Merge tag 'gvt-next-2022-04-21-for-christoph' of...

Merge tag 'gvt-next-2022-04-21-for-christoph' of https://github.com/intel/gvt-linux into drm-intel-next

gvt-next-2022-04-21-for-christoph

- Separating the MMIO table from GVT-g. (Zhi)
- GVT-g re-factor. (Christoph)
- GVT-g mdev API cleanup. (Jason)
- GVT-g trace/makefile cleanup. (Jani)

[Jani: added #include to adapt to header refactoring in drm-intel-next]
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
From: "Wang, Zhi A" <zhi.a.wang@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/25a713cd-0b7d-4c09-7d91-4f4ef6c9eb11@intel.com

Documentation/driver-api/vfio-mediated-device.rst

@@ -105,6 +105,7 @@ structure to represent a mediated device's driver::
      struct mdev_driver {
 	     int  (*probe)  (struct mdev_device *dev);
 	     void (*remove) (struct mdev_device *dev);
+	     struct attribute_group **supported_type_groups;
 	     struct device_driver    driver;
      };
 
@@ -119,33 +120,15 @@ to register and unregister itself with the core driver:
 
     extern void mdev_unregister_driver(struct mdev_driver *drv);
 
-The mediated bus driver is responsible for adding mediated devices to the VFIO
-group when devices are bound to the driver and removing mediated devices from
-the VFIO when devices are unbound from the driver.
-
-
-Physical Device Driver Interface
---------------------------------
-
-The physical device driver interface provides the mdev_parent_ops[3] structure
-to define the APIs to manage work in the mediated core driver that is related
-to the physical device.
-
-The structures in the mdev_parent_ops structure are as follows:
-
-* dev_attr_groups: attributes of the parent device
-* mdev_attr_groups: attributes of the mediated device
-* supported_config: attributes to define supported configurations
-* device_driver: device driver to bind for mediated device instances
-
-The mdev_parent_ops also still has various functions pointers.  Theses exist
-for historical reasons only and shall not be used for new drivers.
+The mediated bus driver's probe function should create a vfio_device on top of
+the mdev_device and connect it to an appropriate implementation of
+vfio_device_ops.
 
 When a driver wants to add the GUID creation sysfs to an existing device it has
 probe'd to then it should call::
 
 	extern int  mdev_register_device(struct device *dev,
-	                                 const struct mdev_parent_ops *ops);
+	                                 struct mdev_driver *mdev_driver);
 
 This will provide the 'mdev_supported_types/XX/create' files which can then be
 used to trigger the creation of a mdev_device. The created mdev_device will be

drivers/gpu/drm/i915/Kconfig

@@ -102,40 +102,30 @@ config DRM_I915_USERPTR
 	  If in doubt, say "Y".
 
 config DRM_I915_GVT
-	bool "Enable Intel GVT-g graphics virtualization host support"
+	bool
+
+config DRM_I915_GVT_KVMGT
+	tristate "Enable KVM host support Intel GVT-g graphics virtualization"
 	depends on DRM_I915
 	depends on X86
 	depends on 64BIT
-	default n
+	depends on KVM
+	depends on VFIO_MDEV
+	select DRM_I915_GVT
+	select KVM_EXTERNAL_WRITE_TRACKING
+
 	help
 	  Choose this option if you want to enable Intel GVT-g graphics
 	  virtualization technology host support with integrated graphics.
 	  With GVT-g, it's possible to have one integrated graphics
-	  device shared by multiple VMs under different hypervisors.
-
-	  Note that at least one hypervisor like Xen or KVM is required for
-	  this driver to work, and it only supports newer device from
-	  Broadwell+. For further information and setup guide, you can
-	  visit: http://01.org/igvt-g.
+	  device shared by multiple VMs under KVM.
 
-	  Now it's just a stub to support the modifications of i915 for
-	  GVT device model. It requires at least one MPT modules for Xen/KVM
-	  and other components of GVT device model to work. Use it under
-	  you own risk.
+	  Note that this driver only supports newer device from Broadwell on.
+	  For further information and setup guide, you can visit:
+	  http://01.org/igvt-g.
 
 	  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
-	depends on VFIO_MDEV
-	select KVM_EXTERNAL_WRITE_TRACKING
-	default n
-	help
-	  Choose this option if you want to enable KVMGT support for
-	  Intel GVT-g.
-
 config DRM_I915_PXP
 	bool "Enable Intel PXP support"
 	depends on DRM_I915

drivers/gpu/drm/i915/Makefile

@@ -322,13 +322,13 @@ i915-$(CONFIG_DRM_I915_SELFTEST) += \
 # virtual gpu code
 i915-y += i915_vgpu.o
 
-ifeq ($(CONFIG_DRM_I915_GVT),y)
-i915-y += intel_gvt.o
+i915-$(CONFIG_DRM_I915_GVT) += \
+	intel_gvt.o \
+	intel_gvt_mmio_table.o
 include $(src)/gvt/Makefile
-endif
 
 obj-$(CONFIG_DRM_I915) += i915.o
-obj-$(CONFIG_DRM_I915_GVT_KVMGT) += gvt/kvmgt.o
+obj-$(CONFIG_DRM_I915_GVT_KVMGT) += kvmgt.o
 
 # header test
 

drivers/gpu/drm/i915/gvt/Makefile

 # SPDX-License-Identifier: GPL-2.0
-GVT_DIR := gvt
-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 mmio_context.o cmd_parser.o debugfs.o \
-	fb_decoder.o dmabuf.o page_track.o
 
-ccflags-y				+= -I $(srctree)/$(src) -I $(srctree)/$(src)/$(GVT_DIR)/
-i915-y					+= $(addprefix $(GVT_DIR)/, $(GVT_SOURCE))
+kvmgt-$(CONFIG_DRM_I915_GVT) += \
+	gvt/aperture_gm.o \
+	gvt/cfg_space.o \
+	gvt/cmd_parser.o \
+	gvt/debugfs.o \
+	gvt/display.o \
+	gvt/dmabuf.o \
+	gvt/edid.o \
+	gvt/execlist.o \
+	gvt/fb_decoder.o \
+	gvt/firmware.o \
+	gvt/gtt.o \
+	gvt/handlers.o \
+	gvt/interrupt.o \
+	gvt/kvmgt.o \
+	gvt/mmio.o \
+	gvt/mmio_context.o \
+	gvt/opregion.o \
+	gvt/page_track.o \
+	gvt/sched_policy.o \
+	gvt/scheduler.o \
+	gvt/trace_points.o \
+	gvt/vgpu.o

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

@@ -129,60 +129,16 @@ int intel_vgpu_emulate_cfg_read(struct intel_vgpu *vgpu, unsigned int offset,
 	return 0;
 }
 
-static int map_aperture(struct intel_vgpu *vgpu, bool map)
+static void map_aperture(struct intel_vgpu *vgpu, bool map)
 {
-	phys_addr_t aperture_pa = vgpu_aperture_pa_base(vgpu);
-	unsigned long aperture_sz = vgpu_aperture_sz(vgpu);
-	u64 first_gfn;
-	u64 val;
-	int ret;
-
-	if (map == vgpu->cfg_space.bar[INTEL_GVT_PCI_BAR_APERTURE].tracked)
-		return 0;
-
-	val = vgpu_cfg_space(vgpu)[PCI_BASE_ADDRESS_2];
-	if (val & PCI_BASE_ADDRESS_MEM_TYPE_64)
-		val = *(u64 *)(vgpu_cfg_space(vgpu) + PCI_BASE_ADDRESS_2);
-	else
-		val = *(u32 *)(vgpu_cfg_space(vgpu) + PCI_BASE_ADDRESS_2);
-
-	first_gfn = (val + vgpu_aperture_offset(vgpu)) >> PAGE_SHIFT;
-
-	ret = intel_gvt_hypervisor_map_gfn_to_mfn(vgpu, first_gfn,
-						  aperture_pa >> PAGE_SHIFT,
-						  aperture_sz >> PAGE_SHIFT,
-						  map);
-	if (ret)
-		return ret;
-
-	vgpu->cfg_space.bar[INTEL_GVT_PCI_BAR_APERTURE].tracked = map;
-	return 0;
+	if (map != vgpu->cfg_space.bar[INTEL_GVT_PCI_BAR_APERTURE].tracked)
+		vgpu->cfg_space.bar[INTEL_GVT_PCI_BAR_APERTURE].tracked = map;
 }
 
