Merge branch 'core-iommu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip

* 'core-iommu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  iommu/core: Fix build with INTR_REMAP=y && CONFIG_DMAR=n
  iommu/amd: Don't use MSI address range for DMA addresses
  iommu/amd: Move missing parts to drivers/iommu
  iommu: Move iommu Kconfig entries to submenu
  x86/ia64: intel-iommu: move to drivers/iommu/
  x86: amd_iommu: move to drivers/iommu/
  msm: iommu: move to drivers/iommu/
  drivers: iommu: move to a dedicated folder
  x86/amd-iommu: Store device alias as dev_data pointer
  x86/amd-iommu: Search for existind dev_data before allocting a new one
  x86/amd-iommu: Allow dev_data->alias to be NULL
  x86/amd-iommu: Use only dev_data in low-level domain attach/detach functions
  x86/amd-iommu: Use only dev_data for dte and iotlb flushing routines
  x86/amd-iommu: Store ATS state in dev_data
  x86/amd-iommu: Store devid in dev_data
  x86/amd-iommu: Introduce global dev_data_list
  x86/amd-iommu: Remove redundant device_flush_dte() calls
  iommu-api: Add missing header file

Fix up trivial conflicts (independent additions close to each other) in
drivers/Makefile and include/linux/pci.h

19 files changed, 6308 insertions, 10 deletions

diff --git a/drivers/Kconfig b/drivers/Kconfig
index 258473ce8d0..52e306dd501 100644
--- a/drivers/Kconfig
+++ b/drivers/Kconfig
@@ -124,4 +124,6 @@ source "drivers/hwspinlock/Kconfig"
 
 source "drivers/clocksource/Kconfig"
 
+source "drivers/iommu/Kconfig"
+
 endmenu
diff --git a/drivers/Makefile b/drivers/Makefile
index 1bc896571a3..939fcdeb2d3 100644
--- a/drivers/Makefile
+++ b/drivers/Makefile
@@ -123,3 +123,4 @@ obj-y				+= clk/
 
 obj-$(CONFIG_HWSPINLOCK)	+= hwspinlock/
 obj-$(CONFIG_NFC)		+= nfc/
+obj-$(CONFIG_IOMMU_SUPPORT)	+= iommu/
diff --git a/drivers/base/Makefile b/drivers/base/Makefile
index 4c5701c15f5..5ab0d07c457 100644
--- a/drivers/base/Makefile
+++ b/drivers/base/Makefile
@@ -13,7 +13,6 @@ obj-$(CONFIG_FW_LOADER)	+= firmware_class.o
 obj-$(CONFIG_NUMA)	+= node.o
 obj-$(CONFIG_MEMORY_HOTPLUG_SPARSE) += memory.o
 obj-$(CONFIG_SMP)	+= topology.o
-obj-$(CONFIG_IOMMU_API) += iommu.o
 ifeq ($(CONFIG_SYSFS),y)
 obj-$(CONFIG_MODULES)	+= module.o
 endif
diff --git a/drivers/iommu/Kconfig b/drivers/iommu/Kconfig
new file mode 100644
index 00000000000..b57b3fa492f
--- /dev/null
+++ b/drivers/iommu/Kconfig
@@ -0,0 +1,110 @@
+# IOMMU_API always gets selected by whoever wants it.
+config IOMMU_API
+	bool
+
+menuconfig IOMMU_SUPPORT
+	bool "IOMMU Hardware Support"
+	default y
+	---help---
+	  Say Y here if you want to compile device drivers for IO Memory
+	  Management Units into the kernel. These devices usually allow to
+	  remap DMA requests and/or remap interrupts from other devices on the
+	  system.
+
+if IOMMU_SUPPORT
+
+# MSM IOMMU support
+config MSM_IOMMU
+	bool "MSM IOMMU Support"
+	depends on ARCH_MSM8X60 || ARCH_MSM8960
+	select IOMMU_API
+	help
+	  Support for the IOMMUs found on certain Qualcomm SOCs.
+	  These IOMMUs allow virtualization of the address space used by most
+	  cores within the multimedia subsystem.
+
+	  If unsure, say N here.
+
+config IOMMU_PGTABLES_L2
+	def_bool y
+	depends on MSM_IOMMU && MMU && SMP && CPU_DCACHE_DISABLE=n
+
+# AMD IOMMU support
+config AMD_IOMMU
+	bool "AMD IOMMU support"
+	select SWIOTLB
+	select PCI_MSI
+	select PCI_IOV
+	select IOMMU_API
+	depends on X86_64 && PCI && ACPI
+	---help---
+	  With this option you can enable support for AMD IOMMU hardware in
+	  your system. An IOMMU is a hardware component which provides
+	  remapping of DMA memory accesses from devices. With an AMD IOMMU you
+	  can isolate the the DMA memory of different devices and protect the
+	  system from misbehaving device drivers or hardware.
+
+	  You can find out if your system has an AMD IOMMU if you look into
+	  your BIOS for an option to enable it or if you have an IVRS ACPI
+	  table.
+
+config AMD_IOMMU_STATS
+	bool "Export AMD IOMMU statistics to debugfs"
+	depends on AMD_IOMMU
+	select DEBUG_FS
+	---help---
+	  This option enables code in the AMD IOMMU driver to collect various
+	  statistics about whats happening in the driver and exports that
+	  information to userspace via debugfs.
+	  If unsure, say N.
+
+# Intel IOMMU support
+config DMAR
+	bool "Support for DMA Remapping Devices"
+	depends on PCI_MSI && ACPI && (X86 || IA64_GENERIC)
+	select IOMMU_API
+	help
+	  DMA remapping (DMAR) devices support enables independent address
+	  translations for Direct Memory Access (DMA) from devices.
+	  These DMA remapping devices are reported via ACPI tables
+	  and include PCI device scope covered by these DMA
+	  remapping devices.
+
+config DMAR_DEFAULT_ON
+	def_bool y
+	prompt "Enable DMA Remapping Devices by default"
+	depends on DMAR
+	help
+	  Selecting this option will enable a DMAR device at boot time if
+	  one is found. If this option is not selected, DMAR support can
+	  be enabled by passing intel_iommu=on to the kernel.
+
+config DMAR_BROKEN_GFX_WA
+	bool "Workaround broken graphics drivers (going away soon)"
+	depends on DMAR && BROKEN && X86
+	---help---
+	  Current Graphics drivers tend to use physical address
+	  for DMA and avoid using DMA APIs. Setting this config
+	  option permits the IOMMU driver to set a unity map for
+	  all the OS-visible memory. Hence the driver can continue
+	  to use physical addresses for DMA, at least until this
+	  option is removed in the 2.6.32 kernel.
+
+config DMAR_FLOPPY_WA
+	def_bool y
+	depends on DMAR && X86
+	---help---
+	  Floppy disk drivers are known to bypass DMA API calls
+	  thereby failing to work when IOMMU is enabled. This
+	  workaround will setup a 1:1 mapping for the first
+	  16MiB to make floppy (an ISA device) work.
+
+config INTR_REMAP
+	bool "Support for Interrupt Remapping (EXPERIMENTAL)"
+	depends on X86_64 && X86_IO_APIC && PCI_MSI && ACPI && EXPERIMENTAL
+	---help---
+	  Supports Interrupt remapping for IO-APIC and MSI devices.
+	  To use x2apic mode in the CPU's which support x2APIC enhancements or
+	  to support platforms with CPU's having > 8 bit APIC ID, say Y.
+
+endif # IOMMU_SUPPORT
diff --git a/drivers/iommu/Makefile b/drivers/iommu/Makefile
new file mode 100644
index 00000000000..4d4d77df7ca
--- /dev/null
+++ b/drivers/iommu/Makefile
@@ -0,0 +1,5 @@
+obj-$(CONFIG_IOMMU_API) += iommu.o
+obj-$(CONFIG_MSM_IOMMU) += msm_iommu.o msm_iommu_dev.o
+obj-$(CONFIG_AMD_IOMMU) += amd_iommu.o amd_iommu_init.o
+obj-$(CONFIG_DMAR) += dmar.o iova.o intel-iommu.o
+obj-$(CONFIG_INTR_REMAP) += dmar.o intr_remapping.o
diff --git a/drivers/iommu/amd_iommu.c b/drivers/iommu/amd_iommu.c
new file mode 100644
index 00000000000..a14f8dc2346
--- /dev/null
+++ b/drivers/iommu/amd_iommu.c
@@ -0,0 +1,2824 @@
+/*
+ * Copyright (C) 2007-2010 Advanced Micro Devices, Inc.
+ * Author: Joerg Roedel <joerg.roedel@amd.com>
+ *         Leo Duran <leo.duran@amd.com>
+ *
+ * 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#include <linux/pci.h>
+#include <linux/pci-ats.h>
+#include <linux/bitmap.h>
+#include <linux/slab.h>
+#include <linux/debugfs.h>
+#include <linux/scatterlist.h>
+#include <linux/dma-mapping.h>
+#include <linux/iommu-helper.h>
+#include <linux/iommu.h>
+#include <linux/delay.h>
+#include <linux/amd-iommu.h>
+#include <asm/msidef.h>
+#include <asm/proto.h>
+#include <asm/iommu.h>
+#include <asm/gart.h>
+#include <asm/dma.h>
+
+#include "amd_iommu_proto.h"
+#include "amd_iommu_types.h"
+
+#define CMD_SET_TYPE(cmd, t) ((cmd)->data[1] |= ((t) << 28))
+
+#define LOOP_TIMEOUT	100000
+
+static DEFINE_RWLOCK(amd_iommu_devtable_lock);
+
+/* A list of preallocated protection domains */
+static LIST_HEAD(iommu_pd_list);
+static DEFINE_SPINLOCK(iommu_pd_list_lock);
+
+/* List of all available dev_data structures */
+static LIST_HEAD(dev_data_list);
+static DEFINE_SPINLOCK(dev_data_list_lock);
+
+/*
+ * Domain for untranslated devices - only allocated
+ * if iommu=pt passed on kernel cmd line.
+ */
+static struct protection_domain *pt_domain;
+
+static struct iommu_ops amd_iommu_ops;
+
+/*
+ * general struct to manage commands send to an IOMMU
+ */
+struct iommu_cmd {
+	u32 data[4];
+};
+
+static void update_domain(struct protection_domain *domain);
+
+/****************************************************************************
+ *
+ * Helper functions
+ *
+ ****************************************************************************/
+
+static struct iommu_dev_data *alloc_dev_data(u16 devid)
+{
+	struct iommu_dev_data *dev_data;
+	unsigned long flags;
+
+	dev_data = kzalloc(sizeof(*dev_data), GFP_KERNEL);
+	if (!dev_data)
+		return NULL;
+
+	dev_data->devid = devid;
+	atomic_set(&dev_data->bind, 0);
+
+	spin_lock_irqsave(&dev_data_list_lock, flags);
+	list_add_tail(&dev_data->dev_data_list, &dev_data_list);
+	spin_unlock_irqrestore(&dev_data_list_lock, flags);
+
+	return dev_data;
+}
+
+static void free_dev_data(struct iommu_dev_data *dev_data)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&dev_data_list_lock, flags);
+	list_del(&dev_data->dev_data_list);
+	spin_unlock_irqrestore(&dev_data_list_lock, flags);
+
+	kfree(dev_data);
+}
+
+static struct iommu_dev_data *search_dev_data(u16 devid)
+{
+	struct iommu_dev_data *dev_data;
+	unsigned long flags;
+
+	spin_lock_irqsave(&dev_data_list_lock, flags);
+	list_for_each_entry(dev_data, &dev_data_list, dev_data_list) {
+		if (dev_data->devid == devid)
+			goto out_unlock;
+	}
+
+	dev_data = NULL;
+
+out_unlock:
+	spin_unlock_irqrestore(&dev_data_list_lock, flags);
+
+	return dev_data;
+}
+
+static struct iommu_dev_data *find_dev_data(u16 devid)
+{
+	struct iommu_dev_data *dev_data;
+
+	dev_data = search_dev_data(devid);
+
+	if (dev_data == NULL)
+		dev_data = alloc_dev_data(devid);
+
+	return dev_data;
+}
+
+static inline u16 get_device_id(struct device *dev)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+
+	return calc_devid(pdev->bus->number, pdev->devfn);
+}
+
+static struct iommu_dev_data *get_dev_data(struct device *dev)
+{
+	return dev->archdata.iommu;
+}
+
+/*
+ * In this function the list of preallocated protection domains is traversed to
+ * find the domain for a specific device
+ */
+static struct dma_ops_domain *find_protection_domain(u16 devid)
+{
+	struct dma_ops_domain *entry, *ret = NULL;
+	unsigned long flags;
+	u16 alias = amd_iommu_alias_table[devid];
+
+	if (list_empty(&iommu_pd_list))
+		return NULL;
+
+	spin_lock_irqsave(&iommu_pd_list_lock, flags);
+
+	list_for_each_entry(entry, &iommu_pd_list, list) {
+		if (entry->target_dev == devid ||
+		    entry->target_dev == alias) {
+			ret = entry;
+			break;
+		}
+	}
+
+	spin_unlock_irqrestore(&iommu_pd_list_lock, flags);
+
+	return ret;
+}
+
+/*
+ * This function checks if the driver got a valid device from the caller to
+ * avoid dereferencing invalid pointers.
+ */
+static bool check_device(struct device *dev)
+{
+	u16 devid;
+
+	if (!dev || !dev->dma_mask)
+		return false;
+
+	/* No device or no PCI device */
+	if (dev->bus != &pci_bus_type)
+		return false;
+
+	devid = get_device_id(dev);
+
+	/* Out of our scope? */
+	if (devid > amd_iommu_last_bdf)
+		return false;
+
+	if (amd_iommu_rlookup_table[devid] == NULL)
+		return false;
+
+	return true;
+}
+
+static int iommu_init_device(struct device *dev)
+{
+	struct iommu_dev_data *dev_data;
+	u16 alias;
+
+	if (dev->archdata.iommu)
+		return 0;
+
+	dev_data = find_dev_data(get_device_id(dev));
+	if (!dev_data)
+		return -ENOMEM;
+
+	alias = amd_iommu_alias_table[dev_data->devid];
+	if (alias != dev_data->devid) {
+		struct iommu_dev_data *alias_data;
+
+		alias_data = find_dev_data(alias);
+		if (alias_data == NULL) {
+			pr_err("AMD-Vi: Warning: Unhandled device %s\n",
+					dev_name(dev));
+			free_dev_data(dev_data);
+			return -ENOTSUPP;
+		}
+		dev_data->alias_data = alias_data;
+	}
+
+	dev->archdata.iommu = dev_data;
+
+	return 0;
+}
+
+static void iommu_ignore_device(struct device *dev)
+{
+	u16 devid, alias;
+
+	devid = get_device_id(dev);
+	alias = amd_iommu_alias_table[devid];
+
+	memset(&amd_iommu_dev_table[devid], 0, sizeof(struct dev_table_entry));
+	memset(&amd_iommu_dev_table[alias], 0, sizeof(struct dev_table_entry));
+
+	amd_iommu_rlookup_table[devid] = NULL;
+	amd_iommu_rlookup_table[alias] = NULL;
+}
+
+static void iommu_uninit_device(struct device *dev)
+{
+	/*
+	 * Nothing to do here - we keep dev_data around for unplugged devices
+	 * and reuse it when the device is re-plugged - not doing so would
+	 * introduce a ton of races.
+	 */
+}
+
+void __init amd_iommu_uninit_devices(void)
+{
+	struct iommu_dev_data *dev_data, *n;
+	struct pci_dev *pdev = NULL;
+
+	for_each_pci_dev(pdev) {
+
+		if (!check_device(&pdev->dev))
+			continue;
+
+		iommu_uninit_device(&pdev->dev);
+	}
+
+	/* Free all of our dev_data structures */
+	list_for_each_entry_safe(dev_data, n, &dev_data_list, dev_data_list)
+		free_dev_data(dev_data);
+}
+
+int __init amd_iommu_init_devices(void)
+{
+	struct pci_dev *pdev = NULL;
+	int ret = 0;
+
+	for_each_pci_dev(pdev) {
+
+		if (!check_device(&pdev->dev))
+			continue;
+
+		ret = iommu_init_device(&pdev->dev);
+		if (ret == -ENOTSUPP)
+			iommu_ignore_device(&pdev->dev);
+		else if (ret)
+			goto out_free;
+	}
+
+	return 0;
+
+out_free:
+
+	amd_iommu_uninit_devices();
+
+	return ret;
+}
+#ifdef CONFIG_AMD_IOMMU_STATS
+
+/*
+ * Initialization code for statistics collection
+ */
+
+DECLARE_STATS_COUNTER(compl_wait);
+DECLARE_STATS_COUNTER(cnt_map_single);
+DECLARE_STATS_COUNTER(cnt_unmap_single);
+DECLARE_STATS_COUNTER(cnt_map_sg);
+DECLARE_STATS_COUNTER(cnt_unmap_sg);
+DECLARE_STATS_COUNTER(cnt_alloc_coherent);
+DECLARE_STATS_COUNTER(cnt_free_coherent);
+DECLARE_STATS_COUNTER(cross_page);
+DECLARE_STATS_COUNTER(domain_flush_single);
+DECLARE_STATS_COUNTER(domain_flush_all);
+DECLARE_STATS_COUNTER(alloced_io_mem);
+DECLARE_STATS_COUNTER(total_map_requests);
+
+static struct dentry *stats_dir;
+static struct dentry *de_fflush;
+
+static void amd_iommu_stats_add(struct __iommu_counter *cnt)
+{
+	if (stats_dir == NULL)
+		return;
+
+	cnt->dent = debugfs_create_u64(cnt->name, 0444, stats_dir,
+				       &cnt->value);
+}
+
+static void amd_iommu_stats_init(void)
+{
+	stats_dir = debugfs_create_dir("amd-iommu", NULL);
+	if (stats_dir == NULL)
+		return;
+
+	de_fflush  = debugfs_create_bool("fullflush", 0444, stats_dir,
+					 (u32 *)&amd_iommu_unmap_flush);
+
+	amd_iommu_stats_add(&compl_wait);
+	amd_iommu_stats_add(&cnt_map_single);
+	amd_iommu_stats_add(&cnt_unmap_single);
+	amd_iommu_stats_add(&cnt_map_sg);
+	amd_iommu_stats_add(&cnt_unmap_sg);
+	amd_iommu_stats_add(&cnt_alloc_coherent);
+	amd_iommu_stats_add(&cnt_free_coherent);
+	amd_iommu_stats_add(&cross_page);
+	amd_iommu_stats_add(&domain_flush_single);
+	amd_iommu_stats_add(&domain_flush_all);
+	amd_iommu_stats_add(&alloced_io_mem);
+	amd_iommu_stats_add(&total_map_requests);
+}
+
+#endif
+
+/****************************************************************************
+ *
+ * Interrupt handling functions
+ *
+ ****************************************************************************/
+
+static void dump_dte_entry(u16 devid)
+{
+	int i;
+
+	for (i = 0; i < 8; ++i)
+		pr_err("AMD-Vi: DTE[%d]: %08x\n", i,
+			amd_iommu_dev_table[devid].data[i]);
+}
+
+static void dump_command(unsigned long phys_addr)
+{
+	struct iommu_cmd *cmd = phys_to_virt(phys_addr);
+	int i;
+
+	for (i = 0; i < 4; ++i)
+		pr_err("AMD-Vi: CMD[%d]: %08x\n", i, cmd->data[i]);
+}
+
+static void iommu_print_event(struct amd_iommu *iommu, void *__evt)
+{
+	u32 *event = __evt;
+	int type  = (event[1] >> EVENT_TYPE_SHIFT)  & EVENT_TYPE_MASK;
+	int devid = (event[0] >> EVENT_DEVID_SHIFT) & EVENT_DEVID_MASK;
+	int domid = (event[1] >> EVENT_DOMID_SHIFT) & EVENT_DOMID_MASK;
+	int flags = (event[1] >> EVENT_FLAGS_SHIFT) & EVENT_FLAGS_MASK;
+	u64 address = (u64)(((u64)event[3]) << 32) | event[2];
+
+	printk(KERN_ERR "AMD-Vi: Event logged [");
+
+	switch (type) {
+	case EVENT_TYPE_ILL_DEV:
+		printk("ILLEGAL_DEV_TABLE_ENTRY device=%02x:%02x.%x "
+		       "address=0x%016llx flags=0x%04x]\n",
+		       PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
+		       address, flags);
+		dump_dte_entry(devid);
+		break;
+	case EVENT_TYPE_IO_FAULT:
+		printk("IO_PAGE_FAULT device=%02x:%02x.%x "
+		       "domain=0x%04x address=0x%016llx flags=0x%04x]\n",
+		       PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
+		       domid, address, flags);
+		break;
+	case EVENT_TYPE_DEV_TAB_ERR:
+		printk("DEV_TAB_HARDWARE_ERROR device=%02x:%02x.%x "
+		       "address=0x%016llx flags=0x%04x]\n",
+		       PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
+		       address, flags);
+		break;
+	case EVENT_TYPE_PAGE_TAB_ERR:
+		printk("PAGE_TAB_HARDWARE_ERROR device=%02x:%02x.%x "
+		       "domain=0x%04x address=0x%016llx flags=0x%04x]\n",
+		       PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
+		       domid, address, flags);
+		break;
+	case EVENT_TYPE_ILL_CMD:
+		printk("ILLEGAL_COMMAND_ERROR address=0x%016llx]\n", address);
+		dump_command(address);
+		break;
+	case EVENT_TYPE_CMD_HARD_ERR:
+		printk("COMMAND_HARDWARE_ERROR address=0x%016llx "
+		       "flags=0x%04x]\n", address, flags);
+		break;
+	case EVENT_TYPE_IOTLB_INV_TO:
+		printk("IOTLB_INV_TIMEOUT device=%02x:%02x.%x "
+		       "address=0x%016llx]\n",
+		       PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
+		       address);
+		break;
+	case EVENT_TYPE_INV_DEV_REQ:
+		printk("INVALID_DEVICE_REQUEST device=%02x:%02x.%x "
+		       "address=0x%016llx flags=0x%04x]\n",
+		       PCI_BUS(devid), PCI_SLOT(devid), PCI_FUNC(devid),
+		       address, flags);
+		break;
+	default:
+		printk(KERN_ERR "UNKNOWN type=0x%02x]\n", type);
+	}
+}
+
+static void iommu_poll_events(struct amd_iommu *iommu)
+{
+	u32 head, tail;
+	unsigned long flags;
+
+	spin_lock_irqsave(&iommu->lock, flags);
+
+	head = readl(iommu->mmio_base + MMIO_EVT_HEAD_OFFSET);
+	tail = readl(iommu->mmio_base + MMIO_EVT_TAIL_OFFSET);
+
+	while (head != tail) {
+		iommu_print_event(iommu, iommu->evt_buf + head);
+		head = (head + EVENT_ENTRY_SIZE) % iommu->evt_buf_size;
+	}
+
+	writel(head, iommu->mmio_base + MMIO_EVT_HEAD_OFFSET);
+
+	spin_unlock_irqrestore(&iommu->lock, flags);
+}
+
+irqreturn_t amd_iommu_int_thread(int irq, void *data)
+{
+	struct amd_iommu *iommu;
+
+	for_each_iommu(iommu)
+		iommu_poll_events(iommu);
+
+	return IRQ_HANDLED;
+}
+
+irqreturn_t amd_iommu_int_handler(int irq, void *data)
+{
+	return IRQ_WAKE_THREAD;
+}
+
+/****************************************************************************
+ *
+ * IOMMU command queuing functions
+ *
+ ****************************************************************************/
+
+static int wait_on_sem(volatile u64 *sem)
+{
+	int i = 0;
+
+	while (*sem == 0 && i < LOOP_TIMEOUT) {
+		udelay(1);
+		i += 1;
+	}
+
+	if (i == LOOP_TIMEOUT) {
+		pr_alert("AMD-Vi: Completion-Wait loop timed out\n");
+		return -EIO;
+	}
+
+	return 0;
+}
+
+static void copy_cmd_to_buffer(struct amd_iommu *iommu,
+			       struct iommu_cmd *cmd,
+			       u32 tail)
+{
+	u8 *target;
+
+	target = iommu->cmd_buf + tail;
+	tail   = (tail + sizeof(*cmd)) % iommu->cmd_buf_size;
+
+	/* Copy command to buffer */
+	memcpy(target, cmd, sizeof(*cmd));
+
+	/* Tell the IOMMU about it */
+	writel(tail, iommu->mmio_base + MMIO_CMD_TAIL_OFFSET);
+}
+
+static void build_completion_wait(struct iommu_cmd *cmd, u64 address)
+{
+	WARN_ON(address & 0x7ULL);
+
+	memset(cmd, 0, sizeof(*cmd));
+	cmd->data[0] = lower_32_bits(__pa(address)) | CMD_COMPL_WAIT_STORE_MASK;
+	cmd->data[1] = upper_32_bits(__pa(address));
+	cmd->data[2] = 1;
+	CMD_SET_TYPE(cmd, CMD_COMPL_WAIT);
+}
+
+static void build_inv_dte(struct iommu_cmd *cmd, u16 devid)
+{
+	memset(cmd, 0, sizeof(*cmd));
+	cmd->data[0] = devid;
+	CMD_SET_TYPE(cmd, CMD_INV_DEV_ENTRY);
+}
+
+static void build_inv_iommu_pages(struct iommu_cmd *cmd, u64 address,
+				  size_t size, u16 domid, int pde)
+{
+	u64 pages;
+	int s;
+
+	pages = iommu_num_pages(address, size, PAGE_SIZE);
+	s     = 0;
+
+	if (pages > 1) {
+		/*
+		 * If we have to flush more than one page, flush all
+		 * TLB entries for this domain
+		 */
+		address = CMD_INV_IOMMU_ALL_PAGES_ADDRESS;
+		s = 1;
+	}
+
+	address &= PAGE_MASK;
+
+	memset(cmd, 0, sizeof(*cmd));
+	cmd->data[1] |= domid;
+	cmd->data[2]  = lower_32_bits(address);
+	cmd->data[3]  = upper_32_bits(address);
+	CMD_SET_TYPE(cmd, CMD_INV_IOMMU_PAGES);
+	if (s) /* size bit - we flush more than one 4kb page */
+		cmd->data[2] |= CMD_INV_IOMMU_PAGES_SIZE_MASK;
+	if (pde) /* PDE bit - we wan't flush everything not only the PTEs */
+		cmd->data[2] |= CMD_INV_IOMMU_PAGES_PDE_MASK;
+}
+
+static void build_inv_iotlb_pages(struct iommu_cmd *cmd, u16 devid, int qdep,
+				  u64 address, size_t size)
+{
+	u64 pages;
+	int s;
+
+	pages = iommu_num_pages(address, size, PAGE_SIZE);
+	s     = 0;
+
+	if (pages > 1) {
+		/*
+		 * If we have to flush more than one page, flush all
+		 * TLB entries for this domain
+		 */
+		address = CMD_INV_IOMMU_ALL_PAGES_ADDRESS;
+		s = 1;
+	}
+
+	address &= PAGE_MASK;
+
+	memset(cmd, 0, sizeof(*cmd));
+	cmd->data[0]  = devid;
+	cmd->data[0] |= (qdep & 0xff) << 24;
+	cmd->data[1]  = devid;
+	cmd->data[2]  = lower_32_bits(address);
+	cmd->data[3]  = upper_32_bits(address);
+	CMD_SET_TYPE(cmd, CMD_INV_IOTLB_PAGES);
+	if (s)
+		cmd->data[2] |= CMD_INV_IOMMU_PAGES_SIZE_MASK;
+}
+
+static void build_inv_all(struct iommu_cmd *cmd)
+{
+	memset(cmd, 0, sizeof(*cmd));
+	CMD_SET_TYPE(cmd, CMD_INV_ALL);
+}
+
+/*
+ * Writes the command to the IOMMUs command buffer and informs the
+ * hardware about the new command.
