drm/i915/gvt/kvmgt: add vfio/mdev support to KVMGT
KVMGT leverages vfio/mdev to mediate device accesses from guest,
this patch adds the vfio/mdev support, thereby completes the
functionality. An intel_vgpu is presented as a mdev device,
and full userspace API compatibility with vfio-pci is kept.
An intel_vgpu_ops is provided to mdev framework, methods get
called to create/remove a vgpu, to open/close it, and to
access it.
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>
diff --git a/drivers/gpu/drm/i915/Kconfig b/drivers/gpu/drm/i915/Kconfig
index 5ddde73..183f5dc 100644
--- a/drivers/gpu/drm/i915/Kconfig
+++ b/drivers/gpu/drm/i915/Kconfig
@@ -116,6 +116,7 @@
tristate "Enable KVM/VFIO support for Intel GVT-g"
depends on DRM_I915_GVT
depends on KVM
+ depends on VFIO_MDEV && VFIO_MDEV_DEVICE
default n
help
Choose this option if you want to enable KVMGT support for
diff --git a/drivers/gpu/drm/i915/gvt/Makefile b/drivers/gpu/drm/i915/gvt/Makefile
index 8a46a7f..b123c20 100644
--- a/drivers/gpu/drm/i915/gvt/Makefile
+++ b/drivers/gpu/drm/i915/gvt/Makefile
@@ -5,6 +5,4 @@
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
diff --git a/drivers/gpu/drm/i915/gvt/gvt.h b/drivers/gpu/drm/i915/gvt/gvt.h
index b1a7c8d..ad0e936 100644
--- a/drivers/gpu/drm/i915/gvt/gvt.h
+++ b/drivers/gpu/drm/i915/gvt/gvt.h
@@ -164,15 +164,17 @@
#if IS_ENABLED(CONFIG_DRM_I915_GVT_KVMGT)
struct {
- struct device *mdev;
+ struct mdev_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;
+ struct notifier_block group_notifier;
+ struct kvm *kvm;
+ struct work_struct release_work;
} vdev;
#endif
};
diff --git a/drivers/gpu/drm/i915/gvt/kvmgt.c b/drivers/gpu/drm/i915/gvt/kvmgt.c
index 24496ad..4dd6722 100644
--- a/drivers/gpu/drm/i915/gvt/kvmgt.c
+++ b/drivers/gpu/drm/i915/gvt/kvmgt.c
@@ -40,24 +40,13 @@
#include <linux/uuid.h>
#include <linux/kvm_host.h>
#include <linux/vfio.h>
+#include <linux/mdev.h>
#include "i915_drv.h"
#include "gvt.h"
-static inline long kvmgt_pin_pages(struct device *dev, unsigned long *user_pfn,
- long npage, int prot, unsigned long *phys_pfn)
-{
- return 0;
-}
-static inline long kvmgt_unpin_pages(struct device *dev, unsigned long *pfn,
- long npage)
-{
- return 0;
-}
-
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)
@@ -91,6 +80,15 @@
kvm_pfn_t pfn;
};
+static inline bool handle_valid(unsigned long handle)
+{
+ return !!(handle & ~0xff);
+}
+
+static int kvmgt_guest_init(struct mdev_device *mdev);
+static void intel_vgpu_release_work(struct work_struct *work);
+static bool kvmgt_guest_exit(struct kvmgt_guest_info *info);
+
static struct gvt_dma *__gvt_cache_find(struct intel_vgpu *vgpu, gfn_t gfn)
{
struct rb_node *node = vgpu->vdev.cache.rb_node;
@@ -168,9 +166,10 @@
static void gvt_cache_remove(struct intel_vgpu *vgpu, gfn_t gfn)
{
- struct device *dev = vgpu->vdev.mdev;
+ struct device *dev = &vgpu->vdev.mdev->dev;
struct gvt_dma *this;
- unsigned long pfn;
+ unsigned long g1;
+ int rc;
mutex_lock(&vgpu->vdev.cache_lock);
this = __gvt_cache_find(vgpu, gfn);
@@ -179,8 +178,9 @@
return;
}
- pfn = this->pfn;
- WARN_ON((kvmgt_unpin_pages(dev, &pfn, 1) != 1));
+ g1 = gfn;
+ rc = vfio_unpin_pages(dev, &g1, 1);
+ WARN_ON(rc != 1);
__gvt_cache_remove_entry(vgpu, this);
mutex_unlock(&vgpu->vdev.cache_lock);
}
