drm/i915/gvt: add KVMGT support

KVMGT is the MPT implementation based on VFIO/KVM. It provides
a kvmgt_mpt ops to gvt for vGPU access mediation, e.g. to
mediate and emulate the MMIO accesses, to inject interrupts
to vGPU user, to intercept the GTT writing and replace it with
DMA-able address, to write-protect guest PPGTT table for
shadowing synchronization, etc. This patch provides the MPT
implementation for GVT, not yet functional due to theabsence
of mdev.

It's built as kvmgt.ko, depends on vfio.ko, kvm.ko and mdev.ko,
and being required by i915.ko. To not introduce hard dependency
in i915.ko, we used indirect symbol reference. But that means
users have to include kvmgt.ko into init ramdisk if their
i915.ko is included.
Signed-off-by: Kevin Tian <kevin.tian@intel.com>
Signed-off-by: Xiaoguang Chen <xiaoguang.chen@intel.com>
Signed-off-by: Jike Song <jike.song@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>

drivers/gpu/drm/i915/Kconfig

@@ -107,6 +107,15 @@ config DRM_I915_GVT
 
 	  If in doubt, say "N".
 
+config DRM_I915_GVT_KVMGT
+	tristate "Enable KVM/VFIO support for Intel GVT-g"
+	depends on DRM_I915_GVT
+	depends on KVM
+	default n
+	help
+	  Choose this option if you want to enable KVMGT support for
+	  Intel GVT-g.
+
 menu "drm/i915 Debugging"
 depends on DRM_I915
 depends on EXPERT

drivers/gpu/drm/i915/gvt/Makefile

@@ -3,5 +3,8 @@ GVT_SOURCE := gvt.o aperture_gm.o handlers.o vgpu.o trace_points.o firmware.o \
 	interrupt.o gtt.o cfg_space.o opregion.o mmio.o display.o edid.o \
 	execlist.o scheduler.o sched_policy.o render.o cmd_parser.o
 
-ccflags-y                      += -I$(src) -I$(src)/$(GVT_DIR) -Wall
-i915-y			       += $(addprefix $(GVT_DIR)/, $(GVT_SOURCE))
+ccflags-y				+= -I$(src) -I$(src)/$(GVT_DIR) -Wall
+i915-y					+= $(addprefix $(GVT_DIR)/, $(GVT_SOURCE))
+
+CFLAGS_kvmgt.o				:= -Wno-unused-function
+obj-$(CONFIG_DRM_I915_GVT_KVMGT)	+= $(GVT_DIR)/kvmgt.o

drivers/gpu/drm/i915/gvt/gvt.c

@@ -84,10 +84,12 @@ int intel_gvt_init_host(void)
 				symbol_get(xengt_mpt), "xengt");
 		intel_gvt_host.hypervisor_type = INTEL_GVT_HYPERVISOR_XEN;
 	} else {
+#if IS_ENABLED(CONFIG_DRM_I915_GVT_KVMGT)
 		/* not in Xen. Try KVMGT */
 		intel_gvt_host.mpt = try_then_request_module(
-				symbol_get(kvmgt_mpt), "kvm");
+				symbol_get(kvmgt_mpt), "kvmgt");
 		intel_gvt_host.hypervisor_type = INTEL_GVT_HYPERVISOR_KVM;
+#endif
 	}
 
 	/* Fail to load MPT modules - bail out */

drivers/gpu/drm/i915/gvt/gvt.h

@@ -161,6 +161,20 @@ struct intel_vgpu {
 	DECLARE_BITMAP(tlb_handle_pending, I915_NUM_ENGINES);
 	struct i915_gem_context *shadow_ctx;
 	struct notifier_block shadow_ctx_notifier_block;
+
+#if IS_ENABLED(CONFIG_DRM_I915_GVT_KVMGT)
+	struct {
+		struct device *mdev;
+		struct vfio_region *region;
+		int num_regions;
+		struct eventfd_ctx *intx_trigger;
+		struct eventfd_ctx *msi_trigger;
+		struct rb_root cache;
+		struct mutex cache_lock;
+		void *vfio_group;
+		struct notifier_block iommu_notifier;
+	} vdev;
+#endif
 };
 
 struct intel_gvt_gm {

drivers/gpu/drm/i915/gvt/kvmgt.c

+/*
+ * KVMGT - the implementation of Intel mediated pass-through framework for KVM
+ *
+ * Copyright(c) 2014-2016 Intel Corporation. All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Authors:
+ *    Kevin Tian <kevin.tian@intel.com>
+ *    Jike Song <jike.song@intel.com>
+ *    Xiaoguang Chen <xiaoguang.chen@intel.com>
+ */
+
+#include <linux/init.h>
+#include <linux/device.h>
+#include <linux/mm.h>
+#include <linux/types.h>
+#include <linux/list.h>
+#include <linux/rbtree.h>
+#include <linux/spinlock.h>
+#include <linux/eventfd.h>
+#include <linux/uuid.h>
+#include <linux/kvm_host.h>
+#include <linux/vfio.h>
+
+#include "i915_drv.h"
+#include "gvt.h"
+
+#if IS_ENABLED(CONFIG_VFIO_MDEV)
+#include <linux/mdev.h>
+#else
+static inline long vfio_pin_pages(struct device *dev, unsigned long *user_pfn,
+			long npage, int prot, unsigned long *phys_pfn)
+{
+	return 0;
+}
+static inline long vfio_unpin_pages(struct device *dev, unsigned long *pfn,
+			long npage)
+{
+	return 0;
+}
+#endif
+
+static const struct intel_gvt_ops *intel_gvt_ops;
+
+
+/* helper macros copied from vfio-pci */
+#define VFIO_PCI_OFFSET_SHIFT   40
+#define VFIO_PCI_OFFSET_TO_INDEX(off)   (off >> VFIO_PCI_OFFSET_SHIFT)
+#define VFIO_PCI_INDEX_TO_OFFSET(index) ((u64)(index) << VFIO_PCI_OFFSET_SHIFT)
+#define VFIO_PCI_OFFSET_MASK    (((u64)(1) << VFIO_PCI_OFFSET_SHIFT) - 1)
+
+struct vfio_region {
+	u32				type;
+	u32				subtype;
+	size_t				size;
+	u32				flags;
+};
+
+struct kvmgt_pgfn {
+	gfn_t gfn;
+	struct hlist_node hnode;
+};
+
+struct kvmgt_guest_info {
+	struct kvm *kvm;
+	struct intel_vgpu *vgpu;
+	struct kvm_page_track_notifier_node track_node;
+#define NR_BKT (1 << 18)
+	struct hlist_head ptable[NR_BKT];
+#undef NR_BKT
+};
+
+struct gvt_dma {
+	struct rb_node node;
+	gfn_t gfn;
+	kvm_pfn_t pfn;
+};
+
+static struct gvt_dma *__gvt_cache_find(struct intel_vgpu *vgpu, gfn_t gfn)
+{
+	struct rb_node *node = vgpu->vdev.cache.rb_node;
+	struct gvt_dma *ret = NULL;
+
+	while (node) {
+		struct gvt_dma *itr = rb_entry(node, struct gvt_dma, node);
+
+		if (gfn < itr->gfn)
+			node = node->rb_left;
+		else if (gfn > itr->gfn)
+			node = node->rb_right;
+		else {
+			ret = itr;
+			goto out;
+		}
+	}
+
+out:
+	return ret;
+}
+
+static kvm_pfn_t gvt_cache_find(struct intel_vgpu *vgpu, gfn_t gfn)
+{
+	struct gvt_dma *entry;
+
+	mutex_lock(&vgpu->vdev.cache_lock);
+	entry = __gvt_cache_find(vgpu, gfn);
+	mutex_unlock(&vgpu->vdev.cache_lock);
+
+	return entry == NULL ? 0 : entry->pfn;
+}
+
+static void gvt_cache_add(struct intel_vgpu *vgpu, gfn_t gfn, kvm_pfn_t pfn)
+{
+	struct gvt_dma *new, *itr;
