> From: Alex Williamson <alex.william...@redhat.com>
> Sent: Friday, April 28, 2023 5:55 AM
>
> On Wed, 26 Apr 2023 07:54:19 -0700
> Yi Liu <yi.l....@intel.com> wrote:
>
> > This is the way user to invoke hot-reset for the devices opened by cdev
> > interface. User should check the flag VFIO_PCI_HOT_RESET_FLAG_RESETTABLE
> > in the output of VFIO_DEVICE_GET_PCI_HOT_RESET_INFO ioctl before doing
> > hot-reset for cdev devices.
> >
> > Suggested-by: Jason Gunthorpe <j...@nvidia.com>
> > Signed-off-by: Jason Gunthorpe <j...@nvidia.com>
> > Reviewed-by: Jason Gunthorpe <j...@nvidia.com>
> > Tested-by: Yanting Jiang <yanting.ji...@intel.com>
> > Signed-off-by: Yi Liu <yi.l....@intel.com>
> > ---
> > drivers/vfio/pci/vfio_pci_core.c | 66 +++++++++++++++++++++++++++-----
> > include/uapi/linux/vfio.h | 22 +++++++++++
> > 2 files changed, 79 insertions(+), 9 deletions(-)
> >
> > diff --git a/drivers/vfio/pci/vfio_pci_core.c
> > b/drivers/vfio/pci/vfio_pci_core.c
> > index 43858d471447..f70e3b948b16 100644
> > --- a/drivers/vfio/pci/vfio_pci_core.c
> > +++ b/drivers/vfio/pci/vfio_pci_core.c
> > @@ -180,7 +180,8 @@ static void vfio_pci_probe_mmaps(struct
> > vfio_pci_core_device
> *vdev)
> > struct vfio_pci_group_info;
> > static void vfio_pci_dev_set_try_reset(struct vfio_device_set *dev_set);
> > static int vfio_pci_dev_set_hot_reset(struct vfio_device_set *dev_set,
> > - struct vfio_pci_group_info *groups);
> > + struct vfio_pci_group_info *groups,
> > + struct iommufd_ctx *iommufd_ctx);
> >
> > /*
> > * INTx masking requires the ability to disable INTx signaling via
> > PCI_COMMAND
> > @@ -1364,8 +1365,7 @@ vfio_pci_ioctl_pci_hot_reset_groups(struct
> vfio_pci_core_device *vdev,
> > if (ret)
> > return ret;
> >
> > - /* Somewhere between 1 and count is OK */
> > - if (!array_count || array_count > count)
> > + if (array_count > count)
> > return -EINVAL;
>
> Doesn't this need a || vfio_device_cdev_opened(vdev) test as well?
> It's invalid to pass fds for a cdev device. Presumably it would fail
> later collecting group fds as well, but might as well enforce the
> semantics early.
Yes, it is.
>
> >
> > group_fds = kcalloc(array_count, sizeof(*group_fds), GFP_KERNEL);
> > @@ -1414,7 +1414,7 @@ vfio_pci_ioctl_pci_hot_reset_groups(struct
> vfio_pci_core_device *vdev,
> > info.count = array_count;
> > info.files = files;
> >
> > - ret = vfio_pci_dev_set_hot_reset(vdev->vdev.dev_set, &info);
> > + ret = vfio_pci_dev_set_hot_reset(vdev->vdev.dev_set, &info, NULL);
> >
> > hot_reset_release:
> > for (file_idx--; file_idx >= 0; file_idx--)
> > @@ -1429,6 +1429,7 @@ static int vfio_pci_ioctl_pci_hot_reset(struct
> vfio_pci_core_device *vdev,
> > {
> > unsigned long minsz = offsetofend(struct vfio_pci_hot_reset, count);
> > struct vfio_pci_hot_reset hdr;
> > + struct iommufd_ctx *iommufd;
> > bool slot = false;
> >
> > if (copy_from_user(&hdr, arg, minsz))
> > @@ -1443,7 +1444,12 @@ static int vfio_pci_ioctl_pci_hot_reset(struct
> vfio_pci_core_device *vdev,
> > else if (pci_probe_reset_bus(vdev->pdev->bus))
> > return -ENODEV;
> >
> > - return vfio_pci_ioctl_pci_hot_reset_groups(vdev, hdr.count, slot, arg);
> > + if (hdr.count)
> > + return vfio_pci_ioctl_pci_hot_reset_groups(vdev, hdr.count,
> > slot, arg);
> > +
> > + iommufd = vfio_iommufd_physical_ictx(&vdev->vdev);
> > +
> > + return vfio_pci_dev_set_hot_reset(vdev->vdev.dev_set, NULL, iommufd);
>
> Why did we need to store iommufd in a variable?
