* 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/bitmap.h>
+#include <linux/debugfs.h>
#include <linux/slab.h>
#include <linux/irq.h>
#include <linux/interrupt.h>
#include <linux/dmar.h>
#include <linux/dma-mapping.h>
#include <linux/mempool.h>
-#include "iova.h"
-#include "intel-iommu.h"
-#include <asm/proto.h> /* force_iommu in this header in x86-64*/
+#include <linux/timer.h>
+#include <linux/iova.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 DEFAULT_DOMAIN_ADDRESS_WIDTH 48
-#define DMAR_OPERATION_TIMEOUT (HZ*60) /* 1m */
-
#define DOMAIN_MAX_ADDR(gaw) ((((u64)1) << gaw) - 1)
+
+static void flush_unmaps_timeout(unsigned long data);
+
+DEFINE_TIMER(unmap_timer, flush_unmaps_timeout, 0, 0);
+
+#define HIGH_WATER_MARK 250
+struct deferred_flush_tables {
+ int next;
+ struct iova *iova[HIGH_WATER_MARK];
+ struct dmar_domain *domain[HIGH_WATER_MARK];
+};
+
+static struct deferred_flush_tables *deferred_flush;
+
+/* bitmap for indexing intel_iommus */
+static int g_num_of_iommus;
+
+static DEFINE_SPINLOCK(async_umap_flush_lock);
+static LIST_HEAD(unmaps_to_do);
+
+static int timer_on;
+static long list_size;
+
static void domain_remove_dev_info(struct dmar_domain *domain);
-static int dmar_disabled;
+int dmar_disabled;
static int __initdata dmar_map_gfx = 1;
static int dmar_forcedac;
+static int intel_iommu_strict;
#define DUMMY_DEVICE_DOMAIN_INFO ((struct device_domain_info *)(-1))
static DEFINE_SPINLOCK(device_domain_lock);
printk(KERN_INFO
"Intel-IOMMU: disable GFX device mapping\n");
} else if (!strncmp(str, "forcedac", 8)) {
- printk (KERN_INFO
+ printk(KERN_INFO
"Intel-IOMMU: Forcing DAC for PCI devices\n");
dmar_forcedac = 1;
+ } else if (!strncmp(str, "strict", 6)) {
+ printk(KERN_INFO
+ "Intel-IOMMU: disable batched IOTLB flush\n");
+ intel_iommu_strict = 1;
}
str += strcspn(str, ",");
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 void __iommu_flush_cache(
- struct intel_iommu *iommu, void *addr, int size)
-{
- if (!ecap_coherent(iommu->ecap))
- 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);
return NULL;
}
__iommu_flush_cache(domain->iommu, tmp_page,
- PAGE_SIZE_4K);
+ PAGE_SIZE);
dma_set_pte_addr(*pte, virt_to_phys(tmp_page));
/*
* high level table always sets r/w, last level page
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;
}
}
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;
return 0;
}
-#define IOMMU_WAIT_OP(iommu, offset, op, cond, sts) \
-{\
- unsigned long start_time = jiffies;\
- while (1) {\
- sts = op (iommu->reg + offset);\
- if (cond)\
- break;\
- if (time_after(jiffies, start_time + DMAR_OPERATION_TIMEOUT))\
- panic("DMAR hardware is malfunctioning\n");\
- cpu_relax();\
- }\
-}
-
static void iommu_set_root_entry(struct intel_iommu *iommu)
{
void *addr;
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;
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.
