The IOMMU updates for this round are not very large patch-wise. But

they contain two new IOMMU drivers for the ARM Tegra 2 and 3 platforms.
 Besides that there are also a few patches for the AMD IOMMU which
 prepare the driver for adding intr-remapping support and a couple of
 fixes.
 -----BEGIN PGP SIGNATURE-----
 Version: GnuPG v1.4.11 (GNU/Linux)
 
 iQIcBAABAgAGBQJPa3RMAAoJECvwRC2XARrjgwUQAKYXxRmbiIT2FxWxoIx75Q6N
 vO+Z3kaERBpY1dx2VD5shenIs6P6hr5DxnxbcMhD4wBNXVSRToj2BskLhBz79syi
 ecBCCPPELKypJvNpK55YM9gLnSwT8tbnPrCCMUqLQJQ1sptYMAfbwdRvNAOHF4Kp
 eU9JoJjq37WreX0hgHNqDcEBsBo18NwI+1jkayleTxeWf2gJiCWTs7VwKsa3Xcni
 OkFvNHy1krjqFlcT21p/eEf5dCTHXsxzU4+ilr4S8pg99KMsZuHFc6tFUkby3OC6
 jqbgNqxve2y5KvqAvyqxWkcUbejbN3B6AukiYj70Duy/6OPZ39F6O+vPAoYsxbR1
 0kT3evwbTBRYH/xcV4oJjPgdH7+hzAoe4ZNm3ROditXxdFU7vb8rlArTyVup8zrF
 LiVbLHll9v6vaxk8Ff8ZY9nJSOiNXWBGwsusj33PTE3RkVucpOlPQIX5yAcBYBDO
 JI6Qf7gm7GXdOKA23HxXwQKr8kw+EsDAEqslAMFTd9o0dFejikMXoDk2GYVO5qOK
 b0kGcGwDCcd1QaFDT+GimT2IueoM+YNUtwLpbXLm3syWfKNuqC8uJ32V0gHt4k9X
 1FU0qKSifKiJlDYT+JjHuRdRlZhfreIK7rHLVDyulKOHp1O5DAGeFHigRpEq8i9O
 xQvz5vq1vyO5nO8P6gDl
 =lCBm
 -----END PGP SIGNATURE-----

Merge tag 'iommu-updates-v3.4' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu

Pull IOMMU updates from Joerg Roedel:
 "The IOMMU updates for this round are not very large patch-wise.  But
  they contain two new IOMMU drivers for the ARM Tegra 2 and 3
  platforms.  Besides that there are also a few patches for the AMD
  IOMMU which prepare the driver for adding intr-remapping support and a
  couple of fixes."

* tag 'iommu-updates-v3.4' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
  iommu/amd: Fix section mismatch
  iommu/amd: Move interrupt setup code into seperate function
  iommu/amd: Make sure IOMMU interrupts are re-enabled on resume
  iommu/amd: Fix section warning for prealloc_protection_domains
  iommu/amd: Don't initialize IOMMUv2 resources when not required
  iommu/amd: Update git-tree in MAINTAINERS
  iommu/tegra-gart: fix spin_unlock in map failure path
  iommu/amd: Fix double free of mem-region in error-path
  iommu/amd: Split amd_iommu_init function
  ARM: IOMMU: Tegra30: Add iommu_ops for SMMU driver
  ARM: IOMMU: Tegra20: Add iommu_ops for GART driver

MAINTAINERS

@@ -503,7 +503,7 @@ F:	arch/x86/include/asm/geode.h
 AMD IOMMU (AMD-VI)
 M:	Joerg Roedel <joerg.roedel@amd.com>
 L:	iommu@lists.linux-foundation.org
-T:	git git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git
+T:	git git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu.git
 S:	Supported
 F:	drivers/iommu/amd_iommu*.[ch]
 F:	include/linux/amd-iommu.h

