diff options
Diffstat (limited to 'drivers/iommu')
-rw-r--r-- | drivers/iommu/amd/iommu.c | 4 | ||||
-rw-r--r-- | drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 2 | ||||
-rw-r--r-- | drivers/iommu/dma-iommu.c | 2 | ||||
-rw-r--r-- | drivers/iommu/intel/dmar.c | 18 | ||||
-rw-r--r-- | drivers/iommu/intel/iommu.c | 18 | ||||
-rw-r--r-- | drivers/iommu/intel/iommu.h | 3 | ||||
-rw-r--r-- | drivers/iommu/intel/irq_remapping.c | 2 | ||||
-rw-r--r-- | drivers/iommu/intel/svm.c | 26 | ||||
-rw-r--r-- | drivers/iommu/iommu.c | 79 | ||||
-rw-r--r-- | drivers/iommu/iommufd/device.c | 14 | ||||
-rw-r--r-- | drivers/iommu/iommufd/hw_pagetable.c | 8 | ||||
-rw-r--r-- | drivers/iommu/iommufd/ioas.c | 14 | ||||
-rw-r--r-- | drivers/iommu/iommufd/iommufd_private.h | 70 | ||||
-rw-r--r-- | drivers/iommu/iommufd/main.c | 146 | ||||
-rw-r--r-- | drivers/iommu/iommufd/selftest.c | 14 | ||||
-rw-r--r-- | drivers/iommu/iommufd/vfio_compat.c | 18 | ||||
-rw-r--r-- | drivers/iommu/of_iommu.c | 14 |
17 files changed, 293 insertions, 159 deletions
diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c index fcc987f5d4ed..b9a0523cbb0a 100644 --- a/drivers/iommu/amd/iommu.c +++ b/drivers/iommu/amd/iommu.c @@ -3357,7 +3357,7 @@ static void irq_remapping_prepare_irte(struct amd_ir_data *data, data->irq_2_irte.devid = devid; data->irq_2_irte.index = index + sub_handle; - iommu->irte_ops->prepare(data->entry, apic->delivery_mode, + iommu->irte_ops->prepare(data->entry, APIC_DELIVERY_MODE_FIXED, apic->dest_mode_logical, irq_cfg->vector, irq_cfg->dest_apicid, devid); @@ -3634,7 +3634,7 @@ int amd_iommu_deactivate_guest_mode(void *data) entry->lo.fields_remap.valid = valid; entry->lo.fields_remap.dm = apic->dest_mode_logical; - entry->lo.fields_remap.int_type = apic->delivery_mode; + entry->lo.fields_remap.int_type = APIC_DELIVERY_MODE_FIXED; entry->hi.fields.vector = cfg->vector; entry->lo.fields_remap.destination = APICID_TO_IRTE_DEST_LO(cfg->dest_apicid); diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h index 961205ba86d2..925ac6a47bce 100644 --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h @@ -188,7 +188,7 @@ #ifdef CONFIG_CMA_ALIGNMENT #define Q_MAX_SZ_SHIFT (PAGE_SHIFT + CONFIG_CMA_ALIGNMENT) #else -#define Q_MAX_SZ_SHIFT (PAGE_SHIFT + MAX_ORDER) +#define Q_MAX_SZ_SHIFT (PAGE_SHIFT + MAX_PAGE_ORDER) #endif /* diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c index 85163a83df2f..e59f50e11ea8 100644 --- a/drivers/iommu/dma-iommu.c +++ b/drivers/iommu/dma-iommu.c @@ -884,7 +884,7 @@ static struct page **__iommu_dma_alloc_pages(struct device *dev, struct page **pages; unsigned int i = 0, nid = dev_to_node(dev); - order_mask &= GENMASK(MAX_ORDER, 0); + order_mask &= GENMASK(MAX_PAGE_ORDER, 0); if (!order_mask) return NULL; diff --git a/drivers/iommu/intel/dmar.c b/drivers/iommu/intel/dmar.c index a3414afe11b0..23cb80d62a9a 100644 --- a/drivers/iommu/intel/dmar.c +++ b/drivers/iommu/intel/dmar.c @@ -1522,6 +1522,15 @@ void qi_flush_dev_iotlb(struct intel_iommu *iommu, u16 sid, u16 pfsid, { struct qi_desc desc; + /* + * VT-d spec, section 4.3: + * + * Software is recommended to not submit any Device-TLB invalidation + * requests while address remapping hardware is disabled. + */ + if (!(iommu->gcmd & DMA_GCMD_TE)) + return; + if (mask) { addr |= (1ULL << (VTD_PAGE_SHIFT + mask - 1)) - 1; desc.qw1 = QI_DEV_IOTLB_ADDR(addr) | QI_DEV_IOTLB_SIZE; @@ -1587,6 +1596,15 @@ void qi_flush_dev_iotlb_pasid(struct intel_iommu *iommu, u16 sid, u16 pfsid, unsigned long mask = 1UL << (VTD_PAGE_SHIFT + size_order - 1); struct qi_desc desc = {.qw1 = 0, .qw2 = 0, .qw3 = 0}; + /* + * VT-d spec, section 4.3: + * + * Software is recommended to not submit any Device-TLB invalidation + * requests while address remapping hardware is disabled. + */ + if (!(iommu->gcmd & DMA_GCMD_TE)) + return; + desc.qw0 = QI_DEV_EIOTLB_PASID(pasid) | QI_DEV_EIOTLB_SID(sid) | QI_DEV_EIOTLB_QDEP(qdep) | QI_DEIOTLB_TYPE | QI_DEV_IOTLB_PFSID(pfsid); diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c index 3531b956556c..897159dba47d 100644 --- a/drivers/iommu/intel/iommu.c +++ b/drivers/iommu/intel/iommu.c @@ -299,7 +299,7 @@ static int iommu_skip_te_disable; #define IDENTMAP_AZALIA 4 const struct iommu_ops intel_iommu_ops; -const struct iommu_dirty_ops intel_dirty_ops; +static const struct iommu_dirty_ops intel_dirty_ops; static bool translation_pre_enabled(struct intel_iommu *iommu) { @@ -2207,6 +2207,8 @@ __domain_mapping(struct dmar_domain *domain, unsigned long iov_pfn, attr |= DMA_FL_PTE_DIRTY; } + domain->has_mappings = true; + pteval = ((phys_addr_t)phys_pfn << VTD_PAGE_SHIFT) | attr; while (nr_pages > 0) { @@ -2490,7 +2492,8 @@ static int dmar_domain_attach_device(struct dmar_domain *domain, return ret; } - iommu_enable_pci_caps(info); + if (sm_supported(info->iommu) || !domain_type_is_si(info->domain)) + iommu_enable_pci_caps(info); return 0; } @@ -3925,8 +3928,8 @@ static int domain_context_clear_one_cb(struct pci_dev *pdev, u16 alias, void *op */ static void domain_context_clear(struct device_domain_info *info) { - if (!info->iommu || !info->dev || !dev_is_pci(info->dev)) - return; + if (!dev_is_pci(info->dev)) + domain_context_clear_one(info, info->bus, info->devfn); pci_for_each_dma_alias(to_pci_dev(info->dev), &domain_context_clear_one_cb, info); @@ -4360,7 +4363,8 @@ static bool intel_iommu_enforce_cache_coherency(struct iommu_domain *domain) return true; spin_lock_irqsave(&dmar_domain->lock, flags); - if (!domain_support_force_snooping(dmar_domain)) { + if (!domain_support_force_snooping(dmar_domain) || + (!dmar_domain->use_first_level && dmar_domain->has_mappings)) { spin_unlock_irqrestore(&dmar_domain->lock, flags); return false; } @@ -4925,7 +4929,7 @@ static int intel_iommu_read_and_clear_dirty(struct iommu_domain *domain, return 0; } -const struct iommu_dirty_ops intel_dirty_ops = { +static const struct iommu_dirty_ops intel_dirty_ops = { .set_dirty_tracking = intel_iommu_set_dirty_tracking, .read_and_clear_dirty = intel_iommu_read_and_clear_dirty, }; @@ -5073,7 +5077,7 @@ static void quirk_igfx_skip_te_disable(struct pci_dev *dev) ver = (dev->device >> 8) & 0xff; if (ver != 0x45 && ver != 0x46 && ver != 0x4c && ver != 0x4e && ver != 0x8a && ver != 0x98 && - ver != 0x9a && ver != 0xa7) + ver != 0x9a && ver != 0xa7 && ver != 0x7d) return; if (risky_device(dev)) diff --git a/drivers/iommu/intel/iommu.h b/drivers/iommu/intel/iommu.h index 65d37a138c75..ce030c5b5772 100644 --- a/drivers/iommu/intel/iommu.h +++ b/drivers/iommu/intel/iommu.h @@ -602,6 +602,9 @@ struct dmar_domain { */ u8 dirty_tracking:1; /* Dirty tracking is enabled */ u8 nested_parent:1; /* Has other domains nested on it */ + u8 has_mappings:1; /* Has mappings configured through + * iommu_map() interface. + */ spinlock_t lock; /* Protect device tracking lists */ struct list_head devices; /* all devices' list */ diff --git a/drivers/iommu/intel/irq_remapping.c b/drivers/iommu/intel/irq_remapping.c index 29b9e55dcf26..566297bc87dd 100644 --- a/drivers/iommu/intel/irq_remapping.c +++ b/drivers/iommu/intel/irq_remapping.c @@ -1112,7 +1112,7 @@ static void prepare_irte(struct irte *irte, int vector, unsigned int dest) * irq migration in the presence of interrupt-remapping. */ irte->trigger_mode = 0; - irte->dlvry_mode = apic->delivery_mode; + irte->dlvry_mode = APIC_DELIVERY_MODE_FIXED; irte->vector = vector; irte->dest_id = IRTE_DEST(dest); irte->redir_hint = 1; diff --git a/drivers/iommu/intel/svm.c b/drivers/iommu/intel/svm.c index 50a481c895b8..ac12f76c1212 100644 --- a/drivers/iommu/intel/svm.c +++ b/drivers/iommu/intel/svm.c @@ -216,6 +216,27 @@ static void intel_flush_svm_range(struct intel_svm *svm, unsigned long address, rcu_read_unlock(); } +static void intel_flush_svm_all(struct intel_svm *svm) +{ + struct device_domain_info *info; + struct intel_svm_dev *sdev; + + rcu_read_lock(); + list_for_each_entry_rcu(sdev, &svm->devs, list) { + info = dev_iommu_priv_get(sdev->dev); + + qi_flush_piotlb(sdev->iommu, sdev->did, svm->pasid, 0, -1UL, 0); + if (info->ats_enabled) { + qi_flush_dev_iotlb_pasid(sdev->iommu, sdev->sid, info->pfsid, + svm->pasid, sdev->qdep, + 0, 64 - VTD_PAGE_SHIFT); + quirk_extra_dev_tlb_flush(info, 0, 64 - VTD_PAGE_SHIFT, + svm->pasid, sdev->qdep); + } + } + rcu_read_unlock(); +} + /* Pages have been freed at this point */ static void intel_arch_invalidate_secondary_tlbs(struct mmu_notifier *mn, struct mm_struct *mm, @@ -223,6 +244,11 @@ static void intel_arch_invalidate_secondary_tlbs(struct mmu_notifier *mn, { struct intel_svm *svm = container_of(mn, struct intel_svm, notifier); + if (start == 0 && end == -1UL) { + intel_flush_svm_all(svm); + return; + } + intel_flush_svm_range(svm, start, (end - start + PAGE_SIZE - 1) >> VTD_PAGE_SHIFT, 0); } diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c index f17a1113f3d6..33e2a9b5d339 100644 --- a/drivers/iommu/iommu.c +++ b/drivers/iommu/iommu.c @@ -485,11 +485,12 @@ static