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>

4 files changed, 419 insertions, 0 deletions

diff --git a/samples/cgroup/.gitignore b/samples/cgroup/.gitignore
new file mode 100644
index 000000000..3a0161194
--- /dev/null
+++ b/samples/cgroup/.gitignore
@@ -0,0 +1,3 @@
+/cgroup_event_listener
+/memcg_event_listener
+
diff --git a/samples/cgroup/Makefile b/samples/cgroup/Makefile
new file mode 100644
index 000000000..526c85697
--- /dev/null
+++ b/samples/cgroup/Makefile
@@ -0,0 +1,5 @@
+# SPDX-License-Identifier: GPL-2.0
+
+userprogs-always-y += cgroup_event_listener memcg_event_listener
+
+userccflags += -I usr/include
diff --git a/samples/cgroup/cgroup_event_listener.c b/samples/cgroup/cgroup_event_listener.c
new file mode 100644
index 000000000..3d70dc831
--- /dev/null
+++ b/samples/cgroup/cgroup_event_listener.c
@@ -0,0 +1,83 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * cgroup_event_listener.c - Simple listener of cgroup events
+ *
+ * Copyright (C) Kirill A. Shutemov <kirill@shutemov.name>
+ */
+
+#include <assert.h>
+#include <err.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <libgen.h>
+#include <limits.h>
+#include <stdio.h>
+#include <string.h>
+#include <unistd.h>
+
+#include <sys/eventfd.h>
+
+#define USAGE_STR "Usage: cgroup_event_listener <path-to-control-file> <args>"
+
+int main(int argc, char **argv)
+{
+	int efd = -1;
+	int cfd = -1;
+	int event_control = -1;
+	char event_control_path[PATH_MAX];
+	char line[LINE_MAX];
+	int ret;
+
+	if (argc != 3)
+		errx(1, "%s", USAGE_STR);
+
+	cfd = open(argv[1], O_RDONLY);
+	if (cfd == -1)
+		err(1, "Cannot open %s", argv[1]);
+
+	ret = snprintf(event_control_path, PATH_MAX, "%s/cgroup.event_control",
+			dirname(argv[1]));
+	if (ret >= PATH_MAX)
+		errx(1, "Path to cgroup.event_control is too long");
+
+	event_control = open(event_control_path, O_WRONLY);
+	if (event_control == -1)
+		err(1, "Cannot open %s", event_control_path);
+
+	efd = eventfd(0, 0);
+	if (efd == -1)
+		err(1, "eventfd() failed");
+
+	ret = snprintf(line, LINE_MAX, "%d %d %s", efd, cfd, argv[2]);
+	if (ret >= LINE_MAX)
+		errx(1, "Arguments string is too long");
+
+	ret = write(event_control, line, strlen(line) + 1);
+	if (ret == -1)
+		err(1, "Cannot write to cgroup.event_control");
+
+	while (1) {
+		uint64_t result;
+
+		ret = read(efd, &result, sizeof(result));
+		if (ret == -1) {
+			if (errno == EINTR)
+				continue;
+			err(1, "Cannot read from eventfd");
+		}
+		assert(ret == sizeof(result));
+
+		ret = access(event_control_path, W_OK);
+		if ((ret == -1) && (errno == ENOENT)) {
+			puts("The cgroup seems to have removed.");
+			break;
+		}
+
+		if (ret == -1)
+			err(1, "cgroup.event_control is not accessible any more");
+
+		printf("%s %s: crossed\n", argv[1], argv[2]);
+	}
+
+	return 0;
+}
diff --git a/samples/cgroup/memcg_event_listener.c b/samples/cgroup/memcg_event_listener.c
new file mode 100644
index 000000000..41425edbd
--- /dev/null
+++ b/samples/cgroup/memcg_event_listener.c
@@ -0,0 +1,328 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * memcg_event_listener.c - Simple listener of memcg memory.events
+ *
+ * Copyright (c) 2023, SaluteDevices. All Rights Reserved.
