#include <linux/iommu.h>
#include <linux/intel-iommu.h>
#include <linux/sysdev.h>
+#include <linux/tboot.h>
+#include <linux/dmi.h>
#include <asm/cacheflush.h>
#include <asm/iommu.h>
#include "pci.h"
#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 IS_AZALIA(pdev) ((pdev)->vendor == 0x8086 && (pdev)->device == 0x3a3e)
#define IOAPIC_RANGE_START (0xfee00000)
#define IOAPIC_RANGE_END (0xfeefffff)
#define DEFAULT_DOMAIN_ADDRESS_WIDTH 48
-#define DOMAIN_MAX_ADDR(gaw) ((((u64)1) << gaw) - 1)
+#define MAX_AGAW_WIDTH 64
+
+#define __DOMAIN_MAX_PFN(gaw) ((((uint64_t)1) << (gaw-VTD_PAGE_SHIFT)) - 1)
+#define __DOMAIN_MAX_ADDR(gaw) ((((uint64_t)1) << gaw) - 1)
+
+/* We limit DOMAIN_MAX_PFN to fit in an unsigned long, and DOMAIN_MAX_ADDR
+ to match. That way, we can use 'unsigned long' for PFNs with impunity. */
+#define DOMAIN_MAX_PFN(gaw) ((unsigned long) min_t(uint64_t, \
+ __DOMAIN_MAX_PFN(gaw), (unsigned long)-1))
+#define DOMAIN_MAX_ADDR(gaw) (((uint64_t)__DOMAIN_MAX_PFN(gaw)) << VTD_PAGE_SHIFT)
#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)
+#define DMA_32BIT_PFN IOVA_PFN(DMA_BIT_MASK(32))
+#define DMA_64BIT_PFN IOVA_PFN(DMA_BIT_MASK(64))
+
+
+/* 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;
+static void __init check_tylersburg_isoch(void);
static int rwbf_quirk;
/*
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)
{
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)
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.
+ * 2. It maps to each iommu if successful.
+ * 3. Each iommu mapps to this domain if successful.
+ */
+static struct dmar_domain *si_domain;
+static int hw_pass_through = 1;
+
/* devices under the same p2p bridge are owned in one domain */
#define DOMAIN_FLAG_P2P_MULTIPLE_DEVICES (1 << 0)
*/
#define DOMAIN_FLAG_VIRTUAL_MACHINE (1 << 1)
+/* si_domain contains mulitple devices */
+#define DOMAIN_FLAG_STATIC_IDENTITY (1 << 2)
+
struct dmar_domain {
int id; /* domain id */
unsigned long iommu_bmp; /* bitmap of iommus this domain uses*/
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 */
u8 bus; /* PCI bus number */
u8 devfn; /* PCI devfn number */
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 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)
+static int __iommu_calculate_agaw(struct intel_iommu *iommu, int max_gaw)
{
unsigned long sagaw;
int agaw = -1;
sagaw = cap_sagaw(iommu->cap);
- for (agaw = width_to_agaw(DEFAULT_DOMAIN_ADDRESS_WIDTH);
+ for (agaw = width_to_agaw(max_gaw);
agaw >= 0; agaw--) {
if (test_bit(agaw, &sagaw))
break;
return agaw;
}
-/* in native case, each domain is related to only one iommu */
+/*
+ * Calculate max SAGAW for each iommu.
+ */
+int iommu_calculate_max_sagaw(struct intel_iommu *iommu)
+{
+ return __iommu_calculate_agaw(iommu, MAX_AGAW_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)
+{
+ return __iommu_calculate_agaw(iommu, DEFAULT_DOMAIN_ADDRESS_WIDTH);
+}
+
+/* This functionin only returns single iommu in a domain */
static struct intel_iommu *domain_get_iommu(struct dmar_domain *domain)
{
int iommu_id;
+ /* si_domain and vm domain should not get here. */
BUG_ON(domain->flags & DOMAIN_FLAG_VIRTUAL_MACHINE);
+ BUG_ON(domain->flags & DOMAIN_FLAG_STATIC_IDENTITY);
iommu_id = find_first_bit(&domain->iommu_bmp, g_num_of_iommus);
if (iommu_id < 0 || iommu_id >= g_num_of_iommus)
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 = ((uint64_t)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);
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;
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)
-{
- int addr_width = agaw_to_width(domain->agaw);
- int npages;
+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) - VTD_PAGE_SHIFT;
+ struct dma_pte *first_pte, *pte;
+
+ BUG_ON(addr_width < BITS_PER_LONG && start_pfn >> addr_width);
+ BUG_ON(addr_width < BITS_PER_LONG && last_pfn >> addr_width);
+ BUG_ON(start_pfn > last_pfn);
+
+ /* we don't need lock here; nobody else touches the iova range */
+ do {
+ 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));
- start &= (((u64)1) << addr_width) - 1;
- end &= (((u64)1) << addr_width) - 1;
- /* in case it's partial page */
- start = PAGE_ALIGN(start);
- end &= PAGE_MASK;
- npages = (end - start) / VTD_PAGE_SIZE;
+ domain_flush_cache(domain, first_pte,
+ (void *)pte - (void *)first_pte);
- /* we don't need lock here, nobody else touches the iova range */
- while (npages--) {
- dma_pte_clear_one(domain, start);
- start += VTD_PAGE_SIZE;
- }
+ } while (start_pfn && start_pfn <= last_pfn);
}
/* 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);
+ BUG_ON(start_pfn > last_pfn);
- /* 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));
+ do {
+ 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);
+
+ } while (tmp && tmp + level_size(level) - 1 <= last_pfn);
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;
}
static void iommu_set_root_entry(struct intel_iommu *iommu)
{
void *addr;
- u32 cmd, sts;
+ u32 sts;
unsigned long flag;
addr = iommu->root_entry;
spin_lock_irqsave(&iommu->register_lock, flag);
