* Author: Ashok Raj <ashok.raj@intel.com>
* Author: Shaohua Li <shaohua.li@intel.com>
* Author: Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com>
+ * Author: Fenghua Yu <fenghua.yu@intel.com>
*/
#include <linux/init.h>
#include <linux/slab.h>
#include <linux/irq.h>
#include <linux/interrupt.h>
-#include <linux/sysdev.h>
#include <linux/spinlock.h>
#include <linux/pci.h>
#include <linux/dmar.h>
#include <linux/dma-mapping.h>
#include <linux/mempool.h>
#include <linux/timer.h>
-#include "iova.h"
-#include "intel-iommu.h"
-#include <asm/proto.h> /* force_iommu in this header in x86-64*/
+#include <linux/iova.h>
+#include <linux/iommu.h>
+#include <linux/intel-iommu.h>
#include <asm/cacheflush.h>
-#include <asm/gart.h>
+#include <asm/iommu.h>
#include "pci.h"
+#define ROOT_SIZE VTD_PAGE_SIZE
+#define CONTEXT_SIZE VTD_PAGE_SIZE
+
#define IS_GFX_DEVICE(pdev) ((pdev->class >> 16) == PCI_BASE_CLASS_DISPLAY)
#define IS_ISA_DEVICE(pdev) ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA)
#define DOMAIN_MAX_ADDR(gaw) ((((u64)1) << gaw) - 1)
+#define IOVA_PFN(addr) ((addr) >> PAGE_SHIFT)
+#define DMA_32BIT_PFN IOVA_PFN(DMA_32BIT_MASK)
+#define DMA_64BIT_PFN IOVA_PFN(DMA_64BIT_MASK)
+
+/* global iommu list, set NULL for ignored DMAR units */
+static struct intel_iommu **g_iommus;
+
+/*
+ * 0: Present
+ * 1-11: Reserved
+ * 12-63: Context Ptr (12 - (haw-1))
+ * 64-127: Reserved
+ */
+struct root_entry {
+ u64 val;
+ u64 rsvd1;
+};
+#define ROOT_ENTRY_NR (VTD_PAGE_SIZE/sizeof(struct root_entry))
+static inline bool root_present(struct root_entry *root)
+{
+ return (root->val & 1);
+}
+static inline void set_root_present(struct root_entry *root)
+{
+ root->val |= 1;
+}
+static inline void set_root_value(struct root_entry *root, unsigned long value)
+{
+ root->val |= value & VTD_PAGE_MASK;
+}
+
+static inline struct context_entry *
+get_context_addr_from_root(struct root_entry *root)
+{
+ return (struct context_entry *)
+ (root_present(root)?phys_to_virt(
+ root->val & VTD_PAGE_MASK) :
+ NULL);
+}
+
+/*
+ * low 64 bits:
+ * 0: present
+ * 1: fault processing disable
+ * 2-3: translation type
+ * 12-63: address space root
+ * high 64 bits:
+ * 0-2: address width
+ * 3-6: aval
+ * 8-23: domain id
+ */
+struct context_entry {
+ u64 lo;
+ u64 hi;
+};
+
+static inline bool context_present(struct context_entry *context)
+{
+ return (context->lo & 1);
+}
+static inline void context_set_present(struct context_entry *context)
+{
+ context->lo |= 1;
+}
+
+static inline void context_set_fault_enable(struct context_entry *context)
+{
+ context->lo &= (((u64)-1) << 2) | 1;
+}
+
+#define CONTEXT_TT_MULTI_LEVEL 0
+
+static inline void context_set_translation_type(struct context_entry *context,
+ unsigned long value)
+{
+ context->lo &= (((u64)-1) << 4) | 3;
+ context->lo |= (value & 3) << 2;
+}
+
+static inline void context_set_address_root(struct context_entry *context,
+ unsigned long value)
+{
+ context->lo |= value & VTD_PAGE_MASK;
+}
+
+static inline void context_set_address_width(struct context_entry *context,
+ unsigned long value)
+{
+ context->hi |= value & 7;
+}
+
+static inline void context_set_domain_id(struct context_entry *context,
+ unsigned long value)
+{
+ context->hi |= (value & ((1 << 16) - 1)) << 8;
+}
+
+static inline void context_clear_entry(struct context_entry *context)
+{
+ context->lo = 0;
+ context->hi = 0;
+}
+
+/*
+ * 0: readable
+ * 1: writable
+ * 2-6: reserved
+ * 7: super page
+ * 8-11: available
+ * 12-63: Host physcial address
+ */
+struct dma_pte {
+ u64 val;
+};
+
+static inline void dma_clear_pte(struct dma_pte *pte)
+{
+ pte->val = 0;
+}
+
+static inline void dma_set_pte_readable(struct dma_pte *pte)
+{
+ pte->val |= DMA_PTE_READ;
+}
+
+static inline void dma_set_pte_writable(struct dma_pte *pte)
+{
+ pte->val |= DMA_PTE_WRITE;
+}
+
+static inline void dma_set_pte_prot(struct dma_pte *pte, unsigned long prot)
+{
+ pte->val = (pte->val & ~3) | (prot & 3);
+}
+
+static inline u64 dma_pte_addr(struct dma_pte *pte)
+{
+ return (pte->val & VTD_PAGE_MASK);
+}
+
+static inline void dma_set_pte_addr(struct dma_pte *pte, u64 addr)
+{
+ pte->val |= (addr & VTD_PAGE_MASK);
+}
+
+static inline bool dma_pte_present(struct dma_pte *pte)
+{
+ return (pte->val & 3) != 0;
+}
+
+/* devices under the same p2p bridge are owned in one domain */
+#define DOMAIN_FLAG_P2P_MULTIPLE_DEVICES (1 << 0)
+
+/* domain represents a virtual machine, more than one devices
+ * across iommus may be owned in one domain, e.g. kvm guest.
+ */
+#define DOMAIN_FLAG_VIRTUAL_MACHINE (1 << 1)
+
+struct dmar_domain {
+ int id; /* domain id */
+ unsigned long iommu_bmp; /* bitmap of iommus this domain uses*/
+
+ struct list_head devices; /* all devices' list */
+ 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 */
+ int agaw;
+
+ int flags; /* flags to find out type of domain */
+
+ int iommu_coherency;/* indicate coherency of iommu access */
+ int iommu_count; /* reference count of iommu */
+ spinlock_t iommu_lock; /* protect iommu set in domain */
+ u64 max_addr; /* maximum mapped address */
+};
+
+/* PCI domain-device relationship */
+struct device_domain_info {
+ struct list_head link; /* link to domain siblings */
+ struct list_head global; /* link to global list */
+ u8 bus; /* PCI bus numer */
+ u8 devfn; /* PCI devfn number */
+ struct pci_dev *dev; /* it's NULL for PCIE-to-PCI bridge */
+ struct dmar_domain *domain; /* pointer to domain */
+};
static void flush_unmaps_timeout(unsigned long data);
static void domain_remove_dev_info(struct dmar_domain *domain);
-int dmar_disabled;
+#ifdef CONFIG_DMAR_DEFAULT_ON
+int dmar_disabled = 0;
+#else
+int dmar_disabled = 1;
+#endif /*CONFIG_DMAR_DEFAULT_ON*/
+
static int __initdata dmar_map_gfx = 1;
static int dmar_forcedac;
static int intel_iommu_strict;
static DEFINE_SPINLOCK(device_domain_lock);
static LIST_HEAD(device_domain_list);
+static struct iommu_ops intel_iommu_ops;
+
static int __init intel_iommu_setup(char *str)
{
if (!str)
return -EINVAL;
while (*str) {
- if (!strncmp(str, "off", 3)) {
+ if (!strncmp(str, "on", 2)) {
+ dmar_disabled = 0;
+ printk(KERN_INFO "Intel-IOMMU: enabled\n");
+ } else if (!strncmp(str, "off", 3)) {
dmar_disabled = 1;
- printk(KERN_INFO"Intel-IOMMU: disabled\n");
+ printk(KERN_INFO "Intel-IOMMU: disabled\n");
} else if (!strncmp(str, "igfx_off", 8)) {
dmar_map_gfx = 0;
printk(KERN_INFO
return iommu_kmem_cache_alloc(iommu_domain_cache);
}
-static inline void free_domain_mem(void *vaddr)
+static void free_domain_mem(void *vaddr)
{
kmem_cache_free(iommu_domain_cache, vaddr);
}
kmem_cache_free(iommu_iova_cache, iova);
}
+
+static inline int width_to_agaw(int width);
+
+/* calculate agaw for each iommu.
