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, 222 insertions, 0 deletions

diff --git a/drivers/dma/idxd/submit.c b/drivers/dma/idxd/submit.c
new file mode 100644
index 000000000..03217041b
--- /dev/null
+++ b/drivers/dma/idxd/submit.c
@@ -0,0 +1,222 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright(c) 2019 Intel Corporation. All rights rsvd. */
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include <uapi/linux/idxd.h>
+#include "idxd.h"
+#include "registers.h"
+
+static struct idxd_desc *__get_desc(struct idxd_wq *wq, int idx, int cpu)
+{
+	struct idxd_desc *desc;
+	struct idxd_device *idxd = wq->idxd;
+
+	desc = wq->descs[idx];
+	memset(desc->hw, 0, sizeof(struct dsa_hw_desc));
+	memset(desc->completion, 0, idxd->data->compl_size);
+	desc->cpu = cpu;
+
+	if (device_pasid_enabled(idxd))
+		desc->hw->pasid = idxd->pasid;
+
+	return desc;
+}
+
+struct idxd_desc *idxd_alloc_desc(struct idxd_wq *wq, enum idxd_op_type optype)
+{
+	int cpu, idx;
+	struct idxd_device *idxd = wq->idxd;
+	DEFINE_SBQ_WAIT(wait);
+	struct sbq_wait_state *ws;
+	struct sbitmap_queue *sbq;
+
+	if (idxd->state != IDXD_DEV_ENABLED)
+		return ERR_PTR(-EIO);
+
+	sbq = &wq->sbq;
+	idx = sbitmap_queue_get(sbq, &cpu);
+	if (idx < 0) {
+		if (optype == IDXD_OP_NONBLOCK)
+			return ERR_PTR(-EAGAIN);
+	} else {
+		return __get_desc(wq, idx, cpu);
+	}
+
+	ws = &sbq->ws[0];
+	for (;;) {
+		sbitmap_prepare_to_wait(sbq, ws, &wait, TASK_INTERRUPTIBLE);
+		if (signal_pending_state(TASK_INTERRUPTIBLE, current))
+			break;
+		idx = sbitmap_queue_get(sbq, &cpu);
+		if (idx >= 0)
+			break;
+		schedule();
+	}
+
+	sbitmap_finish_wait(sbq, ws, &wait);
+	if (idx < 0)
+		return ERR_PTR(-EAGAIN);
+
+	return __get_desc(wq, idx, cpu);
+}
+EXPORT_SYMBOL_NS_GPL(idxd_alloc_desc, "IDXD");
+
+void idxd_free_desc(struct idxd_wq *wq, struct idxd_desc *desc)
+{
+	int cpu = desc->cpu;
+
+	desc->cpu = -1;
+	sbitmap_queue_clear(&wq->sbq, desc->id, cpu);
+}
+EXPORT_SYMBOL_NS_GPL(idxd_free_desc, "IDXD");
+
+static struct idxd_desc *list_abort_desc(struct idxd_wq *wq, struct idxd_irq_entry *ie,
+					 struct idxd_desc *desc)
+{
+	struct idxd_desc *d, *n;
+
+	lockdep_assert_held(&ie->list_lock);
+	list_for_each_entry_safe(d, n, &ie->work_list, list) {
+		if (d == desc) {
+			list_del(&d->list);
+			return d;
+		}
+	}
+
+	/*
+	 * At this point, the desc needs to be aborted is held by the completion
+	 * handler where it has taken it off the pending list but has not added to the
+	 * work list. It will be cleaned up by the interrupt handler when it sees the
+	 * IDXD_COMP_DESC_ABORT for completion status.
+	 */
+	return NULL;
+}
+
+static void llist_abort_desc(struct idxd_wq *wq, struct idxd_irq_entry *ie,
+			     struct idxd_desc *desc)
+{
+	struct idxd_desc *d, *t, *found = NULL;
+	struct llist_node *head;
+	LIST_HEAD(flist);
+
+	desc->completion->status = IDXD_COMP_DESC_ABORT;
+	/*
+	 * Grab the list lock so it will block the irq thread handler. This allows the
+	 * abort code to locate the descriptor need to be aborted.
