bpf: Convert bpf_prog refcnt to atomic64_t

Similarly to bpf_map's refcnt/usercnt, convert bpf_prog's refcnt to atomic64
and remove artificial 32k limit. This allows to make bpf_prog's refcounting
non-failing, simplifying logic of users of bpf_prog_add/bpf_prog_inc.

Validated compilation by running allyesconfig kernel build.

Suggested-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20191117172806.2195367-3-andriin@fb.com

drivers/net/ethernet/broadcom/bnxt/bnxt.c

@@ -3171,13 +3171,8 @@ static int bnxt_init_one_rx_ring(struct bnxt *bp, int ring_nr)
 	bnxt_init_rxbd_pages(ring, type);
 
 	if (BNXT_RX_PAGE_MODE(bp) && bp->xdp_prog) {
-		rxr->xdp_prog = bpf_prog_add(bp->xdp_prog, 1);
+		bpf_prog_add(bp->xdp_prog, 1);
-		if (IS_ERR(rxr->xdp_prog)) {
+		rxr->xdp_prog = bp->xdp_prog;
-			int rc = PTR_ERR(rxr->xdp_prog);
-
-			rxr->xdp_prog = NULL;
-			return rc;
-		}
 	}
 	prod = rxr->rx_prod;
 	for (i = 0; i < bp->rx_ring_size; i++) {

drivers/net/ethernet/cavium/thunder/nicvf_main.c

@@ -1876,13 +1876,8 @@ static int nicvf_xdp_setup(struct nicvf *nic, struct bpf_prog *prog)
 
 	if (nic->xdp_prog) {
 		/* Attach BPF program */
-		nic->xdp_prog = bpf_prog_add(nic->xdp_prog, nic->rx_queues - 1);
+		bpf_prog_add(nic->xdp_prog, nic->rx_queues - 1);
-		if (!IS_ERR(nic->xdp_prog)) {
+		bpf_attached = true;
-			bpf_attached = true;
-		} else {
-			ret = PTR_ERR(nic->xdp_prog);
-			nic->xdp_prog = NULL;
-		}
 	}
 
 	/* Calculate Tx queues needed for XDP and network stack */

drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c

@@ -1807,11 +1807,8 @@ static int setup_xdp(struct net_device *dev, struct bpf_prog *prog)
 	if (prog && !xdp_mtu_valid(priv, dev->mtu))
 		return -EINVAL;
 
-	if (prog) {
+	if (prog)
-		prog = bpf_prog_add(prog, priv->num_channels);
+		bpf_prog_add(prog, priv->num_channels);
-		if (IS_ERR(prog))
-			return PTR_ERR(prog);
-	}
 
 	up = netif_running(dev);
 	need_update = (!!priv->xdp_prog != !!prog);

drivers/net/ethernet/mellanox/mlx4/en_netdev.c

@@ -2286,11 +2286,7 @@ int mlx4_en_try_alloc_resources(struct mlx4_en_priv *priv,
 		lockdep_is_held(&priv->mdev->state_lock));
 
 	if (xdp_prog && carry_xdp_prog) {
-		xdp_prog = bpf_prog_add(xdp_prog, tmp->rx_ring_num);
+		bpf_prog_add(xdp_prog, tmp->rx_ring_num);
-		if (IS_ERR(xdp_prog)) {
-			mlx4_en_free_resources(tmp);
-			return PTR_ERR(xdp_prog);
-		}
 		for (i = 0; i < tmp->rx_ring_num; i++)
 			rcu_assign_pointer(tmp->rx_ring[i]->xdp_prog,
 					   xdp_prog);
@@ -2782,11 +2778,9 @@ static int mlx4_xdp_set(struct net_device *dev, struct bpf_prog *prog)
 	 * program for a new one.
 	 */
 	if (priv->tx_ring_num[TX_XDP] == xdp_ring_num) {
-		if (prog) {
+		if (prog)
-			prog = bpf_prog_add(prog, priv->rx_ring_num - 1);
+			bpf_prog_add(prog, priv->rx_ring_num - 1);
-			if (IS_ERR(prog))
+
-				return PTR_ERR(prog);
-		}
 		mutex_lock(&mdev->state_lock);
 		for (i = 0; i < priv->rx_ring_num; i++) {
 			old_prog = rcu_dereference_protected(
@@ -2807,13 +2801,8 @@ static int mlx4_xdp_set(struct net_device *dev, struct bpf_prog *prog)
 	if (!tmp)
 		return -ENOMEM;
 