arch/arm/mach-tegra/include/mach/smmu.h

@@ -0,0 +1,63 @@
+/*
+ * IOMMU API for SMMU in Tegra30
+ *
+ * Copyright (c) 2012, NVIDIA CORPORATION.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope 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, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ */
+
+#ifndef	MACH_SMMU_H
+#define	MACH_SMMU_H
+
+enum smmu_hwgrp {
+	HWGRP_AFI,
+	HWGRP_AVPC,
+	HWGRP_DC,
+	HWGRP_DCB,
+	HWGRP_EPP,
+	HWGRP_G2,
+	HWGRP_HC,
+	HWGRP_HDA,
+	HWGRP_ISP,
+	HWGRP_MPE,
+	HWGRP_NV,
+	HWGRP_NV2,
+	HWGRP_PPCS,
+	HWGRP_SATA,
+	HWGRP_VDE,
+	HWGRP_VI,
+
+	HWGRP_COUNT,
+
+	HWGRP_END = ~0,
+};
+
+#define HWG_AFI		(1 << HWGRP_AFI)
+#define HWG_AVPC	(1 << HWGRP_AVPC)
+#define HWG_DC		(1 << HWGRP_DC)
+#define HWG_DCB		(1 << HWGRP_DCB)
+#define HWG_EPP		(1 << HWGRP_EPP)
+#define HWG_G2		(1 << HWGRP_G2)
+#define HWG_HC		(1 << HWGRP_HC)
+#define HWG_HDA		(1 << HWGRP_HDA)
+#define HWG_ISP		(1 << HWGRP_ISP)
+#define HWG_MPE		(1 << HWGRP_MPE)
+#define HWG_NV		(1 << HWGRP_NV)
+#define HWG_NV2		(1 << HWGRP_NV2)
+#define HWG_PPCS	(1 << HWGRP_PPCS)
+#define HWG_SATA	(1 << HWGRP_SATA)
+#define HWG_VDE		(1 << HWGRP_VDE)
+#define HWG_VI		(1 << HWGRP_VI)
+
+#endif	/* MACH_SMMU_H */

drivers/iommu/Kconfig

@@ -142,4 +142,24 @@ config OMAP_IOMMU_DEBUG
 
          Say N unless you know you need this.
 
+config TEGRA_IOMMU_GART
+	bool "Tegra GART IOMMU Support"
+	depends on ARCH_TEGRA_2x_SOC
+	select IOMMU_API
+	help
+	  Enables support for remapping discontiguous physical memory
+	  shared with the operating system into contiguous I/O virtual
+	  space through the GART (Graphics Address Relocation Table)
+	  hardware included on Tegra SoCs.
+
+config TEGRA_IOMMU_SMMU
+	bool "Tegra SMMU IOMMU Support"
+	depends on ARCH_TEGRA_3x_SOC
+	select IOMMU_API
+	help
+	  Enables support for remapping discontiguous physical memory
+	  shared with the operating system into contiguous I/O virtual
+	  space through the SMMU (System Memory Management Unit)
+	  hardware included on Tegra SoCs.
+
 endif # IOMMU_SUPPORT

drivers/iommu/Makefile

@@ -8,3 +8,5 @@ obj-$(CONFIG_IRQ_REMAP) += intr_remapping.o
 obj-$(CONFIG_OMAP_IOMMU) += omap-iommu.o
 obj-$(CONFIG_OMAP_IOVMM) += omap-iovmm.o
 obj-$(CONFIG_OMAP_IOMMU_DEBUG) += omap-iommu-debug.o
+obj-$(CONFIG_TEGRA_IOMMU_GART) += tegra-gart.o
+obj-$(CONFIG_TEGRA_IOMMU_SMMU) += tegra-smmu.o

drivers/iommu/amd_iommu_init.c

@@ -196,6 +196,8 @@ static u32 rlookup_table_size;	/* size if the rlookup table */
  */
 extern void iommu_flush_all_caches(struct amd_iommu *iommu);
 
+static int amd_iommu_enable_interrupts(void);
+
 static inline void update_last_devid(u16 devid)
 {
 	if (devid > amd_iommu_last_bdf)
@@ -358,8 +360,6 @@ static void iommu_disable(struct amd_iommu *iommu)
  */
 static u8 * __init iommu_map_mmio_space(u64 address)
 {
-	u8 *ret;
-
 	if (!request_mem_region(address, MMIO_REGION_LENGTH, "amd_iommu")) {
 		pr_err("AMD-Vi: Can not reserve memory region %llx for mmio\n",
 			address);
@@ -367,13 +367,7 @@ static u8 * __init iommu_map_mmio_space(u64 address)
 		return NULL;
 	}
 
