctx,
Signature · 1 line
2.43.0
Quoted 34 lines · David Lee
The recent fhandle RCU fix moved the mount namespace capability check into
capable_wrt_mount(), so a non-NULLmnt_namespacesurvives thens_capable()dereference. The helper still assumes the laterREAD_ONCE(mount->mnt_ns)must be non-NULL becausemay_decode_fh()checkedis_mounted()first.That assumption is not stable. A detached mount from
open_tree(..., OPEN_TREE_CLONE)can be dissolved on fput whileopen_by_handle_at()is between those checks, andumount_tree()can clearmount->mnt_ns. If the helper observes NULL, it dereferencesmnt_ns->user_nsand panics.Return false when the RCU read observes a detached mount. This keeps the relaxed permission path conservative: a mount no longer attached to a namespace cannot authorize
open_by_handle_at()access.Fixes 620c266f3949 ("fhandle: relax open_by_handle_at() permission checks")Signed-off-by David Lee <[email protected]>Assisted-by Codex:gpt-5---
Bug found and triaged by David Lee from Trail of Bits.Trail of Bits has a minimal PoC that triggers this crash on a custom kernel build, which can be shared further if needed.
fs/fhandle.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)fs/fhandle.c +0
-0
fs/fhandle.c static bool capable_wrt_mount(struct mount *mount) · line 301
Quoted 8 lines
fs/fhandle.c +1
-1@@ -295,7 +295,7 @@ static bool capable_wrt_mount(struct mount *mount)*/ guard (rcu)(); mnt_ns =READ_ONCE (mount->mnt_ns); -return ns_capable (mnt_ns->user_ns,CAP_SYS_ADMIN ); +return mnt_ns &&ns_capable (mnt_ns->user_ns,CAP_SYS_ADMIN ); } static inline int may_decode_fh(struct handle_to_path_ctx *ctx,
Seems sane.
Quoted 1 line · David Lee
fhandle: reject detached mounts in
capable_wrt_mount()
Applied to the vfs.fixes branch of the vfs/vfs.git tree. Patches in the vfs.fixes branch should appear in linux-next soon.
Please report any outstanding bugs that were missed during review in a new review to the original patch series allowing us to drop it.
It's encouraged to provide Acked-bys and Reviewed-bys even though the patch has now been applied. If possible patch trailers will be updated.
Note that commit hashes shown below are subject to change due to rebase, trailer updates or similar. If in doubt, please check the listed branch.
tree: https://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs.git branch: vfs.fixes
[1/1] fhandle: reject detached mounts in capable_wrt_mount()
https://git.kernel.org/vfs/vfs/c/b2f046af8c6c