-	if (prog) {
+	if (prog)
-		prog = bpf_prog_add(prog, priv->rx_ring_num - 1);
+		bpf_prog_add(prog, priv->rx_ring_num - 1);
-		if (IS_ERR(prog)) {
-			err = PTR_ERR(prog);
-			goto out;
-		}
-	}
 
 	mutex_lock(&mdev->state_lock);
 	memcpy(&new_prof, priv->prof, sizeof(struct mlx4_en_port_profile));
@@ -2862,7 +2851,6 @@ static int mlx4_xdp_set(struct net_device *dev, struct bpf_prog *prog)
 
 unlock_out:
 	mutex_unlock(&mdev->state_lock);
-out:
 	kfree(tmp);
 	return err;
 }

drivers/net/ethernet/mellanox/mlx5/core/en_main.c

@@ -408,12 +408,9 @@ static int mlx5e_alloc_rq(struct mlx5e_channel *c,
 		rq->stats = &c->priv->channel_stats[c->ix].rq;
 	INIT_WORK(&rq->recover_work, mlx5e_rq_err_cqe_work);
 
-	rq->xdp_prog = params->xdp_prog ? bpf_prog_inc(params->xdp_prog) : NULL;
+	if (params->xdp_prog)
-	if (IS_ERR(rq->xdp_prog)) {
+		bpf_prog_inc(params->xdp_prog);
-		err = PTR_ERR(rq->xdp_prog);
+	rq->xdp_prog = params->xdp_prog;
-		rq->xdp_prog = NULL;
-		goto err_rq_wq_destroy;
-	}
 
 	rq_xdp_ix = rq->ix;
 	if (xsk)
@@ -4406,16 +4403,11 @@ static int mlx5e_xdp_set(struct net_device *netdev, struct bpf_prog *prog)
 	/* no need for full reset when exchanging programs */
 	reset = (!priv->channels.params.xdp_prog || !prog);
 
-	if (was_opened && !reset) {
+	if (was_opened && !reset)
 		/* num_channels is invariant here, so we can take the
 		 * batched reference right upfront.
 		 */
-		prog = bpf_prog_add(prog, priv->channels.num);
+		bpf_prog_add(prog, priv->channels.num);
-		if (IS_ERR(prog)) {
-			err = PTR_ERR(prog);
-			goto unlock;
-		}
-	}
 
 	if (was_opened && reset) {
 		struct mlx5e_channels new_channels = {};

drivers/net/ethernet/qlogic/qede/qede_main.c

@@ -2107,12 +2107,8 @@ static int qede_start_queues(struct qede_dev *edev, bool clear_stats)
 			if (rc)
 				goto out;
 
-			fp->rxq->xdp_prog = bpf_prog_add(edev->xdp_prog, 1);
+			bpf_prog_add(edev->xdp_prog, 1);
-			if (IS_ERR(fp->rxq->xdp_prog)) {
+			fp->rxq->xdp_prog = edev->xdp_prog;
-				rc = PTR_ERR(fp->rxq->xdp_prog);
-				fp->rxq->xdp_prog = NULL;
-				goto out;
-			}
 		}
 
 		if (fp->type & QEDE_FASTPATH_TX) {

drivers/net/virtio_net.c

@@ -2445,11 +2445,8 @@ static int virtnet_xdp_set(struct net_device *dev, struct bpf_prog *prog,
 	if (!prog && !old_prog)
 		return 0;
 