+ */
+static int iommu_queue_command(struct amd_iommu *iommu, struct iommu_cmd *cmd)
+{
+	u32 left, tail, head, next_tail;
+	unsigned long flags;
+
+	WARN_ON(iommu->cmd_buf_size & CMD_BUFFER_UNINITIALIZED);
+
+again:
+	spin_lock_irqsave(&iommu->lock, flags);
+
+	head      = readl(iommu->mmio_base + MMIO_CMD_HEAD_OFFSET);
+	tail      = readl(iommu->mmio_base + MMIO_CMD_TAIL_OFFSET);
+	next_tail = (tail + sizeof(*cmd)) % iommu->cmd_buf_size;
+	left      = (head - next_tail) % iommu->cmd_buf_size;
+
+	if (left <= 2) {
+		struct iommu_cmd sync_cmd;
+		volatile u64 sem = 0;
+		int ret;
+
+		build_completion_wait(&sync_cmd, (u64)&sem);
+		copy_cmd_to_buffer(iommu, &sync_cmd, tail);
+
+		spin_unlock_irqrestore(&iommu->lock, flags);
+
+		if ((ret = wait_on_sem(&sem)) != 0)
+			return ret;
+
+		goto again;
+	}
+
+	copy_cmd_to_buffer(iommu, cmd, tail);
+
+	/* We need to sync now to make sure all commands are processed */
+	iommu->need_sync = true;
+
+	spin_unlock_irqrestore(&iommu->lock, flags);
+
+	return 0;
+}
+
+/*
+ * This function queues a completion wait command into the command
+ * buffer of an IOMMU
+ */
+static int iommu_completion_wait(struct amd_iommu *iommu)
+{
+	struct iommu_cmd cmd;
+	volatile u64 sem = 0;
+	int ret;
+
+	if (!iommu->need_sync)
+		return 0;
+
+	build_completion_wait(&cmd, (u64)&sem);
+
+	ret = iommu_queue_command(iommu, &cmd);
+	if (ret)
+		return ret;
+
+	return wait_on_sem(&sem);
+}
+
+static int iommu_flush_dte(struct amd_iommu *iommu, u16 devid)
+{
+	struct iommu_cmd cmd;
+
+	build_inv_dte(&cmd, devid);
+
+	return iommu_queue_command(iommu, &cmd);
+}
+
+static void iommu_flush_dte_all(struct amd_iommu *iommu)
+{
+	u32 devid;
+
+	for (devid = 0; devid <= 0xffff; ++devid)
+		iommu_flush_dte(iommu, devid);
+
+	iommu_completion_wait(iommu);
+}
+
+/*
+ * This function uses heavy locking and may disable irqs for some time. But
+ * this is no issue because it is only called during resume.
+ */
+static void iommu_flush_tlb_all(struct amd_iommu *iommu)
+{
+	u32 dom_id;
+
+	for (dom_id = 0; dom_id <= 0xffff; ++dom_id) {
+		struct iommu_cmd cmd;
+		build_inv_iommu_pages(&cmd, 0, CMD_INV_IOMMU_ALL_PAGES_ADDRESS,
+				      dom_id, 1);
+		iommu_queue_command(iommu, &cmd);
+	}
+
+	iommu_completion_wait(iommu);
+}
+
+static void iommu_flush_all(struct amd_iommu *iommu)
+{
+	struct iommu_cmd cmd;
+
+	build_inv_all(&cmd);
+
+	iommu_queue_command(iommu, &cmd);
+	iommu_completion_wait(iommu);
+}
+
+void iommu_flush_all_caches(struct amd_iommu *iommu)
+{
+	if (iommu_feature(iommu, FEATURE_IA)) {
+		iommu_flush_all(iommu);
+	} else {
+		iommu_flush_dte_all(iommu);
+		iommu_flush_tlb_all(iommu);
+	}
+}
+
+/*
+ * Command send function for flushing on-device TLB
+ */
+static int device_flush_iotlb(struct iommu_dev_data *dev_data,
+			      u64 address, size_t size)
+{
+	struct amd_iommu *iommu;
+	struct iommu_cmd cmd;
+	int qdep;
+
+	qdep     = dev_data->ats.qdep;
+	iommu    = amd_iommu_rlookup_table[dev_data->devid];
+
+	build_inv_iotlb_pages(&cmd, dev_data->devid, qdep, address, size);
+
+	return iommu_queue_command(iommu, &cmd);
+}
+
+/*
+ * Command send function for invalidating a device table entry
+ */
+static int device_flush_dte(struct iommu_dev_data *dev_data)
+{
+	struct amd_iommu *iommu;
+	int ret;
+
+	iommu = amd_iommu_rlookup_table[dev_data->devid];
+
+	ret = iommu_flush_dte(iommu, dev_data->devid);
+	if (ret)
+		return ret;
+
+	if (dev_data->ats.enabled)
+		ret = device_flush_iotlb(dev_data, 0, ~0UL);
+
+	return ret;
+}
+
+/*
+ * TLB invalidation function which is called from the mapping functions.
+ * It invalidates a single PTE if the range to flush is within a single
+ * page. Otherwise it flushes the whole TLB of the IOMMU.
+ */
+static void __domain_flush_pages(struct protection_domain *domain,
+				 u64 address, size_t size, int pde)
+{
+	struct iommu_dev_data *dev_data;
+	struct iommu_cmd cmd;
+	int ret = 0, i;
+
+	build_inv_iommu_pages(&cmd, address, size, domain->id, pde);
+
+	for (i = 0; i < amd_iommus_present; ++i) {
+		if (!domain->dev_iommu[i])
+			continue;
+
+		/*
+		 * Devices of this domain are behind this IOMMU
+		 * We need a TLB flush
+		 */
+		ret |= iommu_queue_command(amd_iommus[i], &cmd);
+	}
+
+	list_for_each_entry(dev_data, &domain->dev_list, list) {
+
+		if (!dev_data->ats.enabled)
+			continue;
+
+		ret |= device_flush_iotlb(dev_data, address, size);
+	}
+
+	WARN_ON(ret);
+}
+
+static void domain_flush_pages(struct protection_domain *domain,
+			       u64 address, size_t size)
+{
+	__domain_flush_pages(domain, address, size, 0);
+}
+
+/* Flush the whole IO/TLB for a given protection domain */
+static void domain_flush_tlb(struct protection_domain *domain)
+{
+	__domain_flush_pages(domain, 0, CMD_INV_IOMMU_ALL_PAGES_ADDRESS, 0);
+}
+
+/* Flush the whole IO/TLB for a given protection domain - including PDE */
+static void domain_flush_tlb_pde(struct protection_domain *domain)
+{
+	__domain_flush_pages(domain, 0, CMD_INV_IOMMU_ALL_PAGES_ADDRESS, 1);
+}
+
+static void domain_flush_complete(struct protection_domain *domain)
+{
+	int i;
+
+	for (i = 0; i < amd_iommus_present; ++i) {
+		if (!domain->dev_iommu[i])
+			continue;
+
+		/*
+		 * Devices of this domain are behind this IOMMU
+		 * We need to wait for completion of all commands.
+		 */
+		iommu_completion_wait(amd_iommus[i]);
+	}
+}
+
+
+/*
+ * This function flushes the DTEs for all devices in domain
+ */
+static void domain_flush_devices(struct protection_domain *domain)
+{
+	struct iommu_dev_data *dev_data;
+	unsigned long flags;
+
+	spin_lock_irqsave(&domain->lock, flags);
+
+	list_for_each_entry(dev_data, &domain->dev_list, list)
+		device_flush_dte(dev_data);
+
+	spin_unlock_irqrestore(&domain->lock, flags);
+}
+
+/****************************************************************************
+ *
+ * The functions below are used the create the page table mappings for
+ * unity mapped regions.
+ *
+ ****************************************************************************/
+
+/*
+ * This function is used to add another level to an IO page table. Adding
+ * another level increases the size of the address space by 9 bits to a size up
+ * to 64 bits.
+ */
+static bool increase_address_space(struct protection_domain *domain,
+				   gfp_t gfp)
+{
+	u64 *pte;
+
+	if (domain->mode == PAGE_MODE_6_LEVEL)
+		/* address space already 64 bit large */
+		return false;
+
+	pte = (void *)get_zeroed_page(gfp);
+	if (!pte)
+		return false;
+
+	*pte             = PM_LEVEL_PDE(domain->mode,
+					virt_to_phys(domain->pt_root));
+	domain->pt_root  = pte;
+	domain->mode    += 1;
+	domain->updated  = true;
+
+	return true;
+}
+
+static u64 *alloc_pte(struct protection_domain *domain,
+		      unsigned long address,
+		      unsigned long page_size,
+		      u64 **pte_page,
+		      gfp_t gfp)
+{
+	int level, end_lvl;
+	u64 *pte, *page;
+
+	BUG_ON(!is_power_of_2(page_size));
+
+	while (address > PM_LEVEL_SIZE(domain->mode))
+		increase_address_space(domain, gfp);
+
+	level   = domain->mode - 1;
+	pte     = &domain->pt_root[PM_LEVEL_INDEX(level, address)];
+	address = PAGE_SIZE_ALIGN(address, page_size);
+	end_lvl = PAGE_SIZE_LEVEL(page_size);
+
+	while (level > end_lvl) {
+		if (!IOMMU_PTE_PRESENT(*pte)) {
+			page = (u64 *)get_zeroed_page(gfp);
+			if (!page)
+				return NULL;
+			*pte = PM_LEVEL_PDE(level, virt_to_phys(page));
+		}
+
+		/* No level skipping support yet */
+		if (PM_PTE_LEVEL(*pte) != level)
+			return NULL;
+
+		level -= 1;
+
+		pte = IOMMU_PTE_PAGE(*pte);
+
+		if (pte_page && level == end_lvl)
+			*pte_page = pte;
+
+		pte = &pte[PM_LEVEL_INDEX(level, address)];
+	}
+
+	return pte;
+}
+
+/*
+ * This function checks if there is a PTE for a given dma address. If
+ * there is one, it returns the pointer to it.
+ */
+static u64 *fetch_pte(struct protection_domain *domain, unsigned long address)
+{
+	int level;
+	u64 *pte;
+
+	if (address > PM_LEVEL_SIZE(domain->mode))
+		return NULL;
+
+	level   =  domain->mode - 1;
+	pte     = &domain->pt_root[PM_LEVEL_INDEX(level, address)];
+
+	while (level > 0) {
+
+		/* Not Present */
+		if (!IOMMU_PTE_PRESENT(*pte))
+			return NULL;
+
+		/* Large PTE */
+		if (PM_PTE_LEVEL(*pte) == 0x07) {
+			unsigned long pte_mask, __pte;
+
+			/*
+			 * If we have a series of large PTEs, make
+			 * sure to return a pointer to the first one.
+			 */
+			pte_mask = PTE_PAGE_SIZE(*pte);
+			pte_mask = ~((PAGE_SIZE_PTE_COUNT(pte_mask) << 3) - 1);
+			__pte    = ((unsigned long)pte) & pte_mask;
+
+			return (u64 *)__pte;
+		}
+
+		/* No level skipping support yet */
+		if (PM_PTE_LEVEL(*pte) != level)
+			return NULL;
+
+		level -= 1;
+
+		/* Walk to the next level */
+		pte = IOMMU_PTE_PAGE(*pte);
+		pte = &pte[PM_LEVEL_INDEX(level, address)];
+	}
+
+	return pte;
+}
+
+/*
+ * Generic mapping functions. It maps a physical address into a DMA
+ * address space. It allocates the page table pages if necessary.
+ * In the future it can be extended to a generic mapping function
+ * supporting all features of AMD IOMMU page tables like level skipping
+ * and full 64 bit address spaces.
+ */
+static int iommu_map_page(struct protection_domain *dom,
+			  unsigned long bus_addr,
+			  unsigned long phys_addr,
+			  int prot,
+			  unsigned long page_size)
+{
+	u64 __pte, *pte;
+	int i, count;
+
+	if (!(prot & IOMMU_PROT_MASK))
+		return -EINVAL;
+
+	bus_addr  = PAGE_ALIGN(bus_addr);
+	phys_addr = PAGE_ALIGN(phys_addr);
+	count     = PAGE_SIZE_PTE_COUNT(page_size);
+	pte       = alloc_pte(dom, bus_addr, page_size, NULL, GFP_KERNEL);
+
+	for (i = 0; i < count; ++i)
+		if (IOMMU_PTE_PRESENT(pte[i]))
+			return -EBUSY;
+
+	if (page_size > PAGE_SIZE) {
+		__pte = PAGE_SIZE_PTE(phys_addr, page_size);
+		__pte |= PM_LEVEL_ENC(7) | IOMMU_PTE_P | IOMMU_PTE_FC;
+	} else
+		__pte = phys_addr | IOMMU_PTE_P | IOMMU_PTE_FC;
+
+	if (prot & IOMMU_PROT_IR)
+		__pte |= IOMMU_PTE_IR;
+	if (prot & IOMMU_PROT_IW)
+		__pte |= IOMMU_PTE_IW;
+
+	for (i = 0; i < count; ++i)
+		pte[i] = __pte;
+
+	update_domain(dom);
+
+	return 0;
+}
+
+static unsigned long iommu_unmap_page(struct protection_domain *dom,
+				      unsigned long bus_addr,
+				      unsigned long page_size)
+{
+	unsigned long long unmap_size, unmapped;
+	u64 *pte;
+
+	BUG_ON(!is_power_of_2(page_size));
+
+	unmapped = 0;
+
+	while (unmapped < page_size) {
+
+		pte = fetch_pte(dom, bus_addr);
+
+		if (!pte) {
+			/*
+			 * No PTE for this address
+			 * move forward in 4kb steps
+			 */
+			unmap_size = PAGE_SIZE;
+		} else if (PM_PTE_LEVEL(*pte) == 0) {
+			/* 4kb PTE found for this address */
+			unmap_size = PAGE_SIZE;
+			*pte       = 0ULL;
+		} else {
+			int count, i;
+
+			/* Large PTE found which maps this address */
+			unmap_size = PTE_PAGE_SIZE(*pte);
+			count      = PAGE_SIZE_PTE_COUNT(unmap_size);
+			for (i = 0; i < count; i++)
+				pte[i] = 0ULL;
+		}
+
+		bus_addr  = (bus_addr & ~(unmap_size - 1)) + unmap_size;
+		unmapped += unmap_size;
+	}
+
+	BUG_ON(!is_power_of_2(unmapped));
+
+	return unmapped;
+}
+
+/*
+ * This function checks if a specific unity mapping entry is needed for
+ * this specific IOMMU.
+ */
+static int iommu_for_unity_map(struct amd_iommu *iommu,
+			       struct unity_map_entry *entry)
+{
+	u16 bdf, i;
+
+	for (i = entry->devid_start; i <= entry->devid_end; ++i) {
+		bdf = amd_iommu_alias_table[i];
+		if (amd_iommu_rlookup_table[bdf] == iommu)
+			return 1;
+	}
+
+	return 0;
+}
+
+/*
+ * This function actually applies the mapping to the page table of the
+ * dma_ops domain.
+ */
+static int dma_ops_unity_map(struct dma_ops_domain *dma_dom,
+			     struct unity_map_entry *e)
+{
+	u64 addr;
+	int ret;
+
+	for (addr = e->address_start; addr < e->address_end;
+	     addr += PAGE_SIZE) {
+		ret = iommu_map_page(&dma_dom->domain, addr, addr, e->prot,
+				     PAGE_SIZE);
+		if (ret)
+			return ret;
+		/*
+		 * if unity mapping is in aperture range mark the page
+		 * as allocated in the aperture
+		 */
+		if (addr < dma_dom->aperture_size)
+			__set_bit(addr >> PAGE_SHIFT,
+				  dma_dom->aperture[0]->bitmap);
+	}
+
+	return 0;
+}
+
+/*
+ * Init the unity mappings for a specific IOMMU in the system
+ *
+ * Basically iterates over all unity mapping entries and applies them to
+ * the default domain DMA of that IOMMU if necessary.
+ */
+static int iommu_init_unity_mappings(struct amd_iommu *iommu)
+{
+	struct unity_map_entry *entry;
+	int ret;
+
+	list_for_each_entry(entry, &amd_iommu_unity_map, list) {
+		if (!iommu_for_unity_map(iommu, entry))
+			continue;
+		ret = dma_ops_unity_map(iommu->default_dom, entry);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+/*
+ * Inits the unity mappings required for a specific device
+ */
+static int init_unity_mappings_for_device(struct dma_ops_domain *dma_dom,
+					  u16 devid)
+{
+	struct unity_map_entry *e;
+	int ret;
+
+	list_for_each_entry(e, &amd_iommu_unity_map, list) {
+		if (!(devid >= e->devid_start && devid <= e->devid_end))
+			continue;
+		ret = dma_ops_unity_map(dma_dom, e);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+/****************************************************************************
+ *
+ * The next functions belong to the address allocator for the dma_ops
+ * interface functions. They work like the allocators in the other IOMMU
+ * drivers. Its basically a bitmap which marks the allocated pages in
+ * the aperture. Maybe it could be enhanced in the future to a more
+ * efficient allocator.
+ *
+ ****************************************************************************/
+
+/*
+ * The address allocator core functions.
+ *
+ * called with domain->lock held
+ */
+
+/*
+ * Used to reserve address ranges in the aperture (e.g. for exclusion
+ * ranges.
+ */
+static void dma_ops_reserve_addresses(struct dma_ops_domain *dom,
+				      unsigned long start_page,
+				      unsigned int pages)
+{
+	unsigned int i, last_page = dom->aperture_size >> PAGE_SHIFT;
+
+	if (start_page + pages > last_page)
+		pages = last_page - start_page;
+
+	for (i = start_page; i < start_page + pages; ++i) {
+		int index = i / APERTURE_RANGE_PAGES;
+		int page  = i % APERTURE_RANGE_PAGES;
+		__set_bit(page, dom->aperture[index]->bitmap);
+	}
+}
+
+/*
+ * This function is used to add a new aperture range to an existing
+ * aperture in case of dma_ops domain allocation or address allocation
+ * failure.
+ */
+static int alloc_new_range(struct dma_ops_domain *dma_dom,
+			   bool populate, gfp_t gfp)
+{
+	int index = dma_dom->aperture_size >> APERTURE_RANGE_SHIFT;
+	struct amd_iommu *iommu;
+	unsigned long i, old_size;
+
+#ifdef CONFIG_IOMMU_STRESS
+	populate = false;
+#endif
+
+	if (index >= APERTURE_MAX_RANGES)
+		return -ENOMEM;
+
+	dma_dom->aperture[index] = kzalloc(sizeof(struct aperture_range), gfp);
+	if (!dma_dom->aperture[index])
+		return -ENOMEM;
+
+	dma_dom->aperture[index]->bitmap = (void *)get_zeroed_page(gfp);
+	if (!dma_dom->aperture[index]->bitmap)
+		goto out_free;
+
+	dma_dom->aperture[index]->offset = dma_dom->aperture_size;
+
+	if (populate) {
+		unsigned long address = dma_dom->aperture_size;
+		int i, num_ptes = APERTURE_RANGE_PAGES / 512;
+		u64 *pte, *pte_page;
+
+		for (i = 0; i < num_ptes; ++i) {
+			pte = alloc_pte(&dma_dom->domain, address, PAGE_SIZE,
+					&pte_page, gfp);
+			if (!pte)
+				goto out_free;
+
+			dma_dom->aperture[index]->pte_pages[i] = pte_page;
+
+			address += APERTURE_RANGE_SIZE / 64;
+		}
+	}
+
+	old_size                = dma_dom->aperture_size;
+	dma_dom->aperture_size += APERTURE_RANGE_SIZE;
+
+	/* Reserve address range used for MSI messages */
+	if (old_size < MSI_ADDR_BASE_LO &&
+	    dma_dom->aperture_size > MSI_ADDR_BASE_LO) {
+		unsigned long spage;
+		int pages;
+
+		pages = iommu_num_pages(MSI_ADDR_BASE_LO, 0x10000, PAGE_SIZE);
+		spage = MSI_ADDR_BASE_LO >> PAGE_SHIFT;
+
+		dma_ops_reserve_addresses(dma_dom, spage, pages);
+	}
+
+	/* Initialize the exclusion range if necessary */
+	for_each_iommu(iommu) {
+		if (iommu->exclusion_start &&
+		    iommu->exclusion_start >= dma_dom->aperture[index]->offset
+		    && iommu->exclusion_start < dma_dom->aperture_size) {
+			unsigned long startpage;
+			int pages = iommu_num_pages(iommu->exclusion_start,
+						    iommu->exclusion_length,
+						    PAGE_SIZE);
+			startpage = iommu->exclusion_start >> PAGE_SHIFT;
+			dma_ops_reserve_addresses(dma_dom, startpage, pages);
+		}
+	}
+
+	/*
+	 * Check for areas already mapped as present in the new aperture
+	 * range and mark those pages as reserved in the allocator. Such
+	 * mappings may already exist as a result of requested unity
+	 * mappings for devices.