+ * "SAGAW" may be different across iommus, use a default agaw, and
+ * get a supported less agaw for iommus that don't support the default agaw.
+ */
+int iommu_calculate_agaw(struct intel_iommu *iommu)
+{
+ unsigned long sagaw;
+ int agaw = -1;
+
+ sagaw = cap_sagaw(iommu->cap);
+ for (agaw = width_to_agaw(DEFAULT_DOMAIN_ADDRESS_WIDTH);
+ agaw >= 0; agaw--) {
+ if (test_bit(agaw, &sagaw))
+ break;
+ }
+
+ return agaw;
+}
+
+/* in native case, each domain is related to only one iommu */
+static struct intel_iommu *domain_get_iommu(struct dmar_domain *domain)
+{
+ int iommu_id;
+
+ BUG_ON(domain->flags & DOMAIN_FLAG_VIRTUAL_MACHINE);
+
+ iommu_id = find_first_bit(&domain->iommu_bmp, g_num_of_iommus);
+ if (iommu_id < 0 || iommu_id >= g_num_of_iommus)
+ return NULL;
+
+ return g_iommus[iommu_id];
+}
+
+/* "Coherency" capability may be different across iommus */
+static void domain_update_iommu_coherency(struct dmar_domain *domain)
+{
+ int i;
+
+ domain->iommu_coherency = 1;
+
+ i = find_first_bit(&domain->iommu_bmp, g_num_of_iommus);
+ for (; i < 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);
+ }
+}
+
+static struct intel_iommu *device_to_iommu(u8 bus, u8 devfn)
+{
+ struct dmar_drhd_unit *drhd = NULL;
+ int i;
+
+ for_each_drhd_unit(drhd) {
+ if (drhd->ignored)
+ continue;
+
+ for (i = 0; i < drhd->devices_cnt; i++)
+ if (drhd->devices[i] &&
+ drhd->devices[i]->bus->number == bus &&
+ drhd->devices[i]->devfn == devfn)
+ return drhd->iommu;
+
+ if (drhd->include_all)
+ return drhd->iommu;
+ }
+
+ return NULL;
+}
+
+static void domain_flush_cache(struct dmar_domain *domain,
+ void *addr, int size)
+{
+ if (!domain->iommu_coherency)
+ clflush_cache_range(addr, size);
+}
+
/* Gets context entry for a given bus and devfn */
static struct context_entry * device_to_context_entry(struct intel_iommu *iommu,
u8 bus, u8 devfn)
spin_unlock_irqrestore(&iommu->lock, flags);
return NULL;
}
- __iommu_flush_cache(iommu, (void *)context, PAGE_SIZE_4K);
+ __iommu_flush_cache(iommu, (void *)context, CONTEXT_SIZE);
phy_addr = virt_to_phys((void *)context);
set_root_value(root, phy_addr);
set_root_present(root);
ret = 0;
goto out;
}
- ret = context_present(context[devfn]);
+ ret = context_present(&context[devfn]);
out:
spin_unlock_irqrestore(&iommu->lock, flags);
return ret;
root = &iommu->root_entry[bus];
context = get_context_addr_from_root(root);
if (context) {
- context_clear_entry(context[devfn]);
+ context_clear_entry(&context[devfn]);
__iommu_flush_cache(iommu, &context[devfn], \
sizeof(*context));
}
if (level == 1)
break;
- if (!dma_pte_present(*pte)) {
+ if (!dma_pte_present(pte)) {
tmp_page = alloc_pgtable_page();
if (!tmp_page) {
flags);
return NULL;
}
- __iommu_flush_cache(domain->iommu, tmp_page,
- PAGE_SIZE_4K);
- dma_set_pte_addr(*pte, virt_to_phys(tmp_page));
+ 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);
- __iommu_flush_cache(domain->iommu, pte, sizeof(*pte));
+ 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));
+ parent = phys_to_virt(dma_pte_addr(pte));
level--;
}
if (level == total)
return pte;
- if (!dma_pte_present(*pte))
+ if (!dma_pte_present(pte))
break;
- parent = phys_to_virt(dma_pte_addr(*pte));
+ parent = phys_to_virt(dma_pte_addr(pte));
total--;
}
return NULL;
pte = dma_addr_level_pte(domain, addr, 1);
if (pte) {
- dma_clear_pte(*pte);
- __iommu_flush_cache(domain->iommu, pte, sizeof(*pte));
+ dma_clear_pte(pte);
+ domain_flush_cache(domain, pte, sizeof(*pte));
}
}
start &= (((u64)1) << addr_width) - 1;
end &= (((u64)1) << addr_width) - 1;
/* in case it's partial page */
- start = PAGE_ALIGN_4K(start);
- end &= PAGE_MASK_4K;
+ start = PAGE_ALIGN(start);
+ end &= PAGE_MASK;
/* we don't need lock here, nobody else touches the iova range */
while (start < end) {
dma_pte_clear_one(domain, start);
- start += PAGE_SIZE_4K;
+ start += VTD_PAGE_SIZE;
}
}
pte = dma_addr_level_pte(domain, tmp, level);
if (pte) {
free_pgtable_page(
- phys_to_virt(dma_pte_addr(*pte)));
- dma_clear_pte(*pte);
- __iommu_flush_cache(domain->iommu,
- pte, sizeof(*pte));
+ phys_to_virt(dma_pte_addr(pte)));
+ dma_clear_pte(pte);
+ domain_flush_cache(domain, pte, sizeof(*pte));
}
tmp += level_size(level);
}
if (!root)
return -ENOMEM;
- __iommu_flush_cache(iommu, root, PAGE_SIZE_4K);
+ __iommu_flush_cache(iommu, root, ROOT_SIZE);
spin_lock_irqsave(&iommu->lock, flags);
iommu->root_entry = root;
spin_unlock_irqrestore(&iommu->register_lock, flag);
- /* flush context entry will implictly flush write buffer */
+ /* flush context entry will implicitly flush write buffer */
return 0;
}
-static int inline iommu_flush_context_global(struct intel_iommu *iommu,
- int non_present_entry_flush)
-{
- return __iommu_flush_context(iommu, 0, 0, 0, DMA_CCMD_GLOBAL_INVL,
- non_present_entry_flush);
-}
-
-static int inline iommu_flush_context_domain(struct intel_iommu *iommu, u16 did,
- int non_present_entry_flush)
-{
- return __iommu_flush_context(iommu, did, 0, 0, DMA_CCMD_DOMAIN_INVL,
- non_present_entry_flush);
-}
-
-static int inline iommu_flush_context_device(struct intel_iommu *iommu,
- u16 did, u16 source_id, u8 function_mask, int non_present_entry_flush)
-{
- return __iommu_flush_context(iommu, did, source_id, function_mask,
- DMA_CCMD_DEVICE_INVL, non_present_entry_flush);
-}
-