@@ -195,15 +195,15 @@
{
struct gvt_dma *dma;
struct rb_node *node = NULL;
- struct device *dev = vgpu->vdev.mdev;
- unsigned long pfn;
+ struct device *dev = &vgpu->vdev.mdev->dev;
+ unsigned long gfn;
mutex_lock(&vgpu->vdev.cache_lock);
while ((node = rb_first(&vgpu->vdev.cache))) {
dma = rb_entry(node, struct gvt_dma, node);
- pfn = dma->pfn;
+ gfn = dma->gfn;
- kvmgt_unpin_pages(dev, &pfn, 1);
+ vfio_unpin_pages(dev, &gfn, 1);
__gvt_cache_remove_entry(vgpu, dma);
}
mutex_unlock(&vgpu->vdev.cache_lock);
@@ -227,7 +227,53 @@
return NULL;
}
+static ssize_t available_instance_show(struct kobject *kobj, struct device *dev,
+ char *buf)
+{
+ struct intel_vgpu_type *type;
+ unsigned int num = 0;
+ void *gvt = kdev_to_i915(dev)->gvt;
+
+ type = intel_gvt_find_vgpu_type(gvt, kobject_name(kobj));
+ if (!type)
+ num = 0;
+ else
+ num = type->avail_instance;
+
+ return sprintf(buf, "%u\n", num);
+}
+
+static ssize_t device_api_show(struct kobject *kobj, struct device *dev,
+ char *buf)
+{
+ return sprintf(buf, "%s\n", VFIO_DEVICE_API_PCI_STRING);
+}
+
+static ssize_t description_show(struct kobject *kobj, struct device *dev,
+ char *buf)
+{
+ struct intel_vgpu_type *type;
+ void *gvt = kdev_to_i915(dev)->gvt;
+
+ type = intel_gvt_find_vgpu_type(gvt, kobject_name(kobj));
+ if (!type)
+ return 0;
+
+ return sprintf(buf, "low_gm_size: %dMB\nhigh_gm_size: %dMB\n"
+ "fence: %d\n",
+ BYTES_TO_MB(type->low_gm_size),
+ BYTES_TO_MB(type->high_gm_size),
+ type->fence);
+}
+
+static MDEV_TYPE_ATTR_RO(available_instance);
+static MDEV_TYPE_ATTR_RO(device_api);
+static MDEV_TYPE_ATTR_RO(description);
+
static struct attribute *type_attrs[] = {
+ &mdev_type_attr_available_instance.attr,
+ &mdev_type_attr_device_api.attr,
+ &mdev_type_attr_description.attr,
NULL,
};
@@ -343,6 +389,720 @@
}
}
+static int intel_vgpu_create(struct kobject *kobj, struct mdev_device *mdev)
+{
+ struct intel_vgpu *vgpu;
+ struct intel_vgpu_type *type;
+ struct device *pdev;
+ void *gvt;
+
+ pdev = mdev->parent->dev;
+ gvt = kdev_to_i915(pdev)->gvt;
+
+ type = intel_gvt_find_vgpu_type(gvt, kobject_name(kobj));
+ if (!type) {
+ gvt_err("failed to find type %s to create\n",
+ kobject_name(kobj));
+ return -EINVAL;
+ }
+
+ vgpu = intel_gvt_ops->vgpu_create(gvt, type);
+ if (IS_ERR_OR_NULL(vgpu)) {
+ gvt_err("create intel vgpu failed\n");
+ return -EINVAL;
+ }
+
+ INIT_WORK(&vgpu->vdev.release_work, intel_vgpu_release_work);
+
+ vgpu->vdev.mdev = mdev;
+ mdev_set_drvdata(mdev, vgpu);
+
+ gvt_dbg_core("intel_vgpu_create succeeded for mdev: %s\n",
+ dev_name(&mdev->dev));
+ return 0;
+}
+
+static int intel_vgpu_remove(struct mdev_device *mdev)
+{
+ struct intel_vgpu *vgpu = mdev_get_drvdata(mdev);
+
+ if (handle_valid(vgpu->handle))
+ return -EBUSY;
+
+ intel_gvt_ops->vgpu_destroy(vgpu);
+ return 0;
+}
+
+static int intel_vgpu_iommu_notifier(struct notifier_block *nb,
+ unsigned long action, void *data)
+{
+ struct intel_vgpu *vgpu = container_of(nb,
+ struct intel_vgpu,
+ vdev.iommu_notifier);
+
+ if (action == VFIO_IOMMU_NOTIFY_DMA_UNMAP) {
+ struct vfio_iommu_type1_dma_unmap *unmap = data;
+ unsigned long gfn, end_gfn;
+
+ gfn = unmap->iova >> PAGE_SHIFT;
+ end_gfn = gfn + unmap->size / PAGE_SIZE;
+
+ while (gfn < end_gfn)
+ gvt_cache_remove(vgpu, gfn++);
+ }
+
+ return NOTIFY_OK;
+}