+ *
+ * Author: Dmitry Rokosov <ddrokosov@salutedevices.com>
+ */
+
+#include <err.h>
+#include <errno.h>
+#include <limits.h>
+#include <poll.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/inotify.h>
+#include <unistd.h>
+
+/* Size of buffer to use when reading inotify events */
+#define INOTIFY_BUFFER_SIZE 8192
+
+#define INOTIFY_EVENT_NEXT(event, length) ({         \
+	(length) -= sizeof(*(event)) + (event)->len; \
+	(event)++;                                   \
+})
+
+#define INOTIFY_EVENT_OK(event, length) ((length) >= (ssize_t)sizeof(*(event)))
+
+#define ARRAY_SIZE(arr) (sizeof(arr) / sizeof(arr[0]))
+
+struct memcg_counters {
+	long low;
+	long high;
+	long max;
+	long oom;
+	long oom_kill;
+	long oom_group_kill;
+};
+
+struct memcg_events {
+	struct memcg_counters counters;
+	char path[PATH_MAX];
+	int inotify_fd;
+	int inotify_wd;
+};
+
+static void print_memcg_counters(const struct memcg_counters *counters)
+{
+	printf("MEMCG events:\n");
+	printf("\tlow: %ld\n", counters->low);
+	printf("\thigh: %ld\n", counters->high);
+	printf("\tmax: %ld\n", counters->max);
+	printf("\toom: %ld\n", counters->oom);
+	printf("\toom_kill: %ld\n", counters->oom_kill);
+	printf("\toom_group_kill: %ld\n", counters->oom_group_kill);
+}
+
+static int get_memcg_counter(char *line, const char *name, long *counter)
+{
+	size_t len = strlen(name);
+	char *endptr;
+	long tmp;
+
+	if (memcmp(line, name, len)) {
+		warnx("Counter line %s has wrong name, %s is expected",
+		      line, name);
+		return -EINVAL;
+	}
+
+	/* skip the whitespace delimiter */
+	len += 1;
+
+	errno = 0;
+	tmp = strtol(&line[len], &endptr, 10);
+	if (((tmp == LONG_MAX || tmp == LONG_MIN) && errno == ERANGE) ||
+	    (errno && !tmp)) {
+		warnx("Failed to parse: %s", &line[len]);
+		return -ERANGE;
+	}
+
+	if (endptr == &line[len]) {
+		warnx("Not digits were found in line %s", &line[len]);
+		return -EINVAL;
+	}
+
+	if (!(*endptr == '\0' || (*endptr == '\n' && *++endptr == '\0'))) {
+		warnx("Further characters after number: %s", endptr);
+		return -EINVAL;
+	}
+
+	*counter = tmp;
+
+	return 0;
+}
+
+static int read_memcg_events(struct memcg_events *events, bool show_diff)
+{
+	FILE *fp = fopen(events->path, "re");
+	size_t i;
+	int ret = 0;
+	bool any_new_events = false;
+	char *line = NULL;
+	size_t len = 0;
+	struct memcg_counters new_counters;
+	struct memcg_counters *counters = &events->counters;
+	struct {
+		const char *name;
+		long *new;
+		long *old;
+	} map[] = {
+		{
+			.name = "low",
+			.new = &new_counters.low,
+			.old = &counters->low,
+		},
+		{
+			.name = "high",
+			.new = &new_counters.high,
+			.old = &counters->high,
+		},
+		{
+			.name = "max",
+			.new = &new_counters.max,
+			.old = &counters->max,
+		},
+		{
+			.name = "oom",
+			.new = &new_counters.oom,
+			.old = &counters->oom,
+		},
+		{
+			.name = "oom_kill",
+			.new = &new_counters.oom_kill,
+			.old = &counters->oom_kill,
+		},
+		{
+			.name = "oom_group_kill",
+			.new = &new_counters.oom_group_kill,
+			.old = &counters->oom_group_kill,
+		},
+	};
+
+	if (!fp) {
+		warn("Failed to open memcg events file %s", events->path);
+		return -EBADF;
+	}
+
+	/* Read new values for memcg counters */
+	for (i = 0; i < ARRAY_SIZE(map); ++i) {
+		ssize_t nread;
+
+		errno = 0;
+		nread = getline(&line, &len, fp);
+		if (nread == -1) {
+			if (errno) {
+				warn("Failed to read line for counter %s",
+				     map[i].name);
+				ret = -EIO;
+				goto exit;
+			}
+
+			break;
+		}
+
+		ret = get_memcg_counter(line, map[i].name, map[i].new);
+		if (ret) {
+			warnx("Failed to get counter value from line %s", line);
+			goto exit;
+		}
+	}
+
