drm/i915/gvt: Add opregion support

Windows guest driver needs vbt in opregion, to configure the setting
for display. Without opregion support, the display registers won't
be set and this blocks display model to get the correct information
of the guest display plane.

This patch is to provide a virtual opregion for guest. The original
author of this patch is Xiaoguang Chen.

This patch is split from the "Dma-buf support for GVT-g" patch set,
with being rebased to the latest gvt-staging branch.

v3:
- add checking region index during intel_vgpu_rw. (Xiong)

v2:
- refine intel_vgpu_reg_release_opregion. (Xiong)

Here are the previous version comments:

v18:
- unmap vgpu's opregion when destroying vgpu.

v16:
- rebase to 4.14.0-rc6.

Signed-off-by: Bing Niu <bing.niu@intel.com>
Signed-off-by: Tina Zhang <tina.zhang@intel.com>
Tested-by: Xiong Zhang <xiong.y.zhang@intel.com>
Cc: Zhenyu Wang <zhenyuw@linux.intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>

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

@@ -125,6 +125,7 @@ struct intel_vgpu_irq {
 struct intel_vgpu_opregion {
 	bool mapped;
 	void *va;
+	void *va_gopregion;
 	u32 gfn[INTEL_GVT_OPREGION_PAGES];
 };
 

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

@@ -55,6 +55,7 @@ struct intel_gvt_mpt {
 			      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);
 };
 
 extern struct intel_gvt_mpt xengt_mpt;

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

@@ -53,11 +53,23 @@ static const struct intel_gvt_ops *intel_gvt_ops;
 #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)
 
+#define OPREGION_SIGNATURE "IntelGraphicsMem"
+
+struct vfio_region;
+struct intel_vgpu_regops {
+	size_t (*rw)(struct intel_vgpu *vgpu, char *buf,
+			size_t count, loff_t *ppos, bool iswrite);
+	void (*release)(struct intel_vgpu *vgpu,
+			struct vfio_region *region);
+};
+
 struct vfio_region {
 	u32				type;
 	u32				subtype;
 	size_t				size;
 	u32				flags;
+	const struct intel_vgpu_regops	*ops;
+	void				*data;
 };
 
 struct kvmgt_pgfn {
@@ -316,6 +328,87 @@ static void kvmgt_protect_table_del(struct kvmgt_guest_info *info,
 	}
 }
 
+static size_t intel_vgpu_reg_rw_opregion(struct intel_vgpu *vgpu, char *buf,
+		size_t count, loff_t *ppos, bool iswrite)
+{
+	unsigned int i = VFIO_PCI_OFFSET_TO_INDEX(*ppos) -
+			VFIO_PCI_NUM_REGIONS;
+	void *base = vgpu->vdev.region[i].data;
+	loff_t pos = *ppos & VFIO_PCI_OFFSET_MASK;
+
+	if (pos >= vgpu->vdev.region[i].size || iswrite) {
+		gvt_vgpu_err("invalid op or offset for Intel vgpu OpRegion\n");
+		return -EINVAL;
+	}
+	count = min(count, (size_t)(vgpu->vdev.region[i].size - pos));
+	memcpy(buf, base + pos, count);
+
+	return count;
+}
+
+static void intel_vgpu_reg_release_opregion(struct intel_vgpu *vgpu,
+		struct vfio_region *region)
+{
+}
+
+static const struct intel_vgpu_regops intel_vgpu_regops_opregion = {
+	.rw = intel_vgpu_reg_rw_opregion,
+	.release = intel_vgpu_reg_release_opregion,
+};
+
+static int intel_vgpu_register_reg(struct intel_vgpu *vgpu,
+		unsigned int type, unsigned int subtype,
+		const struct intel_vgpu_regops *ops,
+		size_t size, u32 flags, void *data)
+{
+	struct vfio_region *region;
+
+	region = krealloc(vgpu->vdev.region,
+			(vgpu->vdev.num_regions + 1) * sizeof(*region),
+			GFP_KERNEL);
+	if (!region)
+		return -ENOMEM;
+
+	vgpu->vdev.region = region;
+	vgpu->vdev.region[vgpu->vdev.num_regions].type = type;
+	vgpu->vdev.region[vgpu->vdev.num_regions].subtype = subtype;
+	vgpu->vdev.region[vgpu->vdev.num_regions].ops = ops;
+	vgpu->vdev.region[vgpu->vdev.num_regions].size = size;
+	vgpu->vdev.region[vgpu->vdev.num_regions].flags = flags;
+	vgpu->vdev.region[vgpu->vdev.num_regions].data = data;
+	vgpu->vdev.num_regions++;
+
+	return 0;
+}
+
+static int kvmgt_set_opregion(void *p_vgpu)
+{
+	struct intel_vgpu *vgpu = (struct intel_vgpu *)p_vgpu;
+	void *base;
+	int ret;
+
+	/* Each vgpu has its own opregion, although VFIO would create another
+	 * one later. This one is used to expose opregion to VFIO. And the
+	 * other one created by VFIO later, is used by guest actually.
+	 */
+	base = vgpu_opregion(vgpu)->va;
+	if (!base)
+		return -ENOMEM;
+
+	if (memcmp(base, OPREGION_SIGNATURE, 16)) {
+		memunmap(base);
+		return -EINVAL;
+	}
+
+	ret = intel_vgpu_register_reg(vgpu,
+			PCI_VENDOR_ID_INTEL | VFIO_REGION_TYPE_PCI_VENDOR_TYPE,
+			VFIO_REGION_SUBTYPE_INTEL_IGD_OPREGION,
+			&intel_vgpu_regops_opregion, OPREGION_SIZE,
+			VFIO_REGION_INFO_FLAG_READ, base);
+
+	return ret;
+}
+
 static int intel_vgpu_create(struct kobject *kobj, struct mdev_device *mdev)
 {
 	struct intel_vgpu *vgpu = NULL;
@@ -546,7 +639,7 @@ static ssize_t intel_vgpu_rw(struct mdev_device *mdev, char *buf,
 	int ret = -EINVAL;
 