-static int trap_gttmmio(struct intel_vgpu *vgpu, bool trap)
+static void trap_gttmmio(struct intel_vgpu *vgpu, bool trap)
 {
-	u64 start, end;
-	u64 val;
-	int ret;
-
-	if (trap == vgpu->cfg_space.bar[INTEL_GVT_PCI_BAR_GTTMMIO].tracked)
-		return 0;
-
-	val = vgpu_cfg_space(vgpu)[PCI_BASE_ADDRESS_0];
-	if (val & PCI_BASE_ADDRESS_MEM_TYPE_64)
-		start = *(u64 *)(vgpu_cfg_space(vgpu) + PCI_BASE_ADDRESS_0);
-	else
-		start = *(u32 *)(vgpu_cfg_space(vgpu) + PCI_BASE_ADDRESS_0);
-
-	start &= ~GENMASK(3, 0);
-	end = start + vgpu->cfg_space.bar[INTEL_GVT_PCI_BAR_GTTMMIO].size - 1;
-
-	ret = intel_gvt_hypervisor_set_trap_area(vgpu, start, end, trap);
-	if (ret)
-		return ret;
-
-	vgpu->cfg_space.bar[INTEL_GVT_PCI_BAR_GTTMMIO].tracked = trap;
-	return 0;
+	if (trap != vgpu->cfg_space.bar[INTEL_GVT_PCI_BAR_GTTMMIO].tracked)
+		vgpu->cfg_space.bar[INTEL_GVT_PCI_BAR_GTTMMIO].tracked = trap;
 }
 
 static int emulate_pci_command_write(struct intel_vgpu *vgpu,
@@ -191,26 +147,17 @@ static int emulate_pci_command_write(struct intel_vgpu *vgpu,
 	u8 old = vgpu_cfg_space(vgpu)[offset];
 	u8 new = *(u8 *)p_data;
 	u8 changed = old ^ new;
-	int ret;
 
 	vgpu_pci_cfg_mem_write(vgpu, offset, p_data, bytes);
 	if (!(changed & PCI_COMMAND_MEMORY))
 		return 0;
 
 	if (old & PCI_COMMAND_MEMORY) {
-		ret = trap_gttmmio(vgpu, false);
-		if (ret)
-			return ret;
-		ret = map_aperture(vgpu, false);
-		if (ret)
-			return ret;
+		trap_gttmmio(vgpu, false);
+		map_aperture(vgpu, false);
 	} else {
-		ret = trap_gttmmio(vgpu, true);
-		if (ret)
-			return ret;
-		ret = map_aperture(vgpu, true);
-		if (ret)
-			return ret;
+		trap_gttmmio(vgpu, true);
+		map_aperture(vgpu, true);
 	}
 
 	return 0;
@@ -230,13 +177,12 @@ static int emulate_pci_rom_bar_write(struct intel_vgpu *vgpu,
 	return 0;
 }
 
-static int emulate_pci_bar_write(struct intel_vgpu *vgpu, unsigned int offset,
+static void emulate_pci_bar_write(struct intel_vgpu *vgpu, unsigned int offset,
 	void *p_data, unsigned int bytes)
 {
 	u32 new = *(u32 *)(p_data);
 	bool lo = IS_ALIGNED(offset, 8);
 	u64 size;
-	int ret = 0;
 	bool mmio_enabled =
 		vgpu_cfg_space(vgpu)[PCI_COMMAND] & PCI_COMMAND_MEMORY;
 	struct intel_vgpu_pci_bar *bars = vgpu->cfg_space.bar;
@@ -259,14 +205,14 @@ static int emulate_pci_bar_write(struct intel_vgpu *vgpu, unsigned int offset,
 			 * Untrap the BAR, since guest hasn't configured a
 			 * valid GPA
 			 */
-			ret = trap_gttmmio(vgpu, false);
+			trap_gttmmio(vgpu, false);
 			break;
 		case PCI_BASE_ADDRESS_2:
 		case PCI_BASE_ADDRESS_3:
 			size = ~(bars[INTEL_GVT_PCI_BAR_APERTURE].size -1);
 			intel_vgpu_write_pci_bar(vgpu, offset,
 						size >> (lo ? 0 : 32), lo);
-			ret = map_aperture(vgpu, false);
+			map_aperture(vgpu, false);
 			break;
 		default:
 			/* Unimplemented BARs */
@@ -282,19 +228,18 @@ static int emulate_pci_bar_write(struct intel_vgpu *vgpu, unsigned int offset,
 			 */
 			trap_gttmmio(vgpu, false);
 			intel_vgpu_write_pci_bar(vgpu, offset, new, lo);
-			ret = trap_gttmmio(vgpu, mmio_enabled);
+			trap_gttmmio(vgpu, mmio_enabled);
 			break;
 		case PCI_BASE_ADDRESS_2:
 		case PCI_BASE_ADDRESS_3:
 			map_aperture(vgpu, false);
 			intel_vgpu_write_pci_bar(vgpu, offset, new, lo);
-			ret = map_aperture(vgpu, mmio_enabled);
+			map_aperture(vgpu, mmio_enabled);
 			break;
 		default:
 			intel_vgpu_write_pci_bar(vgpu, offset, new, lo);
 		}
 	}
-	return ret;
 }
 
 /**
@@ -336,8 +281,8 @@ int intel_vgpu_emulate_cfg_write(struct intel_vgpu *vgpu, unsigned int offset,
 	case PCI_BASE_ADDRESS_0 ... PCI_BASE_ADDRESS_5:
 		if (drm_WARN_ON(&i915->drm, !IS_ALIGNED(offset, 4)))
 			return -EINVAL;
-		return emulate_pci_bar_write(vgpu, offset, p_data, bytes);
-
+		emulate_pci_bar_write(vgpu, offset, p_data, bytes);
+		break;
 	case INTEL_GVT_PCI_SWSCI:
 		if (drm_WARN_ON(&i915->drm, !IS_ALIGNED(offset, 4)))
 			return -EINVAL;

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

@@ -1011,7 +1011,7 @@ static int cmd_reg_handler(struct parser_exec_state *s,
 	if (GRAPHICS_VER(s->engine->i915) == 9 &&
 	    intel_gvt_mmio_is_sr_in_ctx(gvt, offset) &&
 	    !strncmp(cmd, "lri", 3)) {
-		intel_gvt_hypervisor_read_gpa(s->vgpu,
+		intel_gvt_read_gpa(s->vgpu,
 			s->workload->ring_context_gpa + 12, &ctx_sr_ctl, 4);
 		/* check inhibit context */
 		if (ctx_sr_ctl & 1) {
@@ -1775,7 +1775,7 @@ static int copy_gma_to_hva(struct intel_vgpu *vgpu, struct intel_vgpu_mm *mm,
 		copy_len = (end_gma - gma) >= (I915_GTT_PAGE_SIZE - offset) ?
 			I915_GTT_PAGE_SIZE - offset : end_gma - gma;
 
-		intel_gvt_hypervisor_read_gpa(vgpu, gpa, va + len, copy_len);
+		intel_gvt_read_gpa(vgpu, gpa, va + len, copy_len);
 
 		len += copy_len;
 		gma += copy_len;

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

@@ -29,7 +29,7 @@
  */
 
 #include <linux/dma-buf.h>
-#include <linux/vfio.h>
+#include <linux/mdev.h>
 
 #include <drm/drm_fourcc.h>
 #include <drm/drm_plane.h>
@@ -42,24 +42,6 @@
 
 #define GEN8_DECODE_PTE(pte) (pte & GENMASK_ULL(63, 12))
 