+
+static int intel_vgpu_group_notifier(struct notifier_block *nb,
+ unsigned long action, void *data)
+{
+ struct intel_vgpu *vgpu = container_of(nb,
+ struct intel_vgpu,
+ vdev.group_notifier);
+
+ /* the only action we care about */
+ if (action == VFIO_GROUP_NOTIFY_SET_KVM) {
+ vgpu->vdev.kvm = data;
+
+ if (!data)
+ schedule_work(&vgpu->vdev.release_work);
+ }
+
+ return NOTIFY_OK;
+}
+
+static int intel_vgpu_open(struct mdev_device *mdev)
+{
+ struct intel_vgpu *vgpu = mdev_get_drvdata(mdev);
+ unsigned long events;
+ int ret;
+
+ vgpu->vdev.iommu_notifier.notifier_call = intel_vgpu_iommu_notifier;
+ vgpu->vdev.group_notifier.notifier_call = intel_vgpu_group_notifier;
+
+ events = VFIO_IOMMU_NOTIFY_DMA_UNMAP;
+ ret = vfio_register_notifier(&mdev->dev, VFIO_IOMMU_NOTIFY, &events,
+ &vgpu->vdev.iommu_notifier);
+ if (ret != 0) {
+ gvt_err("vfio_register_notifier for iommu failed: %d\n", ret);
+ goto out;
+ }
+
+ events = VFIO_GROUP_NOTIFY_SET_KVM;
+ ret = vfio_register_notifier(&mdev->dev, VFIO_GROUP_NOTIFY, &events,
+ &vgpu->vdev.group_notifier);
+ if (ret != 0) {
+ gvt_err("vfio_register_notifier for group failed: %d\n", ret);
+ goto undo_iommu;
+ }
+
+ return kvmgt_guest_init(mdev);
+
+undo_iommu:
+ vfio_unregister_notifier(&mdev->dev, VFIO_IOMMU_NOTIFY,
+ &vgpu->vdev.iommu_notifier);
+out:
+ return ret;
+}
+
+static void __intel_vgpu_release(struct intel_vgpu *vgpu)
+{
+ struct kvmgt_guest_info *info;
+
+ if (!handle_valid(vgpu->handle))
+ return;
+
+ vfio_unregister_notifier(&vgpu->vdev.mdev->dev, VFIO_IOMMU_NOTIFY,
+ &vgpu->vdev.iommu_notifier);
+ vfio_unregister_notifier(&vgpu->vdev.mdev->dev, VFIO_GROUP_NOTIFY,
+ &vgpu->vdev.group_notifier);
+
+ info = (struct kvmgt_guest_info *)vgpu->handle;
+ kvmgt_guest_exit(info);
+ vgpu->handle = 0;
+}
+
+static void intel_vgpu_release(struct mdev_device *mdev)
+{
+ struct intel_vgpu *vgpu = mdev_get_drvdata(mdev);
+
+ __intel_vgpu_release(vgpu);
+}
+
+static void intel_vgpu_release_work(struct work_struct *work)
+{
+ struct intel_vgpu *vgpu = container_of(work, struct intel_vgpu,
+ vdev.release_work);
+ __intel_vgpu_release(vgpu);
+}
+
+static uint64_t intel_vgpu_get_bar0_addr(struct intel_vgpu *vgpu)
+{
+ u32 start_lo, start_hi;
+ u32 mem_type;
+ int pos = PCI_BASE_ADDRESS_0;
+
+ start_lo = (*(u32 *)(vgpu->cfg_space.virtual_cfg_space + pos)) &
+ PCI_BASE_ADDRESS_MEM_MASK;
+ mem_type = (*(u32 *)(vgpu->cfg_space.virtual_cfg_space + pos)) &
+ PCI_BASE_ADDRESS_MEM_TYPE_MASK;
+
+ switch (mem_type) {
+ case PCI_BASE_ADDRESS_MEM_TYPE_64:
+ start_hi = (*(u32 *)(vgpu->cfg_space.virtual_cfg_space
+ + pos + 4));
+ break;
+ case PCI_BASE_ADDRESS_MEM_TYPE_32:
+ case PCI_BASE_ADDRESS_MEM_TYPE_1M:
+ /* 1M mem BAR treated as 32-bit BAR */
+ default:
+ /* mem unknown type treated as 32-bit BAR */
+ start_hi = 0;
+ break;
+ }
+
+ return ((u64)start_hi << 32) | start_lo;
+}
+
+static ssize_t intel_vgpu_rw(struct mdev_device *mdev, char *buf,
+ size_t count, loff_t *ppos, bool is_write)
+{
+ struct intel_vgpu *vgpu = mdev_get_drvdata(mdev);
+ unsigned int index = VFIO_PCI_OFFSET_TO_INDEX(*ppos);
+ uint64_t pos = *ppos & VFIO_PCI_OFFSET_MASK;
+ int ret = -EINVAL;
+
+
+ if (index >= VFIO_PCI_NUM_REGIONS) {
+ gvt_err("invalid index: %u\n", index);
+ return -EINVAL;
+ }
+
+ switch (index) {