-	if (prog) {
+	if (prog)
-		prog = bpf_prog_add(prog, vi->max_queue_pairs - 1);
+		bpf_prog_add(prog, vi->max_queue_pairs - 1);
-		if (IS_ERR(prog))
-			return PTR_ERR(prog);
-	}
 
 	/* Make sure NAPI is not using any XDP TX queues for RX. */
 	if (netif_running(dev)) {

include/linux/bpf.h

@@ -485,7 +485,7 @@ struct bpf_func_info_aux {
 };
 
 struct bpf_prog_aux {
-	atomic_t refcnt;
+	atomic64_t refcnt;
 	u32 used_map_cnt;
 	u32 max_ctx_offset;
 	u32 max_pkt_offset;
@@ -770,9 +770,9 @@ extern const struct bpf_verifier_ops xdp_analyzer_ops;
 struct bpf_prog *bpf_prog_get(u32 ufd);
 struct bpf_prog *bpf_prog_get_type_dev(u32 ufd, enum bpf_prog_type type,
 				       bool attach_drv);
-struct bpf_prog * __must_check bpf_prog_add(struct bpf_prog *prog, int i);
+void bpf_prog_add(struct bpf_prog *prog, int i);
 void bpf_prog_sub(struct bpf_prog *prog, int i);
-struct bpf_prog * __must_check bpf_prog_inc(struct bpf_prog *prog);
+void bpf_prog_inc(struct bpf_prog *prog);
 struct bpf_prog * __must_check bpf_prog_inc_not_zero(struct bpf_prog *prog);
 void bpf_prog_put(struct bpf_prog *prog);
 int __bpf_prog_charge(struct user_struct *user, u32 pages);
@@ -912,10 +912,8 @@ static inline struct bpf_prog *bpf_prog_get_type_dev(u32 ufd,
 	return ERR_PTR(-EOPNOTSUPP);
 }
 
-static inline struct bpf_prog * __must_check bpf_prog_add(struct bpf_prog *prog,
+static inline void bpf_prog_add(struct bpf_prog *prog, int i)
-							  int i)
 {
-	return ERR_PTR(-EOPNOTSUPP);
 }
 
 static inline void bpf_prog_sub(struct bpf_prog *prog, int i)
@@ -926,9 +924,8 @@ static inline void bpf_prog_put(struct bpf_prog *prog)
 {
 }
 
-static inline struct bpf_prog * __must_check bpf_prog_inc(struct bpf_prog *prog)
+static inline void bpf_prog_inc(struct bpf_prog *prog)
 {
-	return ERR_PTR(-EOPNOTSUPP);
 }
 
 static inline struct bpf_prog *__must_check

kernel/bpf/inode.c

@@ -31,7 +31,7 @@ static void *bpf_any_get(void *raw, enum bpf_type type)
 {
 	switch (type) {
 	case BPF_TYPE_PROG:
-		raw = bpf_prog_inc(raw);
+		bpf_prog_inc(raw);
 		break;
 	case BPF_TYPE_MAP:
 		bpf_map_inc_with_uref(raw);
@@ -534,7 +534,8 @@ static struct bpf_prog *__get_prog_inode(struct inode *inode, enum bpf_prog_type
 	if (!bpf_prog_get_ok(prog, &type, false))
 		return ERR_PTR(-EINVAL);
 
-	return bpf_prog_inc(prog);
+	bpf_prog_inc(prog);
+	return prog;
 }
 
 struct bpf_prog *bpf_prog_get_type_path(const char *name, enum bpf_prog_type type)

kernel/bpf/syscall.c

@@ -1339,7 +1339,7 @@ static void __bpf_prog_put_noref(struct bpf_prog *prog, bool deferred)
 
 static void __bpf_prog_put(struct bpf_prog *prog, bool do_idr_lock)
 {
-	if (atomic_dec_and_test(&prog->aux->refcnt)) {
+	if (atomic64_dec_and_test(&prog->aux->refcnt)) {
 		perf_event_bpf_event(prog, PERF_BPF_EVENT_PROG_UNLOAD, 0);
 		/* bpf_prog_free_id() must be called first */
 		bpf_prog_free_id(prog, do_idr_lock);
@@ -1445,16 +1445,9 @@ static struct bpf_prog *____bpf_prog_get(struct fd f)
 	return f.file->private_data;
 }
 
