diff options
Diffstat (limited to 'drivers/pci')
| -rw-r--r-- | drivers/pci/hotplug/pci_hotplug_core.c | 2 | ||||
| -rw-r--r-- | drivers/pci/intel-iommu.c | 791 | ||||
| -rw-r--r-- | drivers/pci/iova.c | 26 | ||||
| -rw-r--r-- | drivers/pci/msi.c | 64 | ||||
| -rw-r--r-- | drivers/pci/msi.h | 10 | ||||
| -rw-r--r-- | drivers/pci/pci.c | 15 | ||||
| -rw-r--r-- | drivers/pci/pcie/aer/ecrc.c | 2 | ||||
| -rw-r--r-- | drivers/pci/quirks.c | 5 | ||||
| -rw-r--r-- | drivers/pci/slot.c | 4 |
9 files changed, 519 insertions, 400 deletions
diff --git a/drivers/pci/hotplug/pci_hotplug_core.c b/drivers/pci/hotplug/pci_hotplug_core.c index 844580489d4..5c5043f239c 100644 --- a/drivers/pci/hotplug/pci_hotplug_core.c +++ b/drivers/pci/hotplug/pci_hotplug_core.c @@ -555,6 +555,8 @@ static struct hotplug_slot *get_slot_from_name (const char *name) * @slot: pointer to the &struct hotplug_slot to register * @devnr: device number * @name: name registered with kobject core + * @owner: caller module owner + * @mod_name: caller module name * * Registers a hotplug slot with the pci hotplug subsystem, which will allow * userspace interaction to the slot. diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c index e53eacd75c8..ebc9b8dca88 100644 --- a/drivers/pci/intel-iommu.c +++ b/drivers/pci/intel-iommu.c @@ -39,7 +39,6 @@ #include <linux/sysdev.h> #include <asm/cacheflush.h> #include <asm/iommu.h> -#include <asm/e820.h> #include "pci.h" #define ROOT_SIZE VTD_PAGE_SIZE @@ -57,14 +56,32 @@ #define MAX_AGAW_WIDTH 64 #define DOMAIN_MAX_ADDR(gaw) ((((u64)1) << gaw) - 1) +#define DOMAIN_MAX_PFN(gaw) ((((u64)1) << (gaw-VTD_PAGE_SHIFT)) - 1) #define IOVA_PFN(addr) ((addr) >> PAGE_SHIFT) #define DMA_32BIT_PFN IOVA_PFN(DMA_BIT_MASK(32)) #define DMA_64BIT_PFN IOVA_PFN(DMA_BIT_MASK(64)) -#ifndef PHYSICAL_PAGE_MASK -#define PHYSICAL_PAGE_MASK PAGE_MASK -#endif + +/* VT-d pages must always be _smaller_ than MM pages. Otherwise things + are never going to work. */ +static inline unsigned long dma_to_mm_pfn(unsigned long dma_pfn) +{ + return dma_pfn >> (PAGE_SHIFT - VTD_PAGE_SHIFT); +} + +static inline unsigned long mm_to_dma_pfn(unsigned long mm_pfn) +{ + return mm_pfn << (PAGE_SHIFT - VTD_PAGE_SHIFT); +} +static inline unsigned long page_to_dma_pfn(struct page *pg) +{ + return mm_to_dma_pfn(page_to_pfn(pg)); +} +static inline unsigned long virt_to_dma_pfn(void *p) +{ + return page_to_dma_pfn(virt_to_page(p)); +} /* global iommu list, set NULL for ignored DMAR units */ static struct intel_iommu **g_iommus; @@ -205,12 +222,17 @@ static inline void dma_set_pte_prot(struct dma_pte *pte, unsigned long prot) static inline u64 dma_pte_addr(struct dma_pte *pte) { - return (pte->val & VTD_PAGE_MASK); +#ifdef CONFIG_64BIT + return pte->val & VTD_PAGE_MASK; +#else + /* Must have a full atomic 64-bit read */ + return __cmpxchg64(pte, 0ULL, 0ULL) & VTD_PAGE_MASK; +#endif } -static inline void dma_set_pte_addr(struct dma_pte *pte, u64 addr) +static inline void dma_set_pte_pfn(struct dma_pte *pte, unsigned long pfn) { - pte->val |= (addr & VTD_PAGE_MASK); + pte->val |= (uint64_t)pfn << VTD_PAGE_SHIFT; } static inline bool dma_pte_present(struct dma_pte *pte) @@ -218,6 +240,11 @@ static inline bool dma_pte_present(struct dma_pte *pte) return (pte->val & 3) != 0; } +static inline int first_pte_in_page(struct dma_pte *pte) +{ + return !((unsigned long)pte & ~VTD_PAGE_MASK); +} + /* * This domain is a statically identity mapping domain. * 1. This domain creats a static 1:1 mapping to all usable memory. @@ -245,7 +272,6 @@ struct dmar_domain { struct iova_domain iovad; /* iova's that belong to this domain */ struct dma_pte *pgd; /* virtual address */ - spinlock_t mapping_lock; /* page table lock */ int gaw; /* max guest address width */ /* adjusted guest address width, 0 is level 2 30-bit */ @@ -649,80 +675,78 @@ static inline int width_to_agaw(int width) static inline unsigned int level_to_offset_bits(int level) { - return (12 + (level - 1) * LEVEL_STRIDE); + return (level - 1) * LEVEL_STRIDE; } -static inline int address_level_offset(u64 addr, int level) +static inline int pfn_level_offset(unsigned long pfn, int level) { - return ((addr >> level_to_offset_bits(level)) & LEVEL_MASK); + return (pfn >> level_to_offset_bits(level)) & LEVEL_MASK; } -static inline u64 level_mask(int level) +static inline unsigned long level_mask(int level) { - return ((u64)-1 << level_to_offset_bits(level)); + return -1UL << level_to_offset_bits(level); } -static inline u64 level_size(int level) +static inline unsigned long level_size(int level) { - return ((u64)1 << level_to_offset_bits(level)); + return 1UL << level_to_offset_bits(level); } -static inline u64 align_to_level(u64 addr, int level) +static inline unsigned long align_to_level(unsigned long pfn, int level) { - return ((addr + level_size(level) - 1) & level_mask(level)); + return (pfn + level_size(level) - 1) & level_mask(level); } -static struct dma_pte * addr_to_dma_pte(struct dmar_domain *domain, u64 addr) +static struct dma_pte *pfn_to_dma_pte(struct dmar_domain *domain, + unsigned long pfn) { - int addr_width = agaw_to_width(domain->agaw); + int addr_width = agaw_to_width(domain->agaw) - VTD_PAGE_SHIFT; struct dma_pte *parent, *pte = NULL; int level = agaw_to_level(domain->agaw); int offset; - unsigned long flags; BUG_ON(!domain->pgd); - - addr &= (((u64)1) << addr_width) - 1; + BUG_ON(addr_width < BITS_PER_LONG && pfn >> addr_width); parent = domain->pgd; - spin_lock_irqsave(&domain->mapping_lock, flags); while (level > 0) { void *tmp_page; - offset = address_level_offset(addr, level); + offset = pfn_level_offset(pfn, level); pte = &parent[offset]; if (level == 1) break; if (!dma_pte_present(pte)) { + uint64_t pteval; + tmp_page = alloc_pgtable_page(); - if (!tmp_page) { - spin_unlock_irqrestore(&domain->mapping_lock, - flags); + if (!tmp_page) return NULL; + + domain_flush_cache(domain, tmp_page, VTD_PAGE_SIZE); + pteval = (virt_to_dma_pfn(tmp_page) << VTD_PAGE_SHIFT) | DMA_PTE_READ | DMA_PTE_WRITE; + if (cmpxchg64(&pte->val, 0ULL, pteval)) { + /* Someone else set it while we were thinking; use