Merge branch 'for-joerg/arm-smmu/updates' of git://git.kernel.org/pub/scm/linux/kernel/git/will/linux into arm/smmu

9 files changed, 1706 insertions, 620 deletions

diff --git a/MAINTAINERS b/MAINTAINERS
index 2ebb056cbe0a..c437f7e9e9a4 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1582,6 +1582,7 @@ M:	Will Deacon <will.deacon@arm.com>
 L:	linux-arm-kernel@lists.infradead.org (moderated for non-subscribers)
 S:	Maintained
 F:	drivers/iommu/arm-smmu.c
+F:	drivers/iommu/io-pgtable-arm.c
 
 ARM64 PORT (AARCH64 ARCHITECTURE)
 M:	Catalin Marinas <catalin.marinas@arm.com>
diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index b1f9a20a3677..528c3fd2d4c1 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -349,7 +349,6 @@ config ARM64_VA_BITS_42
 
 config ARM64_VA_BITS_48
 	bool "48-bit"
-	depends on !ARM_SMMU
 
 endchoice
 
diff --git a/drivers/iommu/Kconfig b/drivers/iommu/Kconfig
index 325188eef1c1..87060ad6829d 100644
--- a/drivers/iommu/Kconfig
+++ b/drivers/iommu/Kconfig
@@ -13,6 +13,32 @@ menuconfig IOMMU_SUPPORT
 
 if IOMMU_SUPPORT
 
+menu "Generic IOMMU Pagetable Support"
+
+# Selected by the actual pagetable implementations
+config IOMMU_IO_PGTABLE
+	bool
+
+config IOMMU_IO_PGTABLE_LPAE
+	bool "ARMv7/v8 Long Descriptor Format"
+	select IOMMU_IO_PGTABLE
+	help
+	  Enable support for the ARM long descriptor pagetable format.
+	  This allocator supports 4K/2M/1G, 16K/32M and 64K/512M page
+	  sizes at both stage-1 and stage-2, as well as address spaces
+	  up to 48-bits in size.
+
+config IOMMU_IO_PGTABLE_LPAE_SELFTEST
+	bool "LPAE selftests"
+	depends on IOMMU_IO_PGTABLE_LPAE
+	help
+	  Enable self-tests for LPAE page table allocator. This performs
+	  a series of page-table consistency checks during boot.
+
+	  If unsure, say N here.
+
+endmenu
+
 config OF_IOMMU
        def_bool y
        depends on OF && IOMMU_API
@@ -304,13 +330,13 @@ config SPAPR_TCE_IOMMU
 
 config ARM_SMMU
 	bool "ARM Ltd. System MMU (SMMU) Support"
-	depends on ARM64 || (ARM_LPAE && OF)
+	depends on ARM64 || ARM
 	select IOMMU_API
+	select IOMMU_IO_PGTABLE_LPAE
 	select ARM_DMA_USE_IOMMU if ARM
 	help
 	  Support for implementations of the ARM System MMU architecture
-	  versions 1 and 2. The driver supports both v7l and v8l table
-	  formats with 4k and 64k page sizes.
+	  versions 1 and 2.
 
 	  Say Y here if your SoC includes an IOMMU device implementing
 	  the ARM SMMU architecture.
diff --git a/drivers/iommu/Makefile b/drivers/iommu/Makefile
index 7b976f294a69..d6889b487d55 100644
--- a/drivers/iommu/Makefile
+++ b/drivers/iommu/Makefile
@@ -1,6 +1,8 @@
 obj-$(CONFIG_IOMMU_API) += iommu.o
 obj-$(CONFIG_IOMMU_API) += iommu-traces.o
 obj-$(CONFIG_IOMMU_API) += iommu-sysfs.o
+obj-$(CONFIG_IOMMU_IO_PGTABLE) += io-pgtable.o
+obj-$(CONFIG_IOMMU_IO_PGTABLE_LPAE) += io-pgtable-arm.o
 obj-$(CONFIG_OF_IOMMU)	+= of_iommu.o
 obj-$(CONFIG_MSM_IOMMU) += msm_iommu.o msm_iommu_dev.o
 obj-$(CONFIG_AMD_IOMMU) += amd_iommu.o amd_iommu_init.o
diff --git a/drivers/iommu/arm-smmu.c b/drivers/iommu/arm-smmu.c
index 6cd47b75286f..1d6d43bb3395 100644
--- a/drivers/iommu/arm-smmu.c
+++ b/drivers/iommu/arm-smmu.c
@@ -23,8 +23,6 @@
  *	- Stream-matching and stream-indexing
  *	- v7/v8 long-descriptor format
  *	- Non-secure access to the SMMU
- *	- 4k and 64k pages, with contiguous pte hints.
- *	- Up to 48-bit addressing (dependent on VA_BITS)
  *	- Context fault reporting
  */
 
@@ -36,7 +34,7 @@
 #include <linux/interrupt.h>
 #include <linux/io.h>
 #include <linux/iommu.h>
-#include <linux/mm.h>
+#include <linux/iopoll.h>
 #include <linux/module.h>
 #include <linux/of.h>
 #include <linux/pci.h>
@@ -46,7 +44,7 @@
 
 #include <linux/amba/bus.h>
 
-#include <asm/pgalloc.h>
+#include "io-pgtable.h"
 
 /* Maximum number of stream IDs assigned to a single device */
 #define MAX_MASTER_STREAMIDS		MAX_PHANDLE_ARGS
@@ -71,40 +69,6 @@
 		((smmu->options & ARM_SMMU_OPT_SECURE_CFG_ACCESS)	\
 			? 0x400 : 0))
 
-/* Page table bits */
-#define ARM_SMMU_PTE_XN			(((pteval_t)3) << 53)
-#define ARM_SMMU_PTE_CONT		(((pteval_t)1) << 52)
-#define ARM_SMMU_PTE_AF			(((pteval_t)1) << 10)
-#define ARM_SMMU_PTE_SH_NS		(((pteval_t)0) << 8)
-#define ARM_SMMU_PTE_SH_OS		(((pteval_t)2) << 8)
-#define ARM_SMMU_PTE_SH_IS		(((pteval_t)3) << 8)
-#define ARM_SMMU_PTE_PAGE		(((pteval_t)3) << 0)
-
-#if PAGE_SIZE == SZ_4K
-#define ARM_SMMU_PTE_CONT_ENTRIES	16
-#elif PAGE_SIZE == SZ_64K
-#define ARM_SMMU_PTE_CONT_ENTRIES	32
-#else
-#define ARM_SMMU_PTE_CONT_ENTRIES	1
-#endif
-
-#define ARM_SMMU_PTE_CONT_SIZE		(PAGE_SIZE * ARM_SMMU_PTE_CONT_ENTRIES)
-#define ARM_SMMU_PTE_CONT_MASK		(~(ARM_SMMU_PTE_CONT_SIZE - 1))
-
-/* Stage-1 PTE */
-#define ARM_SMMU_PTE_AP_UNPRIV		(((pteval_t)1) << 6)
-#define ARM_SMMU_PTE_AP_RDONLY		(((pteval_t)2) << 6)
-#define ARM_SMMU_PTE_ATTRINDX_SHIFT	2
-#define ARM_SMMU_PTE_nG			(((pteval_t)1) << 11)
-
-/* Stage-2 PTE */
-#define ARM_SMMU_PTE_HAP_FAULT		(((pteval_t)0) << 6)
-#define ARM_SMMU_PTE_HAP_READ		(((pteval_t)1) << 6)
-#define ARM_SMMU_PTE_HAP_WRITE		(((pteval_t)2) << 6)
-#define ARM_SMMU_PTE_MEMATTR_OIWB	(((pteval_t)0xf) << 2)
-#define ARM_SMMU_PTE_MEMATTR_NC		(((pteval_t)0x5) << 2)
-#define ARM_SMMU_PTE_MEMATTR_DEV	(((pteval_t)0x1) << 2)
-
 /* Configuration registers */
 #define ARM_SMMU_GR0_sCR0		0x0
 #define sCR0_CLIENTPD			(1 << 0)
@@ -132,17 +96,12 @@
 #define ARM_SMMU_GR0_sGFSYNR0		0x50
 #define ARM_SMMU_GR0_sGFSYNR1		0x54
 #define ARM_SMMU_GR0_sGFSYNR2		0x58
-#define ARM_SMMU_GR0_PIDR0		0xfe0
-#define ARM_SMMU_GR0_PIDR1		0xfe4
-#define ARM_SMMU_GR0_PIDR2		0xfe8
 
 #define ID0_S1TS			(1 << 30)
 #define ID0_S2TS			(1 << 29)
 #define ID0_NTS				(1 << 28)
 #define ID0_SMS				(1 << 27)
-#define ID0_PTFS_SHIFT			24
-#define ID0_PTFS_MASK			0x2
-#define ID0_PTFS_V8_ONLY		0x2
+#define ID0_ATOSNS			(1 << 26)
 #define ID0_CTTW			(1 << 14)
 #define ID0_NUMIRPT_SHIFT		16
 #define ID0_NUMIRPT_MASK		0xff
@@ -169,11 +128,7 @@
 #define ID2_PTFS_16K			(1 << 13)
 #define ID2_PTFS_64K			(1 << 14)
 
-#define PIDR2_ARCH_SHIFT		4
-#define PIDR2_ARCH_MASK			0xf
-
 /* Global TLB invalidation */
-#define ARM_SMMU_GR0_STLBIALL		0x60
 #define ARM_SMMU_GR0_TLBIVMID		0x64
 #define ARM_SMMU_GR0_TLBIALLNSNH	0x68
 #define ARM_SMMU_GR0_TLBIALLH		0x6c
@@ -231,13 +186,25 @@
 #define ARM_SMMU_CB_TTBCR2		0x10
 #define ARM_SMMU_CB_TTBR0_LO		0x20
 #define ARM_SMMU_CB_TTBR0_HI		0x24
+#define ARM_SMMU_CB_TTBR1_LO		0x28
+#define ARM_SMMU_CB_TTBR1_HI		0x2c
 #define ARM_SMMU_CB_TTBCR		0x30
 #define ARM_SMMU_CB_S1_MAIR0		0x38
+#define ARM_SMMU_CB_S1_MAIR1		0x3c
+#define ARM_SMMU_CB_PAR_LO		0x50
+#define ARM_SMMU_CB_PAR_HI		0x54
 #define ARM_SMMU_CB_FSR			0x58
 #define ARM_SMMU_CB_FAR_LO		0x60
 #define ARM_SMMU_CB_FAR_HI		0x64
 #define ARM_SMMU_CB_FSYNR0		0x68
+#define ARM_SMMU_CB_S1_TLBIVA		0x600
 #define ARM_SMMU_CB_S1_TLBIASID		0x610
+#define ARM_SMMU_CB_S1_TLBIVAL		0x620
+#define ARM_SMMU_CB_S2_TLBIIPAS2	0x630
+#define ARM_SMMU_CB_S2_TLBIIPAS2L	0x638
+#define ARM_SMMU_CB_ATS1PR_LO		0x800
+#define ARM_SMMU_CB_ATS1PR_HI		0x804
+#define ARM_SMMU_CB_ATSR		0x8f0
 
 #define SCTLR_S1_ASIDPNE		(1 << 12)
 #define SCTLR_CFCFG			(1 << 7)
@@ -249,47 +216,16 @@
 #define SCTLR_M				(1 << 0)
 #define SCTLR_EAE_SBOP			(SCTLR_AFE | SCTLR_TRE)
 
-#define RESUME_RETRY			(0 << 0)
-#define RESUME_TERMINATE		(1 << 0)
-
-#define TTBCR_EAE			(1 << 31)
+#define CB_PAR_F			(1 << 0)
 
-#define TTBCR_PASIZE_SHIFT		16
-#define TTBCR_PASIZE_MASK		0x7
+#define ATSR_ACTIVE			(1 << 0)
 
-#define TTBCR_TG0_4K			(0 << 14)
-#define TTBCR_TG0_64K			(1 << 14)
-
-#define TTBCR_SH0_SHIFT			12
-#define TTBCR_SH0_MASK			0x3
-#define TTBCR_SH_NS			0
-#define TTBCR_SH_OS			2
-#define TTBCR_SH_IS			3
-
-#define TTBCR_ORGN0_SHIFT		10
-#define TTBCR_IRGN0_SHIFT		8
-#define TTBCR_RGN_MASK			0x3
-#define TTBCR_RGN_NC			0
-#define TTBCR_RGN_WBWA			1
-#define TTBCR_RGN_WT			2
-#define TTBCR_RGN_WB			3
-
-#define TTBCR_SL0_SHIFT			6
-#define TTBCR_SL0_MASK			0x3
-#define TTBCR_SL0_LVL_2			0
-#define TTBCR_SL0_LVL_1			1
-
-#define TTBCR_T1SZ_SHIFT		16
-#define TTBCR_T0SZ_SHIFT		0
-#define TTBCR_SZ_MASK			0xf
+#define RESUME_RETRY			(0 << 0)
+#define RESUME_TERMINATE		(1 << 0)
 
 #define TTBCR2_SEP_SHIFT		15
 #define TTBCR2_SEP_MASK			0x7
 
-#define TTBCR2_PASIZE_SHIFT		0
-#define TTBCR2_PASIZE_MASK		0x7
-
-/* Common definitions for PASize and SEP fields */
 #define TTBCR2_ADDR_32			0
 #define TTBCR2_ADDR_36			1
 #define TTBCR2_ADDR_40			2
@@ -297,16 +233,7 @@
 #define TTBCR2_ADDR_44			4
 #define TTBCR2_ADDR_48			5
 
-#define TTBRn_HI_ASID_SHIFT		16
-
-#define MAIR_ATTR_SHIFT(n)		((n) << 3)
-#define MAIR_ATTR_MASK			0xff
-#define MAIR_ATTR_DEVICE		0x04
-#define MAIR_ATTR_NC			0x44
-#define MAIR_ATTR_WBRWA			0xff
-#define MAIR_ATTR_IDX_NC		0
-#define MAIR_ATTR_IDX_CACHE		1
-#define MAIR_ATTR_IDX_DEV		2
+#define TTBRn_HI_ASID_SHIFT            16
 
 #define FSR_MULTI			(1 << 31)
 #define FSR_SS				(1 << 30)
@@ -366,6 +293,7 @@ struct arm_smmu_device {
 #define ARM_SMMU_FEAT_TRANS_S1		(1 << 2)
 #define ARM_SMMU_FEAT_TRANS_S2		(1 << 3)
 #define ARM_SMMU_FEAT_TRANS_NESTED	(1 << 4)
+#define ARM_SMMU_FEAT_TRANS_OPS		(1 << 5)
 	u32				features;
 
 #define ARM_SMMU_OPT_SECURE_CFG_ACCESS (1 << 0)
@@ -380,10 +308,9 @@ struct arm_smmu_device {
 	u32				num_mapping_groups;
 	DECLARE_BITMAP(smr_map, ARM_SMMU_MAX_SMRS);
 