+	 */
+	spin_lock(&ie->list_lock);
+	head = llist_del_all(&ie->pending_llist);
+	if (head) {
+		llist_for_each_entry_safe(d, t, head, llnode) {
+			if (d == desc) {
+				found = desc;
+				continue;
+			}
+
+			if (d->completion->status)
+				list_add_tail(&d->list, &flist);
+			else
+				list_add_tail(&d->list, &ie->work_list);
+		}
+	}
+
+	if (!found)
+		found = list_abort_desc(wq, ie, desc);
+	spin_unlock(&ie->list_lock);
+
+	if (found)
+		idxd_dma_complete_txd(found, IDXD_COMPLETE_ABORT, false,
+				      NULL, NULL);
+
+	/*
+	 * completing the descriptor will return desc to allocator and
+	 * the desc can be acquired by a different process and the
+	 * desc->list can be modified.  Delete desc from list so the
+	 * list traversing does not get corrupted by the other process.
+	 */
+	list_for_each_entry_safe(d, t, &flist, list) {
+		list_del_init(&d->list);
+		idxd_dma_complete_txd(d, IDXD_COMPLETE_ABORT, true,
+				      NULL, NULL);
+	}
+}
+
+/*
+ * ENQCMDS typically fail when the WQ is inactive or busy. On host submission, the driver
+ * has better control of number of descriptors being submitted to a shared wq by limiting
+ * the number of driver allocated descriptors to the wq size. However, when the swq is
+ * exported to a guest kernel, it may be shared with multiple guest kernels. This means
+ * the likelihood of getting busy returned on the swq when submitting goes significantly up.
+ * Having a tunable retry mechanism allows the driver to keep trying for a bit before giving
+ * up. The sysfs knob can be tuned by the system administrator.
+ */
+int idxd_enqcmds(struct idxd_wq *wq, void __iomem *portal, const void *desc)
+{
+	unsigned int retries = wq->enqcmds_retries;
+	int rc;
+
+	do {
+		rc = enqcmds(portal, desc);
+		if (rc == 0)
+			break;
+		cpu_relax();
+	} while (retries--);
+
+	return rc;
+}
+
+int idxd_submit_desc(struct idxd_wq *wq, struct idxd_desc *desc)
+{
+	struct idxd_device *idxd = wq->idxd;
+	struct idxd_irq_entry *ie = NULL;
+	u32 desc_flags = desc->hw->flags;
+	void __iomem *portal;
+	int rc;
+
+	if (idxd->state != IDXD_DEV_ENABLED)
+		return -EIO;
+
+	if (!percpu_ref_tryget_live(&wq->wq_active)) {
+		wait_for_completion(&wq->wq_resurrect);
+		if (!percpu_ref_tryget_live(&wq->wq_active))
+			return -ENXIO;
+	}
+
+	portal = idxd_wq_portal_addr(wq);
+
+	/*
+	 * Pending the descriptor to the lockless list for the irq_entry
+	 * that we designated the descriptor to.
+	 */
+	if (desc_flags & IDXD_OP_FLAG_RCI) {
+		ie = &wq->ie;
+		desc->hw->int_handle = ie->int_handle;
+		llist_add(&desc->llnode, &ie->pending_llist);
+	}
+
+	/*
+	 * The wmb() flushes writes to coherent DMA data before
+	 * possibly triggering a DMA read. The wmb() is necessary
+	 * even on UP because the recipient is a device.
+	 */
+	wmb();
+
+	if (wq_dedicated(wq)) {
+		iosubmit_cmds512(portal, desc->hw, 1);
+	} else {
+		rc = idxd_enqcmds(wq, portal, desc->hw);
+		if (rc < 0) {
+			percpu_ref_put(&wq->wq_active);
+			/* abort operation frees the descriptor */
+			if (ie)
+				llist_abort_desc(wq, ie, desc);
+			return rc;
+		}
+	}
+
+	percpu_ref_put(&wq->wq_active);
+	return 0;
+}
+EXPORT_SYMBOL_NS_GPL(idxd_submit_desc, "IDXD");