KVM: SEV: Do cache maintenance on the source VM during intra-host migrationgrafted

Manually perform cache maintenance on the source VM during intra-host
migration to ensure no stale data is left in CPU caches after the VM is
destroyed.  Because the source VM is "converted" to a non-SEV VM, KVM's
memory reclaim flows won't trigger cache maintenance, e.g. when all guest
memory is reclaimed in response to detaching from the mmu_notifier.

Note, relying on the destination VM to do cache maintenance isn't an option
as KVM doesn't require identical guest memory configurations, i.e. the
source VM may have access to memory that the destination VM does not.
Enforcing equivalent memory configurations is infeasible, as it would
require a *deep* comparison of memslots, e.g. to verify that not only are
the memslot identical, but what the memslots point at is also identical.

Fixes: b56639318bb2 ("KVM: SEV: Add support for SEV intra host migration")
Cc: stable@vger.kernel.org
Reported-by: Stefan Teodorescu <fane@google.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260923163721.1584779-3-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

1 files changed, 88 insertions, 0 deletions

diff --git a/include/net/cls_cgroup.h b/include/net/cls_cgroup.h
new file mode 100644
index 000000000..668aeee9b
--- /dev/null
+++ b/include/net/cls_cgroup.h
@@ -0,0 +1,88 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * cls_cgroup.h			Control Group Classifier
+ *
+ * Authors:	Thomas Graf <tgraf@suug.ch>
+ */
+
+#ifndef _NET_CLS_CGROUP_H
+#define _NET_CLS_CGROUP_H
+
+#include <linux/cgroup.h>
+#include <linux/hardirq.h>
+#include <linux/rcupdate.h>
+#include <net/sock.h>
+#include <net/inet_sock.h>
+
+#ifdef CONFIG_CGROUP_NET_CLASSID
+struct cgroup_cls_state {
+	struct cgroup_subsys_state css;
+	u32 classid;
+};
+
+struct cgroup_cls_state *task_cls_state(struct task_struct *p);
+
+static inline u32 task_cls_classid(struct task_struct *p)
+{
+	u32 classid;
+
+	if (in_interrupt())
+		return 0;
+
+	rcu_read_lock();
+	classid = container_of(task_css(p, net_cls_cgrp_id),
+			       struct cgroup_cls_state, css)->classid;
+	rcu_read_unlock();
+
+	return classid;
+}
+
+static inline void sock_update_classid(struct sock_cgroup_data *skcd)
+{
+	u32 classid;
+
+	classid = task_cls_classid(current);
+	sock_cgroup_set_classid(skcd, classid);
+}
+
+static inline u32 __task_get_classid(struct task_struct *task)
+{
+	return task_cls_state(task)->classid;
+}
+
+static inline u32 task_get_classid(const struct sk_buff *skb)
+{
+	u32 classid = __task_get_classid(current);
+
+	/* Due to the nature of the classifier it is required to ignore all
+	 * packets originating from softirq context as accessing `current'
+	 * would lead to false results.
+	 *
+	 * This test assumes that all callers of dev_queue_xmit() explicitly
+	 * disable bh. Knowing this, it is possible to detect softirq based
+	 * calls by looking at the number of nested bh disable calls because
+	 * softirqs always disables bh.
+	 */
+	if (softirq_count()) {
+		struct sock *sk = skb_to_full_sk(skb);
+
+		/* If there is an sock_cgroup_classid we'll use that. */
+		if (!sk || !sk_fullsock(sk))
+			return 0;
+
+		classid = sock_cgroup_classid(&sk->sk_cgrp_data);
+	}
+
+	return classid;
+}
+#else /* !CONFIG_CGROUP_NET_CLASSID */
+static inline void sock_update_classid(struct sock_cgroup_data *skcd)
+{
+}
+
+static inline u32 task_get_classid(const struct sk_buff *skb)
+{
+	return 0;
+}
+#endif /* CONFIG_CGROUP_NET_CLASSID */
+#endif  /* _NET_CLS_CGROUP_H */