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, 192 insertions, 0 deletions

diff --git a/drivers/reset/sti/reset-syscfg.c b/drivers/reset/sti/reset-syscfg.c
new file mode 100644
index 000000000..38f78d78f
--- /dev/null
+++ b/drivers/reset/sti/reset-syscfg.c
@@ -0,0 +1,192 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2013 STMicroelectronics Limited
+ * Author: Stephen Gallimore <stephen.gallimore@st.com>
+ *
+ * Inspired by mach-imx/src.c
+ */
+#include <linux/kernel.h>
+#include <linux/platform_device.h>
+#include <linux/property.h>
+#include <linux/module.h>
+#include <linux/err.h>
+#include <linux/types.h>
+#include <linux/of.h>
+#include <linux/regmap.h>
+#include <linux/mfd/syscon.h>
+
+#include "reset-syscfg.h"
+
+/**
+ * struct syscfg_reset_channel - Reset channel regmap configuration
+ *
+ * @reset: regmap field for the channel's reset bit.
+ * @ack: regmap field for the channel's ack bit (optional).
+ */
+struct syscfg_reset_channel {
+	struct regmap_field *reset;
+	struct regmap_field *ack;
+};
+
+/**
+ * struct syscfg_reset_controller - A reset controller which groups together
+ * a set of related reset bits, which may be located in different system
+ * configuration registers.
+ *
+ * @rst: base reset controller structure.
+ * @active_low: are the resets in this controller active low, i.e. clearing
+ *              the reset bit puts the hardware into reset.
+ * @channels: An array of reset channels for this controller.
+ */
+struct syscfg_reset_controller {
+	struct reset_controller_dev rst;
+	bool active_low;
+	struct syscfg_reset_channel channels[];
+};
+
+#define to_syscfg_reset_controller(_rst) \
+	container_of(_rst, struct syscfg_reset_controller, rst)
+
+static int syscfg_reset_program_hw(struct reset_controller_dev *rcdev,
+				   unsigned long idx, int assert)
+{
+	struct syscfg_reset_controller *rst = to_syscfg_reset_controller(rcdev);
+	const struct syscfg_reset_channel *ch;
+	u32 ctrl_val = rst->active_low ? !assert : !!assert;
+	int err;
+
+	if (idx >= rcdev->nr_resets)
+		return -EINVAL;
+
+	ch = &rst->channels[idx];
+
+	err = regmap_field_write(ch->reset, ctrl_val);
+	if (err)
+		return err;
+
+	if (ch->ack) {
+		u32 ack_val;
+
+		err = regmap_field_read_poll_timeout(ch->ack, ack_val, (ack_val == ctrl_val),
+						     100, USEC_PER_SEC);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int syscfg_reset_assert(struct reset_controller_dev *rcdev,
+			       unsigned long idx)
+{
+	return syscfg_reset_program_hw(rcdev, idx, true);
+}
+
+static int syscfg_reset_deassert(struct reset_controller_dev *rcdev,
+				 unsigned long idx)
+{
+	return syscfg_reset_program_hw(rcdev, idx, false);
+}
+
+static int syscfg_reset_dev(struct reset_controller_dev *rcdev,
+			    unsigned long idx)
+{
+	int err;
+
+	err = syscfg_reset_assert(rcdev, idx);
+	if (err)
+		return err;
+
+	return syscfg_reset_deassert(rcdev, idx);
+}
+
+static int syscfg_reset_status(struct reset_controller_dev *rcdev,
+			       unsigned long idx)
+{
+	struct syscfg_reset_controller *rst = to_syscfg_reset_controller(rcdev);
+	const struct syscfg_reset_channel *ch;
+	u32 ret_val = 0;
+	int err;
+
+	if (idx >= rcdev->nr_resets)
+		return -EINVAL;
+
+	ch = &rst->channels[idx];
+	if (ch->ack)
+		err = regmap_field_read(ch->ack, &ret_val);
+	else
+		err = regmap_field_read(ch->reset, &ret_val);
+	if (err)
+		return err;
+
+	return rst->active_low ? !ret_val : !!ret_val;
+}
+
+static const struct reset_control_ops syscfg_reset_ops = {
+	.reset    = syscfg_reset_dev,
+	.assert   = syscfg_reset_assert,
+	.deassert = syscfg_reset_deassert,
+	.status   = syscfg_reset_status,
+};
+
+static int syscfg_reset_controller_register(struct device *dev,
+				const struct syscfg_reset_controller_data *data)
+{
+	struct syscfg_reset_controller *rc;
+	int i, err;
+
+	rc = devm_kzalloc(dev, struct_size(rc, channels, data->nr_channels), GFP_KERNEL);
+	if (!rc)
+		return -ENOMEM;
+
+	rc->rst.ops = &syscfg_reset_ops;
+	rc->rst.of_node = dev->of_node;
+	rc->rst.nr_resets = data->nr_channels;
+	rc->active_low = data->active_low;
+
+	for (i = 0; i < data->nr_channels; i++) {
+		struct regmap *map;
+		struct regmap_field *f;
+		const char *compatible = data->channels[i].compatible;
+
+		map = syscon_regmap_lookup_by_compatible(compatible);
+		if (IS_ERR(map))
+			return PTR_ERR(map);
+
+		f = devm_regmap_field_alloc(dev, map, data->channels[i].reset);
+		if (IS_ERR(f))
+			return PTR_ERR(f);
+
+		rc->channels[i].reset = f;
+
+		if (!data->wait_for_ack)
+			continue;
+
+		f = devm_regmap_field_alloc(dev, map, data->channels[i].ack);
+		if (IS_ERR(f))
+			return PTR_ERR(f);
+
+		rc->channels[i].ack = f;
+	}
+
+	err = reset_controller_register(&rc->rst);
+	if (!err)
+		dev_info(dev, "registered\n");
+
+	return err;
+}
+
+int syscfg_reset_probe(struct platform_device *pdev)
+{
+	struct device *dev = pdev ? &pdev->dev : NULL;
+	const void *data;
+
+	if (!dev || !dev->driver)
+		return -ENODEV;
+
+	data = device_get_match_data(&pdev->dev);
+	if (!data)
+		return -EINVAL;
+
+	return syscfg_reset_controller_register(dev, data);
+}