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, 620 insertions, 0 deletions

diff --git a/lib/test_lockup.c b/lib/test_lockup.c
new file mode 100644
index 000000000..c3fd87d6c
--- /dev/null
+++ b/lib/test_lockup.c
@@ -0,0 +1,620 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Test module to generate lockups
+ */
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/delay.h>
+#include <linux/sched.h>
+#include <linux/sched/signal.h>
+#include <linux/sched/clock.h>
+#include <linux/cpu.h>
+#include <linux/nmi.h>
+#include <linux/mm.h>
+#include <linux/uaccess.h>
+#include <linux/file.h>
+
+static unsigned int time_secs;
+module_param(time_secs, uint, 0600);
+MODULE_PARM_DESC(time_secs, "lockup time in seconds, default 0");
+
+static unsigned int time_nsecs;
+module_param(time_nsecs, uint, 0600);
+MODULE_PARM_DESC(time_nsecs, "nanoseconds part of lockup time, default 0");
+
+static unsigned int cooldown_secs;
+module_param(cooldown_secs, uint, 0600);
+MODULE_PARM_DESC(cooldown_secs, "cooldown time between iterations in seconds, default 0");
+
+static unsigned int cooldown_nsecs;
+module_param(cooldown_nsecs, uint, 0600);
+MODULE_PARM_DESC(cooldown_nsecs, "nanoseconds part of cooldown, default 0");
+
+static unsigned int iterations = 1;
+module_param(iterations, uint, 0600);
+MODULE_PARM_DESC(iterations, "lockup iterations, default 1");
+
+static bool all_cpus;
+module_param(all_cpus, bool, 0400);
+MODULE_PARM_DESC(all_cpus, "trigger lockup at all cpus at once");
+
+static int wait_state;
+static char *state = "R";
+module_param(state, charp, 0400);
+MODULE_PARM_DESC(state, "wait in 'R' running (default), 'D' uninterruptible, 'K' killable, 'S' interruptible state");
+
+static bool use_hrtimer;
+module_param(use_hrtimer, bool, 0400);
+MODULE_PARM_DESC(use_hrtimer, "use high-resolution timer for sleeping");
+
+static bool iowait;
+module_param(iowait, bool, 0400);
+MODULE_PARM_DESC(iowait, "account sleep time as iowait");
+
+static bool lock_read;
+module_param(lock_read, bool, 0400);
+MODULE_PARM_DESC(lock_read, "lock read-write locks for read");
+
+static bool lock_single;
+module_param(lock_single, bool, 0400);
+MODULE_PARM_DESC(lock_single, "acquire locks only at one cpu");
+
+static bool reacquire_locks;
+module_param(reacquire_locks, bool, 0400);
+MODULE_PARM_DESC(reacquire_locks, "release and reacquire locks/irq/preempt between iterations");
+
+static bool touch_softlockup;
+module_param(touch_softlockup, bool, 0600);
+MODULE_PARM_DESC(touch_softlockup, "touch soft-lockup watchdog between iterations");
+
+static bool touch_hardlockup;
+module_param(touch_hardlockup, bool, 0600);
+MODULE_PARM_DESC(touch_hardlockup, "touch hard-lockup watchdog between iterations");
+
+static bool call_cond_resched;
+module_param(call_cond_resched, bool, 0600);
+MODULE_PARM_DESC(call_cond_resched, "call cond_resched() between iterations");
+
+static bool measure_lock_wait;
+module_param(measure_lock_wait, bool, 0400);