-	unsigned long			s1_input_size;
-	unsigned long			s1_output_size;
-	unsigned long			s2_input_size;
-	unsigned long			s2_output_size;
+	unsigned long			va_size;
+	unsigned long			ipa_size;
+	unsigned long			pa_size;
 
 	u32				num_global_irqs;
 	u32				num_context_irqs;
@@ -397,7 +324,6 @@ struct arm_smmu_cfg {
 	u8				cbndx;
 	u8				irptndx;
 	u32				cbar;
-	pgd_t				*pgd;
 };
 #define INVALID_IRPTNDX			0xff
 
@@ -412,11 +338,15 @@ enum arm_smmu_domain_stage {
 
 struct arm_smmu_domain {
 	struct arm_smmu_device		*smmu;
+	struct io_pgtable_ops		*pgtbl_ops;
+	spinlock_t			pgtbl_lock;
 	struct arm_smmu_cfg		cfg;
 	enum arm_smmu_domain_stage	stage;
-	spinlock_t			lock;
+	struct mutex			init_mutex; /* Protects smmu pointer */
 };
 
+static struct iommu_ops arm_smmu_ops;
+
 static DEFINE_SPINLOCK(arm_smmu_devices_lock);
 static LIST_HEAD(arm_smmu_devices);
 
@@ -597,7 +527,7 @@ static void __arm_smmu_free_bitmap(unsigned long *map, int idx)
 }
 
 /* Wait for any pending TLB invalidations to complete */
-static void arm_smmu_tlb_sync(struct arm_smmu_device *smmu)
+static void __arm_smmu_tlb_sync(struct arm_smmu_device *smmu)
 {
 	int count = 0;
 	void __iomem *gr0_base = ARM_SMMU_GR0(smmu);
@@ -615,12 +545,19 @@ static void arm_smmu_tlb_sync(struct arm_smmu_device *smmu)
 	}
 }
 
-static void arm_smmu_tlb_inv_context(struct arm_smmu_domain *smmu_domain)
+static void arm_smmu_tlb_sync(void *cookie)
 {
+	struct arm_smmu_domain *smmu_domain = cookie;
+	__arm_smmu_tlb_sync(smmu_domain->smmu);
+}
+
+static void arm_smmu_tlb_inv_context(void *cookie)
+{
+	struct arm_smmu_domain *smmu_domain = cookie;
 	struct arm_smmu_cfg *cfg = &smmu_domain->cfg;
 	struct arm_smmu_device *smmu = smmu_domain->smmu;
-	void __iomem *base = ARM_SMMU_GR0(smmu);
 	bool stage1 = cfg->cbar != CBAR_TYPE_S2_TRANS;
+	void __iomem *base;
 
 	if (stage1) {
 		base = ARM_SMMU_CB_BASE(smmu) + ARM_SMMU_CB(smmu, cfg->cbndx);
@@ -632,9 +569,76 @@ static void arm_smmu_tlb_inv_context(struct arm_smmu_domain *smmu_domain)
 			       base + ARM_SMMU_GR0_TLBIVMID);
 	}
 
-	arm_smmu_tlb_sync(smmu);
+	__arm_smmu_tlb_sync(smmu);
+}
+
+static void arm_smmu_tlb_inv_range_nosync(unsigned long iova, size_t size,
+					  bool leaf, void *cookie)
+{
+	struct arm_smmu_domain *smmu_domain = cookie;
+	struct arm_smmu_cfg *cfg = &smmu_domain->cfg;
+	struct arm_smmu_device *smmu = smmu_domain->smmu;
+	bool stage1 = cfg->cbar != CBAR_TYPE_S2_TRANS;
+	void __iomem *reg;
+
+	if (stage1) {
+		reg = ARM_SMMU_CB_BASE(smmu) + ARM_SMMU_CB(smmu, cfg->cbndx);
+		reg += leaf ? ARM_SMMU_CB_S1_TLBIVAL : ARM_SMMU_CB_S1_TLBIVA;
+
+		if (!IS_ENABLED(CONFIG_64BIT) || smmu->version == ARM_SMMU_V1) {
+			iova &= ~12UL;
+			iova |= ARM_SMMU_CB_ASID(cfg);
+			writel_relaxed(iova, reg);
+#ifdef CONFIG_64BIT
+		} else {
+			iova >>= 12;
+			iova |= (u64)ARM_SMMU_CB_ASID(cfg) << 48;
+			writeq_relaxed(iova, reg);
+#endif
+		}
+#ifdef CONFIG_64BIT
+	} else if (smmu->version == ARM_SMMU_V2) {
+		reg = ARM_SMMU_CB_BASE(smmu) + ARM_SMMU_CB(smmu, cfg->cbndx);
+		reg += leaf ? ARM_SMMU_CB_S2_TLBIIPAS2L :
+			      ARM_SMMU_CB_S2_TLBIIPAS2;
+		writeq_relaxed(iova >> 12, reg);
+#endif
+	} else {
+		reg = ARM_SMMU_GR0(smmu) + ARM_SMMU_GR0_TLBIVMID;
+		writel_relaxed(ARM_SMMU_CB_VMID(cfg), reg);
+	}
+}
+
+static void arm_smmu_flush_pgtable(void *addr, size_t size, void *cookie)
+{
+	struct arm_smmu_domain *smmu_domain = cookie;
+	struct arm_smmu_device *smmu = smmu_domain->smmu;
+	unsigned long offset = (unsigned long)addr & ~PAGE_MASK;
+
+
+	/* Ensure new page tables are visible to the hardware walker */
+	if (smmu->features & ARM_SMMU_FEAT_COHERENT_WALK) {
+		dsb(ishst);
+	} else {
+		/*
+		 * If the SMMU can't walk tables in the CPU caches, treat them
+		 * like non-coherent DMA since we need to flush the new entries
+		 * all the way out to memory. There's no possibility of
+		 * recursion here as the SMMU table walker will not be wired
+		 * through another SMMU.
+		 */
+		dma_map_page(smmu->dev, virt_to_page(addr), offset, size,
+			     DMA_TO_DEVICE);
+	}
 }
 
+static struct iommu_gather_ops arm_smmu_gather_ops = {
+	.tlb_flush_all	= arm_smmu_tlb_inv_context,
+	.tlb_add_flush	= arm_smmu_tlb_inv_range_nosync,
+	.tlb_sync	= arm_smmu_tlb_sync,
+	.flush_pgtable	= arm_smmu_flush_pgtable,
+};
+
 static irqreturn_t arm_smmu_context_fault(int irq, void *dev)
 {
 	int flags, ret;
@@ -712,29 +716,8 @@ static irqreturn_t arm_smmu_global_fault(int irq, void *dev)
 	return IRQ_HANDLED;
 }
 
-static void arm_smmu_flush_pgtable(struct arm_smmu_device *smmu, void *addr,
-				   size_t size)
-{
-	unsigned long offset = (unsigned long)addr & ~PAGE_MASK;
-
-
-	/* Ensure new page tables are visible to the hardware walker */
-	if (smmu->features & ARM_SMMU_FEAT_COHERENT_WALK) {
-		dsb(ishst);
-	} else {
-		/*
-		 * If the SMMU can't walk tables in the CPU caches, treat them
-		 * like non-coherent DMA since we need to flush the new entries
-		 * all the way out to memory. There's no possibility of
-		 * recursion here as the SMMU table walker will not be wired
-		 * through another SMMU.
-		 */
-		dma_map_page(smmu->dev, virt_to_page(addr), offset, size,
-				DMA_TO_DEVICE);
-	}
-}
-
-static void arm_smmu_init_context_bank(struct arm_smmu_domain *smmu_domain)
+static void arm_smmu_init_context_bank(struct arm_smmu_domain *smmu_domain,
+				       struct io_pgtable_cfg *pgtbl_cfg)
 {
 	u32 reg;
 	bool stage1;
@@ -771,124 +754,68 @@ static void arm_smmu_init_context_bank(struct arm_smmu_domain *smmu_domain)
 #else
 		reg = CBA2R_RW64_32BIT;
 #endif
-		writel_relaxed(reg,
-			       gr1_base + ARM_SMMU_GR1_CBA2R(cfg->cbndx));
-
-		/* TTBCR2 */
-		switch (smmu->s1_input_size) {
-		case 32:
-			reg = (TTBCR2_ADDR_32 << TTBCR2_SEP_SHIFT);
-			break;
-		case 36:
-			reg = (TTBCR2_ADDR_36 << TTBCR2_SEP_SHIFT);
-			break;
-		case 39:
-		case 40:
-			reg = (TTBCR2_ADDR_40 << TTBCR2_SEP_SHIFT);
-			break;
-		case 42:
-			reg = (TTBCR2_ADDR_42 << TTBCR2_SEP_SHIFT);
-			break;
-		case 44:
-			reg = (TTBCR2_ADDR_44 << TTBCR2_SEP_SHIFT);
-			break;
-		case 48:
-			reg = (TTBCR2_ADDR_48 << TTBCR2_SEP_SHIFT);
-			break;
-		}
-
-		switch (smmu->s1_output_size) {
-		case 32:
-			reg |= (TTBCR2_ADDR_32 << TTBCR2_PASIZE_SHIFT);
-			break;
-		case 36:
-			reg |= (TTBCR2_ADDR_36 << TTBCR2_PASIZE_SHIFT);
-			break;
-		case 39:
-		case 40:
-			reg |= (TTBCR2_ADDR_40 << TTBCR2_PASIZE_SHIFT);
-			break;
-		case 42:
-			reg |= (TTBCR2_ADDR_42 << TTBCR2_PASIZE_SHIFT);
-			break;
-		case 44:
-			reg |= (TTBCR2_ADDR_44 << TTBCR2_PASIZE_SHIFT);
-			break;
-		case 48:
-			reg |= (TTBCR2_ADDR_48 << TTBCR2_PASIZE_SHIFT);
-			break;
-		}
-
-		if (stage1)
-			writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBCR2);
+		writel_relaxed(reg, gr1_base + ARM_SMMU_GR1_CBA2R(cfg->cbndx));
 	}
 
-	/* TTBR0 */
-	arm_smmu_flush_pgtable(smmu, cfg->pgd,
-			       PTRS_PER_PGD * sizeof(pgd_t));
-	reg = __pa(cfg->pgd);
-	writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBR0_LO);
-	reg = (phys_addr_t)__pa(cfg->pgd) >> 32;
-	if (stage1)
+	/* TTBRs */
+	if (stage1) {
+		reg = pgtbl_cfg->arm_lpae_s1_cfg.ttbr[0];
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBR0_LO);
+		reg = pgtbl_cfg->arm_lpae_s1_cfg.ttbr[0] >> 32;
 		reg |= ARM_SMMU_CB_ASID(cfg) << TTBRn_HI_ASID_SHIFT;
-	writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBR0_HI);
-
-	/*
-	 * TTBCR
-	 * We use long descriptor, with inner-shareable WBWA tables in TTBR0.
-	 */
-	if (smmu->version > ARM_SMMU_V1) {
-		if (PAGE_SIZE == SZ_4K)
-			reg = TTBCR_TG0_4K;
-		else
-			reg = TTBCR_TG0_64K;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBR0_HI);
 
-		if (!stage1) {
-			reg |= (64 - smmu->s2_input_size) << TTBCR_T0SZ_SHIFT;
+		reg = pgtbl_cfg->arm_lpae_s1_cfg.ttbr[1];
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBR1_LO);
+		reg = pgtbl_cfg->arm_lpae_s1_cfg.ttbr[1] >> 32;
+		reg |= ARM_SMMU_CB_ASID(cfg) << TTBRn_HI_ASID_SHIFT;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBR1_HI);
+	} else {
+		reg = pgtbl_cfg->arm_lpae_s2_cfg.vttbr;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBR0_LO);
+		reg = pgtbl_cfg->arm_lpae_s2_cfg.vttbr >> 32;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBR0_HI);
+	}
 
-			switch (smmu->s2_output_size) {
+	/* TTBCR */
+	if (stage1) {
+		reg = pgtbl_cfg->arm_lpae_s1_cfg.tcr;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBCR);
+		if (smmu->version > ARM_SMMU_V1) {
+			reg = pgtbl_cfg->arm_lpae_s1_cfg.tcr >> 32;
+			switch (smmu->va_size) {
 			case 32:
-				reg |= (TTBCR2_ADDR_32 << TTBCR_PASIZE_SHIFT);
+				reg |= (TTBCR2_ADDR_32 << TTBCR2_SEP_SHIFT);
 				break;
 			case 36:
-				reg |= (TTBCR2_ADDR_36 << TTBCR_PASIZE_SHIFT);
+				reg |= (TTBCR2_ADDR_36 << TTBCR2_SEP_SHIFT);
 				break;
 			case 40:
-				reg |= (TTBCR2_ADDR_40 << TTBCR_PASIZE_SHIFT);
+				reg |= (TTBCR2_ADDR_40 << TTBCR2_SEP_SHIFT);
 				break;
 			case 42:
-				reg |= (TTBCR2_ADDR_42 << TTBCR_PASIZE_SHIFT);
+				reg |= (TTBCR2_ADDR_42 << TTBCR2_SEP_SHIFT);
 				break;
 			case 44:
-				reg |= (TTBCR2_ADDR_44 << TTBCR_PASIZE_SHIFT);
+				reg |= (TTBCR2_ADDR_44 << TTBCR2_SEP_SHIFT);
 				break;
 			case 48:
-				reg |= (TTBCR2_ADDR_48 << TTBCR_PASIZE_SHIFT);
+				reg |= (TTBCR2_ADDR_48 << TTBCR2_SEP_SHIFT);
 				break;
 			}
-		} else {
-			reg |= (64 - smmu->s1_input_size) << TTBCR_T0SZ_SHIFT;
+			writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBCR2);
 		}
 	} else {
-		reg = 0;
+		reg = pgtbl_cfg->arm_lpae_s2_cfg.vtcr;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBCR);
 	}
 
-	reg |= TTBCR_EAE |
-	      (TTBCR_SH_IS << TTBCR_SH0_SHIFT) |
-	      (TTBCR_RGN_WBWA << TTBCR_ORGN0_SHIFT) |
-	      (TTBCR_RGN_WBWA << TTBCR_IRGN0_SHIFT);
-
-	if (!stage1)
-		reg |= (TTBCR_SL0_LVL_1 << TTBCR_SL0_SHIFT);
-
-	writel_relaxed(reg, cb_base + ARM_SMMU_CB_TTBCR);
-
-	/* MAIR0 (stage-1 only) */
+	/* MAIRs (stage-1 only) */
 	if (stage1) {
-		reg = (MAIR_ATTR_NC << MAIR_ATTR_SHIFT(MAIR_ATTR_IDX_NC)) |
-		      (MAIR_ATTR_WBRWA << MAIR_ATTR_SHIFT(MAIR_ATTR_IDX_CACHE)) |
-		      (MAIR_ATTR_DEVICE << MAIR_ATTR_SHIFT(MAIR_ATTR_IDX_DEV));
+		reg = pgtbl_cfg->arm_lpae_s1_cfg.mair[0];
 		writel_relaxed(reg, cb_base + ARM_SMMU_CB_S1_MAIR0);
+		reg = pgtbl_cfg->arm_lpae_s1_cfg.mair[1];
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_S1_MAIR1);
 	}
 