void iommu_deinit_device(struct device *dev) dev_iommu_free(dev); } +DEFINE_MUTEX(iommu_probe_device_lock); + static int __iommu_probe_device(struct device *dev, struct list_head *group_list) { const struct iommu_ops *ops = dev->bus->iommu_ops; struct iommu_group *group; - static DEFINE_MUTEX(iommu_probe_device_lock); struct group_device *gdev; int ret; @@ -502,17 +503,15 @@ static int __iommu_probe_device(struct device *dev, struct list_head *group_list * probably be able to use device_lock() here to minimise the scope, * but for now enforcing a simple global ordering is fine. */ - mutex_lock(&iommu_probe_device_lock); + lockdep_assert_held(&iommu_probe_device_lock); /* Device is probed already if in a group */ - if (dev->iommu_group) { - ret = 0; - goto out_unlock; - } + if (dev->iommu_group) + return 0; ret = iommu_init_device(dev, ops); if (ret) - goto out_unlock; + return ret; group = dev->iommu_group; gdev = iommu_group_alloc_device(group, dev); @@ -548,7 +547,6 @@ static int __iommu_probe_device(struct device *dev, struct list_head *group_list list_add_tail(&group->entry, group_list); } mutex_unlock(&group->mutex); - mutex_unlock(&iommu_probe_device_lock); if (dev_is_pci(dev)) iommu_dma_set_pci_32bit_workaround(dev); @@ -562,8 +560,6 @@ err_put_group: iommu_deinit_device(dev); mutex_unlock(&group->mutex); iommu_group_put(group); -out_unlock: - mutex_unlock(&iommu_probe_device_lock); return ret; } @@ -573,7 +569,9 @@ int iommu_probe_device(struct device *dev) const struct iommu_ops *ops; int ret; + mutex_lock(&iommu_probe_device_lock); ret = __iommu_probe_device(dev, NULL); + mutex_unlock(&iommu_probe_device_lock); if (ret) return ret; @@ -1788,7 +1786,7 @@ iommu_group_alloc_default_domain(struct iommu_group *group, int req_type) */ if (ops->default_domain) { if (req_type) - return NULL; + return ERR_PTR(-EINVAL); return ops->default_domain; } @@ -1797,15 +1795,15 @@ iommu_group_alloc_default_domain(struct iommu_group *group, int req_type) /* The driver gave no guidance on what type to use, try the default */ dom = __iommu_group_alloc_default_domain(group, iommu_def_domain_type); - if (dom) + if (!IS_ERR(dom)) return dom; /* Otherwise IDENTITY and DMA_FQ defaults will try DMA */ if (iommu_def_domain_type == IOMMU_DOMAIN_DMA) - return NULL; + return ERR_PTR(-EINVAL); dom = __iommu_group_alloc_default_domain(group, IOMMU_DOMAIN_DMA); - if (!dom) - return NULL; + if (IS_ERR(dom)) + return dom; pr_warn("Failed to allocate default IOMMU domain of type %u for group %s - Falling back to IOMMU_DOMAIN_DMA", iommu_def_domain_type, group->name); @@ -1822,7 +1820,9 @@ static int probe_iommu_group(struct device *dev, void *data) struct list_head *group_list = data; int ret; + mutex_lock(&iommu_probe_device_lock); ret = __iommu_probe_device(dev, group_list); + mutex_unlock(&iommu_probe_device_lock); if (ret == -ENODEV) ret = 0; @@ -2094,10 +2094,17 @@ static struct iommu_domain *__iommu_domain_alloc(const struct iommu_ops *ops, else if (ops->domain_alloc) domain = ops->domain_alloc(alloc_type); else - return NULL; + return ERR_PTR(-EOPNOTSUPP); + /* + * Many domain_alloc ops now return ERR_PTR, make things easier for the + * driver by accepting ERR_PTR from all domain_alloc ops instead of + * having two rules. + */ + if (IS_ERR(domain)) + return domain; if (!domain) - return NULL; + return ERR_PTR(-ENOMEM); domain->type = type; /* @@ -2110,9 +2117,14 @@ static struct iommu_domain *__iommu_domain_alloc(const struct iommu_ops *ops, if (!domain->ops) domain->ops = ops->default_domain_ops; - if (iommu_is_dma_domain(domain) && iommu_get_dma_cookie(domain)) { - iommu_domain_free(domain); - domain = NULL; + if (iommu_is_dma_domain(domain)) { + int rc; + + rc = iommu_get_dma_cookie(domain); + if (rc) { + iommu_domain_free(domain); + return ERR_PTR(rc); + } } return domain; } @@ -2129,10 +2141,15 @@ __iommu_group_domain_alloc(struct iommu_group *group, unsigned int type) struct iommu_domain *iommu_domain_alloc(const struct bus_type *bus) { + struct iommu_domain *domain; + if (bus == NULL || bus->iommu_ops == NULL) return NULL; - return __iommu_domain_alloc(bus->iommu_ops, NULL, + domain = __iommu_domain_alloc(bus->iommu_ops, NULL, IOMMU_DOMAIN_UNMANAGED); + if (IS_ERR(domain)) + return NULL; + return domain; } EXPORT_SYMBOL_GPL(iommu_domain_alloc); @@ -3041,8 +3058,8 @@ static int iommu_setup_default_domain(struct iommu_group *group, return -EINVAL; dom = iommu_group_alloc_default_domain(group, req_type); - if (!dom) - return -ENODEV; + if (IS_ERR(dom)) + return PTR_ERR(dom); if (group->default_domain == dom) return 0; @@ -3243,21 +3260,23 @@ void iommu_device_unuse_default_domain(struct device *dev) static int __iommu_group_alloc_blocking_domain(struct iommu_group *group) { + struct iommu_domain *domain; + if (group->blocking_domain) return 0; - group->blocking_domain = - __iommu_group_domain_alloc(group, IOMMU_DOMAIN_BLOCKED); - if (!group->blocking_domain) { + domain = __iommu_group_domain_alloc(group, IOMMU_DOMAIN_BLOCKED); + if (IS_ERR(domain)) { /* * For drivers that do not yet understand IOMMU_DOMAIN_BLOCKED * create an empty domain instead. */ - group->blocking_domain = __iommu_group_domain_alloc( - group, IOMMU_DOMAIN_UNMANAGED); - if (!group->blocking_domain) - return -EINVAL; + domain = __iommu_group_domain_alloc(group, + IOMMU_DOMAIN_UNMANAGED); + if (IS_ERR(domain)) + return PTR_ERR(domain); } + group->blocking_domain = domain; return 0; } diff --git a/drivers/iommu/iommufd/device.c b/drivers/iommu/iommufd/device.c index 59d3a07300d9..873630c111c1 100644 --- a/drivers/iommu/iommufd/device.c +++ b/drivers/iommu/iommufd/device.c @@ -571,7 +571,7 @@ iommufd_device_auto_get_domain(struct iommufd_device *idev, continue; destroy_hwpt = (*do_attach)(idev, hwpt); if (IS_ERR(destroy_hwpt)) { - iommufd_put_object(&hwpt->obj); + iommufd_put_object(idev->ictx, &hwpt->obj); /* * -EINVAL means the domain is incompatible with the * device. Other error codes should propagate to @@ -583,7 +583,7 @@ iommufd_device_auto_get_domain(struct iommufd_device *idev, goto out_unlock; } *pt_id = hwpt->obj.id; - iommufd_put_object(&hwpt->obj); + iommufd_put_object(idev->ictx, &hwpt->obj); goto out_unlock; } @@ -652,7 +652,7 @@ static int iommufd_device_change_pt(struct iommufd_device *idev, u32 *pt_id, destroy_hwpt = ERR_PTR(-EINVAL); goto out_put_pt_obj; } - iommufd_put_object(pt_obj); + iommufd_put_object(idev->ictx, pt_obj); /* This destruction has to be after we unlock everything */ if (destroy_hwpt) @@ -660,7 +660,7 @@ static int iommufd_device_change_pt(struct iommufd_device *idev, u32 *pt_id, return 0; out_put_pt_obj: - iommufd_put_object(pt_obj); + iommufd_put_object(idev->ictx, pt_obj); return PTR_ERR(destroy_hwpt); } @@ -792,7 +792,7 @@ static int iommufd_access_change_ioas_id(struct iommufd_access *access, u32 id) if (IS_ERR(ioas)) return PTR_ERR(ioas); rc = iommufd_access_change_ioas(access, ioas); - iommufd_put_object(&ioas->obj); + iommufd_put_object(access->ictx, &ioas->obj); return rc; } @@ -941,7 +941,7 @@ void iommufd_access_notify_unmap(struct io_pagetable *iopt, unsigned long iova, access->ops->unmap(access->data, iova, length); - iommufd_put_object(&access->obj); + iommufd_put_object(access->ictx, &access->obj); xa_lock(&ioas->iopt.access_list); } xa_unlock(&ioas->iopt.access_list); @@ -1243,6 +1243,6 @@ int iommufd_get_hw_info(struct iommufd_ucmd *ucmd) out_free: kfree(data); out_put: - iommufd_put_object(&idev->obj); + iommufd_put_object(ucmd->ictx, &idev->obj); return rc; } diff --git a/drivers/iommu/iommufd/hw_pagetable.c b/drivers/iommu/iommufd/hw_pagetable.c index 2abbeafdbd22..cbb5df0a6c32 100644 --- a/drivers/iommu/iommufd/hw_pagetable.c +++ b/drivers/iommu/iommufd/hw_pagetable.c @@ -318,9 +318,9 @@ out_unlock: if (ioas) mutex_unlock(&ioas->mutex); out_put_pt: - iommufd_put_object(pt_obj); + iommufd_put_object(ucmd->ictx, pt_obj); out_put_idev: - iommufd_put_object(&idev->obj); + iommufd_put_object(ucmd->ictx, &idev->obj); return rc; } @@ -345,7 +345,7 @@ int iommufd_hwpt_set_dirty_tracking(struct iommufd_ucmd *ucmd) rc = iopt_set_dirty_tracking(&ioas->iopt, hwpt_paging->common.domain, enable); - iommufd_put_object(&hwpt_paging->common.obj); + iommufd_put_object(ucmd->ictx, &hwpt_paging->common.obj); return rc; } @@ -368,6 +368,6 @@ int iommufd_hwpt_get_dirty_bitmap(struct iommufd_ucmd *ucmd) rc = iopt_read_and_clear_dirty_data( &ioas->iopt, hwpt_paging->common.domain, cmd->flags, cmd); - iommufd_put_object(&hwpt_paging->common.obj); + iommufd_put_object(ucmd->ictx, &hwpt_paging->common.obj); return rc; } diff --git a/drivers/iommu/iommufd/ioas.c b/drivers/iommu/iommufd/ioas.c index d5624577f79f..742248276548 100644 --- a/drivers/iommu/iommufd/ioas.c +++ b/drivers/iommu/iommufd/ioas.c @@ -105,7 +105,7 @@ int iommufd_ioas_iova_ranges(struct iommufd_ucmd *ucmd) rc = -EMSGSIZE; out_put: up_read(&ioas->iopt.iova_rwsem); - iommufd_put_object(&ioas->obj); + iommufd_put_object(ucmd->ictx, &ioas->obj); return