will remove it.
> > }
> >
> > static int vfio_pci_ioctl_ioeventfd(struct vfio_pci_core_device *vdev,
> > @@ -2415,6 +2421,9 @@ static bool vfio_dev_in_groups(struct
> > vfio_pci_core_device
> *vdev,
> > {
> > unsigned int i;
> >
> > + if (!groups)
> > + return false;
> > +
> > for (i = 0; i < groups->count; i++)
> > if (vfio_file_has_dev(groups->files[i], &vdev->vdev))
> > return true;
> > @@ -2488,13 +2497,38 @@ static int vfio_pci_dev_set_pm_runtime_get(struct
> vfio_device_set *dev_set)
> > return ret;
> > }
> >
> > +static bool vfio_dev_in_iommufd_ctx(struct vfio_pci_core_device *vdev,
> > + struct iommufd_ctx *iommufd_ctx)
> > +{
> > + struct iommufd_ctx *iommufd = vfio_iommufd_physical_ictx(&vdev->vdev);
> > + struct iommu_group *iommu_group;
> > +
> > + if (!iommufd_ctx)
> > + return false;
> > +
> > + if (iommufd == iommufd_ctx)
> > + return true;
> > +
> > + iommu_group = iommu_group_get(vdev->vdev.dev);
> > + if (!iommu_group)
> > + return false;
> > +
> > + /*
> > + * Try to check if any device within iommu_group is bound with
> > + * the input iommufd_ctx.
> > + */
> > + return vfio_devset_iommufd_has_group(vdev->vdev.dev_set,
> > + iommufd_ctx, iommu_group);
> > +}
>
> This last test makes this not do what the function name suggests it
> does. If it were true, the device is not in the iommufd_ctx, it simply
> cannot be within another iommu ctx.
Yes. it actually means not possible to be in another iommufd_ctx.
>
> > +
> > /*
> > * We need to get memory_lock for each device, but devices can share
> > mmap_lock,
> > * therefore we need to zap and hold the vma_lock for each device, and
> > only then
> > * get each memory_lock.
> > */
> > static int vfio_pci_dev_set_hot_reset(struct vfio_device_set *dev_set,
> > - struct vfio_pci_group_info *groups)
> > + struct vfio_pci_group_info *groups,
> > + struct iommufd_ctx *iommufd_ctx)
> > {
> > struct vfio_pci_core_device *cur_mem;
> > struct vfio_pci_core_device *cur_vma;
> > @@ -2525,10 +2559,24 @@ static int vfio_pci_dev_set_hot_reset(struct
> vfio_device_set *dev_set,
> >
> > list_for_each_entry(cur_vma, &dev_set->device_list, vdev.dev_set_list) {
> > /*
> > - * Test whether all the affected devices are contained by the
> > - * set of groups provided by the user.
> > + * Test whether all the affected devices can be reset by the
> > + * user.
> > + *
> > + * If user provides a set of groups, all the opened devices
> > + * in the dev_set should be contained by the set of groups
> > + * provided by the user.