+MODULE_PARM_DESC(measure_lock_wait, "measure lock wait time");
+
+static unsigned long lock_wait_threshold = ULONG_MAX;
+module_param(lock_wait_threshold, ulong, 0400);
+MODULE_PARM_DESC(lock_wait_threshold, "print lock wait time longer than this in nanoseconds, default off");
+
+static bool test_disable_irq;
+module_param_named(disable_irq, test_disable_irq, bool, 0400);
+MODULE_PARM_DESC(disable_irq, "disable interrupts: generate hard-lockups");
+
+static bool disable_softirq;
+module_param(disable_softirq, bool, 0400);
+MODULE_PARM_DESC(disable_softirq, "disable bottom-half irq handlers");
+
+static bool disable_preempt;
+module_param(disable_preempt, bool, 0400);
+MODULE_PARM_DESC(disable_preempt, "disable preemption: generate soft-lockups");
+
+static bool lock_rcu;
+module_param(lock_rcu, bool, 0400);
+MODULE_PARM_DESC(lock_rcu, "grab rcu_read_lock: generate rcu stalls");
+
+static bool lock_mmap_sem;
+module_param(lock_mmap_sem, bool, 0400);
+MODULE_PARM_DESC(lock_mmap_sem, "lock mm->mmap_lock: block procfs interfaces");
+
+static unsigned long lock_rwsem_ptr;
+module_param_unsafe(lock_rwsem_ptr, ulong, 0400);
+MODULE_PARM_DESC(lock_rwsem_ptr, "lock rw_semaphore at address");
+
+static unsigned long lock_mutex_ptr;
+module_param_unsafe(lock_mutex_ptr, ulong, 0400);
+MODULE_PARM_DESC(lock_mutex_ptr, "lock mutex at address");
+
+static unsigned long lock_spinlock_ptr;
+module_param_unsafe(lock_spinlock_ptr, ulong, 0400);
+MODULE_PARM_DESC(lock_spinlock_ptr, "lock spinlock at address");
+
+static unsigned long lock_rwlock_ptr;
+module_param_unsafe(lock_rwlock_ptr, ulong, 0400);
+MODULE_PARM_DESC(lock_rwlock_ptr, "lock rwlock at address");
+
+static unsigned int alloc_pages_nr;
+module_param_unsafe(alloc_pages_nr, uint, 0600);
+MODULE_PARM_DESC(alloc_pages_nr, "allocate and free pages under locks");
+
+static unsigned int alloc_pages_order;
+module_param(alloc_pages_order, uint, 0400);
+MODULE_PARM_DESC(alloc_pages_order, "page order to allocate");
+
+static gfp_t alloc_pages_gfp = GFP_KERNEL;
+module_param_unsafe(alloc_pages_gfp, uint, 0400);
+MODULE_PARM_DESC(alloc_pages_gfp, "allocate pages with this gfp_mask, default GFP_KERNEL");
+
+static bool alloc_pages_atomic;
+module_param(alloc_pages_atomic, bool, 0400);
+MODULE_PARM_DESC(alloc_pages_atomic, "allocate pages with GFP_ATOMIC");
+
+static bool reallocate_pages;
+module_param(reallocate_pages, bool, 0400);
+MODULE_PARM_DESC(reallocate_pages, "free and allocate pages between iterations");
+
+struct file *test_file;
+static struct inode *test_inode;
+static char test_file_path[256];
+module_param_string(file_path, test_file_path, sizeof(test_file_path), 0400);
+MODULE_PARM_DESC(file_path, "file path to test");
+
+static bool test_lock_inode;
+module_param_named(lock_inode, test_lock_inode, bool, 0400);
+MODULE_PARM_DESC(lock_inode, "lock file -> inode -> i_rwsem");
+
+static bool test_lock_mapping;