/* return value determine if we need a write buffer flush */
static int __iommu_flush_iotlb(struct intel_iommu *iommu, u16 did,
u64 addr, unsigned int size_order, u64 type,
printk(KERN_ERR"IOMMU: flush IOTLB failed\n");
if (DMA_TLB_IAIG(val) != DMA_TLB_IIRG(type))
pr_debug("IOMMU: tlb flush request %Lx, actual %Lx\n",
- DMA_TLB_IIRG(type), DMA_TLB_IAIG(val));
- /* flush context entry will implictly flush write buffer */
+ (unsigned long long)DMA_TLB_IIRG(type),
+ (unsigned long long)DMA_TLB_IAIG(val));
+ /* flush iotlb entry will implicitly flush write buffer */
return 0;
}
-static int inline iommu_flush_iotlb_global(struct intel_iommu *iommu,
- int non_present_entry_flush)
-{
- return __iommu_flush_iotlb(iommu, 0, 0, 0, DMA_TLB_GLOBAL_FLUSH,
- non_present_entry_flush);
-}
-
-static int inline iommu_flush_iotlb_dsi(struct intel_iommu *iommu, u16 did,
- int non_present_entry_flush)
-{
- return __iommu_flush_iotlb(iommu, did, 0, 0, DMA_TLB_DSI_FLUSH,
- non_present_entry_flush);
-}
-
static int iommu_flush_iotlb_psi(struct intel_iommu *iommu, u16 did,
u64 addr, unsigned int pages, int non_present_entry_flush)
{
unsigned int mask;
- BUG_ON(addr & (~PAGE_MASK_4K));
+ BUG_ON(addr & (~VTD_PAGE_MASK));
BUG_ON(pages == 0);
/* Fallback to domain selective flush if no PSI support */
if (!cap_pgsel_inv(iommu->cap))
- return iommu_flush_iotlb_dsi(iommu, did,
- non_present_entry_flush);
+ return iommu->flush.flush_iotlb(iommu, did, 0, 0,
+ DMA_TLB_DSI_FLUSH,
+ non_present_entry_flush);
/*
* PSI requires page size to be 2 ^ x, and the base address is naturally
mask = ilog2(__roundup_pow_of_two(pages));
/* Fallback to domain selective flush if size is too big */
if (mask > cap_max_amask_val(iommu->cap))
- return iommu_flush_iotlb_dsi(iommu, did,
- non_present_entry_flush);
+ return iommu->flush.flush_iotlb(iommu, did, 0, 0,
+ DMA_TLB_DSI_FLUSH, non_present_entry_flush);
- return __iommu_flush_iotlb(iommu, did, addr, mask,
- DMA_TLB_PSI_FLUSH, non_present_entry_flush);
+ return iommu->flush.flush_iotlb(iommu, did, addr, mask,
+ DMA_TLB_PSI_FLUSH,
+ non_present_entry_flush);
}
static void iommu_disable_protect_mem_regions(struct intel_iommu *iommu)
}
static int iommu_page_fault_do_one(struct intel_iommu *iommu, int type,
- u8 fault_reason, u16 source_id, u64 addr)
+ u8 fault_reason, u16 source_id, unsigned long long addr)
{
const char *reason;
static void domain_exit(struct dmar_domain *domain);
+static void vm_domain_exit(struct dmar_domain *domain);
void free_dmar_iommu(struct intel_iommu *iommu)
{
struct dmar_domain *domain;
int i;
+ unsigned long flags;
i = find_first_bit(iommu->domain_ids, cap_ndoms(iommu->cap));
for (; i < cap_ndoms(iommu->cap); ) {
domain = iommu->domains[i];
clear_bit(i, iommu->domain_ids);
- domain_exit(domain);
+
+ spin_lock_irqsave(&domain->iommu_lock, flags);
+ if (--domain->iommu_count == 0) {
+ if (domain->flags & DOMAIN_FLAG_VIRTUAL_MACHINE)
+ vm_domain_exit(domain);
+ else
+ domain_exit(domain);
+ }
+ spin_unlock_irqrestore(&domain->iommu_lock, flags);
+
i = find_next_bit(iommu->domain_ids,
cap_ndoms(iommu->cap), i+1);
}
kfree(iommu->domains);
kfree(iommu->domain_ids);
+ g_iommus[iommu->seq_id] = NULL;
+
+ /* if all iommus are freed, free g_iommus */
+ for (i = 0; i < g_num_of_iommus; i++) {
+ if (g_iommus[i])
+ break;
+ }
+
+ if (i == g_num_of_iommus)
+ kfree(g_iommus);
+
/* free context mapping */
free_context_table(iommu);
}
set_bit(num, iommu->domain_ids);
domain->id = num;
- domain->iommu = iommu;
+ memset(&domain->iommu_bmp, 0, sizeof(unsigned long));
+ set_bit(iommu->seq_id, &domain->iommu_bmp);
+ domain->flags = 0;
iommu->domains[num] = domain;
spin_unlock_irqrestore(&iommu->lock, flags);
static void iommu_free_domain(struct dmar_domain *domain)
{
unsigned long flags;
+ struct intel_iommu *iommu;
+
+ iommu = domain_get_iommu(domain);
- spin_lock_irqsave(&domain->iommu->lock, flags);
- clear_bit(domain->id, domain->iommu->domain_ids);
- spin_unlock_irqrestore(&domain->iommu->lock, flags);
+ spin_lock_irqsave(&iommu->lock, flags);
+ clear_bit(domain->id, iommu->domain_ids);
+ spin_unlock_irqrestore(&iommu->lock, flags);
}
static struct iova_domain reserved_iova_list;
if (!r->flags || !(r->flags & IORESOURCE_MEM))
continue;
addr = r->start;
- addr &= PAGE_MASK_4K;
+ addr &= PAGE_MASK;
size = r->end - addr;
- size = PAGE_ALIGN_4K(size);
+ size = PAGE_ALIGN(size);
iova = reserve_iova(&reserved_iova_list, IOVA_PFN(addr),
IOVA_PFN(size + addr) - 1);
if (!iova)
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);
/* calculate AGAW */
- iommu = domain->iommu;
+ iommu = domain_get_iommu(domain);
if (guest_width > cap_mgaw(iommu->cap))
guest_width = cap_mgaw(iommu->cap);
domain->gaw = guest_width;
domain->agaw = agaw;
INIT_LIST_HEAD(&domain->devices);
+ if (ecap_coherent(iommu->ecap))
+ domain->iommu_coherency = 1;
+ else
+ domain->iommu_coherency = 0;
+
+ domain->iommu_count = 1;
+
/* always allocate the top pgd */
domain->pgd = (struct dma_pte *)alloc_pgtable_page();
if (!domain->pgd)
return -ENOMEM;
- __iommu_flush_cache(iommu, domain->pgd, PAGE_SIZE_4K);
+ __iommu_flush_cache(iommu, domain->pgd, PAGE_SIZE);
return 0;
}
/* destroy iovas */
put_iova_domain(&domain->iovad);
end = DOMAIN_MAX_ADDR(domain->gaw);
- end = end & (~PAGE_MASK_4K);
+ end = end & (~PAGE_MASK);
/* clear ptes */
dma_pte_clear_range(domain, 0, end);
