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, 120 insertions, 0 deletions

diff --git a/samples/pidfd/pidfd-metadata.c b/samples/pidfd/pidfd-metadata.c
new file mode 100644
index 000000000..c459155da
--- /dev/null
+++ b/samples/pidfd/pidfd-metadata.c
@@ -0,0 +1,120 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#define _GNU_SOURCE
+#include <err.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <inttypes.h>
+#include <limits.h>
+#include <sched.h>
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/stat.h>
+#include <sys/syscall.h>
+#include <sys/types.h>
+#include <sys/wait.h>
+#include <unistd.h>
+
+#ifndef CLONE_PIDFD
+#define CLONE_PIDFD 0x00001000
+#endif
+
+#ifndef __NR_pidfd_send_signal
+#define __NR_pidfd_send_signal -1
+#endif
+
+static int do_child(void *args)
+{
+	printf("%d\n", getpid());
+	_exit(EXIT_SUCCESS);
+}
+
+static pid_t pidfd_clone(int flags, int *pidfd)
+{
+	size_t stack_size = 1024;
+	char *stack[1024] = { 0 };
+
+#ifdef __ia64__
+	return __clone2(do_child, stack, stack_size, flags | SIGCHLD, NULL, pidfd);
+#else
+	return clone(do_child, stack + stack_size, flags | SIGCHLD, NULL, pidfd);
+#endif
+}
+
+static inline int sys_pidfd_send_signal(int pidfd, int sig, siginfo_t *info,
+					unsigned int flags)
+{
+	return syscall(__NR_pidfd_send_signal, pidfd, sig, info, flags);
+}
+
+static int pidfd_metadata_fd(pid_t pid, int pidfd)
+{
+	int procfd, ret;
+	char path[100];
+
+	snprintf(path, sizeof(path), "/proc/%d", pid);
+	procfd = open(path, O_DIRECTORY | O_RDONLY | O_CLOEXEC);
+	if (procfd < 0) {
+		warn("Failed to open %s\n", path);
+		return -1;
+	}
+
+	/*
+	 * Verify that the pid has not been recycled and our /proc/<pid> handle
+	 * is still valid.
+	 */
+	ret = sys_pidfd_send_signal(pidfd, 0, NULL, 0);
+	if (ret < 0) {
+		switch (errno) {
+		case EPERM:
+			/* Process exists, just not allowed to signal it. */
+			break;
+		default:
+			warn("Failed to signal process\n");
+			close(procfd);
+			procfd = -1;
+		}
+	}
+
+	return procfd;
+}
+
+int main(int argc, char *argv[])
+{
+	int pidfd = -1, ret = EXIT_FAILURE;
+	char buf[4096] = { 0 };
+	pid_t pid;
+	int procfd, statusfd;
+	ssize_t bytes;
+
+	pid = pidfd_clone(CLONE_PIDFD, &pidfd);
+	if (pid < 0)
+		err(ret, "CLONE_PIDFD");
+	if (pidfd == -1) {
+		warnx("CLONE_PIDFD is not supported by the kernel");
+		goto out;
+	}
+
+	procfd = pidfd_metadata_fd(pid, pidfd);
+	close(pidfd);
+	if (procfd < 0)
+		goto out;
+
+	statusfd = openat(procfd, "status", O_RDONLY | O_CLOEXEC);
+	close(procfd);
+	if (statusfd < 0)
+		goto out;
+
+	bytes = read(statusfd, buf, sizeof(buf));
+	if (bytes > 0)
+		bytes = write(STDOUT_FILENO, buf, bytes);
+	close(statusfd);
+	ret = EXIT_SUCCESS;
+
+out:
+	(void)wait(NULL);
+
+	exit(ret);
+}