theirs. */ + free_pgtable_page(tmp_page); + } else { + dma_pte_addr(pte); + domain_flush_cache(domain, pte, sizeof(*pte)); } - domain_flush_cache(domain, tmp_page, PAGE_SIZE); - dma_set_pte_addr(pte, virt_to_phys(tmp_page)); - /* - * high level table always sets r/w, last level page - * table control read/write - */ - dma_set_pte_readable(pte); - dma_set_pte_writable(pte); - domain_flush_cache(domain, pte, sizeof(*pte)); } parent = phys_to_virt(dma_pte_addr(pte)); level--; } - spin_unlock_irqrestore(&domain->mapping_lock, flags); return pte; } /* return address's pte at specific level */ -static struct dma_pte *dma_addr_level_pte(struct dmar_domain *domain, u64 addr, - int level) +static struct dma_pte *dma_pfn_level_pte(struct dmar_domain *domain, + unsigned long pfn, + int level) { struct dma_pte *parent, *pte = NULL; int total = agaw_to_level(domain->agaw); @@ -730,7 +754,7 @@ static struct dma_pte *dma_addr_level_pte(struct dmar_domain *domain, u64 addr, parent = domain->pgd; while (level <= total) { - offset = address_level_offset(addr, total); + offset = pfn_level_offset(pfn, total); pte = &parent[offset]; if (level == total) return pte; @@ -743,74 +767,82 @@ static struct dma_pte *dma_addr_level_pte(struct dmar_domain *domain, u64 addr, return NULL; } -/* clear one page's page table */ -static void dma_pte_clear_one(struct dmar_domain *domain, u64 addr) -{ - struct dma_pte *pte = NULL; - - /* get last level pte */ - pte = dma_addr_level_pte(domain, addr, 1); - - if (pte) { - dma_clear_pte(pte); - domain_flush_cache(domain, pte, sizeof(*pte)); - } -} - /* clear last level pte, a tlb flush should be followed */ -static void dma_pte_clear_range(struct dmar_domain *domain, u64 start, u64 end) +static void dma_pte_clear_range(struct dmar_domain *domain, + unsigned long start_pfn, + unsigned long last_pfn) { - int addr_width = agaw_to_width(domain->agaw); - int npages; + int addr_width = agaw_to_width(domain->agaw) - VTD_PAGE_SHIFT; + struct dma_pte *first_pte, *pte; - start &= (((u64)1) << addr_width) - 1; - end &= (((u64)1) << addr_width) - 1; - /* in case it's partial page */ - start &= PAGE_MASK; - end = PAGE_ALIGN(end); - npages = (end - start) / VTD_PAGE_SIZE; + BUG_ON(addr_width < BITS_PER_LONG && start_pfn >> addr_width); + BUG_ON(addr_width < BITS_PER_LONG && last_pfn >> addr_width); - /* we don't need lock here, nobody else touches the iova range */ - while (npages--) { - dma_pte_clear_one(domain, start); - start += VTD_PAGE_SIZE; + /* we don't need lock here; nobody else touches the iova range */ + while (start_pfn <= last_pfn) { + first_pte = pte = dma_pfn_level_pte(domain, start_pfn, 1); + if (!pte) { + start_pfn = align_to_level(start_pfn + 1, 2); + continue; + } + do { + dma_clear_pte(pte); + start_pfn++; + pte++; + } while (start_pfn <= last_pfn && !first_pte_in_page(pte)); + + domain_flush_cache(domain, first_pte, + (void *)pte - (void *)first_pte); } } /* free page table pages. last level pte should already be cleared */ static void dma_pte_free_pagetable(struct dmar_domain *domain, - u64 start, u64 end) + unsigned long start_pfn, + unsigned long last_pfn) { - int addr_width = agaw_to_width(domain->agaw); - struct dma_pte *pte; + int addr_width = agaw_to_width(domain->agaw) - VTD_PAGE_SHIFT; + struct dma_pte *first_pte, *pte; int total = agaw_to_level(domain->agaw); int level; - u64 tmp; + unsigned long tmp; - start &= (((u64)1) << addr_width) - 1; - end &= (((u64)1) << addr_width) - 1; + BUG_ON(addr_width < BITS_PER_LONG && start_pfn >> addr_width); + BUG_ON(addr_width < BITS_PER_LONG && last_pfn >> addr_width); - /* we don't need lock here, nobody else touches the iova range */ + /* We don't need lock here; nobody else touches the iova range */ level = 2; while (level <= total) { - tmp = align_to_level(start, level); - if (tmp >= end || (tmp + level_size(level) > end)) + tmp = align_to_level(start_pfn, level); + + /* If we can't even clear one PTE at this level, we're done */ + if (tmp + level_size(level) - 1 > last_pfn) return; - while (tmp < end) { - pte = dma_addr_level_pte(domain, tmp, level); - if (pte) { - free_pgtable_page( - phys_to_virt(dma_pte_addr(pte))); - dma_clear_pte(pte); - domain_flush_cache(domain, pte, sizeof(*pte)); + while (tmp + level_size(level) - 1 <= last_pfn) { + first_pte = pte = dma_pfn_level_pte(domain, tmp, level); + if (!pte) { + tmp = align_to_level(tmp + 1, level + 1); + continue; } - tmp += level_size(level); + do { + if (dma_pte_present(pte)) { + free_pgtable_page(phys_to_virt(dma_pte_addr(pte))); + dma_clear_pte(pte); + } + pte++; + tmp += level_size(level); + } while (!first_pte_in_page(pte) && + tmp + level_size(level) - 1 <= last_pfn); + + domain_flush_cache(domain, first_pte, + (void *)pte - (void *)first_pte); + } level++; } /* free pgd */ - if (start == 0 && end >= ((((u64)1) << addr_width) - 1)) { + if (start_pfn == 0 && last_pfn == DOMAIN_MAX_PFN(domain->gaw)) { free_pgtable_page(domain->pgd); domain->pgd = NULL; } @@ -1036,11 +1068,11 @@ static void iommu_flush_dev_iotlb(struct dmar_domain *domain, } static void iommu_flush_iotlb_psi(struct intel_iommu *iommu, u16 did, - u64 addr, unsigned int pages) + unsigned long pfn, unsigned int pages) { unsigned int mask = ilog2(__roundup_pow_of_two(pages)); + uint64_t addr = (uint64_t)pfn << VTD_PAGE_SHIFT; - BUG_ON(addr & (~VTD_PAGE_MASK)); BUG_ON(pages == 0); /* @@ -1055,7 +1087,12 @@ static void iommu_flush_iotlb_psi(struct intel_iommu *iommu, u16 did, else iommu->flush.flush_iotlb(iommu, did, addr, mask, DMA_TLB_PSI_FLUSH); - if (did) + + /* + * In caching mode, domain ID 0 is reserved for non-present to present + * mapping flush. Device IOTLB doesn't need to be flushed in this case. + */ + if (!cap_caching_mode(iommu->cap) || did) iommu_flush_dev_iotlb(iommu->domains[did], addr, mask); } @@ -1280,7 +1317,6 @@ static void dmar_init_reserved_ranges(void) struct pci_dev *pdev = NULL; struct iova *iova; int i; - u64 addr, size; init_iova_domain(&reserved_iova_list, DMA_32BIT_PFN); @@ -1303,12 +1339,9 @@ static void dmar_init_reserved_ranges(void) r = &pdev->resource[i]; if (!r->flags || !