dmar_writeq(iommu->reg + DMAR_RTADDR_REG, virt_to_phys(addr));
- cmd = iommu->gcmd | DMA_GCMD_SRTP;
- writel(cmd, iommu->reg + DMAR_GCMD_REG);
+ writel(iommu->gcmd | DMA_GCMD_SRTP, iommu->reg + DMAR_GCMD_REG);
/* Make sure hardware complete it */
IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG,
- readl, (sts & DMA_GSTS_RTPS), sts);
+ readl, (sts & DMA_GSTS_RTPS), sts);
spin_unlock_irqrestore(&iommu->register_lock, flag);
}
if (!rwbf_quirk && !cap_rwbf(iommu->cap))
return;
- val = iommu->gcmd | DMA_GCMD_WBF;
spin_lock_irqsave(&iommu->register_lock, flag);
- writel(val, iommu->reg + DMAR_GCMD_REG);
+ writel(iommu->gcmd | DMA_GCMD_WBF, iommu->reg + DMAR_GCMD_REG);
/* Make sure hardware complete it */
IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG,
- readl, (!(val & DMA_GSTS_WBFS)), val);
+ readl, (!(val & DMA_GSTS_WBFS)), val);
spin_unlock_irqrestore(&iommu->register_lock, flag);
}
/* return value determine if we need a write buffer flush */
-static int __iommu_flush_context(struct intel_iommu *iommu,
- u16 did, u16 source_id, u8 function_mask, u64 type,
- int non_present_entry_flush)
+static void __iommu_flush_context(struct intel_iommu *iommu,
+ u16 did, u16 source_id, u8 function_mask,
+ u64 type)
{
u64 val = 0;
unsigned long flag;
- /*
- * In the non-present entry flush case, if hardware doesn't cache
- * non-present entry we do nothing and if hardware cache non-present
- * entry, we flush entries of domain 0 (the domain id is used to cache
- * any non-present entries)
- */
- if (non_present_entry_flush) {
- if (!cap_caching_mode(iommu->cap))
- return 1;
- else
- did = 0;
- }
-
switch (type) {
case DMA_CCMD_GLOBAL_INVL:
val = DMA_CCMD_GLOBAL_INVL;
dmar_readq, (!(val & DMA_CCMD_ICC)), val);
spin_unlock_irqrestore(&iommu->register_lock, flag);
-
- /* flush context entry will implicitly flush write buffer */
- return 0;
}
/* 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,
- int non_present_entry_flush)
+static void __iommu_flush_iotlb(struct intel_iommu *iommu, u16 did,
+ u64 addr, unsigned int size_order, u64 type)
{
int tlb_offset = ecap_iotlb_offset(iommu->ecap);
u64 val = 0, val_iva = 0;
unsigned long flag;
- /*
- * In the non-present entry flush case, if hardware doesn't cache
- * non-present entry we do nothing and if hardware cache non-present
- * entry, we flush entries of domain 0 (the domain id is used to cache
- * any non-present entries)
- */
- if (non_present_entry_flush) {
- if (!cap_caching_mode(iommu->cap))
- return 1;
- else
- did = 0;
- }
-
switch (type) {
case DMA_TLB_GLOBAL_FLUSH:
/* global flush doesn't need set IVA_REG */
pr_debug("IOMMU: tlb flush request %Lx, actual %Lx\n",
(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 iommu_flush_iotlb_psi(struct intel_iommu *iommu, u16 did,
- u64 addr, unsigned int pages, int non_present_entry_flush)
+static struct device_domain_info *iommu_support_dev_iotlb(
+ struct dmar_domain *domain, int segment, u8 bus, u8 devfn)
{
- unsigned int mask;
+ int found = 0;
+ unsigned long flags;
+ struct device_domain_info *info;
+ struct intel_iommu *iommu = device_to_iommu(segment, bus, devfn);
- BUG_ON(addr & (~VTD_PAGE_MASK));
- BUG_ON(pages == 0);
+ if (!ecap_dev_iotlb_support(iommu->ecap))
+ return NULL;
+
+ if (!iommu->qi)
+ return NULL;
+
+ spin_lock_irqsave(&device_domain_lock, flags);
+ list_for_each_entry(info, &domain->devices, link)
+ if (info->bus == bus && info->devfn == devfn) {
+ found = 1;
+ break;
+ }
+ spin_unlock_irqrestore(&device_domain_lock, flags);
+
+ if (!found || !info->dev)
+ return NULL;
+
+ if (!pci_find_ext_capability(info->dev, PCI_EXT_CAP_ID_ATS))
+ return NULL;
+
+ if (!dmar_find_matched_atsr_unit(info->dev))
+ return NULL;
+
+ info->iommu = iommu;
- /* Fallback to domain selective flush if no PSI support */
- if (!cap_pgsel_inv(iommu->cap))
- return iommu->flush.flush_iotlb(iommu, did, 0, 0,
- DMA_TLB_DSI_FLUSH,
- non_present_entry_flush);
+ return info;
+}
+
+static void iommu_enable_dev_iotlb(struct device_domain_info *info)
+{
+ if (!info)
+ return;
+
+ pci_enable_ats(info->dev, VTD_PAGE_SHIFT);
+}
+
+static void iommu_disable_dev_iotlb(struct device_domain_info *info)
+{
+ if (!info->dev || !pci_ats_enabled(info->dev))
+ return;
+
+ pci_disable_ats(info->dev);
+}
+
+static void iommu_flush_dev_iotlb(struct dmar_domain *domain,
+ u64 addr, unsigned mask)
+{
+ u16 sid, qdep;
+ unsigned long flags;
+ struct device_domain_info *info;
+
+ spin_lock_irqsave(&device_domain_lock, flags);
+ list_for_each_entry(info, &domain->devices, link) {
+ if (!info->dev || !pci_ats_enabled(info->dev))
+ continue;
+
+ sid = info->bus << 8 | info->devfn;
+ qdep = pci_ats_queue_depth(info->dev);
+ qi_flush_dev_iotlb(info->iommu, sid, qdep, addr, mask);
+ }
+ spin_unlock_irqrestore(&device_domain_lock, flags);
+}
+
+static void iommu_flush_iotlb_psi(struct intel_iommu *iommu, u16 did,
+ 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(pages == 0);
/*
+ * Fallback to domain selective flush if no PSI support or the size is
+ * too big.