+	struct rb_node **link = &vgpu->vdev.cache.rb_node, *parent = NULL;
+
+	new = kzalloc(sizeof(struct gvt_dma), GFP_KERNEL);
+	if (!new)
+		return;
+
+	new->gfn = gfn;
+	new->pfn = pfn;
+
+	mutex_lock(&vgpu->vdev.cache_lock);
+	while (*link) {
+		parent = *link;
+		itr = rb_entry(parent, struct gvt_dma, node);
+
+		if (gfn == itr->gfn)
+			goto out;
+		else if (gfn < itr->gfn)
+			link = &parent->rb_left;
+		else
+			link = &parent->rb_right;
+	}
+
+	rb_link_node(&new->node, parent, link);
+	rb_insert_color(&new->node, &vgpu->vdev.cache);
+	mutex_unlock(&vgpu->vdev.cache_lock);
+	return;
+
+out:
+	mutex_unlock(&vgpu->vdev.cache_lock);
+	kfree(new);
+}
+
+static void __gvt_cache_remove_entry(struct intel_vgpu *vgpu,
+				struct gvt_dma *entry)
+{
+	rb_erase(&entry->node, &vgpu->vdev.cache);
+	kfree(entry);
+}
+
+static void gvt_cache_remove(struct intel_vgpu *vgpu, gfn_t gfn)
+{
+	struct device *dev = vgpu->vdev.mdev;
+	struct gvt_dma *this;
+	unsigned long pfn;
+
+	mutex_lock(&vgpu->vdev.cache_lock);
+	this  = __gvt_cache_find(vgpu, gfn);
+	if (!this) {
+		mutex_unlock(&vgpu->vdev.cache_lock);
+		return;
+	}
+
+	pfn = this->pfn;
+	WARN_ON((vfio_unpin_pages(dev, &pfn, 1) != 1));
+	__gvt_cache_remove_entry(vgpu, this);
+	mutex_unlock(&vgpu->vdev.cache_lock);
+}
+
+static void gvt_cache_init(struct intel_vgpu *vgpu)
+{
+	vgpu->vdev.cache = RB_ROOT;
+	mutex_init(&vgpu->vdev.cache_lock);
+}
+
+static void gvt_cache_destroy(struct intel_vgpu *vgpu)
+{
+	struct gvt_dma *dma;
+	struct rb_node *node = NULL;
+	struct device *dev = vgpu->vdev.mdev;
+	unsigned long pfn;
+
+	mutex_lock(&vgpu->vdev.cache_lock);
+	while ((node = rb_first(&vgpu->vdev.cache))) {
+		dma = rb_entry(node, struct gvt_dma, node);
+		pfn = dma->pfn;
+
+		vfio_unpin_pages(dev, &pfn, 1);
+		__gvt_cache_remove_entry(vgpu, dma);
+	}
+	mutex_unlock(&vgpu->vdev.cache_lock);
+}
+
+static struct intel_vgpu_type *intel_gvt_find_vgpu_type(struct intel_gvt *gvt,
+		const char *name)
+{
+	int i;
+	struct intel_vgpu_type *t;
+	const char *driver_name = dev_driver_string(
+			&gvt->dev_priv->drm.pdev->dev);
+
+	for (i = 0; i < gvt->num_types; i++) {
+		t = &gvt->types[i];
+		if (!strncmp(t->name, name + strlen(driver_name) + 1,
+			sizeof(t->name)))
+			return t;
+	}
+
+	return NULL;
+}
+
+static struct attribute *type_attrs[] = {
+	NULL,
+};
+
+static struct attribute_group *intel_vgpu_type_groups[] = {
+	[0 ... NR_MAX_INTEL_VGPU_TYPES - 1] = NULL,
+};
+
+static bool intel_gvt_init_vgpu_type_groups(struct intel_gvt *gvt)
+{
+	int i, j;
+	struct intel_vgpu_type *type;
+	struct attribute_group *group;
+
+	for (i = 0; i < gvt->num_types; i++) {
+		type = &gvt->types[i];
