Hyper-V: support for free page reporting

Linux has support for free page reporting now (36e66c5) for
virtualized environment. On Hyper-V when virtually backed VMs are
configured, Hyper-V will advertise cold memory discard capability,
when supported. This patch adds the support to hook into the free
page reporting infrastructure and leverage the Hyper-V cold memory
discard hint hypercall to report/free these pages back to the host.

Signed-off-by: Sunil Muthuswamy <[email protected]>

Author: sunilmut

arch/arm64/hyperv/mshyperv.c

@@ -83,13 +83,15 @@ void __init hyperv_early_init(void)
 	/* Get the features and hints from Hyper-V */
 	hv_get_vpreg_128(HV_REGISTER_FEATURES, &result);
 	ms_hyperv.features = result.as32.a;
+	ms_hyperv.priv_high = result.as32.b;
 	ms_hyperv.misc_features = result.as32.c;
 
 	hv_get_vpreg_128(HV_REGISTER_ENLIGHTENMENTS, &result);
 	ms_hyperv.hints = result.as32.a;
 
-	pr_info("Hyper-V: Features 0x%x, hints 0x%x, misc 0x%x\n",
-		ms_hyperv.features, ms_hyperv.hints, ms_hyperv.misc_features);
+	pr_info("Hyper-V: privilege flags low 0x%x, high 0x%x, hints 0x%x, misc 0x%x\n",
+		ms_hyperv.features, ms_hyperv.priv_high, ms_hyperv.hints,
+		ms_hyperv.misc_features);
 
 	/*
 	 * If Hyper-V has crash notifications, set crash_kexec_post_notifiers
@@ -152,6 +154,8 @@ static int __init hyperv_init(void)
 		return -EINVAL;
 	}
 
+	/* Query the VMs extended capability once, so that it can be cached. */
+	hv_query_ext_cap(0);
 	return 0;
 }
 

arch/x86/kernel/cpu/mshyperv.c

@@ -234,11 +234,13 @@ static void __init ms_hyperv_init_platform(void)
 	 * Extract the features and hints
 	 */
 	ms_hyperv.features = cpuid_eax(HYPERV_CPUID_FEATURES);
+	ms_hyperv.priv_high = cpuid_ebx(HYPERV_CPUID_FEATURES);
 	ms_hyperv.misc_features = cpuid_edx(HYPERV_CPUID_FEATURES);
 	ms_hyperv.hints    = cpuid_eax(HYPERV_CPUID_ENLIGHTMENT_INFO);
 
-	pr_info("Hyper-V: features 0x%x, hints 0x%x, misc 0x%x\n",
-		ms_hyperv.features, ms_hyperv.hints, ms_hyperv.misc_features);
+	pr_info("Hyper-V: privilege flags low 0x%x, high 0x%x, hints 0x%x, misc 0x%x\n",
+		ms_hyperv.features, ms_hyperv.priv_high, ms_hyperv.hints,
+		ms_hyperv.misc_features);
 
 	ms_hyperv.max_vp_index = cpuid_eax(HYPERV_CPUID_IMPLEMENT_LIMITS);
 	ms_hyperv.max_lp_index = cpuid_ebx(HYPERV_CPUID_IMPLEMENT_LIMITS);

drivers/Makefile

@@ -160,7 +160,7 @@ obj-$(CONFIG_SOUNDWIRE)		+= soundwire/
 
 # Virtualization drivers
 obj-$(CONFIG_VIRT_DRIVERS)	+= virt/
-obj-$(CONFIG_HYPERV)		+= hv/
+obj-$(subst m,y,$(CONFIG_HYPERV))	+= hv/
 
 obj-$(CONFIG_PM_DEVFREQ)	+= devfreq/
 obj-$(CONFIG_EXTCON)		+= extcon/

drivers/hv/Kconfig

@@ -24,6 +24,7 @@ config HYPERV_UTILS
 config HYPERV_BALLOON
 	tristate "Microsoft Hyper-V Balloon driver"
 	depends on HYPERV
+	select PAGE_REPORTING
 	help
 	  Select this option to enable Hyper-V Balloon driver.
 