rc; } @@ -175,7 +175,7 @@ out_free: interval_tree_remove(node, &allowed_iova); kfree(container_of(node, struct iopt_allowed, node)); } - iommufd_put_object(&ioas->obj); + iommufd_put_object(ucmd->ictx, &ioas->obj); return rc; } @@ -228,7 +228,7 @@ int iommufd_ioas_map(struct iommufd_ucmd *ucmd) cmd->iova = iova; rc = iommufd_ucmd_respond(ucmd, sizeof(*cmd)); out_put: - iommufd_put_object(&ioas->obj); + iommufd_put_object(ucmd->ictx, &ioas->obj); return rc; } @@ -258,7 +258,7 @@ int iommufd_ioas_copy(struct iommufd_ucmd *ucmd) return PTR_ERR(src_ioas); rc = iopt_get_pages(&src_ioas->iopt, cmd->src_iova, cmd->length, &pages_list); - iommufd_put_object(&src_ioas->obj); + iommufd_put_object(ucmd->ictx, &src_ioas->obj); if (rc) return rc; @@ -279,7 +279,7 @@ int iommufd_ioas_copy(struct iommufd_ucmd *ucmd) cmd->dst_iova = iova; rc = iommufd_ucmd_respond(ucmd, sizeof(*cmd)); out_put_dst: - iommufd_put_object(&dst_ioas->obj); + iommufd_put_object(ucmd->ictx, &dst_ioas->obj); out_pages: iopt_free_pages_list(&pages_list); return rc; @@ -315,7 +315,7 @@ int iommufd_ioas_unmap(struct iommufd_ucmd *ucmd) rc = iommufd_ucmd_respond(ucmd, sizeof(*cmd)); out_put: - iommufd_put_object(&ioas->obj); + iommufd_put_object(ucmd->ictx, &ioas->obj); return rc; } @@ -393,6 +393,6 @@ int iommufd_ioas_option(struct iommufd_ucmd *ucmd) rc = -EOPNOTSUPP; } - iommufd_put_object(&ioas->obj); + iommufd_put_object(ucmd->ictx, &ioas->obj); return rc; } diff --git a/drivers/iommu/iommufd/iommufd_private.h b/drivers/iommu/iommufd/iommufd_private.h index a74cfefffbc6..abae041e256f 100644 --- a/drivers/iommu/iommufd/iommufd_private.h +++ b/drivers/iommu/iommufd/iommufd_private.h @@ -21,6 +21,7 @@ struct iommufd_ctx { struct file *file; struct xarray objects; struct xarray groups; + wait_queue_head_t destroy_wait; u8 account_mode; /* Compatibility with VFIO no iommu */ @@ -135,7 +136,7 @@ enum iommufd_object_type { /* Base struct for all objects with a userspace ID handle. */ struct iommufd_object { - struct rw_semaphore destroy_rwsem; + refcount_t shortterm_users; refcount_t users; enum iommufd_object_type type; unsigned int id; @@ -143,10 +144,15 @@ struct iommufd_object { static inline bool iommufd_lock_obj(struct iommufd_object *obj) { - if (!down_read_trylock(&obj->destroy_rwsem)) + if (!refcount_inc_not_zero(&obj->users)) return false; - if (!refcount_inc_not_zero(&obj->users)) { - up_read(&obj->destroy_rwsem); + if (!refcount_inc_not_zero(&obj->shortterm_users)) { + /* + * If the caller doesn't already have a ref on obj this must be + * called under the xa_lock. Otherwise the caller is holding a + * ref on users. Thus it cannot be one before this decrement. + */ + refcount_dec(&obj->users); return false; } return true; @@ -154,10 +160,16 @@ static inline bool iommufd_lock_obj(struct iommufd_object *obj) struct iommufd_object *iommufd_get_object(struct iommufd_ctx *ictx, u32 id, enum iommufd_object_type type); -static inline void iommufd_put_object(struct iommufd_object *obj) +static inline void iommufd_put_object(struct iommufd_ctx *ictx, + struct iommufd_object *obj) { + /* + * Users first, then shortterm so that REMOVE_WAIT_SHORTTERM never sees + * a spurious !0 users with a 0 shortterm_users. + */ refcount_dec(&obj->users); - up_read(&obj->destroy_rwsem); + if (refcount_dec_and_test(&obj->shortterm_users)) + wake_up_interruptible_all(&ictx->destroy_wait); } void iommufd_object_abort(struct iommufd_ctx *ictx, struct iommufd_object *obj); @@ -165,17 +177,49 @@ void iommufd_object_abort_and_destroy(struct iommufd_ctx *ictx, struct iommufd_object *obj); void iommufd_object_finalize(struct iommufd_ctx *ictx, struct iommufd_object *obj); -void __iommufd_object_destroy_user(struct iommufd_ctx *ictx, - struct iommufd_object *obj, bool allow_fail); + +enum { + REMOVE_WAIT_SHORTTERM = 1, +}; +int iommufd_object_remove(struct iommufd_ctx *ictx, + struct iommufd_object *to_destroy, u32 id, + unsigned int flags); + +/* + * The caller holds a users refcount and wants to destroy the object. At this + * point the caller has no shortterm_users reference and at least the xarray + * will be holding one. + */ static inline void iommufd_object_destroy_user(struct iommufd_ctx *ictx, struct iommufd_object *obj) { - __iommufd_object_destroy_user(ictx, obj, false); + int ret; + + ret = iommufd_object_remove(ictx, obj, obj->id, REMOVE_WAIT_SHORTTERM); + + /* + * If there is a bug and we couldn't destroy the object then we did put + * back