struct dmar_domain {
int id; /* domain id */
+ int nid; /* node id */
unsigned long iommu_bmp; /* bitmap of iommus this domain uses*/
struct list_head devices; /* all devices' list */
int segment; /* PCI domain */
u8 bus; /* PCI bus number */
u8 devfn; /* PCI devfn number */
- struct pci_dev *dev; /* it's NULL for PCIE-to-PCI bridge */
+ struct pci_dev *dev; /* it's NULL for PCIe-to-PCI bridge */
struct intel_iommu *iommu; /* IOMMU used by this device */
struct dmar_domain *domain; /* pointer to domain */
};
static struct kmem_cache *iommu_devinfo_cache;
static struct kmem_cache *iommu_iova_cache;
-static inline void *iommu_kmem_cache_alloc(struct kmem_cache *cachep)
+static inline void *alloc_pgtable_page(int node)
{
- unsigned int flags;
- void *vaddr;
-
- /* trying to avoid low memory issues */
- flags = current->flags & PF_MEMALLOC;
- current->flags |= PF_MEMALLOC;
- vaddr = kmem_cache_alloc(cachep, GFP_ATOMIC);
- current->flags &= (~PF_MEMALLOC | flags);
- return vaddr;
-}
-
-
-static inline void *alloc_pgtable_page(void)
-{
- unsigned int flags;
- void *vaddr;
+ struct page *page;
+ void *vaddr = NULL;
- /* trying to avoid low memory issues */
- flags = current->flags & PF_MEMALLOC;
- current->flags |= PF_MEMALLOC;
- vaddr = (void *)get_zeroed_page(GFP_ATOMIC);
- current->flags &= (~PF_MEMALLOC | flags);
+ page = alloc_pages_node(node, GFP_ATOMIC | __GFP_ZERO, 0);
+ if (page)
+ vaddr = page_address(page);
return vaddr;
}
static inline void *alloc_domain_mem(void)
{
- return iommu_kmem_cache_alloc(iommu_domain_cache);
+ return kmem_cache_alloc(iommu_domain_cache, GFP_ATOMIC);
}
static void free_domain_mem(void *vaddr)
static inline void * alloc_devinfo_mem(void)
{
- return iommu_kmem_cache_alloc(iommu_devinfo_cache);
+ return kmem_cache_alloc(iommu_devinfo_cache, GFP_ATOMIC);
}
static inline void free_devinfo_mem(void *vaddr)
struct iova *alloc_iova_mem(void)
{
- return iommu_kmem_cache_alloc(iommu_iova_cache);
+ return kmem_cache_alloc(iommu_iova_cache, GFP_ATOMIC);
}
void free_iova_mem(struct iova *iova)
domain->iommu_coherency = 1;
- i = find_first_bit(&domain->iommu_bmp, g_num_of_iommus);
- for (; i < g_num_of_iommus; ) {
+ for_each_set_bit(i, &domain->iommu_bmp, g_num_of_iommus) {
if (!ecap_coherent(g_iommus[i]->ecap)) {
domain->iommu_coherency = 0;
break;
}
- i = find_next_bit(&domain->iommu_bmp, g_num_of_iommus, i+1);
}
}
domain->iommu_snooping = 1;
- i = find_first_bit(&domain->iommu_bmp, g_num_of_iommus);
- for (; i < g_num_of_iommus; ) {
+ for_each_set_bit(i, &domain->iommu_bmp, g_num_of_iommus) {
if (!ecap_sc_support(g_iommus[i]->ecap)) {
domain->iommu_snooping = 0;
break;
}
- i = find_next_bit(&domain->iommu_bmp, g_num_of_iommus, i+1);
}
}
root = &iommu->root_entry[bus];
context = get_context_addr_from_root(root);
if (!context) {
- context = (struct context_entry *)alloc_pgtable_page();
+ context = (struct context_entry *)
+ alloc_pgtable_page(iommu->node);
if (!context) {
spin_unlock_irqrestore(&iommu->lock, flags);
return NULL;
if (!dma_pte_present(pte)) {
uint64_t pteval;
- tmp_page = alloc_pgtable_page();
+ tmp_page = alloc_pgtable_page(domain->nid);
if (!tmp_page)
return NULL;
struct root_entry *root;
unsigned long flags;
- root = (struct root_entry *)alloc_pgtable_page();
+ root = (struct root_entry *)alloc_pgtable_page(iommu->node);
if (!root)
return -ENOMEM;
}
static void iommu_flush_iotlb_psi(struct intel_iommu *iommu, u16 did,
- unsigned long pfn, unsigned int pages)
+ unsigned long pfn, unsigned int pages, int map)
{
unsigned int mask = ilog2(__roundup_pow_of_two(pages));
uint64_t addr = (uint64_t)pfn << VTD_PAGE_SHIFT;
DMA_TLB_PSI_FLUSH);
/*
- * 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.
