netdev - RE: [PATCH bpf-next 5/6] bpf: sockmap: allow update from BPF

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Wed, 19 Aug 2020 15:41:21 -0700
From:   John Fastabend <john.fastabend@...il.com>
To:     John Fastabend <john.fastabend@...il.com>,
        Lorenz Bauer <lmb@...udflare.com>, jakub@...udflare.com,
        john.fastabend@...il.com, Alexei Starovoitov <ast@...nel.org>,
        Daniel Borkmann <daniel@...earbox.net>,
        Lorenz Bauer <lmb@...udflare.com>,
        "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>
Cc:     kernel-team@...udflare.com, netdev@...r.kernel.org,
        bpf@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: RE: [PATCH bpf-next 5/6] bpf: sockmap: allow update from BPF

John Fastabend wrote:
> Lorenz Bauer wrote:
> > Allow calling bpf_map_update_elem on sockmap and sockhash from a BPF
> > context. The synchronization required for this is a bit fiddly: we
> > need to prevent the socket from changing it's state while we add it
> > to the sockmap, since we rely on getting a callback via
> > sk_prot->unhash. However, we can't just lock_sock like in
> > sock_map_sk_acquire because that might sleep. So instead we disable
> > softirq processing and use bh_lock_sock to prevent further
> > modification.
> > 
> > Signed-off-by: Lorenz Bauer <lmb@...udflare.com>
> > ---
> >  kernel/bpf/verifier.c |  6 ++++--
> >  net/core/sock_map.c   | 24 ++++++++++++++++++++++++
> >  2 files changed, 28 insertions(+), 2 deletions(-)
> > 
> > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> > index 47f9b94bb9d4..421fccf18dea 100644
> > --- a/kernel/bpf/verifier.c
> > +++ b/kernel/bpf/verifier.c
> > @@ -4254,7 +4254,8 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env,
> >  		    func_id != BPF_FUNC_map_delete_elem &&
> >  		    func_id != BPF_FUNC_msg_redirect_map &&
> >  		    func_id != BPF_FUNC_sk_select_reuseport &&
> > -		    func_id != BPF_FUNC_map_lookup_elem)
> > +		    func_id != BPF_FUNC_map_lookup_elem &&
> > +		    func_id != BPF_FUNC_map_update_elem)
> >  			goto error;
> >  		break;
> >  	case BPF_MAP_TYPE_SOCKHASH:
> > @@ -4263,7 +4264,8 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env,
> >  		    func_id != BPF_FUNC_map_delete_elem &&
> >  		    func_id != BPF_FUNC_msg_redirect_hash &&
> >  		    func_id != BPF_FUNC_sk_select_reuseport &&
> > -		    func_id != BPF_FUNC_map_lookup_elem)
> > +		    func_id != BPF_FUNC_map_lookup_elem &&
> > +		    func_id != BPF_FUNC_map_update_elem)
> 
> I lost track of a detail here, map_lookup_elem should return
> PTR_TO_MAP_VALUE_OR_NULL but if we want to feed that back into
> the map_update_elem() we need to return PTR_TO_SOCKET_OR_NULL
> and then presumably have a null check to get a PTR_TO_SOCKET
> type as expect.
> 
> Can we use the same logic for expected arg (previous patch) on the
> ret_type. Or did I miss it:/ Need some coffee I guess.

OK, I tracked this down. It looks like we rely on mark_ptr_or_null_reg()
to update the reg->tyype to PTR_TO_SOCKET. I do wonder if it would be
a bit more straight forward to do something similar to the previous
patch and refine it earlier to PTR_TO_SOCKET_OR_NULL, but should be
safe as-is for now.

I still have the below question though.

> 
> >  			goto error;
> >  		break;
> >  	case BPF_MAP_TYPE_REUSEPORT_SOCKARRAY:
> > diff --git a/net/core/sock_map.c b/net/core/sock_map.c
> > index 018367fb889f..b2c886c34566 100644
> > --- a/net/core/sock_map.c
> > +++ b/net/core/sock_map.c
> > @@ -603,6 +603,28 @@ int sock_map_update_elem_sys(struct bpf_map *map, void *key,
> >  	return ret;
> >  }
> >  
> > +static int sock_map_update_elem(struct bpf_map *map, void *key,
> > +				void *value, u64 flags)
> > +{
> > +	struct sock *sk = (struct sock *)value;
> > +	int ret;
> > +
> > +	if (!sock_map_sk_is_suitable(sk))
> > +		return -EOPNOTSUPP;
> > +
> > +	local_bh_disable();
> > +	bh_lock_sock(sk);
> 
> How do ensure we are not being called from some context which
> already has the bh_lock_sock() held? It seems we can call map_update_elem()
> from any context, kprobes, tc, xdp, etc.?
> 
> > +	if (!sock_map_sk_state_allowed(sk))
> > +		ret = -EOPNOTSUPP;
> > +	else if (map->map_type == BPF_MAP_TYPE_SOCKMAP)
> > +		ret = sock_map_update_common(map, *(u32 *)key, sk, flags);
> > +	else
> > +		ret = sock_hash_update_common(map, key, sk, flags);
> > +	bh_unlock_sock(sk);
> > +	local_bh_enable();
> > +	return ret;
> > +}
> > +
> >  BPF_CALL_4(bpf_sock_map_update, struct bpf_sock_ops_kern *, sops,
> >  	   struct bpf_map *, map, void *, key, u64, flags)
> >  {
> > @@ -687,6 +709,7 @@ const struct bpf_map_ops sock_map_ops = {
> >  	.map_free		= sock_map_free,
> >  	.map_get_next_key	= sock_map_get_next_key,
> >  	.map_lookup_elem_sys_only = sock_map_lookup_sys,
> > +	.map_update_elem	= sock_map_update_elem,
> >  	.map_delete_elem	= sock_map_delete_elem,
> >  	.map_lookup_elem	= sock_map_lookup,
> >  	.map_release_uref	= sock_map_release_progs,
> > @@ -1180,6 +1203,7 @@ const struct bpf_map_ops sock_hash_ops = {
> >  	.map_alloc		= sock_hash_alloc,
> >  	.map_free		= sock_hash_free,
> >  	.map_get_next_key	= sock_hash_get_next_key,
> > +	.map_update_elem	= sock_map_update_elem,
> >  	.map_delete_elem	= sock_hash_delete_elem,
> >  	.map_lookup_elem	= sock_hash_lookup,
> >  	.map_lookup_elem_sys_only = sock_hash_lookup_sys,
> > -- 
> > 2.25.1
> > 
> 
>