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, 256 insertions, 0 deletions

diff --git a/scripts/coccinelle/api/kvmalloc.cocci b/scripts/coccinelle/api/kvmalloc.cocci
new file mode 100644
index 000000000..5ddcb76b7
--- /dev/null
+++ b/scripts/coccinelle/api/kvmalloc.cocci
@@ -0,0 +1,256 @@
+// SPDX-License-Identifier: GPL-2.0-only
+///
+/// Find if/else condition with kmalloc/vmalloc calls.
+/// Suggest to use kvmalloc instead. Same for kvfree.
+///
+// Confidence: High
+// Copyright: (C) 2020 Denis Efremov ISPRAS
+// Options: --no-includes --include-headers
+//
+
+virtual patch
+virtual report
+virtual org
+virtual context
+
+@initialize:python@
+@@
+filter = frozenset(['kvfree'])
+
+def relevant(p):
+    return not (filter & {el.current_element for el in p})
+
+@kvmalloc depends on !patch@
+expression E, E1, size;
+identifier flags;
+binary operator cmp = {<=, <, ==, >, >=};
+identifier x;
+type T;
+position p;
+@@
+
+(
+* if (size cmp E1 || ...)@p {
+    ...
+*    E = \(kmalloc\|kzalloc\|kcalloc\|kmalloc_node\|kzalloc_node\|
+*          kmalloc_array\|kmalloc_array_node\|kcalloc_node\)
+*          (..., size, \(flags\|GFP_KERNEL\|\(GFP_KERNEL\|flags\)|__GFP_NOWARN\), ...)
+    ...
+  } else {
+    ...
+*    E = \(vmalloc\|vzalloc\|vmalloc_node\|vzalloc_node\)(..., size, ...)
+    ...
+  }
+|
+* E = \(kmalloc\|kzalloc\|kcalloc\|kmalloc_node\|kzalloc_node\|
+*       kmalloc_array\|kmalloc_array_node\|kcalloc_node\)
+*       (..., size, \(flags\|GFP_KERNEL\|\(GFP_KERNEL\|flags\)|__GFP_NOWARN\), ...)
+  ... when != E = E1
+      when != size = E1
+      when any
+* if (E == NULL)@p {
+    ...
+*   E = \(vmalloc\|vzalloc\|vmalloc_node\|vzalloc_node\)(..., size, ...)
+    ...
+  }
+|
+* T x = \(kmalloc\|kzalloc\|kcalloc\|kmalloc_node\|kzalloc_node\|
+*         kmalloc_array\|kmalloc_array_node\|kcalloc_node\)
+*         (..., size, \(flags\|GFP_KERNEL\|\(GFP_KERNEL\|flags\)|__GFP_NOWARN\), ...);
+  ... when != x = E1
+      when != size = E1
+      when any
+* if (x == NULL)@p {
+    ...
+*   x = \(vmalloc\|vzalloc\|vmalloc_node\|vzalloc_node\)(..., size, ...)
+    ...
+  }
+)
+
+@kvfree depends on !patch@
+expression E;
+position p : script:python() { relevant(p) };
+@@
+
+* if (is_vmalloc_addr(E))@p {
+    ...
+*   vfree(E)
+    ...
+  } else {
+    ... when != krealloc(E, ...)
+        when any
+*   \(kfree\|kfree_sensitive\)(E)
+    ...