 	/* SCTLR */
@@ -905,11 +832,14 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain,
 					struct arm_smmu_device *smmu)
 {
 	int irq, start, ret = 0;
-	unsigned long flags;
+	unsigned long ias, oas;
+	struct io_pgtable_ops *pgtbl_ops;
+	struct io_pgtable_cfg pgtbl_cfg;
+	enum io_pgtable_fmt fmt;
 	struct arm_smmu_domain *smmu_domain = domain->priv;
 	struct arm_smmu_cfg *cfg = &smmu_domain->cfg;
 
-	spin_lock_irqsave(&smmu_domain->lock, flags);
+	mutex_lock(&smmu_domain->init_mutex);
 	if (smmu_domain->smmu)
 		goto out_unlock;
 
@@ -940,6 +870,12 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain,
 	case ARM_SMMU_DOMAIN_S1:
 		cfg->cbar = CBAR_TYPE_S1_TRANS_S2_BYPASS;
 		start = smmu->num_s2_context_banks;
+		ias = smmu->va_size;
+		oas = smmu->ipa_size;
+		if (IS_ENABLED(CONFIG_64BIT))
+			fmt = ARM_64_LPAE_S1;
+		else
+			fmt = ARM_32_LPAE_S1;
 		break;
 	case ARM_SMMU_DOMAIN_NESTED:
 		/*
@@ -949,6 +885,12 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain,
 	case ARM_SMMU_DOMAIN_S2:
 		cfg->cbar = CBAR_TYPE_S2_TRANS;
 		start = 0;
+		ias = smmu->ipa_size;
+		oas = smmu->pa_size;
+		if (IS_ENABLED(CONFIG_64BIT))
+			fmt = ARM_64_LPAE_S2;
+		else
+			fmt = ARM_32_LPAE_S2;
 		break;
 	default:
 		ret = -EINVAL;
@@ -968,10 +910,30 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain,
 		cfg->irptndx = cfg->cbndx;
 	}
 
-	ACCESS_ONCE(smmu_domain->smmu) = smmu;
-	arm_smmu_init_context_bank(smmu_domain);
-	spin_unlock_irqrestore(&smmu_domain->lock, flags);
+	pgtbl_cfg = (struct io_pgtable_cfg) {
+		.pgsize_bitmap	= arm_smmu_ops.pgsize_bitmap,
+		.ias		= ias,
+		.oas		= oas,
+		.tlb		= &arm_smmu_gather_ops,
+	};
+
+	smmu_domain->smmu = smmu;
+	pgtbl_ops = alloc_io_pgtable_ops(fmt, &pgtbl_cfg, smmu_domain);
+	if (!pgtbl_ops) {
+		ret = -ENOMEM;
+		goto out_clear_smmu;
+	}
+
+	/* Update our support page sizes to reflect the page table format */
+	arm_smmu_ops.pgsize_bitmap = pgtbl_cfg.pgsize_bitmap;
 
+	/* Initialise the context bank with our page table cfg */
+	arm_smmu_init_context_bank(smmu_domain, &pgtbl_cfg);
+
+	/*
+	 * Request context fault interrupt. Do this last to avoid the
+	 * handler seeing a half-initialised domain state.
+	 */
 	irq = smmu->irqs[smmu->num_global_irqs + cfg->irptndx];
 	ret = request_irq(irq, arm_smmu_context_fault, IRQF_SHARED,
 			  "arm-smmu-context-fault", domain);
@@ -981,10 +943,16 @@ static int arm_smmu_init_domain_context(struct iommu_domain *domain,
 		cfg->irptndx = INVALID_IRPTNDX;
 	}
 
+	mutex_unlock(&smmu_domain->init_mutex);
+
+	/* Publish page table ops for map/unmap */
+	smmu_domain->pgtbl_ops = pgtbl_ops;
 	return 0;
 
+out_clear_smmu:
+	smmu_domain->smmu = NULL;
 out_unlock:
-	spin_unlock_irqrestore(&smmu_domain->lock, flags);
+	mutex_unlock(&smmu_domain->init_mutex);
 	return ret;
 }
 
@@ -999,23 +967,27 @@ static void arm_smmu_destroy_domain_context(struct iommu_domain *domain)
 	if (!smmu)
 		return;
 
-	/* Disable the context bank and nuke the TLB before freeing it. */
+	/*
+	 * Disable the context bank and free the page tables before freeing
+	 * it.
+	 */
 	cb_base = ARM_SMMU_CB_BASE(smmu) + ARM_SMMU_CB(smmu, cfg->cbndx);
 	writel_relaxed(0, cb_base + ARM_SMMU_CB_SCTLR);
-	arm_smmu_tlb_inv_context(smmu_domain);
 
 	if (cfg->irptndx != INVALID_IRPTNDX) {
 		irq = smmu->irqs[smmu->num_global_irqs + cfg->irptndx];
 		free_irq(irq, domain);
 	}
 
+	if (smmu_domain->pgtbl_ops)
+		free_io_pgtable_ops(smmu_domain->pgtbl_ops);
+
 	__arm_smmu_free_bitmap(smmu->context_map, cfg->cbndx);
 }
 
 static int arm_smmu_domain_init(struct iommu_domain *domain)
 {
 	struct arm_smmu_domain *smmu_domain;
-	pgd_t *pgd;
 
 	/*
 	 * Allocate the domain and initialise some of its data structures.
@@ -1026,81 +998,10 @@ static int arm_smmu_domain_init(struct iommu_domain *domain)
 	if (!smmu_domain)
 		return -ENOMEM;
 
-	pgd = kcalloc(PTRS_PER_PGD, sizeof(pgd_t), GFP_KERNEL);
-	if (!pgd)
-		goto out_free_domain;
-	smmu_domain->cfg.pgd = pgd;
-
-	spin_lock_init(&smmu_domain->lock);
+	mutex_init(&smmu_domain->init_mutex);
+	spin_lock_init(&smmu_domain->pgtbl_lock);
 	domain->priv = smmu_domain;
 	return 0;
-
-out_free_domain:
-	kfree(smmu_domain);
-	return -ENOMEM;
-}
-
-static void arm_smmu_free_ptes(pmd_t *pmd)
-{
-	pgtable_t table = pmd_pgtable(*pmd);
-
-	__free_page(table);
-}
-
-static void arm_smmu_free_pmds(pud_t *pud)
-{
-	int i;
-	pmd_t *pmd, *pmd_base = pmd_offset(pud, 0);
-
-	pmd = pmd_base;
-	for (i = 0; i < PTRS_PER_PMD; ++i) {
-		if (pmd_none(*pmd))
-			continue;
-
-		arm_smmu_free_ptes(pmd);
-		pmd++;
-	}
-
-	pmd_free(NULL, pmd_base);
-}
-
-static void arm_smmu_free_puds(pgd_t *pgd)
-{
-	int i;
-	pud_t *pud, *pud_base = pud_offset(pgd, 0);
-
-	pud = pud_base;
-	for (i = 0; i < PTRS_PER_PUD; ++i) {
-		if (pud_none(*pud))
-			continue;
-
-		arm_smmu_free_pmds(pud);
-		pud++;
-	}
-
-	pud_free(NULL, pud_base);
-}
-
-static void arm_smmu_free_pgtables(struct arm_smmu_domain *smmu_domain)
-{
-	int i;
-	struct arm_smmu_cfg *cfg = &smmu_domain->cfg;
-	pgd_t *pgd, *pgd_base = cfg->pgd;
-
-	/*
-	 * Recursively free the page tables for this domain. We don't
-	 * care about speculative TLB filling because the tables should
-	 * not be active in any context bank at this point (SCTLR.M is 0).
-	 */
-	pgd = pgd_base;
-	for (i = 0; i < PTRS_PER_PGD; ++i) {
-		if (pgd_none(*pgd))
-			continue;
-		arm_smmu_free_puds(pgd);
-		pgd++;
-	}
-
-	kfree(pgd_base);
 }
 
 static void arm_smmu_domain_destroy(struct iommu_domain *domain)
@@ -1112,7 +1013,6 @@ static void arm_smmu_domain_destroy(struct iommu_domain *domain)
 	 * already been detached.
 	 */
 	arm_smmu_destroy_domain_context(domain);
-	arm_smmu_free_pgtables(smmu_domain);
 	kfree(smmu_domain);
 }
 
@@ -1244,7 +1144,7 @@ static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev)
 {
 	int ret;
 	struct arm_smmu_domain *smmu_domain = domain->priv;
-	struct arm_smmu_device *smmu, *dom_smmu;
+	struct arm_smmu_device *smmu;
 	struct arm_smmu_master_cfg *cfg;
 
 	smmu = find_smmu_for_device(dev);
@@ -1258,21 +1158,16 @@ static int arm_smmu_attach_dev(struct iommu_domain *domain, struct device *dev)
 		return -EEXIST;
 	}
 
+	/* Ensure that the domain is finalised */
+	ret = arm_smmu_init_domain_context(domain, smmu);
+	if (IS_ERR_VALUE(ret))
+		return ret;
+
 	/*
 	 * Sanity check the domain. We don't support domains across
 	 * different SMMUs.
 	 */
-	dom_smmu = ACCESS_ONCE(smmu_domain->smmu);
-	if (!dom_smmu) {
-		/* Now that we have a master, we can finalise the domain */
-		ret = arm_smmu_init_domain_context(domain, smmu);
-		if (IS_ERR_VALUE(ret))
-			return ret;
-
-		dom_smmu = smmu_domain->smmu;
-	}
-
-	if (dom_smmu != smmu) {
+	if (smmu_domain->smmu != smmu) {
 		dev_err(dev,
 			"cannot attach to SMMU %s whilst already attached to domain on SMMU %s\n",
 			dev_name(smmu_domain->smmu->dev), dev_name(smmu->dev));
@@ -1303,293 +1198,103 @@ static void arm_smmu_detach_dev(struct iommu_domain *domain, struct device *dev)
 	arm_smmu_domain_remove_master(smmu_domain, cfg);
 }
 
-static bool arm_smmu_pte_is_contiguous_range(unsigned long addr,
-					     unsigned long end)
-{
-	return !(addr & ~ARM_SMMU_PTE_CONT_MASK) &&
-		(addr + ARM_SMMU_PTE_CONT_SIZE <= end);
-}
-
-static int arm_smmu_alloc_init_pte(struct arm_smmu_device *smmu, pmd_t *pmd,
-				   unsigned long addr, unsigned long end,
-				   unsigned long pfn, int prot, int stage)
-{
-	pte_t *pte, *start;
-	pteval_t pteval = ARM_SMMU_PTE_PAGE | ARM_SMMU_PTE_AF;
-
-	if (pmd_none(*pmd)) {
-		/* Allocate a new set of tables */
-		pgtable_t table = alloc_page(GFP_ATOMIC|__GFP_ZERO);
-
-		if (!table)
-			return -ENOMEM;
-
-		arm_smmu_flush_pgtable(smmu, page_address(table), PAGE_SIZE);
-		pmd_populate(NULL, pmd, table);
-		arm_smmu_flush_pgtable(smmu, pmd, sizeof(*pmd));
-	}
-
-	if (stage == 1) {
-		pteval |= ARM_SMMU_PTE_AP_UNPRIV | ARM_SMMU_PTE_nG;
-		if (!(prot & IOMMU_WRITE) && (prot & IOMMU_READ))
-			pteval |= ARM_SMMU_PTE_AP_RDONLY;
-
-		if (prot & IOMMU_CACHE)
-			pteval |= (MAIR_ATTR_IDX_CACHE <<
-				   ARM_SMMU_PTE_ATTRINDX_SHIFT);
-	} else {
-		pteval |= ARM_SMMU_PTE_HAP_FAULT;
-		if (prot & IOMMU_READ)
-			pteval |= ARM_SMMU_PTE_HAP_READ;
-		if (prot & IOMMU_WRITE)
-			pteval |= ARM_SMMU_PTE_HAP_WRITE;
-		if (prot & IOMMU_CACHE)
-			pteval |= ARM_SMMU_PTE_MEMATTR_OIWB;
-		else
-			pteval |= ARM_SMMU_PTE_MEMATTR_NC;
-	}
-
-	if (prot & IOMMU_NOEXEC)
-		pteval |= ARM_SMMU_PTE_XN;
-
-	/* If no access, create a faulting entry to avoid TLB fills */
-	if (!(prot & (IOMMU_READ | IOMMU_WRITE)))
-		pteval &= ~ARM_SMMU_PTE_PAGE;
-
-	pteval |= ARM_SMMU_PTE_SH_IS;
-	start = pmd_page_vaddr(*pmd) + pte_index(addr);
-	pte = start;
-
-	/*
-	 * Install the page table entries. This is fairly complicated
-	 * since we attempt to make use of the contiguous hint in the
-	 * ptes where possible. The contiguous hint indicates a series
-	 * of ARM_SMMU_PTE_CONT_ENTRIES ptes mapping a physically
-	 * contiguous region with the following constraints:
-	 *
-	 *   - The region start is aligned to ARM_SMMU_PTE_CONT_SIZE
-	 *   - Each pte in the region has the contiguous hint bit set
-	 *
-	 * This complicates unmapping (also handled by this code, when
-	 * neither IOMMU_READ or IOMMU_WRITE are set) because it is
-	 * possible, yet highly unlikely, that a client may unmap only
-	 * part of a contiguous range. This requires clearing of the
-	 * contiguous hint bits in the range before installing the new
-	 * faulting entries.
-	 *
-	 * Note that re-mapping an address range without first unmapping
-	 * it is not supported, so TLB invalidation is not required here
-	 * and is instead performed at unmap and domain-init time.
-	 */
-	do {
-		int i = 1;
-
-		pteval &= ~ARM_SMMU_PTE_CONT;
-
-		if (arm_smmu_pte_is_contiguous_range(addr, end)) {
-			i = ARM_SMMU_PTE_CONT_ENTRIES;
-			pteval |= ARM_SMMU_PTE_CONT;
-		} else if (pte_val(*pte) &
-			   (ARM_SMMU_PTE_CONT | ARM_SMMU_PTE_PAGE)) {
-			int j;
-			pte_t *cont_start;
-			unsigned long idx = pte_index(addr);
-
-			idx &= ~(ARM_SMMU_PTE_CONT_ENTRIES - 1);
-			cont_start = pmd_page_vaddr(*pmd) + idx;
-			for (j = 0; j < ARM_SMMU_PTE_CONT_ENTRIES; ++j)
-				pte_val(*(cont_start + j)) &=
-					~ARM_SMMU_PTE_CONT;
-
-			arm_smmu_flush_pgtable(smmu, cont_start,
-					       sizeof(*pte) *
-					       ARM_SMMU_PTE_CONT_ENTRIES);
-		}
-
-		do {
-			*pte = pfn_pte(pfn, __pgprot(pteval));
-		} while (pte++, pfn++, addr += PAGE_SIZE, --i);
-	} while (addr != end);
-
-	arm_smmu_flush_pgtable(smmu, start, sizeof(*pte) * (pte - start));
-	return 0;
-}
-
-static int arm_smmu_alloc_init_pmd(struct arm_smmu_device *smmu, pud_t *pud,
-				   unsigned long addr, unsigned long end,
-				   phys_addr_t phys, int prot, int stage)
+static int arm_smmu_map(struct iommu_domain *domain, unsigned long iova,
+			phys_addr_t paddr, size_t size, int prot)
 {
 	int ret;
-	pmd_t *pmd;
-	unsigned long next, pfn = __phys_to_pfn(phys);
-
-#ifndef __PAGETABLE_PMD_FOLDED
-	if (pud_none(*pud)) {
-		pmd = (pmd_t *)get_zeroed_page(GFP_ATOMIC);
-		if (!pmd)
-			return -ENOMEM;
-
-		arm_smmu_flush_pgtable(smmu, pmd, PAGE_SIZE);
-		pud_populate(NULL, pud, pmd);
-		arm_smmu_flush_pgtable(smmu, pud, sizeof(*pud));
-
-		pmd += pmd_index(addr);
-	} else
-#endif
-		pmd = pmd_offset(pud, addr);
+	unsigned long flags;
+	struct arm_smmu_domain *smmu_domain = domain->priv;
+	struct io_pgtable_ops *ops= smmu_domain->pgtbl_ops;
 