u8 bus, u8 devfn)
{
struct context_entry *context;
- struct intel_iommu *iommu = domain->iommu;
unsigned long flags;
+ struct intel_iommu *iommu;
+ struct dma_pte *pgd;
+ unsigned long num;
+ unsigned long ndomains;
+ int id;
+ int agaw;
pr_debug("Set context mapping for %02x:%02x.%d\n",
bus, PCI_SLOT(devfn), PCI_FUNC(devfn));
BUG_ON(!domain->pgd);
+
+ iommu = device_to_iommu(bus, devfn);
+ if (!iommu)
+ return -ENODEV;
+
context = device_to_context_entry(iommu, bus, devfn);
if (!context)
return -ENOMEM;
spin_lock_irqsave(&iommu->lock, flags);
- if (context_present(*context)) {
+ if (context_present(context)) {
spin_unlock_irqrestore(&iommu->lock, flags);
return 0;
}
- context_set_domain_id(*context, domain->id);
- context_set_address_width(*context, domain->agaw);
- context_set_address_root(*context, virt_to_phys(domain->pgd));
- context_set_translation_type(*context, CONTEXT_TT_MULTI_LEVEL);
- context_set_fault_enable(*context);
- context_set_present(*context);
- __iommu_flush_cache(iommu, context, sizeof(*context));
+ id = domain->id;
+ pgd = domain->pgd;
+
+ if (domain->flags & DOMAIN_FLAG_VIRTUAL_MACHINE) {
+ int found = 0;
+
+ /* 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; ) {
+ 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) {
+ num = find_first_zero_bit(iommu->domain_ids, ndomains);
+ if (num >= ndomains) {
+ spin_unlock_irqrestore(&iommu->lock, flags);
+ printk(KERN_ERR "IOMMU: no free domain ids\n");
+ return -EFAULT;
+ }
+
+ set_bit(num, iommu->domain_ids);
+ iommu->domains[num] = domain;
+ id = num;
+ }
+
+ /* Skip top levels of page tables for
+ * iommu which has less agaw than default.
+ */
+ 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;
+ }
+ }
+ }
+
+ context_set_domain_id(context, id);
+ context_set_address_width(context, iommu->agaw);
+ context_set_address_root(context, virt_to_phys(pgd));
+ context_set_translation_type(context, CONTEXT_TT_MULTI_LEVEL);
+ context_set_fault_enable(context);
+ context_set_present(context);
+ domain_flush_cache(domain, context, sizeof(*context));
/* it's a non-present to present mapping */
- if (iommu_flush_context_device(iommu, domain->id,
- (((u16)bus) << 8) | devfn, DMA_CCMD_MASK_NOBIT, 1))
+ if (iommu->flush.flush_context(iommu, domain->id,
+ (((u16)bus) << 8) | devfn, DMA_CCMD_MASK_NOBIT,
+ DMA_CCMD_DEVICE_INVL, 1))
iommu_flush_write_buffer(iommu);
else
- iommu_flush_iotlb_dsi(iommu, 0, 0);
+ iommu->flush.flush_iotlb(iommu, 0, 0, 0, DMA_TLB_DSI_FLUSH, 0);
+
spin_unlock_irqrestore(&iommu->lock, flags);
+
+ spin_lock_irqsave(&domain->iommu_lock, flags);
+ if (!test_and_set_bit(iommu->seq_id, &domain->iommu_bmp)) {
+ domain->iommu_count++;
+ domain_update_iommu_coherency(domain);
+ }
+ spin_unlock_irqrestore(&domain->iommu_lock, flags);
return 0;
}
tmp->bus->number, tmp->devfn);
}
-static int domain_context_mapped(struct dmar_domain *domain,
- struct pci_dev *pdev)
+static int domain_context_mapped(struct pci_dev *pdev)
{
int ret;
struct pci_dev *tmp, *parent;
+ struct intel_iommu *iommu;
- ret = device_context_mapped(domain->iommu,
+ iommu = device_to_iommu(pdev->bus->number, pdev->devfn);
+ if (!iommu)
+ return -ENODEV;
+
+ ret = device_context_mapped(iommu,
pdev->bus->number, pdev->devfn);
if (!ret)
return ret;
/* Secondary interface's bus number and devfn 0 */
parent = pdev->bus->self;
while (parent != tmp) {
- ret = device_context_mapped(domain->iommu, parent->bus->number,
+ ret = device_context_mapped(iommu, parent->bus->number,
parent->devfn);
if (!ret)
return ret;
parent = parent->bus->self;
}
if (tmp->is_pcie)
- return device_context_mapped(domain->iommu,
+ return device_context_mapped(iommu,
tmp->subordinate->number, 0);
else
- return device_context_mapped(domain->iommu,
+ return device_context_mapped(iommu,
tmp->bus->number, tmp->devfn);
}
u64 start_pfn, end_pfn;
struct dma_pte *pte;
int index;
+ int addr_width = agaw_to_width(domain->agaw);
+
+ hpa &= (((u64)1) << addr_width) - 1;
if ((prot & (DMA_PTE_READ|DMA_PTE_WRITE)) == 0)
return -EINVAL;
- iova &= PAGE_MASK_4K;
- start_pfn = ((u64)hpa) >> PAGE_SHIFT_4K;
- end_pfn = (PAGE_ALIGN_4K(((u64)hpa) + size)) >> PAGE_SHIFT_4K;
+ 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 + PAGE_SIZE_4K * index);
+ pte = addr_to_dma_pte(domain, iova + VTD_PAGE_SIZE * index);
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 << PAGE_SHIFT_4K);
- dma_set_pte_prot(*pte, prot);
- __iommu_flush_cache(domain->iommu, pte, sizeof(*pte));
+ BUG_ON(dma_pte_addr(pte));
+ dma_set_pte_addr(pte, start_pfn << VTD_PAGE_SHIFT);
+ dma_set_pte_prot(pte, prot);
+ domain_flush_cache(domain, pte, sizeof(*pte));
start_pfn++;
index++;
}
return 0;
}
-static void detach_domain_for_dev(struct dmar_domain *domain, u8 bus, u8 devfn)
+static void iommu_detach_dev(struct intel_iommu *iommu, u8 bus, u8 devfn)
{
- clear_context_table(domain->iommu, bus, devfn);
- iommu_flush_context_global(domain->iommu, 0);
- iommu_flush_iotlb_global(domain->iommu, 0);
+ if (!iommu)
+ return;
+
+ clear_context_table(iommu, bus, devfn);
+ iommu->flush.flush_context(iommu, 0, 0, 0,
+ DMA_CCMD_GLOBAL_INVL, 0);
+ iommu->flush.flush_iotlb(iommu, 0, 0, 0,
+ DMA_TLB_GLOBAL_FLUSH, 0);
}
static void domain_remove_dev_info(struct dmar_domain *domain)
{
struct device_domain_info *info;
unsigned long flags;
+ struct intel_iommu *iommu;
spin_lock_irqsave(&device_domain_lock, flags);
while (!list_empty(&domain->devices)) {
info->dev->dev.archdata.iommu = NULL;
spin_unlock_irqrestore(&device_domain_lock, flags);
- detach_domain_for_dev(info->domain, info->bus, info->devfn);
+ iommu = device_to_iommu(info->bus, info->devfn);
+ iommu_detach_dev(iommu, info->bus, info->devfn);
free_devinfo_mem(info);
spin_lock_irqsave(&device_domain_lock, flags);
* find_domain
* Note: we use struct pci_dev->dev.archdata.iommu stores the info
*/
-struct dmar_domain *
+static struct dmar_domain *
find_domain(struct pci_dev *pdev)
{
struct device_domain_info *info;
info->dev = NULL;
info->domain = domain;
/* This domain is shared by devices under p2p bridge */
- domain->flags |= DOMAIN_FLAG_MULTIPLE_DEVICES;
+ domain->flags |= DOMAIN_FLAG_P2P_MULTIPLE_DEVICES;
/* pcie-to-pci bridge already has a domain, uses it */
found = NULL;
return find_domain(pdev);
}
-static int iommu_prepare_identity_map(struct pci_dev *pdev, u64 start, u64 end)
+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;
- u64 base;
+ unsigned long long base;
int ret;
printk(KERN_INFO
return -ENOMEM;
/* The address might not be aligned */
- base = start & PAGE_MASK_4K;
+ base = start & PAGE_MASK;
size = end - base;
- size = PAGE_ALIGN_4K(size);
+ 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");
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
}
#endif /* !CONFIG_DMAR_FLPY_WA */
-int __init init_dmars(void)
+static int __init init_dmars(void)
{
struct dmar_drhd_unit *drhd;
struct dmar_rmrr_unit *rmrr;
*/
}
+ g_iommus = kcalloc(g_num_of_iommus, sizeof(struct intel_iommu *),
+ GFP_KERNEL);
+ if (!g_iommus) {
+ printk(KERN_ERR "Allocating global iommu array failed\n");
+ ret = -ENOMEM;
+ goto error;
+ }
+
deferred_flush = kzalloc(g_num_of_iommus *
sizeof(struct deferred_flush_tables), GFP_KERNEL);
if (!deferred_flush) {
+ kfree(g_iommus);
ret = -ENOMEM;
goto error;
}
continue;
iommu = drhd->iommu;
+ g_iommus[iommu->seq_id] = iommu;
ret = iommu_init_domains(iommu);
if (ret)
}
}
+ for_each_drhd_unit(drhd) {
+ if (drhd->ignored)
+ continue;
+
+ iommu = drhd->iommu;
+ if (dmar_enable_qi(iommu)) {
+ /*
+ * Queued Invalidate not enabled, use Register Based
+ * Invalidate
+ */
+ iommu->flush.flush_context = __iommu_flush_context;
+ iommu->flush.flush_iotlb = __iommu_flush_iotlb;
+ printk(KERN_INFO "IOMMU 0x%Lx: using Register based "
+ "invalidation\n",
+ (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 "
+ "invalidation\n",
+ (unsigned long long)drhd->reg_base_addr);
+ }
+ }
+
/*
* For each rmrr
* for each dev attached to rmrr
iommu_set_root_entry(iommu);
- iommu_flush_context_global(iommu, 0);
- iommu_flush_iotlb_global(iommu, 0);
-
+ iommu->flush.flush_context(iommu, 0, 0, 0, DMA_CCMD_GLOBAL_INVL,
+ 0);
+ iommu->flush.flush_iotlb(iommu, 0, 0, 0, DMA_TLB_GLOBAL_FLUSH,
+ 0);
iommu_disable_protect_mem_regions(iommu);
ret = iommu_enable_translation(iommu);
iommu = drhd->iommu;
free_iommu(iommu);
}
+ kfree(g_iommus);
return ret;
}
static inline u64 aligned_size(u64 host_addr, size_t size)
{
u64 addr;
- addr = (host_addr & (~PAGE_MASK_4K)) + size;
- return PAGE_ALIGN_4K(addr);
+ addr = (host_addr & (~PAGE_MASK)) + size;
+ return PAGE_ALIGN(addr);
}
struct iova *
return NULL;
piova = alloc_iova(&domain->iovad,
- size >> PAGE_SHIFT_4K, IOVA_PFN(end), 1);
+ size >> PAGE_SHIFT, IOVA_PFN(end), 1);
return piova;
}
static struct iova *
__intel_alloc_iova(struct device *dev, struct dmar_domain *domain,
- size_t size)
+ size_t size, u64 dma_mask)
{
struct pci_dev *pdev = to_pci_dev(dev);
struct iova *iova = NULL;
- if ((pdev->dma_mask <= DMA_32BIT_MASK) || (dmar_forcedac)) {
- iova = iommu_alloc_iova(domain, size, pdev->dma_mask);
- } else {
+ if (dma_mask <= DMA_32BIT_MASK || dmar_forcedac)
+ iova = iommu_alloc_iova(domain, size, dma_mask);
+ else {
/*
* First try to allocate an io virtual address in
* DMA_32BIT_MASK and if that fails then try allocating
*/
iova = iommu_alloc_iova(domain, size, DMA_32BIT_MASK);
if (!iova)
- iova = iommu_alloc_iova(domain, size, pdev->dma_mask);
+ iova = iommu_alloc_iova(domain, size, dma_mask);
}
if (!iova) {
}
/* make sure context mapping is ok */
- if (unlikely(!domain_context_mapped(domain, pdev))) {
+ if (unlikely(!domain_context_mapped(pdev))) {
ret = domain_context_mapping(domain, pdev);
if (ret) {
printk(KERN_ERR
return domain;
}
-static dma_addr_t
-intel_map_single(struct device *hwdev, phys_addr_t paddr, size_t size, int dir)
+static dma_addr_t __intel_map_single(struct device *hwdev, phys_addr_t paddr,
+ size_t size, int dir, u64 dma_mask)
{
struct pci_dev *pdev = to_pci_dev(hwdev);
struct dmar_domain *domain;
- unsigned long start_paddr;
+ phys_addr_t start_paddr;
struct iova *iova;
int prot = 0;
int ret;
+ struct intel_iommu *iommu;
BUG_ON(dir == DMA_NONE);
if (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO)
if (!domain)
return 0;
+ iommu = domain_get_iommu(domain);
size = aligned_size((u64)paddr, size);
- iova = __intel_alloc_iova(hwdev, domain, size);
+ iova = __intel_alloc_iova(hwdev, domain, size, pdev->dma_mask);
if (!iova)
goto error;
- start_paddr = iova->pfn_lo << PAGE_SHIFT_4K;
+ start_paddr = (phys_addr_t)iova->pfn_lo << PAGE_SHIFT;
/*
* Check if DMAR supports zero-length reads on write only
* mappings..