drivers/hv/Makefile

@@ -11,3 +11,6 @@ hv_vmbus-y := vmbus_drv.o \
 		 channel_mgmt.o ring_buffer.o hv_trace.o
 hv_vmbus-$(CONFIG_HYPERV_TESTING)	+= hv_debugfs.o
 hv_utils-y := hv_util.o hv_kvp.o hv_snapshot.o hv_fcopy.o hv_utils_transport.o
+
+# Code that must be built-in
+obj-$(subst m,y,$(CONFIG_HYPERV)) += hv_common.o

drivers/hv/hv_balloon.c

@@ -21,6 +21,7 @@
 #include <linux/memory.h>
 #include <linux/notifier.h>
 #include <linux/percpu_counter.h>
+#include <linux/page_reporting.h>
 
 #include <linux/hyperv.h>
 #include <asm/hyperv-tlfs.h>
@@ -563,6 +564,8 @@ struct hv_dynmem_device {
 	 * The negotiated version agreed by host.
 	 */
 	__u32 version;
+
+	struct page_reporting_dev_info pr_dev_info;
 };
 
 static struct hv_dynmem_device dm_device;
@@ -1565,6 +1568,89 @@ static void balloon_onchannelcallback(void *context)
 
 }
 
+/* Hyper-V only supports reporting 2MB pages or higher */
+#define HV_MIN_PAGE_REPORTING_ORDER	9
+#define HV_MIN_PAGE_REPORTING_LEN (HV_HYP_PAGE_SIZE << HV_MIN_PAGE_REPORTING_ORDER)
+static int hv_free_page_report(struct page_reporting_dev_info *pr_dev_info,
+		    struct scatterlist *sgl, unsigned int nents)
+{
+	unsigned long flags;
+	struct hv_memory_hint *hint;
+	int i;
+	u64 status;
+	struct scatterlist *sg;
+
+	WARN_ON_ONCE(nents > HV_MEMORY_HINT_MAX_GPA_PAGE_RANGES);
+	WARN_ON_ONCE(sgl->length < HV_MIN_PAGE_REPORTING_LEN);
+	local_irq_save(flags);
+	hint = *(struct hv_memory_hint **)this_cpu_ptr(hyperv_pcpu_input_arg);
+	if (!hint) {
+		local_irq_restore(flags);
+		return -ENOSPC;
+	}
+
+	hint->type = HV_EXT_MEMORY_HEAT_HINT_TYPE_COLD_DISCARD;
+	hint->reserved = 0;
+	for_each_sg(sgl, sg, nents, i) {
+		union hv_gpa_page_range *range;
+
+		range = &hint->ranges[i];
+		range->address_space = 0;
+		/* page reporting only reports 2MB pages or higher */
+		range->page.largepage = 1;
+		range->page.additional_pages =
+			(sg->length / HV_MIN_PAGE_REPORTING_LEN) - 1;
+		range->page_size = HV_GPA_PAGE_RANGE_PAGE_SIZE_2MB;
+		range->base_large_pfn =
+			page_to_hvpfn(sg_page(sg)) >> HV_MIN_PAGE_REPORTING_ORDER;
+	}
+
+	status = hv_do_rep_hypercall(HV_EXT_CALL_MEMORY_HEAT_HINT, nents, 0,
+				     hint, NULL);
+	local_irq_restore(flags);
+	if ((status & HV_HYPERCALL_RESULT_MASK) != HV_STATUS_SUCCESS) {
+		pr_err("Cold memory discard hypercall failed with status %llx\n",
+			status);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static void enable_page_reporting(void)
+{
+	int ret;
+
+	/* Essentially, validating 'PAGE_REPORTING_MIN_ORDER' is big enough. */
+	if (pageblock_order < HV_MIN_PAGE_REPORTING_ORDER) {
+		pr_debug("Cold memory discard is only supported on 2MB pages and above\n");
+		return;
+	}
+
+	if (!hv_query_ext_cap(HV_EXT_CAPABILITY_MEMORY_COLD_DISCARD_HINT)) {
+		pr_debug("Cold memory discard hint not supported by Hyper-V\n");
+		return;
+	}
+
+	BUILD_BUG_ON(PAGE_REPORTING_CAPACITY > HV_MEMORY_HINT_MAX_GPA_PAGE_RANGES);
+	dm_device.pr_dev_info.report = hv_free_page_report;
+	ret = page_reporting_register(&dm_device.pr_dev_info);
+	if (ret < 0) {
+		dm_device.pr_dev_info.report = NULL;
+		pr_err("Failed to enable cold memory discard: %d\n", ret);
+	} else {
+		pr_info("Cold memory discard hint enabled\n");
+	}
+}
+
+static void disable_page_reporting(void)
+{
+	if (dm_device.pr_dev_info.report) {
+		page_reporting_unregister(&dm_device.pr_dev_info);
+		dm_device.pr_dev_info.report = NULL;
+	}
+}
+
 static int balloon_connect_vsp(struct hv_device *dev)
 {
 	struct dm_version_request version_req;
@@ -1710,6 +1796,7 @@ static int balloon_probe(struct hv_device *dev,
 	if (ret != 0)
 		return ret;
 