-	do {
-		next = pmd_addr_end(addr, end);
-		ret = arm_smmu_alloc_init_pte(smmu, pmd, addr, next, pfn,
-					      prot, stage);
-		phys += next - addr;
-		pfn = __phys_to_pfn(phys);
-	} while (pmd++, addr = next, addr < end);
+	if (!ops)
+		return -ENODEV;
 
+	spin_lock_irqsave(&smmu_domain->pgtbl_lock, flags);
+	ret = ops->map(ops, iova, paddr, size, prot);
+	spin_unlock_irqrestore(&smmu_domain->pgtbl_lock, flags);
 	return ret;
 }
 
-static int arm_smmu_alloc_init_pud(struct arm_smmu_device *smmu, pgd_t *pgd,
-				   unsigned long addr, unsigned long end,
-				   phys_addr_t phys, int prot, int stage)
+static size_t arm_smmu_unmap(struct iommu_domain *domain, unsigned long iova,
+			     size_t size)
 {
-	int ret = 0;
-	pud_t *pud;
-	unsigned long next;
-
-#ifndef __PAGETABLE_PUD_FOLDED
-	if (pgd_none(*pgd)) {
-		pud = (pud_t *)get_zeroed_page(GFP_ATOMIC);
-		if (!pud)
-			return -ENOMEM;
-
-		arm_smmu_flush_pgtable(smmu, pud, PAGE_SIZE);
-		pgd_populate(NULL, pgd, pud);
-		arm_smmu_flush_pgtable(smmu, pgd, sizeof(*pgd));
-
-		pud += pud_index(addr);
-	} else
-#endif
-		pud = pud_offset(pgd, addr);
+	size_t ret;
+	unsigned long flags;
+	struct arm_smmu_domain *smmu_domain = domain->priv;
+	struct io_pgtable_ops *ops= smmu_domain->pgtbl_ops;
 
-	do {
-		next = pud_addr_end(addr, end);
-		ret = arm_smmu_alloc_init_pmd(smmu, pud, addr, next, phys,
-					      prot, stage);
-		phys += next - addr;
-	} while (pud++, addr = next, addr < end);
+	if (!ops)
+		return 0;
 
+	spin_lock_irqsave(&smmu_domain->pgtbl_lock, flags);
+	ret = ops->unmap(ops, iova, size);
+	spin_unlock_irqrestore(&smmu_domain->pgtbl_lock, flags);
 	return ret;
 }
 
-static int arm_smmu_handle_mapping(struct arm_smmu_domain *smmu_domain,
-				   unsigned long iova, phys_addr_t paddr,
-				   size_t size, int prot)
+static phys_addr_t arm_smmu_iova_to_phys_hard(struct iommu_domain *domain,
+					      dma_addr_t iova)
 {
-	int ret, stage;
-	unsigned long end;
-	phys_addr_t input_mask, output_mask;
+	struct arm_smmu_domain *smmu_domain = domain->priv;
 	struct arm_smmu_device *smmu = smmu_domain->smmu;
 	struct arm_smmu_cfg *cfg = &smmu_domain->cfg;
-	pgd_t *pgd = cfg->pgd;
-	unsigned long flags;
+	struct io_pgtable_ops *ops= smmu_domain->pgtbl_ops;
+	struct device *dev = smmu->dev;
+	void __iomem *cb_base;
+	u32 tmp;
+	u64 phys;
 
-	if (cfg->cbar == CBAR_TYPE_S2_TRANS) {
-		stage = 2;
-		input_mask = (1ULL << smmu->s2_input_size) - 1;
-		output_mask = (1ULL << smmu->s2_output_size) - 1;
+	cb_base = ARM_SMMU_CB_BASE(smmu) + ARM_SMMU_CB(smmu, cfg->cbndx);
+
+	if (smmu->version == 1) {
+		u32 reg = iova & ~0xfff;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_ATS1PR_LO);
 	} else {
-		stage = 1;
-		input_mask = (1ULL << smmu->s1_input_size) - 1;
-		output_mask = (1ULL << smmu->s1_output_size) - 1;
+		u32 reg = iova & ~0xfff;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_ATS1PR_LO);
+		reg = (iova & ~0xfff) >> 32;
+		writel_relaxed(reg, cb_base + ARM_SMMU_CB_ATS1PR_HI);
 	}
 
-	if (!pgd)
-		return -EINVAL;
-
-	if (size & ~PAGE_MASK)
-		return -EINVAL;
-
-	if ((phys_addr_t)iova & ~input_mask)
-		return -ERANGE;
-
-	if (paddr & ~output_mask)
-		return -ERANGE;
-
-	spin_lock_irqsave(&smmu_domain->lock, flags);
-	pgd += pgd_index(iova);
-	end = iova + size;
-	do {
-		unsigned long next = pgd_addr_end(iova, end);
-
-		ret = arm_smmu_alloc_init_pud(smmu, pgd, iova, next, paddr,
-					      prot, stage);
-		if (ret)
-			goto out_unlock;
-
-		paddr += next - iova;
-		iova = next;
-	} while (pgd++, iova != end);
-
-out_unlock:
-	spin_unlock_irqrestore(&smmu_domain->lock, flags);
-
-	return ret;
-}
-
-static int arm_smmu_map(struct iommu_domain *domain, unsigned long iova,
-			phys_addr_t paddr, size_t size, int prot)
-{
-	struct arm_smmu_domain *smmu_domain = domain->priv;
-
-	if (!smmu_domain)
-		return -ENODEV;
+	if (readl_poll_timeout_atomic(cb_base + ARM_SMMU_CB_ATSR, tmp,
+				      !(tmp & ATSR_ACTIVE), 5, 50)) {
+		dev_err(dev,
+			"iova to phys timed out on 0x%pad. Falling back to software table walk.\n",
+			&iova);
+		return ops->iova_to_phys(ops, iova);
+	}
 
-	return arm_smmu_handle_mapping(smmu_domain, iova, paddr, size, prot);
-}
+	phys = readl_relaxed(cb_base + ARM_SMMU_CB_PAR_LO);
+	phys |= ((u64)readl_relaxed(cb_base + ARM_SMMU_CB_PAR_HI)) << 32;
 
-static size_t arm_smmu_unmap(struct iommu_domain *domain, unsigned long iova,
-			     size_t size)
-{
-	int ret;
-	struct arm_smmu_domain *smmu_domain = domain->priv;
+	if (phys & CB_PAR_F) {
+		dev_err(dev, "translation fault!\n");
+		dev_err(dev, "PAR = 0x%llx\n", phys);
+		return 0;
+	}
 
-	ret = arm_smmu_handle_mapping(smmu_domain, iova, 0, size, 0);
-	arm_smmu_tlb_inv_context(smmu_domain);
-	return ret ? 0 : size;
+	return (phys & GENMASK_ULL(39, 12)) | (iova & 0xfff);
 }
 
 static phys_addr_t arm_smmu_iova_to_phys(struct iommu_domain *domain,
-					 dma_addr_t iova)
+					dma_addr_t iova)
 {
-	pgd_t *pgdp, pgd;
-	pud_t pud;
-	pmd_t pmd;
-	pte_t pte;
+	phys_addr_t ret;
+	unsigned long flags;
 	struct arm_smmu_domain *smmu_domain = domain->priv;
-	struct arm_smmu_cfg *cfg = &smmu_domain->cfg;
+	struct io_pgtable_ops *ops= smmu_domain->pgtbl_ops;
 
-	pgdp = cfg->pgd;
-	if (!pgdp)
+	if (!ops)
 		return 0;
 
-	pgd = *(pgdp + pgd_index(iova));
-	if (pgd_none(pgd))
-		return 0;
-
-	pud = *pud_offset(&pgd, iova);
-	if (pud_none(pud))
-		return 0;
-
-	pmd = *pmd_offset(&pud, iova);
-	if (pmd_none(pmd))
-		return 0;
+	spin_lock_irqsave(&smmu_domain->pgtbl_lock, flags);
+	if (smmu_domain->smmu->features & ARM_SMMU_FEAT_TRANS_OPS)
+		ret = arm_smmu_iova_to_phys_hard(domain, iova);
+	else
+		ret = ops->iova_to_phys(ops, iova);
+	spin_unlock_irqrestore(&smmu_domain->pgtbl_lock, flags);
 
-	pte = *(pmd_page_vaddr(pmd) + pte_index(iova));
-	if (pte_none(pte))
-		return 0;
-
-	return __pfn_to_phys(pte_pfn(pte)) | (iova & ~PAGE_MASK);
+	return ret;
 }
 
 static bool arm_smmu_capable(enum iommu_cap cap)
@@ -1698,24 +1403,34 @@ static int arm_smmu_domain_get_attr(struct iommu_domain *domain,
 static int arm_smmu_domain_set_attr(struct iommu_domain *domain,
 				    enum iommu_attr attr, void *data)
 {
+	int ret = 0;
 	struct arm_smmu_domain *smmu_domain = domain->priv;
 
+	mutex_lock(&smmu_domain->init_mutex);
+
 	switch (attr) {
 	case DOMAIN_ATTR_NESTING:
-		if (smmu_domain->smmu)
-			return -EPERM;
+		if (smmu_domain->smmu) {
+			ret = -EPERM;
+			goto out_unlock;
+		}
+
 		if (*(int *)data)
 			smmu_domain->stage = ARM_SMMU_DOMAIN_NESTED;
 		else
 			smmu_domain->stage = ARM_SMMU_DOMAIN_S1;
 
-		return 0;
+		break;
 	default:
-		return -ENODEV;
+		ret = -ENODEV;
 	}
+
+out_unlock:
+	mutex_unlock(&smmu_domain->init_mutex);
+	return ret;
 }
 
-static const struct iommu_ops arm_smmu_ops = {
+static struct iommu_ops arm_smmu_ops = {
 	.capable		= arm_smmu_capable,
 	.domain_init		= arm_smmu_domain_init,
 	.domain_destroy		= arm_smmu_domain_destroy,
@@ -1729,9 +1444,7 @@ static const struct iommu_ops arm_smmu_ops = {
 	.remove_device		= arm_smmu_remove_device,
 	.domain_get_attr	= arm_smmu_domain_get_attr,
 	.domain_set_attr	= arm_smmu_domain_set_attr,
-	.pgsize_bitmap		= (SECTION_SIZE |
-				   ARM_SMMU_PTE_CONT_SIZE |
-				   PAGE_SIZE),
+	.pgsize_bitmap		= -1UL, /* Restricted during device attach */
 };
 
 static void arm_smmu_device_reset(struct arm_smmu_device *smmu)
@@ -1760,7 +1473,6 @@ static void arm_smmu_device_reset(struct arm_smmu_device *smmu)
 	}
 
 	/* Invalidate the TLB, just in case */
-	writel_relaxed(0, gr0_base + ARM_SMMU_GR0_STLBIALL);
 	writel_relaxed(0, gr0_base + ARM_SMMU_GR0_TLBIALLH);
 	writel_relaxed(0, gr0_base + ARM_SMMU_GR0_TLBIALLNSNH);
 
@@ -1782,7 +1494,7 @@ static void arm_smmu_device_reset(struct arm_smmu_device *smmu)
 	reg &= ~(sCR0_BSU_MASK << sCR0_BSU_SHIFT);
 
 	/* Push the button */
-	arm_smmu_tlb_sync(smmu);
+	__arm_smmu_tlb_sync(smmu);
 	writel(reg, ARM_SMMU_GR0_NS(smmu) + ARM_SMMU_GR0_sCR0);
 }
 
@@ -1816,12 +1528,6 @@ static int arm_smmu_device_cfg_probe(struct arm_smmu_device *smmu)
 
 	/* ID0 */
 	id = readl_relaxed(gr0_base + ARM_SMMU_GR0_ID0);
-#ifndef CONFIG_64BIT
-	if (((id >> ID0_PTFS_SHIFT) & ID0_PTFS_MASK) == ID0_PTFS_V8_ONLY) {
-		dev_err(smmu->dev, "\tno v7 descriptor support!\n");
-		return -ENODEV;
-	}
-#endif
 
 	/* Restrict available stages based on module parameter */
 	if (force_stage == 1)
@@ -1850,6 +1556,11 @@ static int arm_smmu_device_cfg_probe(struct arm_smmu_device *smmu)
 		return -ENODEV;
 	}
 
+	if (smmu->version == 1 || (!(id & ID0_ATOSNS) && (id & ID0_S1TS))) {
+		smmu->features |= ARM_SMMU_FEAT_TRANS_OPS;
+		dev_notice(smmu->dev, "\taddress translation ops\n");
+	}
+
 	if (id & ID0_CTTW) {
 		smmu->features |= ARM_SMMU_FEAT_COHERENT_WALK;
 		dev_notice(smmu->dev, "\tcoherent table walk\n");
@@ -1894,16 +1605,14 @@ static int arm_smmu_device_cfg_probe(struct arm_smmu_device *smmu)
 	smmu->pgshift = (id & ID1_PAGESIZE) ? 16 : 12;
 
