Mark the various refcount_t functions with __signed_wrap, as we depend
on the wrapping behavior to detect the overflow and perform saturation.
Silences warnings seen with the LKDTM REFCOUNT_* tests:

  UBSAN: signed-integer-overflow in ../include/linux/refcount.h:189:11
  2147483647 + 1 cannot be represented in type 'int'

Signed-off-by: Kees Cook <keescook@chromium.org>
---
Cc: Miguel Ojeda <ojeda@kernel.org>
Cc: Will Deacon <will@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: "Gustavo A. R. Silva" <gustavoars@kernel.org>
Cc: linux-hardening@vger.kernel.org
---
 include/linux/refcount.h | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/include/linux/refcount.h b/include/linux/refcount.h
index 85c6df0d1bef..59b3b752394d 100644
--- a/include/linux/refcount.h
+++ b/include/linux/refcount.h
@@ -136,7 +136,8 @@ static inline unsigned int refcount_read(const refcount_t *r)
 	return atomic_read(&r->refs);
 }
 
-static inline __must_check bool __refcount_add_not_zero(int i, refcount_t *r, int *oldp)
+static inline __must_check __signed_wrap
+bool __refcount_add_not_zero(int i, refcount_t *r, int *oldp)
 {
 	int old = refcount_read(r);
 
@@ -177,7 +178,8 @@ static inline __must_check bool refcount_add_not_zero(int i, refcount_t *r)
 	return __refcount_add_not_zero(i, r, NULL);
 }
 
-static inline void __refcount_add(int i, refcount_t *r, int *oldp)
+static inline __signed_wrap
+void __refcount_add(int i, refcount_t *r, int *oldp)
 {
 	int old = atomic_fetch_add_relaxed(i, &r->refs);
 
@@ -256,7 +258,8 @@ static inline void refcount_inc(refcount_t *r)
 	__refcount_inc(r, NULL);
 }
 
-static inline __must_check bool __refcount_sub_and_test(int i, refcount_t *r, int *oldp)
+static inline __must_check __signed_wrap
+bool __refcount_sub_and_test(int i, refcount_t *r, int *oldp)
 {
 	int old = atomic_fetch_sub_release(i, &r->refs);
 
-- 
2.34.1

On Wed, Feb 21, 2024 at 6:16 AM Kees Cook <keescook@chromium.org> wrote:
>
> Mark the various refcount_t functions with __signed_wrap, as we depend
> on the wrapping behavior to detect the overflow and perform saturation.
> Silences warnings seen with the LKDTM REFCOUNT_* tests:
>
>   UBSAN: signed-integer-overflow in ../include/linux/refcount.h:189:11
>   2147483647 + 1 cannot be represented in type 'int'
>
> Signed-off-by: Kees Cook <keescook@chromium.org>

Not sure why I am the "To:" (i.e. even if it is a change involving
only an addition of an attribute), but it looks good to me (UBSan is
triggering on the few `old + i`s caused by the calls from
`drivers/misc/lkdtm/refcount.c`, right?):

Reviewed-by: Miguel Ojeda <ojeda@kernel.org>

As usual, thanks Kees for keeping up on getting the kernel (un)signed
UBSan-clean :)

Cheers,
Miguel