(r->flags & IORESOURCE_MEM)) continue; - addr = r->start; - addr &= PHYSICAL_PAGE_MASK; - size = r->end - addr; - size = PAGE_ALIGN(size); - iova = reserve_iova(&reserved_iova_list, IOVA_PFN(addr), - IOVA_PFN(size + addr) - 1); + iova = reserve_iova(&reserved_iova_list, + IOVA_PFN(r->start), + IOVA_PFN(r->end)); if (!iova) printk(KERN_ERR "Reserve iova failed\n"); } @@ -1342,7 +1375,6 @@ static int domain_init(struct dmar_domain *domain, int guest_width) unsigned long sagaw; init_iova_domain(&domain->iovad, DMA_32BIT_PFN); - spin_lock_init(&domain->mapping_lock); spin_lock_init(&domain->iommu_lock); domain_reserve_special_ranges(domain); @@ -1389,7 +1421,6 @@ static void domain_exit(struct dmar_domain *domain) { struct dmar_drhd_unit *drhd; struct intel_iommu *iommu; - u64 end; /* Domain 0 is reserved, so dont process it */ if (!domain) @@ -1398,14 +1429,12 @@ static void domain_exit(struct dmar_domain *domain) domain_remove_dev_info(domain); /* destroy iovas */ put_iova_domain(&domain->iovad); - end = DOMAIN_MAX_ADDR(domain->gaw); - end = end & (~PAGE_MASK); /* clear ptes */ - dma_pte_clear_range(domain, 0, end); + dma_pte_clear_range(domain, 0, DOMAIN_MAX_PFN(domain->gaw)); /* free page tables */ - dma_pte_free_pagetable(domain, 0, end); + dma_pte_free_pagetable(domain, 0, DOMAIN_MAX_PFN(domain->gaw)); for_each_active_iommu(iommu, drhd) if (test_bit(iommu->seq_id, &domain->iommu_bmp)) @@ -1619,42 +1648,86 @@ static int domain_context_mapped(struct pci_dev *pdev) tmp->devfn); } -static int -domain_page_mapping(struct dmar_domain *domain, dma_addr_t iova, - u64 hpa, size_t size, int prot) +static int __domain_mapping(struct dmar_domain *domain, unsigned long iov_pfn, + struct scatterlist *sg, unsigned long phys_pfn, + unsigned long nr_pages, int prot) { - u64 start_pfn, end_pfn; - struct dma_pte *pte; - int index; - int addr_width = agaw_to_width(domain->agaw); + struct dma_pte *first_pte = NULL, *pte = NULL; + phys_addr_t uninitialized_var(pteval); + int addr_width = agaw_to_width(domain->agaw) - VTD_PAGE_SHIFT; + unsigned long sg_res; - hpa &= (((u64)1) << addr_width) - 1; + BUG_ON(addr_width < BITS_PER_LONG && (iov_pfn + nr_pages - 1) >> addr_width); if ((prot & (DMA_PTE_READ|DMA_PTE_WRITE)) == 0) return -EINVAL; - iova &= PAGE_MASK; - start_pfn = ((u64)hpa) >> VTD_PAGE_SHIFT; - end_pfn = (VTD_PAGE_ALIGN(((u64)hpa) + size)) >> VTD_PAGE_SHIFT; - index = 0; - while (start_pfn < end_pfn) { - pte = addr_to_dma_pte(domain, iova + VTD_PAGE_SIZE * index); - if (!pte) - return -ENOMEM; + + prot &= DMA_PTE_READ | DMA_PTE_WRITE | DMA_PTE_SNP; + + if (sg) + sg_res = 0; + else { + sg_res = nr_pages + 1; + pteval = ((phys_addr_t)phys_pfn << VTD_PAGE_SHIFT) | prot; + } + + while (nr_pages--) { + uint64_t tmp; + + if (!sg_res) { + sg_res = (sg->offset + sg->length + VTD_PAGE_SIZE - 1) >> VTD_PAGE_SHIFT; + sg->dma_address = ((dma_addr_t)iov_pfn << VTD_PAGE_SHIFT) + sg->offset; + sg->dma_length = sg->length; + pteval = page_to_phys(sg_page(sg)) | prot; + } + if (!pte) { + first_pte = pte = pfn_to_dma_pte(domain, iov_pfn); + if (!pte) + return -ENOMEM; + } /* We don't need lock here, nobody else * touches the iova range */ - BUG_ON(dma_pte_addr(pte)); - dma_set_pte_addr(pte, start_pfn << VTD_PAGE_SHIFT); - dma_set_pte_prot(pte, prot); - if (prot & DMA_PTE_SNP) - dma_set_pte_snp(pte); - domain_flush_cache(domain, pte, sizeof(*pte)); - start_pfn++; - index++; + tmp = cmpxchg64_local(&pte->val, 0ULL, pteval); + if (tmp) { + static int dumps = 5; + printk(KERN_CRIT "ERROR: DMA PTE for vPFN 0x%lx already set (to %llx not %llx)\n", + iov_pfn, tmp, (unsigned long long)pteval); + if (dumps) { + dumps--; + debug_dma_dump_mappings(NULL); + } + WARN_ON(1); + } + pte++; + if (!nr_pages || first_pte_in_page(pte)) { + domain_flush_cache(domain, first_pte, + (void *)pte - (void *)first_pte); + pte = NULL; + } + iov_pfn++; + pteval += VTD_PAGE_SIZE; + sg_res--; + if (!sg_res) + sg = sg_next(sg); } return 0; } +static inline int domain_sg_mapping(struct dmar_domain *domain, unsigned long iov_pfn, + struct scatterlist *sg, unsigned long nr_pages, + int prot) +{ + return __domain_mapping(domain, iov_pfn, sg, 0, nr_pages, prot); +} + +static inline int domain_pfn_mapping(struct dmar_domain *domain, unsigned long iov_pfn, + unsigned long phys_pfn, unsigned long nr_pages, + int prot) +{ + return __domain_mapping(domain, iov_pfn, NULL, phys_pfn, nr_pages, prot); +} + static void iommu_detach_dev(struct intel_iommu *iommu, u8 bus, u8 devfn) { if (!iommu) @@ -1845,58 +1918,61 @@ error: static int iommu_identity_mapping; +static int iommu_domain_identity_map(struct dmar_domain *domain, + unsigned long long start, + unsigned long long end) +{ + unsigned long first_vpfn = start >> VTD_PAGE_SHIFT; + unsigned long last_vpfn = end >> VTD_PAGE_SHIFT; + + if (!reserve_iova(&domain->iovad, dma_to_mm_pfn(first_vpfn), + dma_to_mm_pfn(last_vpfn))) { + printk(KERN_ERR "IOMMU: reserve iova failed\n"); + return -ENOMEM; + } + + pr_debug("Mapping reserved region %llx-%llx for domain %d\n", + start, end, domain->id); + /* + * RMRR range might have overlap with physical memory range, + * clear it first + */ + dma_pte_clear_range(domain, first_vpfn, last_vpfn); + + return domain_pfn_mapping(domain, first_vpfn, first_vpfn, + last_vpfn - first_vpfn + 1, + DMA_PTE_READ|DMA_PTE_WRITE); +} + static int iommu_prepare_identity_map(struct pci_dev *pdev, unsigned long long start, unsigned long long end) { struct dmar_domain *domain; - unsigned long size; - unsigned long long base; int ret; printk(KERN_INFO - "IOMMU: Setting identity map for device %s [0x%Lx - 0x%Lx]\n", - pci_name(pdev), start, end); - if (iommu_identity_mapping) - domain = si_domain; - else - /* page table init */ - domain = get_domain_for_dev(pdev, DEFAULT_DOMAIN_ADDRESS_WIDTH); + "IOMMU: Setting identity map for device %s [0x%Lx - 0x%Lx]\n", + pci_name(pdev), start, end); + + domain = get_domain_for_dev(pdev, DEFAULT_DOMAIN_ADDRESS_WIDTH); if (!domain) return -ENOMEM; - /* The address might not be aligned */ - base = start & PAGE_MASK; - size = end - base; - size = PAGE_ALIGN(size); - if (!reserve_iova(&domain->iovad, IOVA_PFN(base), - IOVA_PFN(base + size) - 1)) { - printk(KERN_ERR "IOMMU: reserve iova failed\n"); - ret = -ENOMEM; - goto error; - } - - pr_debug("Mapping reserved region %lx@%llx for %s\n", - size, base, pci_name(pdev)); - /* - * RMRR range might have overlap with physical memory range, - * clear it first - */ - dma_pte_clear_range(domain, base, base + size); - - ret = domain_page_mapping(domain, base, base, size, - DMA_PTE_READ|DMA_PTE_WRITE); + ret = iommu_domain_identity_map(domain, start, end); if (ret) goto error; /* context entry init */ ret = domain_context_mapping(domain, pdev, CONTEXT_TT_MULTI_LEVEL); - if (!ret) - return 0; -error: + if (ret) + goto error; + + return 0; + + error: domain_exit(domain); return ret; - } static inline int iommu_prepare_rmrr_dev(struct dmar_rmrr_unit *rmrr, @@ -1908,64 +1984,6 @@ static inline int iommu_prepare_rmrr_dev(struct dmar_rmrr_unit *rmrr, rmrr->end_address + 1); } -#ifdef CONFIG_DMAR_GFX_WA -struct iommu_prepare_data { - struct pci_dev *pdev; - int ret; -}; - -static int __init iommu_prepare_work_fn(unsigned long start_pfn, - unsigned long end_pfn, void *datax) -{ - struct iommu_prepare_data *data; - - data = (struct iommu_prepare_data *)datax; - - data->ret = iommu_prepare_identity_map(data->pdev, - start_pfn<<PAGE_SHIFT, end_pfn<<PAGE_SHIFT); - return data->ret; - -} - -static int __init iommu_prepare_with_active_regions(struct pci_dev *pdev) -{ - int nid; - struct iommu_prepare_data data; - - data.pdev = pdev; - data.ret = 0; - - for_each_online_node(nid) { - work_with_active_regions(nid, iommu_prepare_work_fn, &data); - if (data.ret) - return data.ret; - } - return data.ret; -} - -static void __init iommu_prepare_gfx_mapping(void) -{ - struct pci_dev *pdev = NULL; - int ret; - - for_each_pci_dev(pdev) { - if (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO || - !IS_GFX_DEVICE(pdev)) - continue; - printk(KERN_INFO "IOMMU: gfx device %s 1-1 mapping\n", - pci_name(pdev)); - ret = iommu_prepare_with_active_regions(pdev); - if (ret) - printk(KERN_ERR "IOMMU: mapping reserved region failed\n"); - } -} -#else /* !CONFIG_DMAR_GFX_WA */ -static inline void iommu_prepare_gfx_mapping(void) -{ - return; -} -#endif - #ifdef CONFIG_DMAR_FLOPPY_WA static inline void iommu_prepare_isa(void) { @@ -1976,12 +1994,12 @@ static inline void iommu_prepare_isa(void) if (!pdev) return; - printk(KERN_INFO "IOMMU: Prepare 0-16M unity mapping for LPC\n"); + printk(KERN_INFO "IOMMU: Prepare 0-16MiB unity mapping for LPC\n"); ret = iommu_prepare_identity_map(pdev, 0, 16*1024*1024); if (ret) - printk(KERN_ERR "IOMMU: Failed to create 0-64M identity map, " - "floppy might not work\n"); + printk(KERN_ERR "IOMMU: Failed to create 0-16MiB identity map; " + "floppy might not work\n"); } #else @@ -2009,16 +2027,30 @@ static int __init init_context_pass_through(void) } static int md_domain_init(struct dmar_domain *domain, int guest_width); + +static int __init si_domain_work_fn(unsigned long start_pfn, + unsigned long end_pfn, void *datax) +{ + int *ret = datax; + + *ret = iommu_domain_identity_map(si_domain, + (uint64_t)start_pfn << PAGE_SHIFT, + (uint64_t)end_pfn << PAGE_SHIFT); + return *ret; + +} + static int si_domain_init(void) { struct dmar_drhd_unit *drhd; struct intel_iommu *iommu; - int ret = 0; + int nid, ret = 0; si_domain = alloc_domain(); if (!si_domain) return -EFAULT; + pr_debug("Identity mapping domain is domain %d\n", si_domain->id); for_each_active_iommu(iommu, drhd) { ret = iommu_attach_domain(si_domain, iommu); @@ -2035,6 +2067,12 @@ static int si_domain_init(void) si_domain->flags = DOMAIN_FLAG_STATIC_IDENTITY; + for_each_online_node(nid) { + work_with_active_regions(nid, si_domain_work_fn, &ret); + if (ret) + return ret; + } + return 0; } @@ -2079,9 +2117,49 @@ static int domain_add_dev_info(struct dmar_domain *domain, return 0; } +static int iommu_should_identity_map(struct pci_dev *pdev, int startup) +{ + if (iommu_identity_mapping == 2) + return IS_GFX_DEVICE(pdev); + + /* + * We want to start off with all devices in the 1:1 domain, and + * take them out later if we find they can't access all of memory. + * + * However, we can't do this for PCI devices behind bridges, + * because all PCI devices behind the same bridge will end up + * with the same source-id on their transactions. + * + * Practically speaking, we can't change things around for these + * devices at run-time, because we can't be sure there'll be no + * DMA transactions in flight for any of their siblings. + * + * So PCI devices (unless they're on the root bus) as well as + * their parent PCI-PCI or PCIe-PCI bridges must be left _out_ of + * the 1:1 domain, just in _case_ one of their siblings turns out + * not to be able to map all of memory. + */ + if (!pdev->is_pcie) { + if (!pci_is_root_bus(pdev->bus)) + return 0; + if (pdev->class >> 8 == PCI_CLASS_BRIDGE_PCI) + return 0; + } else if (pdev->pcie_type == PCI_EXP_TYPE_PCI_BRIDGE) + return 0; + + /* + * At boot time, we don't yet know if devices will be 64-bit capable. + * Assume that they will -- if they turn out not to be, then we can + * take them out of the 1:1 domain later. + */ + if (!startup) + return pdev->dma_mask > DMA_BIT_MASK(32); + + return 1; +} + static int iommu_prepare_static_identity_mapping(void) { - int i; struct pci_dev *pdev = NULL; int ret; @@ -2089,23 +2167,19 @@ static int iommu_prepare_static_identity_mapping(void) if (ret) return -EFAULT; - printk(KERN_INFO "IOMMU: Setting identity map:\n"); for_each_pci_dev(pdev) { - for (i = 0; i < e820.nr_map; i++) { - struct e820entry *ei = &e820.map[i]; - - if (ei->type == E820_RAM) { - ret = iommu_prepare_identity_map(pdev, - ei->addr, ei->addr + ei->size); - if (ret) { - printk(KERN_INFO "1:1 mapping to one domain failed.