+module_param_named(lock_mapping, test_lock_mapping, bool, 0400);
+MODULE_PARM_DESC(lock_mapping, "lock file -> mapping -> i_mmap_rwsem");
+
+static bool test_lock_sb_umount;
+module_param_named(lock_sb_umount, test_lock_sb_umount, bool, 0400);
+MODULE_PARM_DESC(lock_sb_umount, "lock file -> sb -> s_umount");
+
+static atomic_t alloc_pages_failed = ATOMIC_INIT(0);
+
+static atomic64_t max_lock_wait = ATOMIC64_INIT(0);
+
+static struct task_struct *main_task;
+static int master_cpu;
+
+static void test_lock(bool master, bool verbose)
+{
+	u64 wait_start;
+
+	if (measure_lock_wait)
+		wait_start = local_clock();
+
+	if (lock_mutex_ptr && master) {
+		if (verbose)
+			pr_notice("lock mutex %ps\n", (void *)lock_mutex_ptr);
+		mutex_lock((struct mutex *)lock_mutex_ptr);
+	}
+
+	if (lock_rwsem_ptr && master) {
+		if (verbose)
+			pr_notice("lock rw_semaphore %ps\n",
+				  (void *)lock_rwsem_ptr);
+		if (lock_read)
+			down_read((struct rw_semaphore *)lock_rwsem_ptr);
+		else
+			down_write((struct rw_semaphore *)lock_rwsem_ptr);
+	}
+
+	if (lock_mmap_sem && master) {
+		if (verbose)
+			pr_notice("lock mmap_lock pid=%d\n", main_task->pid);
+		if (lock_read)
+			mmap_read_lock(main_task->mm);
+		else
+			mmap_write_lock(main_task->mm);
+	}
+
+	if (test_disable_irq)
+		local_irq_disable();
+
+	if (disable_softirq)
+		local_bh_disable();
+
+	if (disable_preempt)
+		preempt_disable();
+
+	if (lock_rcu)
+		rcu_read_lock();
+
+	if (lock_spinlock_ptr && master) {
+		if (verbose)
+			pr_notice("lock spinlock %ps\n",
+				  (void *)lock_spinlock_ptr);
+		spin_lock((spinlock_t *)lock_spinlock_ptr);
+	}
+
+	if (lock_rwlock_ptr && master) {
+		if (verbose)
+			pr_notice("lock rwlock %ps\n",
+				  (void *)lock_rwlock_ptr);
+		if (lock_read)
+			read_lock((rwlock_t *)lock_rwlock_ptr);
+		else
+			write_lock((rwlock_t *)lock_rwlock_ptr);
+	}
+
+	if (measure_lock_wait) {
+		s64 cur_wait = local_clock() - wait_start;
+		s64 max_wait = atomic64_read(&max_lock_wait);
+
+		do {
+			if (cur_wait < max_wait)
+				break;
+			max_wait = atomic64_cmpxchg(&max_lock_wait,
+						    max_wait, cur_wait);
+		} while (max_wait != cur_wait);
+
+		if (cur_wait > lock_wait_threshold)
+			pr_notice_ratelimited("lock wait %lld ns\n", cur_wait);
+	}
+}
+
+static void test_unlock(bool master, bool verbose)
+{
+	if (lock_rwlock_ptr && master) {
+		if (lock_read)
+			read_unlock((rwlock_t *)lock_rwlock_ptr);
+		else
+			write_unlock((rwlock_t *)lock_rwlock_ptr);
+		if (verbose)
+			pr_notice("unlock rwlock %ps\n",
+				  (void *)lock_rwlock_ptr);
+	}
+
+	if (lock_spinlock_ptr && master) {
+		spin_unlock((spinlock_t *)lock_spinlock_ptr);
+		if (verbose)
+			pr_notice("unlock spinlock %ps\n",
+				  (void *)lock_spinlock_ptr);
+	}
+
+	if (lock_rcu)
+		rcu_read_unlock();
+
+	if (disable_preempt)
+		preempt_enable();
+
+	if (disable_softirq)
+		local_bh_enable();
+
+	if (test_disable_irq)
+		local_irq_enable();
+
+	if (lock_mmap_sem && master) {