 	/* Check for size mismatch of SMMU address space from mapped region */
-	size = 1 <<
-		(((id >> ID1_NUMPAGENDXB_SHIFT) & ID1_NUMPAGENDXB_MASK) + 1);
+	size = 1 << (((id >> ID1_NUMPAGENDXB_SHIFT) & ID1_NUMPAGENDXB_MASK) + 1);
 	size *= 2 << smmu->pgshift;
 	if (smmu->size != size)
 		dev_warn(smmu->dev,
 			"SMMU address space size (0x%lx) differs from mapped region size (0x%lx)!\n",
 			size, smmu->size);
 
-	smmu->num_s2_context_banks = (id >> ID1_NUMS2CB_SHIFT) &
-				      ID1_NUMS2CB_MASK;
+	smmu->num_s2_context_banks = (id >> ID1_NUMS2CB_SHIFT) & ID1_NUMS2CB_MASK;
 	smmu->num_context_banks = (id >> ID1_NUMCB_SHIFT) & ID1_NUMCB_MASK;
 	if (smmu->num_s2_context_banks > smmu->num_context_banks) {
 		dev_err(smmu->dev, "impossible number of S2 context banks!\n");
@@ -1915,46 +1624,40 @@ static int arm_smmu_device_cfg_probe(struct arm_smmu_device *smmu)
 	/* ID2 */
 	id = readl_relaxed(gr0_base + ARM_SMMU_GR0_ID2);
 	size = arm_smmu_id_size_to_bits((id >> ID2_IAS_SHIFT) & ID2_IAS_MASK);
-	smmu->s1_output_size = min_t(unsigned long, PHYS_MASK_SHIFT, size);
-
-	/* Stage-2 input size limited due to pgd allocation (PTRS_PER_PGD) */
-#ifdef CONFIG_64BIT
-	smmu->s2_input_size = min_t(unsigned long, VA_BITS, size);
-#else
-	smmu->s2_input_size = min(32UL, size);
-#endif
+	smmu->ipa_size = size;
 
-	/* The stage-2 output mask is also applied for bypass */
+	/* The output mask is also applied for bypass */
 	size = arm_smmu_id_size_to_bits((id >> ID2_OAS_SHIFT) & ID2_OAS_MASK);
-	smmu->s2_output_size = min_t(unsigned long, PHYS_MASK_SHIFT, size);
+	smmu->pa_size = size;
 
 	if (smmu->version == ARM_SMMU_V1) {
-		smmu->s1_input_size = 32;
+		smmu->va_size = smmu->ipa_size;
+		size = SZ_4K | SZ_2M | SZ_1G;
 	} else {
-#ifdef CONFIG_64BIT
 		size = (id >> ID2_UBS_SHIFT) & ID2_UBS_MASK;
-		size = min(VA_BITS, arm_smmu_id_size_to_bits(size));
-#else
-		size = 32;
+		smmu->va_size = arm_smmu_id_size_to_bits(size);
+#ifndef CONFIG_64BIT
+		smmu->va_size = min(32UL, smmu->va_size);
 #endif
-		smmu->s1_input_size = size;
-
-		if ((PAGE_SIZE == SZ_4K && !(id & ID2_PTFS_4K)) ||
-		    (PAGE_SIZE == SZ_64K && !(id & ID2_PTFS_64K)) ||
-		    (PAGE_SIZE != SZ_4K && PAGE_SIZE != SZ_64K)) {
-			dev_err(smmu->dev, "CPU page size 0x%lx unsupported\n",
-				PAGE_SIZE);
-			return -ENODEV;
-		}
+		size = 0;
+		if (id & ID2_PTFS_4K)
+			size |= SZ_4K | SZ_2M | SZ_1G;
+		if (id & ID2_PTFS_16K)
+			size |= SZ_16K | SZ_32M;
+		if (id & ID2_PTFS_64K)
+			size |= SZ_64K | SZ_512M;
 	}
 
+	arm_smmu_ops.pgsize_bitmap &= size;
+	dev_notice(smmu->dev, "\tSupported page sizes: 0x%08lx\n", size);
+
 	if (smmu->features & ARM_SMMU_FEAT_TRANS_S1)
 		dev_notice(smmu->dev, "\tStage-1: %lu-bit VA -> %lu-bit IPA\n",
-			   smmu->s1_input_size, smmu->s1_output_size);
+			   smmu->va_size, smmu->ipa_size);
 
 	if (smmu->features & ARM_SMMU_FEAT_TRANS_S2)
 		dev_notice(smmu->dev, "\tStage-2: %lu-bit IPA -> %lu-bit PA\n",
-			   smmu->s2_input_size, smmu->s2_output_size);
+			   smmu->ipa_size, smmu->pa_size);
 