+
+		group = kzalloc(sizeof(struct attribute_group), GFP_KERNEL);
+		if (WARN_ON(!group))
+			goto unwind;
+
+		group->name = type->name;
+		group->attrs = type_attrs;
+		intel_vgpu_type_groups[i] = group;
+	}
+
+	return true;
+
+unwind:
+	for (j = 0; j < i; j++) {
+		group = intel_vgpu_type_groups[j];
+		kfree(group);
+	}
+
+	return false;
+}
+
+static void intel_gvt_cleanup_vgpu_type_groups(struct intel_gvt *gvt)
+{
+	int i;
+	struct attribute_group *group;
+
+	for (i = 0; i < gvt->num_types; i++) {
+		group = intel_vgpu_type_groups[i];
+		kfree(group);
+	}
+}
+
+static void kvmgt_protect_table_init(struct kvmgt_guest_info *info)
+{
+	hash_init(info->ptable);
+}
+
+static void kvmgt_protect_table_destroy(struct kvmgt_guest_info *info)
+{
+	struct kvmgt_pgfn *p;
+	struct hlist_node *tmp;
+	int i;
+
+	hash_for_each_safe(info->ptable, i, tmp, p, hnode) {
+		hash_del(&p->hnode);
+		kfree(p);
+	}
+}
+
+static struct kvmgt_pgfn *
+__kvmgt_protect_table_find(struct kvmgt_guest_info *info, gfn_t gfn)
+{
+	struct kvmgt_pgfn *p, *res = NULL;
+
+	hash_for_each_possible(info->ptable, p, hnode, gfn) {
+		if (gfn == p->gfn) {
+			res = p;
+			break;
+		}
+	}
+
+	return res;
+}
+
+static bool kvmgt_gfn_is_write_protected(struct kvmgt_guest_info *info,
+				gfn_t gfn)
+{
+	struct kvmgt_pgfn *p;
+
+	p = __kvmgt_protect_table_find(info, gfn);
+	return !!p;
+}
+
+static void kvmgt_protect_table_add(struct kvmgt_guest_info *info, gfn_t gfn)
+{
+	struct kvmgt_pgfn *p;
+
+	if (kvmgt_gfn_is_write_protected(info, gfn))
+		return;
+
+	p = kmalloc(sizeof(struct kvmgt_pgfn), GFP_ATOMIC);
+	if (WARN(!p, "gfn: 0x%llx\n", gfn))
+		return;
+
+	p->gfn = gfn;
+	hash_add(info->ptable, &p->hnode, gfn);
+}
+
+static void kvmgt_protect_table_del(struct kvmgt_guest_info *info,
+				gfn_t gfn)
+{
+	struct kvmgt_pgfn *p;
+
+	p = __kvmgt_protect_table_find(info, gfn);
+	if (p) {
+		hash_del(&p->hnode);
+		kfree(p);
+	}
+}
+
+static int kvmgt_host_init(struct device *dev, void *gvt, const void *ops)
+{
+	if (!intel_gvt_init_vgpu_type_groups(gvt))
+		return -EFAULT;
+
+	intel_gvt_ops = ops;
+
+	/* MDEV is not yet available */
+	return -ENODEV;
+}
+
+static void kvmgt_host_exit(struct device *dev, void *gvt)
+{
+	intel_gvt_cleanup_vgpu_type_groups(gvt);
+}
+
+static int kvmgt_write_protect_add(unsigned long handle, u64 gfn)
+{
+	struct kvmgt_guest_info *info = (struct kvmgt_guest_info *)handle;
+	struct kvm *kvm = info->kvm;
+	struct kvm_memory_slot *slot;
+	int idx;
+
+	idx = srcu_read_lock(&kvm->srcu);
+	slot = gfn_to_memslot(kvm, gfn);
+
+	spin_lock(&kvm->mmu_lock);
+
+	if (kvmgt_gfn_is_write_protected(info, gfn))
+		goto out;
+
