net: sched: cls_u32: Avoid memcpy() false-positive warning
authorKees Cook <keescook@chromium.org>
Tue, 27 Sep 2022 15:37:01 +0000 (08:37 -0700)
committerJakub Kicinski <kuba@kernel.org>
Fri, 30 Sep 2022 01:44:07 +0000 (18:44 -0700)
To work around a misbehavior of the compiler's ability to see into
composite flexible array structs (as detailed in the coming memcpy()
hardening series[1]), use unsafe_memcpy(), as the sizing,
bounds-checking, and allocation are all very tightly coupled here.
This silences the false-positive reported by syzbot:

  memcpy: detected field-spanning write (size 80) of single field "&n->sel" at net/sched/cls_u32.c:1043 (size 16)

[1] https://lore.kernel.org/linux-hardening/20220901065914.1417829-2-keescook@chromium.org

Cc: Cong Wang <xiyou.wangcong@gmail.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Reported-by: syzbot+a2c4601efc75848ba321@syzkaller.appspotmail.com
Link: https://lore.kernel.org/lkml/000000000000a96c0b05e97f0444@google.com/
Signed-off-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Jamal Hadi Salim <jhs@mojatatu.com>
Link: https://lore.kernel.org/r/20220927153700.3071688-1-keescook@chromium.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
net/sched/cls_u32.c

index 58c7680..0b3d909 100644 (file)
@@ -1040,7 +1040,11 @@ static int u32_change(struct net *net, struct sk_buff *in_skb,
        }
 #endif
 
-       memcpy(&n->sel, s, sel_size);
+       unsafe_memcpy(&n->sel, s, sel_size,
+                     /* A composite flex-array structure destination,
+                      * which was correctly sized with struct_size(),
+                      * bounds-checked against nla_len(), and allocated
+                      * above. */);
        RCU_INIT_POINTER(n->ht_up, ht);
        n->handle = handle;
        n->fshift = s->hmask ? ffs(ntohl(s->hmask)) - 1 : 0;