+	for (i = 0; i < ARRAY_SIZE(map); ++i) {
+		long diff;
+
+		if (*map[i].new > *map[i].old) {
+			diff = *map[i].new - *map[i].old;
+
+			if (show_diff)
+				printf("*** %ld MEMCG %s event%s, "
+				       "change counter %ld => %ld\n",
+				       diff, map[i].name,
+				       (diff == 1) ? "" : "s",
+				       *map[i].old, *map[i].new);
+
+			*map[i].old += diff;
+			any_new_events = true;
+		}
+	}
+
+	if (show_diff && !any_new_events)
+		printf("*** No new untracked memcg events available\n");
+
+exit:
+	free(line);
+	fclose(fp);
+
+	return ret;
+}
+
+static void process_memcg_events(struct memcg_events *events,
+				 struct inotify_event *event)
+{
+	int ret;
+
+	if (events->inotify_wd != event->wd) {
+		warnx("Unknown inotify event %d, should be %d", event->wd,
+		      events->inotify_wd);
+		return;
+	}
+
+	printf("Received event in %s:\n", events->path);
+
+	if (!(event->mask & IN_MODIFY)) {
+		warnx("No IN_MODIFY event, skip it");
+		return;
+	}
+
+	ret = read_memcg_events(events, /* show_diff = */true);
+	if (ret)
+		warnx("Can't read memcg events");
+}
+
+static void monitor_events(struct memcg_events *events)
+{
+	struct pollfd fds[1];
+	int ret;
+
+	printf("Started monitoring memory events from '%s'...\n", events->path);
+
+	fds[0].fd = events->inotify_fd;
+	fds[0].events = POLLIN;
+
+	for (;;) {
+		ret = poll(fds, ARRAY_SIZE(fds), -1);
+		if (ret < 0 && errno != EAGAIN)
+			err(EXIT_FAILURE, "Can't poll memcg events (%d)", ret);
+
+		if (fds[0].revents & POLLERR)
+			err(EXIT_FAILURE, "Got POLLERR during monitor events");
+
+		if (fds[0].revents & POLLIN) {
+			struct inotify_event *event;
+			char buffer[INOTIFY_BUFFER_SIZE];
+			ssize_t length;
+
+			length = read(fds[0].fd, buffer, INOTIFY_BUFFER_SIZE);
+			if (length <= 0)
+				continue;
+
+			event = (struct inotify_event *)buffer;
+			while (INOTIFY_EVENT_OK(event, length)) {
+				process_memcg_events(events, event);
+				event = INOTIFY_EVENT_NEXT(event, length);
+			}
+		}
+	}
+}
+
+static int initialize_memcg_events(struct memcg_events *events,
+				   const char *cgroup)
+{
+	int ret;
+
+	memset(events, 0, sizeof(struct memcg_events));
+
+	ret = snprintf(events->path, PATH_MAX,
+		       "/sys/fs/cgroup/%s/memory.events", cgroup);
+	if (ret >= PATH_MAX) {
+		warnx("Path to cgroup memory.events is too long");
+		return -EMSGSIZE;
+	} else if (ret < 0) {
+		warn("Can't generate cgroup event full name");
+		return ret;
+	}
+
+	ret = read_memcg_events(events, /* show_diff = */false);
+	if (ret) {
+		warnx("Failed to read initial memcg events state (%d)", ret);
+		return ret;
+	}
+
+	events->inotify_fd = inotify_init();
+	if (events->inotify_fd < 0) {
+		warn("Failed to setup new inotify device");
+		return -EMFILE;
+	}
+
+	events->inotify_wd = inotify_add_watch(events->inotify_fd,
+					       events->path, IN_MODIFY);
+	if (events->inotify_wd < 0) {
+		warn("Couldn't add monitor in dir %s", events->path);
+		return -EIO;
+	}
+
+	printf("Initialized MEMCG events with counters:\n");
+	print_memcg_counters(&events->counters);
+
+	return 0;
+}
+
+static void cleanup_memcg_events(struct memcg_events *events)
+{
+	inotify_rm_watch(events->inotify_fd, events->inotify_wd);
+	close(events->inotify_fd);
+}
+
+int main(int argc, const char **argv)
+{
+	struct memcg_events events;
+	ssize_t ret;
+
+	if (argc != 2)
+		errx(EXIT_FAILURE, "Usage: %s <cgroup>", argv[0]);
+
+	ret = initialize_memcg_events(&events, argv[1]);
+	if (ret)
+		errx(EXIT_FAILURE, "Can't initialize memcg events (%zd)", ret);
+
+	monitor_events(&events);
+
+	cleanup_memcg_events(&events);
+
+	printf("Exiting memcg event listener...\n");
+
+	return EXIT_SUCCESS;
+}