*/
if (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL || \
- !cap_zlr(domain->iommu->cap))
+ !cap_zlr(iommu->cap))
prot |= DMA_PTE_READ;
if (dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL)
prot |= DMA_PTE_WRITE;
* is not a big problem
*/
ret = domain_page_mapping(domain, start_paddr,
- ((u64)paddr) & PAGE_MASK_4K, size, prot);
+ ((u64)paddr) & PAGE_MASK, size, prot);
if (ret)
goto error;
- pr_debug("Device %s request: %lx@%llx mapping: %lx@%llx, dir %d\n",
- pci_name(pdev), size, (u64)paddr,
- size, (u64)start_paddr, dir);
-
/* it's a non-present to present mapping */
- ret = iommu_flush_iotlb_psi(domain->iommu, domain->id,
- start_paddr, size >> PAGE_SHIFT_4K, 1);
+ ret = iommu_flush_iotlb_psi(iommu, domain->id,
+ start_paddr, size >> VTD_PAGE_SHIFT, 1);
if (ret)
- iommu_flush_write_buffer(domain->iommu);
+ iommu_flush_write_buffer(iommu);
- return (start_paddr + ((u64)paddr & (~PAGE_MASK_4K)));
+ return start_paddr + ((u64)paddr & (~PAGE_MASK));
error:
if (iova)
__free_iova(&domain->iovad, iova);
printk(KERN_ERR"Device %s request: %lx@%llx dir %d --- failed\n",
- pci_name(pdev), size, (u64)paddr, dir);
+ pci_name(pdev), size, (unsigned long long)paddr, dir);
return 0;
}
+dma_addr_t intel_map_single(struct device *hwdev, phys_addr_t paddr,
+ size_t size, int dir)
+{
+ return __intel_map_single(hwdev, paddr, size, dir,
+ to_pci_dev(hwdev)->dma_mask);
+}
+
static void flush_unmaps(void)
{
int i, j;
/* just flush them all */
for (i = 0; i < g_num_of_iommus; i++) {
- if (deferred_flush[i].next) {
- struct intel_iommu *iommu =
- deferred_flush[i].domain[0]->iommu;
+ struct intel_iommu *iommu = g_iommus[i];
+ if (!iommu)
+ continue;
- iommu_flush_iotlb_global(iommu, 0);
+ if (deferred_flush[i].next) {
+ iommu->flush.flush_iotlb(iommu, 0, 0, 0,
+ DMA_TLB_GLOBAL_FLUSH, 0);
for (j = 0; j < deferred_flush[i].next; j++) {
__free_iova(&deferred_flush[i].domain[j]->iovad,
deferred_flush[i].iova[j]);
{
unsigned long flags;
int next, iommu_id;
+ struct intel_iommu *iommu;
spin_lock_irqsave(&async_umap_flush_lock, flags);
if (list_size == HIGH_WATER_MARK)
flush_unmaps();
- iommu_id = dom->iommu->seq_id;
+ iommu = domain_get_iommu(dom);
+ iommu_id = iommu->seq_id;
next = deferred_flush[iommu_id].next;
deferred_flush[iommu_id].domain[next] = dom;
spin_unlock_irqrestore(&async_umap_flush_lock, flags);
}
-static void intel_unmap_single(struct device *dev, dma_addr_t dev_addr,
- size_t size, int dir)
+void intel_unmap_single(struct device *dev, dma_addr_t dev_addr, size_t size,
+ int dir)
{
struct pci_dev *pdev = to_pci_dev(dev);
struct dmar_domain *domain;
unsigned long start_addr;
struct iova *iova;
+ struct intel_iommu *iommu;
if (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO)
return;
domain = find_domain(pdev);
BUG_ON(!domain);
+ iommu = domain_get_iommu(domain);
+
iova = find_iova(&domain->iovad, IOVA_PFN(dev_addr));
if (!iova)
return;
- start_addr = iova->pfn_lo << PAGE_SHIFT_4K;
+ start_addr = iova->pfn_lo << PAGE_SHIFT;
size = aligned_size((u64)dev_addr, size);
pr_debug("Device %s unmapping: %lx@%llx\n",
- pci_name(pdev), size, (u64)start_addr);
+ pci_name(pdev), size, (unsigned long long)start_addr);
/* clear the whole page */
dma_pte_clear_range(domain, start_addr, start_addr + size);
/* free page tables */
dma_pte_free_pagetable(domain, start_addr, start_addr + size);
if (intel_iommu_strict) {
- if (iommu_flush_iotlb_psi(domain->iommu,
- domain->id, start_addr, size >> PAGE_SHIFT_4K, 0))
- iommu_flush_write_buffer(domain->iommu);
+ if (iommu_flush_iotlb_psi(iommu,
+ domain->id, start_addr, size >> VTD_PAGE_SHIFT, 0))
+ iommu_flush_write_buffer(iommu);
/* free iova */
__free_iova(&domain->iovad, iova);
} else {
}
}
-static void * intel_alloc_coherent(struct device *hwdev, size_t size,
- dma_addr_t *dma_handle, gfp_t flags)
+void *intel_alloc_coherent(struct device *hwdev, size_t size,
+ dma_addr_t *dma_handle, gfp_t flags)
{
void *vaddr;
int order;
- size = PAGE_ALIGN_4K(size);
+ size = PAGE_ALIGN(size);
order = get_order(size);
flags &= ~(GFP_DMA | GFP_DMA32);
return NULL;
memset(vaddr, 0, size);
- *dma_handle = intel_map_single(hwdev, virt_to_bus(vaddr), size, DMA_BIDIRECTIONAL);
+ *dma_handle = __intel_map_single(hwdev, virt_to_bus(vaddr), size,
+ DMA_BIDIRECTIONAL,
+ hwdev->coherent_dma_mask);
if (*dma_handle)
return vaddr;
free_pages((unsigned long)vaddr, order);
return NULL;
}
-static void intel_free_coherent(struct device *hwdev, size_t size,
- void *vaddr, dma_addr_t dma_handle)
+void intel_free_coherent(struct device *hwdev, size_t size, void *vaddr,
+ dma_addr_t dma_handle)
{
int order;
- size = PAGE_ALIGN_4K(size);
+ size = PAGE_ALIGN(size);
order = get_order(size);
intel_unmap_single(hwdev, dma_handle, size, DMA_BIDIRECTIONAL);
}
#define SG_ENT_VIRT_ADDRESS(sg) (sg_virt((sg)))
-static void intel_unmap_sg(struct device *hwdev, struct scatterlist *sglist,
- int nelems, int dir)
+
+void intel_unmap_sg(struct device *hwdev, struct scatterlist *sglist,
+ int nelems, int dir)
{
int i;
struct pci_dev *pdev = to_pci_dev(hwdev);
size_t size = 0;
void *addr;
struct scatterlist *sg;
+ struct intel_iommu *iommu;
if (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO)
return;
domain = find_domain(pdev);
+ BUG_ON(!domain);
+
+ iommu = domain_get_iommu(domain);
iova = find_iova(&domain->iovad, IOVA_PFN(sglist[0].dma_address));
if (!iova)
size += aligned_size((u64)addr, sg->length);
}
- start_addr = iova->pfn_lo << PAGE_SHIFT_4K;
+ start_addr = iova->pfn_lo << PAGE_SHIFT;
/* clear the whole page */
dma_pte_clear_range(domain, start_addr, start_addr + size);
/* free page tables */
dma_pte_free_pagetable(domain, start_addr, start_addr + size);
- if (iommu_flush_iotlb_psi(domain->iommu, domain->id, start_addr,
- size >> PAGE_SHIFT_4K, 0))
- iommu_flush_write_buffer(domain->iommu);
+ if (iommu_flush_iotlb_psi(iommu, domain->id, start_addr,
+ size >> VTD_PAGE_SHIFT, 0))
+ iommu_flush_write_buffer(iommu);
/* free iova */
__free_iova(&domain->iovad, iova);
return nelems;
}
-static int intel_map_sg(struct device *hwdev, struct scatterlist *sglist,
- int nelems, int dir)
+int intel_map_sg(struct device *hwdev, struct scatterlist *sglist, int nelems,
+ int dir)
{
void *addr;
int i;
int ret;
struct scatterlist *sg;
unsigned long start_addr;
+ struct intel_iommu *iommu;
BUG_ON(dir == DMA_NONE);
if (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO)
if (!domain)
return 0;
+ iommu = domain_get_iommu(domain);
+
for_each_sg(sglist, sg, nelems, i) {
addr = SG_ENT_VIRT_ADDRESS(sg);
addr = (void *)virt_to_phys(addr);
size += aligned_size((u64)addr, sg->length);
}
- iova = __intel_alloc_iova(hwdev, domain, size);
+ iova = __intel_alloc_iova(hwdev, domain, size, pdev->dma_mask);
if (!iova) {
sglist->dma_length = 0;
return 0;
* mappings..