the caller's users refcount and will eventually try to free it + * again during close. + */ + WARN_ON(ret); } -static inline void iommufd_object_deref_user(struct iommufd_ctx *ictx, - struct iommufd_object *obj) + +/* + * The HWPT allocated by autodomains is used in possibly many devices and + * is automatically destroyed when its refcount reaches zero. + * + * If userspace uses the HWPT manually, even for a short term, then it will + * disrupt this refcounting and the auto-free in the kernel will not work. + * Userspace that tries to use the automatically allocated HWPT must be careful + * to ensure that it is consistently destroyed, eg by not racing accesses + * and by not attaching an automatic HWPT to a device manually. + */ +static inline void +iommufd_object_put_and_try_destroy(struct iommufd_ctx *ictx, + struct iommufd_object *obj) { - __iommufd_object_destroy_user(ictx, obj, true); + iommufd_object_remove(ictx, obj, obj->id, 0); } struct iommufd_object *_iommufd_object_alloc(struct iommufd_ctx *ictx, @@ -311,7 +355,7 @@ static inline void iommufd_hw_pagetable_put(struct iommufd_ctx *ictx, lockdep_assert_not_held(&hwpt_paging->ioas->mutex); if (hwpt_paging->auto_domain) { - iommufd_object_deref_user(ictx, &hwpt->obj); + iommufd_object_put_and_try_destroy(ictx, &hwpt->obj); return; } } diff --git a/drivers/iommu/iommufd/main.c b/drivers/iommu/iommufd/main.c index 45b9d40773b1..c9091e46d208 100644 --- a/drivers/iommu/iommufd/main.c +++ b/drivers/iommu/iommufd/main.c @@ -33,7 +33,6 @@ struct iommufd_object *_iommufd_object_alloc(struct iommufd_ctx *ictx, size_t size, enum iommufd_object_type type) { - static struct lock_class_key obj_keys[IOMMUFD_OBJ_MAX]; struct iommufd_object *obj; int rc; @@ -41,15 +40,8 @@ struct iommufd_object *_iommufd_object_alloc(struct iommufd_ctx *ictx, if (!obj) return ERR_PTR(-ENOMEM); obj->type = type; - /* - * In most cases the destroy_rwsem is obtained with try so it doesn't - * interact with lockdep, however on destroy we have to sleep. This - * means if we have to destroy an object while holding a get on another - * object it triggers lockdep. Using one locking class per object type - * is a simple and reasonable way to avoid this. - */ - __init_rwsem(&obj->destroy_rwsem, "iommufd_object::destroy_rwsem", - &obj_keys[type]); + /* Starts out bias'd by 1 until it is removed from the xarray */ + refcount_set(&obj->shortterm_users, 1); refcount_set(&obj->users, 1); /* @@ -129,92 +121,113 @@ struct iommufd_object *iommufd_get_object(struct iommufd_ctx *ictx, u32 id, return obj; } +static int iommufd_object_dec_wait_shortterm(struct iommufd_ctx *ictx, + struct iommufd_object *to_destroy) +{ + if (refcount_dec_and_test(&to_destroy->shortterm_users)) + return 0; + + if (wait_event_timeout(ictx->destroy_wait, + refcount_read(&to_destroy->shortterm_users) == + 0, + msecs_to_jiffies(10000))) + return 0; + + pr_crit("Time out waiting for iommufd object to become free\n"); + refcount_inc(&to_destroy->shortterm_users); + return -EBUSY; +} + /* * Remove the given object id from the xarray if the only reference to the - * object is held by the xarray. The caller must call ops destroy(). + * object is held by the xarray. */ -static struct iommufd_object *iommufd_object_remove(struct iommufd_ctx *ictx, - u32 id, bool extra_put) +int iommufd_object_remove(struct iommufd_ctx *ictx, + struct iommufd_object *to_destroy, u32 id, + unsigned int flags) { struct iommufd_object *obj; XA_STATE(xas, &ictx->objects, id); - - xa_lock(&ictx->objects); - obj = xas_load(&xas); - if (xa_is_zero(obj) || !obj) { - obj = ERR_PTR(-ENOENT); - goto out_xa; - } + bool zerod_shortterm = false; + int ret; /* - * If the caller is holding a ref on obj we put it here under the - * spinlock. + * The purpose of the shortterm_users is to ensure deterministic + * destruction of objects used by external drivers and destroyed by this + * function. Any temporary increment of the refcount must increment + * shortterm_users, such as during ioctl execution. */ - if (extra_put) + if (flags & REMOVE_WAIT_SHORTTERM) { + ret = iommufd_object_dec_wait_shortterm(ictx, to_destroy); + if (ret) { + /* + * We have a bug. Put back the callers reference and + * defer cleaning this object until close. + */ + refcount_dec(&to_destroy->users); + return ret; + } + zerod_shortterm = true; + } + + xa_lock(&ictx->objects); + obj = xas_load(&xas); + if (to_destroy) { + /* + * If the caller is holding a ref on obj we put it here under + * the spinlock. + */ refcount_dec(&obj->users); + if (WARN_ON(obj != to_destroy)) { + ret = -ENOENT; + goto err_xa; + } + } else if (xa_is_zero(obj) || !obj) { + ret = -ENOENT; + goto err_xa; + } + if (!refcount_dec_if_one(&obj->users)) { - obj = ERR_PTR(-EBUSY); - goto out_xa; + ret = -EBUSY; + goto err_xa; } xas_store(&xas, NULL); if (ictx->vfio_ioas == container_of(obj, struct iommufd_ioas, obj)) ictx->vfio_ioas = NULL; - -out_xa: xa_unlock(&ictx->objects); - /* The returned object reference count is zero */ - return obj; -} - -/* - * The caller holds a users refcount and wants to destroy the object. Returns - * true if the object was destroyed. In all cases the caller no longer has a - * reference on obj. - */ -void __iommufd_object_destroy_user(struct iommufd_ctx *ictx, - struct iommufd_object *obj, bool allow_fail) -{ - struct iommufd_object *ret; - /* - * The purpose of the destroy_rwsem is to ensure deterministic - * destruction of objects used by external drivers and destroyed by this - * function. Any temporary increment of the refcount must hold the read - * side of this, such as during ioctl execution. - */ - down_write(&obj->destroy_rwsem); - ret = iommufd_object_remove(ictx, obj->id, true); - up_write(&obj->destroy_rwsem); - - if (allow_fail && IS_ERR(ret)) - return; - - /* - * If there is a bug and we couldn't destroy the object then we did put - * back the caller's refcount and will eventually try to free it again - * during close. + * Since users is zero any positive users_shortterm must be racing + * iommufd_put_object(), or we have a bug. */ - if (WARN_ON(IS_ERR(ret))) - return; + if (!zerod_shortterm) { + ret = iommufd_object_dec_wait_shortterm(ictx, obj); + if (WARN_ON(ret)) + return ret; + } iommufd_object_ops[obj->type].destroy(obj); kfree(obj); + return 0; + +err_xa: + if (zerod_shortterm) { + /* Restore the xarray owned reference */ + refcount_set(&obj->shortterm_users, 1); + } + xa_unlock(&ictx->objects); + + /* The returned object reference count is zero */ + return ret; } static int iommufd_destroy(struct iommufd_ucmd *ucmd) { struct iommu_destroy *cmd = ucmd->cmd; - struct iommufd_object *obj; - obj = iommufd_object_remove(ucmd->ictx, cmd->id, false); - if (IS_ERR(obj)) - return PTR_ERR(obj); - iommufd_object_ops[obj->type].destroy(obj); - kfree(obj); - return 0; + return iommufd_object_remove(ucmd->ictx, NULL, cmd->id, 0); } static int iommufd_fops_open(struct inode *inode, struct file *filp) @@ -238,6 +251,7 @@ static int iommufd_fops_open(struct inode *inode, struct file *filp) xa_init_flags(&ictx->objects, XA_FLAGS_ALLOC1 | XA_FLAGS_ACCOUNT); xa_init(&ictx->groups); ictx->file = filp; + init_waitqueue_head(&ictx->destroy_wait); filp->private_data = ictx; return 0; } diff --git a/drivers/iommu/iommufd/selftest.c b/drivers/iommu/iommufd/selftest.c index 5d93434003d8..022ef8f55088 100644 --- a/drivers/iommu/iommufd/selftest.c +++ b/drivers/iommu/iommufd/selftest.c @@ -86,7 +86,7 @@ void iommufd_test_syz_conv_iova_id(struct iommufd_ucmd *ucmd, if (IS_ERR(ioas)) return; *iova = iommufd_test_syz_conv_iova(&ioas->iopt, iova); - iommufd_put_object(&ioas->obj); + iommufd_put_object(ucmd->ictx, &ioas->obj); } struct mock_iommu_domain { @@ -500,7 +500,7 @@ get_md_pagetable(struct iommufd_ucmd *ucmd, u32 mockpt_id, return hwpt; if (hwpt->domain->type != IOMMU_DOMAIN_UNMANAGED || hwpt->domain->ops != mock_ops.default_domain_ops) { - iommufd_put_object(&hwpt->obj); + iommufd_put_object(ucmd->ictx, &hwpt->obj); return ERR_PTR(-EINVAL); } *mock = container_of(hwpt->domain, struct mock_iommu_domain, domain); @@ -518,7 +518,7 @@ get_md_pagetable_nested(struct iommufd_ucmd *ucmd, u32 mockpt_id, return hwpt; if (hwpt->domain->type != IOMMU_DOMAIN_NESTED || hwpt->domain->ops != &domain_nested_ops) { - iommufd_put_object(&hwpt->obj); + iommufd_put_object(ucmd->ictx, &hwpt->obj); return ERR_PTR(-EINVAL); } *mock_nested = container_of(hwpt->domain, @@ -681,7 +681,7 @@ static int iommufd_test_mock_domain_replace(struct iommufd_ucmd *ucmd, rc = iommufd_ucmd_respond(ucmd, sizeof(*cmd)); out_dev_obj: - iommufd_put_object(dev_obj); + iommufd_put_object(ucmd->ictx, dev_obj); return rc; } @@ -699,7 +699,7 @@ static int iommufd_test_add_reserved(struct iommufd_ucmd *ucmd, down_write(&ioas->iopt.iova_rwsem); rc = iopt_reserve_iova(&ioas->iopt, start, start + length - 1, NULL); up_write(&ioas->iopt.iova_rwsem); - iommufd_put_object(&ioas->obj); + iommufd_put_object(ucmd->ictx, &ioas->obj); return