+	 */
+	for (i = dma_dom->aperture[index]->offset;
+	     i < dma_dom->aperture_size;
+	     i += PAGE_SIZE) {
+		u64 *pte = fetch_pte(&dma_dom->domain, i);
+		if (!pte || !IOMMU_PTE_PRESENT(*pte))
+			continue;
+
+		dma_ops_reserve_addresses(dma_dom, i << PAGE_SHIFT, 1);
+	}
+
+	update_domain(&dma_dom->domain);
+
+	return 0;
+
+out_free:
+	update_domain(&dma_dom->domain);
+
+	free_page((unsigned long)dma_dom->aperture[index]->bitmap);
+
+	kfree(dma_dom->aperture[index]);
+	dma_dom->aperture[index] = NULL;
+
+	return -ENOMEM;
+}
+
+static unsigned long dma_ops_area_alloc(struct device *dev,
+					struct dma_ops_domain *dom,
+					unsigned int pages,
+					unsigned long align_mask,
+					u64 dma_mask,
+					unsigned long start)
+{
+	unsigned long next_bit = dom->next_address % APERTURE_RANGE_SIZE;
+	int max_index = dom->aperture_size >> APERTURE_RANGE_SHIFT;
+	int i = start >> APERTURE_RANGE_SHIFT;
+	unsigned long boundary_size;
+	unsigned long address = -1;
+	unsigned long limit;
+
+	next_bit >>= PAGE_SHIFT;
+
+	boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1,
+			PAGE_SIZE) >> PAGE_SHIFT;
+
+	for (;i < max_index; ++i) {
+		unsigned long offset = dom->aperture[i]->offset >> PAGE_SHIFT;
+
+		if (dom->aperture[i]->offset >= dma_mask)
+			break;
+
+		limit = iommu_device_max_index(APERTURE_RANGE_PAGES, offset,
+					       dma_mask >> PAGE_SHIFT);
+
+		address = iommu_area_alloc(dom->aperture[i]->bitmap,
+					   limit, next_bit, pages, 0,
+					    boundary_size, align_mask);
+		if (address != -1) {
+			address = dom->aperture[i]->offset +
+				  (address << PAGE_SHIFT);
+			dom->next_address = address + (pages << PAGE_SHIFT);
+			break;
+		}
+
+		next_bit = 0;
+	}
+
+	return address;
+}
+
+static unsigned long dma_ops_alloc_addresses(struct device *dev,
+					     struct dma_ops_domain *dom,
+					     unsigned int pages,
+					     unsigned long align_mask,
+					     u64 dma_mask)
+{
+	unsigned long address;
+
+#ifdef CONFIG_IOMMU_STRESS
+	dom->next_address = 0;
+	dom->need_flush = true;
+#endif
+
+	address = dma_ops_area_alloc(dev, dom, pages, align_mask,
+				     dma_mask, dom->next_address);
+
+	if (address == -1) {
+		dom->next_address = 0;
+		address = dma_ops_area_alloc(dev, dom, pages, align_mask,
+					     dma_mask, 0);
+		dom->need_flush = true;
+	}
+
+	if (unlikely(address == -1))
+		address = DMA_ERROR_CODE;
+
+	WARN_ON((address + (PAGE_SIZE*pages)) > dom->aperture_size);
+
+	return address;
+}
+
+/*
+ * The address free function.
+ *
+ * called with domain->lock held
+ */
+static void dma_ops_free_addresses(struct dma_ops_domain *dom,
+				   unsigned long address,
+				   unsigned int pages)
+{
+	unsigned i = address >> APERTURE_RANGE_SHIFT;
+	struct aperture_range *range = dom->aperture[i];
+
+	BUG_ON(i >= APERTURE_MAX_RANGES || range == NULL);
+
+#ifdef CONFIG_IOMMU_STRESS
+	if (i < 4)
+		return;
+#endif
+
+	if (address >= dom->next_address)
+		dom->need_flush = true;
+
+	address = (address % APERTURE_RANGE_SIZE) >> PAGE_SHIFT;
+
+	bitmap_clear(range->bitmap, address, pages);
+
+}
+
+/****************************************************************************
+ *
+ * The next functions belong to the domain allocation. A domain is
+ * allocated for every IOMMU as the default domain. If device isolation
+ * is enabled, every device get its own domain. The most important thing
+ * about domains is the page table mapping the DMA address space they
+ * contain.
+ *
+ ****************************************************************************/
+
+/*
+ * This function adds a protection domain to the global protection domain list
+ */
+static void add_domain_to_list(struct protection_domain *domain)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&amd_iommu_pd_lock, flags);
+	list_add(&domain->list, &amd_iommu_pd_list);
+	spin_unlock_irqrestore(&amd_iommu_pd_lock, flags);
+}
+
+/*
+ * This function removes a protection domain to the global
+ * protection domain list
+ */
+static void del_domain_from_list(struct protection_domain *domain)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&amd_iommu_pd_lock, flags);
+	list_del(&domain->list);
+	spin_unlock_irqrestore(&amd_iommu_pd_lock, flags);
+}
+
+static u16 domain_id_alloc(void)
+{
+	unsigned long flags;
+	int id;
+
+	write_lock_irqsave(&amd_iommu_devtable_lock, flags);
+	id = find_first_zero_bit(amd_iommu_pd_alloc_bitmap, MAX_DOMAIN_ID);
+	BUG_ON(id == 0);
+	if (id > 0 && id < MAX_DOMAIN_ID)
+		__set_bit(id, amd_iommu_pd_alloc_bitmap);
+	else
+		id = 0;
+	write_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
+
+	return id;
+}
+
+static void domain_id_free(int id)
+{
+	unsigned long flags;
+
+	write_lock_irqsave(&amd_iommu_devtable_lock, flags);
+	if (id > 0 && id < MAX_DOMAIN_ID)
+		__clear_bit(id, amd_iommu_pd_alloc_bitmap);
+	write_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
+}
+
+static void free_pagetable(struct protection_domain *domain)
+{
+	int i, j;
+	u64 *p1, *p2, *p3;
+
+	p1 = domain->pt_root;
+
+	if (!p1)
+		return;
+
+	for (i = 0; i < 512; ++i) {
+		if (!IOMMU_PTE_PRESENT(p1[i]))
+			continue;
+
+		p2 = IOMMU_PTE_PAGE(p1[i]);
+		for (j = 0; j < 512; ++j) {
+			if (!IOMMU_PTE_PRESENT(p2[j]))
+				continue;
+			p3 = IOMMU_PTE_PAGE(p2[j]);
+			free_page((unsigned long)p3);
+		}
+
+		free_page((unsigned long)p2);
+	}
+
+	free_page((unsigned long)p1);
+
+	domain->pt_root = NULL;
+}
+
+/*
+ * Free a domain, only used if something went wrong in the
+ * allocation path and we need to free an already allocated page table
+ */
+static void dma_ops_domain_free(struct dma_ops_domain *dom)
+{
+	int i;
+
+	if (!dom)
+		return;
+
+	del_domain_from_list(&dom->domain);
+
+	free_pagetable(&dom->domain);
+
+	for (i = 0; i < APERTURE_MAX_RANGES; ++i) {
+		if (!dom->aperture[i])
+			continue;
+		free_page((unsigned long)dom->aperture[i]->bitmap);
+		kfree(dom->aperture[i]);
+	}
+
+	kfree(dom);
+}
+
+/*
+ * Allocates a new protection domain usable for the dma_ops functions.
+ * It also initializes the page table and the address allocator data
+ * structures required for the dma_ops interface
+ */
+static struct dma_ops_domain *dma_ops_domain_alloc(void)
+{
+	struct dma_ops_domain *dma_dom;
+
+	dma_dom = kzalloc(sizeof(struct dma_ops_domain), GFP_KERNEL);
+	if (!dma_dom)
+		return NULL;
+
+	spin_lock_init(&dma_dom->domain.lock);
+
+	dma_dom->domain.id = domain_id_alloc();
+	if (dma_dom->domain.id == 0)
+		goto free_dma_dom;
+	INIT_LIST_HEAD(&dma_dom->domain.dev_list);
+	dma_dom->domain.mode = PAGE_MODE_2_LEVEL;
+	dma_dom->domain.pt_root = (void *)get_zeroed_page(GFP_KERNEL);
+	dma_dom->domain.flags = PD_DMA_OPS_MASK;
+	dma_dom->domain.priv = dma_dom;
+	if (!dma_dom->domain.pt_root)
+		goto free_dma_dom;
+
+	dma_dom->need_flush = false;
+	dma_dom->target_dev = 0xffff;
+
+	add_domain_to_list(&dma_dom->domain);
+
+	if (alloc_new_range(dma_dom, true, GFP_KERNEL))
+		goto free_dma_dom;
+
+	/*
+	 * mark the first page as allocated so we never return 0 as
+	 * a valid dma-address. So we can use 0 as error value
+	 */
+	dma_dom->aperture[0]->bitmap[0] = 1;
+	dma_dom->next_address = 0;
+
+
+	return dma_dom;
+
+free_dma_dom:
+	dma_ops_domain_free(dma_dom);
+
+	return NULL;
+}
+
+/*
+ * little helper function to check whether a given protection domain is a
+ * dma_ops domain
+ */
+static bool dma_ops_domain(struct protection_domain *domain)
+{
+	return domain->flags & PD_DMA_OPS_MASK;
+}
+
+static void set_dte_entry(u16 devid, struct protection_domain *domain, bool ats)
+{
+	u64 pte_root = virt_to_phys(domain->pt_root);
+	u32 flags = 0;
+
+	pte_root |= (domain->mode & DEV_ENTRY_MODE_MASK)
+		    << DEV_ENTRY_MODE_SHIFT;
+	pte_root |= IOMMU_PTE_IR | IOMMU_PTE_IW | IOMMU_PTE_P | IOMMU_PTE_TV;
+
+	if (ats)
+		flags |= DTE_FLAG_IOTLB;
+
+	amd_iommu_dev_table[devid].data[3] |= flags;
+	amd_iommu_dev_table[devid].data[2]  = domain->id;
+	amd_iommu_dev_table[devid].data[1]  = upper_32_bits(pte_root);
+	amd_iommu_dev_table[devid].data[0]  = lower_32_bits(pte_root);
+}
+
+static void clear_dte_entry(u16 devid)
+{
+	/* remove entry from the device table seen by the hardware */
+	amd_iommu_dev_table[devid].data[0] = IOMMU_PTE_P | IOMMU_PTE_TV;
+	amd_iommu_dev_table[devid].data[1] = 0;
+	amd_iommu_dev_table[devid].data[2] = 0;
+
+	amd_iommu_apply_erratum_63(devid);
+}
+
+static void do_attach(struct iommu_dev_data *dev_data,
+		      struct protection_domain *domain)
+{
+	struct amd_iommu *iommu;
+	bool ats;
+
+	iommu = amd_iommu_rlookup_table[dev_data->devid];
+	ats   = dev_data->ats.enabled;
+
+	/* Update data structures */
+	dev_data->domain = domain;
+	list_add(&dev_data->list, &domain->dev_list);
+	set_dte_entry(dev_data->devid, domain, ats);
+
+	/* Do reference counting */
+	domain->dev_iommu[iommu->index] += 1;
+	domain->dev_cnt                 += 1;
+
+	/* Flush the DTE entry */
+	device_flush_dte(dev_data);
+}
+
+static void do_detach(struct iommu_dev_data *dev_data)
+{
+	struct amd_iommu *iommu;
+
+	iommu = amd_iommu_rlookup_table[dev_data->devid];
+
+	/* decrease reference counters */
+	dev_data->domain->dev_iommu[iommu->index] -= 1;
+	dev_data->domain->dev_cnt                 -= 1;
+
+	/* Update data structures */
+	dev_data->domain = NULL;
+	list_del(&dev_data->list);
+	clear_dte_entry(dev_data->devid);
+
+	/* Flush the DTE entry */
+	device_flush_dte(dev_data);
+}
+
+/*
+ * If a device is not yet associated with a domain, this function does
+ * assigns it visible for the hardware
+ */
+static int __attach_device(struct iommu_dev_data *dev_data,
+			   struct protection_domain *domain)
+{
+	int ret;
+
+	/* lock domain */
+	spin_lock(&domain->lock);
+
+	if (dev_data->alias_data != NULL) {
+		struct iommu_dev_data *alias_data = dev_data->alias_data;
+
+		/* Some sanity checks */
+		ret = -EBUSY;
+		if (alias_data->domain != NULL &&
+				alias_data->domain != domain)
+			goto out_unlock;
+
+		if (dev_data->domain != NULL &&
+				dev_data->domain != domain)
+			goto out_unlock;
+
+		/* Do real assignment */
+		if (alias_data->domain == NULL)
+			do_attach(alias_data, domain);
+
+		atomic_inc(&alias_data->bind);
+	}
+
+	if (dev_data->domain == NULL)
+		do_attach(dev_data, domain);
+
+	atomic_inc(&dev_data->bind);
+
+	ret = 0;
+
+out_unlock:
+
+	/* ready */
+	spin_unlock(&domain->lock);
+
+	return ret;
+}
+
+/*
+ * If a device is not yet associated with a domain, this function does
+ * assigns it visible for the hardware
+ */
+static int attach_device(struct device *dev,
+			 struct protection_domain *domain)
+{
+	struct pci_dev *pdev = to_pci_dev(dev);
+	struct iommu_dev_data *dev_data;
+	unsigned long flags;
+	int ret;
+
+	dev_data = get_dev_data(dev);
+
+	if (amd_iommu_iotlb_sup && pci_enable_ats(pdev, PAGE_SHIFT) == 0) {
+		dev_data->ats.enabled = true;
+		dev_data->ats.qdep    = pci_ats_queue_depth(pdev);
+	}
+
+	write_lock_irqsave(&amd_iommu_devtable_lock, flags);
+	ret = __attach_device(dev_data, domain);
+	write_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
+
+	/*
+	 * We might boot into a crash-kernel here. The crashed kernel
+	 * left the caches in the IOMMU dirty. So we have to flush
+	 * here to evict all dirty stuff.
+	 */
+	domain_flush_tlb_pde(domain);
+
+	return ret;
+}
+
+/*
+ * Removes a device from a protection domain (unlocked)
+ */
+static void __detach_device(struct iommu_dev_data *dev_data)
+{
+	struct protection_domain *domain;
+	unsigned long flags;
+
+	BUG_ON(!dev_data->domain);
+
+	domain = dev_data->domain;
+
+	spin_lock_irqsave(&domain->lock, flags);
+
+	if (dev_data->alias_data != NULL) {
+		struct iommu_dev_data *alias_data = dev_data->alias_data;
+
+		if (atomic_dec_and_test(&alias_data->bind))
+			do_detach(alias_data);
+	}
+
+	if (atomic_dec_and_test(&dev_data->bind))
+		do_detach(dev_data);
+
+	spin_unlock_irqrestore(&domain->lock, flags);
+
+	/*
+	 * If we run in passthrough mode the device must be assigned to the
+	 * passthrough domain if it is detached from any other domain.
+	 * Make sure we can deassign from the pt_domain itself.
+	 */
+	if (iommu_pass_through &&
+	    (dev_data->domain == NULL && domain != pt_domain))
+		__attach_device(dev_data, pt_domain);
+}
+
+/*
+ * Removes a device from a protection domain (with devtable_lock held)
+ */
+static void detach_device(struct device *dev)
+{
+	struct iommu_dev_data *dev_data;
+	unsigned long flags;
+
+	dev_data = get_dev_data(dev);
+
+	/* lock device table */
+	write_lock_irqsave(&amd_iommu_devtable_lock, flags);
+	__detach_device(dev_data);
+	write_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
+
+	if (dev_data->ats.enabled) {
+		pci_disable_ats(to_pci_dev(dev));
+		dev_data->ats.enabled = false;
+	}
+}
+
+/*
+ * Find out the protection domain structure for a given PCI device. This
+ * will give us the pointer to the page table root for example.
+ */
+static struct protection_domain *domain_for_device(struct device *dev)
+{
+	struct iommu_dev_data *dev_data;
+	struct protection_domain *dom = NULL;
+	unsigned long flags;
+
+	dev_data   = get_dev_data(dev);
+
+	if (dev_data->domain)
+		return dev_data->domain;
+
+	if (dev_data->alias_data != NULL) {
+		struct iommu_dev_data *alias_data = dev_data->alias_data;
+
+		read_lock_irqsave(&amd_iommu_devtable_lock, flags);
+		if (alias_data->domain != NULL) {
+			__attach_device(dev_data, alias_data->domain);
+			dom = alias_data->domain;
+		}
+		read_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
+	}
+
+	return dom;
+}
+
+static int device_change_notifier(struct notifier_block *nb,
+				  unsigned long action, void *data)
+{
+	struct device *dev = data;
+	u16 devid;
+	struct protection_domain *domain;
+	struct dma_ops_domain *dma_domain;
+	struct amd_iommu *iommu;
+	unsigned long flags;
+
+	if (!check_device(dev))
+		return 0;
+
+	devid  = get_device_id(dev);
+	iommu  = amd_iommu_rlookup_table[devid];
+
+	switch (action) {
+	case BUS_NOTIFY_UNBOUND_DRIVER:
+
+		domain = domain_for_device(dev);
+
+		if (!domain)
+			goto out;
+		if (iommu_pass_through)
+			break;
+		detach_device(dev);
+		break;
+	case BUS_NOTIFY_ADD_DEVICE:
+
+		iommu_init_device(dev);
+
+		domain = domain_for_device(dev);
+
+		/* allocate a protection domain if a device is added */
+		dma_domain = find_protection_domain(devid);
+		if (dma_domain)
+			goto out;
+		dma_domain = dma_ops_domain_alloc();
+		if (!dma_domain)
+			goto out;
+		dma_domain->target_dev = devid;
+
+		spin_lock_irqsave(&iommu_pd_list_lock, flags);
+		list_add_tail(&dma_domain->list, &iommu_pd_list);
+		spin_unlock_irqrestore(&iommu_pd_list_lock, flags);
+
+		break;
+	case BUS_NOTIFY_DEL_DEVICE:
+
+		iommu_uninit_device(dev);
+
+	default:
+		goto out;
+	}
+
+	iommu_completion_wait(iommu);
+
+out:
+	return 0;
+}
+
+static struct notifier_block device_nb = {
+	.notifier_call = device_change_notifier,
+};
+
+void amd_iommu_init_notifier(void)
+{
+	bus_register_notifier(&pci_bus_type, &device_nb);
+}
+
+/*****************************************************************************
+ *
+ * The next functions belong to the dma_ops mapping/unmapping code.
+ *
+ *****************************************************************************/
+
+/*
+ * In the dma_ops path we only have the struct device. This function
+ * finds the corresponding IOMMU, the protection domain and the
+ * requestor id for a given device.
+ * If the device is not yet associated with a domain this is also done
+ * in this function.
+ */
+static struct protection_domain *get_domain(struct device *dev)
+{
+	struct protection_domain *domain;
+	struct dma_ops_domain *dma_dom;
+	u16 devid = get_device_id(dev);
+
+	if (!check_device(dev))
+		return ERR_PTR(-EINVAL);
+
+	domain = domain_for_device(dev);
+	if (domain != NULL && !dma_ops_domain(domain))
+		return ERR_PTR(-EBUSY);
+
+	if (domain != NULL)
+		return domain;
+
+	/* Device not bount yet - bind it */
+	dma_dom = find_protection_domain(devid);
+	if (!dma_dom)
+		dma_dom = amd_iommu_rlookup_table[devid]->default_dom;
+	attach_device(dev, &dma_dom->domain);
+	DUMP_printk("Using protection domain %d for device %s\n",
+		    dma_dom->domain.id, dev_name(dev));
+
+	return &dma_dom->domain;
+}
+
+static void update_device_table(struct protection_domain *domain)
+{
+	struct iommu_dev_data *dev_data;
+
+	list_for_each_entry(dev_data, &domain->dev_list, list)
+		set_dte_entry(dev_data->devid, domain, dev_data->ats.enabled);
+}
+
+static void update_domain(struct protection_domain *domain)
+{
+	if (!domain->updated)
+		return;
+
+	update_device_table(domain);
+
+	domain_flush_devices(domain);
+	domain_flush_tlb_pde(domain);
+
+	domain->updated = false;
+}
+
+/*
+ * This function fetches the PTE for a given address in the aperture
+ */
+static u64* dma_ops_get_pte(struct dma_ops_domain *dom,
+			    unsigned long address)
+{
+	struct aperture_range *aperture;
+	u64 *pte, *pte_page;
+
+	aperture = dom->aperture[APERTURE_RANGE_INDEX(address)];
+	if (!aperture)
+		return NULL;
+
+	pte = aperture->pte_pages[APERTURE_PAGE_INDEX(address)];
+	if (!pte) {
+		pte = alloc_pte(&dom->domain, address, PAGE_SIZE, &pte_page,
+				GFP_ATOMIC);
+		aperture->pte_pages[APERTURE_PAGE_INDEX(address)] = pte_page;
+	} else
+		pte += PM_LEVEL_INDEX(0, address);
+
+	update_domain(&dom->domain);
+
+	return pte;
+}
+
+/*
+ * This is the generic map function. It maps one 4kb page at paddr to
+ * the given address in the DMA address space for the domain.
+ */
+static dma_addr_t dma_ops_domain_map(struct dma_ops_domain *dom,
+				     unsigned long address,
+				     phys_addr_t paddr,
+				     int direction)
+{
+	u64 *pte, __pte;
+
+	WARN_ON(address > dom->aperture_size);
+
+	paddr &= PAGE_MASK;
+
+	pte  = dma_ops_get_pte(dom, address);
+	if (!pte)
+		return DMA_ERROR_CODE;
+
+	__pte = paddr | IOMMU_PTE_P | IOMMU_PTE_FC;
+
+	if (direction == DMA_TO_DEVICE)
+		__pte |= IOMMU_PTE_IR;
+	else if (direction == DMA_FROM_DEVICE)
+		__pte |= IOMMU_PTE_IW;
+	else if (direction == DMA_BIDIRECTIONAL)
+		__pte |= IOMMU_PTE_IR | IOMMU_PTE_IW;
+
+	WARN_ON(*pte);
+
+	*pte = __pte;
+
+	return (dma_addr_t)address;
+}
+
+/*
+ * The generic unmapping function for on page in the DMA address space.
+ */
+static void dma_ops_domain_unmap(struct dma_ops_domain *dom,
+				 unsigned long address)
+{
+	struct aperture_range *aperture;
+	u64 *pte;
+
+	if (address >= dom->aperture_size)
+		return;
+
+	aperture = dom->aperture[APERTURE_RANGE_INDEX(address)];
+	if (!aperture)
+		return;
+
+	pte  = aperture->pte_pages[APERTURE_PAGE_INDEX(address)];
+	if (!pte)
+		return;
+
+	pte += PM_LEVEL_INDEX(0, address);
+
+	WARN_ON(!*pte);
+
+	*pte = 0ULL;
+}
+
+/*
+ * This function contains common code for mapping of a physically
+ * contiguous memory region into DMA address space. It is used by all
+ * mapping functions provided with this IOMMU driver.
+ * Must be called with the domain lock held.
+ */
+static dma_addr_t __map_single(struct device *dev,
+			       struct dma_ops_domain *dma_dom,
+			       phys_addr_t paddr,
+			       size_t size,
+			       int dir,
+			       bool align,
+			       u64 dma_mask)
+{
+	dma_addr_t offset = paddr & ~PAGE_MASK;
+	dma_addr_t address, start, ret;
+	unsigned int pages;
+	unsigned long align_mask = 0;
+	int i;
+
+	pages = iommu_num_pages(paddr, size, PAGE_SIZE);
+	paddr &= PAGE_MASK;
+
+	INC_STATS_COUNTER(total_map_requests);
+
+	if (pages > 1)
+		INC_STATS_COUNTER(cross_page);
+
+	if (align)
+		align_mask = (1UL << get_order(size)) - 1;
+
+retry:
+	address = dma_ops_alloc_addresses(dev, dma_dom, pages, align_mask,
+					  dma_mask);
+	if (unlikely(address == DMA_ERROR_CODE)) {
+		/*
+		 * setting next_address here will let the address
+		 * allocator only scan the new allocated range in the
+		 * first run. This is a small optimization.
+		 */
+		dma_dom->next_address = dma_dom->aperture_size;
+
+		if (alloc_new_range(dma_dom, false, GFP_ATOMIC))
+			goto out;
+
+		/*
+		 * aperture was successfully enlarged by 128 MB, try
+		 * allocation again
+		 */
+		goto retry;
+	}
+
+	start = address;
+	for (i = 0; i < pages; ++i) {
+		ret = dma_ops_domain_map(dma_dom, start, paddr, dir);
+		if (ret == DMA_ERROR_CODE)
+			goto out_unmap;
+
+		paddr += PAGE_SIZE;
+		start += PAGE_SIZE;
+	}
+	address += offset;
+
+	ADD_STATS_COUNTER(alloced_io_mem, size);
+
+	if (unlikely(dma_dom->need_flush && !amd_iommu_unmap_flush)) {
+		domain_flush_tlb(&dma_dom->domain);
+		dma_dom->need_flush = false;
+	} else if (unlikely(amd_iommu_np_cache))
+		domain_flush_pages(&dma_dom->domain, address, size);
+
+out:
+	return address;
+
+out_unmap:
+
+	for (--i; i >= 0; --i) {
+		start -= PAGE_SIZE;
+		dma_ops_domain_unmap(dma_dom, start);
+	}
+
+	dma_ops_free_addresses(dma_dom, address, pages);
+
+	return DMA_ERROR_CODE;
+}
+
+/*
+ * Does the reverse of the __map_single function. Must be called with
+ * the domain lock held too
+ */
+static void __unmap_single(struct dma_ops_domain *dma_dom,
+			   dma_addr_t dma_addr,
+			   size_t size,
+			   int dir)
+{
+	dma_addr_t flush_addr;
+	dma_addr_t i, start;
+	unsigned int pages;
+
+	if ((dma_addr == DMA_ERROR_CODE) ||
+	    (dma_addr + size > dma_dom->aperture_size))
+		return;
+
+	flush_addr = dma_addr;
+	pages = iommu_num_pages(dma_addr, size, PAGE_SIZE);
+	dma_addr &= PAGE_MASK;
+	start = dma_addr;
+
+	for (i = 0; i < pages; ++i) {
+		dma_ops_domain_unmap(dma_dom, start);
+		start += PAGE_SIZE;
+	}
+
+	SUB_STATS_COUNTER(alloced_io_mem, size);
+
+	dma_ops_free_addresses(dma_dom, dma_addr, pages);
+
+	if (amd_iommu_unmap_flush || dma_dom->need_flush) {
+		domain_flush_pages(&dma_dom->domain, flush_addr, size);
+		dma_dom->need_flush = false;
+	}
+}
+
+/*
+ * The exported map_single function for dma_ops.
+ */
+static dma_addr_t map_page(struct device *dev, struct page *page,
+			   unsigned long offset, size_t size,
+			   enum dma_data_direction dir,
+			   struct dma_attrs *attrs)
+{
+	unsigned long flags;
+	struct protection_domain *domain;
+	dma_addr_t addr;
+	u64 dma_mask;
+	phys_addr_t paddr = page_to_phys(page) + offset;
+
+	INC_STATS_COUNTER(cnt_map_single);
+
+	domain = get_domain(dev);
+	if (PTR_ERR(domain) == -EINVAL)
+		return (dma_addr_t)paddr;
+	else if (IS_ERR(domain))
+		return DMA_ERROR_CODE;
+
+	dma_mask = *dev->dma_mask;
+
+	spin_lock_irqsave(&domain->lock, flags);
+
+	addr = __map_single(dev, domain->priv, paddr, size, dir, false,
+			    dma_mask);
+	if (addr == DMA_ERROR_CODE)
+		goto out;
+
+	domain_flush_complete(domain);
+
+out:
+	spin_unlock_irqrestore(&domain->lock, flags);
+
+	return addr;
+}
+
+/*
+ * The exported unmap_single function for dma_ops.
+ */
+static void unmap_page(struct device *dev, dma_addr_t dma_addr, size_t size,
+		       enum dma_data_direction dir, struct dma_attrs *attrs)
+{
+	unsigned long flags;
+	struct protection_domain *domain;
+
+	INC_STATS_COUNTER(cnt_unmap_single);
+
+	domain = get_domain(dev);
+	if (IS_ERR(domain))
+		return;
+
+	spin_lock_irqsave(&domain->lock, flags);
+
+	__unmap_single(domain->priv, dma_addr, size, dir);
+
+	domain_flush_complete(domain);
+
+	spin_unlock_irqrestore(&domain->lock, flags);
+}
+
+/*
+ * This is a special map_sg function which is used if we should map a
+ * device which is not handled by an AMD IOMMU in the system.
+ */
+static int map_sg_no_iommu(struct device *dev, struct scatterlist *sglist,
+			   int nelems, int dir)
+{
+	struct scatterlist *s;
+	int i;
+
+	for_each_sg(sglist, s, nelems, i) {
+		s->dma_address = (dma_addr_t)sg_phys(s);
+		s->dma_length  = s->length;
+	}
+
+	return nelems;
+}
+
+/*
+ * The exported map_sg function for dma_ops (handles scatter-gather
+ * lists).
+ */
+static int map_sg(struct device *dev, struct scatterlist *sglist,
+		  int nelems, enum dma_data_direction dir,
+		  struct dma_attrs *attrs)
+{
+	unsigned long flags;
+	struct protection_domain *domain;
+	int i;
+	struct scatterlist *s;
+	phys_addr_t paddr;
+	int mapped_elems = 0;
+	u64 dma_mask;
+
+	INC_STATS_COUNTER(cnt_map_sg);
+
+	domain = get_domain(dev);
+	if (PTR_ERR(domain) == -EINVAL)
+		return map_sg_no_iommu(dev, sglist, nelems, dir);
+	else if (IS_ERR(domain))
+		return 0;
+
+	dma_mask = *dev->dma_mask;
+
+	spin_lock_irqsave(&domain->lock, flags);
+
+	for_each_sg(sglist, s, nelems, i) {
+		paddr = sg_phys(s);
+
+		s->dma_address = __map_single(dev, domain->priv,
+					      paddr, s->length, dir, false,
+					      dma_mask);
+
+		if (s->dma_address) {
+			s->dma_length = s->length;
+			mapped_elems++;
+		} else
+			goto unmap;
+	}
+
+	domain_flush_complete(domain);
+
+out:
+	spin_unlock_irqrestore(&domain->lock, flags);
+
+	return mapped_elems;
+unmap:
+	for_each_sg(sglist, s, mapped_elems, i) {
+		if (s->dma_address)
+			__unmap_single(domain->priv, s->dma_address,
+				       s->dma_length, dir);
+		s->dma_address = s->dma_length = 0;
+	}
+
+	mapped_elems = 0;
+
+	goto out;
+}
+
+/*
+ * The exported map_sg function for dma_ops (handles scatter-gather
+ * lists).
+ */
+static void unmap_sg(struct device *dev, struct scatterlist *sglist,
+		     int nelems, enum dma_data_direction dir,
+		     struct dma_attrs *attrs)
+{
+	unsigned long flags;
+	struct protection_domain *domain;
+	struct scatterlist *s;
+	int i;
+
+	INC_STATS_COUNTER(cnt_unmap_sg);
+
+	domain = get_domain(dev);
+	if (IS_ERR(domain))
+		return;
+
+	spin_lock_irqsave(&domain->lock, flags);
+
+	for_each_sg(sglist, s, nelems, i) {
+		__unmap_single(domain->priv, s->dma_address,
+			       s->dma_length, dir);
+		s->dma_address = s->dma_length = 0;
+	}
+
+	domain_flush_complete(domain);
+
+	spin_unlock_irqrestore(&domain->lock, flags);
+}
+
+/*
+ * The exported alloc_coherent function for dma_ops.
+ */
+static void *alloc_coherent(struct device *dev, size_t size,
+			    dma_addr_t *dma_addr, gfp_t flag)
+{
+	unsigned long flags;
+	void *virt_addr;
+	struct protection_domain *domain;
+	phys_addr_t paddr;
+	u64 dma_mask = dev->coherent_dma_mask;
+
+	INC_STATS_COUNTER(cnt_alloc_coherent);
+
+	domain = get_domain(dev);
+	if (PTR_ERR(domain) == -EINVAL) {
+		virt_addr = (void *)__get_free_pages(flag, get_order(size));
+		*dma_addr = __pa(virt_addr);
+		return virt_addr;
+	} else if (IS_ERR(domain))
+		return NULL;
+
+	dma_mask  = dev->coherent_dma_mask;
+	flag     &= ~(__GFP_DMA | __GFP_HIGHMEM | __GFP_DMA32);
+	flag     |= __GFP_ZERO;
+
+	virt_addr = (void *)__get_free_pages(flag, get_order(size));
+	if (!virt_addr)
+		return NULL;
+
+	paddr = virt_to_phys(virt_addr);
+
+	if (!dma_mask)
+		dma_mask = *dev->dma_mask;
+
+	spin_lock_irqsave(&domain->lock, flags);
+
+	*dma_addr = __map_single(dev, domain->priv, paddr,
+				 size, DMA_BIDIRECTIONAL, true, dma_mask);
+
+	if (*dma_addr == DMA_ERROR_CODE) {
+		spin_unlock_irqrestore(&domain->lock, flags);
+		goto out_free;
+	}
+
+	domain_flush_complete(domain);
+
+	spin_unlock_irqrestore(&domain->lock, flags);
+
+	return virt_addr;
+
+out_free:
+
+	free_pages((unsigned long)virt_addr, get_order(size));
+
+	return NULL;
+}
+
+/*
+ * The exported free_coherent function for dma_ops.
+ */
+static void free_coherent(struct device *dev, size_t size,
+			  void *virt_addr, dma_addr_t dma_addr)
+{
+	unsigned long flags;
+	struct protection_domain *domain;
+
+	INC_STATS_COUNTER(cnt_free_coherent);
+
+	domain = get_domain(dev);
+	if (IS_ERR(domain))
+		goto free_mem;
+
+	spin_lock_irqsave(&domain->lock, flags);
+
+	__unmap_single(domain->priv, dma_addr, size, DMA_BIDIRECTIONAL);
+
+	domain_flush_complete(domain);
+
+	spin_unlock_irqrestore(&domain->lock, flags);
+
+free_mem:
+	free_pages((unsigned long)virt_addr, get_order(size));
+}
+
+/*
+ * This function is called by the DMA layer to find out if we can handle a
+ * particular device. It is part of the dma_ops.
+ */
+static int amd_iommu_dma_supported(struct device *dev, u64 mask)
+{
+	return check_device(dev);
+}
+
+/*
+ * The function for pre-allocating protection domains.
+ *
+ * If the driver core informs the DMA layer if a driver grabs a device
+ * we don't need to preallocate the protection domains anymore.
+ * For now we have to.
+ */
+static void prealloc_protection_domains(void)
+{
+	struct pci_dev *dev = NULL;
+	struct dma_ops_domain *dma_dom;
+	u16 devid;
+
+	for_each_pci_dev(dev) {
+
+		/* Do we handle this device? */
+		if (!check_device(&dev->dev))
+			continue;
+
+		/* Is there already any domain for it? */
+		if (domain_for_device(&dev->dev))
+			continue;
+
+		devid = get_device_id(&dev->dev);
+
+		dma_dom = dma_ops_domain_alloc();
+		if (!dma_dom)
+			continue;
+		init_unity_mappings_for_device(dma_dom, devid);
+		dma_dom->target_dev = devid;
+
+		attach_device(&dev->dev, &dma_dom->domain);
+
+		list_add_tail(&dma_dom->list, &iommu_pd_list);
+	}
+}
+
+static struct dma_map_ops amd_iommu_dma_ops = {
+	.alloc_coherent = alloc_coherent,
+	.free_coherent = free_coherent,
+	.map_page = map_page,
+	.unmap_page = unmap_page,
+	.map_sg = map_sg,
+	.unmap_sg = unmap_sg,
+	.dma_supported = amd_iommu_dma_supported,
+};
+
+static unsigned device_dma_ops_init(void)
+{
+	struct pci_dev *pdev = NULL;
+	unsigned unhandled = 0;
+
+	for_each_pci_dev(pdev) {
+		if (!check_device(&pdev->dev)) {
+			unhandled += 1;
+			continue;
+		}
+
+		pdev->dev.archdata.dma_ops = &amd_iommu_dma_ops;
+	}
+
+	return unhandled;
+}
+
+/*
+ * The function which clues the AMD IOMMU driver into dma_ops.
+ */
+
+void __init amd_iommu_init_api(void)
+{
+	register_iommu(&amd_iommu_ops);
+}
+
+int __init amd_iommu_init_dma_ops(void)
+{
+	struct amd_iommu *iommu;
+	int ret, unhandled;
+
+	/*
+	 * first allocate a default protection domain for every IOMMU we
+	 * found in the system. Devices not assigned to any other
+	 * protection domain will be assigned to the default one.
+	 */
+	for_each_iommu(iommu) {
+		iommu->default_dom = dma_ops_domain_alloc();
+		if (iommu->default_dom == NULL)
+			return -ENOMEM;
+		iommu->default_dom->domain.flags |= PD_DEFAULT_MASK;
+		ret = iommu_init_unity_mappings(iommu);
+		if (ret)
+			goto free_domains;
+	}
+
+	/*
+	 * Pre-allocate the protection domains for each device.
+	 */
+	prealloc_protection_domains();
+
+	iommu_detected = 1;
+	swiotlb = 0;
+
+	/* Make the driver finally visible to the drivers */
+	unhandled = device_dma_ops_init();
+	if (unhandled && max_pfn > MAX_DMA32_PFN) {
+		/* There are unhandled devices - initialize swiotlb for them */
+		swiotlb = 1;
+	}
+
+	amd_iommu_stats_init();
+
+	return 0;
+
+free_domains:
+
+	for_each_iommu(iommu) {
+		if (iommu->default_dom)
+			dma_ops_domain_free(iommu->default_dom);
+	}
+
+	return ret;
+}
+
+/*****************************************************************************
+ *
+ * The following functions belong to the exported interface of AMD IOMMU
+ *
+ * This interface allows access to lower level functions of the IOMMU
+ * like protection domain handling and assignement of devices to domains
+ * which is not possible with the dma_ops interface.
+ *
+ *****************************************************************************/
+
+static void cleanup_domain(struct protection_domain *domain)
+{
+	struct iommu_dev_data *dev_data, *next;
+	unsigned long flags;
+
+	write_lock_irqsave(&amd_iommu_devtable_lock, flags);
+
+	list_for_each_entry_safe(dev_data, next, &domain->dev_list, list) {
+		__detach_device(dev_data);
+		atomic_set(&dev_data->bind, 0);
+	}
+
+	write_unlock_irqrestore(&amd_iommu_devtable_lock, flags);
+}
+
+static void protection_domain_free(struct protection_domain *domain)
+{
+	if (!domain)
+		return;
+
+	del_domain_from_list(domain);
+
+	if (domain->id)
+		domain_id_free(domain->id);
+
+	kfree(domain);
+}
+
+static struct protection_domain *protection_domain_alloc(void)
+{
+	struct protection_domain *domain;
+
+	domain = kzalloc(sizeof(*domain), GFP_KERNEL);
+	if (!domain)
+		return NULL;
+
+	spin_lock_init(&domain->lock);
+	mutex_init(&domain->api_lock);
+	domain->id = domain_id_alloc();
+	if (!domain->id)
+		goto out_err;
+	INIT_LIST_HEAD(&domain->dev_list);
+
+	add_domain_to_list(domain);
+
+	return domain;
+
+out_err:
+	kfree(domain);
+
+	return NULL;
+}
+
+static int amd_iommu_domain_init(struct iommu_domain *dom)
+{
+	struct protection_domain *domain;
+
+	domain = protection_domain_alloc();
+	if (!domain)
+		goto out_free;
+
+	domain->mode    = PAGE_MODE_3_LEVEL;
+	domain->pt_root = (void *)get_zeroed_page(GFP_KERNEL);
+	if (!domain->pt_root)
+		goto out_free;
+
+	dom->priv = domain;
+
+	return 0;
+
+out_free:
+	protection_domain_free(domain);
+
+	return -ENOMEM;
+}
+
+static void amd_iommu_domain_destroy(struct iommu_domain *dom)
+{
+	struct protection_domain *domain = dom->priv;
+
+	if (!domain)
+		return;
+
+	if (domain->dev_cnt > 0)
+		cleanup_domain(domain);
+
+	BUG_ON(domain->dev_cnt != 0);
+
+	free_pagetable(domain);
+
+	protection_domain_free(domain);
+
+	dom->priv = NULL;
+}
+
+static void amd_iommu_detach_device(struct iommu_domain *dom,
+				    struct device *dev)
+{
+	struct iommu_dev_data *dev_data = dev->archdata.iommu;
+	struct amd_iommu *iommu;
+	u16 devid;
+
+	if (!check_device(dev))
+		return;
+
+	devid = get_device_id(dev);
+
+	if (dev_data->domain != NULL)
+		detach_device(dev);
+
+	iommu = amd_iommu_rlookup_table[devid];
+	if (!iommu)
+		return;
+
+	iommu_completion_wait(iommu);
+}
+
+static int amd_iommu_attach_device(struct iommu_domain *dom,
+				   struct device *dev)
+{
+	struct protection_domain *domain = dom->priv;
+	struct iommu_dev_data *dev_data;
+	struct amd_iommu *iommu;
+	int ret;
+
+	if (!check_device(dev))
+		return -EINVAL;
+
+	dev_data = dev->archdata.iommu;
+
+	iommu = amd_iommu_rlookup_table[dev_data->devid];
+	if (!iommu)
+		return -EINVAL;
+
+	if (dev_data->domain)
+		detach_device(dev);
+
+	ret = attach_device(dev, domain);
+
+	iommu_completion_wait(iommu);
+
+	return ret;
+}
+
+static int amd_iommu_map(struct iommu_domain *dom, unsigned long iova,
+			 phys_addr_t paddr, int gfp_order, int iommu_prot)
+{
+	unsigned long page_size = 0x1000UL << gfp_order;
+	struct protection_domain *domain = dom->priv;
+	int prot = 0;
+	int ret;
+
+	if (iommu_prot & IOMMU_READ)
+		prot |= IOMMU_PROT_IR;
+	if (iommu_prot & IOMMU_WRITE)
+		prot |= IOMMU_PROT_IW;
+
+	mutex_lock(&domain->api_lock);
+	ret = iommu_map_page(domain, iova, paddr, prot, page_size);
+	mutex_unlock(&domain->api_lock);
+
+	return ret;
+}
+
+static int amd_iommu_unmap(struct iommu_domain *dom, unsigned long iova,
+			   int gfp_order)
+{
+	struct protection_domain *domain = dom->priv;
+	unsigned long page_size, unmap_size;
+
+	page_size  = 0x1000UL << gfp_order;
+
+	mutex_lock(&domain->api_lock);
+	unmap_size = iommu_unmap_page(domain, iova, page_size);
+	mutex_unlock(&domain->api_lock);
+
+	domain_flush_tlb_pde(domain);
+
+	return get_order(unmap_size);
+}
+
+static phys_addr_t amd_iommu_iova_to_phys(struct iommu_domain *dom,
+					  unsigned long iova)
+{
+	struct protection_domain *domain = dom->priv;
+	unsigned long offset_mask;
+	phys_addr_t paddr;
+	u64 *pte, __pte;
+
+	pte = fetch_pte(domain, iova);
+
+	if (!pte || !IOMMU_PTE_PRESENT(*pte))
+		return 0;
+
+	if (PM_PTE_LEVEL(*pte) == 0)
+		offset_mask = PAGE_SIZE - 1;
+	else
+		offset_mask = PTE_PAGE_SIZE(*pte) - 1;
+
+	__pte = *pte & PM_ADDR_MASK;
+	paddr = (__pte & ~offset_mask) | (iova & offset_mask);
+
+	return paddr;
+}
+
+static int amd_iommu_domain_has_cap(struct iommu_domain *domain,
+				    unsigned long cap)
+{
+	switch (cap) {
+	case IOMMU_CAP_CACHE_COHERENCY:
+		return 1;
+	}
+
+	return 0;
+}
+
+static struct iommu_ops amd_iommu_ops = {
+	.domain_init = amd_iommu_domain_init,
+	.domain_destroy = amd_iommu_domain_destroy,
+	.attach_dev = amd_iommu_attach_device,
+	.detach_dev = amd_iommu_detach_device,
+	.map = amd_iommu_map,
+	.unmap = amd_iommu_unmap,
+	.iova_to_phys = amd_iommu_iova_to_phys,
+	.domain_has_cap = amd_iommu_domain_has_cap,
+};
+
+/*****************************************************************************
+ *
+ * The next functions do a basic initialization of IOMMU for pass through
+ * mode
+ *
+ * In passthrough mode the IOMMU is initialized and enabled but not used for
+ * DMA-API translation.
+ *
+ *****************************************************************************/
+
+int __init amd_iommu_init_passthrough(void)
+{
+	struct amd_iommu *iommu;
+	struct pci_dev *dev = NULL;
+	u16 devid;
+
+	/* allocate passthrough domain */
+	pt_domain = protection_domain_alloc();
+	if (!pt_domain)
+		return -ENOMEM;
+
+	pt_domain->mode |= PAGE_MODE_NONE;
+
+	for_each_pci_dev(dev) {
+		if (!check_device(&dev->dev))
+			continue;
+
+		devid = get_device_id(&dev->dev);
+
+		iommu = amd_iommu_rlookup_table[devid];
+		if (!iommu)
+			continue;
+
+		attach_device(&dev->dev, pt_domain);
+	}
+
+	pr_info("AMD-Vi: Initialized for Passthrough Mode\n");
+
+	return 0;
+}
diff --git a/drivers/iommu/amd_iommu_init.c b/drivers/iommu/amd_iommu_init.c
new file mode 100644
index 00000000000..82d2410f420
--- /dev/null
+++ b/drivers/iommu/amd_iommu_init.c
@@ -0,0 +1,1574 @@
+/*
+ * Copyright (C) 2007-2010 Advanced Micro Devices, Inc.