+ * In caching mode, changes of pages from non-present to present require
+ * flush. However, device IOTLB doesn't need to be flushed in this case.
*/
- if (!cap_caching_mode(iommu->cap) || did)
+ if (!cap_caching_mode(iommu->cap) || !map)
iommu_flush_dev_iotlb(iommu->domains[did], addr, mask);
}
unsigned long nlongs;
ndomains = cap_ndoms(iommu->cap);
- pr_debug("Number of Domains supportd <%ld>\n", ndomains);
+ pr_debug("IOMMU %d: Number of Domains supportd <%ld>\n", iommu->seq_id,
+ ndomains);
nlongs = BITS_TO_LONGS(ndomains);
spin_lock_init(&iommu->lock);
unsigned long flags;
if ((iommu->domains) && (iommu->domain_ids)) {
- i = find_first_bit(iommu->domain_ids, cap_ndoms(iommu->cap));
- for (; i < cap_ndoms(iommu->cap); ) {
+ for_each_set_bit(i, iommu->domain_ids, cap_ndoms(iommu->cap)) {
domain = iommu->domains[i];
clear_bit(i, iommu->domain_ids);
domain_exit(domain);
}
spin_unlock_irqrestore(&domain->iommu_lock, flags);
-
- i = find_next_bit(iommu->domain_ids,
- cap_ndoms(iommu->cap), i+1);
}
}
if (!domain)
return NULL;
+ domain->nid = -1;
memset(&domain->iommu_bmp, 0, sizeof(unsigned long));
domain->flags = 0;
spin_lock_irqsave(&iommu->lock, flags);
ndomains = cap_ndoms(iommu->cap);
- num = find_first_bit(iommu->domain_ids, ndomains);
- for (; num < ndomains; ) {
+ for_each_set_bit(num, iommu->domain_ids, ndomains) {
if (iommu->domains[num] == domain) {
found = 1;
break;
}
- num = find_next_bit(iommu->domain_ids,
- cap_ndoms(iommu->cap), num+1);
}
if (found) {
domain->iommu_snooping = 0;
domain->iommu_count = 1;
+ domain->nid = iommu->node;
/* always allocate the top pgd */
- domain->pgd = (struct dma_pte *)alloc_pgtable_page();
+ domain->pgd = (struct dma_pte *)alloc_pgtable_page(domain->nid);
if (!domain->pgd)
return -ENOMEM;
__iommu_flush_cache(iommu, domain->pgd, PAGE_SIZE);
/* find an available domain id for this device in iommu */
ndomains = cap_ndoms(iommu->cap);
- num = find_first_bit(iommu->domain_ids, ndomains);
- for (; num < ndomains; ) {
+ for_each_set_bit(num, iommu->domain_ids, ndomains) {
if (iommu->domains[num] == domain) {
id = num;
found = 1;
break;
}
- num = find_next_bit(iommu->domain_ids,
- cap_ndoms(iommu->cap), num+1);
}
if (found == 0) {
/* Skip top levels of page tables for
* iommu which has less agaw than default.