 
-	if (index >= VFIO_PCI_NUM_REGIONS) {
+	if (index >= VFIO_PCI_NUM_REGIONS + vgpu->vdev.num_regions) {
 		gvt_vgpu_err("invalid index: %u\n", index);
 		return -EINVAL;
 	}
@@ -574,8 +667,14 @@ static ssize_t intel_vgpu_rw(struct mdev_device *mdev, char *buf,
 	case VFIO_PCI_BAR5_REGION_INDEX:
 	case VFIO_PCI_VGA_REGION_INDEX:
 	case VFIO_PCI_ROM_REGION_INDEX:
+		break;
 	default:
-		gvt_vgpu_err("unsupported region: %u\n", index);
+		if (index >= VFIO_PCI_NUM_REGIONS + vgpu->vdev.num_regions)
+			return -EINVAL;
+
+		index -= VFIO_PCI_NUM_REGIONS;
+		return vgpu->vdev.region[index].ops->rw(vgpu, buf, count,
+				ppos, is_write);
 	}
 
 	return ret == 0 ? count : ret;
@@ -838,7 +937,8 @@ static long intel_vgpu_ioctl(struct mdev_device *mdev, unsigned int cmd,
 
 		info.flags = VFIO_DEVICE_FLAGS_PCI;
 		info.flags |= VFIO_DEVICE_FLAGS_RESET;
-		info.num_regions = VFIO_PCI_NUM_REGIONS;
+		info.num_regions = VFIO_PCI_NUM_REGIONS +
+				vgpu->vdev.num_regions;
 		info.num_irqs = VFIO_PCI_NUM_IRQS;
 
 		return copy_to_user((void __user *)arg, &info, minsz) ?
@@ -959,6 +1059,7 @@ static long intel_vgpu_ioctl(struct mdev_device *mdev, unsigned int cmd,
 		}
 
 		if (caps.size) {
+			info.flags |= VFIO_REGION_INFO_FLAG_CAPS;
 			if (info.argsz < sizeof(info) + caps.size) {
 				info.argsz = sizeof(info) + caps.size;
 				info.cap_offset = 0;
@@ -1426,6 +1527,7 @@ struct intel_gvt_mpt kvmgt_mpt = {
 	.read_gpa = kvmgt_read_gpa,
 	.write_gpa = kvmgt_write_gpa,
 	.gfn_to_mfn = kvmgt_gfn_to_pfn,
+	.set_opregion = kvmgt_set_opregion,
 };
 EXPORT_SYMBOL_GPL(kvmgt_mpt);
 

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

@@ -294,4 +294,19 @@ static inline int intel_gvt_hypervisor_set_trap_area(
 	return intel_gvt_host.mpt->set_trap_area(vgpu->handle, start, end, map);
 }
 
+/**
+ * intel_gvt_hypervisor_set_opregion - Set opregion for guest
+ * @vgpu: a vGPU
+ *
+ * Returns:
+ * Zero on success, negative error code if failed.
+ */
+static inline int intel_gvt_hypervisor_set_opregion(struct intel_vgpu *vgpu)
+{
+	if (!intel_gvt_host.mpt->set_opregion)
+		return 0;
+
+	return intel_gvt_host.mpt->set_opregion(vgpu);
+}
+
 #endif /* _GVT_MPT_H_ */

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

@@ -297,19 +297,41 @@ static int map_vgpu_opregion(struct intel_vgpu *vgpu, bool map)
  */
 int intel_vgpu_opregion_base_write_handler(struct intel_vgpu *vgpu, u32 gpa)
 {
-	int i, ret;
 
-	/**
-	 * Wins guest on Xengt will write this register twice: xen hvmloader and
-	 * windows graphic driver.
-	 */
-	if (vgpu_opregion(vgpu)->mapped)
-		map_vgpu_opregion(vgpu, false);
+	int i, ret = 0;
+	unsigned long pfn;
+
+	gvt_dbg_core("emulate opregion from kernel\n");
+
+	switch (intel_gvt_host.hypervisor_type) {
+	case INTEL_GVT_HYPERVISOR_KVM:
+		pfn = intel_gvt_hypervisor_gfn_to_mfn(vgpu, gpa >> PAGE_SHIFT);
+		vgpu_opregion(vgpu)->va_gopregion = memremap(pfn << PAGE_SHIFT,
+						INTEL_GVT_OPREGION_SIZE,
+						MEMREMAP_WB);
+		if (!vgpu_opregion(vgpu)->va_gopregion) {
+			gvt_vgpu_err("failed to map guest opregion\n");
+			ret = -EFAULT;
+		}
+		vgpu_opregion(vgpu)->mapped = true;
+		break;
+	case INTEL_GVT_HYPERVISOR_XEN:
+		/**
+		 * Wins guest on Xengt will write this register twice: xen
+		 * hvmloader and windows graphic driver.
+		 */
+		if (vgpu_opregion(vgpu)->mapped)
+			map_vgpu_opregion(vgpu, false);
 
-	for (i = 0; i < INTEL_GVT_OPREGION_PAGES; i++)
-		vgpu_opregion(vgpu)->gfn[i] = (gpa >> PAGE_SHIFT) + i;
+		for (i = 0; i < INTEL_GVT_OPREGION_PAGES; i++)
+			vgpu_opregion(vgpu)->gfn[i] = (gpa >> PAGE_SHIFT) + i;
 
-	ret = map_vgpu_opregion(vgpu, true);
+		ret = map_vgpu_opregion(vgpu, true);
+		break;
+	default:
+		ret = -EINVAL;
+		gvt_vgpu_err("not supported hypervisor\n");
+	}
 
 	return ret;
 }
@@ -326,13 +348,20 @@ void intel_vgpu_clean_opregion(struct intel_vgpu *vgpu)
 	if (!vgpu_opregion(vgpu)->va)
 		return;
 
-	if (vgpu_opregion(vgpu)->mapped)
-		map_vgpu_opregion(vgpu, false);
-
+	if (intel_gvt_host.hypervisor_type == INTEL_GVT_HYPERVISOR_XEN) {
+		if (vgpu_opregion(vgpu)->mapped)
+			map_vgpu_opregion(vgpu, false);
+	} else if (intel_gvt_host.hypervisor_type == INTEL_GVT_HYPERVISOR_KVM) {
+		if (vgpu_opregion(vgpu)->mapped) {
+			memunmap(vgpu_opregion(vgpu)->va_gopregion);
+			vgpu_opregion(vgpu)->va_gopregion = NULL;
+		}
+	}
 	free_pages((unsigned long)vgpu_opregion(vgpu)->va,
-			get_order(INTEL_GVT_OPREGION_SIZE));
+		   get_order(INTEL_GVT_OPREGION_SIZE));
 
 	vgpu_opregion(vgpu)->va = NULL;
+
 }
 
 
@@ -454,8 +483,21 @@ int intel_vgpu_emulate_opregion_request(struct intel_vgpu *vgpu, u32 swsci)
 	u32 *scic, *parm;
 	u32 func, subfunc;
 
-	scic = vgpu_opregion(vgpu)->va + INTEL_GVT_OPREGION_SCIC;
-	parm = vgpu_opregion(vgpu)->va + INTEL_GVT_OPREGION_PARM;
+	switch (intel_gvt_host.hypervisor_type) {
+	case INTEL_GVT_HYPERVISOR_XEN:
+		scic = vgpu_opregion(vgpu)->va + INTEL_GVT_OPREGION_SCIC;
+		parm = vgpu_opregion(vgpu)->va + INTEL_GVT_OPREGION_PARM;
+		break;
+	case INTEL_GVT_HYPERVISOR_KVM:
+		scic = vgpu_opregion(vgpu)->va_gopregion +
+						INTEL_GVT_OPREGION_SCIC;
+		parm = vgpu_opregion(vgpu)->va_gopregion +
+						INTEL_GVT_OPREGION_PARM;
+		break;
+	default:
+		gvt_vgpu_err("not supported hypervisor\n");
+		return -EINVAL;
+	}
 
 	if (!(swsci & SWSCI_SCI_SELECT)) {
 		gvt_vgpu_err("requesting SMI service\n");

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

@@ -390,6 +390,10 @@ static struct intel_vgpu *__intel_gvt_create_vgpu(struct intel_gvt *gvt,
 	if (ret)
 		goto out_clean_sched_policy;
 
+	ret = intel_gvt_hypervisor_set_opregion(vgpu);
+	if (ret)
+		goto out_clean_sched_policy;
+
 	mutex_unlock(&gvt->lock);
 
 	return vgpu;