-/* prog's refcnt limit */
+void bpf_prog_add(struct bpf_prog *prog, int i)
-#define BPF_MAX_REFCNT 32768
-
-struct bpf_prog *bpf_prog_add(struct bpf_prog *prog, int i)
 {
-	if (atomic_add_return(i, &prog->aux->refcnt) > BPF_MAX_REFCNT) {
+	atomic64_add(i, &prog->aux->refcnt);
-		atomic_sub(i, &prog->aux->refcnt);
-		return ERR_PTR(-EBUSY);
-	}
-	return prog;
 }
 EXPORT_SYMBOL_GPL(bpf_prog_add);
 
@@ -1465,13 +1458,13 @@ void bpf_prog_sub(struct bpf_prog *prog, int i)
 	 * path holds a reference to the program, thus atomic_sub() can
 	 * be safely used in such cases!
 	 */
-	WARN_ON(atomic_sub_return(i, &prog->aux->refcnt) == 0);
+	WARN_ON(atomic64_sub_return(i, &prog->aux->refcnt) == 0);
 }
 EXPORT_SYMBOL_GPL(bpf_prog_sub);
 
-struct bpf_prog *bpf_prog_inc(struct bpf_prog *prog)
+void bpf_prog_inc(struct bpf_prog *prog)
 {
-	return bpf_prog_add(prog, 1);
+	atomic64_inc(&prog->aux->refcnt);
 }
 EXPORT_SYMBOL_GPL(bpf_prog_inc);
 
@@ -1480,12 +1473,7 @@ struct bpf_prog *bpf_prog_inc_not_zero(struct bpf_prog *prog)
 {
 	int refold;
 
-	refold = atomic_fetch_add_unless(&prog->aux->refcnt, 1, 0);
+	refold = atomic64_fetch_add_unless(&prog->aux->refcnt, 1, 0);
-
-	if (refold >= BPF_MAX_REFCNT) {
-		__bpf_prog_put(prog, false);
-		return ERR_PTR(-EBUSY);
-	}
 
 	if (!refold)
 		return ERR_PTR(-ENOENT);
@@ -1523,7 +1511,7 @@ static struct bpf_prog *__bpf_prog_get(u32 ufd, enum bpf_prog_type *attach_type,
 		goto out;
 	}
 
-	prog = bpf_prog_inc(prog);
+	bpf_prog_inc(prog);
 out:
 	fdput(f);
 	return prog;
@@ -1714,7 +1702,7 @@ static int bpf_prog_load(union bpf_attr *attr, union bpf_attr __user *uattr)
 	prog->orig_prog = NULL;
 	prog->jited = 0;
 
-	atomic_set(&prog->aux->refcnt, 1);
+	atomic64_set(&prog->aux->refcnt, 1);
 	prog->gpl_compatible = is_gpl ? 1 : 0;
 
 	if (bpf_prog_is_dev_bound(prog->aux)) {

kernel/events/core.c

@@ -10477,12 +10477,9 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu,
 		context = parent_event->overflow_handler_context;
 #if defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_EVENT_TRACING)
 		if (overflow_handler == bpf_overflow_handler) {
-			struct bpf_prog *prog = bpf_prog_inc(parent_event->prog);
+			struct bpf_prog *prog = parent_event->prog;
 
-			if (IS_ERR(prog)) {
+			bpf_prog_inc(prog);
-				err = PTR_ERR(prog);
-				goto err_ns;
-			}
 			event->prog = prog;
 			event->orig_overflow_handler =
 				parent_event->orig_overflow_handler;