Merge commit 'dwmw2-iommu/master'

Conflicts:
	drivers/pci/intel-iommu.c
	drivers/pci/intr_remapping.c

12 files changed, 244 insertions, 177 deletions

diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index f4edb42f0a1a..e61c4e9a6bf8 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -984,6 +984,7 @@ and is between 256 and 4096 characters. It is defined in the file
 		nomerge
 		forcesac
 		soft
+		pt	[x86, IA64]
 
 	io7=		[HW] IO7 for Marvel based alpha systems
 			See comment before marvel_specify_io7 in
diff --git a/arch/ia64/include/asm/iommu.h b/arch/ia64/include/asm/iommu.h
index 0490794fe4aa..37d41ca5645a 100644
--- a/arch/ia64/include/asm/iommu.h
+++ b/arch/ia64/include/asm/iommu.h
@@ -9,6 +9,7 @@ extern void pci_iommu_shutdown(void);
 extern void no_iommu_init(void);
 extern int force_iommu, no_iommu;
 extern int iommu_detected;
+extern int iommu_pass_through;
 extern void iommu_dma_init(void);
 extern void machvec_init(const char *name);
 
diff --git a/arch/ia64/kernel/pci-dma.c b/arch/ia64/kernel/pci-dma.c
index 1376da45fd08..05695962fe44 100644
--- a/arch/ia64/kernel/pci-dma.c
+++ b/arch/ia64/kernel/pci-dma.c
@@ -32,6 +32,8 @@ int force_iommu __read_mostly = 1;
 int force_iommu __read_mostly;
 #endif
 
+int iommu_pass_through;
+
 /* Dummy device used for NULL arguments (normally ISA). Better would
    be probably a smaller DMA mask, but this is bug-to-bug compatible
    to i386. */
diff --git a/arch/ia64/kernel/pci-swiotlb.c b/arch/ia64/kernel/pci-swiotlb.c
index 285aae8431c6..223abb134105 100644
--- a/arch/ia64/kernel/pci-swiotlb.c
+++ b/arch/ia64/kernel/pci-swiotlb.c
@@ -46,7 +46,7 @@ void __init swiotlb_dma_init(void)
 
 void __init pci_swiotlb_init(void)
 {
-	if (!iommu_detected) {
+	if (!iommu_detected || iommu_pass_through) {
 #ifdef CONFIG_IA64_GENERIC
 		swiotlb = 1;
 		printk(KERN_INFO "PCI-DMA: Re-initialize machine vector.\n");
diff --git a/arch/x86/include/asm/iommu.h b/arch/x86/include/asm/iommu.h
index af326a2975b5..fd6d21bbee6c 100644
--- a/arch/x86/include/asm/iommu.h
+++ b/arch/x86/include/asm/iommu.h
@@ -6,6 +6,7 @@ extern void no_iommu_init(void);
 extern struct dma_map_ops nommu_dma_ops;
 extern int force_iommu, no_iommu;
 extern int iommu_detected;
+extern int iommu_pass_through;
 
 /* 10 seconds */
 #define DMAR_OPERATION_TIMEOUT ((cycles_t) tsc_khz*10*1000)
diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c
index 745579bc8256..049005e82178 100644
--- a/arch/x86/kernel/pci-dma.c
+++ b/arch/x86/kernel/pci-dma.c
@@ -32,6 +32,8 @@ int no_iommu __read_mostly;
 /* Set this to 1 if there is a HW IOMMU in the system */
 int iommu_detected __read_mostly = 0;
 
+int iommu_pass_through;
+
 dma_addr_t bad_dma_address __read_mostly = 0;
 EXPORT_SYMBOL(bad_dma_address);
 
