nfp: reject binding to shared blocks

TC shared blocks allow multiple qdiscs to be grouped together and filters
shared between them. Currently the chains of filters attached to a block
are only flushed when the block is removed. If a qdisc is removed from a
block but the block still exists, flow del messages are not passed to the
callback registered for that qdisc. For the NFP, this presents the
possibility of rules still existing in hw when they should be removed.

Prevent binding to shared blocks until the kernel can send per qdisc del
messages when block unbinds occur.

tcf_block_shared() was not used outside of the core until now, so also
add an empty implementation for builds with CONFIG_NET_CLS=n.

Fixes: 4861738 ("net: sched: introduce shared filter blocks infrastructure")
Signed-off-by: John Hurley <john.hurley@netronome.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/netronome/nfp/bpf/main.c

@@ -202,6 +202,9 @@ static int nfp_bpf_setup_tc_block(struct net_device *netdev,
 	if (f->binder_type != TCF_BLOCK_BINDER_TYPE_CLSACT_INGRESS)
 		return -EOPNOTSUPP;
 
+	if (tcf_block_shared(f->block))
+		return -EOPNOTSUPP;
+
 	switch (f->command) {
 	case TC_BLOCK_BIND:
 		return tcf_block_cb_register(f->block,

drivers/net/ethernet/netronome/nfp/flower/offload.c

@@ -631,6 +631,9 @@ static int nfp_flower_setup_tc_block(struct net_device *netdev,
 	if (f->binder_type != TCF_BLOCK_BINDER_TYPE_CLSACT_INGRESS)
 		return -EOPNOTSUPP;
 
+	if (tcf_block_shared(f->block))
+		return -EOPNOTSUPP;
+
 	switch (f->command) {
 	case TC_BLOCK_BIND:
 		return tcf_block_cb_register(f->block,

include/net/pkt_cls.h

@@ -111,6 +111,11 @@ void tcf_block_put_ext(struct tcf_block *block, struct Qdisc *q,
 {
 }
 
+static inline bool tcf_block_shared(struct tcf_block *block)
+{
+	return false;
+}
+
 static inline struct Qdisc *tcf_block_q(struct tcf_block *block)
 {
 	return NULL;