return 0;
}
-static struct intel_iommu *alloc_iommu(struct dmar_drhd_unit *drhd)
-{
- struct intel_iommu *iommu;
- int ret;
- int map_size;
- u32 ver;
-
- iommu = kzalloc(sizeof(*iommu), GFP_KERNEL);
- if (!iommu)
- return NULL;
- iommu->reg = ioremap(drhd->reg_base_addr, PAGE_SIZE_4K);
- if (!iommu->reg) {
- printk(KERN_ERR "IOMMU: can't map the region\n");
- goto error;
- }
- iommu->cap = dmar_readq(iommu->reg + DMAR_CAP_REG);
- iommu->ecap = dmar_readq(iommu->reg + DMAR_ECAP_REG);
-
- /* the registers might be more than one page */
- map_size = max_t(int, ecap_max_iotlb_offset(iommu->ecap),
- cap_max_fault_reg_offset(iommu->cap));
- map_size = PAGE_ALIGN_4K(map_size);
- if (map_size > PAGE_SIZE_4K) {
- iounmap(iommu->reg);
- iommu->reg = ioremap(drhd->reg_base_addr, map_size);
- if (!iommu->reg) {
- printk(KERN_ERR "IOMMU: can't map the region\n");
- goto error;
- }
- }
-
- ver = readl(iommu->reg + DMAR_VER_REG);
- pr_debug("IOMMU %llx: ver %d:%d cap %llx ecap %llx\n",
- drhd->reg_base_addr, DMAR_VER_MAJOR(ver), DMAR_VER_MINOR(ver),
- iommu->cap, iommu->ecap);
- ret = iommu_init_domains(iommu);
- if (ret)
- goto error_unmap;
- spin_lock_init(&iommu->lock);
- spin_lock_init(&iommu->register_lock);
-
- drhd->iommu = iommu;
- return iommu;
-error_unmap:
- iounmap(iommu->reg);
-error:
- kfree(iommu);
- return NULL;
-}
static void domain_exit(struct dmar_domain *domain);
-static void free_iommu(struct intel_iommu *iommu)
+
+void free_dmar_iommu(struct intel_iommu *iommu)
{
struct dmar_domain *domain;
int i;
- if (!iommu)
- return;
-
i = find_first_bit(iommu->domain_ids, cap_ndoms(iommu->cap));
for (; i < cap_ndoms(iommu->cap); ) {
domain = iommu->domains[i];
/* free context mapping */
free_context_table(iommu);
-
- if (iommu->reg)
- iounmap(iommu->reg);
- kfree(iommu);
}
static struct dmar_domain * iommu_alloc_domain(struct intel_iommu *iommu)
}
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)
{
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);
+
/* IOAPIC ranges shouldn't be accessed by DMA */
iova = reserve_iova(&reserved_iova_list, IOVA_PFN(IOAPIC_RANGE_START),
IOVA_PFN(IOAPIC_RANGE_END));
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)
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);
__iommu_flush_cache(iommu, 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);
return 0;
}
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_addr(*pte, start_pfn << VTD_PAGE_SHIFT);
dma_set_pte_prot(*pte, prot);
__iommu_flush_cache(domain->iommu, pte, sizeof(*pte));
start_pfn++;
static void detach_domain_for_dev(struct dmar_domain *domain, 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);
+ domain->iommu->flush.flush_context(domain->iommu, 0, 0, 0,
+ DMA_CCMD_GLOBAL_INVL, 0);
+ domain->iommu->flush.flush_iotlb(domain->iommu, 0, 0, 0,
+ DMA_TLB_GLOBAL_FLUSH, 0);
}
static void domain_remove_dev_info(struct dmar_domain *domain)
* 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;
return NULL;
}
-static int dmar_pci_device_match(struct pci_dev *devices[], int cnt,
- struct pci_dev *dev)
-{
- int index;
-
- while (dev) {
- for (index = 0; index < cnt; index ++)
- if (dev == devices[index])
- return 1;
-
- /* Check our parent */
- dev = dev->bus->self;
- }
-
- return 0;
-}
-
-static struct dmar_drhd_unit *
-dmar_find_matched_drhd_unit(struct pci_dev *dev)
-{
- struct dmar_drhd_unit *drhd = NULL;
-
- list_for_each_entry(drhd, &dmar_drhd_units, list) {
- if (drhd->include_all || dmar_pci_device_match(drhd->devices,
- drhd->devices_cnt, dev))
- return drhd;
- }
-
- return NULL;
-}
-
/* domain is initialized */