@@ -209,6 +211,10 @@ static __init int iommu_setup(char *p)
 #ifdef CONFIG_SWIOTLB
 		if (!strncmp(p, "soft", 4))
 			swiotlb = 1;
+		if (!strncmp(p, "pt", 2)) {
+			iommu_pass_through = 1;
+			return 1;
+		}
 #endif
 
 		gart_parse_options(p);
diff --git a/arch/x86/kernel/pci-swiotlb.c b/arch/x86/kernel/pci-swiotlb.c
index a1712f2b50f1..6af96ee44200 100644
--- a/arch/x86/kernel/pci-swiotlb.c
+++ b/arch/x86/kernel/pci-swiotlb.c
@@ -71,7 +71,8 @@ void __init pci_swiotlb_init(void)
 {
 	/* don't initialize swiotlb if iommu=off (no_iommu=1) */
 #ifdef CONFIG_X86_64
-	if (!iommu_detected && !no_iommu && max_pfn > MAX_DMA32_PFN)
+	if ((!iommu_detected && !no_iommu && max_pfn > MAX_DMA32_PFN) ||
+		iommu_pass_through)
 	       swiotlb = 1;
 #endif
 	if (swiotlb_force)
diff --git a/drivers/pci/dmar.c b/drivers/pci/dmar.c
index fa3a11365ec3..f23460a5d106 100644
--- a/drivers/pci/dmar.c
+++ b/drivers/pci/dmar.c
@@ -515,6 +515,7 @@ int alloc_iommu(struct dmar_drhd_unit *drhd)
 	u32 ver;
 	static int iommu_allocated = 0;
 	int agaw = 0;
+	int msagaw = 0;
 
 	iommu = kzalloc(sizeof(*iommu), GFP_KERNEL);
 	if (!iommu)
@@ -535,12 +536,20 @@ int alloc_iommu(struct dmar_drhd_unit *drhd)
 	agaw = iommu_calculate_agaw(iommu);
 	if (agaw < 0) {
 		printk(KERN_ERR
-			"Cannot get a valid agaw for iommu (seq_id = %d)\n",
+		       "Cannot get a valid agaw for iommu (seq_id = %d)\n",
+		       iommu->seq_id);
+		goto error;
+	}
+	msagaw = iommu_calculate_max_sagaw(iommu);
+	if (msagaw < 0) {
+		printk(KERN_ERR
+			"Cannot get a valid max agaw for iommu (seq_id = %d)\n",
 			iommu->seq_id);
 		goto error;
 	}
 #endif
 	iommu->agaw = agaw;
+	iommu->msagaw = msagaw;
 
 	/* the registers might be more than one page */
 	map_size = max_t(int, ecap_max_iotlb_offset(iommu->ecap),
@@ -714,41 +723,26 @@ void qi_global_iec(struct intel_iommu *iommu)
 	qi_submit_sync(&desc, iommu);
 }
 
-int qi_flush_context(struct intel_iommu *iommu, u16 did, u16 sid, u8 fm,
-		     u64 type, int non_present_entry_flush)
+void qi_flush_context(struct intel_iommu *iommu, u16 did, u16 sid, u8 fm,
+		      u64 type)
 {
 	struct qi_desc desc;
 
-	if (non_present_entry_flush) {
-		if (!cap_caching_mode(iommu->cap))
-			return 1;
-		else
-			did = 0;
-	}
-
 	desc.low = QI_CC_FM(fm) | QI_CC_SID(sid) | QI_CC_DID(did)
 			| QI_CC_GRAN(type) | QI_CC_TYPE;
 	desc.high = 0;
 
-	return qi_submit_sync(&desc, iommu);
+	qi_submit_sync(&desc, iommu);
 }
 
-int qi_flush_iotlb(struct intel_iommu *iommu, u16 did, u64 addr,
-		   unsigned int size_order, u64 type,
-		   int non_present_entry_flush)
+void qi_flush_iotlb(struct intel_iommu *iommu, u16 did, u64 addr,
+		    unsigned int size_order, u64 type)
 {
 	u8 dw = 0, dr = 0;
 
 	struct qi_desc desc;
 	int ih = 0;
 
-	if (non_present_entry_flush) {
-		if (!cap_caching_mode(iommu->cap))
-			return 1;
-		else
-			did = 0;
-	}
-
 	if (cap_write_drain(iommu->cap))
 		dw = 1;
 
@@ -760,7 +754,7 @@ int qi_flush_iotlb(struct intel_iommu *iommu, u16 did, u64 addr,
 	desc.high = QI_IOTLB_ADDR(addr) | QI_IOTLB_IH(ih)
 		| QI_IOTLB_AM(size_order);
 