rc; } @@ -754,7 +754,7 @@ static int iommufd_test_md_check_pa(struct iommufd_ucmd *ucmd, rc = 0; out_put: - iommufd_put_object(&hwpt->obj); + iommufd_put_object(ucmd->ictx, &hwpt->obj); return rc; } @@ -1233,7 +1233,7 @@ static int iommufd_test_dirty(struct iommufd_ucmd *ucmd, unsigned int mockpt_id, out_free: kvfree(tmp); out_put: - iommufd_put_object(&hwpt->obj); + iommufd_put_object(ucmd->ictx, &hwpt->obj); return rc; } diff --git a/drivers/iommu/iommufd/vfio_compat.c b/drivers/iommu/iommufd/vfio_compat.c index 538fbf76354d..a3ad5f0b6c59 100644 --- a/drivers/iommu/iommufd/vfio_compat.c +++ b/drivers/iommu/iommufd/vfio_compat.c @@ -41,7 +41,7 @@ int iommufd_vfio_compat_ioas_get_id(struct iommufd_ctx *ictx, u32 *out_ioas_id) if (IS_ERR(ioas)) return PTR_ERR(ioas); *out_ioas_id = ioas->obj.id; - iommufd_put_object(&ioas->obj); + iommufd_put_object(ictx, &ioas->obj); return 0; } EXPORT_SYMBOL_NS_GPL(iommufd_vfio_compat_ioas_get_id, IOMMUFD_VFIO); @@ -98,7 +98,7 @@ int iommufd_vfio_compat_ioas_create(struct iommufd_ctx *ictx) if (ictx->vfio_ioas && iommufd_lock_obj(&ictx->vfio_ioas->obj)) { ret = 0; - iommufd_put_object(&ictx->vfio_ioas->obj); + iommufd_put_object(ictx, &ictx->vfio_ioas->obj); goto out_abort; } ictx->vfio_ioas = ioas; @@ -133,7 +133,7 @@ int iommufd_vfio_ioas(struct iommufd_ucmd *ucmd) if (IS_ERR(ioas)) return PTR_ERR(ioas); cmd->ioas_id = ioas->obj.id; - iommufd_put_object(&ioas->obj); + iommufd_put_object(ucmd->ictx, &ioas->obj); return iommufd_ucmd_respond(ucmd, sizeof(*cmd)); case IOMMU_VFIO_IOAS_SET: @@ -143,7 +143,7 @@ int iommufd_vfio_ioas(struct iommufd_ucmd *ucmd) xa_lock(&ucmd->ictx->objects); ucmd->ictx->vfio_ioas = ioas; xa_unlock(&ucmd->ictx->objects); - iommufd_put_object(&ioas->obj); + iommufd_put_object(ucmd->ictx, &ioas->obj); return 0; case IOMMU_VFIO_IOAS_CLEAR: @@ -190,7 +190,7 @@ static int iommufd_vfio_map_dma(struct iommufd_ctx *ictx, unsigned int cmd, iova = map.iova; rc = iopt_map_user_pages(ictx, &ioas->iopt, &iova, u64_to_user_ptr(map.vaddr), map.size, iommu_prot, 0); - iommufd_put_object(&ioas->obj); + iommufd_put_object(ictx, &ioas->obj); return rc; } @@ -249,7 +249,7 @@ static int iommufd_vfio_unmap_dma(struct iommufd_ctx *ictx, unsigned int cmd, rc = -EFAULT; err_put: - iommufd_put_object(&ioas->obj); + iommufd_put_object(ictx, &ioas->obj); return rc; } @@ -272,7 +272,7 @@ static int iommufd_vfio_cc_iommu(struct iommufd_ctx *ictx) } mutex_unlock(&ioas->mutex); - iommufd_put_object(&ioas->obj); + iommufd_put_object(ictx, &ioas->obj); return rc; } @@ -349,7 +349,7 @@ static int iommufd_vfio_set_iommu(struct iommufd_ctx *ictx, unsigned long type) */ if (type == VFIO_TYPE1_IOMMU) rc = iopt_disable_large_pages(&ioas->iopt); - iommufd_put_object(&ioas->obj); + iommufd_put_object(ictx, &ioas->obj); return rc; } @@ -511,7 +511,7 @@ static int iommufd_vfio_iommu_get_info(struct iommufd_ctx *ictx, out_put: up_read(&ioas->iopt.iova_rwsem); - iommufd_put_object(&ioas->obj); + iommufd_put_object(ictx, &ioas->obj); return rc; } diff --git a/drivers/iommu/of_iommu.c b/drivers/iommu/of_iommu.c index 157b286e36bf..35ba090f3b5e 100644 --- a/drivers/iommu/of_iommu.c +++ b/drivers/iommu/of_iommu.c @@ -112,16 +112,20 @@ const struct iommu_ops *of_iommu_configure(struct device *dev, const u32 *id) { const struct iommu_ops *ops = NULL; - struct iommu_fwspec *fwspec = dev_iommu_fwspec_get(dev); + struct iommu_fwspec *fwspec; int err = NO_IOMMU; if (!master_np) return NULL; + /* Serialise to make dev->iommu stable under our potential fwspec */ + mutex_lock(&iommu_probe_device_lock); + fwspec = dev_iommu_fwspec_get(dev); if (fwspec) { - if (fwspec->ops) + if (fwspec->ops) { + mutex_unlock(&iommu_probe_device_lock); return fwspec->ops; - + } /* In the deferred case, start again from scratch */ iommu_fwspec_free(dev); } @@ -155,6 +159,8 @@ const struct iommu_ops *of_iommu_configure(struct device *dev, fwspec = dev_iommu_fwspec_get(dev); ops = fwspec->ops; } + mutex_unlock(&iommu_probe_device_lock); + /* * If we have reason to believe the IOMMU driver missed the initial * probe for dev, replay it to get things in order. @@ -191,7 +197,7 @@ iommu_resv_region_get_type(struct device *dev, if (start == phys->start && end == phys->end) return IOMMU_RESV_DIRECT; - dev_warn(dev, "treating non-direct mapping [%pr] -> [%pap-%pap] as reservation\n", &phys, + dev_warn(dev, "treating non-direct mapping [%pr] -> [%pap-%pap] as reservation\n", phys, &start, &end); return IOMMU_RESV_RESERVED; } |