+ * Unnecessary for PT mode.
*/
- for (agaw = domain->agaw; agaw != iommu->agaw; agaw--) {
- pgd = phys_to_virt(dma_pte_addr(pgd));
- if (!dma_pte_present(pgd)) {
- spin_unlock_irqrestore(&iommu->lock, flags);
- return -ENOMEM;
+ if (translation != CONTEXT_TT_PASS_THROUGH) {
+ for (agaw = domain->agaw; agaw != iommu->agaw; agaw--) {
+ pgd = phys_to_virt(dma_pte_addr(pgd));
+ if (!dma_pte_present(pgd)) {
+ spin_unlock_irqrestore(&iommu->lock, flags);
+ return -ENOMEM;
+ }
}
}
}
(((u16)bus) << 8) | devfn,
DMA_CCMD_MASK_NOBIT,
DMA_CCMD_DEVICE_INVL);
- iommu->flush.flush_iotlb(iommu, 0, 0, 0, DMA_TLB_DSI_FLUSH);
+ iommu->flush.flush_iotlb(iommu, domain->id, 0, 0, DMA_TLB_DSI_FLUSH);
} else {
iommu_flush_write_buffer(iommu);
}
spin_lock_irqsave(&domain->iommu_lock, flags);
if (!test_and_set_bit(iommu->seq_id, &domain->iommu_bmp)) {
domain->iommu_count++;
+ if (domain->iommu_count == 1)
+ domain->nid = iommu->node;
domain_update_iommu_cap(domain);
}
spin_unlock_irqrestore(&domain->iommu_lock, flags);
return ret;
parent = parent->bus->self;
}
- if (pci_is_pcie(tmp)) /* this is a PCIE-to-PCI bridge */
+ if (pci_is_pcie(tmp)) /* this is a PCIe-to-PCI bridge */
return domain_context_mapping_one(domain,
pci_domain_nr(tmp->subordinate),
tmp->subordinate->number, 0,
"IOMMU: Setting identity map for device %s [0x%Lx - 0x%Lx]\n",
pci_name(pdev), start, end);
+ if (end < start) {
+ WARN(1, "Your BIOS is broken; RMRR ends before it starts!\n"
+ "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
+ dmi_get_system_info(DMI_BIOS_VENDOR),
+ dmi_get_system_info(DMI_BIOS_VERSION),
+ dmi_get_system_info(DMI_PRODUCT_VERSION));
+ ret = -EIO;
+ goto error;
+ }
+
if (end >> agaw_to_width(domain->agaw)) {
WARN(1, "Your BIOS is broken; RMRR exceeds permitted address width (%d bits)\n"
"BIOS vendor: %s; Ver: %s; Product Version: %s\n",
*/
iommu->flush.flush_context = __iommu_flush_context;
iommu->flush.flush_iotlb = __iommu_flush_iotlb;
- printk(KERN_INFO "IOMMU 0x%Lx: using Register based "
+ printk(KERN_INFO "IOMMU %d 0x%Lx: using Register based "
"invalidation\n",
+ iommu->seq_id,
(unsigned long long)drhd->reg_base_addr);
} else {
iommu->flush.flush_context = qi_flush_context;
iommu->flush.flush_iotlb = qi_flush_iotlb;
- printk(KERN_INFO "IOMMU 0x%Lx: using Queued "
+ printk(KERN_INFO "IOMMU %d 0x%Lx: using Queued "
"invalidation\n",
+ iommu->seq_id,
(unsigned long long)drhd->reg_base_addr);
}
}
/* 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, mm_to_dma_pfn(iova->pfn_lo), size);
+ iommu_flush_iotlb_psi(iommu, domain->id, mm_to_dma_pfn(iova->pfn_lo), size, 1);
else
iommu_flush_write_buffer(iommu);
if (!deferred_flush[i].next)
continue;
- iommu->flush.flush_iotlb(iommu, 0, 0, 0,
+ /* In caching mode, global flushes turn emulation expensive */
+ if (!cap_caching_mode(iommu->cap))
