X-Git-Url: http://git.cascardo.eti.br/?a=blobdiff_plain;f=drivers%2Fiommu%2Famd_iommu.c;h=96de97a460791ae70efa8f6b0ff91e736363c5e9;hb=cfb89f2e7505c6823020a18bbdc5410284305234;hp=634f636393d5e423b2a6d3bf01659c1643b865c8;hpb=60c07f80b04698f6c054b675cb67ec5e7ee8db27;p=cascardo%2Flinux.git diff --git a/drivers/iommu/amd_iommu.c b/drivers/iommu/amd_iommu.c index 634f636393d5..96de97a46079 100644 --- a/drivers/iommu/amd_iommu.c +++ b/drivers/iommu/amd_iommu.c @@ -21,6 +21,7 @@ #include #include #include +#include #include #include #include @@ -38,6 +39,7 @@ #include #include #include +#include #include #include #include @@ -56,6 +58,17 @@ #define LOOP_TIMEOUT 100000 +/* IO virtual address start page frame number */ +#define IOVA_START_PFN (1) +#define IOVA_PFN(addr) ((addr) >> PAGE_SHIFT) +#define DMA_32BIT_PFN IOVA_PFN(DMA_BIT_MASK(32)) + +/* Reserved IOVA ranges */ +#define MSI_RANGE_START (0xfee00000) +#define MSI_RANGE_END (0xfeefffff) +#define HT_RANGE_START (0xfd00000000ULL) +#define HT_RANGE_END (0xffffffffffULL) + /* * This bitmap is used to advertise the page sizes our hardware support * to the IOMMU core, which will then use this information to split @@ -76,6 +89,25 @@ LIST_HEAD(ioapic_map); LIST_HEAD(hpet_map); LIST_HEAD(acpihid_map); +#define FLUSH_QUEUE_SIZE 256 + +struct flush_queue_entry { + unsigned long iova_pfn; + unsigned long pages; + struct dma_ops_domain *dma_dom; +}; + +struct flush_queue { + spinlock_t lock; + unsigned next; + struct flush_queue_entry *entries; +}; + +DEFINE_PER_CPU(struct flush_queue, flush_queue); + +static atomic_t queue_timer_on; +static struct timer_list queue_timer; + /* * Domain for untranslated devices - only allocated * if iommu=pt passed on kernel cmd line. @@ -120,28 +152,6 @@ static void update_domain(struct protection_domain *domain); static int protection_domain_init(struct protection_domain *domain); static void detach_device(struct device *dev); -/* - * For dynamic growth the aperture size is split into ranges of 128MB of - * DMA address space each. This struct represents one such range. - */ -struct aperture_range { - - spinlock_t bitmap_lock; - - /* address allocation bitmap */ - unsigned long *bitmap; - unsigned long offset; - unsigned long next_bit; - - /* - * Array of PTE pages for the aperture. In this array we save all the - * leaf pages of the domain page table used for the aperture. This way - * we don't need to walk the page table to find a specific PTE. We can - * just calculate its address in constant time. - */ - u64 *pte_pages[64]; -}; - /* * Data container for a dma_ops specific protection domain */ @@ -149,16 +159,13 @@ struct dma_ops_domain { /* generic protection domain information */ struct protection_domain domain; - /* size of the aperture for the mappings */ - unsigned long aperture_size; - - /* aperture index we start searching for free addresses */ - u32 __percpu *next_index; - - /* address space relevant data */ - struct aperture_range *aperture[APERTURE_MAX_RANGES]; + /* IOVA RB-Tree */ + struct iova_domain iovad; }; +static struct iova_domain reserved_iova_ranges; +static struct lock_class_key reserved_rbtree_key; + /**************************************************************************** * * Helper functions @@ -224,6 +231,12 @@ static struct protection_domain *to_pdomain(struct iommu_domain *dom) return container_of(dom, struct protection_domain, domain); } +static struct dma_ops_domain* to_dma_ops_domain(struct protection_domain *domain) +{ + BUG_ON(domain->flags != PD_DMA_OPS_MASK); + return container_of(domain, struct dma_ops_domain, domain); +} + static struct iommu_dev_data *alloc_dev_data(u16 devid) { struct iommu_dev_data *dev_data; @@ -390,43 +403,6 @@ static bool pdev_pri_erratum(struct pci_dev *pdev, u32 erratum) return dev_data->errata & (1 << erratum) ? true : false; } -/* - * This function actually applies the mapping to the page table of the - * dma_ops domain. - */ -static void alloc_unity_mapping(struct dma_ops_domain *dma_dom, - struct unity_map_entry *e) -{ - u64 addr; - - for (addr = e->address_start; addr < e->address_end; - addr += PAGE_SIZE) { - if (addr < dma_dom->aperture_size) - __set_bit(addr >> PAGE_SHIFT, - dma_dom->aperture[0]->bitmap); - } -} - -/* - * Inits the unity mappings required for a specific device - */ -static void init_unity_mappings_for_device(struct device *dev, - struct dma_ops_domain *dma_dom) -{ - struct unity_map_entry *e; - int devid; - - devid = get_device_id(dev); - if (devid < 0) - return; - - list_for_each_entry(e, &amd_iommu_unity_map, list) { - if (!