* PSI requires page size to be 2 ^ x, and the base address is naturally
* aligned to the size
*/
- 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.flush_iotlb(iommu, did, 0, 0,
- DMA_TLB_DSI_FLUSH, non_present_entry_flush);
+ if (!cap_pgsel_inv(iommu->cap) || mask > cap_max_amask_val(iommu->cap))
+ iommu->flush.flush_iotlb(iommu, did, 0, 0,
+ DMA_TLB_DSI_FLUSH);
+ else
+ iommu->flush.flush_iotlb(iommu, did, addr, mask,
+ DMA_TLB_PSI_FLUSH);
- return iommu->flush.flush_iotlb(iommu, did, addr, mask,
- DMA_TLB_PSI_FLUSH,
- non_present_entry_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.
+ */
+ if (!cap_caching_mode(iommu->cap) || did)
+ iommu_flush_dev_iotlb(iommu->domains[did], addr, mask);
}
static void iommu_disable_protect_mem_regions(struct intel_iommu *iommu)
unsigned long flags;
spin_lock_irqsave(&iommu->register_lock, flags);
- writel(iommu->gcmd|DMA_GCMD_TE, iommu->reg + DMAR_GCMD_REG);
+ iommu->gcmd |= DMA_GCMD_TE;
+ writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
/* Make sure hardware complete it */
IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG,
- readl, (sts & DMA_GSTS_TES), sts);
+ readl, (sts & DMA_GSTS_TES), sts);
- iommu->gcmd |= DMA_GCMD_TE;
spin_unlock_irqrestore(&iommu->register_lock, flags);
return 0;
}
/* Make sure hardware complete it */
IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG,
- readl, (!(sts & DMA_GSTS_TES)), sts);
+ readl, (!(sts & DMA_GSTS_TES)), sts);
spin_unlock_irqrestore(&iommu->register_lock, flag);
return 0;
pr_debug("Number of Domains supportd <%ld>\n", ndomains);
nlongs = BITS_TO_LONGS(ndomains);
+ spin_lock_init(&iommu->lock);
+
/* TBD: there might be 64K domains,
* consider other allocation for future chip
*/
GFP_KERNEL);
if (!iommu->domains) {
printk(KERN_ERR "Allocating domain array failed\n");
- kfree(iommu->domain_ids);
return -ENOMEM;
}
- spin_lock_init(&iommu->lock);
-
/*
* if Caching mode is set, then invalid translations are tagged
* with domainid 0. Hence we need to pre-allocate it.
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);
+ if ((iommu->domains) && (iommu->domain_ids)) {
+ 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);
+
+ 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);
- 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);
+ i = find_next_bit(iommu->domain_ids,
+ cap_ndoms(iommu->cap), i+1);
}
- spin_unlock_irqrestore(&domain->iommu_lock, flags);
-
- i = find_next_bit(iommu->domain_ids,
- cap_ndoms(iommu->cap), i+1);
}
if (iommu->gcmd & DMA_GCMD_TE)
free_context_table(iommu);
}
-static struct dmar_domain * iommu_alloc_domain(struct intel_iommu *iommu)
+static struct dmar_domain *alloc_domain(void)
{
- unsigned long num;
- unsigned long ndomains;
struct dmar_domain *domain;
- unsigned long flags;
domain = alloc_domain_mem();
if (!domain)
return NULL;
+ memset(&domain->iommu_bmp, 0, sizeof(unsigned long));
+ domain->flags = 0;
+
+ return domain;
+}
+
+static int iommu_attach_domain(struct dmar_domain *domain,
+ struct intel_iommu *iommu)
+{
+ int num;
+ unsigned long ndomains;
+ unsigned long flags;
+
ndomains = cap_ndoms(iommu->cap);
spin_lock_irqsave(&iommu->lock, flags);
+
num = find_first_zero_bit(iommu->domain_ids, ndomains);
if (num >= ndomains) {
spin_unlock_irqrestore(&iommu->lock, flags);
- free_domain_mem(domain);
printk(KERN_ERR "IOMMU: no free domain ids\n");
- return NULL;
+ return -ENOMEM;
}
- set_bit(num, iommu->domain_ids);
domain->id = num;
- memset(&domain->iommu_bmp, 0, sizeof(unsigned long));
+ set_bit(num, iommu->domain_ids);
set_bit(iommu->seq_id, &domain->iommu_bmp);
- domain->flags = 0;
iommu->domains[num] = domain;
spin_unlock_irqrestore(&iommu->lock, flags);
- return domain;
+ return 0;
}
-static void iommu_free_domain(struct dmar_domain *domain)
+static void iommu_detach_domain(struct dmar_domain *domain,
+ struct intel_iommu *iommu)
{
unsigned long flags;
- struct intel_iommu *iommu;
-
- iommu = domain_get_iommu(domain);
+ int num, ndomains;
+ int found = 0;
spin_lock_irqsave(&iommu->lock, flags);
- clear_bit(domain->id, iommu->domain_ids);
+ ndomains = cap_ndoms(iommu->cap);
+ num = find_first_bit(iommu->domain_ids, ndomains);
+ for (; num < ndomains; ) {
+ if (iommu->domains[num] == domain) {
+ found = 1;
+ break;
+ }
+ num = find_next_bit(iommu->domain_ids,
+ cap_ndoms(iommu->cap), num+1);
+ }
+
+ if (found) {
+ clear_bit(num, iommu->domain_ids);
+ clear_bit(iommu->seq_id, &domain->iommu_bmp);
+ iommu->domains[num] = NULL;
+ }
spin_unlock_irqrestore(&iommu->lock, flags);
}
static struct iova_domain reserved_iova_list;
-static struct lock_class_key reserved_alloc_key;
static struct lock_class_key reserved_rbtree_key;
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);
- lockdep_set_class(&reserved_iova_list.iova_alloc_lock,
- &reserved_alloc_key);
lockdep_set_class(&reserved_iova_list.iova_rbtree_lock,
&reserved_rbtree_key);
r = &pdev->resource[i];
if (!r->flags || !(r->flags & IORESOURCE_MEM))
continue;
- addr = r->start;