+		if (lock_read)
+			mmap_read_unlock(main_task->mm);
+		else
+			mmap_write_unlock(main_task->mm);
+		if (verbose)
+			pr_notice("unlock mmap_lock pid=%d\n", main_task->pid);
+	}
+
+	if (lock_rwsem_ptr && master) {
+		if (lock_read)
+			up_read((struct rw_semaphore *)lock_rwsem_ptr);
+		else
+			up_write((struct rw_semaphore *)lock_rwsem_ptr);
+		if (verbose)
+			pr_notice("unlock rw_semaphore %ps\n",
+				  (void *)lock_rwsem_ptr);
+	}
+
+	if (lock_mutex_ptr && master) {
+		mutex_unlock((struct mutex *)lock_mutex_ptr);
+		if (verbose)
+			pr_notice("unlock mutex %ps\n",
+				  (void *)lock_mutex_ptr);
+	}
+}
+
+static void test_alloc_pages(struct list_head *pages)
+{
+	struct page *page;
+	unsigned int i;
+
+	for (i = 0; i < alloc_pages_nr; i++) {
+		page = alloc_pages(alloc_pages_gfp, alloc_pages_order);
+		if (!page) {
+			atomic_inc(&alloc_pages_failed);
+			break;
+		}
+		list_add(&page->lru, pages);
+	}
+}
+
+static void test_free_pages(struct list_head *pages)
+{
+	struct page *page, *next;
+
+	list_for_each_entry_safe(page, next, pages, lru)
+		__free_pages(page, alloc_pages_order);
+	INIT_LIST_HEAD(pages);
+}
+
+static void test_wait(unsigned int secs, unsigned int nsecs)
+{
+	if (wait_state == TASK_RUNNING) {
+		if (secs)
+			mdelay(secs * MSEC_PER_SEC);
+		if (nsecs)
+			ndelay(nsecs);
+		return;
+	}
+
+	__set_current_state(wait_state);
+	if (use_hrtimer) {
+		ktime_t time;
+
+		time = ns_to_ktime((u64)secs * NSEC_PER_SEC + nsecs);
+		schedule_hrtimeout(&time, HRTIMER_MODE_REL);
+	} else {
+		schedule_timeout(secs * HZ + nsecs_to_jiffies(nsecs));
+	}
+}
+
+static void test_lockup(bool master)
+{
+	u64 lockup_start = local_clock();
+	unsigned int iter = 0;
+	LIST_HEAD(pages);
+
+	pr_notice("Start on CPU%d\n", raw_smp_processor_id());
+
+	test_lock(master, true);
+
+	test_alloc_pages(&pages);
+
+	while (iter++ < iterations && !signal_pending(main_task)) {
+
+		if (iowait)
+			current->in_iowait = 1;
+
+		test_wait(time_secs, time_nsecs);
+
+		if (iowait)
+			current->in_iowait = 0;
+
+		if (reallocate_pages)
+			test_free_pages(&pages);
+
+		if (reacquire_locks)
+			test_unlock(master, false);
+
+		if (touch_softlockup)
+			touch_softlockup_watchdog();
+
+		if (touch_hardlockup)
+			touch_nmi_watchdog();
+
+		if (call_cond_resched)
+			cond_resched();
+
+		test_wait(cooldown_secs, cooldown_nsecs);
+
+		if (reacquire_locks)
+			test_lock(master, false);
+
+		if (reallocate_pages)
+			test_alloc_pages(&pages);
+	}
+
+	pr_notice("Finish on CPU%d in %lld ns\n", raw_smp_processor_id(),
+		  local_clock() - lockup_start);
+
+	test_free_pages(&pages);
+
+	test_unlock(master, true);
+}
+
+static DEFINE_PER_CPU(struct work_struct, test_works);
+
+static void test_work_fn(struct work_struct *work)
+{
+	test_lockup(!lock_single ||
+		    work == per_cpu_ptr(&test_works, master_cpu));
+}
+
+static bool test_kernel_ptr(unsigned long addr, int size)
+{
+	void *ptr = (void *)addr;
+	char buf;
+