(devid >= e->devid_start && devid <= e->devid_end)) - continue; - alloc_unity_mapping(dma_dom, e); - } -} - /* * This function checks if the driver got a valid device from the caller to * avoid dereferencing invalid pointers. @@ -454,22 +430,12 @@ static bool check_device(struct device *dev) static void init_iommu_group(struct device *dev) { - struct dma_ops_domain *dma_domain; - struct iommu_domain *domain; struct iommu_group *group; group = iommu_group_get_for_dev(dev); if (IS_ERR(group)) return; - domain = iommu_group_default_domain(group); - if (!domain) - goto out; - - dma_domain = to_pdomain(domain)->priv; - - init_unity_mappings_for_device(dev, dma_domain); -out: iommu_group_put(group); } @@ -1220,7 +1186,7 @@ static void domain_flush_complete(struct protection_domain *domain) int i; for (i = 0; i < amd_iommus_present; ++i) { - if (!domain->dev_iommu[i]) + if (domain && !domain->dev_iommu[i]) continue; /* @@ -1397,8 +1363,9 @@ static u64 *fetch_pte(struct protection_domain *domain, static int iommu_map_page(struct protection_domain *dom, unsigned long bus_addr, unsigned long phys_addr, + unsigned long page_size, int prot, - unsigned long page_size) + gfp_t gfp) { u64 __pte, *pte; int i, count; @@ -1410,7 +1377,7 @@ static int iommu_map_page(struct protection_domain *dom, return -EINVAL; count = PAGE_SIZE_PTE_COUNT(page_size); - pte = alloc_pte(dom, bus_addr, page_size, NULL, GFP_KERNEL); + pte = alloc_pte(dom, bus_addr, page_size, NULL, gfp); if (!pte) return -ENOMEM; @@ -1474,320 +1441,37 @@ static unsigned long iommu_unmap_page(struct protection_domain *dom, /**************************************************************************** * * The next functions belong to the address allocator for the dma_ops - * interface functions. They work like the allocators in the other IOMMU - * drivers. Its basically a bitmap which marks the allocated pages in - * the aperture. Maybe it could be enhanced in the future to a more - * efficient allocator. + * interface functions. * ****************************************************************************/ -/* - * The address allocator core functions. - * - * called with domain->lock held - */ -/* - * Used to reserve address ranges in the aperture (e.g. for exclusion - * ranges. - */ -static void dma_ops_reserve_addresses(struct dma_ops_domain *dom, - unsigned long start_page, - unsigned int pages) +static unsigned long dma_ops_alloc_iova(struct device *dev, + struct dma_ops_domain *dma_dom, + unsigned int pages, u64 dma_mask) { - unsigned int i, last_page = dom->aperture_size >> PAGE_SHIFT; - - if (start_page + pages > last_page) - pages = last_page - start_page; - - for (i = start_page; i < start_page + pages; ++i) { - int index = i / APERTURE_RANGE_PAGES; - int page = i % APERTURE_RANGE_PAGES; - __set_bit(page, dom->aperture[index]->bitmap); - } -} + unsigned long pfn = 0; -/* - * This function is used to add a new aperture range to an existing - * aperture in case of dma_ops domain allocation or address allocation - * failure. - */ -static int alloc_new_range(struct dma_ops_domain *dma_dom, - bool populate, gfp_t gfp) -{ - int index = dma_dom->aperture_size >> APERTURE_RANGE_SHIFT; - unsigned long i, old_size, pte_pgsize; - struct aperture_range *range; - struct amd_iommu *iommu; - unsigned long flags; + pages = __roundup_pow_of_two(pages); -#ifdef CONFIG_IOMMU_STRESS - populate = false; -#endif + if (dma_mask > DMA_BIT_MASK(32)) + pfn = alloc_iova_fast(&dma_dom->iovad, pages, + IOVA_PFN(DMA_BIT_MASK(32))); - if (index >= APERTURE_MAX_RANGES) - return -ENOMEM; - - range = kzalloc(sizeof(struct aperture_range), gfp); - if (!range) - return -ENOMEM; - - range->bitmap = (void *)get_zeroed_page(gfp); - if (!range->bitmap) - goto out_free; - - range->offset = dma_dom->aperture_size; - - spin_lock_init(&range->bitmap_lock); - - if (populate) { - unsigned long address = dma_dom->aperture_size; - int i, num_ptes = APERTURE_RANGE_PAGES / 512; - u64 *pte, *pte_page; - - for (i = 0; i < num_ptes; ++i) { - pte = alloc_pte(&dma_dom->domain, address, PAGE_SIZE, - &pte_page, gfp); - if (!pte) - goto out_free; - - range->pte_pages[i] = pte_page; - - address += APERTURE_RANGE_SIZE / 64; - } - } + if (!pfn) + pfn = alloc_iova_fast(&dma_dom->iovad, pages, IOVA_PFN(dma_mask)); - spin_lock_irqsave(&dma_dom->domain.lock, flags); - - /* First take the bitmap_lock and then publish the range */ - spin_lock(&range->bitmap_lock); - - old_size = dma_dom->aperture_size; - dma_dom->aperture[index] = range; - dma_dom->aperture_size += APERTURE_RANGE_SIZE; - - /* Reserve address range used for MSI messages */ - if (old_size < MSI_ADDR_BASE_LO && - dma_dom->aperture_size > MSI_ADDR_BASE_LO) { - unsigned long spage; - int pages; - - pages = iommu_num_pages(MSI_ADDR_BASE_LO, 0x10000, PAGE_SIZE); - spage = MSI_ADDR_BASE_LO >> PAGE_SHIFT; - - dma_ops_reserve_addresses(dma_dom, spage, pages); - } - - /* Initialize the exclusion range if necessary */ - for_each_iommu(iommu) { - if (iommu->exclusion_start && - iommu->exclusion_start >= dma_dom->aperture[index]->offset - && iommu->exclusion_start < dma_dom->aperture_size) { - unsigned long startpage; - int pages = iommu_num_pages(iommu->exclusion_start, - iommu->exclusion_length, - PAGE_SIZE); - startpage = iommu->exclusion_start >> PAGE_SHIFT; - dma_ops_reserve_addresses(dma_dom, startpage, pages); - } - } - - /* - * Check for areas already mapped as present in the new aperture - * range and mark those pages as reserved in the allocator. Such - * mappings may already exist as a result of requested unity - * mappings for devices. - */ - for (i = dma_dom->aperture[index]->offset; - i < dma_dom->aperture_size; - i += pte_pgsize) { - u64 *pte = fetch_pte(&dma_dom->domain, i, &pte_pgsize); - if (!pte || !IOMMU_PTE_PRESENT(*pte)) - continue; - - dma_ops_reserve_addresses(dma_dom, i >> PAGE_SHIFT, - pte_pgsize >> 12); - } - - update_domain(&dma_dom->domain); - - spin_unlock(&range->bitmap_lock); - - spin_unlock_irqrestore(&dma_dom->domain.lock, flags); - - return 0; - -out_free: - update_domain(&dma_dom->domain); - - free_page((unsigned long)range->bitmap); - - kfree(range); - - return -ENOMEM; + return (pfn << PAGE_SHIFT); } -static dma_addr_t dma_ops_aperture_alloc(struct dma_ops_domain *dom, - struct aperture_range *range, - unsigned long pages, - unsigned long dma_mask, - unsigned long boundary_size, - unsigned long align_mask, - bool trylock) +static void dma_ops_free_iova(struct dma_ops_domain *dma_dom, + unsigned long address, + unsigned int pages) { - unsigned long offset, limit, flags; - dma_addr_t address; - bool flush = false; - - offset = range->offset >> PAGE_SHIFT; - limit = iommu_device_max_index(APERTURE_RANGE_PAGES, offset, - dma_mask >> PAGE_SHIFT); - - if (trylock) { - if (!spin_trylock_irqsave(&range->bitmap_lock, flags)) - return -1; - } else { - spin_lock_irqsave(&range->bitmap_lock, flags); - } - - address = iommu_area_alloc(range->bitmap, limit, range->next_bit, - pages, offset, boundary_size, align_mask); - if (address == -1) { - /* Nothing found, retry one time */ - address = iommu_area_alloc(range->bitmap, limit, - 0, pages, offset, boundary_size, - align_mask); - flush = true; - } - - if (address != -1) - range->next_bit = address + pages; - - spin_unlock_irqrestore(&range->bitmap_lock, flags); - - if (flush) { - domain_flush_tlb(&dom->domain); - domain_flush_complete(&dom->domain); - } - - return address; -} - -static unsigned long dma_ops_area_alloc(struct device *dev, - struct dma_ops_domain *dom, - unsigned int pages, - unsigned long align_mask, - u64 dma_mask) -{ - unsigned long boundary_size, mask; - unsigned long address = -1; - bool first = true; - u32 start, i; - - preempt_disable(); - - mask = dma_get_seg_boundary(dev); - -again: - start = this_cpu_read(*dom->next_index); - - /* Sanity check - is it really necessary? */ - if (unlikely(start > APERTURE_MAX_RANGES)) { - start = 0; - this_cpu_write(*dom->next_index, 0); - } - - boundary_size = mask + 1 ? ALIGN(mask + 1, PAGE_SIZE) >> PAGE_SHIFT : - 1UL << (BITS_PER_LONG - PAGE_SHIFT); - - for (i = 0; i < APERTURE_MAX_RANGES; ++i) { - struct aperture_range *range; - int index; - - index = (start + i) % APERTURE_MAX_RANGES; - - range = dom->aperture[index]; - - if (!range || range->offset >= dma_mask) - continue; - - address = dma_ops_aperture_alloc(dom, range, pages, - dma_mask, boundary_size, - align_mask, first); - if (address != -1) { - address = range->offset + (address << PAGE_SHIFT); - this_cpu_write(*dom->next_index, index); - break; - } - } - - if (address == -1 && first) { - first = false; - goto again; - } - - preempt_enable(); - - return address; -} - -static unsigned long dma_ops_alloc_addresses(struct device *dev, - struct dma_ops_domain *dom, - unsigned int pages, - unsigned long align_mask, - u64 dma_mask) -{ - unsigned long address = -1; - - while (address == -1) { - address = dma_ops_area_alloc(dev, dom, pages, - align_mask, dma_mask); - - if (address == -1 && alloc_new_range(dom, false, GFP_ATOMIC)) - break; - } - - if (unlikely(address == -1)) - address = DMA_ERROR_CODE; - - WARN_ON((address + (PAGE_SIZE*pages)) > dom->aperture_size); - - return address; -} - -/* - * The address free function. - * - * called with domain->lock held - */ -static void dma_ops_free_addresses(struct dma_ops_domain *dom, - unsigned long address, - unsigned int pages) -{ - unsigned i = address >> APERTURE_RANGE_SHIFT; - struct aperture_range *range = dom->aperture[i]; - unsigned long flags; - - BUG_ON(i >= APERTURE_MAX_RANGES || range == NULL); - -#ifdef