- addr &= 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");
}
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);
static void domain_exit(struct dmar_domain *domain)
{
- u64 end;
+ struct dmar_drhd_unit *drhd;
+ struct intel_iommu *iommu;
/* Domain 0 is reserved, so dont process it */
if (!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))
+ iommu_detach_domain(domain, iommu);
- iommu_free_domain(domain);
free_domain_mem(domain);
}
-static int domain_context_mapping_one(struct dmar_domain *domain,
- int segment, u8 bus, u8 devfn)
+static int domain_context_mapping_one(struct dmar_domain *domain, int segment,
+ u8 bus, u8 devfn, int translation)
{
struct context_entry *context;
unsigned long flags;
unsigned long ndomains;
int id;
int agaw;
+ struct device_domain_info *info = NULL;
pr_debug("Set context mapping for %02x:%02x.%d\n",
bus, PCI_SLOT(devfn), PCI_FUNC(devfn));
+
BUG_ON(!domain->pgd);
+ BUG_ON(translation != CONTEXT_TT_PASS_THROUGH &&
+ translation != CONTEXT_TT_MULTI_LEVEL);
iommu = device_to_iommu(segment, bus, devfn);
if (!iommu)
id = domain->id;
pgd = domain->pgd;
- if (domain->flags & DOMAIN_FLAG_VIRTUAL_MACHINE) {
+ if (domain->flags & DOMAIN_FLAG_VIRTUAL_MACHINE ||
+ domain->flags & DOMAIN_FLAG_STATIC_IDENTITY) {
int found = 0;
/* find an available domain id for this device in iommu */
}
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);
+
+ if (translation != CONTEXT_TT_PASS_THROUGH) {
+ info = iommu_support_dev_iotlb(domain, segment, bus, devfn);
+ translation = info ? CONTEXT_TT_DEV_IOTLB :
+ CONTEXT_TT_MULTI_LEVEL;
+ }
+ /*
+ * In pass through mode, AW must be programmed to indicate the largest
+ * AGAW value supported by hardware. And ASR is ignored by hardware.
+ */
+ if (unlikely(translation == CONTEXT_TT_PASS_THROUGH))
+ context_set_address_width(context, iommu->msagaw);
+ else {
+ context_set_address_root(context, virt_to_phys(pgd));
+ context_set_address_width(context, iommu->agaw);
+ }
+
+ context_set_translation_type(context, translation);
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.flush_context(iommu, domain->id,
- (((u16)bus) << 8) | devfn, DMA_CCMD_MASK_NOBIT,
- DMA_CCMD_DEVICE_INVL, 1))
+ /*
+ * It's a non-present to present mapping. If hardware doesn't cache
+ * non-present entry we only need to flush the write-buffer. If the
+ * _does_ cache non-present entries, then it does so in the special
+ * domain #0, which we have to flush:
+ */
+ if (cap_caching_mode(iommu->cap)) {
+ iommu->flush.flush_context(iommu, 0,
+ (((u16)bus) << 8) | devfn,
+ DMA_CCMD_MASK_NOBIT,
+ DMA_CCMD_DEVICE_INVL);
+ iommu->flush.flush_iotlb(iommu, 0, 0, 0, DMA_TLB_DSI_FLUSH);
+ } else {
iommu_flush_write_buffer(iommu);
- else
- iommu->flush.flush_iotlb(iommu, 0, 0, 0, DMA_TLB_DSI_FLUSH, 0);
-
+ }
+ iommu_enable_dev_iotlb(info);
spin_unlock_irqrestore(&iommu->lock, flags);
spin_lock_irqsave(&domain->iommu_lock, flags);
}
static int
-domain_context_mapping(struct dmar_domain *domain, struct pci_dev *pdev)
+domain_context_mapping(struct dmar_domain *domain, struct pci_dev *pdev,
+ int translation)
{
int ret;
struct pci_dev *tmp, *parent;
ret = domain_context_mapping_one(domain, pci_domain_nr(pdev->bus),
- pdev->bus->number, pdev->devfn);
+ pdev->bus->number, pdev->devfn,
+ translation);
if (ret)
return ret;
ret = domain_context_mapping_one(domain,
pci_domain_nr(parent->bus),
parent->bus->number,
- parent->devfn);
+ parent->devfn, translation);
if (ret)
return ret;
parent = parent->bus->self;
if (tmp->is_pcie) /* this is a PCIE-to-PCI bridge */
return domain_context_mapping_one(domain,
pci_domain_nr(tmp->subordinate),
- tmp->subordinate->number, 0);
+ tmp->subordinate->number, 0,
+ translation);
else /* this is a legacy PCI bridge */
return domain_context_mapping_one(domain,
pci_domain_nr(tmp->bus),
tmp->bus->number,
- tmp->devfn);
+ tmp->devfn,
+ translation);
}
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)
+/* 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 start_pfn, end_pfn;
- struct dma_pte *pte;
- int index;
- int addr_width = agaw_to_width(domain->agaw);
+ host_addr &= ~PAGE_MASK;
+ return PAGE_ALIGN(host_addr + size) >> VTD_PAGE_SHIFT;
+}
- hpa &= (((u64)1) << addr_width) - 1;
+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)
+{
+ 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;
+
+ 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 = aligned_nrpages(sg->offset, sg->length);
+ 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)
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);
+ DMA_CCMD_GLOBAL_INVL);
+ iommu->flush.flush_iotlb(iommu, 0, 0, 0, DMA_TLB_GLOBAL_FLUSH);
}
static void domain_remove_dev_info(struct dmar_domain *domain)
info->dev->dev.archdata.iommu = NULL;
spin_unlock_irqrestore(&device_domain_lock, flags);
+ iommu_disable_dev_iotlb(info);
iommu = device_to_iommu(info->segment, info->bus, info->devfn);
iommu_detach_dev(iommu, info->bus, info->devfn);
free_devinfo_mem(info);
unsigned long flags;
int bus = 0, devfn = 0;