+ * Author: Joerg Roedel <joerg.roedel@amd.com>
+ *         Leo Duran <leo.duran@amd.com>
+ *
+ * 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#include <linux/pci.h>
+#include <linux/acpi.h>
+#include <linux/list.h>
+#include <linux/slab.h>
+#include <linux/syscore_ops.h>
+#include <linux/interrupt.h>
+#include <linux/msi.h>
+#include <linux/amd-iommu.h>
+#include <asm/pci-direct.h>
+#include <asm/iommu.h>
+#include <asm/gart.h>
+#include <asm/x86_init.h>
+#include <asm/iommu_table.h>
+
+#include "amd_iommu_proto.h"
+#include "amd_iommu_types.h"
+
+/*
+ * definitions for the ACPI scanning code
+ */
+#define IVRS_HEADER_LENGTH 48
+
+#define ACPI_IVHD_TYPE                  0x10
+#define ACPI_IVMD_TYPE_ALL              0x20
+#define ACPI_IVMD_TYPE                  0x21
+#define ACPI_IVMD_TYPE_RANGE            0x22
+
+#define IVHD_DEV_ALL                    0x01
+#define IVHD_DEV_SELECT                 0x02
+#define IVHD_DEV_SELECT_RANGE_START     0x03
+#define IVHD_DEV_RANGE_END              0x04
+#define IVHD_DEV_ALIAS                  0x42
+#define IVHD_DEV_ALIAS_RANGE            0x43
+#define IVHD_DEV_EXT_SELECT             0x46
+#define IVHD_DEV_EXT_SELECT_RANGE       0x47
+
+#define IVHD_FLAG_HT_TUN_EN_MASK        0x01
+#define IVHD_FLAG_PASSPW_EN_MASK        0x02
+#define IVHD_FLAG_RESPASSPW_EN_MASK     0x04
+#define IVHD_FLAG_ISOC_EN_MASK          0x08
+
+#define IVMD_FLAG_EXCL_RANGE            0x08
+#define IVMD_FLAG_UNITY_MAP             0x01
+
+#define ACPI_DEVFLAG_INITPASS           0x01
+#define ACPI_DEVFLAG_EXTINT             0x02
+#define ACPI_DEVFLAG_NMI                0x04
+#define ACPI_DEVFLAG_SYSMGT1            0x10
+#define ACPI_DEVFLAG_SYSMGT2            0x20
+#define ACPI_DEVFLAG_LINT0              0x40
+#define ACPI_DEVFLAG_LINT1              0x80
+#define ACPI_DEVFLAG_ATSDIS             0x10000000
+
+/*
+ * ACPI table definitions
+ *
+ * These data structures are laid over the table to parse the important values
+ * out of it.
+ */
+
+/*
+ * structure describing one IOMMU in the ACPI table. Typically followed by one
+ * or more ivhd_entrys.
+ */
+struct ivhd_header {
+	u8 type;
+	u8 flags;
+	u16 length;
+	u16 devid;
+	u16 cap_ptr;
+	u64 mmio_phys;
+	u16 pci_seg;
+	u16 info;
+	u32 reserved;
+} __attribute__((packed));
+
+/*
+ * A device entry describing which devices a specific IOMMU translates and
+ * which requestor ids they use.
+ */
+struct ivhd_entry {
+	u8 type;
+	u16 devid;
+	u8 flags;
+	u32 ext;
+} __attribute__((packed));
+
+/*
+ * An AMD IOMMU memory definition structure. It defines things like exclusion
+ * ranges for devices and regions that should be unity mapped.
+ */
+struct ivmd_header {
+	u8 type;
+	u8 flags;
+	u16 length;
+	u16 devid;
+	u16 aux;
+	u64 resv;
+	u64 range_start;
+	u64 range_length;
+} __attribute__((packed));
+
+bool amd_iommu_dump;
+
+static int __initdata amd_iommu_detected;
+static bool __initdata amd_iommu_disabled;
+
+u16 amd_iommu_last_bdf;			/* largest PCI device id we have
+					   to handle */
+LIST_HEAD(amd_iommu_unity_map);		/* a list of required unity mappings
+					   we find in ACPI */
+bool amd_iommu_unmap_flush;		/* if true, flush on every unmap */
+
+LIST_HEAD(amd_iommu_list);		/* list of all AMD IOMMUs in the
+					   system */
+
+/* Array to assign indices to IOMMUs*/
+struct amd_iommu *amd_iommus[MAX_IOMMUS];
+int amd_iommus_present;
+
+/* IOMMUs have a non-present cache? */
+bool amd_iommu_np_cache __read_mostly;
+bool amd_iommu_iotlb_sup __read_mostly = true;
+
+/*
+ * The ACPI table parsing functions set this variable on an error
+ */
+static int __initdata amd_iommu_init_err;
+
+/*
+ * List of protection domains - used during resume
+ */
+LIST_HEAD(amd_iommu_pd_list);
+spinlock_t amd_iommu_pd_lock;
+
+/*
+ * Pointer to the device table which is shared by all AMD IOMMUs
+ * it is indexed by the PCI device id or the HT unit id and contains
+ * information about the domain the device belongs to as well as the
+ * page table root pointer.
+ */
+struct dev_table_entry *amd_iommu_dev_table;
+
+/*
+ * The alias table is a driver specific data structure which contains the
+ * mappings of the PCI device ids to the actual requestor ids on the IOMMU.
+ * More than one device can share the same requestor id.
+ */
+u16 *amd_iommu_alias_table;
+
+/*
+ * The rlookup table is used to find the IOMMU which is responsible
+ * for a specific device. It is also indexed by the PCI device id.
+ */
+struct amd_iommu **amd_iommu_rlookup_table;
+
+/*
+ * AMD IOMMU allows up to 2^16 differend protection domains. This is a bitmap
+ * to know which ones are already in use.
+ */
+unsigned long *amd_iommu_pd_alloc_bitmap;
+
+static u32 dev_table_size;	/* size of the device table */
+static u32 alias_table_size;	/* size of the alias table */
+static u32 rlookup_table_size;	/* size if the rlookup table */
+
+/*
+ * This function flushes all internal caches of
+ * the IOMMU used by this driver.
+ */
+extern void iommu_flush_all_caches(struct amd_iommu *iommu);
+
+static inline void update_last_devid(u16 devid)
+{
+	if (devid > amd_iommu_last_bdf)
+		amd_iommu_last_bdf = devid;
+}
+
+static inline unsigned long tbl_size(int entry_size)
+{
+	unsigned shift = PAGE_SHIFT +
+			 get_order(((int)amd_iommu_last_bdf + 1) * entry_size);
+
+	return 1UL << shift;
+}
+
+/* Access to l1 and l2 indexed register spaces */
+
+static u32 iommu_read_l1(struct amd_iommu *iommu, u16 l1, u8 address)
+{
+	u32 val;
+
+	pci_write_config_dword(iommu->dev, 0xf8, (address | l1 << 16));
+	pci_read_config_dword(iommu->dev, 0xfc, &val);
+	return val;
+}
+
+static void iommu_write_l1(struct amd_iommu *iommu, u16 l1, u8 address, u32 val)
+{
+	pci_write_config_dword(iommu->dev, 0xf8, (address | l1 << 16 | 1 << 31));
+	pci_write_config_dword(iommu->dev, 0xfc, val);
+	pci_write_config_dword(iommu->dev, 0xf8, (address | l1 << 16));
+}
+
+static u32 iommu_read_l2(struct amd_iommu *iommu, u8 address)
+{
+	u32 val;
+
+	pci_write_config_dword(iommu->dev, 0xf0, address);
+	pci_read_config_dword(iommu->dev, 0xf4, &val);
+	return val;
+}
+
+static void iommu_write_l2(struct amd_iommu *iommu, u8 address, u32 val)
+{
+	pci_write_config_dword(iommu->dev, 0xf0, (address | 1 << 8));
+	pci_write_config_dword(iommu->dev, 0xf4, val);
+}
+
+/****************************************************************************
+ *
+ * AMD IOMMU MMIO register space handling functions
+ *
+ * These functions are used to program the IOMMU device registers in
+ * MMIO space required for that driver.
+ *
+ ****************************************************************************/
+
+/*
+ * This function set the exclusion range in the IOMMU. DMA accesses to the
+ * exclusion range are passed through untranslated
+ */
+static void iommu_set_exclusion_range(struct amd_iommu *iommu)
+{
+	u64 start = iommu->exclusion_start & PAGE_MASK;
+	u64 limit = (start + iommu->exclusion_length) & PAGE_MASK;
+	u64 entry;
+
+	if (!iommu->exclusion_start)
+		return;
+
+	entry = start | MMIO_EXCL_ENABLE_MASK;
+	memcpy_toio(iommu->mmio_base + MMIO_EXCL_BASE_OFFSET,
+			&entry, sizeof(entry));
+
+	entry = limit;
+	memcpy_toio(iommu->mmio_base + MMIO_EXCL_LIMIT_OFFSET,
+			&entry, sizeof(entry));
+}
+
+/* Programs the physical address of the device table into the IOMMU hardware */
+static void __init iommu_set_device_table(struct amd_iommu *iommu)
+{
+	u64 entry;
+
+	BUG_ON(iommu->mmio_base == NULL);
+
+	entry = virt_to_phys(amd_iommu_dev_table);
+	entry |= (dev_table_size >> 12) - 1;
+	memcpy_toio(iommu->mmio_base + MMIO_DEV_TABLE_OFFSET,
+			&entry, sizeof(entry));
+}
+
+/* Generic functions to enable/disable certain features of the IOMMU. */
+static void iommu_feature_enable(struct amd_iommu *iommu, u8 bit)
+{
+	u32 ctrl;
+
+	ctrl = readl(iommu->mmio_base + MMIO_CONTROL_OFFSET);
+	ctrl |= (1 << bit);
+	writel(ctrl, iommu->mmio_base + MMIO_CONTROL_OFFSET);
+}
+
+static void iommu_feature_disable(struct amd_iommu *iommu, u8 bit)
+{
+	u32 ctrl;
+
+	ctrl = readl(iommu->mmio_base + MMIO_CONTROL_OFFSET);
+	ctrl &= ~(1 << bit);
+	writel(ctrl, iommu->mmio_base + MMIO_CONTROL_OFFSET);
+}
+
+/* Function to enable the hardware */
+static void iommu_enable(struct amd_iommu *iommu)
+{
+	static const char * const feat_str[] = {
+		"PreF", "PPR", "X2APIC", "NX", "GT", "[5]",
+		"IA", "GA", "HE", "PC", NULL
+	};
+	int i;
+
+	printk(KERN_INFO "AMD-Vi: Enabling IOMMU at %s cap 0x%hx",
+	       dev_name(&iommu->dev->dev), iommu->cap_ptr);
+
+	if (iommu->cap & (1 << IOMMU_CAP_EFR)) {
+		printk(KERN_CONT " extended features: ");
+		for (i = 0; feat_str[i]; ++i)
+			if (iommu_feature(iommu, (1ULL << i)))
+				printk(KERN_CONT " %s", feat_str[i]);
+	}
+	printk(KERN_CONT "\n");
+
+	iommu_feature_enable(iommu, CONTROL_IOMMU_EN);
+}
+
+static void iommu_disable(struct amd_iommu *iommu)
+{
+	/* Disable command buffer */
+	iommu_feature_disable(iommu, CONTROL_CMDBUF_EN);
+
+	/* Disable event logging and event interrupts */
+	iommu_feature_disable(iommu, CONTROL_EVT_INT_EN);
+	iommu_feature_disable(iommu, CONTROL_EVT_LOG_EN);
+
+	/* Disable IOMMU hardware itself */
+	iommu_feature_disable(iommu, CONTROL_IOMMU_EN);
+}
+
+/*
+ * mapping and unmapping functions for the IOMMU MMIO space. Each AMD IOMMU in
+ * the system has one.
+ */
+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);
+		pr_err("AMD-Vi: This is a BIOS bug. Please contact your hardware vendor\n");
+		return NULL;
+	}
+
+	ret = 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)
+{
+	if (iommu->mmio_base)
+		iounmap(iommu->mmio_base);
+	release_mem_region(iommu->mmio_phys, MMIO_REGION_LENGTH);
+}
+
+/****************************************************************************
+ *
+ * The functions below belong to the first pass of AMD IOMMU ACPI table
+ * parsing. In this pass we try to find out the highest device id this
+ * code has to handle. Upon this information the size of the shared data
+ * structures is determined later.
+ *
+ ****************************************************************************/
+
+/*
+ * This function calculates the length of a given IVHD entry
+ */
+static inline int ivhd_entry_length(u8 *ivhd)
+{
+	return 0x04 << (*ivhd >> 6);
+}
+
+/*
+ * This function reads the last device id the IOMMU has to handle from the PCI
+ * capability header for this IOMMU
+ */
+static int __init find_last_devid_on_pci(int bus, int dev, int fn, int cap_ptr)
+{
+	u32 cap;
+
+	cap = read_pci_config(bus, dev, fn, cap_ptr+MMIO_RANGE_OFFSET);
+	update_last_devid(calc_devid(MMIO_GET_BUS(cap), MMIO_GET_LD(cap)));
+
+	return 0;
+}
+
+/*
+ * After reading the highest device id from the IOMMU PCI capability header
+ * this function looks if there is a higher device id defined in the ACPI table
+ */
+static int __init find_last_devid_from_ivhd(struct ivhd_header *h)
+{
+	u8 *p = (void *)h, *end = (void *)h;
+	struct ivhd_entry *dev;
+
+	p += sizeof(*h);
+	end += h->length;
+
+	find_last_devid_on_pci(PCI_BUS(h->devid),
+			PCI_SLOT(h->devid),
+			PCI_FUNC(h->devid),
+			h->cap_ptr);
+
+	while (p < end) {
+		dev = (struct ivhd_entry *)p;
+		switch (dev->type) {
+		case IVHD_DEV_SELECT:
+		case IVHD_DEV_RANGE_END:
+		case IVHD_DEV_ALIAS:
+		case IVHD_DEV_EXT_SELECT:
+			/* all the above subfield types refer to device ids */
+			update_last_devid(dev->devid);
+			break;
+		default:
+			break;
+		}
+		p += ivhd_entry_length(p);
+	}
+
+	WARN_ON(p != end);
+
+	return 0;
+}
+
+/*
+ * Iterate over all IVHD entries in the ACPI table and find the highest device
+ * id which we need to handle. This is the first of three functions which parse
+ * the ACPI table. So we check the checksum here.
+ */
+static int __init find_last_devid_acpi(struct acpi_table_header *table)
+{
+	int i;
+	u8 checksum = 0, *p = (u8 *)table, *end = (u8 *)table;
+	struct ivhd_header *h;
+
+	/*
+	 * Validate checksum here so we don't need to do it when
+	 * we actually parse the table
+	 */
+	for (i = 0; i < table->length; ++i)
+		checksum += p[i];
+	if (checksum != 0) {
+		/* ACPI table corrupt */
+		amd_iommu_init_err = -ENODEV;
+		return 0;
+	}
+
+	p += IVRS_HEADER_LENGTH;
+
+	end += table->length;
+	while (p < end) {
+		h = (struct ivhd_header *)p;
+		switch (h->type) {
+		case ACPI_IVHD_TYPE:
+			find_last_devid_from_ivhd(h);
+			break;
+		default:
+			break;
+		}
+		p += h->length;
+	}
+	WARN_ON(p != end);
+
+	return 0;
+}
+
+/****************************************************************************
+ *
+ * The following functions belong the the code path which parses the ACPI table
+ * the second time. In this ACPI parsing iteration we allocate IOMMU specific
+ * data structures, initialize the device/alias/rlookup table and also
+ * basically initialize the hardware.
+ *
+ ****************************************************************************/
+
+/*
+ * Allocates the command buffer. This buffer is per AMD IOMMU. We can
+ * write commands to that buffer later and the IOMMU will execute them
+ * asynchronously
+ */
+static u8 * __init alloc_command_buffer(struct amd_iommu *iommu)
+{
+	u8 *cmd_buf = (u8 *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
+			get_order(CMD_BUFFER_SIZE));
+
+	if (cmd_buf == NULL)
+		return NULL;
+
+	iommu->cmd_buf_size = CMD_BUFFER_SIZE | CMD_BUFFER_UNINITIALIZED;
+
+	return cmd_buf;
+}
+
+/*
+ * This function resets the command buffer if the IOMMU stopped fetching
+ * commands from it.
+ */
+void amd_iommu_reset_cmd_buffer(struct amd_iommu *iommu)
+{
+	iommu_feature_disable(iommu, CONTROL_CMDBUF_EN);
+
+	writel(0x00, iommu->mmio_base + MMIO_CMD_HEAD_OFFSET);
+	writel(0x00, iommu->mmio_base + MMIO_CMD_TAIL_OFFSET);
+
+	iommu_feature_enable(iommu, CONTROL_CMDBUF_EN);
+}
+
+/*
+ * This function writes the command buffer address to the hardware and
+ * enables it.
+ */
+static void iommu_enable_command_buffer(struct amd_iommu *iommu)
+{
+	u64 entry;
+
+	BUG_ON(iommu->cmd_buf == NULL);
+
+	entry = (u64)virt_to_phys(iommu->cmd_buf);
+	entry |= MMIO_CMD_SIZE_512;
+
+	memcpy_toio(iommu->mmio_base + MMIO_CMD_BUF_OFFSET,
+		    &entry, sizeof(entry));
+
+	amd_iommu_reset_cmd_buffer(iommu);
+	iommu->cmd_buf_size &= ~(CMD_BUFFER_UNINITIALIZED);
+}
+
+static void __init free_command_buffer(struct amd_iommu *iommu)
+{
+	free_pages((unsigned long)iommu->cmd_buf,
+		   get_order(iommu->cmd_buf_size & ~(CMD_BUFFER_UNINITIALIZED)));
+}
+
+/* allocates the memory where the IOMMU will log its events to */
+static u8 * __init alloc_event_buffer(struct amd_iommu *iommu)
+{
+	iommu->evt_buf = (u8 *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
+						get_order(EVT_BUFFER_SIZE));
+
+	if (iommu->evt_buf == NULL)
+		return NULL;
+
+	iommu->evt_buf_size = EVT_BUFFER_SIZE;
+
+	return iommu->evt_buf;
+}
+
+static void iommu_enable_event_buffer(struct amd_iommu *iommu)
+{
+	u64 entry;
+
+	BUG_ON(iommu->evt_buf == NULL);
+
+	entry = (u64)virt_to_phys(iommu->evt_buf) | EVT_LEN_MASK;
+
+	memcpy_toio(iommu->mmio_base + MMIO_EVT_BUF_OFFSET,
+		    &entry, sizeof(entry));
+
+	/* set head and tail to zero manually */
+	writel(0x00, iommu->mmio_base + MMIO_EVT_HEAD_OFFSET);
+	writel(0x00, iommu->mmio_base + MMIO_EVT_TAIL_OFFSET);
+
+	iommu_feature_enable(iommu, CONTROL_EVT_LOG_EN);
+}
+
+static void __init free_event_buffer(struct amd_iommu *iommu)
+{
+	free_pages((unsigned long)iommu->evt_buf, get_order(EVT_BUFFER_SIZE));
+}
+
+/* sets a specific bit in the device table entry. */
+static void set_dev_entry_bit(u16 devid, u8 bit)
+{
+	int i = (bit >> 5) & 0x07;
+	int _bit = bit & 0x1f;
+
+	amd_iommu_dev_table[devid].data[i] |= (1 << _bit);
+}
+
+static int get_dev_entry_bit(u16 devid, u8 bit)
+{
+	int i = (bit >> 5) & 0x07;
+	int _bit = bit & 0x1f;
+
+	return (amd_iommu_dev_table[devid].data[i] & (1 << _bit)) >> _bit;
+}
+
+
+void amd_iommu_apply_erratum_63(u16 devid)
+{
+	int sysmgt;
+
+	sysmgt = get_dev_entry_bit(devid, DEV_ENTRY_SYSMGT1) |
+		 (get_dev_entry_bit(devid, DEV_ENTRY_SYSMGT2) << 1);
+
+	if (sysmgt == 0x01)
+		set_dev_entry_bit(devid, DEV_ENTRY_IW);
+}
+
+/* Writes the specific IOMMU for a device into the rlookup table */
+static void __init set_iommu_for_device(struct amd_iommu *iommu, u16 devid)
+{
+	amd_iommu_rlookup_table[devid] = iommu;
+}
+
+/*
+ * This function takes the device specific flags read from the ACPI
+ * table and sets up the device table entry with that information
+ */
+static void __init set_dev_entry_from_acpi(struct amd_iommu *iommu,
+					   u16 devid, u32 flags, u32 ext_flags)
+{
+	if (flags & ACPI_DEVFLAG_INITPASS)
+		set_dev_entry_bit(devid, DEV_ENTRY_INIT_PASS);
+	if (flags & ACPI_DEVFLAG_EXTINT)
+		set_dev_entry_bit(devid, DEV_ENTRY_EINT_PASS);
+	if (flags & ACPI_DEVFLAG_NMI)
+		set_dev_entry_bit(devid, DEV_ENTRY_NMI_PASS);
+	if (flags & ACPI_DEVFLAG_SYSMGT1)
+		set_dev_entry_bit(devid, DEV_ENTRY_SYSMGT1);
+	if (flags & ACPI_DEVFLAG_SYSMGT2)
+		set_dev_entry_bit(devid, DEV_ENTRY_SYSMGT2);
+	if (flags & ACPI_DEVFLAG_LINT0)
+		set_dev_entry_bit(devid, DEV_ENTRY_LINT0_PASS);
+	if (flags & ACPI_DEVFLAG_LINT1)
+		set_dev_entry_bit(devid, DEV_ENTRY_LINT1_PASS);
+
+	amd_iommu_apply_erratum_63(devid);
+
+	set_iommu_for_device(iommu, devid);
+}
+
+/*
+ * Reads the device exclusion range from ACPI and initialize IOMMU with
+ * it
+ */
+static void __init set_device_exclusion_range(u16 devid, struct ivmd_header *m)
+{
+	struct amd_iommu *iommu = amd_iommu_rlookup_table[devid];
+
+	if (!(m->flags & IVMD_FLAG_EXCL_RANGE))
+		return;
+
+	if (iommu) {
+		/*
+		 * We only can configure exclusion ranges per IOMMU, not
+		 * per device. But we can enable the exclusion range per
+		 * device. This is done here
+		 */
+		set_dev_entry_bit(m->devid, DEV_ENTRY_EX);
+		iommu->exclusion_start = m->range_start;
+		iommu->exclusion_length = m->range_length;
+	}
+}
+
+/*
+ * This function reads some important data from the IOMMU PCI space and
+ * initializes the driver data structure with it. It reads the hardware
+ * capabilities and the first/last device entries
+ */
+static void __init init_iommu_from_pci(struct amd_iommu *iommu)
+{
+	int cap_ptr = iommu->cap_ptr;
+	u32 range, misc, low, high;
+	int i, j;
+
+	pci_read_config_dword(iommu->dev, cap_ptr + MMIO_CAP_HDR_OFFSET,
+			      &iommu->cap);
+	pci_read_config_dword(iommu->dev, cap_ptr + MMIO_RANGE_OFFSET,
+			      &range);
+	pci_read_config_dword(iommu->dev, cap_ptr + MMIO_MISC_OFFSET,
+			      &misc);
+
+	iommu->first_device = calc_devid(MMIO_GET_BUS(range),
+					 MMIO_GET_FD(range));
+	iommu->last_device = calc_devid(MMIO_GET_BUS(range),
+					MMIO_GET_LD(range));
+	iommu->evt_msi_num = MMIO_MSI_NUM(misc);
+
+	if (!(iommu->cap & (1 << IOMMU_CAP_IOTLB)))
+		amd_iommu_iotlb_sup = false;
+
+	/* read extended feature bits */
+	low  = readl(iommu->mmio_base + MMIO_EXT_FEATURES);
+	high = readl(iommu->mmio_base + MMIO_EXT_FEATURES + 4);
+
+	iommu->features = ((u64)high << 32) | low;
+
+	if (!is_rd890_iommu(iommu->dev))
+		return;
+
+	/*
+	 * Some rd890 systems may not be fully reconfigured by the BIOS, so
+	 * it's necessary for us to store this information so it can be
+	 * reprogrammed on resume
+	 */
+
+	pci_read_config_dword(iommu->dev, iommu->cap_ptr + 4,
+			      &iommu->stored_addr_lo);
+	pci_read_config_dword(iommu->dev, iommu->cap_ptr + 8,
+			      &iommu->stored_addr_hi);
+
+	/* Low bit locks writes to configuration space */
+	iommu->stored_addr_lo &= ~1;
+
+	for (i = 0; i < 6; i++)
+		for (j = 0; j < 0x12; j++)
+			iommu->stored_l1[i][j] = iommu_read_l1(iommu, i, j);
+
+	for (i = 0; i < 0x83; i++)
+		iommu->stored_l2[i] = iommu_read_l2(iommu, i);
+}
+
+/*
+ * Takes a pointer to an AMD IOMMU entry in the ACPI table and
+ * initializes the hardware and our data structures with it.