+	if (!addr)
+		return false;
+
+	/* should be at least readable kernel address */
+	if (!IS_ENABLED(CONFIG_ALTERNATE_USER_ADDRESS_SPACE) &&
+	    (access_ok((void __user *)ptr, 1) ||
+	     access_ok((void __user *)ptr + size - 1, 1))) {
+		pr_err("user space ptr invalid in kernel: %#lx\n", addr);
+		return true;
+	}
+
+	if (get_kernel_nofault(buf, ptr) ||
+	    get_kernel_nofault(buf, ptr + size - 1)) {
+		pr_err("invalid kernel ptr: %#lx\n", addr);
+		return true;
+	}
+
+	return false;
+}
+
+static bool __maybe_unused test_magic(unsigned long addr, int offset,
+				      unsigned int expected)
+{
+	void *ptr = (void *)addr + offset;
+	unsigned int magic = 0;
+
+	if (!addr)
+		return false;
+
+	if (get_kernel_nofault(magic, ptr) || magic != expected) {
+		pr_err("invalid magic at %#lx + %#x = %#x, expected %#x\n",
+		       addr, offset, magic, expected);
+		return true;
+	}
+
+	return false;
+}
+
+static int __init test_lockup_init(void)
+{
+	u64 test_start = local_clock();
+
+	main_task = current;
+
+	switch (state[0]) {
+	case 'S':
+		wait_state = TASK_INTERRUPTIBLE;
+		break;
+	case 'D':
+		wait_state = TASK_UNINTERRUPTIBLE;
+		break;
+	case 'K':
+		wait_state = TASK_KILLABLE;
+		break;
+	case 'R':
+		wait_state = TASK_RUNNING;
+		break;
+	default:
+		pr_err("unknown state=%s\n", state);
+		return -EINVAL;
+	}
+
+	if (alloc_pages_atomic)
+		alloc_pages_gfp = GFP_ATOMIC;
+
+	if (test_kernel_ptr(lock_spinlock_ptr, sizeof(spinlock_t)) ||
+	    test_kernel_ptr(lock_rwlock_ptr, sizeof(rwlock_t)) ||
+	    test_kernel_ptr(lock_mutex_ptr, sizeof(struct mutex)) ||
+	    test_kernel_ptr(lock_rwsem_ptr, sizeof(struct rw_semaphore)))
+		return -EINVAL;
+
+#ifdef CONFIG_DEBUG_SPINLOCK
+#ifdef CONFIG_PREEMPT_RT
+	if (test_magic(lock_spinlock_ptr,
+		       offsetof(spinlock_t, lock.wait_lock.magic),
+		       SPINLOCK_MAGIC) ||
+	    test_magic(lock_rwlock_ptr,
+		       offsetof(rwlock_t, rwbase.rtmutex.wait_lock.magic),
+		       SPINLOCK_MAGIC) ||
+	    test_magic(lock_mutex_ptr,
+		       offsetof(struct mutex, rtmutex.wait_lock.magic),
+		       SPINLOCK_MAGIC) ||
+	    test_magic(lock_rwsem_ptr,
+		       offsetof(struct rw_semaphore, rwbase.rtmutex.wait_lock.magic),
+		       SPINLOCK_MAGIC))
+		return -EINVAL;
+#else
+	if (test_magic(lock_spinlock_ptr,
+		       offsetof(spinlock_t, rlock.magic),
+		       SPINLOCK_MAGIC) ||
+	    test_magic(lock_rwlock_ptr,
+		       offsetof(rwlock_t, magic),
+		       RWLOCK_MAGIC) ||
+	    test_magic(lock_mutex_ptr,
+		       offsetof(struct mutex, wait_lock.magic),
+		       SPINLOCK_MAGIC) ||
+	    test_magic(lock_rwsem_ptr,
+		       offsetof(struct rw_semaphore, wait_lock.magic),
+		       SPINLOCK_MAGIC))
+		return -EINVAL;
+#endif
+#endif
+
+	if ((wait_state != TASK_RUNNING ||
+	     (call_cond_resched && !reacquire_locks) ||
+	     (alloc_pages_nr && gfpflags_allow_blocking(alloc_pages_gfp))) &&
+	    (test_disable_irq || disable_softirq || disable_preempt ||