-	ret = ioremap_nocache(address, MMIO_REGION_LENGTH);
+	return ioremap_nocache(address, MMIO_REGION_LENGTH);
-	if (ret != NULL)
-		return ret;
-
-	release_mem_region(address, MMIO_REGION_LENGTH);
-
-	return NULL;
 }
 
 static void __init iommu_unmap_mmio_space(struct amd_iommu *iommu)
@@ -1131,8 +1125,9 @@ static int iommu_setup_msi(struct amd_iommu *iommu)
 {
 	int r;
 
-	if (pci_enable_msi(iommu->dev))
+	r = pci_enable_msi(iommu->dev);
-		return 1;
+	if (r)
+		return r;
 
 	r = request_threaded_irq(iommu->dev->irq,
 				 amd_iommu_int_handler,
@@ -1142,27 +1137,36 @@ static int iommu_setup_msi(struct amd_iommu *iommu)
 
 	if (r) {
 		pci_disable_msi(iommu->dev);
-		return 1;
+		return r;
 	}
 
 	iommu->int_enabled = true;
-	iommu_feature_enable(iommu, CONTROL_EVT_INT_EN);
-
-	if (iommu->ppr_log != NULL)
-		iommu_feature_enable(iommu, CONTROL_PPFINT_EN);
 
 	return 0;
 }
 
 static int iommu_init_msi(struct amd_iommu *iommu)
 {
+	int ret;
+
 	if (iommu->int_enabled)
-		return 0;
+		goto enable_faults;
 
 	if (pci_find_capability(iommu->dev, PCI_CAP_ID_MSI))
-		return iommu_setup_msi(iommu);
+		ret = iommu_setup_msi(iommu);
+	else
+		ret = -ENODEV;
 
-	return 1;
+	if (ret)
+		return ret;
+
+enable_faults:
+	iommu_feature_enable(iommu, CONTROL_EVT_INT_EN);
+
+	if (iommu->ppr_log != NULL)
+		iommu_feature_enable(iommu, CONTROL_PPFINT_EN);
+
+	return 0;
 }
 
 /****************************************************************************
@@ -1381,7 +1385,6 @@ static void enable_iommus(void)
 		iommu_enable_ppr_log(iommu);
 		iommu_enable_gt(iommu);
 		iommu_set_exclusion_range(iommu);
-		iommu_init_msi(iommu);
 		iommu_enable(iommu);
 		iommu_flush_all_caches(iommu);
 	}
@@ -1409,6 +1412,8 @@ static void amd_iommu_resume(void)
 
 	/* re-load the hardware */
 	enable_iommus();
+
+	amd_iommu_enable_interrupts();
 }
 
 static int amd_iommu_suspend(void)
@@ -1424,10 +1429,40 @@ static struct syscore_ops amd_iommu_syscore_ops = {
 	.resume = amd_iommu_resume,
 };
 
+static void __init free_on_init_error(void)
+{
+	amd_iommu_uninit_devices();
+
+	free_pages((unsigned long)amd_iommu_pd_alloc_bitmap,
+		   get_order(MAX_DOMAIN_ID/8));
+
+	free_pages((unsigned long)amd_iommu_rlookup_table,
+		   get_order(rlookup_table_size));
+
+	free_pages((unsigned long)amd_iommu_alias_table,
+		   get_order(alias_table_size));
+
+	free_pages((unsigned long)amd_iommu_dev_table,
+		   get_order(dev_table_size));
+
+	free_iommu_all();
+
+	free_unity_maps();
+
+#ifdef CONFIG_GART_IOMMU
+	/*
+	 * We failed to initialize the AMD IOMMU - try fallback to GART
+	 * if possible.
+	 */
+	gart_iommu_init();
+
+#endif
+}
+
 /*
- * This is the core init function for AMD IOMMU hardware in the system.
+ * This is the hardware init function for AMD IOMMU in the system.
- * This function is called from the generic x86 DMA layer initialization
+ * This function is called either from amd_iommu_init or from the interrupt
- * code.
+ * remapping setup code.
  *
  * This function basically parses the ACPI table for AMD IOMMU (IVRS)
  * three times:
@@ -1446,16 +1481,21 @@ static struct syscore_ops amd_iommu_syscore_ops = {
  *		remapping requirements parsed out of the ACPI table in
  *		this last pass.
  *
- * After that the hardware is initialized and ready to go. In the last
+ * After everything is set up the IOMMUs are enabled and the necessary
- * step we do some Linux specific things like registering the driver in
+ * hotplug and suspend notifiers are registered.
- * the dma_ops interface and initializing the suspend/resume support
- * functions. Finally it prints some information about AMD IOMMUs and
- * the driver state and enables the hardware.
  */
-static int __init amd_iommu_init(void)
+int __init amd_iommu_init_hardware(void)
 {
 	int i, ret = 0;
 
