Re: [PATCH 4.19 13/36] crypto: af_alg - fix use-after-free in af_alg_accept() due to bh_lock_sock()

[Herbert Xu]

On Tue, Jul 07, 2020 at 11:25:31PM +0200, Pavel Machek wrote:
>
> > @@ -308,12 +302,14 @@ int af_alg_accept(struct sock *sk, struc
> >  
> >  	sk2->sk_family = PF_ALG;
> >  
> > -	if (nokey || !ask->refcnt++)
> > +	if (atomic_inc_return_relaxed(&ask->refcnt) == 1)
> >  		sock_hold(sk);
> > -	ask->nokey_refcnt += nokey;
> > +	if (nokey) {
> > +		atomic_inc(&ask->nokey_refcnt);
> > +		atomic_set(&alg_sk(sk2)->nokey_refcnt, 1);
> > +	}
> 
> Should we set the nokey_refcnt to 0 using atomic_set, too?
> Aternatively, should the nokey_refcnt be initialized using
> ATOMIC_INIT()?

What are you asking for? It's already being set with atomic_set.
Or are you asking it to be set to 0 instead of 1? No it needs to
be 1 for the socket destructor.

Cheers,
-- 
Email: Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt