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, 123 insertions, 0 deletions

diff --git a/include/trace/events/vmalloc.h b/include/trace/events/vmalloc.h
new file mode 100644
index 000000000..ad4e02191
--- /dev/null
+++ b/include/trace/events/vmalloc.h
@@ -0,0 +1,123 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#undef TRACE_SYSTEM
+#define TRACE_SYSTEM vmalloc
+
+#if !defined(_TRACE_VMALLOC_H) || defined(TRACE_HEADER_MULTI_READ)
+#define _TRACE_VMALLOC_H
+
+#include <linux/tracepoint.h>
+
+/**
+ * alloc_vmap_area - called when a new vmap allocation occurs
+ * @addr:	an allocated address
+ * @size:	a requested size
+ * @align:	a requested alignment
+ * @vstart:	a requested start range
+ * @vend:	a requested end range
+ * @failed:	an allocation failed or not
+ *
+ * This event is used for a debug purpose, it can give an extra
+ * information for a developer about how often it occurs and which
+ * parameters are passed for further validation.
+ */
+TRACE_EVENT(alloc_vmap_area,
+
+	TP_PROTO(unsigned long addr, unsigned long size, unsigned long align,
+		unsigned long vstart, unsigned long vend, int failed),
+
+	TP_ARGS(addr, size, align, vstart, vend, failed),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, addr)
+		__field(unsigned long, size)
+		__field(unsigned long, align)
+		__field(unsigned long, vstart)
+		__field(unsigned long, vend)
+		__field(int, failed)
+	),
+
+	TP_fast_assign(
+		__entry->addr = addr;
+		__entry->size = size;
+		__entry->align = align;
+		__entry->vstart = vstart;
+		__entry->vend = vend;
+		__entry->failed = failed;
+	),
+
+	TP_printk("va_start: %lu size=%lu align=%lu vstart=0x%lx vend=0x%lx failed=%d",
+		__entry->addr, __entry->size, __entry->align,
+		__entry->vstart, __entry->vend, __entry->failed)
+);
+
+/**
+ * purge_vmap_area_lazy - called when vmap areas were lazily freed
+ * @start:		purging start address
+ * @end:		purging end address
+ * @npurged:	numbed of purged vmap areas
+ *
+ * This event is used for a debug purpose. It gives some
+ * indication about start:end range and how many objects
+ * are released.
+ */
+TRACE_EVENT(purge_vmap_area_lazy,
+
+	TP_PROTO(unsigned long start, unsigned long end,
+		unsigned int npurged),
+
+	TP_ARGS(start, end, npurged),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, start)
+		__field(unsigned long, end)
+		__field(unsigned int, npurged)
+	),
+
+	TP_fast_assign(
+		__entry->start = start;
+		__entry->end = end;
+		__entry->npurged = npurged;
+	),
+
+	TP_printk("start=0x%lx end=0x%lx num_purged=%u",
+		__entry->start, __entry->end, __entry->npurged)
+);
+
+/**
+ * free_vmap_area_noflush - called when a vmap area is freed
+ * @va_start:		a start address of VA
+ * @nr_lazy:		number of current lazy pages
+ * @nr_lazy_max:	number of maximum lazy pages
+ *
+ * This event is used for a debug purpose. It gives some
+ * indication about a VA that is released, number of current
+ * outstanding areas and a maximum allowed threshold before
+ * dropping all of them.
+ */
+TRACE_EVENT(free_vmap_area_noflush,
+
+	TP_PROTO(unsigned long va_start, unsigned long nr_lazy,
+		unsigned long nr_lazy_max),
+
+	TP_ARGS(va_start, nr_lazy, nr_lazy_max),
+
+	TP_STRUCT__entry(
+		__field(unsigned long, va_start)
+		__field(unsigned long, nr_lazy)
+		__field(unsigned long, nr_lazy_max)
+	),
+
+	TP_fast_assign(
+		__entry->va_start = va_start;
+		__entry->nr_lazy = nr_lazy;
+		__entry->nr_lazy_max = nr_lazy_max;
+	),
+
+	TP_printk("va_start=0x%lx nr_lazy=%lu nr_lazy_max=%lu",
+		__entry->va_start, __entry->nr_lazy, __entry->nr_lazy_max)
+);
+
+#endif /*  _TRACE_VMALLOC_H */
+
+/* This part must be outside protection */
+#include <trace/define_trace.h>