+	if (!amd_iommu_detected)
+		return -ENODEV;
+
+	if (amd_iommu_dev_table != NULL) {
+		/* Hardware already initialized */
+		return 0;
+	}
+
 	/*
 	 * First parse ACPI tables to find the largest Bus/Dev/Func
 	 * we need to handle. Upon this information the shared data
@@ -1472,9 +1512,8 @@ static int __init amd_iommu_init(void)
 	alias_table_size   = tbl_size(ALIAS_TABLE_ENTRY_SIZE);
 	rlookup_table_size = tbl_size(RLOOKUP_TABLE_ENTRY_SIZE);
 
-	ret = -ENOMEM;
-
 	/* Device table - directly used by all IOMMUs */
+	ret = -ENOMEM;
 	amd_iommu_dev_table = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
 				      get_order(dev_table_size));
 	if (amd_iommu_dev_table == NULL)
@@ -1546,20 +1585,65 @@ static int __init amd_iommu_init(void)
 
 	enable_iommus();
 
+	amd_iommu_init_notifier();
+
+	register_syscore_ops(&amd_iommu_syscore_ops);
+
+out:
+	return ret;
+
+free:
+	free_on_init_error();
+
+	return ret;
+}
+
+static int amd_iommu_enable_interrupts(void)
+{
+	struct amd_iommu *iommu;
+	int ret = 0;
+
+	for_each_iommu(iommu) {
+		ret = iommu_init_msi(iommu);
+		if (ret)
+			goto out;
+	}
+
+out:
+	return ret;
+}
+
+/*
+ * This is the core init function for AMD IOMMU hardware in the system.
+ * This function is called from the generic x86 DMA layer initialization
+ * code.
+ *
+ * The function calls amd_iommu_init_hardware() to setup and enable the
+ * IOMMU hardware if this has not happened yet. After that the driver
+ * registers for the DMA-API and for the IOMMU-API as necessary.
+ */
+static int __init amd_iommu_init(void)
+{
+	int ret = 0;
+
+	ret = amd_iommu_init_hardware();
+	if (ret)
+		goto out;
+
+	ret = amd_iommu_enable_interrupts();
+	if (ret)
+		goto free;
+
 	if (iommu_pass_through)
 		ret = amd_iommu_init_passthrough();
 	else
 		ret = amd_iommu_init_dma_ops();
 
 	if (ret)
-		goto free_disable;
+		goto free;
 
 	amd_iommu_init_api();
 
-	amd_iommu_init_notifier();
-
-	register_syscore_ops(&amd_iommu_syscore_ops);
-
 	if (iommu_pass_through)
 		goto out;
 
@@ -1569,39 +1653,14 @@ static int __init amd_iommu_init(void)
 		printk(KERN_INFO "AMD-Vi: Lazy IO/TLB flushing enabled\n");
 
 	x86_platform.iommu_shutdown = disable_iommus;
+
 out:
 	return ret;
 
-free_disable:
+free:
 	disable_iommus();
 
-free:
+	free_on_init_error();
-	amd_iommu_uninit_devices();
-
-	free_pages((unsigned long)amd_iommu_pd_alloc_bitmap,
-		   get_order(MAX_DOMAIN_ID/8));
-
-	free_pages((unsigned long)amd_iommu_rlookup_table,
-		   get_order(rlookup_table_size));
-
-	free_pages((unsigned long)amd_iommu_alias_table,
-		   get_order(alias_table_size));
-
-	free_pages((unsigned long)amd_iommu_dev_table,
-		   get_order(dev_table_size));
-
-	free_iommu_all();
-
-	free_unity_maps();
-
-#ifdef CONFIG_GART_IOMMU
-	/*
-	 * We failed to initialize the AMD IOMMU - try fallback to GART
-	 * if possible.
-	 */
-	gart_iommu_init();
-
-#endif
 
 	goto out;
 }

drivers/iommu/amd_iommu_v2.c

@@ -921,7 +921,16 @@ static int __init amd_iommu_v2_init(void)
 	size_t state_table_size;
 	int ret;
 