> > + *
> > + * If user provides a zero-length group fd array, then all
> > + * the affected devices must be bound to same iommufd_ctx as
> > + * the input iommufd_ctx. If there is device that has not
> > + * been bound to iommufd_ctx yet, shall check if there is any
> > + * device within its iommu_group that has been bound to the
> > + * input iommufd_ctx.
> > + *
> > + * Otherwise, reset is not allowed.
> > */
> > - if (!vfio_dev_in_groups(cur_vma, groups)) {
> > + if (!vfio_dev_in_groups(cur_vma, groups) &&
> > + !vfio_dev_in_iommufd_ctx(cur_vma, iommufd_ctx)) {
>
>
> Rather than mangling vfio_dev_in_groups() and inventing
> vfio_dev_in_iommufd_ctx() that doesn't do what it implies, how about:
>
> bool vfio_device_owned(struct vfio_device *vdev,
> struct vfio_pci_group_info *groups,
> struct iommufd_ctx *iommufd_ctx)
> {
> struct iommu_group *group;
>
> WARN_ON(!!groups == !!iommufd_ctx);
>
> if (groups)
> return vfio_dev_in_groups(vdev, groups));
>
> if (vfio_iommufd_physical_ictx(vdev) == iommufd_ctx)
> return true;
>
> group = iommu_group_get(vdev->dev);
> if (group)
> return vfio_devset_iommufd_has_group(vdev->vdev.dev_set,
> iommufd_ctx, group);
> return false;
> }
Will follow above suggestion.
> Seems like such a function would live in vfio_main.c
It may require to make the struct vfio_pci_group_info visible outside
of vfio-pci. This seems to be strange to make vfio_main.c to refer pci
specific structure.
>
> > ret = -EINVAL;
> > goto err_undo;
> > }
> > diff --git a/include/uapi/linux/vfio.h b/include/uapi/linux/vfio.h
> > index 4b4e2c28984b..1241d02d8701 100644
> > --- a/include/uapi/linux/vfio.h
> > +++ b/include/uapi/linux/vfio.h
> > @@ -710,6 +710,28 @@ struct vfio_pci_hot_reset_info {
> > * VFIO_DEVICE_PCI_HOT_RESET - _IOW(VFIO_TYPE, VFIO_BASE + 13,
> > * struct vfio_pci_hot_reset)
> > *
> > + * Userspace requests hot reset for the devices it operates. Due to the
> > + * underlying topology, multiple devices can be affected in the reset
> > + * while some might be opened by another user. To avoid interference
> > + * the calling user must ensure all affected devices are owned by itself.
> > + * The ownership proof needs to refer the output of
> > + * VFIO_DEVICE_GET_PCI_HOT_RESET_INFO. Ownership can be proved as:
> > + *
> > + * 1) An array of group fds - This is used for the devices opened via
> > + * the group/container interface.
> > + * 2) A zero-length array - This is used for the devices opened via
> > + * the cdev interface. User should check the
> > + * flag VFIO_PCI_HOT_RESET_FLAG_IOMMUFD_DEV_ID
> > + * and flag VFIO_PCI_HOT_RESET_FLAG_RESETTABLE
> > + * before using this method.
> > + *
> > + * In case a non void group fd array is passed, the devices affected by
> > + * the reset must belong to those opened VFIO groups. In case a zero
> > + * length array is passed, the other devices affected by the reset, if
> > + * any, must be either bound to the same iommufd as this VFIO device or
> > + * in the same iommu_group with a device that does. Either of the two
> > + * methods is applied to check the feasibility of the hot reset.
>
> This should probably just refer to the concept of ownership described
> in the INFO ioctl and clarify that cdev opened device must exclusively
> provide an empty array and group opened devices must exclusively use an
> array of group fds for proof of ownership. Mixed access to devices
> between cdev and legacy groups are not supported by this interface.
> Thanks,
Sure. Will make it in next version.
Regards,
Yi Liu
>
> Alex
>
> > + *
> > * Return: 0 on success, -errno on failure.
> > */
> > struct vfio_pci_hot_reset {