*/
if (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL || \
- !cap_zlr(domain->iommu->cap))
+ !cap_zlr(iommu->cap))
prot |= DMA_PTE_READ;
if (dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL)
prot |= DMA_PTE_WRITE;
- start_addr = iova->pfn_lo << PAGE_SHIFT_4K;
+ start_addr = iova->pfn_lo << PAGE_SHIFT;
offset = 0;
for_each_sg(sglist, sg, nelems, i) {
addr = SG_ENT_VIRT_ADDRESS(sg);
addr = (void *)virt_to_phys(addr);
size = aligned_size((u64)addr, sg->length);
ret = domain_page_mapping(domain, start_addr + offset,
- ((u64)addr) & PAGE_MASK_4K,
+ ((u64)addr) & PAGE_MASK,
size, prot);
if (ret) {
/* clear the page */
return 0;
}
sg->dma_address = start_addr + offset +
- ((u64)addr & (~PAGE_MASK_4K));
+ ((u64)addr & (~PAGE_MASK));
sg->dma_length = sg->length;
offset += size;
}
/* it's a non-present to present mapping */
- if (iommu_flush_iotlb_psi(domain->iommu, domain->id,
- start_addr, offset >> PAGE_SHIFT_4K, 1))
- iommu_flush_write_buffer(domain->iommu);
+ if (iommu_flush_iotlb_psi(iommu, domain->id,
+ start_addr, offset >> VTD_PAGE_SHIFT, 1))
+ iommu_flush_write_buffer(iommu);
return nelems;
}
sizeof(struct device_domain_info),
0,
SLAB_HWCACHE_ALIGN,
-
NULL);
if (!iommu_devinfo_cache) {
printk(KERN_ERR "Couldn't create devinfo cache\n");
sizeof(struct iova),
0,
SLAB_HWCACHE_ALIGN,
-
NULL);
if (!iommu_iova_cache) {
printk(KERN_ERR "Couldn't create iova cache\n");
init_timer(&unmap_timer);
force_iommu = 1;
dma_ops = &intel_dma_ops;
+
+ register_iommu(&intel_iommu_ops);
+
+ return 0;
+}
+
+static int vm_domain_add_dev_info(struct dmar_domain *domain,
+ struct pci_dev *pdev)
+{
+ struct device_domain_info *info;
+ unsigned long flags;
+
+ info = alloc_devinfo_mem();
+ if (!info)
+ return -ENOMEM;
+
+ info->bus = pdev->bus->number;
+ info->devfn = pdev->devfn;
+ info->dev = pdev;
+ info->domain = domain;
+
+ spin_lock_irqsave(&device_domain_lock, flags);
+ list_add(&info->link, &domain->devices);
+ list_add(&info->global, &device_domain_list);
+ pdev->dev.archdata.iommu = info;
+ spin_unlock_irqrestore(&device_domain_lock, flags);
+
+ return 0;
+}
+
+static void vm_domain_remove_one_dev_info(struct dmar_domain *domain,
+ struct pci_dev *pdev)
+{
+ struct device_domain_info *info;
+ struct intel_iommu *iommu;
+ unsigned long flags;
+ int found = 0;
+ struct list_head *entry, *tmp;
+
+ iommu = device_to_iommu(pdev->bus->number, pdev->devfn);
+ if (!iommu)
+ return;
+
+ spin_lock_irqsave(&device_domain_lock, flags);
+ list_for_each_safe(entry, tmp, &domain->devices) {
+ info = list_entry(entry, struct device_domain_info, link);
+ if (info->bus == pdev->bus->number &&
+ info->devfn == pdev->devfn) {
+ list_del(&info->link);
+ list_del(&info->global);
+ if (info->dev)
+ info->dev->dev.archdata.iommu = NULL;
+ spin_unlock_irqrestore(&device_domain_lock, flags);
+
+ iommu_detach_dev(iommu, info->bus, info->devfn);
+ free_devinfo_mem(info);
+
+ spin_lock_irqsave(&device_domain_lock, flags);
+
+ if (found)
+ break;
+ else
+ continue;
+ }
+
+ /* if there is no other devices under the same iommu
+ * owned by this domain, clear this iommu in iommu_bmp
+ * update iommu count and coherency
+ */
+ if (device_to_iommu(info->bus, info->devfn) == iommu)
+ found = 1;
+ }
+
+ if (found == 0) {
+ unsigned long tmp_flags;
+ spin_lock_irqsave(&domain->iommu_lock, tmp_flags);
+ clear_bit(iommu->seq_id, &domain->iommu_bmp);
+ domain->iommu_count--;
+ domain_update_iommu_coherency(domain);
+ spin_unlock_irqrestore(&domain->iommu_lock, tmp_flags);
+ }
+
+ spin_unlock_irqrestore(&device_domain_lock, flags);
+}
+
+static void vm_domain_remove_all_dev_info(struct dmar_domain *domain)
+{
+ struct device_domain_info *info;
+ struct intel_iommu *iommu;
+ unsigned long flags1, flags2;
+
+ spin_lock_irqsave(&device_domain_lock, flags1);
+ while (!list_empty(&domain->devices)) {
+ info = list_entry(domain->devices.next,
+ struct device_domain_info, link);
+ list_del(&info->link);
+ list_del(&info->global);
+ if (info->dev)
+ info->dev->dev.archdata.iommu = NULL;
+
+ spin_unlock_irqrestore(&device_domain_lock, flags1);
+
+ iommu = device_to_iommu(info->bus, info->devfn);
+ iommu_detach_dev(iommu, info->bus, info->devfn);
+
+ /* clear this iommu in iommu_bmp, update iommu count
+ * and coherency
+ */
+ spin_lock_irqsave(&domain->iommu_lock, flags2);
+ if (test_and_clear_bit(iommu->seq_id,
+ &domain->iommu_bmp)) {
+ domain->iommu_count--;
+ domain_update_iommu_coherency(domain);
+ }
+ spin_unlock_irqrestore(&domain->iommu_lock, flags2);
+
+ free_devinfo_mem(info);
+ spin_lock_irqsave(&device_domain_lock, flags1);
+ }
+ spin_unlock_irqrestore(&device_domain_lock, flags1);
+}
+
+/* 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;
+
+ domain = alloc_domain_mem();
+ if (!domain)
+ return NULL;
+
+ domain->id = vm_domid++;
+ memset(&domain->iommu_bmp, 0, sizeof(unsigned long));