CONFIG_IOMMU_STRESS - if (i < 4) - return; -#endif - - if (amd_iommu_unmap_flush) { - domain_flush_tlb(&dom->domain); - domain_flush_complete(&dom->domain); - } - - address = (address % APERTURE_RANGE_SIZE) >> PAGE_SHIFT; - - spin_lock_irqsave(&range->bitmap_lock, flags); - if (address + pages > range->next_bit) - range->next_bit = address + pages; - bitmap_clear(range->bitmap, address, pages); - spin_unlock_irqrestore(&range->bitmap_lock, flags); + pages = __roundup_pow_of_two(pages); + address >>= PAGE_SHIFT; + free_iova_fast(&dma_dom->iovad, address, pages); } /**************************************************************************** @@ -1961,44 +1645,18 @@ static void free_gcr3_table(struct protection_domain *domain) */ static void dma_ops_domain_free(struct dma_ops_domain *dom) { - int i; - if (!dom) return; - free_percpu(dom->next_index); - del_domain_from_list(&dom->domain); - free_pagetable(&dom->domain); + put_iova_domain(&dom->iovad); - for (i = 0; i < APERTURE_MAX_RANGES; ++i) { - if (!dom->aperture[i]) - continue; - free_page((unsigned long)dom->aperture[i]->bitmap); - kfree(dom->aperture[i]); - } + free_pagetable(&dom->domain); kfree(dom); } -static int dma_ops_domain_alloc_apertures(struct dma_ops_domain *dma_dom, - int max_apertures) -{ - int ret, i, apertures; - - apertures = dma_dom->aperture_size >> APERTURE_RANGE_SHIFT; - ret = 0; - - for (i = apertures; i < max_apertures; ++i) { - ret = alloc_new_range(dma_dom, false, GFP_KERNEL); - if (ret) - break; - } - - return ret; -} - /* * Allocates a new protection domain usable for the dma_ops functions. * It also initializes the page table and the address allocator data @@ -2007,7 +1665,6 @@ static int dma_ops_domain_alloc_apertures(struct dma_ops_domain *dma_dom, static struct dma_ops_domain *dma_ops_domain_alloc(void) { struct dma_ops_domain *dma_dom; - int cpu; dma_dom = kzalloc(sizeof(struct dma_ops_domain), GFP_KERNEL); if (!dma_dom) @@ -2016,30 +1673,19 @@ static struct dma_ops_domain *dma_ops_domain_alloc(void) if (protection_domain_init(&dma_dom->domain)) goto free_dma_dom; - dma_dom->next_index = alloc_percpu(u32); - if (!dma_dom->next_index) - goto free_dma_dom; - - dma_dom->domain.mode = PAGE_MODE_2_LEVEL; + dma_dom->domain.mode = PAGE_MODE_3_LEVEL; dma_dom->domain.pt_root = (void *)get_zeroed_page(GFP_KERNEL); dma_dom->domain.flags = PD_DMA_OPS_MASK; - dma_dom->domain.priv = dma_dom; if (!dma_dom->domain.pt_root) goto free_dma_dom; - add_domain_to_list(&dma_dom->domain); + init_iova_domain(&dma_dom->iovad, PAGE_SIZE, + IOVA_START_PFN, DMA_32BIT_PFN); - if (alloc_new_range(dma_dom, true, GFP_KERNEL)) - goto free_dma_dom; - - /* - * mark the first page as allocated so we never return 0 as - * a valid dma-address. So we can use 0 as error value - */ - dma_dom->aperture[0]->bitmap[0] = 1; + /* Initialize reserved ranges */ + copy_reserved_iova(&reserved_iova_ranges, &dma_dom->iovad); - for_each_possible_cpu(cpu) - *per_cpu_ptr(dma_dom->next_index, cpu) = 0; + add_domain_to_list(&dma_dom->domain); return dma_dom; @@ -2482,6 +2128,92 @@ static struct iommu_group *amd_iommu_device_group(struct device *dev) * *****************************************************************************/ +static void __queue_flush(struct flush_queue *queue) +{ + struct protection_domain *domain; + unsigned long flags; + int idx; + + /* First flush TLB of all known domains */ + spin_lock_irqsave(&amd_iommu_pd_lock, flags); + list_for_each_entry(domain, &amd_iommu_pd_list, list) + domain_flush_tlb(domain); + spin_unlock_irqrestore(&amd_iommu_pd_lock, flags); + + /* Wait until flushes have completed */ + domain_flush_complete(NULL); + + for (idx = 0; idx < queue->next; ++idx) { + struct flush_queue_entry *entry; + + entry = queue->entries + idx; + + free_iova_fast(&entry->dma_dom->iovad, + entry->iova_pfn, + entry->pages); + + /* Not really necessary, just to make sure we catch any bugs */ + entry->dma_dom = NULL; + } + + queue->next = 0; +} + +static void queue_flush_all(void) +{ + int cpu; + + for_each_possible_cpu(cpu) { + struct flush_queue *queue; + unsigned long flags; + + queue = per_cpu_ptr(&flush_queue, cpu); + spin_lock_irqsave(&queue->lock, flags); + if (queue->next > 0) + __queue_flush(queue); + spin_unlock_irqrestore(&queue->lock, flags); + } +} + +static void queue_flush_timeout(unsigned long unsused) +{ + atomic_set(&queue_timer_on, 0); + queue_flush_all(); +} + +static void queue_add(struct dma_ops_domain *dma_dom, + unsigned long address, unsigned long pages) +{ + struct flush_queue_entry *entry; + struct flush_queue *queue; + unsigned long flags; + int idx; + + pages = __roundup_pow_of_two(pages); + address >>= PAGE_SHIFT; + + queue = get_cpu_ptr(&flush_queue); + spin_lock_irqsave(&queue->lock, flags); + + if (queue->next == FLUSH_QUEUE_SIZE) + __queue_flush(queue); + + idx = queue->next++; + entry = queue->entries + idx; + + entry->iova_pfn = address; + entry->pages = pages; + entry->dma_dom = dma_dom; + + spin_unlock_irqrestore(&queue->lock, flags); + + if (atomic_cmpxchg(&queue_timer_on, 0, 1) == 0) + mod_timer(&queue_timer, jiffies + msecs_to_jiffies(10)); + + put_cpu_ptr(&flush_queue); +} + + /* * In the dma_ops path we only have the struct device. This function * finds the corresponding IOMMU, the protection domain and the @@ -2492,16 +2224,11 @@ static struct iommu_group *amd_iommu_device_group(struct device *dev) static struct protection_domain *get_domain(struct device *dev) { struct protection_domain *domain; - struct iommu_domain *io_domain; if (!check_device(dev)) return ERR_PTR(-EINVAL); - io_domain = iommu_get_domain_for_dev(dev); - if (!io_domain) - return NULL; - - domain = to_pdomain(io_domain); + domain = get_dev_data(dev)->domain; if (!dma_ops_domain(domain)) return ERR_PTR(-EBUSY); @@ -2512,8 +2239,15 @@ static void update_device_table(struct protection_domain *domain) { struct iommu_dev_data *dev_data; - list_for_each_entry(dev_data, &domain->dev_list, list) + list_for_each_entry(dev_data, &domain->dev_list, list) { set_dte_entry(dev_data->devid, domain, dev_data->ats.enabled); + + if (dev_data->devid == dev_data->alias) + continue; + + /* There is an alias, update device table entry for it */ + set_dte_entry(dev_data->alias, domain, dev_data->ats.enabled); + } } static void update_domain(struct protection_domain *domain) @@ -2529,94 +2263,17 @@ static void update_domain(struct protection_domain *domain) domain->updated = false; } -/* - * This function fetches the PTE for a given address in the aperture - */ -static u64* dma_ops_get_pte(struct dma_ops_domain *dom, - unsigned long address) -{ - struct aperture_range *aperture; - u64 *pte, *pte_page; - - aperture = dom->aperture[APERTURE_RANGE_INDEX(address)]; - if (!aperture) - return NULL; - - pte = aperture->pte_pages[APERTURE_PAGE_INDEX(address)]; - if (!pte) { - pte = alloc_pte(&dom->domain, address, PAGE_SIZE, &pte_page, - GFP_ATOMIC); - aperture->pte_pages[APERTURE_PAGE_INDEX(address)] = pte_page; - } else - pte += PM_LEVEL_INDEX(0, address); - - update_domain(&dom->domain); - - return pte; -} - -/* - * This is the generic map function. It maps one 4kb page at paddr to - * the given address in the DMA address space for the domain. - */ -static dma_addr_t dma_ops_domain_map(struct dma_ops_domain *dom, - unsigned long address, - phys_addr_t paddr, - int direction) +static int dir2prot(enum dma_data_direction direction) { - u64 *pte, __pte; - - WARN_ON(address > dom->aperture_size); - - paddr &= PAGE_MASK; - - pte = dma_ops_get_pte(dom, address); - if (!pte) - return DMA_ERROR_CODE; - - __pte = paddr | IOMMU_PTE_P | IOMMU_PTE_FC; - if (direction == DMA_TO_DEVICE) - __pte |= IOMMU_PTE_IR; + return IOMMU_PROT_IR; else if (direction == DMA_FROM_DEVICE) - __pte |= IOMMU_PTE_IW; + return IOMMU_PROT_IW; else if (direction == DMA_BIDIRECTIONAL) - __pte |= IOMMU_PTE_IR | IOMMU_PTE_IW; - - WARN_ON_ONCE(*pte); - - *pte = __pte; - - return (dma_addr_t)address; -} - -/* - * The generic unmapping function for on page in the DMA address space. - */ -static void dma_ops_domain_unmap(struct dma_ops_domain *dom, - unsigned long address) -{ - struct aperture_range *aperture; - u64 *pte; - - if (address >= dom->aperture_size) - return; - - aperture = dom->aperture[APERTURE_RANGE_INDEX(address)]; - if (!aperture) - return; - - pte = aperture->pte_pages[APERTURE_PAGE_INDEX(address)]; - if (!pte) - return; - - pte += PM_LEVEL_INDEX(0, address); - - WARN_ON_ONCE(!*pte); - - *pte = 0ULL; + return IOMMU_PROT_IW | IOMMU_PROT_IR; + else + return 0; } - /* * This function contains common code for mapping of a physically * contiguous memory region into DMA address space. It is used by all @@ -2627,32 +2284,29 @@ static dma_addr_t __map_single(struct device *dev, struct dma_ops_domain *dma_dom, phys_addr_t paddr, size_t size, - int dir, - bool align, + enum dma_data_direction direction, u64 dma_mask) { dma_addr_t offset = paddr & ~PAGE_MASK; dma_addr_t address, start, ret; unsigned int pages; - unsigned long align_mask = 0; + int prot = 0; int i; pages = iommu_num_pages(paddr, size, PAGE_SIZE); paddr &= PAGE_MASK; - if (align) - align_mask = (1UL << get_order(size)) - 1; - - address = dma_ops_alloc_addresses(dev, dma_dom, pages, align_mask, - dma_mask); - + address = dma_ops_alloc_iova(dev, dma_dom, pages, dma_mask); if (address == DMA_ERROR_CODE) goto out; + prot = dir2prot(direction); + start = address; for (i = 0; i < pages; ++i) { - ret = dma_ops_domain_map(dma_dom, start, paddr, dir); - if (ret == DMA_ERROR_CODE) + ret = iommu_map_page(&dma_dom->domain, start, paddr, + PAGE_SIZE, prot, GFP_ATOMIC); + if (ret) goto out_unmap; paddr += PAGE_SIZE; @@ -2672,10 +2326,13 @@ out_unmap: for (--i; i >= 0; --i) { start -= PAGE_SIZE; - dma_ops_domain_unmap(dma_dom, start); + iommu_unmap_page(&dma_dom->domain, start, PAGE_SIZE); } - dma_ops_free_addresses(dma_dom, address, pages); + domain_flush_tlb(&dma_dom->domain); + domain_flush_complete(&dma_dom->domain); + + dma_ops_free_iova(dma_dom, address, pages); return DMA_ERROR_CODE; } @@ -2693,21 +2350,23 @@ static void __unmap_single(struct dma_ops_domain *dma_dom, dma_addr_t i, start; unsigned int pages; - if ((dma_addr == DMA_ERROR_CODE) || - (dma_addr + size > dma_dom->aperture_size)) - return; - flush_addr = dma_addr; pages = iommu_num_pages(dma_addr, size, PAGE_SIZE); dma_addr &= PAGE_MASK; start = dma_addr; for (i = 0; i < pages; ++i) { - dma_ops_domain_unmap(dma_dom, start); + iommu_unmap_page(&dma_dom->domain, start, PAGE_SIZE); start += PAGE_SIZE; } - dma_ops_free_addresses(dma_dom, dma_addr, pages); + if (amd_iommu_unmap_flush) { + dma_ops_free_iova(dma_dom, dma_addr, pages); + domain_flush_tlb(&dma_dom->domain); + domain_flush_complete(&dma_dom->domain); + } else { + queue_add(dma_dom, dma_addr, pages); + } } /* @@ -2716,10 +2375,11 @@ static void __unmap_single(struct dma_ops_domain *dma_dom, static dma_addr_t map_page(struct device *dev, struct page *page, unsigned long offset, size_t size, enum dma_data_direction dir, - struct dma_attrs *attrs) + unsigned long attrs) { phys_addr_t paddr = page_to_phys(page) + offset; struct protection_domain *domain; + struct dma_ops_domain *dma_dom; u64 dma_mask; domain = get_domain(dev); @@ -2729,24 +2389,53 @@ static dma_addr_t map_page(struct device *dev, struct page *page, return DMA_ERROR_CODE; dma_mask = *dev->dma_mask; + dma_dom = to_dma_ops_domain(domain); - return __map_single(dev, domain->priv, paddr, size, dir, false, - dma_mask); + return __map_single(dev, dma_dom, paddr, size, dir, dma_mask); } /* * The exported unmap_single function for dma_ops. */ static void unmap_page(struct device *dev, dma_addr_t dma_addr, size_t size, - enum dma_data_direction dir, struct dma_attrs *attrs) + enum dma_data_direction dir, unsigned long attrs) { struct protection_domain *domain; + struct dma_ops_domain *dma_dom; domain = get_domain(dev); if (IS_ERR(domain)) return; - __unmap_single(domain->priv, dma_addr, size, dir); + dma_dom = to_dma_ops_domain(domain); + + __unmap_single(dma_dom, dma_addr, size, dir); +} + +static int sg_num_pages(struct device *dev, + struct scatterlist *sglist, + int nelems) +{ + unsigned long mask, boundary_size; + struct scatterlist *s; + int i, npages = 0; + + mask = dma_get_seg_boundary(dev); + boundary_size = mask + 1 ? ALIGN(mask + 1, PAGE_SIZE) >> PAGE_SHIFT : + 1UL << (BITS_PER_LONG - PAGE_SHIFT); + + for_each_sg(sglist, s, nelems, i) { + int p, n; + + s->dma_address = npages << PAGE_SHIFT; + p = npages % boundary_size; + n = iommu_num_pages(sg_phys(s), s->length, PAGE_SIZE); + if (p + n > boundary_size) + npages += boundary_size - p; + npages += n; + } + + return npages; } /* @@ -2754,46 +2443,79 @@ static void unmap_page(struct device *dev, dma_addr_t dma_addr, size_t size, * lists). */ static int map_sg(struct device *dev, struct scatterlist *sglist, - int nelems, enum dma_data_direction dir, - struct dma_attrs *attrs) + int nelems, enum dma_data_direction direction, + unsigned long attrs) { + int mapped_pages = 0, npages = 0, prot = 0, i; struct protection_domain *domain; - int i; + struct dma_ops_domain *dma_dom; struct scatterlist *s; - phys_addr_t paddr; - int mapped_elems = 0; + unsigned long address; u64 dma_mask; domain = get_domain(dev); if (IS_ERR(domain)) return 0; + dma_dom = to_dma_ops_domain(domain); dma_mask = *dev->dma_mask; + npages = sg_num_pages(dev, sglist, nelems); + + address = dma_ops_alloc_iova(dev, dma_dom, npages, dma_mask); + if (address == DMA_ERROR_CODE) + goto out_err; + + prot = dir2prot(direction); + + /* Map all sg entries */ for_each_sg(sglist, s, nelems, i) { - paddr = sg_phys(s); + int j, pages = iommu_num_pages(sg_phys(s), s->length, PAGE_SIZE); - s->dma_address = __map_single(dev, domain->priv, - paddr, s->length, dir, false, - dma_mask); + for (j = 0; j < pages; ++j) { + unsigned long bus_addr, phys_addr; + int ret; - if (s->dma_address) { - s->dma_length = s->length; - mapped_elems++; - } else - goto unmap; + bus_addr = address + s->dma_address + (j << PAGE_SHIFT); + phys_addr = (sg_phys(s) & PAGE_MASK) + (j << PAGE_SHIFT); + ret = iommu_map_page(domain, bus_addr, phys_addr, PAGE_SIZE, prot, GFP_ATOMIC); + if (ret) + goto out_unmap; + + mapped_pages += 1; + } } - return mapped_elems; + /* Everything is mapped - write the right values into s->dma_address */ + for_each_sg(sglist, s, nelems, i) { + s->dma_address += address + s->offset; + s->dma_length = s->length; + } + + return nelems; + +out_unmap: + pr_err("%s: IOMMU mapping error in map_sg (io-pages: %d)\n", + dev_name(dev), npages); -unmap: - for_each_sg(sglist, s, mapped_elems, i) { - if (s->dma_address) - __unmap_single(domain->priv, s->dma_address, - s->dma_length, dir); - s->dma_address = s->dma_length = 0; + for_each_sg(sglist, s, nelems, i) { + int j, pages = iommu_num_pages(sg_phys(s), s->length, PAGE_SIZE); + + for (j = 0; j < pages; ++j) { + unsigned long bus_addr; + + bus_addr = address + s->dma_address + (j << PAGE_SHIFT); + iommu_unmap_page(domain, bus_addr, PAGE_SIZE); + + if (--mapped_pages) + goto out_free_iova; + } } +out_free_iova: + free_iova_fast(&dma_dom->iovad, address, npages); + +out_err: return 0; } @@ -2803,21 +2525,22 @@ unmap: */ static void unmap_sg(struct device *dev, struct scatterlist *sglist, int nelems, enum dma_data_direction dir, - struct dma_attrs *attrs) + unsigned long attrs) { struct protection_domain *domain; - struct scatterlist *s; - int i; + struct dma_ops_domain *dma_dom; + unsigned long startaddr; + int npages = 2; domain = get_domain(dev); if (IS_ERR(domain)) return; - for_each_sg(sglist, s, nelems, i) { - __unmap_single(domain->priv, s->dma_address, - s->dma_length, dir); - s->dma_address = s->dma_length = 0; - } + startaddr = sg_dma_address(sglist) & PAGE_MASK; + dma_dom = to_dma_ops_domain(domain); + npages = sg_num_pages(dev, sglist, nelems); + + __unmap_single(dma_dom, startaddr, npages << PAGE_SHIFT, dir); } /* @@ -2825,10 +2548,11 @@ static void unmap_sg(struct device *dev, struct scatterlist *sglist, */ static void *alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_addr, gfp_t flag, - struct dma_attrs *attrs) + unsigned long attrs) { u64 dma_mask = dev->coherent_dma_mask; struct protection_domain *domain; + struct dma_ops_domain *dma_dom; struct page *page; domain = get_domain(dev); @@ -2839,6 +2563,7 @@ static void *alloc_coherent(struct device *dev, size_t size, } else if (IS_ERR(domain)) return NULL; + dma_dom = to_dma_ops_domain(domain); size = PAGE_ALIGN(size); dma_mask = dev->coherent_dma_mask; flag &= ~(__GFP_DMA | __GFP_HIGHMEM | __GFP_DMA32); @@ -2858,8 +2583,8 @@ static void *alloc_coherent(struct device *dev, size_t size, if (!dma_mask) dma_mask = *dev->dma_mask; - *dma_addr = __map_single(dev, domain->priv, page_to_phys(page), - size, DMA_BIDIRECTIONAL, true, dma_mask); + *dma_addr = __map_single(dev, dma_dom, page_to_phys(page), + size, DMA_BIDIRECTIONAL, dma_mask); if (*dma_addr == DMA_ERROR_CODE) goto out_free; @@ -2879,9 +2604,10 @@ out_free: */ static void free_coherent(struct device *dev, size_t size, void *virt_addr, dma_addr_t dma_addr, - struct dma_attrs *attrs) + unsigned long attrs) { struct protection_domain *domain; + struct dma_ops_domain *dma_dom; struct page *page; page = virt_to_page(virt_addr); @@ -2891,7 +2617,9 @@ static void free_coherent(struct device *dev, size_t size, if (IS_ERR(domain)) goto free_mem; - __unmap_single(domain->priv, dma_addr, size, DMA_BIDIRECTIONAL); + dma_dom = to_dma_ops_domain(domain); + + __unmap_single(dma_dom, dma_addr, size, DMA_BIDIRECTIONAL); free_mem: if (!dma_release_from_contiguous(dev, page, size >> PAGE_SHIFT)) @@ -2907,48 +2635,92 @@ static int amd_iommu_dma_supported(struct device *dev, u64 mask) return check_device(dev); } -static int set_dma_mask(struct device *dev, u64 mask) +static struct dma_map_ops amd_iommu_dma_ops = { + .alloc = alloc_coherent, + .free = free_coherent, + .map_page = map_page, + .unmap_page = unmap_page, + .map_sg = map_sg, + .unmap_sg = unmap_sg, + .dma_supported = amd_iommu_dma_supported, +}; + +static int init_reserved_iova_ranges(void) { - struct protection_domain *domain; - int max_apertures = 1; + struct pci_dev *pdev = NULL; + struct iova *val; - domain = get_domain(dev); - if (IS_ERR(domain)) - return PTR_ERR(domain); + init_iova_domain(&reserved_iova_ranges, PAGE_SIZE, + IOVA_START_PFN, DMA_32BIT_PFN); - if (mask == DMA_BIT_MASK(64)) - max_apertures = 8; - else if (mask > DMA_BIT_MASK(32)) - max_apertures = 4; + lockdep_set_class(&reserved_iova_ranges.iova_rbtree_lock, + &reserved_rbtree_key); + + /* MSI memory range */ + val = reserve_iova(&reserved_iova_ranges, + IOVA_PFN(MSI_RANGE_START), IOVA_PFN(MSI_RANGE_END)); + if (!val) { + pr_err("Reserving MSI range failed\n"); + return -ENOMEM; + } + + /* HT memory range */ + val = reserve_iova(&reserved_iova_ranges, + IOVA_PFN(HT_RANGE_START), IOVA_PFN(HT_RANGE_END)); + if (!val) { + pr_err("Reserving HT range failed\n"); + return -ENOMEM; + } /* - * To prevent lock contention it doesn't make sense to allocate more - * apertures than online cpus + * Memory used for PCI resources + * FIXME: Check whether we can reserve the PCI-hole completly */ - if (max_apertures > num_online_cpus()) - max_apertures = num_online_cpus(); + for_each_pci_dev(pdev) { + int i; + + for (i = 0; i < PCI_NUM_RESOURCES; ++i) { + struct resource *r = &pdev->resource[i]; + + if (!