sched_ext: Maintain an online cid mask in the scheduler arenagrafted

Schedulers on the default cid mapping treat [0, nr_online_cids) as the
online set and restart on hotplug. Schedulers that install their own mapping
with scx_bpf_cid_override() have no way to learn which cids are online: the
count no longer identifies members and the CPU-form cpumask is unusable from
cid programs. This is an obvious hole in the cid API.

Add scx_bpf_online_cmask(), a kernel-maintained cmask in the scheduler's
arena, allocated alongside the per-CPU scratch masks and populated after the
cid mapping is finalized and before ops.init(), for child schedulers too.
The pointer stays valid through ops.exit() with no reference to take. It is
the arena offset as a void pointer, the same form struct_ops arena arguments
arrive in. The verifier types the void return as a scalar for the program's
arena cast.

The mask follows the SCX hotplug notifications: seeded from cpu_active_mask
and updated before ops.cid_online/offline() runs, so it lags cpu_online_mask
only inside a hotplug transition. Updates walk the scheduler list under the
lock that also serializes unlinking. Reads are live, not atomic snapshots.
Root initialization excludes hotplug.

v2: Reworded the getter kerneldoc (Andrea Righi).

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>

1 files changed, 229 insertions, 0 deletions

diff --git a/net/ipv4/xfrm4_input.c b/net/ipv4/xfrm4_input.c
new file mode 100644
index 000000000..f6f2a8ef3
--- /dev/null
+++ b/net/ipv4/xfrm4_input.c
@@ -0,0 +1,229 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * xfrm4_input.c
+ *
+ * Changes:
+ *	YOSHIFUJI Hideaki @USAGI
+ *		Split up af-specific portion
+ *	Derek Atkins <derek@ihtfp.com>
+ *		Add Encapsulation support
+ *
+ */
+
+#include <linux/slab.h>
+#include <linux/module.h>
+#include <linux/string.h>
+#include <linux/netfilter.h>
+#include <linux/netfilter_ipv4.h>
+#include <net/ip.h>
+#include <net/xfrm.h>
+#include <net/protocol.h>
+#include <net/gro.h>
+
+static int xfrm4_rcv_encap_finish2(struct net *net, struct sock *sk,
+				   struct sk_buff *skb)
+{
+	return dst_input(skb);
+}
+
+static inline int xfrm4_rcv_encap_finish(struct net *net, struct sock *sk,
+					 struct sk_buff *skb)
+{
+	if (!skb_dst(skb)) {
+		const struct iphdr *iph = ip_hdr(skb);
+
+		if (ip_route_input_noref(skb, iph->daddr, iph->saddr,
+					 ip4h_dscp(iph), skb->dev))
+			goto drop;
+	}
+
+	if (xfrm_trans_queue(skb, xfrm4_rcv_encap_finish2))
+		goto drop;
+
+	return 0;
+drop:
+	kfree_skb(skb);
+	return NET_RX_DROP;
+}
+
+int xfrm4_transport_finish(struct sk_buff *skb, int async)
+{
+	struct xfrm_offload *xo = xfrm_offload(skb);
+	struct iphdr *iph = ip_hdr(skb);
+	struct net_device *dev = skb->dev;
+
+	iph->protocol = XFRM_MODE_SKB_CB(skb)->protocol;
+
+#ifndef CONFIG_NETFILTER
+	if (!async)
+		return -iph->protocol;
+#endif
+
+	__skb_push(skb, -skb_network_offset(skb));
+	iph->tot_len = htons(skb->len);
+	ip_send_check(iph);
+
+	if (xo && (xo->flags & XFRM_GRO)) {
+		/* The full l2 header needs to be preserved so that re-injecting the packet at l2
+		 * works correctly in the presence of vlan tags.
+		 */
+		skb_mac_header_rebuild_full(skb, xo->orig_mac_len);
+		skb_reset_network_header(skb);
+		skb_reset_transport_header(skb);
+		return 0;
+	}
+
+	NF_HOOK(NFPROTO_IPV4, NF_INET_PRE_ROUTING,
+		dev_net(dev), NULL, skb, dev, NULL,
+		xfrm4_rcv_encap_finish);
+	return 0;
+}
+
+static int __xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb, bool pull)
+{
+	struct udp_sock *up = udp_sk(sk);
+	struct udphdr *uh;
+	struct iphdr *iph;
+	int iphlen, len;
+	__u8 *udpdata;
+	__be32 *udpdata32;
