RE: [RFC v1 4/4] vfio/type1: bind guest pasid (guest page tables) to host

["Liu, Yi L" <yi.l.liu@intel.com>]

> From: Auger Eric [mailto:eric.auger@redhat.com]
> Sent: Thursday, July 18, 2019 6:17 PM
> To: Liu, Yi L <yi.l.liu@intel.com>; alex.williamson@redhat.com
> Subject: Re: [RFC v1 4/4] vfio/type1: bind guest pasid (guest page tables) to host
> 
> Yi,
> 
> On 7/5/19 1:06 PM, Liu, Yi L wrote:
> > From: Liu Yi L <yi.l.liu@intel.com>
> >
> > This patch adds vfio support to bind guest translation structure to
> > host iommu. VFIO exposes iommu programming capability to user- space.
> > Guest is a user-space application in host under KVM solution.
> > For SVA usage in Virtual Machine, guest owns GVA->GPA translation
> > structure. And this part should be passdown to host to enable nested
> > translation (or say two stage translation). This patch reuses the
> > VFIO_IOMMU_BIND proposal from Jean-Philippe Brucker, and adds new bind
> > type for binding guest owned translation structure to host.
> >
> > *) Add two new ioctls for VFIO containers.
> >
> >   - VFIO_IOMMU_BIND: for bind request from userspace, it could be
> >                    bind a process to a pasid or bind a guest pasid
> >                    to a device, this is indicated by type
> >   - VFIO_IOMMU_UNBIND: for unbind request from userspace, it could be
> >                    unbind a process to a pasid or unbind a guest pasid
> >                    to a device, also indicated by type
> >   - Bind type:
> > 	VFIO_IOMMU_BIND_PROCESS: user-space request to bind a process
> >                    to a device
> > 	VFIO_IOMMU_BIND_GUEST_PASID: bind guest owned translation
> >                    structure to host iommu. e.g. guest page table
> You may add that only VFIO_IOMMU_BIND_GUEST_PASID gets implemented in this
> patch

Good catch:-).

> >
> > *) Code logic in vfio_iommu_type1_ioctl() to handle
> > VFIO_IOMMU_BIND/UNBIND
> >
> > Cc: Kevin Tian <kevin.tian@intel.com>
> > Signed-off-by: Jean-Philippe Brucker <jean-philippe.brucker@arm.com>
> > Signed-off-by: Liu Yi L <yi.l.liu@intel.com>
> > Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
> > ---
> >  drivers/vfio/vfio_iommu_type1.c | 151
> ++++++++++++++++++++++++++++++++++++++++
> >  include/uapi/linux/vfio.h       |  56 +++++++++++++++
> >  2 files changed, 207 insertions(+)
> >
> > diff --git a/drivers/vfio/vfio_iommu_type1.c
> > b/drivers/vfio/vfio_iommu_type1.c index d5e0c01..57826ed 100644
> > --- a/drivers/vfio/vfio_iommu_type1.c
> > +++ b/drivers/vfio/vfio_iommu_type1.c
> > @@ -1920,6 +1920,119 @@ static int vfio_iommu_type1_pasid_free(struct
> vfio_iommu *iommu, int pasid)
> >  	return ret;
> >  }
> >
> > +static int vfio_bind_gpasid_fn(struct device *dev, void *data) {
> > +	struct iommu_domain *domain = iommu_get_domain_for_dev(dev);
> > +	struct vfio_iommu_type1_bind_guest_pasid *guest_bind = data;
> > +
> > +	return iommu_sva_bind_gpasid(domain, dev, &guest_bind->bind_data); }
> > +
> > +static int vfio_unbind_gpasid_fn(struct device *dev, void *data) {
> > +	struct iommu_domain *domain = iommu_get_domain_for_dev(dev);
> > +	struct vfio_iommu_type1_bind_guest_pasid *guest_bind = data;
> > +
> > +	return iommu_sva_unbind_gpasid(domain, dev,
> > +					guest_bind->bind_data.hpasid);
> > +}
> > +
> > +/*
> > + * unbind specific gpasid, caller of this function requires hold
> > + * vfio_iommu->lock
> > + */
> > +static long vfio_iommu_type1_do_guest_unbind(struct vfio_iommu *iommu,
> > +		  struct vfio_iommu_type1_bind_guest_pasid *guest_bind) {
> > +	struct vfio_domain *domain;
> > +	struct vfio_group *group;
> > +	int ret = 0;
> > +
> > +	if (!IS_IOMMU_CAP_DOMAIN_IN_CONTAINER(iommu)) {
> > +		ret = -EINVAL;
> > +		goto out;
> > +	}
> > +
> > +	list_for_each_entry(domain, &iommu->domain_list, next) {
> > +		list_for_each_entry(group, &domain->group_list, next) {
> > +			ret = iommu_group_for_each_dev(group->iommu_group,
> > +			   guest_bind, vfio_unbind_gpasid_fn);
> can it fail individually, in which case we end up with something half unset or it is safe?
> A comment may be worth.

thanks, good suggestion. Actually, we have an assumption that for devices which
belong to non-singleton group, we should not enable PASID capability. So may not
fail individually. But yes, a comment would be needed here.