-	return qi_submit_sync(&desc, iommu);
+	qi_submit_sync(&desc, iommu);
 }
 
 /*
@@ -790,7 +784,6 @@ void dmar_disable_qi(struct intel_iommu *iommu)
 		cpu_relax();
 
 	iommu->gcmd &= ~DMA_GCMD_QIE;
-
 	writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
 
 	IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl,
@@ -804,7 +797,7 @@ end:
  */
 static void __dmar_enable_qi(struct intel_iommu *iommu)
 {
-	u32 cmd, sts;
+	u32 sts;
 	unsigned long flags;
 	struct q_inval *qi = iommu->qi;
 
@@ -818,9 +811,8 @@ static void __dmar_enable_qi(struct intel_iommu *iommu)
 
 	dmar_writeq(iommu->reg + DMAR_IQA_REG, virt_to_phys(qi->desc));
 
-	cmd = iommu->gcmd | DMA_GCMD_QIE;
 	iommu->gcmd |= DMA_GCMD_QIE;
-	writel(cmd, iommu->reg + DMAR_GCMD_REG);
+	writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
 
 	/* Make sure hardware complete it */
 	IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl, (sts & DMA_GSTS_QIES), sts);
@@ -1096,7 +1088,7 @@ int dmar_set_interrupt(struct intel_iommu *iommu)
 		set_irq_data(irq, NULL);
 		iommu->irq = 0;
 		destroy_irq(irq);
-		return 0;
+		return ret;
 	}
 
 	ret = request_irq(irq, dmar_fault, 0, iommu->name, iommu);
diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
index cd389162735f..8cd4aacd0711 100644
--- a/drivers/pci/intel-iommu.c
+++ b/drivers/pci/intel-iommu.c
@@ -53,6 +53,8 @@
 
 #define DEFAULT_DOMAIN_ADDRESS_WIDTH 48
 
+#define MAX_AGAW_WIDTH 64
+
 #define DOMAIN_MAX_ADDR(gaw) ((((u64)1) << gaw) - 1)
 
 #define IOVA_PFN(addr)		((addr) >> PAGE_SHIFT)
@@ -131,8 +133,6 @@ static inline void context_set_fault_enable(struct context_entry *context)
 	context->lo &= (((u64)-1) << 2) | 1;
 }
 
-#define CONTEXT_TT_MULTI_LEVEL 0
-
 static inline void context_set_translation_type(struct context_entry *context,
 						unsigned long value)
 {
@@ -401,17 +401,13 @@ void free_iova_mem(struct iova *iova)
 
 static inline int width_to_agaw(int width);
 
-/* calculate agaw for each iommu.
- * "SAGAW" may be different across iommus, use a default agaw, and
- * get a supported less agaw for iommus that don't support the default agaw.
- */
-int iommu_calculate_agaw(struct intel_iommu *iommu)
+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;
@@ -420,6 +416,24 @@ int iommu_calculate_agaw(struct intel_iommu *iommu)
 	return agaw;
 }
 
+/*
+ * 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);
+}
+
 /* in native case, each domain is related to only one iommu */
 static struct intel_iommu *domain_get_iommu(struct dmar_domain *domain)
 {
@@ -809,7 +823,7 @@ static int iommu_alloc_root_entry(struct intel_iommu *iommu)
 static void iommu_set_root_entry(struct intel_iommu *iommu)
 {
 	void *addr;
-	u32 cmd, sts;
+	u32 sts;
 	unsigned long flag;
 
 	addr = iommu->root_entry;
@@ -817,12 +831,11 @@ static void iommu_set_root_entry(struct intel_iommu *iommu)
 	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);
 }
@@ -834,39 +847,25 @@ static void iommu_flush_write_buffer(struct intel_iommu *iommu)
 
 	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;
@@ -891,33 +890,16 @@ static int __iommu_flush_context(struct intel_iommu *iommu,
 		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 */
@@ -965,12 +947,10 @@ static int __iommu_flush_iotlb(struct intel_iommu *iommu, u16 did,
 		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 void iommu_flush_iotlb_psi(struct intel_iommu *iommu, u16 did,
+				  u64 addr, unsigned int pages)
 {
 	unsigned int mask;
 
@@ -980,8 +960,7 @@ static int iommu_flush_iotlb_psi(struct intel_iommu *iommu, u16 did,
 	/* 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);
+						DMA_TLB_DSI_FLUSH);
 