 	return 0;
 }
diff --git a/drivers/iommu/io-pgtable-arm.c b/drivers/iommu/io-pgtable-arm.c
new file mode 100644
index 000000000000..5a500edf00cc
--- /dev/null
+++ b/drivers/iommu/io-pgtable-arm.c
@@ -0,0 +1,986 @@
+/*
+ * CPU-agnostic ARM page table allocator.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ * Copyright (C) 2014 ARM Limited
+ *
+ * Author: Will Deacon <will.deacon@arm.com>
+ */
+
+#define pr_fmt(fmt)	"arm-lpae io-pgtable: " fmt
+
+#include <linux/iommu.h>
+#include <linux/kernel.h>
+#include <linux/sizes.h>
+#include <linux/slab.h>
+#include <linux/types.h>
+
+#include "io-pgtable.h"
+
+#define ARM_LPAE_MAX_ADDR_BITS		48
+#define ARM_LPAE_S2_MAX_CONCAT_PAGES	16
+#define ARM_LPAE_MAX_LEVELS		4
+
+/* Struct accessors */
+#define io_pgtable_to_data(x)						\
+	container_of((x), struct arm_lpae_io_pgtable, iop)
+
+#define io_pgtable_ops_to_pgtable(x)					\
+	container_of((x), struct io_pgtable, ops)
+
+#define io_pgtable_ops_to_data(x)					\
+	io_pgtable_to_data(io_pgtable_ops_to_pgtable(x))
+
+/*
+ * For consistency with the architecture, we always consider
+ * ARM_LPAE_MAX_LEVELS levels, with the walk starting at level n >=0
+ */
+#define ARM_LPAE_START_LVL(d)		(ARM_LPAE_MAX_LEVELS - (d)->levels)
+
+/*
+ * Calculate the right shift amount to get to the portion describing level l
+ * in a virtual address mapped by the pagetable in d.
+ */
+#define ARM_LPAE_LVL_SHIFT(l,d)						\
+	((((d)->levels - ((l) - ARM_LPAE_START_LVL(d) + 1))		\
+	  * (d)->bits_per_level) + (d)->pg_shift)
+
+#define ARM_LPAE_PAGES_PER_PGD(d)	((d)->pgd_size >> (d)->pg_shift)
+
+/*
+ * Calculate the index at level l used to map virtual address a using the
+ * pagetable in d.
+ */
+#define ARM_LPAE_PGD_IDX(l,d)						\
+	((l) == ARM_LPAE_START_LVL(d) ? ilog2(ARM_LPAE_PAGES_PER_PGD(d)) : 0)
+
+#define ARM_LPAE_LVL_IDX(a,l,d)						\
+	(((a) >> ARM_LPAE_LVL_SHIFT(l,d)) &				\
+	 ((1 << ((d)->bits_per_level + ARM_LPAE_PGD_IDX(l,d))) - 1))
+
+/* Calculate the block/page mapping size at level l for pagetable in d. */
+#define ARM_LPAE_BLOCK_SIZE(l,d)					\
+	(1 << (ilog2(sizeof(arm_lpae_iopte)) +				\
+		((ARM_LPAE_MAX_LEVELS - (l)) * (d)->bits_per_level)))
+
+/* Page table bits */
+#define ARM_LPAE_PTE_TYPE_SHIFT		0
+#define ARM_LPAE_PTE_TYPE_MASK		0x3
+
+#define ARM_LPAE_PTE_TYPE_BLOCK		1
+#define ARM_LPAE_PTE_TYPE_TABLE		3
+#define ARM_LPAE_PTE_TYPE_PAGE		3
+
+#define ARM_LPAE_PTE_NSTABLE		(((arm_lpae_iopte)1) << 63)
+#define ARM_LPAE_PTE_XN			(((arm_lpae_iopte)3) << 53)
+#define ARM_LPAE_PTE_AF			(((arm_lpae_iopte)1) << 10)
+#define ARM_LPAE_PTE_SH_NS		(((arm_lpae_iopte)0) << 8)
+#define ARM_LPAE_PTE_SH_OS		(((arm_lpae_iopte)2) << 8)
+#define ARM_LPAE_PTE_SH_IS		(((arm_lpae_iopte)3) << 8)
+#define ARM_LPAE_PTE_NS			(((arm_lpae_iopte)1) << 5)
+#define ARM_LPAE_PTE_VALID		(((arm_lpae_iopte)1) << 0)
+
+#define ARM_LPAE_PTE_ATTR_LO_MASK	(((arm_lpae_iopte)0x3ff) << 2)
+/* Ignore the contiguous bit for block splitting */
+#define ARM_LPAE_PTE_ATTR_HI_MASK	(((arm_lpae_iopte)6) << 52)
+#define ARM_LPAE_PTE_ATTR_MASK		(ARM_LPAE_PTE_ATTR_LO_MASK |	\
+					 ARM_LPAE_PTE_ATTR_HI_MASK)
+
+/* Stage-1 PTE */
+#define ARM_LPAE_PTE_AP_UNPRIV		(((arm_lpae_iopte)1) << 6)
+#define ARM_LPAE_PTE_AP_RDONLY		(((arm_lpae_iopte)2) << 6)
+#define ARM_LPAE_PTE_ATTRINDX_SHIFT	2
+#define ARM_LPAE_PTE_nG			(((arm_lpae_iopte)1) << 11)
+
+/* Stage-2 PTE */
+#define ARM_LPAE_PTE_HAP_FAULT		(((arm_lpae_iopte)0) << 6)
+#define ARM_LPAE_PTE_HAP_READ		(((arm_lpae_iopte)1) << 6)
+#define ARM_LPAE_PTE_HAP_WRITE		(((arm_lpae_iopte)2) << 6)
+#define ARM_LPAE_PTE_MEMATTR_OIWB	(((arm_lpae_iopte)0xf) << 2)
+#define ARM_LPAE_PTE_MEMATTR_NC		(((arm_lpae_iopte)0x5) << 2)
+#define ARM_LPAE_PTE_MEMATTR_DEV	(((arm_lpae_iopte)0x1) << 2)
+
+/* Register bits */
+#define ARM_32_LPAE_TCR_EAE		(1 << 31)
+#define ARM_64_LPAE_S2_TCR_RES1		(1 << 31)
+
+#define ARM_LPAE_TCR_TG0_4K		(0 << 14)
+#define ARM_LPAE_TCR_TG0_64K		(1 << 14)
+#define ARM_LPAE_TCR_TG0_16K		(2 << 14)
+
+#define ARM_LPAE_TCR_SH0_SHIFT		12
+#define ARM_LPAE_TCR_SH0_MASK		0x3
+#define ARM_LPAE_TCR_SH_NS		0
+#define ARM_LPAE_TCR_SH_OS		2
+#define ARM_LPAE_TCR_SH_IS		3
+
+#define ARM_LPAE_TCR_ORGN0_SHIFT	10
+#define ARM_LPAE_TCR_IRGN0_SHIFT	8
+#define ARM_LPAE_TCR_RGN_MASK		0x3
+#define ARM_LPAE_TCR_RGN_NC		0
+#define ARM_LPAE_TCR_RGN_WBWA		1
+#define ARM_LPAE_TCR_RGN_WT		2
+#define ARM_LPAE_TCR_RGN_WB		3
+
+#define ARM_LPAE_TCR_SL0_SHIFT		6
+#define ARM_LPAE_TCR_SL0_MASK		0x3
+
+#define ARM_LPAE_TCR_T0SZ_SHIFT		0
+#define ARM_LPAE_TCR_SZ_MASK		0xf
+
+#define ARM_LPAE_TCR_PS_SHIFT		16
+#define ARM_LPAE_TCR_PS_MASK		0x7
+
+#define ARM_LPAE_TCR_IPS_SHIFT		32
+#define ARM_LPAE_TCR_IPS_MASK		0x7
+
+#define ARM_LPAE_TCR_PS_32_BIT		0x0ULL
+#define ARM_LPAE_TCR_PS_36_BIT		0x1ULL
+#define ARM_LPAE_TCR_PS_40_BIT		0x2ULL
+#define ARM_LPAE_TCR_PS_42_BIT		0x3ULL
+#define ARM_LPAE_TCR_PS_44_BIT		0x4ULL
+#define ARM_LPAE_TCR_PS_48_BIT		0x5ULL
+
+#define ARM_LPAE_MAIR_ATTR_SHIFT(n)	((n) << 3)
+#define ARM_LPAE_MAIR_ATTR_MASK		0xff
+#define ARM_LPAE_MAIR_ATTR_DEVICE	0x04
+#define ARM_LPAE_MAIR_ATTR_NC		0x44
+#define ARM_LPAE_MAIR_ATTR_WBRWA	0xff
+#define ARM_LPAE_MAIR_ATTR_IDX_NC	0
+#define ARM_LPAE_MAIR_ATTR_IDX_CACHE	1
+#define ARM_LPAE_MAIR_ATTR_IDX_DEV	2
+
+/* IOPTE accessors */
+#define iopte_deref(pte,d)					\
+	(__va((pte) & ((1ULL << ARM_LPAE_MAX_ADDR_BITS) - 1)	\
+	& ~((1ULL << (d)->pg_shift) - 1)))
+
+#define iopte_type(pte,l)					\
+	(((pte) >> ARM_LPAE_PTE_TYPE_SHIFT) & ARM_LPAE_PTE_TYPE_MASK)
+
+#define iopte_prot(pte)	((pte) & ARM_LPAE_PTE_ATTR_MASK)
+
+#define iopte_leaf(pte,l)					\
+	(l == (ARM_LPAE_MAX_LEVELS - 1) ?			\
+		(iopte_type(pte,l) == ARM_LPAE_PTE_TYPE_PAGE) :	\
+		(iopte_type(pte,l) == ARM_LPAE_PTE_TYPE_BLOCK))
+
+#define iopte_to_pfn(pte,d)					\
+	(((pte) & ((1ULL << ARM_LPAE_MAX_ADDR_BITS) - 1)) >> (d)->pg_shift)
+
+#define pfn_to_iopte(pfn,d)					\
+	(((pfn) << (d)->pg_shift) & ((1ULL << ARM_LPAE_MAX_ADDR_BITS) - 1))
+
+struct arm_lpae_io_pgtable {
+	struct io_pgtable	iop;
+
+	int			levels;
+	size_t			pgd_size;
+	unsigned long		pg_shift;
+	unsigned long		bits_per_level;
+
+	void			*pgd;
+};
+
+typedef u64 arm_lpae_iopte;
+
+static bool selftest_running = false;
+
+static int arm_lpae_init_pte(struct arm_lpae_io_pgtable *data,
+			     unsigned long iova, phys_addr_t paddr,
+			     arm_lpae_iopte prot, int lvl,
+			     arm_lpae_iopte *ptep)
+{
+	arm_lpae_iopte pte = prot;
+
+	/* We require an unmap first */
+	if (iopte_leaf(*ptep, lvl)) {
+		WARN_ON(!selftest_running);
+		return -EEXIST;
+	}
+
+	if (data->iop.cfg.quirks & IO_PGTABLE_QUIRK_ARM_NS)
+		pte |= ARM_LPAE_PTE_NS;
+
+	if (lvl == ARM_LPAE_MAX_LEVELS - 1)
+		pte |= ARM_LPAE_PTE_TYPE_PAGE;
+	else
+		pte |= ARM_LPAE_PTE_TYPE_BLOCK;
+
+	pte |= ARM_LPAE_PTE_AF | ARM_LPAE_PTE_SH_IS;
+	pte |= pfn_to_iopte(paddr >> data->pg_shift, data);
+
+	*ptep = pte;
+	data->iop.cfg.tlb->flush_pgtable(ptep, sizeof(*ptep), data->iop.cookie);
+	return 0;
+}
+
+static int __arm_lpae_map(struct arm_lpae_io_pgtable *data, unsigned long iova,
+			  phys_addr_t paddr, size_t size, arm_lpae_iopte prot,
+			  int lvl, arm_lpae_iopte *ptep)
+{
+	arm_lpae_iopte *cptep, pte;
+	void *cookie = data->iop.cookie;
+	size_t block_size = ARM_LPAE_BLOCK_SIZE(lvl, data);
+
+	/* Find our entry at the current level */
+	ptep += ARM_LPAE_LVL_IDX(iova, lvl, data);
+
+	/* If we can install a leaf entry at this level, then do so */
+	if (size == block_size && (size & data->iop.cfg.pgsize_bitmap))
+		return arm_lpae_init_pte(data, iova, paddr, prot, lvl, ptep);
+
+	/* We can't allocate tables at the final level */
+	if (WARN_ON(lvl >= ARM_LPAE_MAX_LEVELS - 1))
+		return -EINVAL;
+
+	/* Grab a pointer to the next level */
+	pte = *ptep;
+	if (!pte) {
+		cptep = alloc_pages_exact(1UL << data->pg_shift,
+					 GFP_ATOMIC | __GFP_ZERO);
+		if (!cptep)
+			return -ENOMEM;
+
+		data->iop.cfg.tlb->flush_pgtable(cptep, 1UL << data->pg_shift,
+						 cookie);
+		pte = __pa(cptep) | ARM_LPAE_PTE_TYPE_TABLE;
+		if (data->iop.cfg.quirks & IO_PGTABLE_QUIRK_ARM_NS)
+			pte |= ARM_LPAE_PTE_NSTABLE;
+		*ptep = pte;
+		data->iop.cfg.tlb->flush_pgtable(ptep, sizeof(*ptep), cookie);
+	} else {
+		cptep = iopte_deref(pte, data);
+	}
+
+	/* Rinse, repeat */
+	return __arm_lpae_map(data, iova, paddr, size, prot, lvl + 1, cptep);
+}
+
+static arm_lpae_iopte arm_lpae_prot_to_pte(struct arm_lpae_io_pgtable *data,
+					   int prot)
+{
+	arm_lpae_iopte pte;
+
+	if (data->iop.fmt == ARM_64_LPAE_S1 ||
+	    data->iop.fmt == ARM_32_LPAE_S1) {
+		pte = ARM_LPAE_PTE_AP_UNPRIV | ARM_LPAE_PTE_nG;
+
+		if (!(prot & IOMMU_WRITE) && (prot & IOMMU_READ))
+			pte |= ARM_LPAE_PTE_AP_RDONLY;
+
+		if (prot & IOMMU_CACHE)
+			pte |= (ARM_LPAE_MAIR_ATTR_IDX_CACHE
+				<< ARM_LPAE_PTE_ATTRINDX_SHIFT);
+	} else {
+		pte = ARM_LPAE_PTE_HAP_FAULT;
+		if (prot & IOMMU_READ)
+			pte |= ARM_LPAE_PTE_HAP_READ;
+		if (prot & IOMMU_WRITE)
+			pte |= ARM_LPAE_PTE_HAP_WRITE;
+		if (prot & IOMMU_CACHE)
+			pte |= ARM_LPAE_PTE_MEMATTR_OIWB;
+		else
+			pte |= ARM_LPAE_PTE_MEMATTR_NC;
+	}
+
+	if (prot & IOMMU_NOEXEC)
+		pte |= ARM_LPAE_PTE_XN;
+
+	return pte;
+}
+
+static int arm_lpae_map(struct io_pgtable_ops *ops, unsigned long iova,
+			phys_addr_t paddr, size_t size, int iommu_prot)
+{
+	struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(ops);
+	arm_lpae_iopte *ptep = data->pgd;
+	int lvl = ARM_LPAE_START_LVL(data);
+	arm_lpae_iopte prot;
+
+	/* If no access, then nothing to do */
+	if (!(iommu_prot & (IOMMU_READ | IOMMU_WRITE)))
+		return 0;
+
+	prot = arm_lpae_prot_to_pte(data, iommu_prot);
+	return __arm_lpae_map(data, iova, paddr, size, prot, lvl, ptep);
+}
+
+static void __arm_lpae_free_pgtable(struct arm_lpae_io_pgtable *data, int lvl,
+				    arm_lpae_iopte *ptep)
+{
+	arm_lpae_iopte *start, *end;
+	unsigned long table_size;
+
+	/* Only leaf entries at the last level */
+	if (lvl == ARM_LPAE_MAX_LEVELS - 1)
+		return;
+
+	if (lvl == ARM_LPAE_START_LVL(data))
+		table_size = data->pgd_size;
+	else
+		table_size = 1UL << data->pg_shift;
+
+	start = ptep;
+	end = (void *)ptep + table_size;
+
+	while (ptep != end) {
+		arm_lpae_iopte pte = *ptep++;
+
+		if (!pte || iopte_leaf(pte, lvl))
+			continue;
+
+		__arm_lpae_free_pgtable(data, lvl + 1, iopte_deref(pte, data));
+	}
+
+	free_pages_exact(start, table_size);
+}
+
+static void arm_lpae_free_pgtable(struct io_pgtable *iop)
+{
+	struct arm_lpae_io_pgtable *data = io_pgtable_to_data(iop);
+
+	__arm_lpae_free_pgtable(data, ARM_LPAE_START_LVL(data), data->pgd);
+	kfree(data);
+}
+
+static int arm_lpae_split_blk_unmap(struct arm_lpae_io_pgtable *data,
+				    unsigned long iova, size_t size,
+				    arm_lpae_iopte prot, int lvl,
+				    arm_lpae_iopte *ptep, size_t blk_size)
+{
+	unsigned long blk_start, blk_end;
+	phys_addr_t blk_paddr;
+	arm_lpae_iopte table = 0;
+	void *cookie = data->iop.cookie;
+	const struct iommu_gather_ops *tlb = data->iop.cfg.tlb;
+
+	blk_start = iova & ~(blk_size - 1);
+	blk_end = blk_start + blk_size;
+	blk_paddr = iopte_to_pfn(*ptep, data) << data->pg_shift;
+
+	for (; blk_start < blk_end; blk_start += size, blk_paddr += size) {
+		arm_lpae_iopte *tablep;
+
+		/* Unmap! */
+		if (blk_start == iova)
+			continue;
+
+		/* __arm_lpae_map expects a pointer to the start of the table */
+		tablep = &table - ARM_LPAE_LVL_IDX(blk_start, lvl, data);
+		if (__arm_lpae_map(data, blk_start, blk_paddr, size, prot, lvl,
+				   tablep) < 0) {
+			if (table) {
+				/* Free the table we allocated */
+				tablep = iopte_deref(table, data);
+				__arm_lpae_free_pgtable(data, lvl + 1, tablep);
+			}
+			return 0; /* Bytes unmapped */
+		}
+	}
+
+	*ptep = table;
+	tlb->flush_pgtable(ptep, sizeof(*ptep), cookie);
+	iova &= ~(blk_size - 1);
+	tlb->tlb_add_flush(iova, blk_size, true, cookie);
+	return size;
+}
+
+static int __arm_lpae_unmap(struct arm_lpae_io_pgtable *data,
+			    unsigned long iova, size_t size, int lvl,
+			    arm_lpae_iopte *ptep)
+{
+	arm_lpae_iopte pte;
+	const struct iommu_gather_ops *tlb = data->iop.cfg.tlb;
+	void *cookie = data->iop.cookie;
+	size_t blk_size = ARM_LPAE_BLOCK_SIZE(lvl, data);
+
+	ptep += ARM_LPAE_LVL_IDX(iova, lvl, data);
+	pte = *ptep;
+
+	/* Something went horribly wrong and we ran out of page table */
+	if (WARN_ON(!pte || (lvl == ARM_LPAE_MAX_LEVELS)))
+		return 0;
+
+	/* If the size matches this level, we're in the right place */
+	if (size == blk_size) {
+		*ptep = 0;
+		tlb->flush_pgtable(ptep, sizeof(*ptep), cookie);
+
+		if (!iopte_leaf(pte, lvl)) {
+			/* Also flush any partial walks */
+			tlb->tlb_add_flush(iova, size, false, cookie);
+			tlb->tlb_sync(data->iop.cookie);
+			ptep = iopte_deref(pte, data);
+			__arm_lpae_free_pgtable(data, lvl + 1, ptep);
+		} else {
+			tlb->tlb_add_flush(iova, size, true, cookie);
+		}
+
+		return size;
+	} else if (iopte_leaf(pte, lvl)) {
+		/*
+		 * Insert a table at the next level to map the old region,