+ case VFIO_PCI_CONFIG_REGION_INDEX:
+ if (is_write)
+ ret = intel_gvt_ops->emulate_cfg_write(vgpu, pos,
+ buf, count);
+ else
+ ret = intel_gvt_ops->emulate_cfg_read(vgpu, pos,
+ buf, count);
+ break;
+ case VFIO_PCI_BAR0_REGION_INDEX:
+ case VFIO_PCI_BAR1_REGION_INDEX:
+ if (is_write) {
+ uint64_t bar0_start = intel_vgpu_get_bar0_addr(vgpu);
+
+ ret = intel_gvt_ops->emulate_mmio_write(vgpu,
+ bar0_start + pos, buf, count);
+ } else {
+ uint64_t bar0_start = intel_vgpu_get_bar0_addr(vgpu);
+
+ ret = intel_gvt_ops->emulate_mmio_read(vgpu,
+ bar0_start + pos, buf, count);
+ }
+ break;
+ case VFIO_PCI_BAR2_REGION_INDEX:
+ case VFIO_PCI_BAR3_REGION_INDEX:
+ case VFIO_PCI_BAR4_REGION_INDEX:
+ case VFIO_PCI_BAR5_REGION_INDEX:
+ case VFIO_PCI_VGA_REGION_INDEX:
+ case VFIO_PCI_ROM_REGION_INDEX:
+ default:
+ gvt_err("unsupported region: %u\n", index);
+ }
+
+ return ret == 0 ? count : ret;
+}
+
+static ssize_t intel_vgpu_read(struct mdev_device *mdev, char __user *buf,
+ size_t count, loff_t *ppos)
+{
+ unsigned int done = 0;
+ int ret;
+
+ while (count) {
+ size_t filled;
+
+ if (count >= 4 && !(*ppos % 4)) {
+ u32 val;
+
+ ret = intel_vgpu_rw(mdev, (char *)&val, sizeof(val),
+ ppos, false);
+ if (ret <= 0)
+ goto read_err;
+
+ if (copy_to_user(buf, &val, sizeof(val)))
+ goto read_err;
+
+ filled = 4;
+ } else if (count >= 2 && !(*ppos % 2)) {
+ u16 val;
+
+ ret = intel_vgpu_rw(mdev, (char *)&val, sizeof(val),
+ ppos, false);
+ if (ret <= 0)
+ goto read_err;
+
+ if (copy_to_user(buf, &val, sizeof(val)))
+ goto read_err;
+
+ filled = 2;
+ } else {
+ u8 val;
+
+ ret = intel_vgpu_rw(mdev, &val, sizeof(val), ppos,
+ false);
+ if (ret <= 0)
+ goto read_err;
+
+ if (copy_to_user(buf, &val, sizeof(val)))
+ goto read_err;
+
+ filled = 1;
+ }
+
+ count -= filled;
+ done += filled;
+ *ppos += filled;
+ buf += filled;
+ }
+
+ return done;
+
+read_err:
+ return -EFAULT;
+}
+
+static ssize_t intel_vgpu_write(struct mdev_device *mdev,
+ const char __user *buf,
+ size_t count, loff_t *ppos)
+{
+ unsigned int done = 0;
+ int ret;
+
+ while (count) {
+ size_t filled;
+
+ if (count >= 4 && !(*ppos % 4)) {
+ u32 val;
+
+ if (copy_from_user(&val, buf, sizeof(val)))
+ goto write_err;
+
+ ret = intel_vgpu_rw(mdev, (char *)&val, sizeof(val),
+ ppos, true);
+ if (ret <= 0)
+ goto write_err;
+
+ filled = 4;
+ } else if (count >= 2 && !(*ppos % 2)) {
+ u16 val;
+
+ if (copy_from_user(&val, buf, sizeof(val)))
+ goto write_err;
+
+ ret = intel_vgpu_rw(mdev, (char *)&val,
+ sizeof(val), ppos, true);
+ if (ret <= 0)
+ goto write_err;
+
+ filled = 2;
+ } else {
+ u8 val;
+
+ if (copy_from_user(&val, buf, sizeof(val)))
+ goto write_err;
+
+ ret = intel_vgpu_rw(mdev, &val, sizeof(val),
+ ppos, true);
+ if (ret <= 0)
+ goto write_err;
+
+ filled = 1;
+ }
+
+ count -= filled;
+ done += filled;
+ *ppos += filled;
+ buf += filled;
+ }
+
+ return done;
+write_err:
+ return -EFAULT;
+}
+
+static int intel_vgpu_mmap(struct mdev_device *mdev, struct vm_area_struct *vma)
+{
+ unsigned int index;
+ u64 virtaddr;
+ unsigned long req_size, pgoff = 0;
+ pgprot_t pg_prot;
+ struct intel_vgpu *vgpu = mdev_get_drvdata(mdev);
+
+ index = vma->vm_pgoff >> (VFIO_PCI_OFFSET_SHIFT - PAGE_SHIFT);
+ if (index >= VFIO_PCI_ROM_REGION_INDEX)
+ return -EINVAL;
+
+ if (vma->vm_end < vma->vm_start)