+	kvm_slot_page_track_add_page(kvm, slot, gfn, KVM_PAGE_TRACK_WRITE);
+	kvmgt_protect_table_add(info, gfn);
+
+out:
+	spin_unlock(&kvm->mmu_lock);
+	srcu_read_unlock(&kvm->srcu, idx);
+	return 0;
+}
+
+static int kvmgt_write_protect_remove(unsigned long handle, u64 gfn)
+{
+	struct kvmgt_guest_info *info = (struct kvmgt_guest_info *)handle;
+	struct kvm *kvm = info->kvm;
+	struct kvm_memory_slot *slot;
+	int idx;
+
+	idx = srcu_read_lock(&kvm->srcu);
+	slot = gfn_to_memslot(kvm, gfn);
+
+	spin_lock(&kvm->mmu_lock);
+
+	if (!kvmgt_gfn_is_write_protected(info, gfn))
+		goto out;
+
+	kvm_slot_page_track_remove_page(kvm, slot, gfn, KVM_PAGE_TRACK_WRITE);
+	kvmgt_protect_table_del(info, gfn);
+
+out:
+	spin_unlock(&kvm->mmu_lock);
+	srcu_read_unlock(&kvm->srcu, idx);
+	return 0;
+}
+
+static void kvmgt_page_track_write(struct kvm_vcpu *vcpu, gpa_t gpa,
+		const u8 *val, int len,
+		struct kvm_page_track_notifier_node *node)
+{
+	struct kvmgt_guest_info *info = container_of(node,
+					struct kvmgt_guest_info, track_node);
+
+	if (kvmgt_gfn_is_write_protected(info, gpa_to_gfn(gpa)))
+		intel_gvt_ops->emulate_mmio_write(info->vgpu, gpa,
+					(void *)val, len);
+}
+
+static void kvmgt_page_track_flush_slot(struct kvm *kvm,
+		struct kvm_memory_slot *slot,
+		struct kvm_page_track_notifier_node *node)
+{
+	int i;
+	gfn_t gfn;
+	struct kvmgt_guest_info *info = container_of(node,
+					struct kvmgt_guest_info, track_node);
+
+	spin_lock(&kvm->mmu_lock);
+	for (i = 0; i < slot->npages; i++) {
+		gfn = slot->base_gfn + i;
+		if (kvmgt_gfn_is_write_protected(info, gfn)) {
+			kvm_slot_page_track_remove_page(kvm, slot, gfn,
+						KVM_PAGE_TRACK_WRITE);
+			kvmgt_protect_table_del(info, gfn);
+		}
+	}
+	spin_unlock(&kvm->mmu_lock);
+}
+
+static bool kvmgt_check_guest(void)
+{
+	unsigned int eax, ebx, ecx, edx;
+	char s[12];
+	unsigned int *i;
+
+	eax = KVM_CPUID_SIGNATURE;
+	ebx = ecx = edx = 0;
+
+	asm volatile ("cpuid"
+		      : "+a"(eax), "=b"(ebx), "=c"(ecx), "=d"(edx)
+		      :
+		      : "cc", "memory");
+	i = (unsigned int *)s;
+	i[0] = ebx;
+	i[1] = ecx;
+	i[2] = edx;
+
+	return !strncmp(s, "KVMKVMKVM", strlen("KVMKVMKVM"));
+}
+
+/**
+ * NOTE:
+ * It's actually impossible to check if we are running in KVM host,
+ * since the "KVM host" is simply native. So we only dectect guest here.
+ */
+static int kvmgt_detect_host(void)
+{
+#ifdef CONFIG_INTEL_IOMMU
+	if (intel_iommu_gfx_mapped) {
+		gvt_err("Hardware IOMMU compatibility not yet supported, try to boot with intel_iommu=igfx_off\n");
+		return -ENODEV;
+	}
+#endif
+	return kvmgt_check_guest() ? -ENODEV : 0;
+}
+
+static int kvmgt_attach_vgpu(void *vgpu, unsigned long *handle)
