Re: [PATCH] cxgb4: Convert from atomic_t to refcount_t on l2t_entry->refcnt

[Leon Romanovsky <leon@kernel.org>]

On Sat, Jul 17, 2021 at 06:16:54PM +0800, Xiyu Yang wrote:
> refcount_t type and corresponding API can protect refcounters from
> accidental underflow and overflow and further use-after-free situations.
> 
> Signed-off-by: Xiyu Yang <xiyuyang19@fudan.edu.cn>
> Signed-off-by: Xin Tan <tanxin.ctf@gmail.com>
> ---
>  drivers/net/ethernet/chelsio/cxgb4/l2t.c | 31 ++++++++++++++++---------------
>  drivers/net/ethernet/chelsio/cxgb4/l2t.h |  3 ++-
>  2 files changed, 18 insertions(+), 16 deletions(-)
> 
> diff --git a/drivers/net/ethernet/chelsio/cxgb4/l2t.c b/drivers/net/ethernet/chelsio/cxgb4/l2t.c
> index a10a6862a9a4..cb26a5e315b1 100644
> --- a/drivers/net/ethernet/chelsio/cxgb4/l2t.c
> +++ b/drivers/net/ethernet/chelsio/cxgb4/l2t.c
> @@ -69,7 +69,8 @@ static inline unsigned int vlan_prio(const struct l2t_entry *e)
>  
>  static inline void l2t_hold(struct l2t_data *d, struct l2t_entry *e)
>  {
> -	if (atomic_add_return(1, &e->refcnt) == 1)  /* 0 -> 1 transition */
> +	refcount_inc(&e->refcnt);
> +	if (refcount_read(&e->refcnt) == 1)  /* 0 -> 1 transition */
>  		atomic_dec(&d->nfree);
>  }
>  
> @@ -270,10 +271,10 @@ static struct l2t_entry *alloc_l2e(struct l2t_data *d)
>  
>  	/* there's definitely a free entry */
>  	for (e = d->rover, end = &d->l2tab[d->l2t_size]; e != end; ++e)
> -		if (atomic_read(&e->refcnt) == 0)
> +		if (refcount_read(&e->refcnt) == 0)

This is wrong.

Thanks