+		 * minus the part we want to unmap
+		 */
+		return arm_lpae_split_blk_unmap(data, iova, size,
+						iopte_prot(pte), lvl, ptep,
+						blk_size);
+	}
+
+	/* Keep on walkin' */
+	ptep = iopte_deref(pte, data);
+	return __arm_lpae_unmap(data, iova, size, lvl + 1, ptep);
+}
+
+static int arm_lpae_unmap(struct io_pgtable_ops *ops, unsigned long iova,
+			  size_t size)
+{
+	size_t unmapped;
+	struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(ops);
+	struct io_pgtable *iop = &data->iop;
+	arm_lpae_iopte *ptep = data->pgd;
+	int lvl = ARM_LPAE_START_LVL(data);
+
+	unmapped = __arm_lpae_unmap(data, iova, size, lvl, ptep);
+	if (unmapped)
+		iop->cfg.tlb->tlb_sync(iop->cookie);
+
+	return unmapped;
+}
+
+static phys_addr_t arm_lpae_iova_to_phys(struct io_pgtable_ops *ops,
+					 unsigned long iova)
+{
+	struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(ops);
+	arm_lpae_iopte pte, *ptep = data->pgd;
+	int lvl = ARM_LPAE_START_LVL(data);
+
+	do {
+		/* Valid IOPTE pointer? */
+		if (!ptep)
+			return 0;
+
+		/* Grab the IOPTE we're interested in */
+		pte = *(ptep + ARM_LPAE_LVL_IDX(iova, lvl, data));
+
+		/* Valid entry? */
+		if (!pte)
+			return 0;
+
+		/* Leaf entry? */
+		if (iopte_leaf(pte,lvl))
+			goto found_translation;
+
+		/* Take it to the next level */
+		ptep = iopte_deref(pte, data);
+	} while (++lvl < ARM_LPAE_MAX_LEVELS);
+
+	/* Ran out of page tables to walk */
+	return 0;
+
+found_translation:
+	iova &= ((1 << data->pg_shift) - 1);
+	return ((phys_addr_t)iopte_to_pfn(pte,data) << data->pg_shift) | iova;
+}
+
+static void arm_lpae_restrict_pgsizes(struct io_pgtable_cfg *cfg)
+{
+	unsigned long granule;
+
+	/*
+	 * We need to restrict the supported page sizes to match the
+	 * translation regime for a particular granule. Aim to match
+	 * the CPU page size if possible, otherwise prefer smaller sizes.
+	 * While we're at it, restrict the block sizes to match the
+	 * chosen granule.
+	 */
+	if (cfg->pgsize_bitmap & PAGE_SIZE)
+		granule = PAGE_SIZE;
+	else if (cfg->pgsize_bitmap & ~PAGE_MASK)
+		granule = 1UL << __fls(cfg->pgsize_bitmap & ~PAGE_MASK);
+	else if (cfg->pgsize_bitmap & PAGE_MASK)
+		granule = 1UL << __ffs(cfg->pgsize_bitmap & PAGE_MASK);
+	else
+		granule = 0;
+
+	switch (granule) {
+	case SZ_4K:
+		cfg->pgsize_bitmap &= (SZ_4K | SZ_2M | SZ_1G);
+		break;
+	case SZ_16K:
+		cfg->pgsize_bitmap &= (SZ_16K | SZ_32M);
+		break;
+	case SZ_64K:
+		cfg->pgsize_bitmap &= (SZ_64K | SZ_512M);
+		break;
+	default:
+		cfg->pgsize_bitmap = 0;
+	}
+}
+
+static struct arm_lpae_io_pgtable *
+arm_lpae_alloc_pgtable(struct io_pgtable_cfg *cfg)
+{
+	unsigned long va_bits, pgd_bits;
+	struct arm_lpae_io_pgtable *data;
+
+	arm_lpae_restrict_pgsizes(cfg);
+
+	if (!(cfg->pgsize_bitmap & (SZ_4K | SZ_16K | SZ_64K)))
+		return NULL;
+
+	if (cfg->ias > ARM_LPAE_MAX_ADDR_BITS)
+		return NULL;
+
+	if (cfg->oas > ARM_LPAE_MAX_ADDR_BITS)
+		return NULL;
+
+	data = kmalloc(sizeof(*data), GFP_KERNEL);
+	if (!data)
+		return NULL;
+
+	data->pg_shift = __ffs(cfg->pgsize_bitmap);
+	data->bits_per_level = data->pg_shift - ilog2(sizeof(arm_lpae_iopte));
+
+	va_bits = cfg->ias - data->pg_shift;
+	data->levels = DIV_ROUND_UP(va_bits, data->bits_per_level);
+
+	/* Calculate the actual size of our pgd (without concatenation) */
+	pgd_bits = va_bits - (data->bits_per_level * (data->levels - 1));
+	data->pgd_size = 1UL << (pgd_bits + ilog2(sizeof(arm_lpae_iopte)));
+
+	data->iop.ops = (struct io_pgtable_ops) {
+		.map		= arm_lpae_map,
+		.unmap		= arm_lpae_unmap,
+		.iova_to_phys	= arm_lpae_iova_to_phys,
+	};
+
+	return data;
+}
+
+static struct io_pgtable *
+arm_64_lpae_alloc_pgtable_s1(struct io_pgtable_cfg *cfg, void *cookie)
+{
+	u64 reg;
+	struct arm_lpae_io_pgtable *data = arm_lpae_alloc_pgtable(cfg);
+
+	if (!data)
+		return NULL;
+
+	/* TCR */
+	reg = (ARM_LPAE_TCR_SH_IS << ARM_LPAE_TCR_SH0_SHIFT) |
+	      (ARM_LPAE_TCR_RGN_WBWA << ARM_LPAE_TCR_IRGN0_SHIFT) |
+	      (ARM_LPAE_TCR_RGN_WBWA << ARM_LPAE_TCR_ORGN0_SHIFT);
+
+	switch (1 << data->pg_shift) {
+	case SZ_4K:
+		reg |= ARM_LPAE_TCR_TG0_4K;
+		break;
+	case SZ_16K:
+		reg |= ARM_LPAE_TCR_TG0_16K;
+		break;
+	case SZ_64K:
+		reg |= ARM_LPAE_TCR_TG0_64K;
+		break;
+	}
+
+	switch (cfg->oas) {
+	case 32:
+		reg |= (ARM_LPAE_TCR_PS_32_BIT << ARM_LPAE_TCR_IPS_SHIFT);
+		break;
+	case 36:
+		reg |= (ARM_LPAE_TCR_PS_36_BIT << ARM_LPAE_TCR_IPS_SHIFT);
+		break;
+	case 40:
+		reg |= (ARM_LPAE_TCR_PS_40_BIT << ARM_LPAE_TCR_IPS_SHIFT);
+		break;
+	case 42:
+		reg |= (ARM_LPAE_TCR_PS_42_BIT << ARM_LPAE_TCR_IPS_SHIFT);
+		break;
+	case 44:
+		reg |= (ARM_LPAE_TCR_PS_44_BIT << ARM_LPAE_TCR_IPS_SHIFT);
+		break;
+	case 48:
+		reg |= (ARM_LPAE_TCR_PS_48_BIT << ARM_LPAE_TCR_IPS_SHIFT);
+		break;
+	default:
+		goto out_free_data;
+	}
+
+	reg |= (64ULL - cfg->ias) << ARM_LPAE_TCR_T0SZ_SHIFT;
+	cfg->arm_lpae_s1_cfg.tcr = reg;
+
+	/* MAIRs */
+	reg = (ARM_LPAE_MAIR_ATTR_NC
+	       << ARM_LPAE_MAIR_ATTR_SHIFT(ARM_LPAE_MAIR_ATTR_IDX_NC)) |
+	      (ARM_LPAE_MAIR_ATTR_WBRWA
+	       << ARM_LPAE_MAIR_ATTR_SHIFT(ARM_LPAE_MAIR_ATTR_IDX_CACHE)) |
+	      (ARM_LPAE_MAIR_ATTR_DEVICE
+	       << ARM_LPAE_MAIR_ATTR_SHIFT(ARM_LPAE_MAIR_ATTR_IDX_DEV));
+
+	cfg->arm_lpae_s1_cfg.mair[0] = reg;
+	cfg->arm_lpae_s1_cfg.mair[1] = 0;
+
+	/* Looking good; allocate a pgd */
+	data->pgd = alloc_pages_exact(data->pgd_size, GFP_KERNEL | __GFP_ZERO);
+	if (!data->pgd)
+		goto out_free_data;
+
+	cfg->tlb->flush_pgtable(data->pgd, data->pgd_size, cookie);
+
+	/* TTBRs */
+	cfg->arm_lpae_s1_cfg.ttbr[0] = virt_to_phys(data->pgd);
+	cfg->arm_lpae_s1_cfg.ttbr[1] = 0;
+	return &data->iop;
+
+out_free_data:
+	kfree(data);
+	return NULL;
+}
+
+static struct io_pgtable *
+arm_64_lpae_alloc_pgtable_s2(struct io_pgtable_cfg *cfg, void *cookie)
+{
+	u64 reg, sl;
+	struct arm_lpae_io_pgtable *data = arm_lpae_alloc_pgtable(cfg);
+
+	if (!data)
+		return NULL;
+
+	/*
+	 * Concatenate PGDs at level 1 if possible in order to reduce
+	 * the depth of the stage-2 walk.
+	 */
+	if (data->levels == ARM_LPAE_MAX_LEVELS) {
+		unsigned long pgd_pages;
+
+		pgd_pages = data->pgd_size >> ilog2(sizeof(arm_lpae_iopte));
+		if (pgd_pages <= ARM_LPAE_S2_MAX_CONCAT_PAGES) {
+			data->pgd_size = pgd_pages << data->pg_shift;
+			data->levels--;
+		}
+	}
+
+	/* VTCR */
+	reg = ARM_64_LPAE_S2_TCR_RES1 |
+	     (ARM_LPAE_TCR_SH_IS << ARM_LPAE_TCR_SH0_SHIFT) |
+	     (ARM_LPAE_TCR_RGN_WBWA << ARM_LPAE_TCR_IRGN0_SHIFT) |
+	     (ARM_LPAE_TCR_RGN_WBWA << ARM_LPAE_TCR_ORGN0_SHIFT);
+
+	sl = ARM_LPAE_START_LVL(data);
+
+	switch (1 << data->pg_shift) {
+	case SZ_4K:
+		reg |= ARM_LPAE_TCR_TG0_4K;
+		sl++; /* SL0 format is different for 4K granule size */
+		break;
+	case SZ_16K:
+		reg |= ARM_LPAE_TCR_TG0_16K;
+		break;
+	case SZ_64K:
+		reg |= ARM_LPAE_TCR_TG0_64K;
+		break;
+	}
+
+	switch (cfg->oas) {
+	case 32:
+		reg |= (ARM_LPAE_TCR_PS_32_BIT << ARM_LPAE_TCR_PS_SHIFT);
+		break;
+	case 36:
+		reg |= (ARM_LPAE_TCR_PS_36_BIT << ARM_LPAE_TCR_PS_SHIFT);
+		break;
+	case 40:
+		reg |= (ARM_LPAE_TCR_PS_40_BIT << ARM_LPAE_TCR_PS_SHIFT);
+		break;
+	case 42:
+		reg |= (ARM_LPAE_TCR_PS_42_BIT << ARM_LPAE_TCR_PS_SHIFT);
+		break;
+	case 44:
+		reg |= (ARM_LPAE_TCR_PS_44_BIT << ARM_LPAE_TCR_PS_SHIFT);
+		break;
+	case 48:
+		reg |= (ARM_LPAE_TCR_PS_48_BIT << ARM_LPAE_TCR_PS_SHIFT);
+		break;
+	default:
+		goto out_free_data;
+	}
+
+	reg |= (64ULL - cfg->ias) << ARM_LPAE_TCR_T0SZ_SHIFT;
+	reg |= (~sl & ARM_LPAE_TCR_SL0_MASK) << ARM_LPAE_TCR_SL0_SHIFT;
+	cfg->arm_lpae_s2_cfg.vtcr = reg;
+
+	/* Allocate pgd pages */
+	data->pgd = alloc_pages_exact(data->pgd_size, GFP_KERNEL | __GFP_ZERO);
+	if (!data->pgd)
+		goto out_free_data;
+
+	cfg->tlb->flush_pgtable(data->pgd, data->pgd_size, cookie);
+
+	/* VTTBR */
+	cfg->arm_lpae_s2_cfg.vttbr = virt_to_phys(data->pgd);
+	return &data->iop;
+
+out_free_data:
+	kfree(data);
+	return NULL;
+}
+
+static struct io_pgtable *
+arm_32_lpae_alloc_pgtable_s1(struct io_pgtable_cfg *cfg, void *cookie)
+{
+	struct io_pgtable *iop;
+
+	if (cfg->ias > 32 || cfg->oas > 40)
+		return NULL;
+
+	cfg->pgsize_bitmap &= (SZ_4K | SZ_2M | SZ_1G);
+	iop = arm_64_lpae_alloc_pgtable_s1(cfg, cookie);
+	if (iop) {
+		cfg->arm_lpae_s1_cfg.tcr |= ARM_32_LPAE_TCR_EAE;
+		cfg->arm_lpae_s1_cfg.tcr &= 0xffffffff;
+	}
+
+	return iop;
+}
+
+static struct io_pgtable *
+arm_32_lpae_alloc_pgtable_s2(struct io_pgtable_cfg *cfg, void *cookie)
+{
+	struct io_pgtable *iop;
+
+	if (cfg->ias > 40 || cfg->oas > 40)
+		return NULL;
+
+	cfg->pgsize_bitmap &= (SZ_4K | SZ_2M | SZ_1G);
+	iop = arm_64_lpae_alloc_pgtable_s2(cfg, cookie);
+	if (iop)
+		cfg->arm_lpae_s2_cfg.vtcr &= 0xffffffff;
+
+	return iop;
+}
+
+struct io_pgtable_init_fns io_pgtable_arm_64_lpae_s1_init_fns = {
+	.alloc	= arm_64_lpae_alloc_pgtable_s1,
+	.free	= arm_lpae_free_pgtable,
+};
+
+struct io_pgtable_init_fns io_pgtable_arm_64_lpae_s2_init_fns = {
+	.alloc	= arm_64_lpae_alloc_pgtable_s2,
+	.free	= arm_lpae_free_pgtable,
+};
+
+struct io_pgtable_init_fns io_pgtable_arm_32_lpae_s1_init_fns = {
+	.alloc	= arm_32_lpae_alloc_pgtable_s1,
+	.free	= arm_lpae_free_pgtable,
+};
+
+struct io_pgtable_init_fns io_pgtable_arm_32_lpae_s2_init_fns = {
+	.alloc	= arm_32_lpae_alloc_pgtable_s2,
+	.free	= arm_lpae_free_pgtable,
+};
+
+#ifdef CONFIG_IOMMU_IO_PGTABLE_LPAE_SELFTEST
+
+static struct io_pgtable_cfg *cfg_cookie;
+
+static void dummy_tlb_flush_all(void *cookie)
+{
+	WARN_ON(cookie != cfg_cookie);
+}
+
+static void dummy_tlb_add_flush(unsigned long iova, size_t size, bool leaf,
+				void *cookie)
+{
+	WARN_ON(cookie != cfg_cookie);
+	WARN_ON(!(size & cfg_cookie->pgsize_bitmap));
+}
+
+static void dummy_tlb_sync(void *cookie)
+{
+	WARN_ON(cookie != cfg_cookie);
+}
+
+static void dummy_flush_pgtable(void *ptr, size_t size, void *cookie)
+{
+	WARN_ON(cookie != cfg_cookie);
+}
+
+static struct iommu_gather_ops dummy_tlb_ops __initdata = {
+	.tlb_flush_all	= dummy_tlb_flush_all,
+	.tlb_add_flush	= dummy_tlb_add_flush,
+	.tlb_sync	= dummy_tlb_sync,
+	.flush_pgtable	= dummy_flush_pgtable,
+};
+
+static void __init arm_lpae_dump_ops(struct io_pgtable_ops *ops)
+{
+	struct arm_lpae_io_pgtable *data = io_pgtable_ops_to_data(ops);
+	struct io_pgtable_cfg *cfg = &data->iop.cfg;
+
+	pr_err("cfg: pgsize_bitmap 0x%lx, ias %u-bit\n",
+		cfg->pgsize_bitmap, cfg->ias);
+	pr_err("data: %d levels, 0x%zx pgd_size, %lu pg_shift, %lu bits_per_level, pgd @ %p\n",
+		data->levels, data->pgd_size, data->pg_shift,
+		data->bits_per_level, data->pgd);
+}
+
+#define __FAIL(ops, i)	({						\
+		WARN(1, "selftest: test failed for fmt idx %d\n", (i));	\
+		arm_lpae_dump_ops(ops);					\
+		selftest_running = false;				\
+		-EFAULT;						\
+})
+
+static int __init arm_lpae_run_tests(struct io_pgtable_cfg *cfg)
+{
+	static const enum io_pgtable_fmt fmts[] = {
+		ARM_64_LPAE_S1,
+		ARM_64_LPAE_S2,
+	};
+
+	int i, j;
+	unsigned long iova;
+	size_t size;
+	struct io_pgtable_ops *ops;
+
+	selftest_running = true;
+
+	for (i = 0; i < ARRAY_SIZE(fmts); ++i) {
+		cfg_cookie = cfg;
+		ops = alloc_io_pgtable_ops(fmts[i], cfg, cfg);
+		if (!ops) {
+			pr_err("selftest: failed to allocate io pgtable ops\n");
+			return -ENOMEM;
+		}
+
+		/*
+		 * Initial sanity checks.
+		 * Empty page tables shouldn't provide any translations.
+		 */
+		if (ops->iova_to_phys(ops, 42))
+			return __FAIL(ops, i);
+
+		if (ops->iova_to_phys(ops, SZ_1G + 42))
+			return __FAIL(ops, i);
+
+		if (ops->iova_to_phys(ops, SZ_2G + 42))
+			return __FAIL(ops, i);
+
+		/*
+		 * Distinct mappings of different granule sizes.
+		 */
+		iova = 0;
+		j = find_first_bit(&cfg->pgsize_bitmap, BITS_PER_LONG);
+		while (j != BITS_PER_LONG) {
+			size = 1UL << j;
+
+			if (ops->map(ops, iova, iova, size, IOMMU_READ |
+							    IOMMU_WRITE |
+							    IOMMU_NOEXEC |
+							    IOMMU_CACHE))
+				return __FAIL(ops, i);
+
+			/* Overlapping mappings */
+			if (!ops->map(ops, iova, iova + size, size,
+				      IOMMU_READ | IOMMU_NOEXEC))
+				return __FAIL(ops, i);
+
+			if (ops->iova_to_phys(ops, iova + 42) != (iova + 42))
+				return __FAIL(ops, i);
+
+			iova += SZ_1G;
+			j++;
+			j = find_next_bit(&cfg->pgsize_bitmap, BITS_PER_LONG, j);
+		}
+
+		/* Partial unmap */
+		size = 1UL << __ffs(cfg->pgsize_bitmap);
+		if (ops->unmap(ops, SZ_1G + size, size) != size)
+			return __FAIL(ops, i);
+
+		/* Remap of partial unmap */
+		if (ops->map(ops, SZ_1G + size, size, size, IOMMU_READ))
+			return __FAIL(ops, i);
+
+		if (ops->iova_to_phys(ops, SZ_1G + size + 42) != (size + 42))
+			return __FAIL(ops, i);
+
+		/* Full unmap */
+		iova = 0;
+		j = find_first_bit(&cfg->pgsize_bitmap, BITS_PER_LONG);
+		while (j != BITS_PER_LONG) {
+			size = 1UL << j;
+
+			if (ops->unmap(ops, iova, size) != size)
+				return __FAIL(ops, i);
+
+			if (ops->iova_to_phys(ops, iova + 42))
+				return __FAIL(ops, i);
+
+			/* Remap full block */
+			if (ops->map(ops, iova, iova, size, IOMMU_WRITE))
+				return __FAIL(ops, i);
