[v6,07/10] iommufd: Add a nested HW pagetable object

Message ID	20231024150609.46884-8-yi.l.liu@intel.com (mailing list archive)
State	New, archived
Headers	show Return-Path: <kvm-owner@vger.kernel.org> From: Yi Liu <yi.l.liu@intel.com> To: joro@8bytes.org, alex.williamson@redhat.com, jgg@nvidia.com, kevin.tian@intel.com, robin.murphy@arm.com, baolu.lu@linux.intel.com Cc: cohuck@redhat.com, eric.auger@redhat.com, nicolinc@nvidia.com, kvm@vger.kernel.org, mjrosato@linux.ibm.com, chao.p.peng@linux.intel.com, yi.l.liu@intel.com, yi.y.sun@linux.intel.com, peterx@redhat.com, jasowang@redhat.com, shameerali.kolothum.thodi@huawei.com, lulu@redhat.com, suravee.suthikulpanit@amd.com, iommu@lists.linux.dev, linux-kernel@vger.kernel.org, linux-kselftest@vger.kernel.org, zhenzhong.duan@intel.com, joao.m.martins@oracle.com Subject: [PATCH v6 07/10] iommufd: Add a nested HW pagetable object Date: Tue, 24 Oct 2023 08:06:06 -0700 Message-Id: <20231024150609.46884-8-yi.l.liu@intel.com> In-Reply-To: <20231024150609.46884-1-yi.l.liu@intel.com> References: <20231024150609.46884-1-yi.l.liu@intel.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Precedence: bulk
Series	iommufd: Add nesting infrastructure (part 1/2) \| expand [v6,00/10] iommufd: Add nesting infrastructure (part 1/2) [v6,01/10] iommu: Add IOMMU_DOMAIN_NESTED [v6,02/10] iommu: Pass in parent domain with user_data to domain_alloc_user op [v6,03/10] iommufd: Rename IOMMUFD_OBJ_HW_PAGETABLE to IOMMUFD_OBJ_HWPT_PAGING [v6,04/10] iommufd/device: Wrap IOMMUFD_OBJ_HWPT_PAGING-only configurations [v6,05/10] iommufd: Derive iommufd_hwpt_paging from iommufd_hw_pagetable [v6,06/10] iommufd: Share iommufd_hwpt_alloc with IOMMUFD_OBJ_HWPT_NESTED [v6,07/10] iommufd: Add a nested HW pagetable object [v6,08/10] iommu: Add iommu_copy_struct_from_user helper [v6,09/10] iommufd/selftest: Add nested domain allocation for mock domain [v6,10/10] iommufd/selftest: Add coverage for IOMMU_HWPT_ALLOC with nested HWPTs

diff --git a/drivers/iommu/iommufd/device.c b/drivers/iommu/iommufd/device.c index 902288ca910d..6dbd58a718e4 100644 --- a/drivers/iommu/iommufd/device.c +++ b/drivers/iommu/iommufd/device.c @@ -627,6 +627,7 @@ static int iommufd_device_change_pt(struct iommufd_device *idev, u32 *pt_id, return PTR_ERR(pt_obj); switch (pt_obj->type) { + case IOMMUFD_OBJ_HWPT_NESTED: case IOMMUFD_OBJ_HWPT_PAGING: { struct iommufd_hw_pagetable *hwpt = container_of(pt_obj, struct iommufd_hw_pagetable, obj); diff --git a/drivers/iommu/iommufd/hw_pagetable.c b/drivers/iommu/iommufd/hw_pagetable.c index fbe09e2473bc..d41428a0fdc2 100644 --- a/drivers/iommu/iommufd/hw_pagetable.c +++ b/drivers/iommu/iommufd/hw_pagetable.c @@ -44,6 +44,22 @@ void iommufd_hwpt_paging_abort(struct iommufd_object *obj) iommufd_hwpt_paging_destroy(obj); } +void iommufd_hwpt_nested_destroy(struct iommufd_object *obj) +{ + struct iommufd_hwpt_nested *hwpt_nested = + container_of(obj, struct iommufd_hwpt_nested, common.obj); + + if (hwpt_nested->common.domain) + iommu_domain_free(hwpt_nested->common.domain); + + refcount_dec(&hwpt_nested->parent->common.obj.users); +} + +void iommufd_hwpt_nested_abort(struct iommufd_object *obj) +{ + iommufd_hwpt_nested_destroy(obj); +} + static int iommufd_hwpt_paging_enforce_cc(struct iommufd_hwpt_paging *hwpt_paging) { @@ -107,6 +123,7 @@ iommufd_hwpt_paging_alloc(struct iommufd_ctx *ictx, struct iommufd_ioas *ioas, /* Pairs with iommufd_hw_pagetable_destroy() */ refcount_inc(&ioas->obj.users); hwpt_paging->ioas = ioas; + hwpt_paging->nest_parent = flags & IOMMU_HWPT_ALLOC_NEST_PARENT; if (ops->domain_alloc_user) { hwpt->domain = ops->domain_alloc_user(idev->dev, flags, @@ -170,6 +187,73 @@ iommufd_hwpt_paging_alloc(struct iommufd_ctx *ictx, struct iommufd_ioas *ioas, return ERR_PTR(rc); } +/** + * iommufd_hwpt_nested_alloc() - Get a NESTED iommu_domain for a device + * @ictx: iommufd context + * @parent: Parent PAGING-type hwpt to associate the domain with + * @idev: Device to get an iommu_domain for + * @flags: Flags from userspace + * @user_data: user_data pointer. Must be valid + * + * Allocate a new iommu_domain (must be IOMMU_DOMAIN_NESTED) and return it as + * a NESTED hw_pagetable. The given parent PAGING-type hwpt must be capable of + * being a parent. + */ +static struct iommufd_hwpt_nested * +iommufd_hwpt_nested_alloc(struct iommufd_ctx *ictx, + struct iommufd_hwpt_paging *parent, + struct iommufd_device *idev, u32 flags, + const struct iommu_user_data *user_data) +{ + const struct iommu_ops *ops = dev_iommu_ops(idev->dev); + struct iommufd_hwpt_nested *hwpt_nested; + struct iommufd_hw_pagetable *hwpt; + int rc; + + if (flags != 0) + return ERR_PTR(-EOPNOTSUPP); + if (!user_data) + return ERR_PTR(-EINVAL); + if (user_data->type == IOMMU_HWPT_DATA_NONE) + return ERR_PTR(-EINVAL); + if (parent->auto_domain) + return ERR_PTR(-EINVAL); + if (!parent->nest_parent) + return ERR_PTR(-EINVAL); + + if (!ops->domain_alloc_user) + return ERR_PTR(-EOPNOTSUPP); + + hwpt_nested = __iommufd_object_alloc(ictx, hwpt_nested, + IOMMUFD_OBJ_HWPT_NESTED, + common.obj); + if (IS_ERR(hwpt_nested)) + return ERR_CAST(hwpt_nested); + hwpt = &hwpt_nested->common; + + refcount_inc(&parent->common.obj.users); + hwpt_nested->parent = parent; + + hwpt->domain = ops->domain_alloc_user(idev->dev, 0, + parent->common.domain, + user_data); + if (IS_ERR(hwpt->domain)) { + rc = PTR_ERR(hwpt->domain); + hwpt->domain = NULL; + goto out_abort; + } + + if (WARN_ON_ONCE(hwpt->domain->type != IOMMU_DOMAIN_NESTED)) { + rc = -EINVAL; + goto out_abort; + } + return hwpt_nested; + +out_abort: + iommufd_object_abort_and_destroy(ictx, &hwpt->obj); + return ERR_PTR(rc); +} + int iommufd_hwpt_alloc(struct iommufd_ucmd *ucmd) { struct iommu_hwpt_alloc *cmd = ucmd->cmd; @@ -195,6 +279,10 @@ int iommufd_hwpt_alloc(struct iommufd_ucmd *ucmd) if (pt_obj->type == IOMMUFD_OBJ_IOAS) { struct iommufd_hwpt_paging *hwpt_paging; + if (cmd->data_type != IOMMU_HWPT_DATA_NONE) { + rc = -EINVAL; + goto out_put_pt; + } ioas = container_of(pt_obj, struct iommufd_ioas, obj); mutex_lock(&ioas->mutex); hwpt_paging = iommufd_hwpt_paging_alloc(ucmd->ictx, ioas, idev, @@ -204,6 +292,23 @@ int iommufd_hwpt_alloc(struct iommufd_ucmd *ucmd) goto out_unlock; } hwpt = &hwpt_paging->common; + } else if (pt_obj->type == IOMMUFD_OBJ_HWPT_PAGING) { + const struct iommu_user_data user_data = { + .type = cmd->data_type, + .uptr = u64_to_user_ptr(cmd->data_uptr), + .len = cmd->data_len, + }; + struct iommufd_hwpt_nested *hwpt_nested; + struct iommufd_hwpt_paging *parent; + + parent = container_of(pt_obj, typeof(*parent), common.obj); + hwpt_nested = iommufd_hwpt_nested_alloc(ucmd->ictx, parent, idev, + cmd->flags, &user_data); + if (IS_ERR(hwpt_nested)) { + rc = PTR_ERR(hwpt_nested); + goto out_unlock; + } + hwpt = &hwpt_nested->common; } else { rc = -EINVAL; goto out_put_pt; diff --git a/drivers/iommu/iommufd/iommufd_private.h b/drivers/iommu/iommufd/iommufd_private.h index 6493fcdad1b7..6fdad56893af 100644 --- a/drivers/iommu/iommufd/iommufd_private.h +++ b/drivers/iommu/iommufd/iommufd_private.h @@ -124,6 +124,7 @@ enum iommufd_object_type { IOMMUFD_OBJ_ANY = IOMMUFD_OBJ_NONE, IOMMUFD_OBJ_DEVICE, IOMMUFD_OBJ_HWPT_PAGING, + IOMMUFD_OBJ_HWPT_NESTED, IOMMUFD_OBJ_IOAS, IOMMUFD_OBJ_ACCESS, #ifdef CONFIG_IOMMUFD_TEST @@ -255,10 +256,16 @@ struct iommufd_hwpt_paging { bool auto_domain : 1; bool enforce_cache_coherency : 1; bool msi_cookie : 1; + bool nest_parent : 1; /* Head at iommufd_ioas::hwpt_list */ struct list_head hwpt_item; }; +struct iommufd_hwpt_nested { + struct iommufd_hw_pagetable common; + struct iommufd_hwpt_paging *parent; +}; + static inline bool hwpt_is_paging(struct iommufd_hw_pagetable *hwpt) { return hwpt->obj.type == IOMMUFD_OBJ_HWPT_PAGING; @@ -290,18 +297,28 @@ struct iommufd_hw_pagetable * iommufd_hw_pagetable_detach(struct iommufd_device *idev); void iommufd_hwpt_paging_destroy(struct iommufd_object *obj); void iommufd_hwpt_paging_abort(struct iommufd_object *obj); +void iommufd_hwpt_nested_destroy(struct iommufd_object *obj); +void iommufd_hwpt_nested_abort(struct iommufd_object *obj); int iommufd_hwpt_alloc(struct iommufd_ucmd *ucmd); static inline void iommufd_hw_pagetable_put(struct iommufd_ctx *ictx, struct iommufd_hw_pagetable *hwpt) { - struct iommufd_hwpt_paging *hwpt_paging = to_hwpt_paging(hwpt); + if (WARN_ON(hwpt->obj.type != IOMMUFD_OBJ_HWPT_PAGING && + hwpt->obj.type != IOMMUFD_OBJ_HWPT_NESTED)) + return; - lockdep_assert_not_held(&hwpt_paging->ioas->mutex); - if (hwpt_paging->auto_domain) - iommufd_object_deref_user(ictx, &hwpt->obj); - else - refcount_dec(&hwpt->obj.users); + if (hwpt->obj.type == IOMMUFD_OBJ_HWPT_PAGING) { + struct iommufd_hwpt_paging *hwpt_paging = to_hwpt_paging(hwpt); + + lockdep_assert_not_held(&hwpt_paging->ioas->mutex); + + if (hwpt_paging->auto_domain) { + iommufd_object_deref_user(ictx, &hwpt->obj); + return; + } + } + refcount_dec(&hwpt->obj.users); } struct iommufd_group { diff --git a/drivers/iommu/iommufd/main.c b/drivers/iommu/iommufd/main.c index ab6675a7f6b4..45b9d40773b1 100644 --- a/drivers/iommu/iommufd/main.c +++ b/drivers/iommu/iommufd/main.c @@ -492,6 +492,10 @@ static const struct iommufd_object_ops iommufd_object_ops[] = { .destroy = iommufd_hwpt_paging_destroy, .abort = iommufd_hwpt_paging_abort, }, + [IOMMUFD_OBJ_HWPT_NESTED] = { + .destroy = iommufd_hwpt_nested_destroy, + .abort = iommufd_hwpt_nested_abort, + }, #ifdef CONFIG_IOMMUFD_TEST [IOMMUFD_OBJ_SELFTEST] = { .destroy = iommufd_selftest_destroy, diff --git a/include/uapi/linux/iommufd.h b/include/uapi/linux/iommufd.h index fccc6315a520..d816deac906f 100644 --- a/include/uapi/linux/iommufd.h +++ b/include/uapi/linux/iommufd.h @@ -374,15 +374,31 @@ enum iommu_hwpt_data_type { * @size: sizeof(struct iommu_hwpt_alloc) * @flags: Combination of enum iommufd_hwpt_alloc_flags * @dev_id: The device to allocate this HWPT for - * @pt_id: The IOAS to connect this HWPT to + * @pt_id: The IOAS or HWPT to connect this HWPT to * @out_hwpt_id: The ID of the new HWPT * @__reserved: Must be 0 + * @data_type: One of enum iommu_hwpt_data_type + * @data_len: Length of the type specific data + * @data_uptr: User pointer to the type specific data * * Explicitly allocate a hardware page table object. This is the same object * type that is returned by iommufd_device_attach() and represents the * underlying iommu driver's iommu_domain kernel object. * - * A HWPT will be created with the IOVA mappings from the given IOAS. + * A kernel-managed HWPT will be created with the mappings from the given + * IOAS via the @pt_id. The @data_type for this allocation must be set to + * IOMMU_HWPT_DATA_NONE. The HWPT can be allocated as a parent HWPT for a + * nesting configuration by passing IOMMU_HWPT_ALLOC_NEST_PARENT via @flags. + * + * A user-managed nested HWPT will be created from a given parent HWPT via + * @pt_id, in which the parent HWPT must be allocated previously via the + * same ioctl from a given IOAS (@pt_id). In this case, the @data_type + * must be set to a pre-defined type corresponding to an I/O page table + * type supported by the underlying IOMMU hardware. + * + * If the @data_type is set to IOMMU_HWPT_DATA_NONE, @data_len and + * @data_uptr should be zero. Otherwise, both @data_len and @data_uptr + * must be given. */ struct iommu_hwpt_alloc { __u32 size; @@ -391,6 +407,9 @@ struct iommu_hwpt_alloc { __u32 pt_id; __u32 out_hwpt_id; __u32 __reserved; + __u32 data_type; + __u32 data_len; + __aligned_u64 data_uptr; }; #define IOMMU_HWPT_ALLOC _IO(IOMMUFD_TYPE, IOMMUFD_CMD_HWPT_ALLOC)

[v6,07/10] iommufd: Add a nested HW pagetable object

Commit Message

Comments

Patch