+ */
+static void __init init_iommu_from_acpi(struct amd_iommu *iommu,
+					struct ivhd_header *h)
+{
+	u8 *p = (u8 *)h;
+	u8 *end = p, flags = 0;
+	u16 devid = 0, devid_start = 0, devid_to = 0;
+	u32 dev_i, ext_flags = 0;
+	bool alias = false;
+	struct ivhd_entry *e;
+
+	/*
+	 * First save the recommended feature enable bits from ACPI
+	 */
+	iommu->acpi_flags = h->flags;
+
+	/*
+	 * Done. Now parse the device entries
+	 */
+	p += sizeof(struct ivhd_header);
+	end += h->length;
+
+
+	while (p < end) {
+		e = (struct ivhd_entry *)p;
+		switch (e->type) {
+		case IVHD_DEV_ALL:
+
+			DUMP_printk("  DEV_ALL\t\t\t first devid: %02x:%02x.%x"
+				    " last device %02x:%02x.%x flags: %02x\n",
+				    PCI_BUS(iommu->first_device),
+				    PCI_SLOT(iommu->first_device),
+				    PCI_FUNC(iommu->first_device),
+				    PCI_BUS(iommu->last_device),
+				    PCI_SLOT(iommu->last_device),
+				    PCI_FUNC(iommu->last_device),
+				    e->flags);
+
+			for (dev_i = iommu->first_device;
+					dev_i <= iommu->last_device; ++dev_i)
+				set_dev_entry_from_acpi(iommu, dev_i,
+							e->flags, 0);
+			break;
+		case IVHD_DEV_SELECT:
+
+			DUMP_printk("  DEV_SELECT\t\t\t devid: %02x:%02x.%x "
+				    "flags: %02x\n",
+				    PCI_BUS(e->devid),
+				    PCI_SLOT(e->devid),
+				    PCI_FUNC(e->devid),
+				    e->flags);
+
+			devid = e->devid;
+			set_dev_entry_from_acpi(iommu, devid, e->flags, 0);
+			break;
+		case IVHD_DEV_SELECT_RANGE_START:
+
+			DUMP_printk("  DEV_SELECT_RANGE_START\t "
+				    "devid: %02x:%02x.%x flags: %02x\n",
+				    PCI_BUS(e->devid),
+				    PCI_SLOT(e->devid),
+				    PCI_FUNC(e->devid),
+				    e->flags);
+
+			devid_start = e->devid;
+			flags = e->flags;
+			ext_flags = 0;
+			alias = false;
+			break;
+		case IVHD_DEV_ALIAS:
+
+			DUMP_printk("  DEV_ALIAS\t\t\t devid: %02x:%02x.%x "
+				    "flags: %02x devid_to: %02x:%02x.%x\n",
+				    PCI_BUS(e->devid),
+				    PCI_SLOT(e->devid),
+				    PCI_FUNC(e->devid),
+				    e->flags,
+				    PCI_BUS(e->ext >> 8),
+				    PCI_SLOT(e->ext >> 8),
+				    PCI_FUNC(e->ext >> 8));
+
+			devid = e->devid;
+			devid_to = e->ext >> 8;
+			set_dev_entry_from_acpi(iommu, devid   , e->flags, 0);
+			set_dev_entry_from_acpi(iommu, devid_to, e->flags, 0);
+			amd_iommu_alias_table[devid] = devid_to;
+			break;
+		case IVHD_DEV_ALIAS_RANGE:
+
+			DUMP_printk("  DEV_ALIAS_RANGE\t\t "
+				    "devid: %02x:%02x.%x flags: %02x "
+				    "devid_to: %02x:%02x.%x\n",
+				    PCI_BUS(e->devid),
+				    PCI_SLOT(e->devid),
+				    PCI_FUNC(e->devid),
+				    e->flags,
+				    PCI_BUS(e->ext >> 8),
+				    PCI_SLOT(e->ext >> 8),
+				    PCI_FUNC(e->ext >> 8));
+
+			devid_start = e->devid;
+			flags = e->flags;
+			devid_to = e->ext >> 8;
+			ext_flags = 0;
+			alias = true;
+			break;
+		case IVHD_DEV_EXT_SELECT:
+
+			DUMP_printk("  DEV_EXT_SELECT\t\t devid: %02x:%02x.%x "
+				    "flags: %02x ext: %08x\n",
+				    PCI_BUS(e->devid),
+				    PCI_SLOT(e->devid),
+				    PCI_FUNC(e->devid),
+				    e->flags, e->ext);
+
+			devid = e->devid;
+			set_dev_entry_from_acpi(iommu, devid, e->flags,
+						e->ext);
+			break;
+		case IVHD_DEV_EXT_SELECT_RANGE:
+
+			DUMP_printk("  DEV_EXT_SELECT_RANGE\t devid: "
+				    "%02x:%02x.%x flags: %02x ext: %08x\n",
+				    PCI_BUS(e->devid),
+				    PCI_SLOT(e->devid),
+				    PCI_FUNC(e->devid),
+				    e->flags, e->ext);
+
+			devid_start = e->devid;
+			flags = e->flags;
+			ext_flags = e->ext;
+			alias = false;
+			break;
+		case IVHD_DEV_RANGE_END:
+
+			DUMP_printk("  DEV_RANGE_END\t\t devid: %02x:%02x.%x\n",
+				    PCI_BUS(e->devid),
+				    PCI_SLOT(e->devid),
+				    PCI_FUNC(e->devid));
+
+			devid = e->devid;
+			for (dev_i = devid_start; dev_i <= devid; ++dev_i) {
+				if (alias) {
+					amd_iommu_alias_table[dev_i] = devid_to;
+					set_dev_entry_from_acpi(iommu,
+						devid_to, flags, ext_flags);
+				}
+				set_dev_entry_from_acpi(iommu, dev_i,
+							flags, ext_flags);
+			}
+			break;
+		default:
+			break;
+		}
+
+		p += ivhd_entry_length(p);
+	}
+}
+
+/* Initializes the device->iommu mapping for the driver */
+static int __init init_iommu_devices(struct amd_iommu *iommu)
+{
+	u32 i;
+
+	for (i = iommu->first_device; i <= iommu->last_device; ++i)
+		set_iommu_for_device(iommu, i);
+
+	return 0;
+}
+
+static void __init free_iommu_one(struct amd_iommu *iommu)
+{
+	free_command_buffer(iommu);
+	free_event_buffer(iommu);
+	iommu_unmap_mmio_space(iommu);
+}
+
+static void __init free_iommu_all(void)
+{
+	struct amd_iommu *iommu, *next;
+
+	for_each_iommu_safe(iommu, next) {
+		list_del(&iommu->list);
+		free_iommu_one(iommu);
+		kfree(iommu);
+	}
+}
+
+/*
+ * This function clues the initialization function for one IOMMU
+ * together and also allocates the command buffer and programs the
+ * hardware. It does NOT enable the IOMMU. This is done afterwards.
+ */
+static int __init init_iommu_one(struct amd_iommu *iommu, struct ivhd_header *h)
+{
+	spin_lock_init(&iommu->lock);
+
+	/* Add IOMMU to internal data structures */
+	list_add_tail(&iommu->list, &amd_iommu_list);
+	iommu->index             = amd_iommus_present++;
+
+	if (unlikely(iommu->index >= MAX_IOMMUS)) {
+		WARN(1, "AMD-Vi: System has more IOMMUs than supported by this driver\n");
+		return -ENOSYS;
+	}
+
+	/* Index is fine - add IOMMU to the array */
+	amd_iommus[iommu->index] = iommu;
+
+	/*
+	 * Copy data from ACPI table entry to the iommu struct
+	 */
+	iommu->dev = pci_get_bus_and_slot(PCI_BUS(h->devid), h->devid & 0xff);
+	if (!iommu->dev)
+		return 1;
+
+	iommu->cap_ptr = h->cap_ptr;
+	iommu->pci_seg = h->pci_seg;
+	iommu->mmio_phys = h->mmio_phys;
+	iommu->mmio_base = iommu_map_mmio_space(h->mmio_phys);
+	if (!iommu->mmio_base)
+		return -ENOMEM;
+
+	iommu->cmd_buf = alloc_command_buffer(iommu);
+	if (!iommu->cmd_buf)
+		return -ENOMEM;
+
+	iommu->evt_buf = alloc_event_buffer(iommu);
+	if (!iommu->evt_buf)
+		return -ENOMEM;
+
+	iommu->int_enabled = false;
+
+	init_iommu_from_pci(iommu);
+	init_iommu_from_acpi(iommu, h);
+	init_iommu_devices(iommu);
+
+	if (iommu->cap & (1UL << IOMMU_CAP_NPCACHE))
+		amd_iommu_np_cache = true;
+
+	return pci_enable_device(iommu->dev);
+}
+
+/*
+ * Iterates over all IOMMU entries in the ACPI table, allocates the
+ * IOMMU structure and initializes it with init_iommu_one()
+ */
+static int __init init_iommu_all(struct acpi_table_header *table)
+{
+	u8 *p = (u8 *)table, *end = (u8 *)table;
+	struct ivhd_header *h;
+	struct amd_iommu *iommu;
+	int ret;
+
+	end += table->length;
+	p += IVRS_HEADER_LENGTH;
+
+	while (p < end) {
+		h = (struct ivhd_header *)p;
+		switch (*p) {
+		case ACPI_IVHD_TYPE:
+
+			DUMP_printk("device: %02x:%02x.%01x cap: %04x "
+				    "seg: %d flags: %01x info %04x\n",
+				    PCI_BUS(h->devid), PCI_SLOT(h->devid),
+				    PCI_FUNC(h->devid), h->cap_ptr,
+				    h->pci_seg, h->flags, h->info);
+			DUMP_printk("       mmio-addr: %016llx\n",
+				    h->mmio_phys);
+
+			iommu = kzalloc(sizeof(struct amd_iommu), GFP_KERNEL);
+			if (iommu == NULL) {
+				amd_iommu_init_err = -ENOMEM;
+				return 0;
+			}
+
+			ret = init_iommu_one(iommu, h);
+			if (ret) {
+				amd_iommu_init_err = ret;
+				return 0;
+			}
+			break;
+		default:
+			break;
+		}
+		p += h->length;
+
+	}
+	WARN_ON(p != end);
+
+	return 0;
+}
+
+/****************************************************************************
+ *
+ * The following functions initialize the MSI interrupts for all IOMMUs
+ * in the system. Its a bit challenging because there could be multiple
+ * IOMMUs per PCI BDF but we can call pci_enable_msi(x) only once per
+ * pci_dev.
+ *
+ ****************************************************************************/
+
+static int iommu_setup_msi(struct amd_iommu *iommu)
+{
+	int r;
+
+	if (pci_enable_msi(iommu->dev))
+		return 1;
+
+	r = request_threaded_irq(iommu->dev->irq,
+				 amd_iommu_int_handler,
+				 amd_iommu_int_thread,
+				 0, "AMD-Vi",
+				 iommu->dev);
+
+	if (r) {
+		pci_disable_msi(iommu->dev);
+		return 1;
+	}
+
+	iommu->int_enabled = true;
+	iommu_feature_enable(iommu, CONTROL_EVT_INT_EN);
+
+	return 0;
+}
+
+static int iommu_init_msi(struct amd_iommu *iommu)
+{
+	if (iommu->int_enabled)
+		return 0;
+
+	if (pci_find_capability(iommu->dev, PCI_CAP_ID_MSI))
+		return iommu_setup_msi(iommu);
+
+	return 1;
+}
+
+/****************************************************************************
+ *
+ * The next functions belong to the third pass of parsing the ACPI
+ * table. In this last pass the memory mapping requirements are
+ * gathered (like exclusion and unity mapping reanges).
+ *
+ ****************************************************************************/
+
+static void __init free_unity_maps(void)
+{
+	struct unity_map_entry *entry, *next;
+
+	list_for_each_entry_safe(entry, next, &amd_iommu_unity_map, list) {
+		list_del(&entry->list);
+		kfree(entry);
+	}
+}
+
+/* called when we find an exclusion range definition in ACPI */
+static int __init init_exclusion_range(struct ivmd_header *m)
+{
+	int i;
+
+	switch (m->type) {
+	case ACPI_IVMD_TYPE:
+		set_device_exclusion_range(m->devid, m);
+		break;
+	case ACPI_IVMD_TYPE_ALL:
+		for (i = 0; i <= amd_iommu_last_bdf; ++i)
+			set_device_exclusion_range(i, m);
+		break;
+	case ACPI_IVMD_TYPE_RANGE:
+		for (i = m->devid; i <= m->aux; ++i)
+			set_device_exclusion_range(i, m);
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+/* called for unity map ACPI definition */
+static int __init init_unity_map_range(struct ivmd_header *m)
+{
+	struct unity_map_entry *e = 0;
+	char *s;
+
+	e = kzalloc(sizeof(*e), GFP_KERNEL);
+	if (e == NULL)
+		return -ENOMEM;
+
+	switch (m->type) {
+	default:
+		kfree(e);
+		return 0;
+	case ACPI_IVMD_TYPE:
+		s = "IVMD_TYPEi\t\t\t";
+		e->devid_start = e->devid_end = m->devid;
+		break;
+	case ACPI_IVMD_TYPE_ALL:
+		s = "IVMD_TYPE_ALL\t\t";
+		e->devid_start = 0;
+		e->devid_end = amd_iommu_last_bdf;
+		break;
+	case ACPI_IVMD_TYPE_RANGE:
+		s = "IVMD_TYPE_RANGE\t\t";
+		e->devid_start = m->devid;
+		e->devid_end = m->aux;
+		break;
+	}
+	e->address_start = PAGE_ALIGN(m->range_start);
+	e->address_end = e->address_start + PAGE_ALIGN(m->range_length);
+	e->prot = m->flags >> 1;
+
+	DUMP_printk("%s devid_start: %02x:%02x.%x devid_end: %02x:%02x.%x"
+		    " range_start: %016llx range_end: %016llx flags: %x\n", s,
+		    PCI_BUS(e->devid_start), PCI_SLOT(e->devid_start),
+		    PCI_FUNC(e->devid_start), PCI_BUS(e->devid_end),
+		    PCI_SLOT(e->devid_end), PCI_FUNC(e->devid_end),
+		    e->address_start, e->address_end, m->flags);
+
+	list_add_tail(&e->list, &amd_iommu_unity_map);
+
+	return 0;
+}
+
+/* iterates over all memory definitions we find in the ACPI table */
+static int __init init_memory_definitions(struct acpi_table_header *table)
+{
+	u8 *p = (u8 *)table, *end = (u8 *)table;
+	struct ivmd_header *m;
+
+	end += table->length;
+	p += IVRS_HEADER_LENGTH;
+
+	while (p < end) {
+		m = (struct ivmd_header *)p;
+		if (m->flags & IVMD_FLAG_EXCL_RANGE)
+			init_exclusion_range(m);
+		else if (m->flags & IVMD_FLAG_UNITY_MAP)
+			init_unity_map_range(m);
+
+		p += m->length;
+	}
+
+	return 0;
+}
+
+/*
+ * Init the device table to not allow DMA access for devices and
+ * suppress all page faults
+ */
+static void init_device_table(void)
+{
+	u32 devid;
+
+	for (devid = 0; devid <= amd_iommu_last_bdf; ++devid) {
+		set_dev_entry_bit(devid, DEV_ENTRY_VALID);
+		set_dev_entry_bit(devid, DEV_ENTRY_TRANSLATION);
+	}
+}
+
+static void iommu_init_flags(struct amd_iommu *iommu)
+{
+	iommu->acpi_flags & IVHD_FLAG_HT_TUN_EN_MASK ?
+		iommu_feature_enable(iommu, CONTROL_HT_TUN_EN) :
+		iommu_feature_disable(iommu, CONTROL_HT_TUN_EN);
+
+	iommu->acpi_flags & IVHD_FLAG_PASSPW_EN_MASK ?
+		iommu_feature_enable(iommu, CONTROL_PASSPW_EN) :
+		iommu_feature_disable(iommu, CONTROL_PASSPW_EN);
+
+	iommu->acpi_flags & IVHD_FLAG_RESPASSPW_EN_MASK ?
+		iommu_feature_enable(iommu, CONTROL_RESPASSPW_EN) :
+		iommu_feature_disable(iommu, CONTROL_RESPASSPW_EN);
+
+	iommu->acpi_flags & IVHD_FLAG_ISOC_EN_MASK ?
+		iommu_feature_enable(iommu, CONTROL_ISOC_EN) :
+		iommu_feature_disable(iommu, CONTROL_ISOC_EN);
+
+	/*
+	 * make IOMMU memory accesses cache coherent
+	 */
+	iommu_feature_enable(iommu, CONTROL_COHERENT_EN);
+}
+
+static void iommu_apply_resume_quirks(struct amd_iommu *iommu)
+{
+	int i, j;
+	u32 ioc_feature_control;
+	struct pci_dev *pdev = NULL;
+
+	/* RD890 BIOSes may not have completely reconfigured the iommu */
+	if (!is_rd890_iommu(iommu->dev))
+		return;
+
+	/*
+	 * First, we need to ensure that the iommu is enabled. This is
+	 * controlled by a register in the northbridge
+	 */
+	pdev = pci_get_bus_and_slot(iommu->dev->bus->number, PCI_DEVFN(0, 0));
+
+	if (!pdev)
+		return;
+
+	/* Select Northbridge indirect register 0x75 and enable writing */
+	pci_write_config_dword(pdev, 0x60, 0x75 | (1 << 7));
+	pci_read_config_dword(pdev, 0x64, &ioc_feature_control);
+
+	/* Enable the iommu */
+	if (!(ioc_feature_control & 0x1))
+		pci_write_config_dword(pdev, 0x64, ioc_feature_control | 1);
+
+	pci_dev_put(pdev);
+
+	/* Restore the iommu BAR */
+	pci_write_config_dword(iommu->dev, iommu->cap_ptr + 4,
+			       iommu->stored_addr_lo);
+	pci_write_config_dword(iommu->dev, iommu->cap_ptr + 8,
+			       iommu->stored_addr_hi);
+
+	/* Restore the l1 indirect regs for each of the 6 l1s */
+	for (i = 0; i < 6; i++)
+		for (j = 0; j < 0x12; j++)
+			iommu_write_l1(iommu, i, j, iommu->stored_l1[i][j]);
+
+	/* Restore the l2 indirect regs */
+	for (i = 0; i < 0x83; i++)
+		iommu_write_l2(iommu, i, iommu->stored_l2[i]);
+
+	/* Lock PCI setup registers */
+	pci_write_config_dword(iommu->dev, iommu->cap_ptr + 4,
+			       iommu->stored_addr_lo | 1);
+}
+
+/*
+ * This function finally enables all IOMMUs found in the system after
+ * they have been initialized
+ */
+static void enable_iommus(void)
+{
+	struct amd_iommu *iommu;
+
+	for_each_iommu(iommu) {
+		iommu_disable(iommu);
+		iommu_init_flags(iommu);
+		iommu_set_device_table(iommu);
+		iommu_enable_command_buffer(iommu);
+		iommu_enable_event_buffer(iommu);
+		iommu_set_exclusion_range(iommu);
+		iommu_init_msi(iommu);
+		iommu_enable(iommu);
+		iommu_flush_all_caches(iommu);
+	}
+}
+
+static void disable_iommus(void)
+{
+	struct amd_iommu *iommu;
+
+	for_each_iommu(iommu)
+		iommu_disable(iommu);
+}
+
+/*
+ * Suspend/Resume support
+ * disable suspend until real resume implemented
+ */
+
+static void amd_iommu_resume(void)
+{
+	struct amd_iommu *iommu;
+
+	for_each_iommu(iommu)
+		iommu_apply_resume_quirks(iommu);
+
+	/* re-load the hardware */
+	enable_iommus();
+
+	/*
+	 * we have to flush after the IOMMUs are enabled because a
+	 * disabled IOMMU will never execute the commands we send
+	 */
+	for_each_iommu(iommu)
+		iommu_flush_all_caches(iommu);
+}
+
+static int amd_iommu_suspend(void)
+{
+	/* disable IOMMUs to go out of the way for BIOS */
+	disable_iommus();
+
+	return 0;
+}
+
+static struct syscore_ops amd_iommu_syscore_ops = {
+	.suspend = amd_iommu_suspend,
+	.resume = amd_iommu_resume,
+};
+
+/*
+ * 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.
+ *
+ * This function basically parses the ACPI table for AMD IOMMU (IVRS)
+ * three times:
+ *
+ *	1 pass) Find the highest PCI device id the driver has to handle.
+ *		Upon this information the size of the data structures is
+ *		determined that needs to be allocated.