-	pr_info("AMD IOMMUv2 driver by Joerg Roedel <joerg.roedel@amd.com>");
+	pr_info("AMD IOMMUv2 driver by Joerg Roedel <joerg.roedel@amd.com>\n");
+
+	if (!amd_iommu_v2_supported()) {
+		pr_info("AMD IOMMUv2 functionality not available on this sytem\n");
+		/*
+		 * Load anyway to provide the symbols to other modules
+		 * which may use AMD IOMMUv2 optionally.
+		 */
+		return 0;
+	}
 
 	spin_lock_init(&state_lock);
 
@@ -961,6 +970,9 @@ static void __exit amd_iommu_v2_exit(void)
 	size_t state_table_size;
 	int i;
 
+	if (!amd_iommu_v2_supported())
+		return;
+
 	profile_event_unregister(PROFILE_TASK_EXIT, &profile_nb);
 	amd_iommu_unregister_ppr_notifier(&ppr_nb);
 

drivers/iommu/tegra-gart.c

@@ -0,0 +1,451 @@
+/*
+ * IOMMU API for GART in Tegra20
+ *
+ * Copyright (c) 2010-2012, NVIDIA CORPORATION.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope 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, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ */
+
+#define pr_fmt(fmt)	"%s(): " fmt, __func__
+
+#include <linux/module.h>
+#include <linux/platform_device.h>
+#include <linux/spinlock.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/mm.h>
+#include <linux/list.h>
+#include <linux/device.h>
+#include <linux/io.h>
+#include <linux/iommu.h>
+
+#include <asm/cacheflush.h>
+
+/* bitmap of the page sizes currently supported */
+#define GART_IOMMU_PGSIZES	(SZ_4K)
+
+#define GART_CONFIG		0x24
+#define GART_ENTRY_ADDR		0x28
+#define GART_ENTRY_DATA		0x2c
+#define GART_ENTRY_PHYS_ADDR_VALID	(1 << 31)
+
+#define GART_PAGE_SHIFT		12
+#define GART_PAGE_SIZE		(1 << GART_PAGE_SHIFT)
+#define GART_PAGE_MASK						\
+	(~(GART_PAGE_SIZE - 1) & ~GART_ENTRY_PHYS_ADDR_VALID)
+
+struct gart_client {
+	struct device		*dev;
+	struct list_head	list;
+};
+
+struct gart_device {
+	void __iomem		*regs;
+	u32			*savedata;
+	u32			page_count;	/* total remappable size */
+	dma_addr_t		iovmm_base;	/* offset to vmm_area */
+	spinlock_t		pte_lock;	/* for pagetable */
+	struct list_head	client;
+	spinlock_t		client_lock;	/* for client list */
+	struct device		*dev;
+};
+
+static struct gart_device *gart_handle; /* unique for a system */
+
+#define GART_PTE(_pfn)						\
+	(GART_ENTRY_PHYS_ADDR_VALID | ((_pfn) << PAGE_SHIFT))
+
+/*
+ * Any interaction between any block on PPSB and a block on APB or AHB
+ * must have these read-back to ensure the APB/AHB bus transaction is
+ * complete before initiating activity on the PPSB block.
+ */
+#define FLUSH_GART_REGS(gart)	((void)readl((gart)->regs + GART_CONFIG))
+
+#define for_each_gart_pte(gart, iova)					\
+	for (iova = gart->iovmm_base;					\
+	     iova < gart->iovmm_base + GART_PAGE_SIZE * gart->page_count; \
+	     iova += GART_PAGE_SIZE)
+
+static inline void gart_set_pte(struct gart_device *gart,
+				unsigned long offs, u32 pte)
+{
+	writel(offs, gart->regs + GART_ENTRY_ADDR);
+	writel(pte, gart->regs + GART_ENTRY_DATA);
+
+	dev_dbg(gart->dev, "%s %08lx:%08x\n",
+		 pte ? "map" : "unmap", offs, pte & GART_PAGE_MASK);
+}
+