\n"); - return -EFAULT; - } - } + if (iommu_should_identity_map(pdev, 1)) { + printk(KERN_INFO "IOMMU: identity mapping for device %s\n", + pci_name(pdev)); + + ret = domain_context_mapping(si_domain, pdev, + CONTEXT_TT_MULTI_LEVEL); + if (ret) + return ret; + ret = domain_add_dev_info(si_domain, pdev); + if (ret) + return ret; } - ret = domain_add_dev_info(si_domain, pdev); - if (ret) - return ret; } return 0; @@ -2260,6 +2334,10 @@ int __init init_dmars(void) * identity mapping if iommu_identity_mapping is set. */ if (!iommu_pass_through) { +#ifdef CONFIG_DMAR_BROKEN_GFX_WA + if (!iommu_identity_mapping) + iommu_identity_mapping = 2; +#endif if (iommu_identity_mapping) iommu_prepare_static_identity_mapping(); /* @@ -2293,8 +2371,6 @@ int __init init_dmars(void) } } - iommu_prepare_gfx_mapping(); - iommu_prepare_isa(); } @@ -2339,50 +2415,40 @@ error: return ret; } -static inline u64 aligned_size(u64 host_addr, size_t size) +/* Returns a number of VTD pages, but aligned to MM page size */ +static inline unsigned long aligned_nrpages(unsigned long host_addr, + size_t size) { - u64 addr; - addr = (host_addr & (~PAGE_MASK)) + size; - return PAGE_ALIGN(addr); -} - -struct iova * -iommu_alloc_iova(struct dmar_domain *domain, size_t size, u64 end) -{ - struct iova *piova; - - /* Make sure it's in range */ - end = min_t(u64, DOMAIN_MAX_ADDR(domain->gaw), end); - if (!size || (IOVA_START_ADDR + size > end)) - return NULL; - - piova = alloc_iova(&domain->iovad, - size >> PAGE_SHIFT, IOVA_PFN(end), 1); - return piova; + host_addr &= ~PAGE_MASK; + return PAGE_ALIGN(host_addr + size) >> VTD_PAGE_SHIFT; } -static struct iova * -__intel_alloc_iova(struct device *dev, struct dmar_domain *domain, - size_t size, u64 dma_mask) +/* This takes a number of _MM_ pages, not VTD pages */ +static struct iova *intel_alloc_iova(struct device *dev, + struct dmar_domain *domain, + unsigned long nrpages, uint64_t dma_mask) { struct pci_dev *pdev = to_pci_dev(dev); struct iova *iova = NULL; - if (dma_mask <= DMA_BIT_MASK(32) || dmar_forcedac) - iova = iommu_alloc_iova(domain, size, dma_mask); - else { + /* Restrict dma_mask to the width that the iommu can handle */ + dma_mask = min_t(uint64_t, DOMAIN_MAX_ADDR(domain->gaw), dma_mask); + + if (!dmar_forcedac && dma_mask > DMA_BIT_MASK(32)) { /* * First try to allocate an io virtual address in * DMA_BIT_MASK(32) and if that fails then try allocating * from higher range */ - iova = iommu_alloc_iova(domain, size, DMA_BIT_MASK(32)); - if (!iova) - iova = iommu_alloc_iova(domain, size, dma_mask); - } - - if (!iova) { - printk(KERN_ERR"Allocating iova for %s failed", pci_name(pdev)); + iova = alloc_iova(&domain->iovad, nrpages, + IOVA_PFN(DMA_BIT_MASK(32)), 1); + if (iova) + return iova; + } + iova = alloc_iova(&domain->iovad, nrpages, IOVA_PFN(dma_mask), 1); + if (unlikely(!iova)) { + printk(KERN_ERR "Allocating %ld-page iova for %s failed", + nrpages, pci_name(pdev)); return NULL; } @@ -2424,16 +2490,24 @@ static int iommu_dummy(struct pci_dev *pdev) } /* Check if the pdev needs to go through non-identity map and unmap process.*/ -static int iommu_no_mapping(struct pci_dev *pdev) +static int iommu_no_mapping(struct device *dev) { + struct pci_dev *pdev; int found; + if (unlikely(dev->bus != &pci_bus_type)) + return 1; + + pdev = to_pci_dev(dev); + if (iommu_dummy(pdev)) + return 1; + if (!iommu_identity_mapping) - return iommu_dummy(pdev); + return 0; found = identity_mapping(pdev); if (found) { - if (pdev->dma_mask > DMA_BIT_MASK(32)) + if (iommu_should_identity_map(pdev, 0)) return 1; else { /* @@ -2450,9 +2524,12 @@ static int iommu_no_mapping(struct pci_dev *pdev) * In case of a detached 64 bit DMA device from vm, the device * is put into si_domain for identity mapping. */ - if (pdev->dma_mask > DMA_BIT_MASK(32)) { + if (iommu_should_identity_map(pdev, 0)) { int ret; ret = domain_add_dev_info(si_domain, pdev); + if (ret) + return 0; + ret = domain_context_mapping(si_domain, pdev, CONTEXT_TT_MULTI_LEVEL); if (!ret) { printk(KERN_INFO "64bit %s uses identity mapping\n", pci_name(pdev)); @@ -2461,7 +2538,7 @@ static int iommu_no_mapping(struct pci_dev *pdev) } } - return iommu_dummy(pdev); + return 0; } static dma_addr_t __intel_map_single(struct device *hwdev, phys_addr_t paddr, @@ -2477,7 +2554,7 @@ static dma_addr_t __intel_map_single(struct device *hwdev, phys_addr_t paddr, BUG_ON(dir == DMA_NONE); - if (iommu_no_mapping(pdev)) + if (iommu_no_mapping(hwdev)) return paddr; domain = get_valid_domain_for_dev(pdev); @@ -2485,14 +2562,13 @@ static dma_addr_t __intel_map_single(struct device *hwdev, phys_addr_t paddr, return 0; iommu = domain_get_iommu(domain); - size = aligned_size((u64)paddr, size); + size = aligned_nrpages(paddr, size); - iova = __intel_alloc_iova(hwdev, domain, size, pdev->dma_mask); + iova = intel_alloc_iova(hwdev, domain, dma_to_mm_pfn(size), + pdev->dma_mask); if (!iova) goto error; - start_paddr = (phys_addr_t)iova->pfn_lo << PAGE_SHIFT; - /* * Check if DMAR supports zero-length reads on write only * mappings.. @@ -2508,20 +2584,20 @@ static dma_addr_t __intel_map_single(struct device *hwdev, phys_addr_t paddr, * might have two guest_addr mapping to the same host paddr, but this * is not a big problem */ - ret = domain_page_mapping(domain, start_paddr, - ((u64)paddr) & PHYSICAL_PAGE_MASK, - size, prot); + ret = domain_pfn_mapping(domain, mm_to_dma_pfn(iova->pfn_lo), + paddr >> VTD_PAGE_SHIFT, size, prot); if (ret) goto error; /* it's a non-present to present mapping. Only flush if caching mode */ if (cap_caching_mode(iommu->cap)) - iommu_flush_iotlb_psi(iommu, 0, start_paddr, - size >> VTD_PAGE_SHIFT); + iommu_flush_iotlb_psi(iommu, 0, mm_to_dma_pfn(iova->pfn_lo), size); else iommu_flush_write_buffer(iommu); - return start_paddr + ((u64)paddr & (~PAGE_MASK)); + start_paddr = (phys_addr_t)iova->pfn_lo << PAGE_SHIFT; + start_paddr += paddr & ~PAGE_MASK; + return start_paddr; error: if (iova) @@ -2614,11 +2690,11 @@ static void intel_unmap_page(struct device *dev, dma_addr_t dev_addr, { struct pci_dev *pdev = to_pci_dev(dev); struct dmar_domain *domain; - unsigned long start_addr; + unsigned long start_pfn, last_pfn; struct iova *iova; struct intel_iommu *iommu; - if (iommu_no_mapping(pdev)) + if (iommu_no_mapping(dev)) return; domain = find_domain(pdev); @@ -2627,22 +2703,25 @@ static void intel_unmap_page(struct device *dev, dma_addr_t