static struct dmar_domain *get_domain_for_dev(struct pci_dev *pdev, int gaw)
{
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");
}
#ifdef CONFIG_DMAR_GFX_WA
-extern int arch_get_ram_range(int slot, u64 *addr, u64 *size);
+struct iommu_prepare_data {
+ struct pci_dev *pdev;
+ int ret;
+};
+
+static int __init iommu_prepare_work_fn(unsigned long start_pfn,
+ unsigned long end_pfn, void *datax)
+{
+ struct iommu_prepare_data *data;
+
+ data = (struct iommu_prepare_data *)datax;
+
+ data->ret = iommu_prepare_identity_map(data->pdev,
+ start_pfn<<PAGE_SHIFT, end_pfn<<PAGE_SHIFT);
+ return data->ret;
+
+}
+
+static int __init iommu_prepare_with_active_regions(struct pci_dev *pdev)
+{
+ int nid;
+ struct iommu_prepare_data data;
+
+ data.pdev = pdev;
+ data.ret = 0;
+
+ for_each_online_node(nid) {
+ work_with_active_regions(nid, iommu_prepare_work_fn, &data);
+ if (data.ret)
+ return data.ret;
+ }
+ return data.ret;
+}
+
static void __init iommu_prepare_gfx_mapping(void)
{
struct pci_dev *pdev = NULL;
- u64 base, size;
- int slot;
int ret;
for_each_pci_dev(pdev) {
continue;
printk(KERN_INFO "IOMMU: gfx device %s 1-1 mapping\n",
pci_name(pdev));
- slot = arch_get_ram_range(0, &base, &size);
- while (slot >= 0) {
- ret = iommu_prepare_identity_map(pdev,
- base, base + size);
- if (ret)
- goto error;
- slot = arch_get_ram_range(slot, &base, &size);
- }
- continue;
-error:
- printk(KERN_ERR "IOMMU: mapping reserved region failed\n");
+ ret = iommu_prepare_with_active_regions(pdev);
+ if (ret)
+ printk(KERN_ERR "IOMMU: mapping reserved region failed\n");
}
}
#endif
struct dmar_rmrr_unit *rmrr;
struct pci_dev *pdev;
struct intel_iommu *iommu;
- int ret, unit = 0;
+ int i, ret, unit = 0;
/*
* for each drhd
* endfor
*/
for_each_drhd_unit(drhd) {
+ g_num_of_iommus++;
+ /*
+ * lock not needed as this is only incremented in the single
+ * threaded kernel __init code path all other access are read
+ * only
+ */
+ }
+
+ deferred_flush = kzalloc(g_num_of_iommus *
+ sizeof(struct deferred_flush_tables), GFP_KERNEL);
+ if (!deferred_flush) {
+ ret = -ENOMEM;
+ goto error;
+ }
+
+ for_each_drhd_unit(drhd) {
if (drhd->ignored)
continue;
- iommu = alloc_iommu(drhd);
- if (!iommu) {
- ret = -ENOMEM;
+
+ iommu = drhd->iommu;
+
+ ret = iommu_init_domains(iommu);
+ if (ret)
goto error;
- }
/*
* TBD:
}
}
+ 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
* endfor
*/
for_each_rmrr_units(rmrr) {
- int i;
for (i = 0; i < rmrr->devices_cnt; i++) {
pdev = rmrr->devices[i];
/* some BIOS lists non-exist devices in DMAR table */
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);
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) {
return domain;
}
-static dma_addr_t intel_map_single(struct device *hwdev, void *addr,
- 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);
- int ret;
struct dmar_domain *domain;
- unsigned long start_addr;
+ phys_addr_t start_paddr;
struct iova *iova;
int prot = 0;
+ int ret;
BUG_ON(dir == DMA_NONE);
if (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO)
- return virt_to_bus(addr);
+ return paddr;
domain = get_valid_domain_for_dev(pdev);
if (!domain)
return 0;
- addr = (void *)virt_to_phys(addr);
- size = aligned_size((u64)addr, size);
+ 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_addr = 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
if (dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL)
prot |= DMA_PTE_WRITE;
/*
- * addr - (addr + size) might be partial page, we should map the whole
+ * paddr - (paddr + size) might be partial page, we should map the whole
* page. Note: if two part of one page are separately mapped, we
- * might have two guest_addr mapping to the same host addr, but this
+ * might have two guest_addr mapping to the same host paddr, but this
* is not a big problem
*/
- ret = domain_page_mapping(domain, start_addr,
- ((u64)addr) & PAGE_MASK_4K, size, prot);
+ ret = domain_page_mapping(domain, start_paddr,
+ ((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)addr,
- size, (u64)start_addr, dir);
-
/* it's a non-present to present mapping */
ret = iommu_flush_iotlb_psi(domain->iommu, domain->id,
- start_addr, size >> PAGE_SHIFT_4K, 1);
+ start_paddr, size >> VTD_PAGE_SHIFT, 1);
if (ret)
iommu_flush_write_buffer(domain->iommu);
- return (start_addr + ((u64)addr & (~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)addr, dir);
+ pci_name(pdev), size, (unsigned long long)paddr, dir);
return 0;
}
-static void intel_unmap_single(struct device *dev, dma_addr_t dev_addr,
- size_t size, int dir)
+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;
+
+ timer_on = 0;
+
+ /* 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;
+
+ 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;
+ }
+ }
+
+ list_size = 0;
+}
+
+static void flush_unmaps_timeout(unsigned long data)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&async_umap_flush_lock, flags);
+ flush_unmaps();
+ spin_unlock_irqrestore(&async_umap_flush_lock, flags);
+}
+
+static void add_unmap(struct dmar_domain *dom, struct iova *iova)
+{
+ unsigned long flags;
+ int next, iommu_id;
+
+ spin_lock_irqsave(&async_umap_flush_lock, flags);
+ if (list_size == HIGH_WATER_MARK)
+ flush_unmaps();
+
+ iommu_id = dom->iommu->seq_id;
+
+ next = deferred_flush[iommu_id].next;
+ deferred_flush[iommu_id].domain[next] = dom;
+ deferred_flush[iommu_id].iova[next] = iova;
+ deferred_flush[iommu_id].next++;
+
+ if (!timer_on) {
+ mod_timer(&unmap_timer, jiffies + msecs_to_jiffies(10));
+ timer_on = 1;
+ }
+ list_size++;
+ spin_unlock_irqrestore(&async_umap_flush_lock, flags);
+}
+
+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;
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 (iommu_flush_iotlb_psi(domain->iommu, domain->id, start_addr,
- size >> PAGE_SHIFT_4K, 0))
- iommu_flush_write_buffer(domain->iommu);
-
- /* free iova */
- __free_iova(&domain->iovad, iova);
+ if (intel_iommu_strict) {
+ if (iommu_flush_iotlb_psi(domain->iommu,
+ domain->id, start_addr, size >> VTD_PAGE_SHIFT, 0))
+ iommu_flush_write_buffer(domain->iommu);
+ /* 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 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, 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 += 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);
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))
+ size >> VTD_PAGE_SHIFT, 0))
iommu_flush_write_buffer(domain->iommu);
/* free 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;
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;
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))
+ start_addr, offset >> VTD_PAGE_SHIFT, 1))
iommu_flush_write_buffer(domain->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");
}
-void __init detect_intel_iommu(void)
-{
- if (swiotlb || no_iommu || iommu_detected || dmar_disabled)
- return;
- if (early_dmar_detect()) {
- iommu_detected = 1;
- }
-}
-
static void __init init_no_remapping_devices(void)
{
struct dmar_drhd_unit *drhd;
{
int ret = 0;
- if (no_iommu || swiotlb || dmar_disabled)
- return -ENODEV;
-
if (dmar_table_init())
return -ENODEV;
+ if (dmar_dev_scope_init())
+ return -ENODEV;
+
+ /*
+ * Check the need for DMA-remapping initialization now.