+static inline unsigned long gart_read_pte(struct gart_device *gart,
+					  unsigned long offs)
+{
+	unsigned long pte;
+
+	writel(offs, gart->regs + GART_ENTRY_ADDR);
+	pte = readl(gart->regs + GART_ENTRY_DATA);
+
+	return pte;
+}
+
+static void do_gart_setup(struct gart_device *gart, const u32 *data)
+{
+	unsigned long iova;
+
+	for_each_gart_pte(gart, iova)
+		gart_set_pte(gart, iova, data ? *(data++) : 0);
+
+	writel(1, gart->regs + GART_CONFIG);
+	FLUSH_GART_REGS(gart);
+}
+
+#ifdef DEBUG
+static void gart_dump_table(struct gart_device *gart)
+{
+	unsigned long iova;
+	unsigned long flags;
+
+	spin_lock_irqsave(&gart->pte_lock, flags);
+	for_each_gart_pte(gart, iova) {
+		unsigned long pte;
+
+		pte = gart_read_pte(gart, iova);
+
+		dev_dbg(gart->dev, "%s %08lx:%08lx\n",
+			(GART_ENTRY_PHYS_ADDR_VALID & pte) ? "v" : " ",
+			iova, pte & GART_PAGE_MASK);
+	}
+	spin_unlock_irqrestore(&gart->pte_lock, flags);
+}
+#else
+static inline void gart_dump_table(struct gart_device *gart)
+{
+}
+#endif
+
+static inline bool gart_iova_range_valid(struct gart_device *gart,
+					 unsigned long iova, size_t bytes)
+{
+	unsigned long iova_start, iova_end, gart_start, gart_end;
+
+	iova_start = iova;
+	iova_end = iova_start + bytes - 1;
+	gart_start = gart->iovmm_base;
+	gart_end = gart_start + gart->page_count * GART_PAGE_SIZE - 1;
+
+	if (iova_start < gart_start)
+		return false;
+	if (iova_end > gart_end)
+		return false;
+	return true;
+}
+
+static int gart_iommu_attach_dev(struct iommu_domain *domain,
+				 struct device *dev)
+{
+	struct gart_device *gart;
+	struct gart_client *client, *c;
+	int err = 0;
+
+	gart = dev_get_drvdata(dev->parent);
+	if (!gart)
+		return -EINVAL;
+	domain->priv = gart;
+
+	client = devm_kzalloc(gart->dev, sizeof(*c), GFP_KERNEL);
+	if (!client)
+		return -ENOMEM;
+	client->dev = dev;
+
+	spin_lock(&gart->client_lock);
+	list_for_each_entry(c, &gart->client, list) {
+		if (c->dev == dev) {
+			dev_err(gart->dev,
+				"%s is already attached\n", dev_name(dev));
+			err = -EINVAL;
+			goto fail;
+		}
+	}
+	list_add(&client->list, &gart->client);
+	spin_unlock(&gart->client_lock);
+	dev_dbg(gart->dev, "Attached %s\n", dev_name(dev));
+	return 0;
+
+fail:
+	devm_kfree(gart->dev, client);
+	spin_unlock(&gart->client_lock);
+	return err;
+}
+
+static void gart_iommu_detach_dev(struct iommu_domain *domain,
+				  struct device *dev)
+{
+	struct gart_device *gart = domain->priv;
+	struct gart_client *c;
+
+	spin_lock(&gart->client_lock);
+
+	list_for_each_entry(c, &gart->client, list) {
+		if (c->dev == dev) {
+			list_del(&c->list);
+			devm_kfree(gart->dev, c);
+			dev_dbg(gart->dev, "Detached %s\n", dev_name(dev));
+			goto out;
+		}
+	}
+	dev_err(gart->dev, "Couldn't find\n");
+out:
+	spin_unlock(&gart->client_lock);
+}
+
+static int gart_iommu_domain_init(struct iommu_domain *domain)
+{
+	return 0;
+}
+
+static void gart_iommu_domain_destroy(struct iommu_domain *domain)
+{
+	struct gart_device *gart = domain->priv;
+
+	if (!gart)
+		return;
+
+	spin_lock(&gart->client_lock);