+  }
+
+@depends on patch@
+expression E, E1, size, node;
+binary operator cmp = {<=, <, ==, >, >=};
+identifier flags, x;
+type T;
+@@
+
+(
+- if (size cmp E1)
+-    E = kmalloc(size, flags);
+- else
+-    E = vmalloc(size);
++ E = kvmalloc(size, flags);
+|
+- if (size cmp E1)
+-    E = kmalloc(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\));
+- else
+-    E = vmalloc(size);
++ E = kvmalloc(size, GFP_KERNEL);
+|
+- E = kmalloc(size, flags | __GFP_NOWARN);
+- if (E == NULL)
+-   E = vmalloc(size);
++ E = kvmalloc(size, flags);
+|
+- E = kmalloc(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\));
+- if (E == NULL)
+-   E = vmalloc(size);
++ E = kvmalloc(size, GFP_KERNEL);
+|
+- T x = kmalloc(size, flags | __GFP_NOWARN);
+- if (x == NULL)
+-   x = vmalloc(size);
++ T x = kvmalloc(size, flags);
+|
+- T x = kmalloc(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\));
+- if (x == NULL)
+-   x = vmalloc(size);
++ T x = kvmalloc(size, GFP_KERNEL);
+|
+- if (size cmp E1)
+-    E = kzalloc(size, flags);
+- else
+-    E = vzalloc(size);
++ E = kvzalloc(size, flags);
+|
+- if (size cmp E1)
+-    E = kzalloc(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\));
+- else
+-    E = vzalloc(size);
++ E = kvzalloc(size, GFP_KERNEL);
+|
+- E = kzalloc(size, flags | __GFP_NOWARN);
+- if (E == NULL)
+-   E = vzalloc(size);
++ E = kvzalloc(size, flags);
+|
+- E = kzalloc(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\));
+- if (E == NULL)
+-   E = vzalloc(size);
++ E = kvzalloc(size, GFP_KERNEL);
+|
+- T x = kzalloc(size, flags | __GFP_NOWARN);
+- if (x == NULL)
+-   x = vzalloc(size);
++ T x = kvzalloc(size, flags);
+|
+- T x = kzalloc(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\));
+- if (x == NULL)
+-   x = vzalloc(size);
++ T x = kvzalloc(size, GFP_KERNEL);
+|
+- if (size cmp E1)
+-    E = kmalloc_node(size, flags, node);
+- else
+-    E = vmalloc_node(size, node);
++ E = kvmalloc_node(size, flags, node);
+|
+- if (size cmp E1)
+-    E = kmalloc_node(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\), node);
+- else
+-    E = vmalloc_node(size, node);
++ E = kvmalloc_node(size, GFP_KERNEL, node);
+|
+- E = kmalloc_node(size, flags | __GFP_NOWARN, node);
+- if (E == NULL)
+-   E = vmalloc_node(size, node);
++ E = kvmalloc_node(size, flags, node);
+|
+- E = kmalloc_node(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\), node);
+- if (E == NULL)
+-   E = vmalloc_node(size, node);
++ E = kvmalloc_node(size, GFP_KERNEL, node);
+|
+- T x = kmalloc_node(size, flags | __GFP_NOWARN, node);
+- if (x == NULL)
+-   x = vmalloc_node(size, node);
++ T x = kvmalloc_node(size, flags, node);
+|
+- T x = kmalloc_node(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\), node);
+- if (x == NULL)
+-   x = vmalloc_node(size, node);
++ T x = kvmalloc_node(size, GFP_KERNEL, node);
+|
+- if (size cmp E1)
+-    E = kvzalloc_node(size, flags, node);
+- else
+-    E = vzalloc_node(size, node);
++ E = kvzalloc_node(size, flags, node);
+|
+- if (size cmp E1)
+-    E = kvzalloc_node(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\), node);
+- else
+-    E = vzalloc_node(size, node);
++ E = kvzalloc_node(size, GFP_KERNEL, node);
+|
+- E = kvzalloc_node(size, flags | __GFP_NOWARN, node);
+- if (E == NULL)
+-   E = vzalloc_node(size, node);
++ E = kvzalloc_node(size, flags, node);
+|
+- E = kvzalloc_node(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\), node);
+- if (E == NULL)
+-   E = vzalloc_node(size, node);
++ E = kvzalloc_node(size, GFP_KERNEL, node);
+|
+- T x = kvzalloc_node(size, flags | __GFP_NOWARN, node);
+- if (x == NULL)
+-   x = vzalloc_node(size, node);
++ T x = kvzalloc_node(size, flags, node);
+|
+- T x = kvzalloc_node(size, \(GFP_KERNEL\|GFP_KERNEL|__GFP_NOWARN\), node);
+- if (x == NULL)
+-   x = vzalloc_node(size, node);
++ T x = kvzalloc_node(size, GFP_KERNEL, node);
+)
+
+@depends on patch@
+expression E;
+position p : script:python() { relevant(p) };
+@@
+
+- if (is_vmalloc_addr(E))@p
+-   vfree(E);
+- else
+-   kfree(E);
++ kvfree(E);
+
+@script: python depends on report@
+p << kvmalloc.p;
+@@
+
+coccilib.report.print_report(p[0], "WARNING opportunity for kvmalloc")
+
+@script: python depends on org@
+p << kvmalloc.p;
+@@
+
+coccilib.org.print_todo(p[0], "WARNING opportunity for kvmalloc")
+
+@script: python depends on report@
+p << kvfree.p;
+@@
+
+coccilib.report.print_report(p[0], "WARNING opportunity for kvfree")
+
+@script: python depends on org@
+p << kvfree.p;
+@@
+
+coccilib.org.print_todo(p[0], "WARNING opportunity for kvfree")