+ return -EINVAL;
+ if ((vma->vm_flags & VM_SHARED) == 0)
+ return -EINVAL;
+ if (index != VFIO_PCI_BAR2_REGION_INDEX)
+ return -EINVAL;
+
+ pg_prot = vma->vm_page_prot;
+ virtaddr = vma->vm_start;
+ req_size = vma->vm_end - vma->vm_start;
+ pgoff = vgpu_aperture_pa_base(vgpu) >> PAGE_SHIFT;
+
+ return remap_pfn_range(vma, virtaddr, pgoff, req_size, pg_prot);
+}
+
+static int intel_vgpu_get_irq_count(struct intel_vgpu *vgpu, int type)
+{
+ if (type == VFIO_PCI_INTX_IRQ_INDEX || type == VFIO_PCI_MSI_IRQ_INDEX)
+ return 1;
+
+ return 0;
+}
+
+static int intel_vgpu_set_intx_mask(struct intel_vgpu *vgpu,
+ unsigned int index, unsigned int start,
+ unsigned int count, uint32_t flags,
+ void *data)
+{
+ return 0;
+}
+
+static int intel_vgpu_set_intx_unmask(struct intel_vgpu *vgpu,
+ unsigned int index, unsigned int start,
+ unsigned int count, uint32_t flags, void *data)
+{
+ return 0;
+}
+
+static int intel_vgpu_set_intx_trigger(struct intel_vgpu *vgpu,
+ unsigned int index, unsigned int start, unsigned int count,
+ uint32_t flags, void *data)
+{
+ return 0;
+}
+
+static int intel_vgpu_set_msi_trigger(struct intel_vgpu *vgpu,
+ unsigned int index, unsigned int start, unsigned int count,
+ uint32_t flags, void *data)
+{
+ struct eventfd_ctx *trigger;
+
+ if (flags & VFIO_IRQ_SET_DATA_EVENTFD) {
+ int fd = *(int *)data;
+
+ trigger = eventfd_ctx_fdget(fd);
+ if (IS_ERR(trigger)) {
+ gvt_err("eventfd_ctx_fdget failed\n");
+ return PTR_ERR(trigger);
+ }
+ vgpu->vdev.msi_trigger = trigger;
+ }
+
+ return 0;
+}
+
+static int intel_vgpu_set_irqs(struct intel_vgpu *vgpu, uint32_t flags,
+ unsigned int index, unsigned int start, unsigned int count,
+ void *data)
+{
+ int (*func)(struct intel_vgpu *vgpu, unsigned int index,
+ unsigned int start, unsigned int count, uint32_t flags,
+ void *data) = NULL;
+
+ switch (index) {
+ case VFIO_PCI_INTX_IRQ_INDEX:
+ switch (flags & VFIO_IRQ_SET_ACTION_TYPE_MASK) {
+ case VFIO_IRQ_SET_ACTION_MASK:
+ func = intel_vgpu_set_intx_mask;
+ break;
+ case VFIO_IRQ_SET_ACTION_UNMASK:
+ func = intel_vgpu_set_intx_unmask;
+ break;
+ case VFIO_IRQ_SET_ACTION_TRIGGER:
+ func = intel_vgpu_set_intx_trigger;
+ break;
+ }
+ break;
+ case VFIO_PCI_MSI_IRQ_INDEX:
+ switch (flags & VFIO_IRQ_SET_ACTION_TYPE_MASK) {
+ case VFIO_IRQ_SET_ACTION_MASK:
+ case VFIO_IRQ_SET_ACTION_UNMASK:
+ /* XXX Need masking support exported */
+ break;
+ case VFIO_IRQ_SET_ACTION_TRIGGER:
+ func = intel_vgpu_set_msi_trigger;
+ break;
+ }
+ break;
+ }
+
+ if (!func)
+ return -ENOTTY;
+
+ return func(vgpu, index, start, count, flags, data);
+}
+
+static long intel_vgpu_ioctl(struct mdev_device *mdev, unsigned int cmd,
+ unsigned long arg)
+{
+ struct intel_vgpu *vgpu = mdev_get_drvdata(mdev);
+ unsigned long minsz;
+
+ gvt_dbg_core("vgpu%d ioctl, cmd: %d\n", vgpu->id, cmd);
+
+ if (cmd == VFIO_DEVICE_GET_INFO) {
+ struct vfio_device_info info;
+
+ minsz = offsetofend(struct vfio_device_info, num_irqs);
+
+ if (copy_from_user(&info, (void __user *)arg, minsz))
+ return -EFAULT;
+
+ if (info.argsz < minsz)
+ return -EINVAL;
+
+ info.flags = VFIO_DEVICE_FLAGS_PCI;
+ info.flags |= VFIO_DEVICE_FLAGS_RESET;
+ info.num_regions = VFIO_PCI_NUM_REGIONS;
+ info.num_irqs = VFIO_PCI_NUM_IRQS;
+
+ return copy_to_user((void __user *)arg, &info, minsz) ?