int segment;
+ int ret;
domain = find_domain(pdev);
if (domain)
}
}
+ domain = alloc_domain();
+ if (!domain)
+ goto error;
+
/* Allocate new domain for the device */
drhd = dmar_find_matched_drhd_unit(pdev);
if (!drhd) {
}
iommu = drhd->iommu;
- domain = iommu_alloc_domain(iommu);
- if (!domain)
+ ret = iommu_attach_domain(domain, iommu);
+ if (ret) {
+ domain_exit(domain);
goto error;
+ }
if (domain_init(domain, gaw)) {
domain_exit(domain);
return find_domain(pdev);
}
+static int iommu_identity_mapping;
+#define IDENTMAP_ALL 1
+#define IDENTMAP_GFX 2
+#define IDENTMAP_AZALIA 4
+
+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);
- /* page table init */
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;
+ /* For _hardware_ passthrough, don't bother. But for software
+ passthrough, we do it anyway -- it may indicate a memory
+ range which is reserved in E820, so which didn't get set
+ up to start with in si_domain */
+ if (domain == si_domain && hw_pass_through) {
+ printk("Ignoring identity map for HW passthrough device %s [0x%Lx - 0x%Lx]\n",
+ pci_name(pdev), start, end);
+ return 0;
}
- 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);
+ printk(KERN_INFO
+ "IOMMU: Setting identity map for device %s [0x%Lx - 0x%Lx]\n",
+ pci_name(pdev), start, end);
+
+ 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",
+ agaw_to_width(domain->agaw),
+ 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;
+ }
- 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);
- if (!ret)
- return 0;
-error:
+ ret = domain_context_mapping(domain, pdev, CONTEXT_TT_MULTI_LEVEL);
+ if (ret)
+ goto error;
+
+ return 0;
+
+ error:
domain_exit(domain);
return ret;
-
}
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 {
+#ifdef CONFIG_DMAR_FLOPPY_WA
+static inline void iommu_prepare_isa(void)
+{
struct pci_dev *pdev;
int ret;
-};
-static int __init iommu_prepare_work_fn(unsigned long start_pfn,
- unsigned long end_pfn, void *datax)
+ pdev = pci_get_class(PCI_CLASS_BRIDGE_ISA << 8, NULL);
+ if (!pdev)
+ return;
+
+ 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 0-16MiB identity map; "
+ "floppy might not work\n");
+
+}
+#else
+static inline void iommu_prepare_isa(void)
{
- struct iommu_prepare_data *data;
+ return;
+}
+#endif /* !CONFIG_DMAR_FLPY_WA */
+
+static int md_domain_init(struct dmar_domain *domain, int guest_width);
- data = (struct iommu_prepare_data *)datax;
+static int __init si_domain_work_fn(unsigned long start_pfn,
+ unsigned long end_pfn, void *datax)
+{
+ int *ret = datax;
- data->ret = iommu_prepare_identity_map(data->pdev,
- start_pfn<<PAGE_SHIFT, end_pfn<<PAGE_SHIFT);
- return data->ret;
+ *ret = iommu_domain_identity_map(si_domain,
+ (uint64_t)start_pfn << PAGE_SHIFT,
+ (uint64_t)end_pfn << PAGE_SHIFT);
+ return *ret;
}
-static int __init iommu_prepare_with_active_regions(struct pci_dev *pdev)
+static int __init si_domain_init(int hw)
{
- int nid;
- struct iommu_prepare_data data;
+ struct dmar_drhd_unit *drhd;
+ struct intel_iommu *iommu;
+ int nid, ret = 0;
+
+ si_domain = alloc_domain();
+ if (!si_domain)
+ return -EFAULT;
- data.pdev = pdev;
- data.ret = 0;
+ 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);
+ if (ret) {
+ domain_exit(si_domain);
+ return -EFAULT;
+ }
+ }
+
+ if (md_domain_init(si_domain, DEFAULT_DOMAIN_ADDRESS_WIDTH)) {
+ domain_exit(si_domain);
+ return -EFAULT;
+ }
+
+ si_domain->flags = DOMAIN_FLAG_STATIC_IDENTITY;
+
+ if (hw)
+ return 0;
for_each_online_node(nid) {
- work_with_active_regions(nid, iommu_prepare_work_fn, &data);
- if (data.ret)
- return data.ret;
+ work_with_active_regions(nid, si_domain_work_fn, &ret);
+ if (ret)
+ return ret;
}
- return data.ret;
+
+ return 0;
}
-static void __init iommu_prepare_gfx_mapping(void)
+static void domain_remove_one_dev_info(struct dmar_domain *domain,
+ struct pci_dev *pdev);
+static int identity_mapping(struct pci_dev *pdev)
{
- struct pci_dev *pdev = NULL;
+ struct device_domain_info *info;
+
+ if (likely(!iommu_identity_mapping))
+ return 0;
+
+
+ list_for_each_entry(info, &si_domain->devices, link)
+ if (info->dev == pdev)
+ return 1;
+ return 0;
+}
+
+static int domain_add_dev_info(struct dmar_domain *domain,
+ struct pci_dev *pdev,
+ int translation)
+{
+ struct device_domain_info *info;
+ unsigned long flags;
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");
+ info = alloc_devinfo_mem();
+ if (!info)
+ return -ENOMEM;
+
+ ret = domain_context_mapping(domain, pdev, translation);
+ if (ret) {
+ free_devinfo_mem(info);
+ return ret;
}
+
+ info->segment = pci_domain_nr(pdev->bus);
+ 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;
}
-#else /* !CONFIG_DMAR_GFX_WA */
-static inline void iommu_prepare_gfx_mapping(void)
+
+static int iommu_should_identity_map(struct pci_dev *pdev, int startup)
{
- return;
+ if ((iommu_identity_mapping & IDENTMAP_AZALIA) && IS_AZALIA(pdev))
+ return 1;
+
+ if ((iommu_identity_mapping & IDENTMAP_GFX) && IS_GFX_DEVICE(pdev))
+ return 1;
+