+ * Above initialization will also be used by Interrupt-remapping.
+ */
+ if (no_iommu || swiotlb || dmar_disabled)
+ return -ENODEV;
+
iommu_init_mempool();
dmar_init_reserved_ranges();
printk(KERN_INFO
"PCI-DMA: Intel(R) Virtualization Technology for Directed I/O\n");
+ init_timer(&unmap_timer);
force_iommu = 1;
dma_ops = &intel_dma_ops;
return 0;
}
+void intel_iommu_domain_exit(struct dmar_domain *domain)
+{
+ u64 end;
+
+ /* Domain 0 is reserved, so dont process it */
+ if (!domain)
+ return;
+
+ 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_domain(domain);
+ free_domain_mem(domain);
+}
+EXPORT_SYMBOL_GPL(intel_iommu_domain_exit);
+
+struct dmar_domain *intel_iommu_domain_alloc(struct pci_dev *pdev)
+{
+ struct dmar_drhd_unit *drhd;
+ struct dmar_domain *domain;
+ struct intel_iommu *iommu;
+
+ drhd = dmar_find_matched_drhd_unit(pdev);
+ if (!drhd) {
+ printk(KERN_ERR "intel_iommu_domain_alloc: drhd == NULL\n");
+ return NULL;
+ }
+
+ iommu = drhd->iommu;
+ if (!iommu) {
+ printk(KERN_ERR
+ "intel_iommu_domain_alloc: iommu == NULL\n");
+ return NULL;
+ }
+ domain = iommu_alloc_domain(iommu);
+ if (!domain) {
+ printk(KERN_ERR
+ "intel_iommu_domain_alloc: domain == NULL\n");
+ return NULL;
+ }
+ if (domain_init(domain, DEFAULT_DOMAIN_ADDRESS_WIDTH)) {
+ printk(KERN_ERR
+ "intel_iommu_domain_alloc: domain_init() failed\n");
+ intel_iommu_domain_exit(domain);
+ return NULL;
+ }
+ return domain;
+}
+EXPORT_SYMBOL_GPL(intel_iommu_domain_alloc);
+
+int intel_iommu_context_mapping(
+ struct dmar_domain *domain, struct pci_dev *pdev)
+{
+ int rc;
+ rc = domain_context_mapping(domain, pdev);
+ return rc;
+}
+EXPORT_SYMBOL_GPL(intel_iommu_context_mapping);
+
+int intel_iommu_page_mapping(
+ struct dmar_domain *domain, dma_addr_t iova,
+ u64 hpa, size_t size, int prot)
+{
+ int rc;
+ rc = domain_page_mapping(domain, iova, hpa, size, prot);
+ return rc;
+}
+EXPORT_SYMBOL_GPL(intel_iommu_page_mapping);
+
+void intel_iommu_detach_dev(struct dmar_domain *domain, u8 bus, u8 devfn)
+{
+ detach_domain_for_dev(domain, bus, devfn);
+}
+EXPORT_SYMBOL_GPL(intel_iommu_detach_dev);
+
+struct dmar_domain *
+intel_iommu_find_domain(struct pci_dev *pdev)
+{
+ return find_domain(pdev);
+}
+EXPORT_SYMBOL_GPL(intel_iommu_find_domain);
+
+int intel_iommu_found(void)
+{
+ return g_num_of_iommus;
+}
+EXPORT_SYMBOL_GPL(intel_iommu_found);
+
+u64 intel_iommu_iova_to_pfn(struct dmar_domain *domain, u64 iova)
+{
+ struct dma_pte *pte;
+ u64 pfn;
+
+ pfn = 0;
+ pte = addr_to_dma_pte(domain, iova);
+
+ if (pte)
+ pfn = dma_pte_addr(*pte);
+
+ return pfn >> VTD_PAGE_SHIFT;
+}
+EXPORT_SYMBOL_GPL(intel_iommu_iova_to_pfn);
git://ftp.safe.ca / safe / jmp / linux-2.6 / blobdiff