Re: [PATCH bpf-next] bpf: Use struct_size() helper

[Yonghong Song]

On 12/20/21 3:30 AM, Xiu Jianfeng wrote:
> In an effort to avoid open-coded arithmetic in the kernel, use the
> struct_size() helper instead of open-coded calculation.
>
> Link: https://github.com/KSPP/linux/issues/160
> Signed-off-by: Xiu Jianfeng <xiujianfeng@xxxxxxxxxx>

Ack with a minor comments below.

Acked-by: Yonghong Song <yhs@xxxxxx>

> ---
>   kernel/bpf/local_storage.c   | 3 +--
>   kernel/bpf/reuseport_array.c | 6 +-----
>   2 files changed, 2 insertions(+), 7 deletions(-)
>
> diff --git a/kernel/bpf/local_storage.c b/kernel/bpf/local_storage.c
> index 035e9e3a7132..23f7f9d08a62 100644
> --- a/kernel/bpf/local_storage.c
> +++ b/kernel/bpf/local_storage.c
> @@ -163,8 +163,7 @@ static int cgroup_storage_update_elem(struct bpf_map *map, void *key,
>   		return 0;
>   	}
>   
> -	new = bpf_map_kmalloc_node(map, sizeof(struct bpf_storage_buffer) +
> -				   map->value_size,
> +	new = bpf_map_kmalloc_node(map, struct_size(new, data, map->value_size),
>   				   __GFP_ZERO | GFP_ATOMIC | __GFP_NOWARN,
>   				   map->numa_node);
>   	if (!new)
> diff --git a/kernel/bpf/reuseport_array.c b/kernel/bpf/reuseport_array.c
> index 93a55391791a..556a769b5b80 100644
> --- a/kernel/bpf/reuseport_array.c
> +++ b/kernel/bpf/reuseport_array.c
> @@ -152,16 +152,12 @@ static struct bpf_map *reuseport_array_alloc(union bpf_attr *attr)
>   {
>   	int numa_node = bpf_map_attr_numa_node(attr);
>   	struct reuseport_array *array;
> -	u64 array_size;
>   
>   	if (!bpf_capable())
>   		return ERR_PTR(-EPERM);
>   
> -	array_size = sizeof(*array);
> -	array_size += (u64)attr->max_entries * sizeof(struct sock *);

We have u64 type conversion here while struct_size used type 'size_t'. 
So we won't have problems for 64bit system. But for 32bit system, we may
have a slight different behavior here as the current code can return a
u64 value and the struct_size(...) returns a size_t type value with max 
value SIZE_MAX.

In __bpf_map_area_alloc() we have a check:
        if (size >= SIZE_MAX)
                return NULL;
so we should be fine for 32bit system.

> -
>   	/* allocate all map elements and zero-initialize them */
> -	array = bpf_map_area_alloc(array_size, numa_node);
> +	array = bpf_map_area_alloc(struct_size(array, ptrs, attr->max_entries), numa_node);
>   	if (!array)
>   		return ERR_PTR(-ENOMEM);
>