 	/*
 	 * PSI requires page size to be 2 ^ x, and the base address is naturally
@@ -991,11 +970,10 @@ static int iommu_flush_iotlb_psi(struct intel_iommu *iommu, u16 did,
 	/* 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);
+						DMA_TLB_DSI_FLUSH);
 
 	return iommu->flush.flush_iotlb(iommu, did, addr, mask,
-					DMA_TLB_PSI_FLUSH,
-					non_present_entry_flush);
+					DMA_TLB_PSI_FLUSH);
 }
 
 static void iommu_disable_protect_mem_regions(struct intel_iommu *iommu)
@@ -1021,13 +999,13 @@ static int iommu_enable_translation(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;
 }
@@ -1043,7 +1021,7 @@ static int iommu_disable_translation(struct intel_iommu *iommu)
 
 	/* 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;
@@ -1325,8 +1303,8 @@ static void domain_exit(struct dmar_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;
@@ -1339,7 +1317,10 @@ static int domain_context_mapping_one(struct dmar_domain *domain,
 
 	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)
@@ -1399,21 +1380,37 @@ static int domain_context_mapping_one(struct dmar_domain *domain,
 	}
 
 	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);
+
+	/*
+	 * In pass through mode, AW must be programmed to indicate the largest
+	 * AGAW value supported by hardware. And ASR is ignored by hardware.
+	 */
+	if (likely(translation == CONTEXT_TT_MULTI_LEVEL)) {
+		context_set_address_width(context, iommu->agaw);
+		context_set_address_root(context, virt_to_phys(pgd));
+	} else
+		context_set_address_width(context, iommu->msagaw);
+
+	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);
-
+	}
 	spin_unlock_irqrestore(&iommu->lock, flags);
 
 	spin_lock_irqsave(&domain->iommu_lock, flags);
@@ -1426,13 +1423,15 @@ static int domain_context_mapping_one(struct dmar_domain *domain,
 }
 
 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;
 
@@ -1446,7 +1445,7 @@ domain_context_mapping(struct dmar_domain *domain, struct pci_dev *pdev)
 		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;
@@ -1454,12 +1453,14 @@ domain_context_mapping(struct dmar_domain *domain, struct pci_dev *pdev)
 	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)
@@ -1540,9 +1541,8 @@ static void iommu_detach_dev(struct intel_iommu *iommu, u8 bus, u8 devfn)
 
 	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)
@@ -1756,7 +1756,7 @@ static int iommu_prepare_identity_map(struct pci_dev *pdev,
 		goto error;
 
 	/* context entry init */
-	ret = domain_context_mapping(domain, pdev);
+	ret = domain_context_mapping(domain, pdev, CONTEXT_TT_MULTI_LEVEL);
 	if (!ret)
 		return 0;
 error:
@@ -1857,6 +1857,23 @@ static inline void iommu_prepare_isa(void)
 }
 #endif /* !CONFIG_DMAR_FLPY_WA */
 
+/* Initialize each context entry as pass through.*/
+static int __init init_context_pass_through(void)
+{
+	struct pci_dev *pdev = NULL;
+	struct dmar_domain *domain;
+	int ret;
+
+	for_each_pci_dev(pdev) {
+		domain = get_domain_for_dev(pdev, DEFAULT_DOMAIN_ADDRESS_WIDTH);
+		ret = domain_context_mapping(domain, pdev,
+					     CONTEXT_TT_PASS_THROUGH);
+		if (ret)
+			return ret;
+	}
+	return 0;
+}
+
 static int __init init_dmars(void)
 {
 	struct dmar_drhd_unit *drhd;
@@ -1864,6 +1881,7 @@ static int __init init_dmars(void)
 	struct pci_dev *pdev;
 	struct intel_iommu *iommu;
 	int i, ret;
+	int pass_through = 1;
 
 	/*
 	 * for each drhd
@@ -1917,7 +1935,15 @@ static int __init init_dmars(void)
 			printk(KERN_ERR "IOMMU: allocate root entry failed\n");
 			goto error;
 		}
+		if (!ecap_pass_through(iommu->ecap))
+			pass_through = 0;
 	}
+	if (iommu_pass_through)
+		if (!pass_through) {
+			printk(KERN_INFO
+			       "Pass Through is not supported by hardware.\n");
+			iommu_pass_through = 0;
+		}
 
 	/*
 	 * Start from the sane iommu hardware state.
@@ -1973,35 +1999,56 @@ static int __init init_dmars(void)
 	}
 
 	/*
-	 * For each rmrr
-	 *   for each dev attached to rmrr
-	 *   do
-	 *     locate drhd for dev, alloc domain for dev
-	 *     allocate free domain
-	 *     allocate page table entries for rmrr
-	 *     if context not allocated for bus
-	 *           allocate and init context
-	 *           set present in root table for this bus
-	 *     init context with domain, translation etc
-	 *    endfor
-	 * endfor
+	 * If pass through is set and enabled, context entries of all pci
+	 * devices are intialized by pass through translation type.
 	 */
-	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 */
-			if (!pdev)
-				continue;
-			ret = iommu_prepare_rmrr_dev(rmrr, pdev);
-			if (ret)
-				printk(KERN_ERR
-				 "IOMMU: mapping reserved region failed\n");
+	if (iommu_pass_through) {
+		ret = init_context_pass_through();
+		if (ret) {
+			printk(KERN_ERR "IOMMU: Pass through init failed.\n");
+			iommu_pass_through = 0;
 		}
 	}
 
-	iommu_prepare_gfx_mapping();
+	/*
+	 * If pass through is not set or not enabled, setup context entries for
+	 * identity mappings for rmrr, gfx, and isa.
+	 */
+	if (!iommu_pass_through) {
+		/*
+		 * For each rmrr
+		 *   for each dev attached to rmrr
+		 *   do
+		 *     locate drhd for dev, alloc domain for dev
+		 *     allocate free domain
+		 *     allocate page table entries for rmrr
+		 *     if context not allocated for bus
+		 *           allocate and init context
+		 *           set present in root table for this bus
+		 *     init context with domain, translation etc
+		 *    endfor
+		 * endfor
+		 */
+		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.
+				 */
+				if (!pdev)
+					continue;
+				ret = iommu_prepare_rmrr_dev(rmrr, pdev);
+				if (ret)
+					printk(KERN_ERR
+				 "IOMMU: mapping reserved region failed\n");
+			}
+		}
+
+		iommu_prepare_gfx_mapping();
 
-	iommu_prepare_isa();
+		iommu_prepare_isa();
+	}
 
 	/*
 	 * for each drhd
@@ -2023,10 +2070,8 @@ static int __init init_dmars(void)
 
 		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->flush.flush_context(iommu, 0, 0, 0, DMA_CCMD_GLOBAL_INVL);
+		iommu->flush.flush_iotlb(iommu, 0, 0, 0, DMA_TLB_GLOBAL_FLUSH);
 		iommu_disable_protect_mem_regions(iommu);
 
 		ret = iommu_enable_translation(iommu);
@@ -2112,7 +2157,8 @@ get_valid_domain_for_dev(struct pci_dev *pdev)
 
 	/* 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",
@@ -2173,10 +2219,11 @@ static dma_addr_t __intel_map_single(struct device *hwdev, phys_addr_t paddr,
 	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, start_paddr,
+				      size >> VTD_PAGE_SHIFT);
+	else
 		iommu_flush_write_buffer(iommu);
 
 	return start_paddr + ((u64)paddr & (~PAGE_MASK));
@@ -2212,7 +2259,7 @@ static void flush_unmaps(void)
 
 		if (deferred_flush[i].next) {
 			iommu->flush.flush_iotlb(iommu, 0, 0, 0,
-						 DMA_TLB_GLOBAL_FLUSH, 0);
+						 DMA_TLB_GLOBAL_FLUSH);
 			for (j = 0; j < deferred_flush[i].next; j++) {
 				__free_iova(&deferred_flush[i].domain[j]->iovad,
 						deferred_flush[i].iova[j]);
@@ -2291,9 +2338,8 @@ static void intel_unmap_page(struct device *dev, dma_addr_t dev_addr,
 	/* free page tables */
 	dma_pte_free_pagetable(domain, start_addr, start_addr + size);
 	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_addr,
+				      size >> VTD_PAGE_SHIFT);
 		/* free iova */
 		__free_iova(&domain->iovad, iova);
 	} else {
@@ -2384,9 +2430,8 @@ static void intel_unmap_sg(struct device *hwdev, struct scatterlist *sglist,
 	/* free page tables */
 	dma_pte_free_pagetable(domain, start_addr, start_addr + size);
 