+{
+	/* nothing to do here */
+	return 0;
+}
+
+static void kvmgt_detach_vgpu(unsigned long handle)
+{
+	/* nothing to do here */
+}
+
+static int kvmgt_inject_msi(unsigned long handle, u32 addr, u16 data)
+{
+	struct kvmgt_guest_info *info = (struct kvmgt_guest_info *)handle;
+	struct intel_vgpu *vgpu = info->vgpu;
+
+	if (vgpu->vdev.msi_trigger)
+		return eventfd_signal(vgpu->vdev.msi_trigger, 1) == 1;
+
+	return false;
+}
+
+static unsigned long kvmgt_gfn_to_pfn(unsigned long handle, unsigned long gfn)
+{
+	unsigned long pfn;
+	struct kvmgt_guest_info *info = (struct kvmgt_guest_info *)handle;
+	int rc;
+
+	pfn = gvt_cache_find(info->vgpu, gfn);
+	if (pfn != 0)
+		return pfn;
+
+	rc = vfio_pin_pages(info->vgpu->vdev.mdev, &gfn, 1,
+				IOMMU_READ | IOMMU_WRITE, &pfn);
+	if (rc != 1) {
+		gvt_err("vfio_pin_pages failed for gfn: 0x%lx\n", gfn);
+		return 0;
+	}
+
+	gvt_cache_add(info->vgpu, gfn, pfn);
+	return pfn;
+}
+
+static void *kvmgt_gpa_to_hva(unsigned long handle, unsigned long gpa)
+{
+	unsigned long pfn;
+	gfn_t gfn = gpa_to_gfn(gpa);
+
+	pfn = kvmgt_gfn_to_pfn(handle, gfn);
+	if (!pfn)
+		return NULL;
+
+	return (char *)pfn_to_kaddr(pfn) + offset_in_page(gpa);
+}
+
+static int kvmgt_rw_gpa(unsigned long handle, unsigned long gpa,
+			void *buf, unsigned long len, bool write)
+{
+	void *hva = NULL;
+
+	hva = kvmgt_gpa_to_hva(handle, gpa);
+	if (!hva)
+		return -EFAULT;
+
+	if (write)
+		memcpy(hva, buf, len);
+	else
+		memcpy(buf, hva, len);
+
+	return 0;
+}
+
+static int kvmgt_read_gpa(unsigned long handle, unsigned long gpa,
+			void *buf, unsigned long len)
+{
+	return kvmgt_rw_gpa(handle, gpa, buf, len, false);
+}
+
+static int kvmgt_write_gpa(unsigned long handle, unsigned long gpa,
+			void *buf, unsigned long len)
+{
+	return kvmgt_rw_gpa(handle, gpa, buf, len, true);
+}
+
+static unsigned long kvmgt_virt_to_pfn(void *addr)
+{
+	return PFN_DOWN(__pa(addr));
+}
+
+struct intel_gvt_mpt kvmgt_mpt = {
+	.detect_host = kvmgt_detect_host,
+	.host_init = kvmgt_host_init,
+	.host_exit = kvmgt_host_exit,
+	.attach_vgpu = kvmgt_attach_vgpu,
+	.detach_vgpu = kvmgt_detach_vgpu,
+	.inject_msi = kvmgt_inject_msi,
+	.from_virt_to_mfn = kvmgt_virt_to_pfn,
+	.set_wp_page = kvmgt_write_protect_add,
+	.unset_wp_page = kvmgt_write_protect_remove,
+	.read_gpa = kvmgt_read_gpa,
+	.write_gpa = kvmgt_write_gpa,
+	.gfn_to_mfn = kvmgt_gfn_to_pfn,
+};
+EXPORT_SYMBOL_GPL(kvmgt_mpt);
+
+static int __init kvmgt_init(void)
+{
+	return 0;
+}
+
+static void __exit kvmgt_exit(void)
+{
+}
+
+module_init(kvmgt_init);
+module_exit(kvmgt_exit);
+
+MODULE_LICENSE("GPL and additional rights");
+MODULE_AUTHOR("Intel Corporation");