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, 175 insertions, 0 deletions

diff --git a/Documentation/netlink/specs/dev-energymodel.yaml b/Documentation/netlink/specs/dev-energymodel.yaml
new file mode 100644
index 000000000..11faabfdf
--- /dev/null
+++ b/Documentation/netlink/specs/dev-energymodel.yaml
@@ -0,0 +1,175 @@
+# SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause)
+#
+# Copyright (c) 2025 Valve Corporation.
+#
+---
+name: dev-energymodel
+
+doc: |
+  Energy model netlink interface to notify its changes.
+
+protocol: genetlink
+
+uapi-header: linux/dev_energymodel.h
+
+definitions:
+  -
+    type: flags
+    name: perf-state-flags
+    entries:
+      -
+        name: perf-state-inefficient
+        doc: >-
+          The performance state is inefficient. There is in this perf-domain,
+          another performance state with a higher frequency but a lower or
+          equal power cost.
+  -
+    type: flags
+    name: perf-domain-flags
+    entries:
+      -
+        name: perf-domain-microwatts
+        doc: >-
+          The power values are in micro-Watts or some other scale.
+      -
+        name: perf-domain-skip-inefficiencies
+        doc: >-
+          Skip inefficient states when estimating energy consumption.
+      -
+        name: perf-domain-artificial
+        doc: >-
+          The power values are artificial and might be created by platform
+          missing real power information.
+
+attribute-sets:
+  -
+    name: perf-domain
+    doc: >-
+      Information on a single performance domains.
+    attributes:
+      -
+        name: pad
+        type: pad
+      -
+        name: perf-domain-id
+        type: u32
+        doc: >-
+          A unique ID number for each performance domain.
+      -
+        name: flags
+        type: u64
+        doc: >-
+          Bitmask of performance domain flags.
+        enum: perf-domain-flags
+      -
+        name: cpus
+        type: u64
+        multi-attr: true
+        doc: >-
+          CPUs that belong to this performance domain.
+  -
+    name: perf-table
+    doc: >-
+      Performance states table.
+    attributes:
+      -
+        name: perf-domain-id
+        type: u32
+        doc: >-
+          A unique ID number for each performance domain.
+      -
+        name: perf-state
+        type: nest
+        nested-attributes: perf-state
+        multi-attr: true
+  -
+    name: perf-state
+    doc: >-
+      Performance state of a performance domain.
+    attributes:
+      -
+        name: pad
+        type: pad
+      -
+        name: performance
+        type: u64
+        doc: >-
+          CPU performance (capacity) at a given frequency.
+      -
+        name: frequency
+        type: u64
+        doc: >-
+          The frequency in KHz, for consistency with CPUFreq.
+      -
+        name: power
+        type: u64
+        doc: >-
+          The power consumed at this level (by 1 CPU or by a registered
+          device). It can be a total power: static and dynamic.
+      -
+        name: cost
+        type: u64
+        doc: >-
+          The cost coefficient associated with this level, used during energy
+          calculation. Equal to: power * max_frequency / frequency.
+      -
+        name: flags
+        type: u64
+        doc: >-
+          Bitmask of performance state flags.
+        enum: perf-state-flags
+
+operations:
+  list:
+    -
+      name: get-perf-domains
+      attribute-set: perf-domain
+      doc: Get the list of information for all performance domains.
+      do:
+        request:
+          attributes:
+            - perf-domain-id
+        reply:
+          attributes: &perf-domain-attrs
+            - pad
+            - perf-domain-id
+            - flags
+            - cpus
+      dump:
+        reply:
+          attributes: *perf-domain-attrs
+    -
+      name: get-perf-table
+      attribute-set: perf-table
+      doc: Get the energy model table of a performance domain.
+      do:
+        request:
+          attributes:
+            - perf-domain-id
+        reply:
+          attributes:
+            - perf-domain-id
+            - perf-state
+    -
+      name: perf-domain-created
+      doc: A performance domain is created.
+      notify: get-perf-table
+      mcgrp: event
+    -
+      name: perf-domain-updated
+      doc: A performance domain is updated.
+      notify: get-perf-table
+      mcgrp: event
+    -
+      name: perf-domain-deleted
+      doc: A performance domain is deleted.
+      attribute-set: perf-table
+      event:
+        attributes:
+          - perf-domain-id
+      mcgrp: event
+
+mcast-groups:
+  list:
+    -
+      name: event