+ iommu->flush.flush_iotlb(iommu, 0, 0, 0,
DMA_TLB_GLOBAL_FLUSH);
for (j = 0; j < deferred_flush[i].next; j++) {
unsigned long mask;
struct iova *iova = deferred_flush[i].iova[j];
-
- mask = ilog2(mm_to_dma_pfn(iova->pfn_hi - iova->pfn_lo + 1));
- iommu_flush_dev_iotlb(deferred_flush[i].domain[j],
- (uint64_t)iova->pfn_lo << PAGE_SHIFT, mask);
+ struct dmar_domain *domain = deferred_flush[i].domain[j];
+
+ /* On real hardware multiple invalidations are expensive */
+ if (cap_caching_mode(iommu->cap))
+ iommu_flush_iotlb_psi(iommu, domain->id,
+ iova->pfn_lo, iova->pfn_hi - iova->pfn_lo + 1, 0);
+ else {
+ mask = ilog2(mm_to_dma_pfn(iova->pfn_hi - iova->pfn_lo + 1));
+ iommu_flush_dev_iotlb(deferred_flush[i].domain[j],
+ (uint64_t)iova->pfn_lo << PAGE_SHIFT, mask);
+ }
__free_iova(&deferred_flush[i].domain[j]->iovad, iova);
}
deferred_flush[i].next = 0;
if (intel_iommu_strict) {
iommu_flush_iotlb_psi(iommu, domain->id, start_pfn,
- last_pfn - start_pfn + 1);
+ last_pfn - start_pfn + 1, 0);
/* free iova */
__free_iova(&domain->iovad, iova);
} else {
size = PAGE_ALIGN(size);
order = get_order(size);
- flags &= ~(GFP_DMA | GFP_DMA32);
+
+ if (!iommu_no_mapping(hwdev))
+ flags &= ~(GFP_DMA | GFP_DMA32);
+ else if (hwdev->coherent_dma_mask < dma_get_required_mask(hwdev)) {
+ if (hwdev->coherent_dma_mask < DMA_BIT_MASK(32))
+ flags |= GFP_DMA;
+ else
+ flags |= GFP_DMA32;
+ }
vaddr = (void *)__get_free_pages(flags, order);
if (!vaddr)
if (intel_iommu_strict) {
iommu_flush_iotlb_psi(iommu, domain->id, start_pfn,
- last_pfn - start_pfn + 1);
+ last_pfn - start_pfn + 1, 0);
/* free iova */
__free_iova(&domain->iovad, iova);
} else {
struct dmar_domain *domain;
size_t size = 0;
int prot = 0;
- size_t offset_pfn = 0;
struct iova *iova = NULL;
int ret;
struct scatterlist *sg;
/* 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_vpfn, offset_pfn);
+ iommu_flush_iotlb_psi(iommu, domain->id, start_vpfn, size, 1);
else
iommu_flush_write_buffer(iommu);
}
#endif /* CONFIG_PM */
+/*
+ * Here we only respond to action of unbound device from driver.
+ *
+ * Added device is not attached to its DMAR domain here yet. That will happen
+ * when mapping the device to iova.
+ */
+static int device_notifier(struct notifier_block *nb,
+ unsigned long action, void *data)
+{
+ struct device *dev = data;
+ struct pci_dev *pdev = to_pci_dev(dev);
+ struct dmar_domain *domain;
+
+ if (iommu_no_mapping(dev))
+ return 0;
+
+ domain = find_domain(pdev);
+ if (!domain)
+ return 0;
+
+ if (action == BUS_NOTIFY_UNBOUND_DRIVER && !iommu_pass_through)
+ domain_remove_one_dev_info(domain, pdev);
+
+ return 0;
+}
+
+static struct notifier_block device_nb = {
+ .notifier_call = device_notifier,
+};
+
int __init intel_iommu_init(void)
{
int ret = 0;
* Check the need for DMA-remapping initialization now.