> > +			if (ret)
> > +				goto out;
> > +		}
> > +	}
> you may use vfio_iommu_lookup_dev() introduced in [RFC v1 2/4] vfio:
> VFIO_IOMMU_CACHE_INVALIDATE

yes, let me do it next version. :-)

> > +
> > +	return 0;
> not needed
> > +
> > +out:
> > +	return ret;
> > +}
> > +
> > +static long vfio_iommu_type1_bind_gpasid(struct vfio_iommu *iommu,
> > +					    void __user *arg,
> > +					    struct vfio_iommu_type1_bind *bind) {
> > +	struct vfio_iommu_type1_bind_guest_pasid guest_bind;
> > +	struct vfio_domain *domain;
> > +	struct vfio_group *group;
> > +	unsigned long minsz;
> > +	int ret = 0;
> > +
> > +	minsz = sizeof(*bind) + sizeof(guest_bind);
> > +	if (bind->argsz < minsz)
> > +		return -EINVAL;
> > +
> > +	if (copy_from_user(&guest_bind, arg, sizeof(guest_bind)))
> > +		return -EFAULT;
> > +
> > +	mutex_lock(&iommu->lock);
> > +	if (!IS_IOMMU_CAP_DOMAIN_IN_CONTAINER(iommu)) {
> > +		ret = -EINVAL;
> > +		goto out_unlock;
> > +	}
> > +
> > +	list_for_each_entry(domain, &iommu->domain_list, next) {
> > +		list_for_each_entry(group, &domain->group_list, next) {
> > +			ret = iommu_group_for_each_dev(group->iommu_group,
> > +			   &guest_bind, vfio_bind_gpasid_fn);
> > +			if (ret)
> > +				goto out_unbind;
> use vfio_iommu_lookup_dev

got it. ~

> > +		}
> 
> > +	}
> > +
> > +	mutex_unlock(&iommu->lock);
> > +	return 0;
> > +
> > +out_unbind:
> > +	/* Undo all binds that already succeeded */
> > +	vfio_iommu_type1_do_guest_unbind(iommu, &guest_bind);
> > +
> > +out_unlock:
> > +	mutex_unlock(&iommu->lock);
> > +	return ret;
> > +}
> > +
> > +static long vfio_iommu_type1_unbind_gpasid(struct vfio_iommu *iommu,
> > +					    void __user *arg,
> > +					    struct vfio_iommu_type1_bind *bind) {
> > +	struct vfio_iommu_type1_bind_guest_pasid guest_bind;
> > +	unsigned long minsz;
> > +	int ret = 0;
> nit: init not needed

yeah, nice catch.

> > +
> > +	minsz = sizeof(*bind) + sizeof(guest_bind);
> > +	if (bind->argsz < minsz)
> > +		return -EINVAL;
> > +
> > +	if (copy_from_user(&guest_bind, arg, sizeof(guest_bind)))
> > +		return -EFAULT;
> > +
> > +	mutex_lock(&iommu->lock);
> > +	ret = vfio_iommu_type1_do_guest_unbind(iommu, &guest_bind);
> > +	mutex_unlock(&iommu->lock);
> > +	return ret;
> > +}
> > +
> >  static long vfio_iommu_type1_ioctl(void *iommu_data,
> >  				   unsigned int cmd, unsigned long arg)  { @@ -
> 2061,6 +2174,44 @@
> > static long vfio_iommu_type1_ioctl(void *iommu_data,
> >  		default:
> >  			return -EINVAL;
> >  		}
> > +
> > +	} else if (cmd == VFIO_IOMMU_BIND) {
> > +		struct vfio_iommu_type1_bind bind;
> > +
> > +		minsz = offsetofend(struct vfio_iommu_type1_bind, bind_type);
> > +
> > +		if (copy_from_user(&bind, (void __user *)arg, minsz))
> > +			return -EFAULT;
> > +
> > +		if (bind.argsz < minsz)
> > +			return -EINVAL;
> > +
> > +		switch (bind.bind_type) {
> > +		case VFIO_IOMMU_BIND_GUEST_PASID:
> > +			return vfio_iommu_type1_bind_gpasid(iommu,
> > +					(void __user *)(arg + minsz), &bind);
> > +		default:
> > +			return -EINVAL;
> > +		}
> > +
> > +	} else if (cmd == VFIO_IOMMU_UNBIND) {
> > +		struct vfio_iommu_type1_bind bind;
> > +
> > +		minsz = offsetofend(struct vfio_iommu_type1_bind, bind_type);
> > +
> > +		if (copy_from_user(&bind, (void __user *)arg, minsz))
> > +			return -EFAULT;
> > +
> > +		if (bind.argsz < minsz)
> > +			return -EINVAL;
> > +
> > +		switch (bind.bind_type) {
> > +		case VFIO_IOMMU_BIND_GUEST_PASID:
> > +			return vfio_iommu_type1_unbind_gpasid(iommu,
> > +					(void __user *)(arg + minsz), &bind);
> > +		default:
> > +			return -EINVAL;
> > +		}
> >  	}
> >
> >  	return -ENOTTY;
> > diff --git a/include/uapi/linux/vfio.h b/include/uapi/linux/vfio.h
> > index af03c9f..4167bbd 100644
> > --- a/include/uapi/linux/vfio.h
> > +++ b/include/uapi/linux/vfio.h
> > @@ -823,6 +823,62 @@ struct vfio_iommu_type1_pasid_request {
> >   */
> >  #define VFIO_IOMMU_PASID_REQUEST	_IO(VFIO_TYPE, VFIO_BASE + 27)
> >
> > +/*
> > + * In guest use of SVA, the first level page tables is managed by the guest.
> > + * we can either bind guest PASID table or explicitly bind a PASID
> > +with guest
> > + * page table.
> > + */
> > +struct vfio_iommu_type1_bind_guest_pasid {
> > +	struct gpasid_bind_data bind_data;
> > +};
> do you need this encapsulation? Why not directly using iommu.h uapi struct?