+ -EFAULT : 0;
+
+ } else if (cmd == VFIO_DEVICE_GET_REGION_INFO) {
+ struct vfio_region_info info;
+ struct vfio_info_cap caps = { .buf = NULL, .size = 0 };
+ int i, ret;
+ struct vfio_region_info_cap_sparse_mmap *sparse = NULL;
+ size_t size;
+ int nr_areas = 1;
+ int cap_type_id;
+
+ minsz = offsetofend(struct vfio_region_info, offset);
+
+ if (copy_from_user(&info, (void __user *)arg, minsz))
+ return -EFAULT;
+
+ if (info.argsz < minsz)
+ return -EINVAL;
+
+ switch (info.index) {
+ case VFIO_PCI_CONFIG_REGION_INDEX:
+ info.offset = VFIO_PCI_INDEX_TO_OFFSET(info.index);
+ info.size = INTEL_GVT_MAX_CFG_SPACE_SZ;
+ info.flags = VFIO_REGION_INFO_FLAG_READ |
+ VFIO_REGION_INFO_FLAG_WRITE;
+ break;
+ case VFIO_PCI_BAR0_REGION_INDEX:
+ info.offset = VFIO_PCI_INDEX_TO_OFFSET(info.index);
+ info.size = vgpu->cfg_space.bar[info.index].size;
+ if (!info.size) {
+ info.flags = 0;
+ break;
+ }
+
+ info.flags = VFIO_REGION_INFO_FLAG_READ |
+ VFIO_REGION_INFO_FLAG_WRITE;
+ break;
+ case VFIO_PCI_BAR1_REGION_INDEX:
+ info.offset = VFIO_PCI_INDEX_TO_OFFSET(info.index);
+ info.size = 0;
+ info.flags = 0;
+ break;
+ case VFIO_PCI_BAR2_REGION_INDEX:
+ info.offset = VFIO_PCI_INDEX_TO_OFFSET(info.index);
+ info.flags = VFIO_REGION_INFO_FLAG_CAPS |
+ VFIO_REGION_INFO_FLAG_MMAP |
+ VFIO_REGION_INFO_FLAG_READ |
+ VFIO_REGION_INFO_FLAG_WRITE;
+ info.size = gvt_aperture_sz(vgpu->gvt);
+
+ size = sizeof(*sparse) +
+ (nr_areas * sizeof(*sparse->areas));
+ sparse = kzalloc(size, GFP_KERNEL);
+ if (!sparse)
+ return -ENOMEM;
+
+ sparse->nr_areas = nr_areas;
+ cap_type_id = VFIO_REGION_INFO_CAP_SPARSE_MMAP;
+ sparse->areas[0].offset =
+ PAGE_ALIGN(vgpu_aperture_offset(vgpu));
+ sparse->areas[0].size = vgpu_aperture_sz(vgpu);
+ if (!caps.buf) {
+ kfree(caps.buf);
+ caps.buf = NULL;
+ caps.size = 0;
+ }
+ break;
+
+ case VFIO_PCI_BAR3_REGION_INDEX ... VFIO_PCI_BAR5_REGION_INDEX:
+ info.offset = VFIO_PCI_INDEX_TO_OFFSET(info.index);
+ info.size = 0;
+
+ info.flags = 0;
+ gvt_dbg_core("get region info bar:%d\n", info.index);
+ break;
+
+ case VFIO_PCI_ROM_REGION_INDEX:
+ case VFIO_PCI_VGA_REGION_INDEX:
+ gvt_dbg_core("get region info index:%d\n", info.index);
+ break;
+ default:
+ {
+ struct vfio_region_info_cap_type cap_type;
+
+ if (info.index >= VFIO_PCI_NUM_REGIONS +
+ vgpu->vdev.num_regions)
+ return -EINVAL;
+
+ i = info.index - VFIO_PCI_NUM_REGIONS;
+
+ info.offset =
+ VFIO_PCI_INDEX_TO_OFFSET(info.index);
+ info.size = vgpu->vdev.region[i].size;
+ info.flags = vgpu->vdev.region[i].flags;
+
+ cap_type.type = vgpu->vdev.region[i].type;
+ cap_type.subtype = vgpu->vdev.region[i].subtype;
+
+ ret = vfio_info_add_capability(&caps,
+ VFIO_REGION_INFO_CAP_TYPE,
+ &cap_type);
+ if (ret)
+ return ret;
+ }
+ }
+
+ if ((info.flags & VFIO_REGION_INFO_FLAG_CAPS) && sparse) {
+ switch (cap_type_id) {
+ case VFIO_REGION_INFO_CAP_SPARSE_MMAP:
+ ret = vfio_info_add_capability(&caps,
+ VFIO_REGION_INFO_CAP_SPARSE_MMAP,
+ sparse);
+ kfree(sparse);
+ if (ret)
+ return ret;
+ break;
+ default:
+ return -EINVAL;
+ }
+ }
+
+ if (caps.size) {
+ if (info.argsz < sizeof(info) + caps.size) {
+ info.argsz = sizeof(info) + caps.size;
+ info.cap_offset = 0;
+ } else {
+ vfio_info_cap_shift(&caps, sizeof(info));
+ if (copy_to_user((void __user *)arg +
+ sizeof(info), caps.buf,
+ caps.size)) {
+ kfree(caps.buf);
+ return -EFAULT;
+ }
+ info.cap_offset = sizeof(info);
+ }
+
+ kfree(caps.buf);
+ }
+
+ return copy_to_user((void __user *)arg, &info, minsz) ?