+	if (!list_empty(&gart->client)) {
+		struct gart_client *c;
+
+		list_for_each_entry(c, &gart->client, list)
+			gart_iommu_detach_dev(domain, c->dev);
+	}
+	spin_unlock(&gart->client_lock);
+	domain->priv = NULL;
+}
+
+static int gart_iommu_map(struct iommu_domain *domain, unsigned long iova,
+			  phys_addr_t pa, size_t bytes, int prot)
+{
+	struct gart_device *gart = domain->priv;
+	unsigned long flags;
+	unsigned long pfn;
+
+	if (!gart_iova_range_valid(gart, iova, bytes))
+		return -EINVAL;
+
+	spin_lock_irqsave(&gart->pte_lock, flags);
+	pfn = __phys_to_pfn(pa);
+	if (!pfn_valid(pfn)) {
+		dev_err(gart->dev, "Invalid page: %08x\n", pa);
+		spin_unlock_irqrestore(&gart->pte_lock, flags);
+		return -EINVAL;
+	}
+	gart_set_pte(gart, iova, GART_PTE(pfn));
+	FLUSH_GART_REGS(gart);
+	spin_unlock_irqrestore(&gart->pte_lock, flags);
+	return 0;
+}
+
+static size_t gart_iommu_unmap(struct iommu_domain *domain, unsigned long iova,
+			       size_t bytes)
+{
+	struct gart_device *gart = domain->priv;
+	unsigned long flags;
+
+	if (!gart_iova_range_valid(gart, iova, bytes))
+		return 0;
+
+	spin_lock_irqsave(&gart->pte_lock, flags);
+	gart_set_pte(gart, iova, 0);
+	FLUSH_GART_REGS(gart);
+	spin_unlock_irqrestore(&gart->pte_lock, flags);
+	return 0;
+}
+
+static phys_addr_t gart_iommu_iova_to_phys(struct iommu_domain *domain,
+					   unsigned long iova)
+{
+	struct gart_device *gart = domain->priv;
+	unsigned long pte;
+	phys_addr_t pa;
+	unsigned long flags;
+
+	if (!gart_iova_range_valid(gart, iova, 0))
+		return -EINVAL;
+
+	spin_lock_irqsave(&gart->pte_lock, flags);
+	pte = gart_read_pte(gart, iova);
+	spin_unlock_irqrestore(&gart->pte_lock, flags);
+
+	pa = (pte & GART_PAGE_MASK);
+	if (!pfn_valid(__phys_to_pfn(pa))) {
+		dev_err(gart->dev, "No entry for %08lx:%08x\n", iova, pa);
+		gart_dump_table(gart);
+		return -EINVAL;
+	}
+	return pa;
+}
+
+static int gart_iommu_domain_has_cap(struct iommu_domain *domain,
+				     unsigned long cap)
+{
+	return 0;
+}
+
+static struct iommu_ops gart_iommu_ops = {
+	.domain_init	= gart_iommu_domain_init,
+	.domain_destroy	= gart_iommu_domain_destroy,
+	.attach_dev	= gart_iommu_attach_dev,
+	.detach_dev	= gart_iommu_detach_dev,
+	.map		= gart_iommu_map,
+	.unmap		= gart_iommu_unmap,
+	.iova_to_phys	= gart_iommu_iova_to_phys,
+	.domain_has_cap	= gart_iommu_domain_has_cap,
+	.pgsize_bitmap	= GART_IOMMU_PGSIZES,
+};
+
+static int tegra_gart_suspend(struct device *dev)
+{
+	struct gart_device *gart = dev_get_drvdata(dev);
+	unsigned long iova;
+	u32 *data = gart->savedata;
+	unsigned long flags;
+
+	spin_lock_irqsave(&gart->pte_lock, flags);
+	for_each_gart_pte(gart, iova)
+		*(data++) = gart_read_pte(gart, iova);
+	spin_unlock_irqrestore(&gart->pte_lock, flags);
+	return 0;
+}
+
+static int tegra_gart_resume(struct device *dev)
+{
+	struct gart_device *gart = dev_get_drvdata(dev);
+	unsigned long flags;
+
+	spin_lock_irqsave(&gart->pte_lock, flags);
+	do_gart_setup(gart, gart->savedata);
+	spin_unlock_irqrestore(&gart->pte_lock, flags);