-static int vgpu_pin_dma_address(struct intel_vgpu *vgpu,
-				unsigned long size,
-				dma_addr_t dma_addr)
-{
-	int ret = 0;
-
-	if (intel_gvt_hypervisor_dma_pin_guest_page(vgpu, dma_addr))
-		ret = -EINVAL;
-
-	return ret;
-}
-
-static void vgpu_unpin_dma_address(struct intel_vgpu *vgpu,
-				   dma_addr_t dma_addr)
-{
-	intel_gvt_hypervisor_dma_unmap_guest_page(vgpu, dma_addr);
-}
-
 static int vgpu_gem_get_pages(
 		struct drm_i915_gem_object *obj)
 {
@@ -95,7 +77,7 @@ static int vgpu_gem_get_pages(
 	for_each_sg(st->sgl, sg, page_num, i) {
 		dma_addr_t dma_addr =
 			GEN8_DECODE_PTE(readq(&gtt_entries[i]));
-		if (vgpu_pin_dma_address(vgpu, PAGE_SIZE, dma_addr)) {
+		if (intel_gvt_dma_pin_guest_page(vgpu, dma_addr)) {
 			ret = -EINVAL;
 			goto out;
 		}
@@ -114,7 +96,7 @@ static int vgpu_gem_get_pages(
 		for_each_sg(st->sgl, sg, i, j) {
 			dma_addr = sg_dma_address(sg);
 			if (dma_addr)
-				vgpu_unpin_dma_address(vgpu, dma_addr);
+				intel_gvt_dma_unmap_guest_page(vgpu, dma_addr);
 		}
 		sg_free_table(st);
 		kfree(st);
@@ -136,7 +118,7 @@ static void vgpu_gem_put_pages(struct drm_i915_gem_object *obj,
 		int i;
 
 		for_each_sg(pages->sgl, sg, fb_info->size, i)
-			vgpu_unpin_dma_address(vgpu,
+			intel_gvt_dma_unmap_guest_page(vgpu,
 					       sg_dma_address(sg));
 	}
 
@@ -157,7 +139,6 @@ static void dmabuf_gem_object_free(struct kref *kref)
 			dmabuf_obj = list_entry(pos, struct intel_vgpu_dmabuf_obj, list);
 			if (dmabuf_obj == obj) {
 				list_del(pos);
-				intel_gvt_hypervisor_put_vfio_device(vgpu);
 				idr_remove(&vgpu->object_idr,
 					   dmabuf_obj->dmabuf_id);
 				kfree(dmabuf_obj->info);
@@ -491,14 +472,6 @@ int intel_vgpu_query_plane(struct intel_vgpu *vgpu, void *args)
 
 	kref_init(&dmabuf_obj->kref);
 
-	mutex_lock(&vgpu->dmabuf_lock);
-	if (intel_gvt_hypervisor_get_vfio_device(vgpu)) {
-		gvt_vgpu_err("get vfio device failed\n");
-		mutex_unlock(&vgpu->dmabuf_lock);
-		goto out_free_info;
-	}
-	mutex_unlock(&vgpu->dmabuf_lock);
-
 	update_fb_info(gfx_plane_info, &fb_info);
 
 	INIT_LIST_HEAD(&dmabuf_obj->list);
@@ -603,7 +576,6 @@ void intel_vgpu_dmabuf_cleanup(struct intel_vgpu *vgpu)
 		dmabuf_obj->vgpu = NULL;
 
 		idr_remove(&vgpu->object_idr, dmabuf_obj->dmabuf_id);
-		intel_gvt_hypervisor_put_vfio_device(vgpu);
 		list_del(pos);
 
 		/* dmabuf_obj might be freed in dmabuf_obj_put */

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

@@ -159,12 +159,12 @@ static void emulate_csb_update(struct intel_vgpu_execlist *execlist,
 	hwsp_gpa = intel_vgpu_gma_to_gpa(vgpu->gtt.ggtt_mm,
 					 vgpu->hws_pga[execlist->engine->id]);
 	if (hwsp_gpa != INTEL_GVT_INVALID_ADDR) {
-		intel_gvt_hypervisor_write_gpa(vgpu,
-					       hwsp_gpa + I915_HWS_CSB_BUF0_INDEX * 4 + write_pointer * 8,
-					       status, 8);
-		intel_gvt_hypervisor_write_gpa(vgpu,
-					       hwsp_gpa + INTEL_HWS_CSB_WRITE_INDEX(execlist->engine->i915) * 4,
-					       &write_pointer, 4);
+		intel_gvt_write_gpa(vgpu,
+			hwsp_gpa + I915_HWS_CSB_BUF0_INDEX * 4 + write_pointer * 8,
+			status, 8);
+		intel_gvt_write_gpa(vgpu,
+			hwsp_gpa + INTEL_HWS_CSB_WRITE_INDEX(execlist->engine->i915) * 4,
+			&write_pointer, 4);
 	}
 
 	gvt_dbg_el("vgpu%d: w pointer %u reg %x csb l %x csb h %x\n",

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

@@ -66,22 +66,16 @@ static struct bin_attribute firmware_attr = {
 	.mmap = NULL,
 };
 
-static int mmio_snapshot_handler(struct intel_gvt *gvt, u32 offset, void *data)
-{
-	*(u32 *)(data + offset) = intel_uncore_read_notrace(gvt->gt->uncore,
-							    _MMIO(offset));
-	return 0;
-}
-
 static int expose_firmware_sysfs(struct intel_gvt *gvt)
 {
 	struct intel_gvt_device_info *info = &gvt->device_info;
-	struct pci_dev *pdev = to_pci_dev(gvt->gt->i915->drm.dev);
+	struct drm_i915_private *i915 = gvt->gt->i915;
+	struct pci_dev *pdev = to_pci_dev(i915->drm.dev);
 	struct gvt_firmware_header *h;
 	void *firmware;
 	void *p;
 	unsigned long size, crc32_start;
-	int i, ret;
+	int ret;
 
 	size = sizeof(*h) + info->mmio_size + info->cfg_space_size;
 	firmware = vzalloc(size);
@@ -99,17 +93,16 @@ static int expose_firmware_sysfs(struct intel_gvt *gvt)
 
 	p = firmware + h->cfg_space_offset;
 
-	for (i = 0; i < h->cfg_space_size; i += 4)
-		pci_read_config_dword(pdev, i, p + i);
-
-	memcpy(gvt->firmware.cfg_space, p, info->cfg_space_size);
+	memcpy(gvt->firmware.cfg_space, i915->vgpu.initial_cfg_space,
+	       info->cfg_space_size);
+	memcpy(p, gvt->firmware.cfg_space, info->cfg_space_size);
 
 	p = firmware + h->mmio_offset;
 
-	/* Take a snapshot of hw mmio registers. */
-	intel_gvt_for_each_tracked_mmio(gvt, mmio_snapshot_handler, p);
+	memcpy(gvt->firmware.mmio, i915->vgpu.initial_mmio,
+	       info->mmio_size);
 
-	memcpy(gvt->firmware.mmio, p, info->mmio_size);
+	memcpy(p, gvt->firmware.mmio, info->mmio_size);
 
 	crc32_start = offsetof(struct gvt_firmware_header, crc32) + 4;
 	h->crc32 = crc32_le(0, firmware + crc32_start, size - crc32_start);

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

@@ -49,6 +49,22 @@
 static bool enable_out_of_sync = false;
 static int preallocated_oos_pages = 8192;
 
+static bool intel_gvt_is_valid_gfn(struct intel_vgpu *vgpu, unsigned long gfn)
+{
+	struct kvm *kvm = vgpu->kvm;
+	int idx;
+	bool ret;
+
+	if (!vgpu->attached)
+		return false;
+
+	idx = srcu_read_lock(&kvm->srcu);
+	ret = kvm_is_visible_gfn(kvm, gfn);
+	srcu_read_unlock(&kvm->srcu, idx);
+
+	return ret;
+}
+
 /*
  * validate a gm address and related range size,
  * translate it to host gm address
@@ -314,7 +330,7 @@ static inline int gtt_get_entry64(void *pt,
 		return -EINVAL;
 
 	if (hypervisor_access) {
-		ret = intel_gvt_hypervisor_read_gpa(vgpu, gpa +
+		ret = intel_gvt_read_gpa(vgpu, gpa +
 				(index << info->gtt_entry_size_shift),
 				&e->val64, 8);
 		if (WARN_ON(ret))
@@ -339,7 +355,7 @@ static inline int gtt_set_entry64(void *pt,
 		return -EINVAL;
 
 	if (hypervisor_access) {
-		ret = intel_gvt_hypervisor_write_gpa(vgpu, gpa +
+		ret = intel_gvt_write_gpa(vgpu, gpa +
 				(index << info->gtt_entry_size_shift),
 				&e->val64, 8);
 		if (WARN_ON(ret))
@@ -997,7 +1013,7 @@ static inline void ppgtt_invalidate_pte(struct intel_vgpu_ppgtt_spt *spt,
 	if (!pfn || pfn == vgpu->gtt.scratch_pt[type].page_mfn)
 		return;
 
-	intel_gvt_hypervisor_dma_unmap_guest_page(vgpu, pfn << PAGE_SHIFT);
+	intel_gvt_dma_unmap_guest_page(vgpu, pfn << PAGE_SHIFT);
 }
 
 static int ppgtt_invalidate_spt(struct intel_vgpu_ppgtt_spt *spt)
@@ -1162,15 +1178,16 @@ static int is_2MB_gtt_possible(struct intel_vgpu *vgpu,
 	struct intel_gvt_gtt_entry *entry)
 {
 	const struct intel_gvt_gtt_pte_ops *ops = vgpu->gvt->gtt.pte_ops;
-	unsigned long pfn;
+	kvm_pfn_t pfn;
 
 	if (!HAS_PAGE_SIZES(vgpu->gvt->gt->i915, I915_GTT_PAGE_SIZE_2M))
 		return 0;
 
-	pfn = intel_gvt_hypervisor_gfn_to_mfn(vgpu, ops->get_pfn(entry));
-	if (pfn == INTEL_GVT_INVALID_ADDR)
+	if (!vgpu->attached)
+		return -EINVAL;
+	pfn = gfn_to_pfn(vgpu->kvm, ops->get_pfn(entry));
+	if (is_error_noslot_pfn(pfn))
 		return -EINVAL;
-
 	return PageTransHuge(pfn_to_page(pfn));
 }
 
@@ -1195,8 +1212,8 @@ static int split_2MB_gtt_entry(struct intel_vgpu *vgpu,
 		return PTR_ERR(sub_spt);
 
 	for_each_shadow_entry(sub_spt, &sub_se, sub_index) {
-		ret = intel_gvt_hypervisor_dma_map_guest_page(vgpu,
-				start_gfn + sub_index, PAGE_SIZE, &dma_addr);
+		ret = intel_gvt_dma_map_guest_page(vgpu, start_gfn + sub_index,
+						   PAGE_SIZE, &dma_addr);
 		if (ret) {
 			ppgtt_invalidate_spt(spt);
 			return ret;
@@ -1241,8 +1258,8 @@ static int split_64KB_gtt_entry(struct intel_vgpu *vgpu,
 	ops->set_64k_splited(&entry);
 
 	for (i = 0; i < GTT_64K_PTE_STRIDE; i++) {
-		ret = intel_gvt_hypervisor_dma_map_guest_page(vgpu,
-					start_gfn + i, PAGE_SIZE, &dma_addr);
+		ret = intel_gvt_dma_map_guest_page(vgpu, start_gfn + i,
+						   PAGE_SIZE, &dma_addr);
 		if (ret)
 			return ret;
 
@@ -1296,8 +1313,7 @@ static int ppgtt_populate_shadow_entry(struct intel_vgpu *vgpu,
 	}
 
 	/* direct shadow */
-	ret = intel_gvt_hypervisor_dma_map_guest_page(vgpu, gfn, page_size,
-						      &dma_addr);
+	ret = intel_gvt_dma_map_guest_page(vgpu, gfn, page_size, &dma_addr);
 	if (ret)
 		return -ENXIO;
 
@@ -1331,7 +1347,7 @@ static int ppgtt_populate_spt(struct intel_vgpu_ppgtt_spt *spt)
 			ppgtt_set_shadow_entry(spt, &se, i);
 		} else {
 			gfn = ops->get_pfn(&ge);
-			if (!intel_gvt_hypervisor_is_valid_gfn(vgpu, gfn)) {
+			if (!intel_gvt_is_valid_gfn(vgpu, gfn)) {
 				ops->set_pfn(&se, gvt->gtt.scratch_mfn);
 				ppgtt_set_shadow_entry(spt, &se, i);
 				continue;
@@ -1497,7 +1513,7 @@ static int attach_oos_page(struct intel_vgpu_oos_page *oos_page,
 	struct intel_gvt *gvt = spt->vgpu->gvt;
 	int ret;
 
-	ret = intel_gvt_hypervisor_read_gpa(spt->vgpu,
+	ret = intel_gvt_read_gpa(spt->vgpu,
 			spt->guest_page.gfn << I915_GTT_PAGE_SHIFT,
 			oos_page->mem, I915_GTT_PAGE_SIZE);
 	if (ret)
@@ -2228,8 +2244,7 @@ static void ggtt_invalidate_pte(struct intel_vgpu *vgpu,
 
 	pfn = pte_ops->get_pfn(entry);
 	if (pfn != vgpu->gvt->gtt.scratch_mfn)
-		intel_gvt_hypervisor_dma_unmap_guest_page(vgpu,
-						pfn << PAGE_SHIFT);
+		intel_gvt_dma_unmap_guest_page(vgpu, pfn << PAGE_SHIFT);
 }
 
 static int emulate_ggtt_mmio_write(struct intel_vgpu *vgpu, unsigned int off,
@@ -2315,13 +2330,13 @@ static int emulate_ggtt_mmio_write(struct intel_vgpu *vgpu, unsigned int off,
 		/* one PTE update may be issued in multiple writes and the
 		 * first write may not construct a valid gfn
 		 */
-		if (!intel_gvt_hypervisor_is_valid_gfn(vgpu, gfn)) {
+		if (!intel_gvt_is_valid_gfn(vgpu, gfn)) {
 			ops->set_pfn(&m, gvt->gtt.scratch_mfn);
 			goto out;
 		}
 
-		ret = intel_gvt_hypervisor_dma_map_guest_page(vgpu, gfn,
-							PAGE_SIZE, &dma_addr);
+		ret = intel_gvt_dma_map_guest_page(vgpu, gfn, PAGE_SIZE,
+						   &dma_addr);
 		if (ret) {
 			gvt_vgpu_err("fail to populate guest ggtt entry\n");
 			/* guest driver may read/write the entry when partial

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

-/*
- * Copyright(c) 2011-2016 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>
- *    Eddie Dong <eddie.dong@intel.com>
- *
- * Contributors:
- *    Niu Bing <bing.niu@intel.com>
- *    Zhi Wang <zhi.a.wang@intel.com>
- *
- */
-
-#include <linux/types.h>
-#include <linux/kthread.h>
-
-#include "i915_drv.h"
-#include "intel_gvt.h"
-#include "gvt.h"
-#include <linux/vfio.h>
-#include <linux/mdev.h>
-
-struct intel_gvt_host intel_gvt_host;
-
-static const char * const supported_hypervisors[] = {
-	[INTEL_GVT_HYPERVISOR_XEN] = "XEN",
-	[INTEL_GVT_HYPERVISOR_KVM] = "KVM",
-};
-
-static const struct intel_gvt_ops intel_gvt_ops = {
-	.emulate_cfg_read = intel_vgpu_emulate_cfg_read,
-	.emulate_cfg_write = intel_vgpu_emulate_cfg_write,
-	.emulate_mmio_read = intel_vgpu_emulate_mmio_read,
-	.emulate_mmio_write = intel_vgpu_emulate_mmio_write,
-	.vgpu_create = intel_gvt_create_vgpu,
-	.vgpu_destroy = intel_gvt_destroy_vgpu,
-	.vgpu_release = intel_gvt_release_vgpu,
-	.vgpu_reset = intel_gvt_reset_vgpu,
-	.vgpu_activate = intel_gvt_activate_vgpu,
-	.vgpu_deactivate = intel_gvt_deactivate_vgpu,
-	.vgpu_query_plane = intel_vgpu_query_plane,
-	.vgpu_get_dmabuf = intel_vgpu_get_dmabuf,
-	.write_protect_handler = intel_vgpu_page_track_handler,
-	.emulate_hotplug = intel_vgpu_emulate_hotplug,
-};
-
-static void init_device_info(struct intel_gvt *gvt)
-{
-	struct intel_gvt_device_info *info = &gvt->device_info;
-	struct pci_dev *pdev = to_pci_dev(gvt->gt->i915->drm.dev);
-
-	info->max_support_vgpus = 8;
-	info->cfg_space_size = PCI_CFG_SPACE_EXP_SIZE;
-	info->mmio_size = 2 * 1024 * 1024;
-	info->mmio_bar = 0;
-	info->gtt_start_offset = 8 * 1024 * 1024;
-	info->gtt_entry_size = 8;
-	info->gtt_entry_size_shift = 3;
-	info->gmadr_bytes_in_cmd = 8;
-	info->max_surface_size = 36 * 1024 * 1024;
-	info->msi_cap_offset = pdev->msi_cap;
-}
-
-static void intel_gvt_test_and_emulate_vblank(struct intel_gvt *gvt)
-{
-	struct intel_vgpu *vgpu;
-	int id;
-
-	mutex_lock(&gvt->lock);
-	idr_for_each_entry((&(gvt)->vgpu_idr), (vgpu), (id)) {
-		if (test_and_clear_bit(INTEL_GVT_REQUEST_EMULATE_VBLANK + id,
-				       (void *)&gvt->service_request)) {
-			if (vgpu->active)
-				intel_vgpu_emulate_vblank(vgpu);
-		}
-	}
-	mutex_unlock(&gvt->lock);
-}
-
-static int gvt_service_thread(void *data)
-{
-	struct intel_gvt *gvt = (struct intel_gvt *)data;
-	int ret;
-
-	gvt_dbg_core("service thread start\n");
-
-	while (!kthread_should_stop()) {
-		ret = wait_event_interruptible(gvt->service_thread_wq,
-				kthread_should_stop() || gvt->service_request);
-
-		if (kthread_should_stop())
-			break;
-
-		if (WARN_ONCE(ret, "service thread is waken up by signal.\n"))
-			continue;
-
-		intel_gvt_test_and_emulate_vblank(gvt);
-
-		if (test_bit(INTEL_GVT_REQUEST_SCHED,
-				(void *)&gvt->service_request) ||
-			test_bit(INTEL_GVT_REQUEST_EVENT_SCHED,
-					(void *)&gvt->service_request)) {
-			intel_gvt_schedule(gvt);
-		}
-	}
-
-	return 0;
-}
-
-static void clean_service_thread(struct intel_gvt *gvt)
-{
-	kthread_stop(gvt->service_thread);
-}
-
-static int init_service_thread(struct intel_gvt *gvt)
-{
-	init_waitqueue_head(&gvt->service_thread_wq);
-
-	gvt->service_thread = kthread_run(gvt_service_thread,
-			gvt, "gvt_service_thread");
-	if (IS_ERR(gvt->service_thread)) {
-		gvt_err("fail to start service thread.\n");
-		return PTR_ERR(gvt->service_thread);
-	}
-	return 0;
-}
-
-/**
- * intel_gvt_clean_device - clean a GVT device
- * @i915: i915 private
- *
- * This function is called at the driver unloading stage, to free the
- * resources owned by a GVT device.
- *
- */
-void intel_gvt_clean_device(struct drm_i915_private *i915)
-{
-	struct intel_gvt *gvt = fetch_and_zero(&i915->gvt);
-
-	if (drm_WARN_ON(&i915->drm, !gvt))
-		return;
-
-	intel_gvt_destroy_idle_vgpu(gvt->idle_vgpu);
-	intel_gvt_clean_vgpu_types(gvt);
-
-	intel_gvt_debugfs_clean(gvt);
-	clean_service_thread(gvt);
-	intel_gvt_clean_cmd_parser(gvt);
-	intel_gvt_clean_sched_policy(gvt);
-	intel_gvt_clean_workload_scheduler(gvt);
-	intel_gvt_clean_gtt(gvt);
-	intel_gvt_free_firmware(gvt);
-	intel_gvt_clean_mmio_info(gvt);
-	idr_destroy(&gvt->vgpu_idr);
-
-	kfree(i915->gvt);
-}
-
-/**
- * intel_gvt_init_device - initialize a GVT device
- * @i915: drm i915 private data
- *
- * This function is called at the initialization stage, to initialize
- * necessary GVT components.
- *
- * Returns:
- * Zero on success, negative error code if failed.
- *
- */
-int intel_gvt_init_device(struct drm_i915_private *i915)
-{
-	struct intel_gvt *gvt;
-	struct intel_vgpu *vgpu;
-	int ret;
-
-	if (drm_WARN_ON(&i915->drm, i915->gvt))
-		return -EEXIST;
-
-	gvt = kzalloc(sizeof(struct intel_gvt), GFP_KERNEL);
-	if (!gvt)
-		return -ENOMEM;
-
-	gvt_dbg_core("init gvt device\n");
-
-	idr_init_base(&gvt->vgpu_idr, 1);
-	spin_lock_init(&gvt->scheduler.mmio_context_lock);
-	mutex_init(&gvt->lock);
-	mutex_init(&gvt->sched_lock);
-	gvt->gt = to_gt(i915);
-	i915->gvt = gvt;
-
-	init_device_info(gvt);
-
-	ret = intel_gvt_setup_mmio_info(gvt);
-	if (ret)
-		goto out_clean_idr;
-
-	intel_gvt_init_engine_mmio_context(gvt);
-
-	ret = intel_gvt_load_firmware(gvt);
-	if (ret)
-		goto out_clean_mmio_info;
-
-	ret = intel_gvt_init_irq(gvt);
-	if (ret)
-		goto out_free_firmware;
-
-	ret = intel_gvt_init_gtt(gvt);
-	if (ret)
-		goto out_free_firmware;
-
-	ret = intel_gvt_init_workload_scheduler(gvt);
-	if (ret)
-		goto out_clean_gtt;
-
-	ret = intel_gvt_init_sched_policy(gvt);
-	if (ret)
-		goto out_clean_workload_scheduler;
-
-	ret = intel_gvt_init_cmd_parser(gvt);
-	if (ret)
-		goto out_clean_sched_policy;
-
-	ret = init_service_thread(gvt);
-	if (ret)
-		goto out_clean_cmd_parser;
-
-	ret = intel_gvt_init_vgpu_types(gvt);
-	if (ret)
-		goto out_clean_thread;
-
-	vgpu = intel_gvt_create_idle_vgpu(gvt);
-	if (IS_ERR(vgpu)) {
-		ret = PTR_ERR(vgpu);
-		gvt_err("failed to create idle vgpu\n");
-		goto out_clean_types;
-	}
-	gvt->idle_vgpu = vgpu;
-
-	intel_gvt_debugfs_init(gvt);
-
-	gvt_dbg_core("gvt device initialization is done\n");
-	intel_gvt_host.dev = i915->drm.dev;
-	intel_gvt_host.initialized = true;
-	return 0;
-
-out_clean_types:
-	intel_gvt_clean_vgpu_types(gvt);
-out_clean_thread:
-	clean_service_thread(gvt);
-out_clean_cmd_parser:
-	intel_gvt_clean_cmd_parser(gvt);
-out_clean_sched_policy:
-	intel_gvt_clean_sched_policy(gvt);
-out_clean_workload_scheduler:
-	intel_gvt_clean_workload_scheduler(gvt);
-out_clean_gtt:
-	intel_gvt_clean_gtt(gvt);
-out_free_firmware:
-	intel_gvt_free_firmware(gvt);
-out_clean_mmio_info:
-	intel_gvt_clean_mmio_info(gvt);
-out_clean_idr:
-	idr_destroy(&gvt->vgpu_idr);
-	kfree(gvt);
-	i915->gvt = NULL;
-	return ret;
-}
-
-int
-intel_gvt_pm_resume(struct intel_gvt *gvt)
-{
-	intel_gvt_restore_fence(gvt);
-	intel_gvt_restore_mmio(gvt);
-	intel_gvt_restore_ggtt(gvt);
-	return 0;
-}
-
-int
-intel_gvt_register_hypervisor(const struct intel_gvt_mpt *m)
-{
-	int ret;
-	void *gvt;
-
-	if (!intel_gvt_host.initialized)
-		return -ENODEV;
-
-	if (m->type != INTEL_GVT_HYPERVISOR_KVM &&
-	    m->type != INTEL_GVT_HYPERVISOR_XEN)
-		return -EINVAL;
-
-	/* Get a reference for device model module */
-	if (!try_module_get(THIS_MODULE))
-		return -ENODEV;
-
-	intel_gvt_host.mpt = m;
-	intel_gvt_host.hypervisor_type = m->type;
-	gvt = (void *)kdev_to_i915(intel_gvt_host.dev)->gvt;
-
-	ret = intel_gvt_hypervisor_host_init(intel_gvt_host.dev, gvt,
-					     &intel_gvt_ops);
-	if (ret < 0) {
-		gvt_err("Failed to init %s hypervisor module\n",
-			supported_hypervisors[intel_gvt_host.hypervisor_type]);
-		module_put(THIS_MODULE);
-		return -ENODEV;
-	}
-	gvt_dbg_core("Running with hypervisor %s in host mode\n",
-		     supported_hypervisors[intel_gvt_host.hypervisor_type]);
-	return 0;
-}
-EXPORT_SYMBOL_GPL(intel_gvt_register_hypervisor);
-
-void
-intel_gvt_unregister_hypervisor(void)
-{
-	void *gvt = (void *)kdev_to_i915(intel_gvt_host.dev)->gvt;
-	intel_gvt_hypervisor_host_exit(intel_gvt_host.dev, gvt);
-	module_put(THIS_MODULE);
-}
-EXPORT_SYMBOL_GPL(intel_gvt_unregister_hypervisor);

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

@@ -34,11 +34,13 @@
 #define _GVT_H_
 
 #include <uapi/linux/pci_regs.h>
+#include <linux/kvm_host.h>
+#include <linux/vfio.h>
 
 #include "i915_drv.h"
+#include "intel_gvt.h"
 
 #include "debug.h"
-#include "hypercall.h"
 #include "mmio.h"
 #include "reg.h"
 #include "interrupt.h"
@@ -56,15 +58,6 @@
 
 #define GVT_MAX_VGPU 8
 
-struct intel_gvt_host {
-	struct device *dev;
-	bool initialized;
-	int hypervisor_type;
-	const struct intel_gvt_mpt *mpt;
-};
-
-extern struct intel_gvt_host intel_gvt_host;
-
 /* Describe per-platform limitations. */
 struct intel_gvt_device_info {
 	u32 max_support_vgpus;
@@ -176,12 +169,14 @@ struct intel_vgpu_submission {
 	} last_ctx[I915_NUM_ENGINES];
 };
 
+#define KVMGT_DEBUGFS_FILENAME		"kvmgt_nr_cache_entries"
+
 struct intel_vgpu {
 	struct intel_gvt *gvt;
 	struct mutex vgpu_lock;
 	int id;
-	unsigned long handle; /* vGPU handle used by hypervisor MPT modules */
 	bool active;
+	bool attached;
 	bool pv_notified;
 	bool failsafe;
 	unsigned int resetting_eng;
@@ -209,21 +204,40 @@ struct intel_vgpu {
 
 	struct dentry *debugfs;
 
-	/* Hypervisor-specific device state. */
-	void *vdev;
-
 	struct list_head dmabuf_obj_list_head;
 	struct mutex dmabuf_lock;
 	struct idr object_idr;
 	struct intel_vgpu_vblank_timer vblank_timer;
 
 	u32 scan_nonprivbb;
-};
 
-static inline void *intel_vgpu_vdev(struct intel_vgpu *vgpu)
-{
-	return vgpu->vdev;
-}
+	struct vfio_device vfio_device;
+	struct vfio_region *region;
+	int num_regions;
+	struct eventfd_ctx *intx_trigger;
+	struct eventfd_ctx *msi_trigger;
+
+	/*
+	 * Two caches are used to avoid mapping duplicated pages (eg.
+	 * scratch pages). This help to reduce dma setup overhead.
+	 */
+	struct rb_root gfn_cache;
+	struct rb_root dma_addr_cache;
+	unsigned long nr_cache_entries;
+	struct mutex cache_lock;
+
+	struct notifier_block iommu_notifier;
+	struct notifier_block group_notifier;
+	struct kvm *kvm;
+	struct work_struct release_work;
+	atomic_t released;
+	struct vfio_group *vfio_group;
+
+	struct kvm_page_track_notifier_node track_node;
+#define NR_BKT (1 << 18)
+	struct hlist_head ptable[NR_BKT];
+#undef NR_BKT
+};
 
 /* validating GM healthy status*/
 #define vgpu_is_vm_unhealthy(ret_val) \
@@ -272,7 +286,7 @@ struct intel_gvt_mmio {
 /* Value of command write of this reg needs to be patched */
 #define F_CMD_WRITE_PATCH	(1 << 8)
 
-	const struct gvt_mmio_block *mmio_block;
+	struct gvt_mmio_block *mmio_block;
 	unsigned int num_mmio_block;
 
 	DECLARE_HASHTABLE(mmio_info_table, INTEL_GVT_MMIO_HASH_BITS);
@@ -428,7 +442,6 @@ int intel_gvt_load_firmware(struct intel_gvt *gvt);
 #define RING_CTX_SIZE 320
 
 struct intel_vgpu_creation_params {
-	__u64 handle;
 	__u64 low_gm_sz;  /* in MB */
 	__u64 high_gm_sz; /* in MB */
 	__u64 fence_sz;
@@ -496,6 +509,9 @@ void intel_gvt_reset_vgpu(struct intel_vgpu *vgpu);
 void intel_gvt_activate_vgpu(struct intel_vgpu *vgpu);
 void intel_gvt_deactivate_vgpu(struct intel_vgpu *vgpu);
 
+int intel_gvt_set_opregion(struct intel_vgpu *vgpu);
+int intel_gvt_set_edid(struct intel_vgpu *vgpu, int port_num);
+
 /* validating GM functions */
 #define vgpu_gmadr_is_aperture(vgpu, gmadr) \
 	((gmadr >= vgpu_aperture_gmadr_base(vgpu)) && \
@@ -557,30 +573,7 @@ void populate_pvinfo_page(struct intel_vgpu *vgpu);
 
 int intel_gvt_scan_and_shadow_workload(struct intel_vgpu_workload *workload);
 void enter_failsafe_mode(struct intel_vgpu *vgpu, int reason);
-
-struct intel_gvt_ops {
-	int (*emulate_cfg_read)(struct intel_vgpu *, unsigned int, void *,
-				unsigned int);
-	int (*emulate_cfg_write)(struct intel_vgpu *, unsigned int, void *,
-				unsigned int);
-	int (*emulate_mmio_read)(struct intel_vgpu *, u64, void *,
-				unsigned int);
-	int (*emulate_mmio_write)(struct intel_vgpu *, u64, void *,
-				unsigned int);
-	struct intel_vgpu *(*vgpu_create)(struct intel_gvt *,
-				struct intel_vgpu_type *);
-	void (*vgpu_destroy)(struct intel_vgpu *vgpu);
-	void (*vgpu_release)(struct intel_vgpu *vgpu);
-	void (*vgpu_reset)(struct intel_vgpu *);
-	void (*vgpu_activate)(struct intel_vgpu *);
-	void (*vgpu_deactivate)(struct intel_vgpu *);
-	int (*vgpu_query_plane)(struct intel_vgpu *vgpu, void *);
-	int (*vgpu_get_dmabuf)(struct intel_vgpu *vgpu, unsigned int);
-	int (*write_protect_handler)(struct intel_vgpu *, u64, void *,
-				     unsigned int);
-	void (*emulate_hotplug)(struct intel_vgpu *vgpu, bool connected);
-};
-
+void intel_vgpu_detach_regions(struct intel_vgpu *vgpu);
 
 enum {
 	GVT_FAILSAFE_UNSUPPORTED_GUEST,
@@ -724,13 +717,54 @@ static inline bool intel_gvt_mmio_is_cmd_write_patch(
 	return gvt->mmio.mmio_attribute[offset >> 2] & F_CMD_WRITE_PATCH;
 }
 
+/**
+ * intel_gvt_read_gpa - copy data from GPA to host data buffer
+ * @vgpu: a vGPU
+ * @gpa: guest physical address
+ * @buf: host data buffer
+ * @len: data length
+ *
+ * Returns:
+ * Zero on success, negative error code if failed.
+ */
+static inline int intel_gvt_read_gpa(struct intel_vgpu *vgpu, unsigned long gpa,
+		void *buf, unsigned long len)
+{
+	if (!vgpu->attached)
+		return -ESRCH;
+	return vfio_dma_rw(vgpu->vfio_group, gpa, buf, len, false);
+}
+
+/**
+ * intel_gvt_write_gpa - copy data from host data buffer to GPA
+ * @vgpu: a vGPU
+ * @gpa: guest physical address
+ * @buf: host data buffer
+ * @len: data length
+ *
+ * Returns:
+ * Zero on success, negative error code if failed.
+ */
+static inline int intel_gvt_write_gpa(struct intel_vgpu *vgpu,
+		unsigned long gpa, void *buf, unsigned long len)
+{
+	if (!vgpu->attached)
+		return -ESRCH;
+	return vfio_dma_rw(vgpu->vfio_group, gpa, buf, len, true);
+}
+
 void intel_gvt_debugfs_remove_vgpu(struct intel_vgpu *vgpu);
 void intel_gvt_debugfs_init(struct intel_gvt *gvt);
 void intel_gvt_debugfs_clean(struct intel_gvt *gvt);
 
-int intel_gvt_pm_resume(struct intel_gvt *gvt);
+int intel_gvt_page_track_add(struct intel_vgpu *info, u64 gfn);
+int intel_gvt_page_track_remove(struct intel_vgpu *info, u64 gfn);
+int intel_gvt_dma_pin_guest_page(struct intel_vgpu *vgpu, dma_addr_t dma_addr);
+int intel_gvt_dma_map_guest_page(struct intel_vgpu *vgpu, unsigned long gfn,
+		unsigned long size, dma_addr_t *dma_addr);
+void intel_gvt_dma_unmap_guest_page(struct intel_vgpu *vgpu,
+		dma_addr_t dma_addr);
 
 #include "trace.h"
-#include "mpt.h"
 
 #endif

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

-/*
- * Copyright(c) 2011-2016 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:
- *    Eddie Dong <eddie.dong@intel.com>
- *    Dexuan Cui
- *    Jike Song <jike.song@intel.com>
- *
- * Contributors:
- *    Zhi Wang <zhi.a.wang@intel.com>
- *
- */
-
-#ifndef _GVT_HYPERCALL_H_
-#define _GVT_HYPERCALL_H_
-
-#include <linux/types.h>
-
-struct device;
-
-enum hypervisor_type {
-	INTEL_GVT_HYPERVISOR_XEN = 0,
-	INTEL_GVT_HYPERVISOR_KVM,
-};
-
-/*
- * Specific GVT-g MPT modules function collections. Currently GVT-g supports
- * both Xen and KVM by providing dedicated hypervisor-related MPT modules.
- */
-struct intel_gvt_mpt {
-	enum hypervisor_type type;
-	int (*host_init)(struct device *dev, void *gvt, const void *ops);
-	void (*host_exit)(struct device *dev, void *gvt);
-	int (*attach_vgpu)(void *vgpu, unsigned long *handle);
-	void (*detach_vgpu)(void *vgpu);
-	int (*inject_msi)(unsigned long handle, u32 addr, u16 data);
-	unsigned long (*from_virt_to_mfn)(void *p);
-	int (*enable_page_track)(unsigned long handle, u64 gfn);
-	int (*disable_page_track)(unsigned long handle, u64 gfn);
-	int (*read_gpa)(unsigned long handle, unsigned long gpa, void *buf,
-			unsigned long len);
-	int (*write_gpa)(unsigned long handle, unsigned long gpa, void *buf,
-			 unsigned long len);
-	unsigned long (*gfn_to_mfn)(unsigned long handle, unsigned long gfn);
-
-	int (*dma_map_guest_page)(unsigned long handle, unsigned long gfn,
-				  unsigned long size, dma_addr_t *dma_addr);
-	void (*dma_unmap_guest_page)(unsigned long handle, dma_addr_t dma_addr);
-
-	int (*dma_pin_guest_page)(unsigned long handle, dma_addr_t dma_addr);
-
-	int (*map_gfn_to_mfn)(unsigned long handle, unsigned long gfn,
-			      unsigned long mfn, unsigned int nr, bool map);
-	int (*set_trap_area)(unsigned long handle, u64 start, u64 end,
-			     bool map);
-	int (*set_opregion)(void *vgpu);
-	int (*set_edid)(void *vgpu, int port_num);
-	int (*get_vfio_device)(void *vgpu);
-	void (*put_vfio_device)(void *vgpu);
-	bool (*is_valid_gfn)(unsigned long handle, unsigned long gfn);
-};
-
-#endif /* _GVT_HYPERCALL_H_ */

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

@@ -29,6 +29,8 @@
  *
  */
 
+#include <linux/eventfd.h>
+
 #include "i915_drv.h"
 #include "i915_reg.h"
 #include "gvt.h"
@@ -397,9 +399,45 @@ static void init_irq_map(struct intel_gvt_irq *irq)
 }
 
 /* =======================vEvent injection===================== */
+
+#define MSI_CAP_CONTROL(offset) (offset + 2)
+#define MSI_CAP_ADDRESS(offset) (offset + 4)
+#define MSI_CAP_DATA(offset) (offset + 8)
+#define MSI_CAP_EN 0x1
+
 static int inject_virtual_interrupt(struct intel_vgpu *vgpu)
 {
-	return intel_gvt_hypervisor_inject_msi(vgpu);
+	unsigned long offset = vgpu->gvt->device_info.msi_cap_offset;
+	u16 control, data;
+	u32 addr;
+
+	control = *(u16 *)(vgpu_cfg_space(vgpu) + MSI_CAP_CONTROL(offset));
+	addr = *(u32 *)(vgpu_cfg_space(vgpu) + MSI_CAP_ADDRESS(offset));
+	data = *(u16 *)(vgpu_cfg_space(vgpu) + MSI_CAP_DATA(offset));
+
+	/* Do not generate MSI if MSIEN is disabled */
+	if (!(control & MSI_CAP_EN))
+		return 0;
+
+	if (WARN(control & GENMASK(15, 1), "only support one MSI format\n"))
+		return -EINVAL;
+
+	trace_inject_msi(vgpu->id, addr, data);
+
+	/*
+	 * When guest is powered off, msi_trigger is set to NULL, but vgpu's
+	 * config and mmio register isn't restored to default during guest
+	 * poweroff. If this vgpu is still used in next vm, this vgpu's pipe
+	 * may be enabled, then once this vgpu is active, it will get inject
+	 * vblank interrupt request. But msi_trigger is null until msi is
+	 * enabled by guest. so if msi_trigger is null, success is still
+	 * returned and don't inject interrupt into guest.
+	 */
+	if (!vgpu->attached)
+		return -ESRCH;
+	if (vgpu->msi_trigger && eventfd_signal(vgpu->msi_trigger, 1) != 1)
+		return -EFAULT;
+	return 0;
 }
 
 static void propagate_event(struct intel_gvt_irq *irq,

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

@@ -139,7 +139,7 @@ int intel_vgpu_emulate_mmio_read(struct intel_vgpu *vgpu, u64 pa,
 	}
 
 	if (drm_WARN_ON_ONCE(&i915->drm, !reg_is_mmio(gvt, offset))) {
-		ret = intel_gvt_hypervisor_read_gpa(vgpu, pa, p_data, bytes);
+		ret = intel_gvt_read_gpa(vgpu, pa, p_data, bytes);
 		goto out;
 	}
 
@@ -215,7 +215,7 @@ int intel_vgpu_emulate_mmio_write(struct intel_vgpu *vgpu, u64 pa,
 	}
 
 	if (drm_WARN_ON_ONCE(&i915->drm, !reg_is_mmio(gvt, offset))) {
-		ret = intel_gvt_hypervisor_write_gpa(vgpu, pa, p_data, bytes);
+		ret = intel_gvt_write_gpa(vgpu, pa, p_data, bytes);
 		goto out;
 	}
 

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

@@ -72,7 +72,6 @@ struct intel_gvt_mmio_info {
 const struct intel_engine_cs *
 intel_gvt_render_mmio_to_engine(struct intel_gvt *gvt, unsigned int reg);
 unsigned long intel_gvt_get_device_type(struct intel_gvt *gvt);
-bool intel_gvt_match_device(struct intel_gvt *gvt, unsigned long device);
 
 int intel_gvt_setup_mmio_info(struct intel_gvt *gvt);
 void intel_gvt_clean_mmio_info(struct intel_gvt *gvt);

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

@@ -87,7 +87,7 @@ void intel_vgpu_unregister_page_track(struct intel_vgpu *vgpu,
 	track = radix_tree_delete(&vgpu->page_track_tree, gfn);
 	if (track) {
 		if (track->tracked)
-			intel_gvt_hypervisor_disable_page_track(vgpu, gfn);
+			intel_gvt_page_track_remove(vgpu, gfn);
 		kfree(track);
 	}
 }
@@ -112,7 +112,7 @@ int intel_vgpu_enable_page_track(struct intel_vgpu *vgpu, unsigned long gfn)
 	if (track->tracked)
 		return 0;
 
-	ret = intel_gvt_hypervisor_enable_page_track(vgpu, gfn);
+	ret = intel_gvt_page_track_add(vgpu, gfn);
 	if (ret)
 		return ret;
 	track->tracked = true;
@@ -139,7 +139,7 @@ int intel_vgpu_disable_page_track(struct intel_vgpu *vgpu, unsigned long gfn)
 	if (!track->tracked)
 		return 0;
 
-	ret = intel_gvt_hypervisor_disable_page_track(vgpu, gfn);
+	ret = intel_gvt_page_track_remove(vgpu, gfn);
 	if (ret)
 		return ret;
 	track->tracked = false;
@@ -172,7 +172,7 @@ int intel_vgpu_page_track_handler(struct intel_vgpu *vgpu, u64 gpa,
 
 	if (unlikely(vgpu->failsafe)) {
 		/* Remove write protection to prevent furture traps. */
-		intel_vgpu_disable_page_track(vgpu, gpa >> PAGE_SHIFT);
+		intel_gvt_page_track_remove(vgpu, gpa >> PAGE_SHIFT);
 	} else {
 		ret = page_track->handler(page_track, gpa, data, bytes);
 		if (ret)

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

@@ -132,6 +132,13 @@
 #define RING_GFX_MODE(base)	_MMIO((base) + 0x29c)
 #define VF_GUARDBAND		_MMIO(0x83a4)
 
-
 #define BCS_TILE_REGISTER_VAL_OFFSET (0x43*4)
+
+/* XXX FIXME i915 has changed PP_XXX definition */
+#define PCH_PP_STATUS  _MMIO(0xc7200)
+#define PCH_PP_CONTROL _MMIO(0xc7204)
+#define PCH_PP_ON_DELAYS _MMIO(0xc7208)
+#define PCH_PP_OFF_DELAYS _MMIO(0xc720c)
+#define PCH_PP_DIVISOR _MMIO(0xc7210)
+
 #endif

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

@@ -377,7 +377,7 @@ TRACE_EVENT(render_mmio,
 
 /* This part must be out of protection */
 #undef TRACE_INCLUDE_PATH
-#define TRACE_INCLUDE_PATH .
 #undef TRACE_INCLUDE_FILE
+#define TRACE_INCLUDE_PATH ../../drivers/gpu/drm/i915/gvt
 #define TRACE_INCLUDE_FILE trace
 #include <trace/define_trace.h>

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

@@ -293,7 +293,7 @@ void intel_gvt_destroy_vgpu(struct intel_vgpu *vgpu)
 	intel_vgpu_clean_opregion(vgpu);
 	intel_vgpu_reset_ggtt(vgpu, true);
 	intel_vgpu_clean_gtt(vgpu);
-	intel_gvt_hypervisor_detach_vgpu(vgpu);
+	intel_vgpu_detach_regions(vgpu);
 	intel_vgpu_free_resource(vgpu);
 	intel_vgpu_clean_mmio(vgpu);
 	intel_vgpu_dmabuf_cleanup(vgpu);
@@ -370,8 +370,8 @@ static struct intel_vgpu *__intel_gvt_create_vgpu(struct intel_gvt *gvt,
 	struct intel_vgpu *vgpu;
 	int ret;
 
-	gvt_dbg_core("handle %llu low %llu MB high %llu MB fence %llu\n",
-			param->handle, param->low_gm_sz, param->high_gm_sz,
+	gvt_dbg_core("low %llu MB high %llu MB fence %llu\n",
+			param->low_gm_sz, param->high_gm_sz,
 			param->fence_sz);
 
 	vgpu = vzalloc(sizeof(*vgpu));
@@ -384,7 +384,6 @@ static struct intel_vgpu *__intel_gvt_create_vgpu(struct intel_gvt *gvt,
 		goto out_free_vgpu;
 
 	vgpu->id = ret;
-	vgpu->handle = param->handle;
 	vgpu->gvt = gvt;
 	vgpu->sched_ctl.weight = param->weight;
 	mutex_init(&vgpu->vgpu_lock);
@@ -405,13 +404,9 @@ static struct intel_vgpu *__intel_gvt_create_vgpu(struct intel_gvt *gvt,
 
 	populate_pvinfo_page(vgpu);
 
-	ret = intel_gvt_hypervisor_attach_vgpu(vgpu);
-	if (ret)
-		goto out_clean_vgpu_resource;
-
 	ret = intel_vgpu_init_gtt(vgpu);
 	if (ret)
-		goto out_detach_hypervisor_vgpu;
+		goto out_clean_vgpu_resource;
 
 	ret = intel_vgpu_init_opregion(vgpu);
 	if (ret)
@@ -431,14 +426,14 @@ static struct intel_vgpu *__intel_gvt_create_vgpu(struct intel_gvt *gvt,
 
 	intel_gvt_debugfs_add_vgpu(vgpu);
 
-	ret = intel_gvt_hypervisor_set_opregion(vgpu);
+	ret = intel_gvt_set_opregion(vgpu);
 	if (ret)
 		goto out_clean_sched_policy;
 
 	if (IS_BROADWELL(dev_priv) || IS_BROXTON(dev_priv))
-		ret = intel_gvt_hypervisor_set_edid(vgpu, PORT_B);
+		ret = intel_gvt_set_edid(vgpu, PORT_B);
 	else
-		ret = intel_gvt_hypervisor_set_edid(vgpu, PORT_D);
+		ret = intel_gvt_set_edid(vgpu, PORT_D);
 	if (ret)
 		goto out_clean_sched_policy;
 
@@ -454,8 +449,6 @@ static struct intel_vgpu *__intel_gvt_create_vgpu(struct intel_gvt *gvt,
 	intel_vgpu_clean_opregion(vgpu);
 out_clean_gtt:
 	intel_vgpu_clean_gtt(vgpu);
-out_detach_hypervisor_vgpu:
-	intel_gvt_hypervisor_detach_vgpu(vgpu);
 out_clean_vgpu_resource:
 	intel_vgpu_free_resource(vgpu);
 out_clean_vgpu_mmio:
@@ -483,7 +476,6 @@ struct intel_vgpu *intel_gvt_create_vgpu(struct intel_gvt *gvt,
 	struct intel_vgpu_creation_params param;
 	struct intel_vgpu *vgpu;
 
-	param.handle = 0;
 	param.primary = 1;
 	param.low_gm_sz = type->low_gm_size;
 	param.high_gm_sz = type->high_gm_size;

drivers/gpu/drm/i915/i915_driver.c

@@ -470,11 +470,6 @@ static void i915_driver_mmio_release(struct drm_i915_private *dev_priv)
 	pci_dev_put(dev_priv->bridge_dev);
 }
 
-static void intel_sanitize_options(struct drm_i915_private *dev_priv)
-{
-	intel_gvt_sanitize_options(dev_priv);
-}
-
 /**
  * i915_set_dma_info - set all relevant PCI dma info as configured for the
  * platform
@@ -568,8 +563,6 @@ static int i915_driver_hw_probe(struct drm_i915_private *dev_priv)
 		}
 	}
 
-	intel_sanitize_options(dev_priv);
-
 	/* needs to be done before ggtt probe */
 	intel_dram_edram_detect(dev_priv);
 

drivers/gpu/drm/i915/i915_drv.h

@@ -400,6 +400,9 @@ struct i915_virtual_gpu {
 	struct mutex lock; /* serialises sending of g2v_notify command pkts */
 	bool active;
 	u32 caps;
+	u32 *initial_mmio;
+	u8 *initial_cfg_space;
+	struct list_head entry;
 };
 
 struct i915_selftest_stash {