* Above initialization will also be used by Interrupt-remapping.
*/
- if (no_iommu || swiotlb || dmar_disabled)
+ if (no_iommu || dmar_disabled)
return -ENODEV;
iommu_init_mempool();
"PCI-DMA: Intel(R) Virtualization Technology for Directed I/O\n");
init_timer(&unmap_timer);
- force_iommu = 1;
+#ifdef CONFIG_SWIOTLB
+ swiotlb = 0;
+#endif
dma_ops = &intel_dma_ops;
init_iommu_sysfs();
register_iommu(&intel_iommu_ops);
+ bus_register_notifier(&pci_bus_type, &device_nb);
+
return 0;
}
parent->devfn);
parent = parent->bus->self;
}
- if (pci_is_pcie(tmp)) /* this is a PCIE-to-PCI bridge */
+ if (pci_is_pcie(tmp)) /* this is a PCIe-to-PCI bridge */
iommu_detach_dev(iommu,
tmp->subordinate->number, 0);
else /* this is a legacy PCI bridge */
/* domain id for virtual machine, it won't be set in context */
static unsigned long vm_domid;
-static int vm_domain_min_agaw(struct dmar_domain *domain)
-{
- int i;
- int min_agaw = domain->agaw;
-
- i = find_first_bit(&domain->iommu_bmp, g_num_of_iommus);
- for (; i < g_num_of_iommus; ) {
- if (min_agaw > g_iommus[i]->agaw)
- min_agaw = g_iommus[i]->agaw;
-
- i = find_next_bit(&domain->iommu_bmp, g_num_of_iommus, i+1);
- }
-
- return min_agaw;
-}
-
static struct dmar_domain *iommu_alloc_vm_domain(void)
{
struct dmar_domain *domain;
return NULL;
domain->id = vm_domid++;
+ domain->nid = -1;
memset(&domain->iommu_bmp, 0, sizeof(unsigned long));
domain->flags = DOMAIN_FLAG_VIRTUAL_MACHINE;
domain->iommu_coherency = 0;
domain->iommu_snooping = 0;
domain->max_addr = 0;
+ domain->nid = -1;
/* always allocate the top pgd */
- domain->pgd = (struct dma_pte *)alloc_pgtable_page();
+ domain->pgd = (struct dma_pte *)alloc_pgtable_page(domain->nid);
if (!domain->pgd)
return -ENOMEM;
domain_flush_cache(domain, domain->pgd, PAGE_SIZE);
iommu = drhd->iommu;
ndomains = cap_ndoms(iommu->cap);
- i = find_first_bit(iommu->domain_ids, ndomains);
- for (; i < ndomains; ) {
+ for_each_set_bit(i, iommu->domain_ids, ndomains) {
if (iommu->domains[i] == domain) {
spin_lock_irqsave(&iommu->lock, flags);
clear_bit(i, iommu->domain_ids);
spin_unlock_irqrestore(&iommu->lock, flags);
break;
}
- i = find_next_bit(iommu->domain_ids, ndomains, i+1);
}
}
}
struct pci_dev *pdev = to_pci_dev(dev);
struct intel_iommu *iommu;
int addr_width;
- u64 end;
/* normally pdev is not mapped */
if (unlikely(domain_context_mapped(pdev))) {
/* check if this iommu agaw is sufficient for max mapped address */
addr_width = agaw_to_width(iommu->agaw);
- end = DOMAIN_MAX_ADDR(addr_width);
- end = end & VTD_PAGE_MASK;
- if (end < dmar_domain->max_addr) {
- printk(KERN_ERR "%s: iommu agaw (%d) is not "
+ if (addr_width > cap_mgaw(iommu->cap))
+ addr_width = cap_mgaw(iommu->cap);
+
+ if (dmar_domain->max_addr > (1LL << addr_width)) {