+	enable_page_reporting();
 	dm_device.state = DM_INITIALIZED;
 
 	dm_device.thread =
@@ -1724,6 +1811,7 @@ static int balloon_probe(struct hv_device *dev,
 probe_error:
 	dm_device.state = DM_INIT_ERROR;
 	dm_device.thread  = NULL;
+	disable_page_reporting();
 	vmbus_close(dev->channel);
 #ifdef CONFIG_MEMORY_HOTPLUG
 	unregister_memory_notifier(&hv_memory_nb);
@@ -1746,6 +1834,7 @@ static int balloon_remove(struct hv_device *dev)
 	cancel_work_sync(&dm->ha_wrk.wrk);
 
 	kthread_stop(dm->thread);
+	disable_page_reporting();
 	vmbus_close(dev->channel);
 #ifdef CONFIG_MEMORY_HOTPLUG
 	unregister_memory_notifier(&hv_memory_nb);

drivers/hv/hv_common.c

@@ -0,0 +1,67 @@
+// SPDX-License-Identifier: GPL-2.0
+
+/*
+ * Architecture neutral utility routines for interacting with
+ * Hyper-V. This file is specifically for code that must be
+ * built-in to the kernel image when CONFIG_HYPERV is set
+ * (vs. being in a module) because it is called from architecture
+ * specific code under arch/.
+ *
+ * Copyright (C) 2021, Microsoft, Inc.
+ *
+ * Author : Michael Kelley <[email protected]>
+ */
+
+#include <linux/types.h>
+#include <linux/export.h>
+#include <linux/bitfield.h>
+#include <asm/hyperv-tlfs.h>
+#include <asm/mshyperv.h>
+
+
+/* Bit mask of the extended capability to query: see HV_EXT_CAPABILITY_xxx */
+bool hv_query_ext_cap(u64 cap_query)
+{
+	/*
+	 * The address of the 'hv_extended_cap' variable will be used as an
+	 * output parameter to the hypercall below and so it should be
+	 * compatible with 'virt_to_phys'. Which means, it's address should be
+	 * directly mapped. Use 'static' to keep it compatible; stack variables
+	 * can be virtually mapped, making them imcompatible with
+	 * 'virt_to_phys'.
+	 * Hypercall input/output addresses should also be 8-byte aligned.
+	 */
+	static u64 hv_extended_cap __aligned(8);
+	static bool hv_extended_cap_queried;
+	u64 status;
+
+	/*
+	 * Querying extended capabilities is an extended hypercall. Check if the
+	 * partition supports extended hypercall, first.
+	 */
+	if (!(ms_hyperv.priv_high & HV_ENABLE_EXTENDED_HYPERCALLS))
+		return false;
+
+	/* Extended capabilities do not change at runtime. */
+	if (hv_extended_cap_queried)
+		return hv_extended_cap & cap_query;
+
+	status = hv_do_hypercall(HV_EXT_CALL_QUERY_CAPABILITIES, NULL,
+				 &hv_extended_cap);
+
+	/*
+	 * The query extended capabilities hypercall should not fail under
+	 * any normal circumstances. Avoid repeatedly making the hypercall, on
+	 * error.
+	 */
+	hv_extended_cap_queried = true;
+	status &= HV_HYPERCALL_RESULT_MASK;
+	if (status != HV_STATUS_SUCCESS) {
+		pr_err("Hyper-V: Extended query capabilities hypercall failed 0x%llx\n",
+		       status);
+		return false;
+	}
+
+	return hv_extended_cap & cap_query;
+}
+EXPORT_SYMBOL_GPL(hv_query_ext_cap);

include/asm-generic/hyperv-tlfs.h

@@ -89,6 +89,7 @@
 #define HV_ACCESS_STATS				BIT(8)
 #define HV_DEBUGGING				BIT(11)
 #define HV_CPU_POWER_MANAGEMENT			BIT(12)
+#define HV_ENABLE_EXTENDED_HYPERCALLS		BIT(20)
 