(r->flags & IORESOURCE_MEM)) + continue; - if (dma_ops_domain_alloc_apertures(domain->priv, max_apertures)) - dev_err(dev, "Can't allocate %d iommu apertures\n", - max_apertures); + val = reserve_iova(&reserved_iova_ranges, + IOVA_PFN(r->start), + IOVA_PFN(r->end)); + if (!val) { + pr_err("Reserve pci-resource range failed\n"); + return -ENOMEM; + } + } + } return 0; } -static struct dma_map_ops amd_iommu_dma_ops = { - .alloc = alloc_coherent, - .free = free_coherent, - .map_page = map_page, - .unmap_page = unmap_page, - .map_sg = map_sg, - .unmap_sg = unmap_sg, - .dma_supported = amd_iommu_dma_supported, - .set_dma_mask = set_dma_mask, -}; - int __init amd_iommu_init_api(void) { - int err = 0; + int ret, cpu, err = 0; + + ret = iova_cache_get(); + if (ret) + return ret; + + ret = init_reserved_iova_ranges(); + if (ret) + return ret; + + for_each_possible_cpu(cpu) { + struct flush_queue *queue = per_cpu_ptr(&flush_queue, cpu); + + queue->entries = kzalloc(FLUSH_QUEUE_SIZE * + sizeof(*queue->entries), + GFP_KERNEL); + if (!queue->entries) + goto out_put_iova; + + spin_lock_init(&queue->lock); + } err = bus_set_iommu(&pci_bus_type, &amd_iommu_ops); if (err) @@ -2958,11 +2730,26 @@ int __init amd_iommu_init_api(void) if (err) return err; #endif + err = bus_set_iommu(&platform_bus_type, &amd_iommu_ops); + if (err) + return err; return 0; + +out_put_iova: + for_each_possible_cpu(cpu) { + struct flush_queue *queue = per_cpu_ptr(&flush_queue, cpu); + + kfree(queue->entries); + } + + return -ENOMEM; } int __init amd_iommu_init_dma_ops(void) { + setup_timer(&queue_timer, queue_flush_timeout, 0); + atomic_set(&queue_timer_on, 0); + swiotlb = iommu_pass_through ? 1 : 0; iommu_detected = 1; @@ -2981,6 +2768,7 @@ int __init amd_iommu_init_dma_ops(void) pr_info("AMD-Vi: Lazy IO/TLB flushing enabled\n"); return 0; + } /***************************************************************************** @@ -3103,9 +2891,7 @@ static struct iommu_domain *amd_iommu_domain_alloc(unsigned type) static void amd_iommu_domain_free(struct iommu_domain *dom) { struct protection_domain *domain; - - if (!dom) - return; + struct dma_ops_domain *dma_dom; domain = to_pdomain(dom); @@ -3114,13 +2900,31 @@ static void amd_iommu_domain_free(struct iommu_domain *dom) BUG_ON(domain->dev_cnt != 0); - if (domain->mode != PAGE_MODE_NONE) - free_pagetable(domain); + if (!dom) + return; + + switch (dom->type) { + case IOMMU_DOMAIN_DMA: + /* + * First make sure the domain is no longer referenced from the + * flush queue + */ + queue_flush_all(); - if (domain->flags & PD_IOMMUV2_MASK) - free_gcr3_table(domain); + /* Now release the domain */ + dma_dom = to_dma_ops_domain(domain); + dma_ops_domain_free(dma_dom); + break; + default: + if (domain->mode != PAGE_MODE_NONE) + free_pagetable(domain); + + if (domain->flags & PD_IOMMUV2_MASK) + free_gcr3_table(domain); - protection_domain_free(domain); + protection_domain_free(domain); + break; + } } static void amd_iommu_detach_device(struct iommu_domain *dom, @@ -3190,7 +2994,7 @@ static int amd_iommu_map(struct iommu_domain *dom, unsigned long iova, prot |= IOMMU_PROT_IW; mutex_lock(&domain->api_lock); - ret = iommu_map_page(domain, iova, paddr, prot, page_size); + ret = iommu_map_page(domain, iova, paddr, page_size, prot, GFP_KERNEL); mutex_unlock(&domain->api_lock); return ret; @@ -3292,6 +3096,19 @@ static void amd_iommu_put_dm_regions(struct device *dev, kfree(entry); } +static void amd_iommu_apply_dm_region(struct device *dev, + struct iommu_domain *domain, + struct iommu_dm_region *region) +{ + struct dma_ops_domain *dma_dom = to_dma_ops_domain(to_pdomain(domain)); + unsigned long start, end; + + start = IOVA_PFN(region->start); + end = IOVA_PFN(region->start + region->length); + + WARN_ON_ONCE(reserve_iova(&dma_dom->iovad, start, end) == NULL); +} + static const struct iommu_ops amd_iommu_ops = { .capable = amd_iommu_capable, .domain_alloc = amd_iommu_domain_alloc, @@ -3307,6 +3124,7 @@ static const struct iommu_ops amd_iommu_ops = { .device_group = amd_iommu_device_group, .get_dm_regions = amd_iommu_get_dm_regions, .put_dm_regions = amd_iommu_put_dm_regions, + .apply_dm_region = amd_iommu_apply_dm_region, .pgsize_bitmap = AMD_IOMMU_PGSIZES, };