dev_addr, iommu = domain_get_iommu(domain); iova = find_iova(&domain->iovad, IOVA_PFN(dev_addr)); - if (!iova) + if (WARN_ONCE(!iova, "Driver unmaps unmatched page at PFN %llx\n", + (unsigned long long)dev_addr)) return; - start_addr = iova->pfn_lo << PAGE_SHIFT; - size = aligned_size((u64)dev_addr, size); + start_pfn = mm_to_dma_pfn(iova->pfn_lo); + last_pfn = mm_to_dma_pfn(iova->pfn_hi + 1) - 1; - pr_debug("Device %s unmapping: %zx@%llx\n", - pci_name(pdev), size, (unsigned long long)start_addr); + pr_debug("Device %s unmapping: pfn %lx-%lx\n", + pci_name(pdev), start_pfn, last_pfn); /* clear the whole page */ - dma_pte_clear_range(domain, start_addr, start_addr + size); + dma_pte_clear_range(domain, start_pfn, last_pfn); + /* free page tables */ - dma_pte_free_pagetable(domain, start_addr, start_addr + size); + dma_pte_free_pagetable(domain, start_pfn, last_pfn); + if (intel_iommu_strict) { - iommu_flush_iotlb_psi(iommu, domain->id, start_addr, - size >> VTD_PAGE_SHIFT); + iommu_flush_iotlb_psi(iommu, domain->id, start_pfn, + last_pfn - start_pfn + 1); /* free iova */ __free_iova(&domain->iovad, iova); } else { @@ -2700,17 +2779,13 @@ static void intel_unmap_sg(struct device *hwdev, struct scatterlist *sglist, int nelems, enum dma_data_direction dir, struct dma_attrs *attrs) { - int i; struct pci_dev *pdev = to_pci_dev(hwdev); struct dmar_domain *domain; - unsigned long start_addr; + unsigned long start_pfn, last_pfn; struct iova *iova; - size_t size = 0; - phys_addr_t addr; - struct scatterlist *sg; struct intel_iommu *iommu; - if (iommu_no_mapping(pdev)) + if (iommu_no_mapping(hwdev)) return; domain = find_domain(pdev); @@ -2719,22 +2794,21 @@ static void intel_unmap_sg(struct device *hwdev, struct scatterlist *sglist, iommu = domain_get_iommu(domain); iova = find_iova(&domain->iovad, IOVA_PFN(sglist[0].dma_address)); - if (!iova) + if (WARN_ONCE(!iova, "Driver unmaps unmatched sglist at PFN %llx\n", + (unsigned long long)sglist[0].dma_address)) return; - for_each_sg(sglist, sg, nelems, i) { - addr = page_to_phys(sg_page(sg)) + sg->offset; - size += aligned_size((u64)addr, sg->length); - } - start_addr = iova->pfn_lo << PAGE_SHIFT; + start_pfn = mm_to_dma_pfn(iova->pfn_lo); + last_pfn = mm_to_dma_pfn(iova->pfn_hi + 1) - 1; /* clear the whole page */ - dma_pte_clear_range(domain, start_addr, start_addr + size); + dma_pte_clear_range(domain, start_pfn, last_pfn); + /* free page tables */ - dma_pte_free_pagetable(domain, start_addr, start_addr + size); + dma_pte_free_pagetable(domain, start_pfn, last_pfn); - iommu_flush_iotlb_psi(iommu, domain->id, start_addr, - size >> VTD_PAGE_SHIFT); + iommu_flush_iotlb_psi(iommu, domain->id, start_pfn, + (last_pfn - start_pfn + 1)); /* free iova */ __free_iova(&domain->iovad, iova); @@ -2757,21 +2831,20 @@ static int intel_nontranslate_map_sg(struct device *hddev, static int intel_map_sg(struct device *hwdev, struct scatterlist *sglist, int nelems, enum dma_data_direction dir, struct dma_attrs *attrs) { - phys_addr_t addr; int i; struct pci_dev *pdev = to_pci_dev(hwdev); struct dmar_domain *domain; size_t size = 0; int prot = 0; - size_t offset = 0; + size_t offset_pfn = 0; struct iova *iova = NULL; int ret; struct scatterlist *sg; - unsigned long start_addr; + unsigned long start_vpfn; struct intel_iommu *iommu; BUG_ON(dir == DMA_NONE); - if (iommu_no_mapping(pdev)) + if (iommu_no_mapping(hwdev)) return intel_nontranslate_map_sg(hwdev, sglist, nelems, dir); domain = get_valid_domain_for_dev(pdev); @@ -2780,12 +2853,11 @@ static int intel_map_sg(struct device *hwdev, struct scatterlist *sglist, int ne iommu = domain_get_iommu(domain); - for_each_sg(sglist, sg, nelems, i) { - addr = page_to_phys(sg_page(sg)) + sg->offset; - size += aligned_size((u64)addr, sg->length); - } + for_each_sg(sglist, sg, nelems, i) + size += aligned_nrpages(sg->offset, sg->length); - iova = __intel_alloc_iova(hwdev, domain, size, pdev->dma_mask); + iova = intel_alloc_iova(hwdev, domain, dma_to_mm_pfn(size), + pdev->dma_mask); if (!iova) { sglist->dma_length = 0; return 0; @@ -2801,35 +2873,24 @@ static int intel_map_sg(struct device *hwdev, struct scatterlist *sglist, int ne if (dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL) prot |= DMA_PTE_WRITE; - start_addr = iova->pfn_lo << PAGE_SHIFT; - offset = 0; - for_each_sg(sglist, sg, nelems, i) { - addr = page_to_phys(sg_page(sg)) + sg->offset; - size = aligned_size((u64)addr, sg->length); - ret = domain_page_mapping(domain, start_addr + offset, - ((u64)addr) & PHYSICAL_PAGE_MASK, - size, prot); - if (ret) { - /* clear the page */ - dma_pte_clear_range(domain, start_addr, - start_addr + offset); - /* free page tables */ - dma_pte_free_pagetable(domain, start_addr, - start_addr + offset); - /* free iova */ - __free_iova(&domain->iovad, iova); - return 0; - } - sg->dma_address = start_addr + offset + - ((u64)addr & (~PAGE_MASK)); - sg->dma_length = sg->length; - offset += size; + start_vpfn = mm_to_dma_pfn(iova->pfn_lo); + + ret = domain_sg_mapping(domain, start_vpfn, sglist, mm_to_dma_pfn(size), prot); + if (unlikely(ret)) { + /* clear the page */ + dma_pte_clear_range(domain, start_vpfn, + start_vpfn + size - 1); + /* free page tables */ + dma_pte_free_pagetable(domain, start_vpfn, + start_vpfn + size - 1); + /* free iova */ + __free_iova(&domain->iovad, iova); + return 0; } /* it's a non-present to present mapping. Only flush if caching mode */ if (cap_caching_mode(iommu->cap)) - iommu_flush_iotlb_psi(iommu, 0, start_addr, - offset >> VTD_PAGE_SHIFT); + iommu_flush_iotlb_psi(iommu, 0, start_vpfn, offset_pfn); else iommu_flush_write_buffer(iommu); @@ -3334,7 +3395,6 @@ static int md_domain_init(struct dmar_domain *domain, int guest_width) int adjust_width; init_iova_domain(&domain->iovad, DMA_32BIT_PFN); - spin_lock_init(&domain->mapping_lock); spin_lock_init(&domain->iommu_lock); domain_reserve_special_ranges(domain); @@ -3388,8 +3448,6 @@ static void iommu_free_vm_domain(struct dmar_domain *domain) static void vm_domain_exit(struct dmar_domain *domain) { - u64 end; - /* Domain 0 is reserved, so dont process it */ if (!domain) return; @@ -3397,14 +3455,12 @@ static void vm_domain_exit(struct dmar_domain *domain) vm_domain_remove_all_dev_info(domain); /* destroy iovas */ put_iova_domain(&domain->iovad); - end = DOMAIN_MAX_ADDR(domain->gaw); - end = end & (~VTD_PAGE_MASK); /* clear ptes */ - dma_pte_clear_range(domain, 0, end); + dma_pte_clear_range(domain, 0, DOMAIN_MAX_PFN(domain->gaw)); /* free page tables */ - dma_pte_free_pagetable(domain, 0, end); + dma_pte_free_pagetable(domain, 0, DOMAIN_MAX_PFN(domain->gaw)); iommu_free_vm_domain(domain); free_domain_mem(domain); @@ -3513,7 +3569,7 @@ static int intel_iommu_map_range(struct iommu_domain *domain, if ((iommu_prot & IOMMU_CACHE) && dmar_domain->iommu_snooping) prot |= DMA_PTE_SNP; - max_addr = (iova & VTD_PAGE_MASK) + VTD_PAGE_ALIGN(size); + max_addr = iova + size; if (dmar_domain->max_addr < max_addr) { int min_agaw; u64 end; @@ -3531,8 +3587,11 @@ static int intel_iommu_map_range(struct iommu_domain *domain, } dmar_domain->max_addr = max_addr; } - - ret = domain_page_mapping(dmar_domain, iova, hpa, size, prot); + /* Round up size to next multiple of PAGE_SIZE, if it and + the low bits of hpa would take us onto the next page */ + size = aligned_nrpages(hpa, size); + ret = domain_pfn_mapping(dmar_domain, iova >> VTD_PAGE_SHIFT, + hpa >> VTD_PAGE_SHIFT, size, prot); return ret; } @@ -3540,15 +3599,15 @@ static void intel_iommu_unmap_range(struct iommu_domain *domain, unsigned long iova, size_t size) { struct dmar_domain *dmar_domain = domain->priv; - dma_addr_t base; - /* The address might not be aligned */ - base = iova & VTD_PAGE_MASK; - size = VTD_PAGE_ALIGN(size); - dma_pte_clear_range(dmar_domain, base, base + size); + if (!size) + return; + + dma_pte_clear_range(dmar_domain, iova >> VTD_PAGE_SHIFT, + (iova + size - 1) >> VTD_PAGE_SHIFT); - if (dmar_domain->max_addr == base + size) - dmar_domain->max_addr = base; + if (dmar_domain->max_addr == iova + size) + dmar_domain->max_addr = iova; } static phys_addr_t intel_iommu_iova_to_phys(struct iommu_domain *domain, @@ -3558,7 +3617,7 @@ static phys_addr_t intel_iommu_iova_to_phys(struct iommu_domain *domain, struct dma_pte *pte; u64 phys = 0; - pte = addr_to_dma_pte(dmar_domain, iova); + pte = pfn_to_dma_pte(dmar_domain, iova >> VTD_PAGE_SHIFT); if (pte) phys = dma_pte_addr(pte); diff --git a/drivers/pci/iova.c b/drivers/pci/iova.c index 2287116e982..46dd440e231 100644 --- a/drivers/pci/iova.c +++ b/drivers/pci/iova.c @@ -1,9 +1,19 @@ /* - * Copyright (c) 2006, Intel Corporation. + * Copyright © 2006-2009, Intel Corporation. * - * This file is released under the GPLv2. + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., 59 Temple + * Place - Suite 330, Boston, MA 02111-1307 USA. * - * Copyright (C) 2006-2008 Intel Corporation * Author: Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com> */ @@ -123,7 +133,15 @@ move_left: /* Insert the new_iova into domain rbtree by holding writer lock */ /* Add new node and rebalance tree. */ { - struct rb_node **entry = &((prev)), *parent = NULL; + struct rb_node **entry, *parent = NULL; + + /* If we have 'prev', it's a valid place to start the + insertion. Otherwise, start from the root. */ + if (prev) + entry = &prev; + else + entry = &iovad->rbroot.rb_node; + /* Figure out where to put new node */ while (*entry) { struct iova *this = container_of(*entry, diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c index d9f06fbfa0b..d986afb7032 100644 --- a/drivers/pci/msi.c +++ b/drivers/pci/msi.c @@ -127,17 +127,23 @@ static inline __attribute_const__ u32 msi_enabled_mask(u16 control) * reliably as devices without an INTx disable bit will then generate a * level IRQ which will never be cleared. */ -static void msi_mask_irq(struct msi_desc *desc, u32 mask, u32 flag) +static u32 __msi_mask_irq(struct msi_desc *desc, u32 mask, u32 flag) { u32 mask_bits = desc->masked; if (!desc->msi_attrib.maskbit) - return; + return 0; mask_bits &= ~mask; mask_bits |= flag; pci_write_config_dword(desc->dev, desc->mask_pos, mask_bits); - desc->masked = mask_bits; + + return mask_bits; +} + +static void msi_mask_irq(struct msi_desc *desc, u32 mask, u32 flag) +{ + desc->masked = __msi_mask_irq(desc, mask, flag); } /* @@ -147,15 +153,21 @@ static void msi_mask_irq(struct msi_desc *desc, u32 mask, u32 flag) * file. This saves a few milliseconds when initialising devices with lots * of MSI-X interrupts. */ -static void msix_mask_irq(struct msi_desc *desc, u32 flag) +static u32 __msix_mask_irq(struct msi_desc *desc, u32 flag) { u32 mask_bits = desc->masked; unsigned offset = desc->msi_attrib.entry_nr * PCI_MSIX_ENTRY_SIZE + - PCI_MSIX_ENTRY_VECTOR_CTRL_OFFSET; + PCI_MSIX_ENTRY_VECTOR_CTRL; mask_bits &= ~1; mask_bits |= flag; writel(mask_bits, desc->mask_base + offset); - desc->masked = mask_bits; + + return mask_bits; +} + +static void msix_mask_irq(struct msi_desc *desc, u32 flag) +{ + desc->masked = __msix_mask_irq(desc, flag); } static void msi_set_mask_bit(unsigned irq, u32 flag) @@ -188,9 +200,9 @@ void read_msi_msg_desc(struct irq_desc *desc, struct msi_msg *msg) void __iomem *base = entry->mask_base + entry->msi_attrib.entry_nr * PCI_MSIX_ENTRY_SIZE; - msg->address_lo = readl(base + PCI_MSIX_ENTRY_LOWER_ADDR_OFFSET); - msg->address_hi = readl(base + PCI_MSIX_ENTRY_UPPER_ADDR_OFFSET); - msg->data = readl(base + PCI_MSIX_ENTRY_DATA_OFFSET); + msg->address_lo = readl(base + PCI_MSIX_ENTRY_LOWER_ADDR); + msg->address_hi = readl(base + PCI_MSIX_ENTRY_UPPER_ADDR); + msg->data = readl(base + PCI_MSIX_ENTRY_DATA); } else { struct pci_dev *dev = entry->dev; int pos = entry->msi_attrib.pos; @@ -225,11 +237,9 @@ void write_msi_msg_desc(struct irq_desc *desc, struct msi_msg *msg) base = entry->mask_base + entry->msi_attrib.entry_nr * PCI_MSIX_ENTRY_SIZE; - writel(msg->address_lo, - base + PCI_MSIX_ENTRY_LOWER_ADDR_OFFSET); - writel(msg->address_hi, - base + PCI_MSIX_ENTRY_UPPER_ADDR_OFFSET); - writel(msg->data, base + PCI_MSIX_ENTRY_DATA_OFFSET); + writel(msg->address_lo, base + PCI_MSIX_ENTRY_LOWER_ADDR); + writel(msg->address_hi, base + PCI_MSIX_ENTRY_UPPER_ADDR); + writel(msg->data, base + PCI_MSIX_ENTRY_DATA); } else { struct pci_dev *dev = entry->dev; int pos = entry->msi_attrib.pos; @@ -385,6 +395,7 @@ static int msi_capability_init(struct pci_dev *dev, int nvec) /* Configure MSI capability structure */ ret = arch_setup_msi_irqs(dev, nvec, PCI_CAP_ID_MSI); if (ret) { + msi_mask_irq(entry, mask, ~mask); msi_free_irqs(dev); return ret; } @@ -439,8 +450,14 @@ static int msix_capability_init(struct pci_dev *dev, for (i = 0; i < nvec; i++) { entry = alloc_msi_entry(dev); - if (!entry) - break; + if (!entry) { + if (!i) + iounmap(base); + else + msi_free_irqs(dev); + /* No enough memory. Don't try again */ + return -ENOMEM; + } j = entries[i].entry; entry->msi_attrib.is_msix = 1; @@ -487,7 +504,7 @@ static int msix_capability_init(struct pci_dev *dev, set_irq_msi(entry->irq, entry); j = entries[i].entry; entry->masked = readl(base + j * PCI_MSIX_ENTRY_SIZE + - PCI_MSIX_ENTRY_VECTOR_CTRL_OFFSET); + PCI_MSIX_ENTRY_VECTOR_CTRL); msix_mask_irq(entry, 1); i++; } @@ -611,9 +628,11 @@ void pci_msi_shutdown(struct pci_dev *dev) pci_intx_for_msi(dev, 1); dev->msi_enabled = 0; + /* Return the device with MSI unmasked as initial states */ pci_read_config_word(dev, pos + PCI_MSI_FLAGS, &ctrl); mask = msi_capable_mask(ctrl); - msi_mask_irq(desc, mask, ~mask); + /* Keep cached state to be restored */ + __msi_mask_irq(desc, mask, ~mask); /* Restore dev->irq to its default pin-assertion irq */ dev->irq = desc->msi_attrib.default_irq; @@ -653,7 +672,6 @@ static int msi_free_irqs(struct pci_dev* dev) list_for_each_entry_safe(entry, tmp, &dev->msi_list, list) { if (entry->msi_attrib.is_msix) { - msix_mask_irq(entry, 1); if (list_is_last(&entry->list, &dev->msi_list)) iounmap(entry->mask_base); } @@ -741,9 +759,17 @@ static void msix_free_all_irqs(struct pci_dev *dev) void pci_msix_shutdown(struct pci_dev* dev) { + struct msi_desc *entry; + if (!pci_msi_enable || !dev || !dev->msix_enabled) return; + /* Return the device with MSI-X masked as initial states */ + list_for_each_entry(entry, &dev->msi_list, list) { + /* Keep cached states to be restored */ + __msix_mask_irq(entry, 1); + } + msix_set_enable(dev, 0); pci_intx_for_msi(dev, 1); dev->msix_enabled = 0; diff --git a/drivers/pci/msi.h b/drivers/pci/msi.h index a0662842550..de27c1cb5a2 100644 --- a/drivers/pci/msi.h +++ b/drivers/pci/msi.h @@ -6,11 +6,11 @@ #ifndef MSI_H #define MSI_H -#define PCI_MSIX_ENTRY_SIZE 16 -#define PCI_MSIX_ENTRY_LOWER_ADDR_OFFSET 0 -#define PCI_MSIX_ENTRY_UPPER_ADDR_OFFSET 4 -#define PCI_MSIX_ENTRY_DATA_OFFSET 8 -#define PCI_MSIX_ENTRY_VECTOR_CTRL_OFFSET 12 +#define PCI_MSIX_ENTRY_SIZE 16 +#define PCI_MSIX_ENTRY_LOWER_ADDR 0 +#define PCI_MSIX_ENTRY_UPPER_ADDR 4 +#define PCI_MSIX_ENTRY_DATA 8 +#define PCI_MSIX_ENTRY_VECTOR_CTRL 12 #define msi_control_reg(base) (base + PCI_MSI_FLAGS) #define msi_lower_address_reg(base) (base + PCI_MSI_ADDRESS_LO) diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c index 6c93af5ced1..dbd0f947f49 100644 --- a/drivers/pci/pci.c +++ b/drivers/pci/pci.c @@ -1517,11 +1517,20 @@ void pci_enable_ari(struct pci_dev *dev) * * Perform INTx swizzling for a device behind one level of bridge. This is * required by section 9.1 of the PCI-to-PCI bridge specification for devices - * behind bridges on add-in cards. + * behind bridges on add-in cards. For devices with ARI enabled, the slot + * number is always 0 (see the Implementation Note in section 2.2.8.1 of + * the PCI Express Base Specification, Revision 2.1) */ u8 pci_swizzle_interrupt_pin(struct pci_dev *dev, u8 pin) { - return (((pin - 1) + PCI_SLOT(dev->devfn)) % 4) + 1; + int slot; + + if (pci_ari_enabled(dev->bus)) + slot = 0; + else + slot = PCI_SLOT(dev->devfn); + + return (((pin - 1) + slot) % 4) + 1; } int @@ -2171,7 +2180,7 @@ static int pci_parent_bus_reset(struct pci_dev *dev, int probe) u16 ctrl; struct pci_dev *pdev; - if (dev->subordinate) + if (pci_is_root_bus(dev->bus) || dev->subordinate || !dev->bus->self) return -ENOTTY; list_for_each_entry(pdev, &dev->bus->devices, bus_list) diff --git a/drivers/pci/pcie/aer/ecrc.c b/drivers/pci/pcie/aer/ecrc.c index ece97df4df6..a928d8ab6bd 100644 --- a/drivers/pci/pcie/aer/ecrc.c +++ b/drivers/pci/pcie/aer/ecrc.c @@ -106,7 +106,7 @@ void pcie_set_ecrc_checking(struct pci_dev *dev) disable_ecrc_checking(dev); break; case ECRC_POLICY_ON: - enable_ecrc_checking(dev);; + enable_ecrc_checking(dev); break; default: return; diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c index 56552d74abe..06b96562396 100644 --- a/drivers/pci/quirks.c +++ b/drivers/pci/quirks.c @@ -1058,6 +1058,11 @@ static void __devinit quirk_no_ata_d3(struct pci_dev *pdev) } DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_SERVERWORKS, PCI_ANY_ID, quirk_no_ata_d3); DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_ATI, PCI_ANY_ID, quirk_no_ata_d3); +/* ALi loses some register settings that we cannot then restore */ +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_AL, PCI_ANY_ID, quirk_no_ata_d3); +/* VIA comes back fine but we need to keep it alive or ACPI GTM failures + occur when mode detecting */ +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_VIA, PCI_ANY_ID, quirk_no_ata_d3); /* This was originally an Alpha specific thing, but it really fits here. * The i82375 PCI/EISA bridge appears as non-classified. Fix that. diff --git a/drivers/pci/slot.c b/drivers/pci/slot.c index eddb0748b0e..8c02b6c53bd 100644 --- a/drivers/pci/slot.c +++ b/drivers/pci/slot.c @@ -311,7 +311,7 @@ EXPORT_SYMBOL_GPL(pci_destroy_slot); #include <linux/pci_hotplug.h> /** * pci_hp_create_link - create symbolic link to the hotplug driver module. - * @slot: struct pci_slot + * @pci_slot: struct pci_slot * * Helper function for pci_hotplug_core.c to create symbolic link to * the hotplug driver module. @@ -334,7 +334,7 @@ EXPORT_SYMBOL_GPL(pci_hp_create_module_link); /** * pci_hp_remove_link - remove symbolic link to the hotplug driver module. - * @slot: struct pci_slot + * @pci_slot: struct pci_slot * * Helper function for pci_hotplug_core.c to remove symbolic link to * the hotplug driver module. |