+ if (!(iommu_identity_mapping & IDENTMAP_ALL))
+ return 0;
+
+ /*
+ * 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;
}
-#endif
-#ifdef CONFIG_DMAR_FLOPPY_WA
-static inline void iommu_prepare_isa(void)
+static int __init iommu_prepare_static_identity_mapping(int hw)
{
- struct pci_dev *pdev;
+ struct pci_dev *pdev = NULL;
int ret;
- pdev = pci_get_class(PCI_CLASS_BRIDGE_ISA << 8, NULL);
- if (!pdev)
- return;
+ ret = si_domain_init(hw);
+ if (ret)
+ return -EFAULT;
- printk(KERN_INFO "IOMMU: Prepare 0-16M unity mapping for LPC\n");
- ret = iommu_prepare_identity_map(pdev, 0, 16*1024*1024);
+ for_each_pci_dev(pdev) {
+ if (iommu_should_identity_map(pdev, 1)) {
+ printk(KERN_INFO "IOMMU: %s identity mapping for device %s\n",
+ hw ? "hardware" : "software", pci_name(pdev));
- if (ret)
- printk(KERN_ERR "IOMMU: Failed to create 0-64M identity map, "
- "floppy might not work\n");
+ ret = domain_add_dev_info(si_domain, pdev,
+ hw ? CONTEXT_TT_PASS_THROUGH :
+ CONTEXT_TT_MULTI_LEVEL);
+ if (ret)
+ return ret;
+ }
+ }
+ return 0;
}
-#else
-static inline void iommu_prepare_isa(void)
-{
- return;
-}
-#endif /* !CONFIG_DMAR_FLPY_WA */
-static int __init init_dmars(void)
+int __init init_dmars(void)
{
struct dmar_drhd_unit *drhd;
struct dmar_rmrr_unit *rmrr;
deferred_flush = kzalloc(g_num_of_iommus *
sizeof(struct deferred_flush_tables), GFP_KERNEL);
if (!deferred_flush) {
- kfree(g_iommus);
ret = -ENOMEM;
goto error;
}
printk(KERN_ERR "IOMMU: allocate root entry failed\n");
goto error;
}
+ if (!ecap_pass_through(iommu->ecap))
+ hw_pass_through = 0;
}
/*
}
}
-#ifdef CONFIG_INTR_REMAP
- if (!intr_remapping_enabled) {
- ret = enable_intr_remapping(0);
- if (ret)
- printk(KERN_ERR
- "IOMMU: enable interrupt remapping failed\n");
- }
+ if (iommu_pass_through)
+ iommu_identity_mapping |= IDENTMAP_ALL;
+
+#ifdef CONFIG_DMAR_BROKEN_GFX_WA
+ iommu_identity_mapping |= IDENTMAP_GFX;
#endif
+ check_tylersburg_isoch();
+
+ /*
+ * If pass through is not set or not enabled, setup context entries for
+ * identity mappings for rmrr, gfx, and isa and may fall back to static
+ * identity mapping if iommu_identity_mapping is set.
+ */
+ if (iommu_identity_mapping) {
+ ret = iommu_prepare_static_identity_mapping(hw_pass_through);
+ if (ret) {
+ printk(KERN_CRIT "Failed to setup IOMMU pass-through\n");
+ goto error;
+ }
+ }
/*
* For each rmrr
* for each dev attached to rmrr
* endfor
* endfor
*/
+ printk(KERN_INFO "IOMMU: Setting RMRR:\n");
for_each_rmrr_units(rmrr) {
for (i = 0; i < rmrr->devices_cnt; i++) {
pdev = rmrr->devices[i];
- /* some BIOS lists non-exist devices in DMAR table */
+ /*
+ * some BIOS lists non-exist devices in DMAR
+ * table.
+ */
if (!pdev)
continue;
ret = iommu_prepare_rmrr_dev(rmrr, pdev);
if (ret)
printk(KERN_ERR
- "IOMMU: mapping reserved region failed\n");
+ "IOMMU: mapping reserved region failed\n");
}
}
- iommu_prepare_gfx_mapping();
-
iommu_prepare_isa();
/*
iommu_set_root_entry(iommu);
- 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);
+ iommu->flush.flush_context(iommu, 0, 0, 0, DMA_CCMD_GLOBAL_INVL);
+ iommu->flush.flush_iotlb(iommu, 0, 0, 0, DMA_TLB_GLOBAL_FLUSH);
ret = iommu_enable_translation(iommu);
if (ret)
goto error;
+
+ iommu_disable_protect_mem_regions(iommu);
}
return 0;
return ret;
}
-static inline u64 aligned_size(u64 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;
-}
-
-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_32BIT_MASK || 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_32BIT_MASK and if that fails then try allocating
+ * DMA_BIT_MASK(32) and if that fails then try allocating
* from higher range
*/
- iova = iommu_alloc_iova(domain, size, DMA_32BIT_MASK);
- 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;
}
return iova;
}
-static struct dmar_domain *
-get_valid_domain_for_dev(struct pci_dev *pdev)
+static struct dmar_domain *__get_valid_domain_for_dev(struct pci_dev *pdev)
{
struct dmar_domain *domain;
int ret;
/* make sure context mapping is ok */
if (unlikely(!domain_context_mapped(pdev))) {
- ret = domain_context_mapping(domain, pdev);
+ ret = domain_context_mapping(domain, pdev,
+ CONTEXT_TT_MULTI_LEVEL);
if (ret) {
printk(KERN_ERR
"Domain context map for %s failed",
return domain;
}
+static inline struct dmar_domain *get_valid_domain_for_dev(struct pci_dev *dev)
+{
+ struct device_domain_info *info;
+
+ /* No lock here, assumes no domain exit in normal case */
+ info = dev->dev.archdata.iommu;
+ if (likely(info))
+ return info->domain;
+
+ return __get_valid_domain_for_dev(dev);
+}
+
+static int iommu_dummy(struct pci_dev *pdev)
+{
+ return pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO;
+}
+
+/* Check if the pdev needs to go through non-identity map and unmap process.*/
+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 0;
+
+ found = identity_mapping(pdev);
+ if (found) {
+ if (iommu_should_identity_map(pdev, 0))
+ return 1;
+ else {
+ /*
+ * 32 bit DMA is removed from si_domain and fall back
+ * to non-identity mapping.