yeah, at first I want to cover guest pasid table bind. but now, I agree this may
be not necessary. :-)

> > +
> > +enum vfio_iommu_bind_type {
> > +	VFIO_IOMMU_BIND_PROCESS,
> > +	VFIO_IOMMU_BIND_GUEST_PASID,
> > +};
> > +
> > +/*
> > + * Supported types:
> > + *     - VFIO_IOMMU_BIND_PROCESS: bind native process, which takes
> > + *                      vfio_iommu_type1_bind_process in data.
> > + *     - VFIO_IOMMU_BIND_GUEST_PASID: bind guest pasid, which invoked
> > + *                      by guest process binding, it takes
> > + *                      vfio_iommu_type1_bind_guest_pasid in data.
> > + */
> > +struct vfio_iommu_type1_bind {
> > +	__u32				argsz;
> > +	enum vfio_iommu_bind_type	bind_type;
> The rest of the API does not use enum directly in structs. __u8/__u32?

If using __u8/__u32, it would introduce some bit flags. e.g. bit 0 for bind_gpasid,
bit 1 for bind process. If so, it may be possible that user space pass both bit 0
and bit 1 as set. Then it would be a trouble. Enum may avoid it. Not sure if such
case is critical. If not, I'm willing to use __u8/__u32. :-)

Thanks very much for your review, Eric. Let me address the comments in next
version.

Regards,
Yi Liu

> > +	__u8				data[];
> > +};
> > +
> > +/*
> > + * VFIO_IOMMU_BIND - _IOWR(VFIO_TYPE, VFIO_BASE + 28, struct
> > +vfio_iommu_bind)
> > + *
> > + * Manage address spaces of devices in this container. Initially a
> > +TYPE1
> > + * container can only have one address space, managed with
> > + * VFIO_IOMMU_MAP/UNMAP_DMA.
> > + *
> > + * An IOMMU of type VFIO_TYPE1_NESTING_IOMMU can be managed by both
> > +MAP/UNMAP
> > + * and BIND ioctls at the same time. MAP/UNMAP acts on the stage-2
> > +(host) page
> > + * tables, and BIND manages the stage-1 (guest) page tables. Other
> > +types of
> > + * IOMMU may allow MAP/UNMAP and BIND to coexist, where MAP/UNMAP
> > +controls
> > + * non-PASID traffic and BIND controls PASID traffic. But this
> > +depends on the
> > + * underlying IOMMU architecture and isn't guaranteed.
> > + *
> > + * Availability of this feature depends on the device, its bus, the
> > +underlying
> > + * IOMMU and the CPU architecture.
> > + *
> > + * returns: 0 on success, -errno on failure.
> > + */
> > +#define VFIO_IOMMU_BIND		_IO(VFIO_TYPE, VFIO_BASE + 28)
> > +
> > +/*
> > + * VFIO_IOMMU_UNBIND - _IOWR(VFIO_TYPE, VFIO_BASE + 29, struct
> > +vfio_iommu_bind)
> > + *
> > + * Undo what was done by the corresponding VFIO_IOMMU_BIND ioctl.
> > + */
> > +#define VFIO_IOMMU_UNBIND	_IO(VFIO_TYPE, VFIO_BASE + 29)
> > +
> >  /* -------- Additional API for SPAPR TCE (Server POWERPC) IOMMU
> > -------- */
> >
> >  /*
> >
> 
> Thanks
> 
> Eric
_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu