net/af_xdp: fix integer overflow in umem size calculation
authorMartin Weiser <martin.weiser@allegro-packets.com>
Thu, 29 Oct 2020 11:25:42 +0000 (12:25 +0100)
committerFerruh Yigit <ferruh.yigit@intel.com>
Tue, 3 Nov 2020 22:35:06 +0000 (23:35 +0100)
The multiplication of two u32 integers may cause an overflow with large
mempool sizes.

Fixes: 74b46340e2d4 ("net/af_xdp: support shared UMEM")

Signed-off-by: Martin Weiser <martin.weiser@allegro-packets.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
drivers/net/af_xdp/rte_eth_af_xdp.c

index df2767b..4076ff7 100644 (file)
@@ -968,7 +968,8 @@ xsk_umem_info *xdp_umem_configure(struct pmd_internals *internals,
 
                umem->mb_pool = mb_pool;
                base_addr = (void *)get_base_addr(mb_pool, &align);
-               umem_size = mb_pool->populated_size * usr_config.frame_size +
+               umem_size = (uint64_t)mb_pool->populated_size *
+                               (uint64_t)usr_config.frame_size +
                                align;
 
                ret = xsk_umem__create(&umem->umem, base_addr, umem_size,