+ domain->flags = DOMAIN_FLAG_VIRTUAL_MACHINE;
+
+ return domain;
+}
+
+static int vm_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);
+
+ /* calculate AGAW */
+ domain->gaw = guest_width;
+ adjust_width = guestwidth_to_adjustwidth(guest_width);
+ domain->agaw = width_to_agaw(adjust_width);
+
+ INIT_LIST_HEAD(&domain->devices);
+
+ domain->iommu_count = 0;
+ domain->iommu_coherency = 0;
+ domain->max_addr = 0;
+
+ /* always allocate the top pgd */
+ domain->pgd = (struct dma_pte *)alloc_pgtable_page();
+ if (!domain->pgd)
+ return -ENOMEM;
+ domain_flush_cache(domain, domain->pgd, PAGE_SIZE);
+ return 0;
+}
+
+static void iommu_free_vm_domain(struct dmar_domain *domain)
+{
+ unsigned long flags;
+ struct dmar_drhd_unit *drhd;
+ struct intel_iommu *iommu;
+ unsigned long i;
+ unsigned long ndomains;
+
+ for_each_drhd_unit(drhd) {
+ if (drhd->ignored)
+ continue;
+ iommu = drhd->iommu;
+
+ ndomains = cap_ndoms(iommu->cap);
+ i = find_first_bit(iommu->domain_ids, ndomains);
+ for (; i < ndomains; ) {
+ if (iommu->domains[i] == domain) {
+ spin_lock_irqsave(&iommu->lock, flags);
+ clear_bit(i, iommu->domain_ids);
+ iommu->domains[i] = NULL;
+ spin_unlock_irqrestore(&iommu->lock, flags);
+ break;
+ }
+ i = find_next_bit(iommu->domain_ids, ndomains, i+1);
+ }
+ }
+}
+
+static void vm_domain_exit(struct dmar_domain *domain)
+{
+ u64 end;
+
+ /* Domain 0 is reserved, so dont process it */
+ if (!domain)
+ return;
+
+ 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);
+
+ /* free page tables */
+ dma_pte_free_pagetable(domain, 0, end);
+
+ iommu_free_vm_domain(domain);
+ free_domain_mem(domain);
+}
+
+static int intel_iommu_domain_init(struct iommu_domain *domain)
+{
+ struct dmar_domain *dmar_domain;
+
+ dmar_domain = iommu_alloc_vm_domain();
+ if (!dmar_domain) {
+ printk(KERN_ERR
+ "intel_iommu_domain_init: dmar_domain == NULL\n");
+ return -ENOMEM;
+ }
+ if (vm_domain_init(dmar_domain, DEFAULT_DOMAIN_ADDRESS_WIDTH)) {
+ printk(KERN_ERR
+ "intel_iommu_domain_init() failed\n");
+ vm_domain_exit(dmar_domain);
+ return -ENOMEM;
+ }
+ domain->priv = dmar_domain;
+
return 0;
}
+static void intel_iommu_domain_destroy(struct iommu_domain *domain)
+{
+ struct dmar_domain *dmar_domain = domain->priv;
+
+ domain->priv = NULL;
+ vm_domain_exit(dmar_domain);
+}
+
+static int intel_iommu_attach_device(struct iommu_domain *domain,
+ struct device *dev)
+{
+ struct dmar_domain *dmar_domain = domain->priv;
+ struct pci_dev *pdev = to_pci_dev(dev);
+ struct intel_iommu *iommu;
+ int addr_width;
+ u64 end;
+ int ret;
+
+ /* normally pdev is not mapped */
+ if (unlikely(domain_context_mapped(pdev))) {
+ struct dmar_domain *old_domain;
+
+ old_domain = find_domain(pdev);
+ if (old_domain) {
+ if (dmar_domain->flags & DOMAIN_FLAG_VIRTUAL_MACHINE)
+ vm_domain_remove_one_dev_info(old_domain, pdev);
+ else
+ domain_remove_dev_info(old_domain);
+ }
+ }
+
+ iommu = device_to_iommu(pdev->bus->number, pdev->devfn);
+ if (!iommu)
+ return -ENODEV;
+
+ /* 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 "
+ "sufficient for the mapped address (%llx)\n",
+ __func__, iommu->agaw, dmar_domain->max_addr);
+ return -EFAULT;
+ }
+
+ ret = domain_context_mapping(dmar_domain, pdev);
+ if (ret)
+ return ret;
+
+ ret = vm_domain_add_dev_info(dmar_domain, pdev);
+ return ret;
+}
+
+static void intel_iommu_detach_device(struct iommu_domain *domain,
+ struct device *dev)
+{
+ struct dmar_domain *dmar_domain = domain->priv;
+ struct pci_dev *pdev = to_pci_dev(dev);
+
+ vm_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)
+{
+ struct dmar_domain *dmar_domain = domain->priv;
+ u64 max_addr;
+ int addr_width;
+ int prot = 0;
+ int ret;
+
+ if (iommu_prot & IOMMU_READ)
+ prot |= DMA_PTE_READ;
+ if (iommu_prot & IOMMU_WRITE)
+ prot |= DMA_PTE_WRITE;
+
+ max_addr = (iova & VTD_PAGE_MASK) + VTD_PAGE_ALIGN(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;
+ if (end < max_addr) {
+ printk(KERN_ERR "%s: iommu agaw (%d) is not "
+ "sufficient for the mapped address (%llx)\n",
+ __func__, min_agaw, max_addr);
+ return -EFAULT;
+ }
+ dmar_domain->max_addr = max_addr;
+ }
+
+ ret = domain_page_mapping(dmar_domain, iova, hpa, size, prot);
+ return ret;
+}
+
+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 (dmar_domain->max_addr == base + size)
+ dmar_domain->max_addr = base;
+}
+
+static phys_addr_t intel_iommu_iova_to_phys(struct iommu_domain *domain,
+ unsigned long iova)
+{
+ struct dmar_domain *dmar_domain = domain->priv;
+ struct dma_pte *pte;
+ u64 phys = 0;
+
+ pte = addr_to_dma_pte(dmar_domain, iova);
+ if (pte)
+ phys = dma_pte_addr(pte);
+
+ return phys;
+}
+
+static struct iommu_ops intel_iommu_ops = {
+ .domain_init = intel_iommu_domain_init,
+ .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,
+ .iova_to_phys = intel_iommu_iova_to_phys,
+};