+	u16 encap_type;
+
+	encap_type = READ_ONCE(up->encap_type);
+	/* if this is not encapsulated socket, then just return now */
+	if (!encap_type)
+		return 1;
+
+	/* If this is a paged skb, make sure we pull up
+	 * whatever data we need to look at. */
+	len = skb->len - sizeof(struct udphdr);
+	if (!pskb_may_pull(skb, sizeof(struct udphdr) + min(len, 8)))
+		return 1;
+
+	/* Now we can get the pointers */
+	uh = udp_hdr(skb);
+	udpdata = (__u8 *)uh + sizeof(struct udphdr);
+	udpdata32 = (__be32 *)udpdata;
+
+	switch (encap_type) {
+	default:
+	case UDP_ENCAP_ESPINUDP:
+		/* Check if this is a keepalive packet.  If so, eat it. */
+		if (len == 1 && udpdata[0] == 0xff) {
+			return -EINVAL;
+		} else if (len > sizeof(struct ip_esp_hdr) && udpdata32[0] != 0) {
+			/* ESP Packet without Non-ESP header */
+			len = sizeof(struct udphdr);
+		} else
+			/* Must be an IKE packet.. pass it through */
+			return 1;
+		break;
+	}
+
+	/* At this point we are sure that this is an ESPinUDP packet,
+	 * so we need to remove 'len' bytes from the packet (the UDP
+	 * header and optional ESP marker bytes) and then modify the
+	 * protocol to ESP, and then call into the transform receiver.
+	 */
+	if (skb_unclone(skb, GFP_ATOMIC))
+		return -EINVAL;
+
+	/* Now we can update and verify the packet length... */
+	iph = ip_hdr(skb);
+	iphlen = iph->ihl << 2;
+	iph->tot_len = htons(ntohs(iph->tot_len) - len);
+	if (skb->len < iphlen + len) {
+		/* packet is too small!?! */
+		return -EINVAL;
+	}
+
+	/* pull the data buffer up to the ESP header and set the
+	 * transport header to point to ESP.  Keep UDP on the stack
+	 * for later.
+	 */
+	if (pull) {
+		__skb_pull(skb, len);
+		skb_reset_transport_header(skb);
+	} else {
+		skb_set_transport_header(skb, len);
+	}
+
+	/* process ESP */
+	return 0;
+}
+
+/* If it's a keepalive packet, then just eat it.
+ * If it's an encapsulated packet, then pass it to the
+ * IPsec xfrm input.
+ * Returns 0 if skb passed to xfrm or was dropped.
+ * Returns >0 if skb should be passed to UDP.
+ * Returns <0 if skb should be resubmitted (-ret is protocol)
+ */
+int xfrm4_udp_encap_rcv(struct sock *sk, struct sk_buff *skb)
+{
+	int ret;
+
+	ret = __xfrm4_udp_encap_rcv(sk, skb, true);
+	if (!ret)
+		return xfrm4_rcv_encap(skb, IPPROTO_ESP, 0,
+				       udp_sk(sk)->encap_type);
+
+	if (ret < 0) {
+		kfree_skb(skb);
+		return 0;
+	}
+
+	return ret;
+}
+EXPORT_SYMBOL(xfrm4_udp_encap_rcv);
+
+struct sk_buff *xfrm4_gro_udp_encap_rcv(struct sock *sk, struct list_head *head,
+					struct sk_buff *skb)
+{
+	int offset = skb_gro_offset(skb);
+	const struct net_offload *ops;
+	struct sk_buff *pp = NULL;
+	int len, dlen;
+	__u8 *udpdata;
+	__be32 *udpdata32;
+
+	len = skb->len - offset;
+	dlen = offset + min(len, 8);
+	udpdata = skb_gro_header(skb, dlen, offset);
+	udpdata32 = (__be32 *)udpdata;
+	if (unlikely(!udpdata))
+		return NULL;
+
+	rcu_read_lock();
+	ops = rcu_dereference(inet_offloads[IPPROTO_ESP]);
+	if (!ops || !ops->callbacks.gro_receive)
+		goto out;
+
+	/* check if it is a keepalive or IKE packet */
+	if (len <= sizeof(struct ip_esp_hdr) || udpdata32[0] == 0)
+		goto out;
+
+	/* set the transport header to ESP */
+	skb_set_transport_header(skb, offset);
+
+	NAPI_GRO_CB(skb)->proto = IPPROTO_UDP;
+
+	pp = call_gro_receive(ops->callbacks.gro_receive, head, skb);
+	rcu_read_unlock();
+
+	return pp;
+
+out:
+	rcu_read_unlock();
+	NAPI_GRO_CB(skb)->same_flow = 0;
+	NAPI_GRO_CB(skb)->flush = 1;
+
+	return NULL;
+}
+EXPORT_SYMBOL(xfrm4_gro_udp_encap_rcv);
+
+int xfrm4_rcv(struct sk_buff *skb)
+{
+	return xfrm4_rcv_spi(skb, ip_hdr(skb)->protocol, 0);
+}
+EXPORT_SYMBOL(xfrm4_rcv);