+ printk(KERN_ERR "%s: iommu width (%d) is not "
"sufficient for the mapped address (%llx)\n",
- __func__, iommu->agaw, dmar_domain->max_addr);
+ __func__, addr_width, dmar_domain->max_addr);
return -EFAULT;
}
+ dmar_domain->gaw = addr_width;
+
+ /*
+ * Knock out extra levels of page tables if necessary
+ */
+ while (iommu->agaw < dmar_domain->agaw) {
+ struct dma_pte *pte;
+
+ pte = dmar_domain->pgd;
+ if (dma_pte_present(pte)) {
+ free_pgtable_page(dmar_domain->pgd);
+ dmar_domain->pgd = (struct dma_pte *)dma_pte_addr(pte);
+ }
+ dmar_domain->agaw--;
+ }
return domain_add_dev_info(dmar_domain, pdev, CONTEXT_TT_MULTI_LEVEL);
}
domain_remove_one_dev_info(dmar_domain, pdev);
}
-static int intel_iommu_map_range(struct iommu_domain *domain,
- unsigned long iova, phys_addr_t hpa,
- size_t size, int iommu_prot)
+static int intel_iommu_map(struct iommu_domain *domain,
+ unsigned long iova, phys_addr_t hpa,
+ int gfp_order, int iommu_prot)
{
struct dmar_domain *dmar_domain = domain->priv;
u64 max_addr;
- int addr_width;
int prot = 0;
+ size_t size;
int ret;
if (iommu_prot & IOMMU_READ)
if ((iommu_prot & IOMMU_CACHE) && dmar_domain->iommu_snooping)
prot |= DMA_PTE_SNP;
+ size = PAGE_SIZE << gfp_order;
max_addr = iova + size;
if (dmar_domain->max_addr < max_addr) {
- int min_agaw;
u64 end;
/* check if minimum agaw is sufficient for mapped address */
- min_agaw = vm_domain_min_agaw(dmar_domain);
- addr_width = agaw_to_width(min_agaw);
- end = DOMAIN_MAX_ADDR(addr_width);
- end = end & VTD_PAGE_MASK;
+ end = __DOMAIN_MAX_ADDR(dmar_domain->gaw) + 1;
if (end < max_addr) {
- printk(KERN_ERR "%s: iommu agaw (%d) is not "
+ printk(KERN_ERR "%s: iommu width (%d) is not "
"sufficient for the mapped address (%llx)\n",
- __func__, min_agaw, max_addr);
+ __func__, dmar_domain->gaw, max_addr);
return -EFAULT;
}
dmar_domain->max_addr = max_addr;
return ret;
}
-static void intel_iommu_unmap_range(struct iommu_domain *domain,
- unsigned long iova, size_t size)
+static int intel_iommu_unmap(struct iommu_domain *domain,
+ unsigned long iova, int gfp_order)
{
struct dmar_domain *dmar_domain = domain->priv;
-
- if (!size)
- return;
+ size_t size = PAGE_SIZE << gfp_order;
dma_pte_clear_range(dmar_domain, iova >> VTD_PAGE_SHIFT,
(iova + size - 1) >> VTD_PAGE_SHIFT);
if (dmar_domain->max_addr == iova + size)
dmar_domain->max_addr = iova;
+
+ return gfp_order;
}
static phys_addr_t intel_iommu_iova_to_phys(struct iommu_domain *domain,
.domain_destroy = intel_iommu_domain_destroy,
.attach_dev = intel_iommu_attach_device,
.detach_dev = intel_iommu_detach_device,
- .map = intel_iommu_map_range,
- .unmap = intel_iommu_unmap_range,
+ .map = intel_iommu_map,
+ .unmap = intel_iommu_unmap,
.iova_to_phys = intel_iommu_iova_to_phys,
.domain_has_cap = intel_iommu_domain_has_cap,
};