-	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_addr,
+			      size >> VTD_PAGE_SHIFT);
 
 	/* free iova */
 	__free_iova(&domain->iovad, iova);
@@ -2478,10 +2523,13 @@ static int intel_map_sg(struct device *hwdev, struct scatterlist *sglist, int ne
 		offset += size;
 	}
 
-	/* 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_addr,
+				      offset >> VTD_PAGE_SHIFT);
+	else
 		iommu_flush_write_buffer(iommu);
+
 	return nelems;
 }
 
@@ -2640,9 +2688,9 @@ static int init_iommu_hw(void)
 		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);
+					 DMA_TLB_GLOBAL_FLUSH);
 		iommu_disable_protect_mem_regions(iommu);
 		iommu_enable_translation(iommu);
 	}
@@ -2657,9 +2705,9 @@ static void iommu_flush_all(void)
 
 	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);
 	}
 }
 
@@ -2782,7 +2830,7 @@ int __init intel_iommu_init(void)
 	 * Check the need for DMA-remapping initialization now.
 	 * Above initialization will also be used by Interrupt-remapping.
 	 */
-	if (no_iommu || swiotlb || dmar_disabled)
+	if (no_iommu || (swiotlb && !iommu_pass_through) || dmar_disabled)
 		return -ENODEV;
 
 	iommu_init_mempool();
@@ -2802,7 +2850,15 @@ int __init intel_iommu_init(void)
 
 	init_timer(&unmap_timer);
 	force_iommu = 1;
-	dma_ops = &intel_dma_ops;
+
+	if (!iommu_pass_through) {
+		printk(KERN_INFO
+		       "Multi-level page-table translation for DMAR.\n");
+		dma_ops = &intel_dma_ops;
+	} else
+		printk(KERN_INFO
+		       "DMAR: Pass through translation for DMAR.\n");
+
 	init_iommu_sysfs();
 
 	register_iommu(&intel_iommu_ops);
@@ -3142,7 +3198,7 @@ static int intel_iommu_attach_device(struct iommu_domain *domain,
 		return -EFAULT;
 	}
 
-	ret = domain_context_mapping(dmar_domain, pdev);
+	ret = domain_context_mapping(dmar_domain, pdev, CONTEXT_TT_MULTI_LEVEL);
 	if (ret)
 		return ret;
 
diff --git a/drivers/pci/intr_remapping.c b/drivers/pci/intr_remapping.c
index 3a0cb0bb0593..1e83c8c5f985 100644
--- a/drivers/pci/intr_remapping.c
+++ b/drivers/pci/intr_remapping.c
@@ -409,7 +409,7 @@ int free_irte(int irq)
 static void iommu_set_intr_remapping(struct intel_iommu *iommu, int mode)
 {
 	u64 addr;
-	u32 cmd, sts;
+	u32 sts;
 	unsigned long flags;
 
 	addr = virt_to_phys((void *)iommu->ir_table->base);
@@ -420,9 +420,8 @@ static void iommu_set_intr_remapping(struct intel_iommu *iommu, int mode)
 		    (addr) | IR_X2APIC_MODE(mode) | INTR_REMAP_TABLE_REG_SIZE);
 
 	/* Set interrupt-remapping table pointer */
-	cmd = iommu->gcmd | DMA_GCMD_SIRTP;
 	iommu->gcmd |= DMA_GCMD_SIRTP;
-	writel(cmd, iommu->reg + DMAR_GCMD_REG);
+	writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
 
 	IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG,
 		      readl, (sts & DMA_GSTS_IRTPS), sts);
@@ -437,9 +436,8 @@ static void iommu_set_intr_remapping(struct intel_iommu *iommu, int mode)
 	spin_lock_irqsave(&iommu->register_lock, flags);
 
 	/* Enable interrupt-remapping */
-	cmd = iommu->gcmd | DMA_GCMD_IRE;
 	iommu->gcmd |= DMA_GCMD_IRE;
-	writel(cmd, iommu->reg + DMAR_GCMD_REG);
+	writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
 