+
+			if (ops->iova_to_phys(ops, iova + 42) != (iova + 42))
+				return __FAIL(ops, i);
+
+			iova += SZ_1G;
+			j++;
+			j = find_next_bit(&cfg->pgsize_bitmap, BITS_PER_LONG, j);
+		}
+
+		free_io_pgtable_ops(ops);
+	}
+
+	selftest_running = false;
+	return 0;
+}
+
+static int __init arm_lpae_do_selftests(void)
+{
+	static const unsigned long pgsize[] = {
+		SZ_4K | SZ_2M | SZ_1G,
+		SZ_16K | SZ_32M,
+		SZ_64K | SZ_512M,
+	};
+
+	static const unsigned int ias[] = {
+		32, 36, 40, 42, 44, 48,
+	};
+
+	int i, j, pass = 0, fail = 0;
+	struct io_pgtable_cfg cfg = {
+		.tlb = &dummy_tlb_ops,
+		.oas = 48,
+	};
+
+	for (i = 0; i < ARRAY_SIZE(pgsize); ++i) {
+		for (j = 0; j < ARRAY_SIZE(ias); ++j) {
+			cfg.pgsize_bitmap = pgsize[i];
+			cfg.ias = ias[j];
+			pr_info("selftest: pgsize_bitmap 0x%08lx, IAS %u\n",
+				pgsize[i], ias[j]);
+			if (arm_lpae_run_tests(&cfg))
+				fail++;
+			else
+				pass++;
+		}
+	}
+
+	pr_info("selftest: completed with %d PASS %d FAIL\n", pass, fail);
+	return fail ? -EFAULT : 0;
+}
+subsys_initcall(arm_lpae_do_selftests);
+#endif
diff --git a/drivers/iommu/io-pgtable.c b/drivers/iommu/io-pgtable.c
new file mode 100644
index 000000000000..6436fe24bc2f
--- /dev/null
+++ b/drivers/iommu/io-pgtable.c
@@ -0,0 +1,82 @@
+/*
+ * Generic page table allocator for IOMMUs.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ * Copyright (C) 2014 ARM Limited
+ *
+ * Author: Will Deacon <will.deacon@arm.com>
+ */
+
+#include <linux/bug.h>
+#include <linux/kernel.h>
+#include <linux/types.h>
+
+#include "io-pgtable.h"
+
+extern struct io_pgtable_init_fns io_pgtable_arm_32_lpae_s1_init_fns;
+extern struct io_pgtable_init_fns io_pgtable_arm_32_lpae_s2_init_fns;
+extern struct io_pgtable_init_fns io_pgtable_arm_64_lpae_s1_init_fns;
+extern struct io_pgtable_init_fns io_pgtable_arm_64_lpae_s2_init_fns;
+
+static const struct io_pgtable_init_fns *
+io_pgtable_init_table[IO_PGTABLE_NUM_FMTS] =
+{
+#ifdef CONFIG_IOMMU_IO_PGTABLE_LPAE
+	[ARM_32_LPAE_S1] = &io_pgtable_arm_32_lpae_s1_init_fns,
+	[ARM_32_LPAE_S2] = &io_pgtable_arm_32_lpae_s2_init_fns,
+	[ARM_64_LPAE_S1] = &io_pgtable_arm_64_lpae_s1_init_fns,
+	[ARM_64_LPAE_S2] = &io_pgtable_arm_64_lpae_s2_init_fns,
+#endif
+};
+
+struct io_pgtable_ops *alloc_io_pgtable_ops(enum io_pgtable_fmt fmt,
+					    struct io_pgtable_cfg *cfg,
+					    void *cookie)
+{
+	struct io_pgtable *iop;
+	const struct io_pgtable_init_fns *fns;
+
+	if (fmt >= IO_PGTABLE_NUM_FMTS)
+		return NULL;
+
+	fns = io_pgtable_init_table[fmt];
+	if (!fns)
+		return NULL;
+
+	iop = fns->alloc(cfg, cookie);
+	if (!iop)
+		return NULL;
+
+	iop->fmt	= fmt;
+	iop->cookie	= cookie;
+	iop->cfg	= *cfg;
+
+	return &iop->ops;
+}
+
+/*
+ * It is the IOMMU driver's responsibility to ensure that the page table
+ * is no longer accessible to the walker by this point.
+ */
+void free_io_pgtable_ops(struct io_pgtable_ops *ops)
+{
+	struct io_pgtable *iop;
+
+	if (!ops)
+		return;
+
+	iop = container_of(ops, struct io_pgtable, ops);
+	iop->cfg.tlb->tlb_flush_all(iop->cookie);
+	io_pgtable_init_table[iop->fmt]->free(iop);
+}
diff --git a/drivers/iommu/io-pgtable.h b/drivers/iommu/io-pgtable.h
new file mode 100644
index 000000000000..10e32f69c668
--- /dev/null
+++ b/drivers/iommu/io-pgtable.h
@@ -0,0 +1,143 @@
+#ifndef __IO_PGTABLE_H
+#define __IO_PGTABLE_H
+
+/*
+ * Public API for use by IOMMU drivers
+ */
+enum io_pgtable_fmt {
+	ARM_32_LPAE_S1,
+	ARM_32_LPAE_S2,
+	ARM_64_LPAE_S1,
+	ARM_64_LPAE_S2,
+	IO_PGTABLE_NUM_FMTS,
+};
+
+/**
+ * struct iommu_gather_ops - IOMMU callbacks for TLB and page table management.
+ *
+ * @tlb_flush_all: Synchronously invalidate the entire TLB context.
+ * @tlb_add_flush: Queue up a TLB invalidation for a virtual address range.
+ * @tlb_sync:      Ensure any queue TLB invalidation has taken effect.
+ * @flush_pgtable: Ensure page table updates are visible to the IOMMU.
+ *
+ * Note that these can all be called in atomic context and must therefore
+ * not block.
+ */
+struct iommu_gather_ops {
+	void (*tlb_flush_all)(void *cookie);
+	void (*tlb_add_flush)(unsigned long iova, size_t size, bool leaf,
+			      void *cookie);
+	void (*tlb_sync)(void *cookie);
+	void (*flush_pgtable)(void *ptr, size_t size, void *cookie);
+};
+
+/**
+ * struct io_pgtable_cfg - Configuration data for a set of page tables.
+ *
+ * @quirks:        A bitmap of hardware quirks that require some special
+ *                 action by the low-level page table allocator.
+ * @pgsize_bitmap: A bitmap of page sizes supported by this set of page
+ *                 tables.
+ * @ias:           Input address (iova) size, in bits.
+ * @oas:           Output address (paddr) size, in bits.
+ * @tlb:           TLB management callbacks for this set of tables.
+ */
+struct io_pgtable_cfg {
+	#define IO_PGTABLE_QUIRK_ARM_NS	(1 << 0)	/* Set NS bit in PTEs */
+	int				quirks;
+	unsigned long			pgsize_bitmap;
+	unsigned int			ias;
+	unsigned int			oas;
+	const struct iommu_gather_ops	*tlb;
+
+	/* Low-level data specific to the table format */
+	union {
+		struct {
+			u64	ttbr[2];
+			u64	tcr;
+			u64	mair[2];
+		} arm_lpae_s1_cfg;
+
+		struct {
+			u64	vttbr;
+			u64	vtcr;
+		} arm_lpae_s2_cfg;
+	};
+};
+
+/**
+ * struct io_pgtable_ops - Page table manipulation API for IOMMU drivers.
+ *
+ * @map:          Map a physically contiguous memory region.
+ * @unmap:        Unmap a physically contiguous memory region.
+ * @iova_to_phys: Translate iova to physical address.
+ *
+ * These functions map directly onto the iommu_ops member functions with
+ * the same names.
+ */
+struct io_pgtable_ops {
+	int (*map)(struct io_pgtable_ops *ops, unsigned long iova,
+		   phys_addr_t paddr, size_t size, int prot);
+	int (*unmap)(struct io_pgtable_ops *ops, unsigned long iova,
+		     size_t size);
+	phys_addr_t (*iova_to_phys)(struct io_pgtable_ops *ops,
+				    unsigned long iova);
+};
+
+/**
+ * alloc_io_pgtable_ops() - Allocate a page table allocator for use by an IOMMU.
+ *
+ * @fmt:    The page table format.
+ * @cfg:    The page table configuration. This will be modified to represent
+ *          the configuration actually provided by the allocator (e.g. the
+ *          pgsize_bitmap may be restricted).
+ * @cookie: An opaque token provided by the IOMMU driver and passed back to
+ *          the callback routines in cfg->tlb.
+ */
+struct io_pgtable_ops *alloc_io_pgtable_ops(enum io_pgtable_fmt fmt,
+					    struct io_pgtable_cfg *cfg,
+					    void *cookie);
+
+/**
+ * free_io_pgtable_ops() - Free an io_pgtable_ops structure. The caller
+ *                         *must* ensure that the page table is no longer
+ *                         live, but the TLB can be dirty.
+ *
+ * @ops: The ops returned from alloc_io_pgtable_ops.
+ */
+void free_io_pgtable_ops(struct io_pgtable_ops *ops);
+
+
+/*
+ * Internal structures for page table allocator implementations.
+ */
+
+/**
+ * struct io_pgtable - Internal structure describing a set of page tables.
+ *
+ * @fmt:    The page table format.
+ * @cookie: An opaque token provided by the IOMMU driver and passed back to
+ *          any callback routines.
+ * @cfg:    A copy of the page table configuration.
+ * @ops:    The page table operations in use for this set of page tables.
+ */
+struct io_pgtable {
+	enum io_pgtable_fmt	fmt;
+	void			*cookie;
+	struct io_pgtable_cfg	cfg;
+	struct io_pgtable_ops	ops;
+};
+
+/**
+ * struct io_pgtable_init_fns - Alloc/free a set of page tables for a
+ *                              particular format.
+ *
+ * @alloc: Allocate a set of page tables described by cfg.
+ * @free:  Free the page tables associated with iop.
+ */
+struct io_pgtable_init_fns {
+	struct io_pgtable *(*alloc)(struct io_pgtable_cfg *cfg, void *cookie);
+	void (*free)(struct io_pgtable *iop);
+};
+
+#endif /* __IO_PGTABLE_H */
diff --git a/include/linux/iopoll.h b/include/linux/iopoll.h
new file mode 100644
index 000000000000..1c30014ed176
--- /dev/null
+++ b/include/linux/iopoll.h
@@ -0,0 +1,144 @@
+/*
+ * Copyright (c) 2012-2014 The Linux Foundation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ */
+
+#ifndef _LINUX_IOPOLL_H
+#define _LINUX_IOPOLL_H
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/hrtimer.h>
+#include <linux/delay.h>
+#include <linux/errno.h>
+#include <linux/io.h>
+
+/**
+ * readx_poll_timeout - Periodically poll an address until a condition is met or a timeout occurs
+ * @op: accessor function (takes @addr as its only argument)
+ * @addr: Address to poll
+ * @val: Variable to read the value into
+ * @cond: Break condition (usually involving @val)
+ * @sleep_us: Maximum time to sleep between reads in us (0
+ *            tight-loops).  Should be less than ~20ms since usleep_range
+ *            is used (see Documentation/timers/timers-howto.txt).
+ * @timeout_us: Timeout in us, 0 means never timeout
+ *
+ * Returns 0 on success and -ETIMEDOUT upon a timeout. In either
+ * case, the last read value at @addr is stored in @val. Must not
+ * be called from atomic context if sleep_us or timeout_us are used.
+ *
+ * When available, you'll probably want to use one of the specialized
+ * macros defined below rather than this macro directly.
+ */
+#define readx_poll_timeout(op, addr, val, cond, sleep_us, timeout_us)	\
+({ \
+	ktime_t timeout = ktime_add_us(ktime_get(), timeout_us); \
+	might_sleep_if(sleep_us); \
+	for (;;) { \
+		(val) = op(addr); \
+		if (cond) \
+			break; \
+		if (timeout_us && ktime_compare(ktime_get(), timeout) > 0) { \
+			(val) = op(addr); \
+			break; \
+		} \
+		if (sleep_us) \
+			usleep_range((sleep_us >> 2) + 1, sleep_us); \
+	} \
+	(cond) ? 0 : -ETIMEDOUT; \
+})
+
+/**
+ * readx_poll_timeout_atomic - Periodically poll an address until a condition is met or a timeout occurs
+ * @op: accessor function (takes @addr as its only argument)
+ * @addr: Address to poll
+ * @val: Variable to read the value into
+ * @cond: Break condition (usually involving @val)
+ * @delay_us: Time to udelay between reads in us (0 tight-loops).  Should
+ *            be less than ~10us since udelay is used (see
+ *            Documentation/timers/timers-howto.txt).
+ * @timeout_us: Timeout in us, 0 means never timeout
+ *
+ * Returns 0 on success and -ETIMEDOUT upon a timeout. In either
+ * case, the last read value at @addr is stored in @val.
+ *
+ * When available, you'll probably want to use one of the specialized
+ * macros defined below rather than this macro directly.
+ */
+#define readx_poll_timeout_atomic(op, addr, val, cond, delay_us, timeout_us) \
+({ \
+	ktime_t timeout = ktime_add_us(ktime_get(), timeout_us); \
+	for (;;) { \
+		(val) = op(addr); \
+		if (cond) \
+			break; \
+		if (timeout_us && ktime_compare(ktime_get(), timeout) > 0) { \
+			(val) = op(addr); \
+			break; \
+		} \
+		if (delay_us) \
+			udelay(delay_us);	\
+	} \
+	(cond) ? 0 : -ETIMEDOUT; \
+})
+
+
+#define readb_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readb, addr, val, cond, delay_us, timeout_us)
+
+#define readb_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readb, addr, val, cond, delay_us, timeout_us)
+
+#define readw_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readw, addr, val, cond, delay_us, timeout_us)
+
+#define readw_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readw, addr, val, cond, delay_us, timeout_us)
+
+#define readl_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readl, addr, val, cond, delay_us, timeout_us)
+
+#define readl_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readl, addr, val, cond, delay_us, timeout_us)
+
+#define readq_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readq, addr, val, cond, delay_us, timeout_us)
+
+#define readq_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readq, addr, val, cond, delay_us, timeout_us)
+
+#define readb_relaxed_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readb_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readb_relaxed_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readb_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readw_relaxed_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readw_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readw_relaxed_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readw_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readl_relaxed_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readl_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readl_relaxed_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readl_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readq_relaxed_poll_timeout(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout(readq_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#define readq_relaxed_poll_timeout_atomic(addr, val, cond, delay_us, timeout_us) \
+	readx_poll_timeout_atomic(readq_relaxed, addr, val, cond, delay_us, timeout_us)
+
+#endif /* _LINUX_IOPOLL_H */