+ -EFAULT : 0;
+ } else if (cmd == VFIO_DEVICE_GET_IRQ_INFO) {
+ struct vfio_irq_info info;
+
+ minsz = offsetofend(struct vfio_irq_info, count);
+
+ if (copy_from_user(&info, (void __user *)arg, minsz))
+ return -EFAULT;
+
+ if (info.argsz < minsz || info.index >= VFIO_PCI_NUM_IRQS)
+ return -EINVAL;
+
+ switch (info.index) {
+ case VFIO_PCI_INTX_IRQ_INDEX:
+ case VFIO_PCI_MSI_IRQ_INDEX:
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ info.flags = VFIO_IRQ_INFO_EVENTFD;
+
+ info.count = intel_vgpu_get_irq_count(vgpu, info.index);
+
+ if (info.index == VFIO_PCI_INTX_IRQ_INDEX)
+ info.flags |= (VFIO_IRQ_INFO_MASKABLE |
+ VFIO_IRQ_INFO_AUTOMASKED);
+ else
+ info.flags |= VFIO_IRQ_INFO_NORESIZE;
+
+ return copy_to_user((void __user *)arg, &info, minsz) ?
+ -EFAULT : 0;
+ } else if (cmd == VFIO_DEVICE_SET_IRQS) {
+ struct vfio_irq_set hdr;
+ u8 *data = NULL;
+ int ret = 0;
+ size_t data_size = 0;
+
+ minsz = offsetofend(struct vfio_irq_set, count);
+
+ if (copy_from_user(&hdr, (void __user *)arg, minsz))
+ return -EFAULT;
+
+ if (!(hdr.flags & VFIO_IRQ_SET_DATA_NONE)) {
+ int max = intel_vgpu_get_irq_count(vgpu, hdr.index);
+
+ ret = vfio_set_irqs_validate_and_prepare(&hdr, max,
+ VFIO_PCI_NUM_IRQS, &data_size);
+ if (ret) {
+ gvt_err("intel:vfio_set_irqs_validate_and_prepare failed\n");
+ return -EINVAL;
+ }
+ if (data_size) {
+ data = memdup_user((void __user *)(arg + minsz),
+ data_size);
+ if (IS_ERR(data))
+ return PTR_ERR(data);
+ }
+ }
+
+ ret = intel_vgpu_set_irqs(vgpu, hdr.flags, hdr.index,
+ hdr.start, hdr.count, data);
+ kfree(data);
+
+ return ret;
+ } else if (cmd == VFIO_DEVICE_RESET) {
+ intel_gvt_ops->vgpu_reset(vgpu);
+ return 0;
+ }
+
+ return 0;
+}
+
+static const struct parent_ops intel_vgpu_ops = {
+ .supported_type_groups = intel_vgpu_type_groups,
+ .create = intel_vgpu_create,
+ .remove = intel_vgpu_remove,
+
+ .open = intel_vgpu_open,
+ .release = intel_vgpu_release,
+
+ .read = intel_vgpu_read,
+ .write = intel_vgpu_write,
+ .mmap = intel_vgpu_mmap,
+ .ioctl = intel_vgpu_ioctl,
+};
+
static int kvmgt_host_init(struct device *dev, void *gvt, const void *ops)
{
if (!intel_gvt_init_vgpu_type_groups(gvt))
@@ -350,22 +1110,28 @@
intel_gvt_ops = ops;
- /* MDEV is not yet available */
- return -ENODEV;
+ return mdev_register_device(dev, &intel_vgpu_ops);
}
static void kvmgt_host_exit(struct device *dev, void *gvt)
{
intel_gvt_cleanup_vgpu_type_groups(gvt);
+ mdev_unregister_device(dev);
}
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 kvmgt_guest_info *info;
+ struct kvm *kvm;
struct kvm_memory_slot *slot;
int idx;
+ if (!handle_valid(handle))
+ return -ESRCH;
+
+ info = (struct kvmgt_guest_info *)handle;
+ kvm = info->kvm;
+
idx = srcu_read_lock(&kvm->srcu);
slot = gfn_to_memslot(kvm, gfn);
@@ -385,11 +1151,17 @@
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 kvmgt_guest_info *info;
+ struct kvm *kvm;
struct kvm_memory_slot *slot;
int idx;
+ if (!handle_valid(handle))
+ return 0;
+
+ info = (struct kvmgt_guest_info *)handle;
+ kvm = info->kvm;
+
idx = srcu_read_lock(&kvm->srcu);
slot = gfn_to_memslot(kvm, gfn);
