On Thu, 4 Dec 2014 16:20:52 +0100
Rasmus Villemoes <linux@xxxxxxxxxxxxxxxxxx> wrote:
> Bugs similar to the one in acbbe6fbb240 (kcmp: fix standard comparison
> bug) are in rich supply.
>
> In this variant, the problem is that struct xdr_netobj::len has type
> unsigned int, so the expression o1->len - o2->len _also_ has type
> unsigned int; it has completely well-defined semantics, and the result
> is some non-negative integer, which is always representable in a long
> long. But this means that if the conditional triggers, we are
> guaranteed to return a positive value from compare_blob.
>
> In this case it could be fixed by
>
> - res = o1->len - o2->len;
> + res = (long long)o1->len - (long long)o2->len;
>
> but I'd rather eliminate the usually broken 'return a - b;' idiom.
>
> Signed-off-by: Rasmus Villemoes <linux@xxxxxxxxxxxxxxxxxx>
> ---
>
> Notes:
> How this could ever have worked is beyond me - compare_blob seems to
> be used to maintain an rbtree, and I wouldn't expect rbtrees to behave
> well if the comparison function doesn't satisfy the basic invariant
> sign(cmp(a, b)) == -sign(cmp(b, a)).
>
> fs/nfsd/nfs4state.c | 15 +++++++--------
> 1 file changed, 7 insertions(+), 8 deletions(-)
>
> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> index e9c3afe4b5d3..d504cd6927f8 100644
> --- a/fs/nfsd/nfs4state.c
> +++ b/fs/nfsd/nfs4state.c
> @@ -1711,15 +1711,14 @@ static int copy_cred(struct svc_cred *target, struct svc_cred *source)
> return 0;
> }
>
> -static long long
> +static int
> compare_blob(const struct xdr_netobj *o1, const struct xdr_netobj *o2)
> {
> - long long res;
> -
> - res = o1->len - o2->len;
> - if (res)
> - return res;
> - return (long long)memcmp(o1->data, o2->data, o1->len);
> + if (o1->len < o2->len)
> + return -1;
> + if (o1->len > o2->len)
> + return 1;
> + return memcmp(o1->data, o2->data, o1->len);
> }
>
> static int same_name(const char *n1, const char *n2)
> @@ -1907,7 +1906,7 @@ add_clp_to_name_tree(struct nfs4_client *new_clp, struct rb_root *root)
> static struct nfs4_client *
> find_clp_in_name_tree(struct xdr_netobj *name, struct rb_root *root)
> {
> - long long cmp;
> + int cmp;
> struct rb_node *node = root->rb_node;
> struct nfs4_client *clp;
>
Ouch! Well spotted. I'd have probably stuck with just casting the
lengths to long longs, but this is not a particularly hot codepath and
so I'm ok with this.
This should probably also go to stable.
Reviewed-by: Jeff Layton <jlayton@xxxxxxxxxxxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