+	     lock_rcu || lock_spinlock_ptr || lock_rwlock_ptr)) {
+		pr_err("refuse to sleep in atomic context\n");
+		return -EINVAL;
+	}
+
+	if (lock_mmap_sem && !main_task->mm) {
+		pr_err("no mm to lock mmap_lock\n");
+		return -EINVAL;
+	}
+
+	if (test_file_path[0]) {
+		test_file = filp_open(test_file_path, O_RDONLY, 0);
+		if (IS_ERR(test_file)) {
+			pr_err("failed to open %s: %ld\n", test_file_path, PTR_ERR(test_file));
+			return PTR_ERR(test_file);
+		}
+		test_inode = file_inode(test_file);
+	} else if (test_lock_inode ||
+		   test_lock_mapping ||
+		   test_lock_sb_umount) {
+		pr_err("no file to lock\n");
+		return -EINVAL;
+	}
+
+	if (test_lock_inode && test_inode)
+		lock_rwsem_ptr = (unsigned long)&test_inode->i_rwsem;
+
+	if (test_lock_mapping && test_file && test_file->f_mapping)
+		lock_rwsem_ptr = (unsigned long)&test_file->f_mapping->i_mmap_rwsem;
+
+	if (test_lock_sb_umount && test_inode)
+		lock_rwsem_ptr = (unsigned long)&test_inode->i_sb->s_umount;
+
+	pr_notice("START pid=%d time=%u +%u ns cooldown=%u +%u ns iterations=%u state=%s %s%s%s%s%s%s%s%s%s%s%s\n",
+		  main_task->pid, time_secs, time_nsecs,
+		  cooldown_secs, cooldown_nsecs, iterations, state,
+		  all_cpus ? "all_cpus " : "",
+		  iowait ? "iowait " : "",
+		  test_disable_irq ? "disable_irq " : "",
+		  disable_softirq ? "disable_softirq " : "",
+		  disable_preempt ? "disable_preempt " : "",
+		  lock_rcu ? "lock_rcu " : "",
+		  lock_read ? "lock_read " : "",
+		  touch_softlockup ? "touch_softlockup " : "",
+		  touch_hardlockup ? "touch_hardlockup " : "",
+		  call_cond_resched ? "call_cond_resched " : "",
+		  reacquire_locks ? "reacquire_locks " : "");
+
+	if (alloc_pages_nr)
+		pr_notice("ALLOCATE PAGES nr=%u order=%u gfp=%pGg %s\n",
+			  alloc_pages_nr, alloc_pages_order, &alloc_pages_gfp,
+			  reallocate_pages ? "reallocate_pages " : "");
+
+	if (all_cpus) {
+		unsigned int cpu;
+
+		cpus_read_lock();
+
+		preempt_disable();
+		master_cpu = smp_processor_id();
+		for_each_online_cpu(cpu) {
+			INIT_WORK(per_cpu_ptr(&test_works, cpu), test_work_fn);
+			queue_work_on(cpu, system_highpri_wq,
+				      per_cpu_ptr(&test_works, cpu));
+		}
+		preempt_enable();
+
+		for_each_online_cpu(cpu)
+			flush_work(per_cpu_ptr(&test_works, cpu));
+
+		cpus_read_unlock();
+	} else {
+		test_lockup(true);
+	}
+
+	if (measure_lock_wait)
+		pr_notice("Maximum lock wait: %lld ns\n",
+			  atomic64_read(&max_lock_wait));
+
+	if (alloc_pages_nr)
+		pr_notice("Page allocation failed %u times\n",
+			  atomic_read(&alloc_pages_failed));
+
+	pr_notice("FINISH in %llu ns\n", local_clock() - test_start);
+
+	if (test_file)
+		fput(test_file);
+
+	if (signal_pending(main_task))
+		return -EINTR;
+
+	return -EAGAIN;
+}
+module_init(test_lockup_init);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Konstantin Khlebnikov <khlebnikov@yandex-team.ru>");
+MODULE_DESCRIPTION("Test module to generate lockups");