 	IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG,
 		      readl, (sts & DMA_GSTS_IRES), sts);
diff --git a/include/linux/dma_remapping.h b/include/linux/dma_remapping.h
index 1a455f1f86d7..e0a03aff63d9 100644
--- a/include/linux/dma_remapping.h
+++ b/include/linux/dma_remapping.h
@@ -13,6 +13,9 @@
 #define DMA_PTE_WRITE (2)
 #define DMA_PTE_SNP (1 << 11)
 
+#define CONTEXT_TT_MULTI_LEVEL	0
+#define CONTEXT_TT_PASS_THROUGH 2
+
 struct intel_iommu;
 struct dmar_domain;
 struct root_entry;
@@ -21,11 +24,16 @@ extern void free_dmar_iommu(struct intel_iommu *iommu);
 
 #ifdef CONFIG_DMAR
 extern int iommu_calculate_agaw(struct intel_iommu *iommu);
+extern int iommu_calculate_max_sagaw(struct intel_iommu *iommu);
 #else
 static inline int iommu_calculate_agaw(struct intel_iommu *iommu)
 {
 	return 0;
 }
+static inline int iommu_calculate_max_sagaw(struct intel_iommu *iommu)
+{
+	return 0;
+}
 #endif
 
 extern int dmar_disabled;
diff --git a/include/linux/intel-iommu.h b/include/linux/intel-iommu.h
index aa8c53171233..29e05a034c09 100644
--- a/include/linux/intel-iommu.h
+++ b/include/linux/intel-iommu.h
@@ -120,6 +120,7 @@ static inline void dmar_writeq(void __iomem *addr, u64 val)
 	(ecap_iotlb_offset(e) + ecap_niotlb_iunits(e) * 16)
 #define ecap_coherent(e)	((e) & 0x1)
 #define ecap_qis(e)		((e) & 0x2)
+#define ecap_pass_through(e)	((e >> 6) & 0x1)
 #define ecap_eim_support(e)	((e >> 4) & 0x1)
 #define ecap_ir_support(e)	((e >> 3) & 0x1)
 #define ecap_max_handle_mask(e) ((e >> 20) & 0xf)
@@ -280,10 +281,10 @@ struct ir_table {
 #endif
 
 struct iommu_flush {
-	int (*flush_context)(struct intel_iommu *iommu, u16 did, u16 sid, u8 fm,
-		u64 type, int non_present_entry_flush);
-	int (*flush_iotlb)(struct intel_iommu *iommu, u16 did, u64 addr,
-		unsigned int size_order, u64 type, int non_present_entry_flush);
+	void (*flush_context)(struct intel_iommu *iommu, u16 did, u16 sid,
+			      u8 fm, u64 type);
+	void (*flush_iotlb)(struct intel_iommu *iommu, u16 did, u64 addr,
+			    unsigned int size_order, u64 type);
 };
 
 enum {
@@ -302,6 +303,7 @@ struct intel_iommu {
 	spinlock_t	register_lock; /* protect register handling */
 	int		seq_id;	/* sequence id of the iommu */
 	int		agaw; /* agaw of this iommu */
+	int		msagaw; /* max sagaw of this iommu */
 	unsigned int 	irq;
 	unsigned char 	name[13];    /* Device Name */
 
@@ -337,11 +339,10 @@ extern void dmar_disable_qi(struct intel_iommu *iommu);
 extern int dmar_reenable_qi(struct intel_iommu *iommu);
 extern void qi_global_iec(struct intel_iommu *iommu);
 
-extern int qi_flush_context(struct intel_iommu *iommu, u16 did, u16 sid,
-			        u8 fm, u64 type, int non_present_entry_flush);
-extern int qi_flush_iotlb(struct intel_iommu *iommu, u16 did, u64 addr,
-			  unsigned int size_order, u64 type,
-			  int non_present_entry_flush);
+extern void qi_flush_context(struct intel_iommu *iommu, u16 did, u16 sid,
+			     u8 fm, u64 type);
+extern void qi_flush_iotlb(struct intel_iommu *iommu, u16 did, u64 addr,
+			  unsigned int size_order, u64 type);
 
 extern int qi_submit_sync(struct qi_desc *desc, struct intel_iommu *iommu);