 
 /*
@@ -152,11 +153,18 @@ struct ms_hyperv_tsc_page {
 #define HVCALL_FLUSH_GUEST_PHYSICAL_ADDRESS_SPACE 0x00af
 #define HVCALL_FLUSH_GUEST_PHYSICAL_ADDRESS_LIST 0x00b0
 
+/* Extended hypercalls */
+#define HV_EXT_CALL_QUERY_CAPABILITIES		0x8001
+#define HV_EXT_CALL_MEMORY_HEAT_HINT		0x8003
+
 #define HV_FLUSH_ALL_PROCESSORS			BIT(0)
 #define HV_FLUSH_ALL_VIRTUAL_ADDRESS_SPACES	BIT(1)
 #define HV_FLUSH_NON_GLOBAL_MAPPINGS_ONLY	BIT(2)
 #define HV_FLUSH_USE_EXTENDED_RANGE_FORMAT	BIT(3)
 
+/* Extended capability bits */
+#define HV_EXT_CAPABILITY_MEMORY_COLD_DISCARD_HINT BIT(8)
+
 enum HV_GENERIC_SET_FORMAT {
 	HV_GENERIC_SET_SPARSE_4K,
 	HV_GENERIC_SET_ALL,
@@ -401,15 +409,23 @@ struct hv_guest_mapping_flush {
  *  by the bitwidth of "additional_pages" in union hv_gpa_page_range.
  */
 #define HV_MAX_FLUSH_PAGES (2048)
+#define HV_GPA_PAGE_RANGE_PAGE_SIZE_2MB		0
+#define HV_GPA_PAGE_RANGE_PAGE_SIZE_1GB		1
 
-/* HvFlushGuestPhysicalAddressList hypercall */
+/* HvFlushGuestPhysicalAddressList, HvExtCallMemoryHeatHint hypercall */
 union hv_gpa_page_range {
 	u64 address_space;
 	struct {
 		u64 additional_pages:11;
 		u64 largepage:1;
 		u64 basepfn:52;
 	} page;
+	struct {
+		u64 reserved:12;
+		u64 page_size:1;
+		u64 reserved1:8;
+		u64 base_large_pfn:43;
+	};
 };
 
 /*
@@ -529,4 +545,20 @@ struct hv_set_vp_registers_input {
 	} element[];
 } __packed;
 
+/*
+ * The whole argument should fit in a page to be able to pass to the hypervisor
+ * in one hypercall.
+ */
+#define HV_MEMORY_HINT_MAX_GPA_PAGE_RANGES  \
+	((HV_HYP_PAGE_SIZE - sizeof(struct hv_memory_hint)) / \
+		sizeof(union hv_gpa_page_range))
+
+/* HvExtCallMemoryHeatHint hypercall */
+#define HV_EXT_MEMORY_HEAT_HINT_TYPE_COLD_DISCARD	2
+struct hv_memory_hint {
+	u64 type:2;
+	u64 reserved:62;
+	union hv_gpa_page_range ranges[];
+} __packed;
+
 #endif

include/asm-generic/mshyperv.h

@@ -28,6 +28,7 @@
 
 struct ms_hyperv_info {
 	u32 features;
+	u32 priv_high;
 	u32 misc_features;
 	u32 hints;
 	u32 nested_features;
@@ -205,6 +206,7 @@ void hyperv_report_panic(struct pt_regs *regs, long err, bool in_die);
 bool hv_is_hyperv_initialized(void);
 bool hv_is_hibernation_supported(void);
 void hyperv_cleanup(void);
+bool hv_query_ext_cap(u64 cap_query);
 #else /* CONFIG_HYPERV */
 static inline bool hv_is_hyperv_initialized(void) { return false; }
 static inline bool hv_is_hibernation_supported(void) { return false; }