+ *
+ *	2 pass) Initialize the data structures just allocated with the
+ *		information in the ACPI table about available AMD IOMMUs
+ *		in the system. It also maps the PCI devices in the
+ *		system to specific IOMMUs
+ *
+ *	3 pass) After the basic data structures are allocated and
+ *		initialized we update them with information about memory
+ *		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
+ * step we do some Linux specific things like registering the driver in
+ * 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 i, ret = 0;
+
+	/*
+	 * First parse ACPI tables to find the largest Bus/Dev/Func
+	 * we need to handle. Upon this information the shared data
+	 * structures for the IOMMUs in the system will be allocated
+	 */
+	if (acpi_table_parse("IVRS", find_last_devid_acpi) != 0)
+		return -ENODEV;
+
+	ret = amd_iommu_init_err;
+	if (ret)
+		goto out;
+
+	dev_table_size     = tbl_size(DEV_TABLE_ENTRY_SIZE);
+	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 */
+	amd_iommu_dev_table = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
+				      get_order(dev_table_size));
+	if (amd_iommu_dev_table == NULL)
+		goto out;
+
+	/*
+	 * Alias table - map PCI Bus/Dev/Func to Bus/Dev/Func the
+	 * IOMMU see for that device
+	 */
+	amd_iommu_alias_table = (void *)__get_free_pages(GFP_KERNEL,
+			get_order(alias_table_size));
+	if (amd_iommu_alias_table == NULL)
+		goto free;
+
+	/* IOMMU rlookup table - find the IOMMU for a specific device */
+	amd_iommu_rlookup_table = (void *)__get_free_pages(
+			GFP_KERNEL | __GFP_ZERO,
+			get_order(rlookup_table_size));
+	if (amd_iommu_rlookup_table == NULL)
+		goto free;
+
+	amd_iommu_pd_alloc_bitmap = (void *)__get_free_pages(
+					    GFP_KERNEL | __GFP_ZERO,
+					    get_order(MAX_DOMAIN_ID/8));
+	if (amd_iommu_pd_alloc_bitmap == NULL)
+		goto free;
+
+	/* init the device table */
+	init_device_table();
+
+	/*
+	 * let all alias entries point to itself
+	 */
+	for (i = 0; i <= amd_iommu_last_bdf; ++i)
+		amd_iommu_alias_table[i] = i;
+
+	/*
+	 * never allocate domain 0 because its used as the non-allocated and
+	 * error value placeholder
+	 */
+	amd_iommu_pd_alloc_bitmap[0] = 1;
+
+	spin_lock_init(&amd_iommu_pd_lock);
+
+	/*
+	 * now the data structures are allocated and basically initialized
+	 * start the real acpi table scan
+	 */
+	ret = -ENODEV;
+	if (acpi_table_parse("IVRS", init_iommu_all) != 0)
+		goto free;
+
+	if (amd_iommu_init_err) {
+		ret = amd_iommu_init_err;
+		goto free;
+	}
+
+	if (acpi_table_parse("IVRS", init_memory_definitions) != 0)
+		goto free;
+
+	if (amd_iommu_init_err) {
+		ret = amd_iommu_init_err;
+		goto free;
+	}
+
+	ret = amd_iommu_init_devices();
+	if (ret)
+		goto free;
+
+	enable_iommus();
+
+	if (iommu_pass_through)
+		ret = amd_iommu_init_passthrough();
+	else
+		ret = amd_iommu_init_dma_ops();
+
+	if (ret)
+		goto free_disable;
+
+	amd_iommu_init_api();
+
+	amd_iommu_init_notifier();
+
+	register_syscore_ops(&amd_iommu_syscore_ops);
+
+	if (iommu_pass_through)
+		goto out;
+
+	if (amd_iommu_unmap_flush)
+		printk(KERN_INFO "AMD-Vi: IO/TLB flush on unmap enabled\n");
+	else
+		printk(KERN_INFO "AMD-Vi: Lazy IO/TLB flushing enabled\n");
+
+	x86_platform.iommu_shutdown = disable_iommus;
+out:
+	return ret;
+
+free_disable:
+	disable_iommus();
+
+free:
+	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;
+}
+
+/****************************************************************************
+ *
+ * Early detect code. This code runs at IOMMU detection time in the DMA
+ * layer. It just looks if there is an IVRS ACPI table to detect AMD
+ * IOMMUs
+ *
+ ****************************************************************************/
+static int __init early_amd_iommu_detect(struct acpi_table_header *table)
+{
+	return 0;
+}
+
+int __init amd_iommu_detect(void)
+{
+	if (no_iommu || (iommu_detected && !gart_iommu_aperture))
+		return -ENODEV;
+
+	if (amd_iommu_disabled)
+		return -ENODEV;
+
+	if (acpi_table_parse("IVRS", early_amd_iommu_detect) == 0) {
+		iommu_detected = 1;
+		amd_iommu_detected = 1;
+		x86_init.iommu.iommu_init = amd_iommu_init;
+
+		/* Make sure ACS will be enabled */
+		pci_request_acs();
+		return 1;
+	}
+	return -ENODEV;
+}
+
+/****************************************************************************
+ *
+ * Parsing functions for the AMD IOMMU specific kernel command line
+ * options.
+ *
+ ****************************************************************************/
+
+static int __init parse_amd_iommu_dump(char *str)
+{
+	amd_iommu_dump = true;
+
+	return 1;
+}
+
+static int __init parse_amd_iommu_options(char *str)
+{
+	for (; *str; ++str) {
+		if (strncmp(str, "fullflush", 9) == 0)
+			amd_iommu_unmap_flush = true;
+		if (strncmp(str, "off", 3) == 0)
+			amd_iommu_disabled = true;
+	}
+
+	return 1;
+}
+
+__setup("amd_iommu_dump", parse_amd_iommu_dump);
+__setup("amd_iommu=", parse_amd_iommu_options);
+
+IOMMU_INIT_FINISH(amd_iommu_detect,
+		  gart_iommu_hole_init,
+		  0,
+		  0);
diff --git a/drivers/iommu/amd_iommu_proto.h b/drivers/iommu/amd_iommu_proto.h
new file mode 100644
index 00000000000..7ffaa64410b
--- /dev/null
+++ b/drivers/iommu/amd_iommu_proto.h
@@ -0,0 +1,54 @@
+/*
+ * Copyright (C) 2009-2010 Advanced Micro Devices, Inc.
+ * Author: Joerg Roedel <joerg.roedel@amd.com>
+ *
+ * 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#ifndef _ASM_X86_AMD_IOMMU_PROTO_H
+#define _ASM_X86_AMD_IOMMU_PROTO_H
+
+#include "amd_iommu_types.h"
+
+extern int amd_iommu_init_dma_ops(void);
+extern int amd_iommu_init_passthrough(void);
+extern irqreturn_t amd_iommu_int_thread(int irq, void *data);
+extern irqreturn_t amd_iommu_int_handler(int irq, void *data);
+extern void amd_iommu_apply_erratum_63(u16 devid);
+extern void amd_iommu_reset_cmd_buffer(struct amd_iommu *iommu);
+extern int amd_iommu_init_devices(void);
+extern void amd_iommu_uninit_devices(void);
+extern void amd_iommu_init_notifier(void);
+extern void amd_iommu_init_api(void);
+#ifndef CONFIG_AMD_IOMMU_STATS
+
+static inline void amd_iommu_stats_init(void) { }
+
+#endif /* !CONFIG_AMD_IOMMU_STATS */
+
+static inline bool is_rd890_iommu(struct pci_dev *pdev)
+{
+	return (pdev->vendor == PCI_VENDOR_ID_ATI) &&
+	       (pdev->device == PCI_DEVICE_ID_RD890_IOMMU);
+}
+
+static inline bool iommu_feature(struct amd_iommu *iommu, u64 f)
+{
+	if (!(iommu->cap & (1 << IOMMU_CAP_EFR)))
+		return false;
+
+	return !!(iommu->features & f);
+}
+
+#endif /* _ASM_X86_AMD_IOMMU_PROTO_H  */
diff --git a/drivers/iommu/amd_iommu_types.h b/drivers/iommu/amd_iommu_types.h
new file mode 100644
index 00000000000..5b9c5075e81
--- /dev/null
+++ b/drivers/iommu/amd_iommu_types.h
@@ -0,0 +1,585 @@
+/*
+ * Copyright (C) 2007-2010 Advanced Micro Devices, Inc.
+ * Author: Joerg Roedel <joerg.roedel@amd.com>
+ *         Leo Duran <leo.duran@amd.com>
+ *
+ * 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, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
+ */
+
+#ifndef _ASM_X86_AMD_IOMMU_TYPES_H
+#define _ASM_X86_AMD_IOMMU_TYPES_H
+
+#include <linux/types.h>
+#include <linux/mutex.h>
+#include <linux/list.h>
+#include <linux/spinlock.h>
+
+/*
+ * Maximum number of IOMMUs supported
+ */
+#define MAX_IOMMUS	32
+
+/*
+ * some size calculation constants
+ */
+#define DEV_TABLE_ENTRY_SIZE		32
+#define ALIAS_TABLE_ENTRY_SIZE		2
+#define RLOOKUP_TABLE_ENTRY_SIZE	(sizeof(void *))
+
+/* Length of the MMIO region for the AMD IOMMU */
+#define MMIO_REGION_LENGTH       0x4000
+
+/* Capability offsets used by the driver */
+#define MMIO_CAP_HDR_OFFSET	0x00
+#define MMIO_RANGE_OFFSET	0x0c
+#define MMIO_MISC_OFFSET	0x10
+
+/* Masks, shifts and macros to parse the device range capability */
+#define MMIO_RANGE_LD_MASK	0xff000000
+#define MMIO_RANGE_FD_MASK	0x00ff0000
+#define MMIO_RANGE_BUS_MASK	0x0000ff00
+#define MMIO_RANGE_LD_SHIFT	24
+#define MMIO_RANGE_FD_SHIFT	16
+#define MMIO_RANGE_BUS_SHIFT	8
+#define MMIO_GET_LD(x)  (((x) & MMIO_RANGE_LD_MASK) >> MMIO_RANGE_LD_SHIFT)
+#define MMIO_GET_FD(x)  (((x) & MMIO_RANGE_FD_MASK) >> MMIO_RANGE_FD_SHIFT)
+#define MMIO_GET_BUS(x) (((x) & MMIO_RANGE_BUS_MASK) >> MMIO_RANGE_BUS_SHIFT)
+#define MMIO_MSI_NUM(x)	((x) & 0x1f)
+
+/* Flag masks for the AMD IOMMU exclusion range */
+#define MMIO_EXCL_ENABLE_MASK 0x01ULL
+#define MMIO_EXCL_ALLOW_MASK  0x02ULL
+
+/* Used offsets into the MMIO space */
+#define MMIO_DEV_TABLE_OFFSET   0x0000
+#define MMIO_CMD_BUF_OFFSET     0x0008
+#define MMIO_EVT_BUF_OFFSET     0x0010
+#define MMIO_CONTROL_OFFSET     0x0018
+#define MMIO_EXCL_BASE_OFFSET   0x0020
+#define MMIO_EXCL_LIMIT_OFFSET  0x0028
+#define MMIO_EXT_FEATURES	0x0030
+#define MMIO_CMD_HEAD_OFFSET	0x2000
+#define MMIO_CMD_TAIL_OFFSET	0x2008
+#define MMIO_EVT_HEAD_OFFSET	0x2010
+#define MMIO_EVT_TAIL_OFFSET	0x2018
+#define MMIO_STATUS_OFFSET	0x2020
+
+
+/* Extended Feature Bits */
+#define FEATURE_PREFETCH	(1ULL<<0)
+#define FEATURE_PPR		(1ULL<<1)
+#define FEATURE_X2APIC		(1ULL<<2)
+#define FEATURE_NX		(1ULL<<3)
+#define FEATURE_GT		(1ULL<<4)
+#define FEATURE_IA		(1ULL<<6)
+#define FEATURE_GA		(1ULL<<7)
+#define FEATURE_HE		(1ULL<<8)
+#define FEATURE_PC		(1ULL<<9)
+
+/* MMIO status bits */
+#define MMIO_STATUS_COM_WAIT_INT_MASK	0x04
+
+/* event logging constants */
+#define EVENT_ENTRY_SIZE	0x10
+#define EVENT_TYPE_SHIFT	28
+#define EVENT_TYPE_MASK		0xf
+#define EVENT_TYPE_ILL_DEV	0x1
+#define EVENT_TYPE_IO_FAULT	0x2
+#define EVENT_TYPE_DEV_TAB_ERR	0x3
+#define EVENT_TYPE_PAGE_TAB_ERR	0x4
+#define EVENT_TYPE_ILL_CMD	0x5
+#define EVENT_TYPE_CMD_HARD_ERR	0x6
+#define EVENT_TYPE_IOTLB_INV_TO	0x7
+#define EVENT_TYPE_INV_DEV_REQ	0x8
+#define EVENT_DEVID_MASK	0xffff
+#define EVENT_DEVID_SHIFT	0
+#define EVENT_DOMID_MASK	0xffff
+#define EVENT_DOMID_SHIFT	0
+#define EVENT_FLAGS_MASK	0xfff
+#define EVENT_FLAGS_SHIFT	0x10
+
+/* feature control bits */
+#define CONTROL_IOMMU_EN        0x00ULL
+#define CONTROL_HT_TUN_EN       0x01ULL
+#define CONTROL_EVT_LOG_EN      0x02ULL
+#define CONTROL_EVT_INT_EN      0x03ULL
+#define CONTROL_COMWAIT_EN      0x04ULL
+#define CONTROL_PASSPW_EN       0x08ULL
+#define CONTROL_RESPASSPW_EN    0x09ULL
+#define CONTROL_COHERENT_EN     0x0aULL
+#define CONTROL_ISOC_EN         0x0bULL
+#define CONTROL_CMDBUF_EN       0x0cULL
+#define CONTROL_PPFLOG_EN       0x0dULL
+#define CONTROL_PPFINT_EN       0x0eULL
+
+/* command specific defines */
+#define CMD_COMPL_WAIT          0x01
+#define CMD_INV_DEV_ENTRY       0x02
+#define CMD_INV_IOMMU_PAGES	0x03
+#define CMD_INV_IOTLB_PAGES	0x04
+#define CMD_INV_ALL		0x08
+
+#define CMD_COMPL_WAIT_STORE_MASK	0x01
+#define CMD_COMPL_WAIT_INT_MASK		0x02
+#define CMD_INV_IOMMU_PAGES_SIZE_MASK	0x01
+#define CMD_INV_IOMMU_PAGES_PDE_MASK	0x02
+
+#define CMD_INV_IOMMU_ALL_PAGES_ADDRESS	0x7fffffffffffffffULL
+
+/* macros and definitions for device table entries */
+#define DEV_ENTRY_VALID         0x00
+#define DEV_ENTRY_TRANSLATION   0x01
+#define DEV_ENTRY_IR            0x3d
+#define DEV_ENTRY_IW            0x3e
+#define DEV_ENTRY_NO_PAGE_FAULT	0x62
+#define DEV_ENTRY_EX            0x67
+#define DEV_ENTRY_SYSMGT1       0x68
+#define DEV_ENTRY_SYSMGT2       0x69
+#define DEV_ENTRY_INIT_PASS     0xb8
+#define DEV_ENTRY_EINT_PASS     0xb9
+#define DEV_ENTRY_NMI_PASS      0xba
+#define DEV_ENTRY_LINT0_PASS    0xbe
+#define DEV_ENTRY_LINT1_PASS    0xbf
+#define DEV_ENTRY_MODE_MASK	0x07
+#define DEV_ENTRY_MODE_SHIFT	0x09
+
+/* constants to configure the command buffer */
+#define CMD_BUFFER_SIZE    8192
+#define CMD_BUFFER_UNINITIALIZED 1
+#define CMD_BUFFER_ENTRIES 512
+#define MMIO_CMD_SIZE_SHIFT 56
+#define MMIO_CMD_SIZE_512 (0x9ULL << MMIO_CMD_SIZE_SHIFT)
+
+/* constants for event buffer handling */
+#define EVT_BUFFER_SIZE		8192 /* 512 entries */
+#define EVT_LEN_MASK		(0x9ULL << 56)
+
+#define PAGE_MODE_NONE    0x00
+#define PAGE_MODE_1_LEVEL 0x01
+#define PAGE_MODE_2_LEVEL 0x02
+#define PAGE_MODE_3_LEVEL 0x03
+#define PAGE_MODE_4_LEVEL 0x04
+#define PAGE_MODE_5_LEVEL 0x05
+#define PAGE_MODE_6_LEVEL 0x06
+
+#define PM_LEVEL_SHIFT(x)	(12 + ((x) * 9))
+#define PM_LEVEL_SIZE(x)	(((x) < 6) ? \
+				  ((1ULL << PM_LEVEL_SHIFT((x))) - 1): \
+				   (0xffffffffffffffffULL))
+#define PM_LEVEL_INDEX(x, a)	(((a) >> PM_LEVEL_SHIFT((x))) & 0x1ffULL)
+#define PM_LEVEL_ENC(x)		(((x) << 9) & 0xe00ULL)
+#define PM_LEVEL_PDE(x, a)	((a) | PM_LEVEL_ENC((x)) | \
+				 IOMMU_PTE_P | IOMMU_PTE_IR | IOMMU_PTE_IW)
+#define PM_PTE_LEVEL(pte)	(((pte) >> 9) & 0x7ULL)
+
+#define PM_MAP_4k		0
+#define PM_ADDR_MASK		0x000ffffffffff000ULL
+#define PM_MAP_MASK(lvl)	(PM_ADDR_MASK & \
+				(~((1ULL << (12 + ((lvl) * 9))) - 1)))
+#define PM_ALIGNED(lvl, addr)	((PM_MAP_MASK(lvl) & (addr)) == (addr))
+
+/*
+ * Returns the page table level to use for a given page size
+ * Pagesize is expected to be a power-of-two
+ */
+#define PAGE_SIZE_LEVEL(pagesize) \
+		((__ffs(pagesize) - 12) / 9)
+/*
+ * Returns the number of ptes to use for a given page size
+ * Pagesize is expected to be a power-of-two
+ */
+#define PAGE_SIZE_PTE_COUNT(pagesize) \
+		(1ULL << ((__ffs(pagesize) - 12) % 9))
+
+/*
+ * Aligns a given io-virtual address to a given page size
+ * Pagesize is expected to be a power-of-two
+ */
+#define PAGE_SIZE_ALIGN(address, pagesize) \
+		((address) & ~((pagesize) - 1))
+/*
+ * Creates an IOMMU PTE for an address an a given pagesize
+ * The PTE has no permission bits set
+ * Pagesize is expected to be a power-of-two larger than 4096
+ */
+#define PAGE_SIZE_PTE(address, pagesize)		\
+		(((address) | ((pagesize) - 1)) &	\
+		 (~(pagesize >> 1)) & PM_ADDR_MASK)
+
+/*
+ * Takes a PTE value with mode=0x07 and returns the page size it maps
+ */
+#define PTE_PAGE_SIZE(pte) \
+	(1ULL << (1 + ffz(((pte) | 0xfffULL))))
+
+#define IOMMU_PTE_P  (1ULL << 0)
+#define IOMMU_PTE_TV (1ULL << 1)
+#define IOMMU_PTE_U  (1ULL << 59)
+#define IOMMU_PTE_FC (1ULL << 60)
+#define IOMMU_PTE_IR (1ULL << 61)
+#define IOMMU_PTE_IW (1ULL << 62)
+
+#define DTE_FLAG_IOTLB	0x01
+
+#define IOMMU_PAGE_MASK (((1ULL << 52) - 1) & ~0xfffULL)
+#define IOMMU_PTE_PRESENT(pte) ((pte) & IOMMU_PTE_P)
+#define IOMMU_PTE_PAGE(pte) (phys_to_virt((pte) & IOMMU_PAGE_MASK))
+#define IOMMU_PTE_MODE(pte) (((pte) >> 9) & 0x07)
+
+#define IOMMU_PROT_MASK 0x03
+#define IOMMU_PROT_IR 0x01
+#define IOMMU_PROT_IW 0x02
+
+/* IOMMU capabilities */
+#define IOMMU_CAP_IOTLB   24
+#define IOMMU_CAP_NPCACHE 26
+#define IOMMU_CAP_EFR     27
+
+#define MAX_DOMAIN_ID 65536
+
+/* FIXME: move this macro to <linux/pci.h> */
+#define PCI_BUS(x) (((x) >> 8) & 0xff)
+
+/* Protection domain flags */
+#define PD_DMA_OPS_MASK		(1UL << 0) /* domain used for dma_ops */
+#define PD_DEFAULT_MASK		(1UL << 1) /* domain is a default dma_ops
+					      domain for an IOMMU */
+#define PD_PASSTHROUGH_MASK	(1UL << 2) /* domain has no page
+					      translation */
+
+extern bool amd_iommu_dump;
+#define DUMP_printk(format, arg...)					\
+	do {								\
+		if (amd_iommu_dump)						\
+			printk(KERN_INFO "AMD-Vi: " format, ## arg);	\
+	} while(0);
+
+/* global flag if IOMMUs cache non-present entries */
+extern bool amd_iommu_np_cache;
+/* Only true if all IOMMUs support device IOTLBs */
+extern bool amd_iommu_iotlb_sup;
+
+/*
+ * Make iterating over all IOMMUs easier
+ */
+#define for_each_iommu(iommu) \
+	list_for_each_entry((iommu), &amd_iommu_list, list)
+#define for_each_iommu_safe(iommu, next) \
+	list_for_each_entry_safe((iommu), (next), &amd_iommu_list, list)
+
+#define APERTURE_RANGE_SHIFT	27	/* 128 MB */
+#define APERTURE_RANGE_SIZE	(1ULL << APERTURE_RANGE_SHIFT)
+#define APERTURE_RANGE_PAGES	(APERTURE_RANGE_SIZE >> PAGE_SHIFT)
+#define APERTURE_MAX_RANGES	32	/* allows 4GB of DMA address space */
+#define APERTURE_RANGE_INDEX(a)	((a) >> APERTURE_RANGE_SHIFT)
+#define APERTURE_PAGE_INDEX(a)	(((a) >> 21) & 0x3fULL)
+
+/*
+ * This structure contains generic data for  IOMMU protection domains
+ * independent of their use.
+ */
+struct protection_domain {
+	struct list_head list;  /* for list of all protection domains */
+	struct list_head dev_list; /* List of all devices in this domain */
+	spinlock_t lock;	/* mostly used to lock the page table*/
+	struct mutex api_lock;	/* protect page tables in the iommu-api path */
+	u16 id;			/* the domain id written to the device table */
+	int mode;		/* paging mode (0-6 levels) */
+	u64 *pt_root;		/* page table root pointer */
+	unsigned long flags;	/* flags to find out type of domain */
+	bool updated;		/* complete domain flush required */
+	unsigned dev_cnt;	/* devices assigned to this domain */
+	unsigned dev_iommu[MAX_IOMMUS]; /* per-IOMMU reference count */
+	void *priv;		/* private data */
+
+};
+
+/*
+ * This struct contains device specific data for the IOMMU
+ */
+struct iommu_dev_data {
+	struct list_head list;		  /* For domain->dev_list */
+	struct list_head dev_data_list;	  /* For global dev_data_list */
+	struct iommu_dev_data *alias_data;/* The alias dev_data */
+	struct protection_domain *domain; /* Domain the device is bound to */
+	atomic_t bind;			  /* Domain attach reverent count */
+	u16 devid;			  /* PCI Device ID */
+	struct {
+		bool enabled;
+		int qdep;
+	} ats;				  /* ATS state */
+};
+
+/*
+ * For dynamic growth the aperture size is split into ranges of 128MB of
+ * DMA address space each. This struct represents one such range.
+ */
+struct aperture_range {
+
+	/* address allocation bitmap */
+	unsigned long *bitmap;
+
+	/*
+	 * Array of PTE pages for the aperture. In this array we save all the
+	 * leaf pages of the domain page table used for the aperture. This way
+	 * we don't need to walk the page table to find a specific PTE. We can
+	 * just calculate its address in constant time.
+	 */
+	u64 *pte_pages[64];
+
+	unsigned long offset;
+};
+
+/*
+ * Data container for a dma_ops specific protection domain
+ */
+struct dma_ops_domain {
+	struct list_head list;
+
+	/* generic protection domain information */
+	struct protection_domain domain;
+
+	/* size of the aperture for the mappings */
+	unsigned long aperture_size;
+
+	/* address we start to search for free addresses */
+	unsigned long next_address;
+
+	/* address space relevant data */
+	struct aperture_range *aperture[APERTURE_MAX_RANGES];
+
+	/* This will be set to true when TLB needs to be flushed */
+	bool need_flush;
+
+	/*
+	 * if this is a preallocated domain, keep the device for which it was
+	 * preallocated in this variable
+	 */
+	u16 target_dev;
+};
+
+/*
+ * Structure where we save information about one hardware AMD IOMMU in the
+ * system.
+ */
+struct amd_iommu {
+	struct list_head list;
+
+	/* Index within the IOMMU array */
+	int index;
+
+	/* locks the accesses to the hardware */
+	spinlock_t lock;
+
+	/* Pointer to PCI device of this IOMMU */
+	struct pci_dev *dev;
+
+	/* physical address of MMIO space */
+	u64 mmio_phys;
+	/* virtual address of MMIO space */
+	u8 *mmio_base;
+
+	/* capabilities of that IOMMU read from ACPI */
+	u32 cap;
+
+	/* flags read from acpi table */
+	u8 acpi_flags;
+
+	/* Extended features */
+	u64 features;
+
+	/*
+	 * Capability pointer. There could be more than one IOMMU per PCI
+	 * device function if there are more than one AMD IOMMU capability
+	 * pointers.