+	return 0;
+}
+
+static int tegra_gart_probe(struct platform_device *pdev)
+{
+	struct gart_device *gart;
+	struct resource *res, *res_remap;
+	void __iomem *gart_regs;
+	int err;
+	struct device *dev = &pdev->dev;
+
+	if (gart_handle)
+		return -EIO;
+
+	BUILD_BUG_ON(PAGE_SHIFT != GART_PAGE_SHIFT);
+
+	/* the GART memory aperture is required */
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	res_remap = platform_get_resource(pdev, IORESOURCE_MEM, 1);
+	if (!res || !res_remap) {
+		dev_err(dev, "GART memory aperture expected\n");
+		return -ENXIO;
+	}
+
+	gart = devm_kzalloc(dev, sizeof(*gart), GFP_KERNEL);
+	if (!gart) {
+		dev_err(dev, "failed to allocate gart_device\n");
+		return -ENOMEM;
+	}
+
+	gart_regs = devm_ioremap(dev, res->start, resource_size(res));
+	if (!gart_regs) {
+		dev_err(dev, "failed to remap GART registers\n");
+		err = -ENXIO;
+		goto fail;
+	}
+
+	gart->dev = &pdev->dev;
+	spin_lock_init(&gart->pte_lock);
+	spin_lock_init(&gart->client_lock);
+	INIT_LIST_HEAD(&gart->client);
+	gart->regs = gart_regs;
+	gart->iovmm_base = (dma_addr_t)res_remap->start;
+	gart->page_count = (resource_size(res_remap) >> GART_PAGE_SHIFT);
+
+	gart->savedata = vmalloc(sizeof(u32) * gart->page_count);
+	if (!gart->savedata) {
+		dev_err(dev, "failed to allocate context save area\n");
+		err = -ENOMEM;
+		goto fail;
+	}
+
+	platform_set_drvdata(pdev, gart);
+	do_gart_setup(gart, NULL);
+
+	gart_handle = gart;
+	return 0;
+
+fail:
+	if (gart_regs)
+		devm_iounmap(dev, gart_regs);
+	if (gart && gart->savedata)
+		vfree(gart->savedata);
+	devm_kfree(dev, gart);
+	return err;
+}
+
+static int tegra_gart_remove(struct platform_device *pdev)
+{
+	struct gart_device *gart = platform_get_drvdata(pdev);
+	struct device *dev = gart->dev;
+
+	writel(0, gart->regs + GART_CONFIG);
+	if (gart->savedata)
+		vfree(gart->savedata);
+	if (gart->regs)
+		devm_iounmap(dev, gart->regs);
+	devm_kfree(dev, gart);
+	gart_handle = NULL;
+	return 0;
+}
+
+const struct dev_pm_ops tegra_gart_pm_ops = {
+	.suspend	= tegra_gart_suspend,
+	.resume		= tegra_gart_resume,
+};
+
+static struct platform_driver tegra_gart_driver = {
+	.probe		= tegra_gart_probe,
+	.remove		= tegra_gart_remove,
+	.driver = {
+		.owner	= THIS_MODULE,
+		.name	= "tegra-gart",
+		.pm	= &tegra_gart_pm_ops,
+	},
+};
+
+static int __devinit tegra_gart_init(void)
+{
+	bus_set_iommu(&platform_bus_type, &gart_iommu_ops);
+	return platform_driver_register(&tegra_gart_driver);
+}
+
+static void __exit tegra_gart_exit(void)
+{
+	platform_driver_unregister(&tegra_gart_driver);
+}
+
+subsys_initcall(tegra_gart_init);
+module_exit(tegra_gart_exit);
+
+MODULE_DESCRIPTION("IOMMU API for GART in Tegra20");
+MODULE_AUTHOR("Hiroshi DOYU <hdoyu@nvidia.com>");
+MODULE_LICENSE("GPL v2");

include/linux/amd-iommu.h

@@ -28,7 +28,7 @@ struct task_struct;
 struct pci_dev;
 
 extern int amd_iommu_detect(void);
-
+extern int amd_iommu_init_hardware(void);
 
 /**
  * amd_iommu_enable_device_erratum() - Enable erratum workaround for device