@@ -477,6 +1249,85 @@
return kvmgt_check_guest() ? -ENODEV : 0;
}
+static bool __kvmgt_vgpu_exist(struct intel_vgpu *vgpu, struct kvm *kvm)
+{
+ struct intel_vgpu *itr;
+ struct kvmgt_guest_info *info;
+ int id;
+ bool ret = false;
+
+ mutex_lock(&vgpu->gvt->lock);
+ for_each_active_vgpu(vgpu->gvt, itr, id) {
+ if (!handle_valid(itr->handle))
+ continue;
+
+ info = (struct kvmgt_guest_info *)itr->handle;
+ if (kvm && kvm == info->kvm) {
+ ret = true;
+ goto out;
+ }
+ }
+out:
+ mutex_unlock(&vgpu->gvt->lock);
+ return ret;
+}
+
+static int kvmgt_guest_init(struct mdev_device *mdev)
+{
+ struct kvmgt_guest_info *info;
+ struct intel_vgpu *vgpu;
+ struct kvm *kvm;
+
+ vgpu = mdev_get_drvdata(mdev);
+ if (handle_valid(vgpu->handle))
+ return -EEXIST;
+
+ kvm = vgpu->vdev.kvm;
+ if (!kvm || kvm->mm != current->mm) {
+ gvt_err("KVM is required to use Intel vGPU\n");
+ return -ESRCH;
+ }
+
+ if (__kvmgt_vgpu_exist(vgpu, kvm))
+ return -EEXIST;
+
+ info = vzalloc(sizeof(struct kvmgt_guest_info));
+ if (!info)
+ return -ENOMEM;
+
+ vgpu->handle = (unsigned long)info;
+ info->vgpu = vgpu;
+ info->kvm = kvm;
+
+ kvmgt_protect_table_init(info);
+ gvt_cache_init(vgpu);
+
+ info->track_node.track_write = kvmgt_page_track_write;
+ info->track_node.track_flush_slot = kvmgt_page_track_flush_slot;
+ kvm_page_track_register_notifier(kvm, &info->track_node);
+
+ return 0;
+}
+
+static bool kvmgt_guest_exit(struct kvmgt_guest_info *info)
+{
+ struct intel_vgpu *vgpu;
+
+ if (!info) {
+ gvt_err("kvmgt_guest_info invalid\n");
+ return false;
+ }
+
+ vgpu = info->vgpu;
+
+ kvm_page_track_unregister_notifier(info->kvm, &info->track_node);
+ kvmgt_protect_table_destroy(info);
+ gvt_cache_destroy(vgpu);
+ vfree(info);
+
+ return true;
+}
+
static int kvmgt_attach_vgpu(void *vgpu, unsigned long *handle)
{
/* nothing to do here */
@@ -490,30 +1341,42 @@
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;
+ struct kvmgt_guest_info *info;
+ struct intel_vgpu *vgpu;
- if (vgpu->vdev.msi_trigger)
- return eventfd_signal(vgpu->vdev.msi_trigger, 1) == 1;
+ if (!handle_valid(handle))
+ return -ESRCH;
- return false;
+ info = (struct kvmgt_guest_info *)handle;
+ vgpu = info->vgpu;
+
+ if (eventfd_signal(vgpu->vdev.msi_trigger, 1) == 1)
+ return 0;
+
+ return -EFAULT;
}
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;
+ struct kvmgt_guest_info *info;
+ struct device *dev;
int rc;
+ if (!handle_valid(handle))
+ return INTEL_GVT_INVALID_ADDR;
+
+ info = (struct kvmgt_guest_info *)handle;
pfn = gvt_cache_find(info->vgpu, gfn);
if (pfn != 0)
return pfn;
- rc = kvmgt_pin_pages(info->vgpu->vdev.mdev, &gfn, 1,
- IOMMU_READ | IOMMU_WRITE, &pfn);
+ pfn = INTEL_GVT_INVALID_ADDR;
+ dev = &info->vgpu->vdev.mdev->dev;
+ rc = vfio_pin_pages(dev, &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_err("vfio_pin_pages failed for gfn 0x%lx: %d\n", gfn, rc);
+ return INTEL_GVT_INVALID_ADDR;
}
gvt_cache_add(info->vgpu, gfn, pfn);
@@ -528,6 +1391,9 @@
int ret;
bool kthread = current->mm == NULL;
+ if (!handle_valid(handle))
+ return -ESRCH;
+
info = (struct kvmgt_guest_info *)handle;
kvm = info->kvm;