+ */
+ domain_remove_one_dev_info(si_domain, pdev);
+ printk(KERN_INFO "32bit %s uses non-identity mapping\n",
+ pci_name(pdev));
+ return 0;
+ }
+ } else {
+ /*
+ * In case of a detached 64 bit DMA device from vm, the device
+ * is put into si_domain for identity mapping.
+ */
+ if (iommu_should_identity_map(pdev, 0)) {
+ int ret;
+ ret = domain_add_dev_info(si_domain, pdev,
+ hw_pass_through ?
+ CONTEXT_TT_PASS_THROUGH :
+ CONTEXT_TT_MULTI_LEVEL);
+ if (!ret) {
+ printk(KERN_INFO "64bit %s uses identity mapping\n",
+ pci_name(pdev));
+ return 1;
+ }
+ }
+ }
+
+ return 0;
+}
+
static dma_addr_t __intel_map_single(struct device *hwdev, phys_addr_t paddr,
size_t size, int dir, u64 dma_mask)
{
int prot = 0;
int ret;
struct intel_iommu *iommu;
+ unsigned long paddr_pfn = paddr >> PAGE_SHIFT;
BUG_ON(dir == DMA_NONE);
- if (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO)
+
+ if (iommu_no_mapping(hwdev))
return paddr;
domain = get_valid_domain_for_dev(pdev);
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..
* 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) & PAGE_MASK, size, prot);
+ ret = domain_pfn_mapping(domain, mm_to_dma_pfn(iova->pfn_lo),
+ mm_to_dma_pfn(paddr_pfn), size, prot);
if (ret)
goto error;
- /* it's a non-present to present mapping */
- ret = iommu_flush_iotlb_psi(iommu, domain->id,
- start_paddr, size >> VTD_PAGE_SHIFT, 1);
- if (ret)
+ /* 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);
+ 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)
if (!iommu)
continue;
- 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]);
- }
- deferred_flush[i].next = 0;
+ if (!deferred_flush[i].next)
+ continue;
+
+ 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);
+ __free_iova(&deferred_flush[i].domain[j]->iovad, iova);
}
+ deferred_flush[i].next = 0;
}
list_size = 0;
{
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 (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO)
+ if (iommu_no_mapping(dev))
return;
+
domain = find_domain(pdev);
BUG_ON(!domain);
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) {
- if (iommu_flush_iotlb_psi(iommu,
- domain->id, start_addr, size >> VTD_PAGE_SHIFT, 0))
- iommu_flush_write_buffer(iommu);
+ iommu_flush_iotlb_psi(iommu, domain->id, start_pfn,
+ last_pfn - start_pfn + 1);
/* free iova */
__free_iova(&domain->iovad, iova);
} else {
}
}
-static void intel_unmap_single(struct device *dev, dma_addr_t dev_addr, size_t size,
- int dir)
-{
- intel_unmap_page(dev, dev_addr, size, dir, NULL);
-}
-
static void *intel_alloc_coherent(struct device *hwdev, size_t size,
dma_addr_t *dma_handle, gfp_t flags)
{
size = PAGE_ALIGN(size);
order = get_order(size);
- intel_unmap_single(hwdev, dma_handle, size, DMA_BIDIRECTIONAL);
+ intel_unmap_page(hwdev, dma_handle, size, DMA_BIDIRECTIONAL, NULL);
free_pages((unsigned long)vaddr, order);
}
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 (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO)
+ if (iommu_no_mapping(hwdev))
return;
domain = find_domain(pdev);
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);
- /* free page tables */
- dma_pte_free_pagetable(domain, start_addr, start_addr + size);
+ dma_pte_clear_range(domain, start_pfn, last_pfn);
- if (iommu_flush_iotlb_psi(iommu, domain->id, start_addr,
- size >> VTD_PAGE_SHIFT, 0))
- iommu_flush_write_buffer(iommu);
+ /* free page tables */
+ dma_pte_free_pagetable(domain, start_pfn, last_pfn);
- /* free iova */
- __free_iova(&domain->iovad, iova);
+ if (intel_iommu_strict) {
+ iommu_flush_iotlb_psi(iommu, domain->id, start_pfn,
+ last_pfn - start_pfn + 1);
+ /* free iova */
+ __free_iova(&domain->iovad, iova);
+ } else {
+ add_unmap(domain, iova);
+ /*
+ * queue up the release of the unmap to save the 1/6th of the
+ * cpu used up by the iotlb flush operation...
+ */
+ }
}
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 (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO)
+ if (iommu_no_mapping(hwdev))
return intel_nontranslate_map_sg(hwdev, sglist, nelems, dir);
domain = get_valid_domain_for_dev(pdev);
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;
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) & 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, 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 */
- if (iommu_flush_iotlb_psi(iommu, domain->id,
- start_addr, offset >> VTD_PAGE_SHIFT, 1))
+ /* 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);
+ else
iommu_flush_write_buffer(iommu);
+
return nelems;
}
iommu_set_root_entry(iommu);
iommu->flush.flush_context(iommu, 0, 0, 0,
- DMA_CCMD_GLOBAL_INVL, 0);
+ DMA_CCMD_GLOBAL_INVL);
iommu->flush.flush_iotlb(iommu, 0, 0, 0,
- DMA_TLB_GLOBAL_FLUSH, 0);
- iommu_disable_protect_mem_regions(iommu);
+ DMA_TLB_GLOBAL_FLUSH);
iommu_enable_translation(iommu);
+ iommu_disable_protect_mem_regions(iommu);
}
return 0;
for_each_active_iommu(iommu, drhd) {
iommu->flush.flush_context(iommu, 0, 0, 0,
- DMA_CCMD_GLOBAL_INVL, 0);
+ DMA_CCMD_GLOBAL_INVL);
iommu->flush.flush_iotlb(iommu, 0, 0, 0,
- DMA_TLB_GLOBAL_FLUSH, 0);
+ DMA_TLB_GLOBAL_FLUSH);
}
}
int __init intel_iommu_init(void)
{
int ret = 0;
+ int force_on = 0;
- if (dmar_table_init())
+ /* VT-d is required for a TXT/tboot launch, so enforce that */
+ force_on = tboot_force_iommu();
+
+ if (dmar_table_init()) {
+ if (force_on)
+ panic("tboot: Failed to initialize DMAR table\n");
return -ENODEV;
+ }
- if (dmar_dev_scope_init())
+ if (dmar_dev_scope_init()) {
+ if (force_on)
+ panic("tboot: Failed to initialize DMAR device scope\n");
return -ENODEV;
+ }
/*
* Check the need for DMA-remapping initialization now.
ret = init_dmars();
if (ret) {
+ if (force_on)
+ panic("tboot: Failed to initialize DMARs\n");
printk(KERN_ERR "IOMMU: dmar init failed\n");
put_iova_domain(&reserved_iova_list);
iommu_exit_mempool();
init_timer(&unmap_timer);
force_iommu = 1;
dma_ops = &intel_dma_ops;
+
init_iommu_sysfs();
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->segment = pci_domain_nr(pdev->bus);
- 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 iommu_detach_dependent_devices(struct intel_iommu *iommu,
struct pci_dev *pdev)
{
}
}
-static void vm_domain_remove_one_dev_info(struct dmar_domain *domain,
+static void domain_remove_one_dev_info(struct dmar_domain *domain,
struct pci_dev *pdev)
{
struct device_domain_info *info;
info->dev->dev.archdata.iommu = NULL;
spin_unlock_irqrestore(&device_domain_lock, flags);
+ iommu_disable_dev_iotlb(info);
iommu_detach_dev(iommu, info->bus, info->devfn);
iommu_detach_dependent_devices(iommu, pdev);
free_devinfo_mem(info);
spin_unlock_irqrestore(&device_domain_lock, flags1);
+ iommu_disable_dev_iotlb(info);
iommu = device_to_iommu(info->segment, info->bus, info->devfn);
iommu_detach_dev(iommu, info->bus, info->devfn);
iommu_detach_dependent_devices(iommu, info->dev);
return domain;
}
-static int vm_domain_init(struct dmar_domain *domain, int guest_width)
+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);
domain->iommu_count = 0;
domain->iommu_coherency = 0;
+ domain->iommu_snooping = 0;
domain->max_addr = 0;
/* always allocate the top pgd */
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);
+ 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);
"intel_iommu_domain_init: dmar_domain == NULL\n");
return -ENOMEM;
}
- if (vm_domain_init(dmar_domain, DEFAULT_DOMAIN_ADDRESS_WIDTH)) {
+ if (md_domain_init(dmar_domain, DEFAULT_DOMAIN_ADDRESS_WIDTH)) {
printk(KERN_ERR
"intel_iommu_domain_init() failed\n");
vm_domain_exit(dmar_domain);
struct intel_iommu *iommu;
int addr_width;
u64 end;
- int ret;
/* normally pdev is not mapped */
if (unlikely(domain_context_mapped(pdev))) {
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);
+ if (dmar_domain->flags & DOMAIN_FLAG_VIRTUAL_MACHINE ||
+ dmar_domain->flags & DOMAIN_FLAG_STATIC_IDENTITY)
+ domain_remove_one_dev_info(old_domain, pdev);
else
domain_remove_dev_info(old_domain);
}
return -EFAULT;
}
- ret = domain_context_mapping(dmar_domain, pdev);
- if (ret)
- return ret;
-
- ret = vm_domain_add_dev_info(dmar_domain, pdev);
- return ret;
+ return domain_add_dev_info(dmar_domain, pdev, CONTEXT_TT_MULTI_LEVEL);
}
static void intel_iommu_detach_device(struct iommu_domain *domain,
struct dmar_domain *dmar_domain = domain->priv;
struct pci_dev *pdev = to_pci_dev(dev);
- vm_domain_remove_one_dev_info(dmar_domain, pdev);
+ domain_remove_one_dev_info(dmar_domain, pdev);
}
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;
}
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;
}
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,
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);
}
DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x2a40, quirk_iommu_rwbf);
+
+/* On Tylersburg chipsets, some BIOSes have been known to enable the
+ ISOCH DMAR unit for the Azalia sound device, but not give it any
+ TLB entries, which causes it to deadlock. Check for that. We do
+ this in a function called from init_dmars(), instead of in a PCI
+ quirk, because we don't want to print the obnoxious "BIOS broken"
+ message if VT-d is actually disabled.
+*/
+static void __init check_tylersburg_isoch(void)
+{
+ struct pci_dev *pdev;
+ uint32_t vtisochctrl;
+
+ /* If there's no Azalia in the system anyway, forget it. */
+ pdev = pci_get_device(PCI_VENDOR_ID_INTEL, 0x3a3e, NULL);
+ if (!pdev)
+ return;
+ pci_dev_put(pdev);
+
+ /* System Management Registers. Might be hidden, in which case
+ we can't do the sanity check. But that's OK, because the
+ known-broken BIOSes _don't_ actually hide it, so far. */
+ pdev = pci_get_device(PCI_VENDOR_ID_INTEL, 0x342e, NULL);
+ if (!pdev)
+ return;
+
+ if (pci_read_config_dword(pdev, 0x188, &vtisochctrl)) {
+ pci_dev_put(pdev);
+ return;
+ }
+
+ pci_dev_put(pdev);
+
+ /* If Azalia DMA is routed to the non-isoch DMAR unit, fine. */
+ if (vtisochctrl & 1)
+ return;
+
+ /* Drop all bits other than the number of TLB entries */
+ vtisochctrl &= 0x1c;
+
+ /* If we have the recommended number of TLB entries (16), fine. */
+ if (vtisochctrl == 0x10)
+ return;
+
+ /* Zero TLB entries? You get to ride the short bus to school. */
+ if (!vtisochctrl) {
+ WARN(1, "Your BIOS is broken; DMA routed to ISOCH DMAR unit but no TLB space.\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));
+ iommu_identity_mapping |= IDENTMAP_AZALIA;
+ return;
+ }
+
+ printk(KERN_WARNING "DMAR: Recommended TLB entries for ISOCH unit is 16; your BIOS set %d\n",
+ vtisochctrl);
+}