+	 */
+	u16 cap_ptr;
+
+	/* pci domain of this IOMMU */
+	u16 pci_seg;
+
+	/* first device this IOMMU handles. read from PCI */
+	u16 first_device;
+	/* last device this IOMMU handles. read from PCI */
+	u16 last_device;
+
+	/* start of exclusion range of that IOMMU */
+	u64 exclusion_start;
+	/* length of exclusion range of that IOMMU */
+	u64 exclusion_length;
+
+	/* command buffer virtual address */
+	u8 *cmd_buf;
+	/* size of command buffer */
+	u32 cmd_buf_size;
+
+	/* size of event buffer */
+	u32 evt_buf_size;
+	/* event buffer virtual address */
+	u8 *evt_buf;
+	/* MSI number for event interrupt */
+	u16 evt_msi_num;
+
+	/* true if interrupts for this IOMMU are already enabled */
+	bool int_enabled;
+
+	/* if one, we need to send a completion wait command */
+	bool need_sync;
+
+	/* default dma_ops domain for that IOMMU */
+	struct dma_ops_domain *default_dom;
+
+	/*
+	 * We can't rely on the BIOS to restore all values on reinit, so we
+	 * need to stash them
+	 */
+
+	/* The iommu BAR */
+	u32 stored_addr_lo;
+	u32 stored_addr_hi;
+
+	/*
+	 * Each iommu has 6 l1s, each of which is documented as having 0x12
+	 * registers
+	 */
+	u32 stored_l1[6][0x12];
+
+	/* The l2 indirect registers */
+	u32 stored_l2[0x83];
+};
+
+/*
+ * List with all IOMMUs in the system. This list is not locked because it is
+ * only written and read at driver initialization or suspend time
+ */
+extern struct list_head amd_iommu_list;
+
+/*
+ * Array with pointers to each IOMMU struct
+ * The indices are referenced in the protection domains
+ */
+extern struct amd_iommu *amd_iommus[MAX_IOMMUS];
+
+/* Number of IOMMUs present in the system */
+extern int amd_iommus_present;
+
+/*
+ * Declarations for the global list of all protection domains
+ */
+extern spinlock_t amd_iommu_pd_lock;
+extern struct list_head amd_iommu_pd_list;
+
+/*
+ * Structure defining one entry in the device table
+ */
+struct dev_table_entry {
+	u32 data[8];
+};
+
+/*
+ * One entry for unity mappings parsed out of the ACPI table.
+ */
+struct unity_map_entry {
+	struct list_head list;
+
+	/* starting device id this entry is used for (including) */
+	u16 devid_start;
+	/* end device id this entry is used for (including) */
+	u16 devid_end;
+
+	/* start address to unity map (including) */
+	u64 address_start;
+	/* end address to unity map (including) */
+	u64 address_end;
+
+	/* required protection */
+	int prot;
+};
+
+/*
+ * List of all unity mappings. It is not locked because as runtime it is only
+ * read. It is created at ACPI table parsing time.
+ */
+extern struct list_head amd_iommu_unity_map;
+
+/*
+ * Data structures for device handling
+ */
+
+/*
+ * Device table used by hardware. Read and write accesses by software are
+ * locked with the amd_iommu_pd_table lock.
+ */
+extern struct dev_table_entry *amd_iommu_dev_table;
+
+/*
+ * Alias table to find requestor ids to device ids. Not locked because only
+ * read on runtime.
+ */
+extern u16 *amd_iommu_alias_table;
+
+/*
+ * Reverse lookup table to find the IOMMU which translates a specific device.
+ */
+extern struct amd_iommu **amd_iommu_rlookup_table;
+
+/* size of the dma_ops aperture as power of 2 */
+extern unsigned amd_iommu_aperture_order;
+
+/* largest PCI device id we expect translation requests for */
+extern u16 amd_iommu_last_bdf;
+
+/* allocation bitmap for domain ids */
+extern unsigned long *amd_iommu_pd_alloc_bitmap;
+
+/*
+ * If true, the addresses will be flushed on unmap time, not when
+ * they are reused
+ */
+extern bool amd_iommu_unmap_flush;
+
+/* takes bus and device/function and returns the device id
+ * FIXME: should that be in generic PCI code? */
+static inline u16 calc_devid(u8 bus, u8 devfn)
+{
+	return (((u16)bus) << 8) | devfn;
+}
+
+#ifdef CONFIG_AMD_IOMMU_STATS
+
+struct __iommu_counter {
+	char *name;
+	struct dentry *dent;
+	u64 value;
+};
+
+#define DECLARE_STATS_COUNTER(nm) \
+	static struct __iommu_counter nm = {	\
+		.name = #nm,			\
+	}
+
+#define INC_STATS_COUNTER(name)		name.value += 1
+#define ADD_STATS_COUNTER(name, x)	name.value += (x)
+#define SUB_STATS_COUNTER(name, x)	name.value -= (x)
+
+#else /* CONFIG_AMD_IOMMU_STATS */
+
+#define DECLARE_STATS_COUNTER(name)
+#define INC_STATS_COUNTER(name)
+#define ADD_STATS_COUNTER(name, x)
+#define SUB_STATS_COUNTER(name, x)
+
+#endif /* CONFIG_AMD_IOMMU_STATS */
+
+#endif /* _ASM_X86_AMD_IOMMU_TYPES_H */
diff --git a/drivers/pci/dmar.c b/drivers/iommu/dmar.c
index 3dc9befa5ae..3dc9befa5ae 100644
--- a/drivers/pci/dmar.c
+++ b/drivers/iommu/dmar.c
diff --git a/drivers/pci/intel-iommu.c b/drivers/iommu/intel-iommu.c
index f02c34d26d1..c621c98c99d 100644
--- a/drivers/pci/intel-iommu.c
+++ b/drivers/iommu/intel-iommu.c
@@ -42,7 +42,6 @@
 #include <linux/pci-ats.h>
 #include <asm/cacheflush.h>
 #include <asm/iommu.h>
-#include "pci.h"
 
 #define ROOT_SIZE		VTD_PAGE_SIZE
 #define CONTEXT_SIZE		VTD_PAGE_SIZE
diff --git a/drivers/pci/intr_remapping.c b/drivers/iommu/intr_remapping.c
index 3607faf28a4..1a89d4a2cad 100644
--- a/drivers/pci/intr_remapping.c
+++ b/drivers/iommu/intr_remapping.c
@@ -13,7 +13,6 @@
 #include "intr_remapping.h"
 #include <acpi/acpi.h>
 #include <asm/pci-direct.h>
-#include "pci.h"
 
 static struct ioapic_scope ir_ioapic[MAX_IO_APICS];
 static struct hpet_scope ir_hpet[MAX_HPET_TBS];
diff --git a/drivers/pci/intr_remapping.h b/drivers/iommu/intr_remapping.h
index 5662fecfee6..5662fecfee6 100644
--- a/drivers/pci/intr_remapping.h
+++ b/drivers/iommu/intr_remapping.h
diff --git a/drivers/base/iommu.c b/drivers/iommu/iommu.c
index 6e6b6a11b3c..6e6b6a11b3c 100644
--- a/drivers/base/iommu.c
+++ b/drivers/iommu/iommu.c
diff --git a/drivers/pci/iova.c b/drivers/iommu/iova.c
index c5c274ab5c5..c5c274ab5c5 100644
--- a/drivers/pci/iova.c
+++ b/drivers/iommu/iova.c
diff --git a/drivers/iommu/msm_iommu.c b/drivers/iommu/msm_iommu.c
new file mode 100644
index 00000000000..1a584e077c6
--- /dev/null
+++ b/drivers/iommu/msm_iommu.c
@@ -0,0 +1,731 @@
+/* Copyright (c) 2010-2011, Code Aurora Forum. 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.
+ *
+ * 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 Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#define pr_fmt(fmt)	KBUILD_MODNAME ": " fmt
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+#include <linux/errno.h>
+#include <linux/io.h>
+#include <linux/interrupt.h>
+#include <linux/list.h>
+#include <linux/spinlock.h>
+#include <linux/slab.h>
+#include <linux/iommu.h>
+#include <linux/clk.h>
+
+#include <asm/cacheflush.h>
+#include <asm/sizes.h>
+
+#include <mach/iommu_hw-8xxx.h>
+#include <mach/iommu.h>
+
+#define MRC(reg, processor, op1, crn, crm, op2)				\
+__asm__ __volatile__ (							\
+"   mrc   "   #processor "," #op1 ", %0,"  #crn "," #crm "," #op2 "\n"  \
+: "=r" (reg))
+
+#define RCP15_PRRR(reg)		MRC(reg, p15, 0, c10, c2, 0)
+#define RCP15_NMRR(reg)		MRC(reg, p15, 0, c10, c2, 1)
+
+static int msm_iommu_tex_class[4];
+
+DEFINE_SPINLOCK(msm_iommu_lock);
+
+struct msm_priv {
+	unsigned long *pgtable;
+	struct list_head list_attached;
+};
+
+static int __enable_clocks(struct msm_iommu_drvdata *drvdata)
+{
+	int ret;
+
+	ret = clk_enable(drvdata->pclk);
+	if (ret)
+		goto fail;
+
+	if (drvdata->clk) {
+		ret = clk_enable(drvdata->clk);
+		if (ret)
+			clk_disable(drvdata->pclk);
+	}
+fail:
+	return ret;
+}
+
+static void __disable_clocks(struct msm_iommu_drvdata *drvdata)
+{
+	if (drvdata->clk)
+		clk_disable(drvdata->clk);
+	clk_disable(drvdata->pclk);
+}
+
+static int __flush_iotlb(struct iommu_domain *domain)
+{
+	struct msm_priv *priv = domain->priv;
+	struct msm_iommu_drvdata *iommu_drvdata;
+	struct msm_iommu_ctx_drvdata *ctx_drvdata;
+	int ret = 0;
+#ifndef CONFIG_IOMMU_PGTABLES_L2
+	unsigned long *fl_table = priv->pgtable;
+	int i;
+
+	if (!list_empty(&priv->list_attached)) {
+		dmac_flush_range(fl_table, fl_table + SZ_16K);
+
+		for (i = 0; i < NUM_FL_PTE; i++)
+			if ((fl_table[i] & 0x03) == FL_TYPE_TABLE) {
+				void *sl_table = __va(fl_table[i] &
+								FL_BASE_MASK);
+				dmac_flush_range(sl_table, sl_table + SZ_4K);
+			}
+	}
+#endif
+
+	list_for_each_entry(ctx_drvdata, &priv->list_attached, attached_elm) {
+		if (!ctx_drvdata->pdev || !ctx_drvdata->pdev->dev.parent)
+			BUG();
+
+		iommu_drvdata = dev_get_drvdata(ctx_drvdata->pdev->dev.parent);
+		BUG_ON(!iommu_drvdata);
+
+		ret = __enable_clocks(iommu_drvdata);
+		if (ret)
+			goto fail;
+
+		SET_CTX_TLBIALL(iommu_drvdata->base, ctx_drvdata->num, 0);
+		__disable_clocks(iommu_drvdata);
+	}
+fail:
+	return ret;
+}
+
+static void __reset_context(void __iomem *base, int ctx)
+{
+	SET_BPRCOSH(base, ctx, 0);
+	SET_BPRCISH(base, ctx, 0);
+	SET_BPRCNSH(base, ctx, 0);
+	SET_BPSHCFG(base, ctx, 0);
+	SET_BPMTCFG(base, ctx, 0);
+	SET_ACTLR(base, ctx, 0);
+	SET_SCTLR(base, ctx, 0);
+	SET_FSRRESTORE(base, ctx, 0);
+	SET_TTBR0(base, ctx, 0);
+	SET_TTBR1(base, ctx, 0);
+	SET_TTBCR(base, ctx, 0);
+	SET_BFBCR(base, ctx, 0);
+	SET_PAR(base, ctx, 0);
+	SET_FAR(base, ctx, 0);
+	SET_CTX_TLBIALL(base, ctx, 0);
+	SET_TLBFLPTER(base, ctx, 0);
+	SET_TLBSLPTER(base, ctx, 0);
+	SET_TLBLKCR(base, ctx, 0);
+	SET_PRRR(base, ctx, 0);
+	SET_NMRR(base, ctx, 0);
+}
+
+static void __program_context(void __iomem *base, int ctx, phys_addr_t pgtable)
+{
+	unsigned int prrr, nmrr;
+	__reset_context(base, ctx);
+
+	/* Set up HTW mode */
+	/* TLB miss configuration: perform HTW on miss */
+	SET_TLBMCFG(base, ctx, 0x3);
+
+	/* V2P configuration: HTW for access */
+	SET_V2PCFG(base, ctx, 0x3);
+
+	SET_TTBCR(base, ctx, 0);
+	SET_TTBR0_PA(base, ctx, (pgtable >> 14));
+
+	/* Invalidate the TLB for this context */
+	SET_CTX_TLBIALL(base, ctx, 0);
+
+	/* Set interrupt number to "secure" interrupt */
+	SET_IRPTNDX(base, ctx, 0);
+
+	/* Enable context fault interrupt */
+	SET_CFEIE(base, ctx, 1);
+
+	/* Stall access on a context fault and let the handler deal with it */
+	SET_CFCFG(base, ctx, 1);
+
+	/* Redirect all cacheable requests to L2 slave port. */
+	SET_RCISH(base, ctx, 1);
+	SET_RCOSH(base, ctx, 1);
+	SET_RCNSH(base, ctx, 1);
+
+	/* Turn on TEX Remap */
+	SET_TRE(base, ctx, 1);
+
+	/* Set TEX remap attributes */
+	RCP15_PRRR(prrr);
+	RCP15_NMRR(nmrr);
+	SET_PRRR(base, ctx, prrr);
+	SET_NMRR(base, ctx, nmrr);
+
+	/* Turn on BFB prefetch */
+	SET_BFBDFE(base, ctx, 1);
+
+#ifdef CONFIG_IOMMU_PGTABLES_L2
+	/* Configure page tables as inner-cacheable and shareable to reduce
+	 * the TLB miss penalty.
+	 */
+	SET_TTBR0_SH(base, ctx, 1);
+	SET_TTBR1_SH(base, ctx, 1);
+
+	SET_TTBR0_NOS(base, ctx, 1);
+	SET_TTBR1_NOS(base, ctx, 1);
+
+	SET_TTBR0_IRGNH(base, ctx, 0); /* WB, WA */
+	SET_TTBR0_IRGNL(base, ctx, 1);
+
+	SET_TTBR1_IRGNH(base, ctx, 0); /* WB, WA */
+	SET_TTBR1_IRGNL(base, ctx, 1);
+
+	SET_TTBR0_ORGN(base, ctx, 1); /* WB, WA */
+	SET_TTBR1_ORGN(base, ctx, 1); /* WB, WA */
+#endif
+
+	/* Enable the MMU */
+	SET_M(base, ctx, 1);
+}
+
+static int msm_iommu_domain_init(struct iommu_domain *domain)
+{
+	struct msm_priv *priv = kzalloc(sizeof(*priv), GFP_KERNEL);
+
+	if (!priv)
+		goto fail_nomem;
+
+	INIT_LIST_HEAD(&priv->list_attached);
+	priv->pgtable = (unsigned long *)__get_free_pages(GFP_KERNEL,
+							  get_order(SZ_16K));
+
+	if (!priv->pgtable)
+		goto fail_nomem;
+
+	memset(priv->pgtable, 0, SZ_16K);
+	domain->priv = priv;
+	return 0;
+
+fail_nomem:
+	kfree(priv);
+	return -ENOMEM;
+}
+
+static void msm_iommu_domain_destroy(struct iommu_domain *domain)
+{
+	struct msm_priv *priv;
+	unsigned long flags;
+	unsigned long *fl_table;
+	int i;
+
+	spin_lock_irqsave(&msm_iommu_lock, flags);
+	priv = domain->priv;
+	domain->priv = NULL;
+
+	if (priv) {
+		fl_table = priv->pgtable;
+
+		for (i = 0; i < NUM_FL_PTE; i++)
+			if ((fl_table[i] & 0x03) == FL_TYPE_TABLE)
+				free_page((unsigned long) __va(((fl_table[i]) &
+								FL_BASE_MASK)));
+
+		free_pages((unsigned long)priv->pgtable, get_order(SZ_16K));
+		priv->pgtable = NULL;
+	}
+
+	kfree(priv);
+	spin_unlock_irqrestore(&msm_iommu_lock, flags);
+}
+
+static int msm_iommu_attach_dev(struct iommu_domain *domain, struct device *dev)
+{
+	struct msm_priv *priv;
+	struct msm_iommu_ctx_dev *ctx_dev;
+	struct msm_iommu_drvdata *iommu_drvdata;
+	struct msm_iommu_ctx_drvdata *ctx_drvdata;
+	struct msm_iommu_ctx_drvdata *tmp_drvdata;
+	int ret = 0;
+	unsigned long flags;
+
+	spin_lock_irqsave(&msm_iommu_lock, flags);
+
+	priv = domain->priv;
+
+	if (!priv || !dev) {
+		ret = -EINVAL;
+		goto fail;
+	}
+
+	iommu_drvdata = dev_get_drvdata(dev->parent);
+	ctx_drvdata = dev_get_drvdata(dev);
+	ctx_dev = dev->platform_data;
+
+	if (!iommu_drvdata || !ctx_drvdata || !ctx_dev) {
+		ret = -EINVAL;
+		goto fail;
+	}
+
+	if (!list_empty(&ctx_drvdata->attached_elm)) {
+		ret = -EBUSY;
+		goto fail;
+	}
+
+	list_for_each_entry(tmp_drvdata, &priv->list_attached, attached_elm)
+		if (tmp_drvdata == ctx_drvdata) {
+			ret = -EBUSY;
+			goto fail;
+		}
+
+	ret = __enable_clocks(iommu_drvdata);
+	if (ret)
+		goto fail;
+
+	__program_context(iommu_drvdata->base, ctx_dev->num,
+			  __pa(priv->pgtable));
+
+	__disable_clocks(iommu_drvdata);
+	list_add(&(ctx_drvdata->attached_elm), &priv->list_attached);
+	ret = __flush_iotlb(domain);
+
+fail:
+	spin_unlock_irqrestore(&msm_iommu_lock, flags);
+	return ret;
+}
+
+static void msm_iommu_detach_dev(struct iommu_domain *domain,
+				 struct device *dev)
+{
+	struct msm_priv *priv;
+	struct msm_iommu_ctx_dev *ctx_dev;
+	struct msm_iommu_drvdata *iommu_drvdata;
+	struct msm_iommu_ctx_drvdata *ctx_drvdata;
+	unsigned long flags;
+	int ret;
+
+	spin_lock_irqsave(&msm_iommu_lock, flags);
+	priv = domain->priv;
+
+	if (!priv || !dev)
+		goto fail;
+
+	iommu_drvdata = dev_get_drvdata(dev->parent);
+	ctx_drvdata = dev_get_drvdata(dev);
+	ctx_dev = dev->platform_data;
+
+	if (!iommu_drvdata || !ctx_drvdata || !ctx_dev)
+		goto fail;
+
+	ret = __flush_iotlb(domain);
+	if (ret)
+		goto fail;
+
+	ret = __enable_clocks(iommu_drvdata);
+	if (ret)
+		goto fail;
+
+	__reset_context(iommu_drvdata->base, ctx_dev->num);
+	__disable_clocks(iommu_drvdata);
+	list_del_init(&ctx_drvdata->attached_elm);
+
+fail:
+	spin_unlock_irqrestore(&msm_iommu_lock, flags);
+}
+
+static int msm_iommu_map(struct iommu_domain *domain, unsigned long va,
+			 phys_addr_t pa, int order, int prot)
+{
+	struct msm_priv *priv;
+	unsigned long flags;
+	unsigned long *fl_table;
+	unsigned long *fl_pte;
+	unsigned long fl_offset;
+	unsigned long *sl_table;
+	unsigned long *sl_pte;
+	unsigned long sl_offset;
+	unsigned int pgprot;
+	size_t len = 0x1000UL << order;
+	int ret = 0, tex, sh;
+
+	spin_lock_irqsave(&msm_iommu_lock, flags);
+
+	sh = (prot & MSM_IOMMU_ATTR_SH) ? 1 : 0;
+	tex = msm_iommu_tex_class[prot & MSM_IOMMU_CP_MASK];
+
+	if (tex < 0 || tex > NUM_TEX_CLASS - 1) {
+		ret = -EINVAL;
+		goto fail;
+	}
+
+	priv = domain->priv;
+	if (!priv) {
+		ret = -EINVAL;
+		goto fail;
+	}
+
+	fl_table = priv->pgtable;
+
+	if (len != SZ_16M && len != SZ_1M &&
+	    len != SZ_64K && len != SZ_4K) {
+		pr_debug("Bad size: %d\n", len);
+		ret = -EINVAL;
+		goto fail;
+	}
+
+	if (!fl_table) {
+		pr_debug("Null page table\n");
+		ret = -EINVAL;
+		goto fail;
+	}
+
+	if (len == SZ_16M || len == SZ_1M) {
+		pgprot = sh ? FL_SHARED : 0;
+		pgprot |= tex & 0x01 ? FL_BUFFERABLE : 0;
+		pgprot |= tex & 0x02 ? FL_CACHEABLE : 0;
+		pgprot |= tex & 0x04 ? FL_TEX0 : 0;
+	} else	{
+		pgprot = sh ? SL_SHARED : 0;
+		pgprot |= tex & 0x01 ? SL_BUFFERABLE : 0;
+		pgprot |= tex & 0x02 ? SL_CACHEABLE : 0;
+		pgprot |= tex & 0x04 ? SL_TEX0 : 0;
+	}
+
+	fl_offset = FL_OFFSET(va);	/* Upper 12 bits */
+	fl_pte = fl_table + fl_offset;	/* int pointers, 4 bytes */
+
+	if (len == SZ_16M) {
+		int i = 0;
+		for (i = 0; i < 16; i++)
+			*(fl_pte+i) = (pa & 0xFF000000) | FL_SUPERSECTION |
+				  FL_AP_READ | FL_AP_WRITE | FL_TYPE_SECT |
+				  FL_SHARED | FL_NG | pgprot;
+	}
+
+	if (len == SZ_1M)
+		*fl_pte = (pa & 0xFFF00000) | FL_AP_READ | FL_AP_WRITE | FL_NG |
+					    FL_TYPE_SECT | FL_SHARED | pgprot;
+
+	/* Need a 2nd level table */
+	if ((len == SZ_4K || len == SZ_64K) && (*fl_pte) == 0) {
+		unsigned long *sl;
+		sl = (unsigned long *) __get_free_pages(GFP_ATOMIC,
+							get_order(SZ_4K));
+
+		if (!sl) {
+			pr_debug("Could not allocate second level table\n");
+			ret = -ENOMEM;
+			goto fail;
+		}
+
+		memset(sl, 0, SZ_4K);
+		*fl_pte = ((((int)__pa(sl)) & FL_BASE_MASK) | FL_TYPE_TABLE);
+	}
+
+	sl_table = (unsigned long *) __va(((*fl_pte) & FL_BASE_MASK));
+	sl_offset = SL_OFFSET(va);
+	sl_pte = sl_table + sl_offset;
+
+
+	if (len == SZ_4K)
+		*sl_pte = (pa & SL_BASE_MASK_SMALL) | SL_AP0 | SL_AP1 | SL_NG |
+					  SL_SHARED | SL_TYPE_SMALL | pgprot;
+
+	if (len == SZ_64K) {
+		int i;
+
+		for (i = 0; i < 16; i++)
+			*(sl_pte+i) = (pa & SL_BASE_MASK_LARGE) | SL_AP0 |
+			    SL_NG | SL_AP1 | SL_SHARED | SL_TYPE_LARGE | pgprot;
+	}
+
+	ret = __flush_iotlb(domain);
+fail:
+	spin_unlock_irqrestore(&msm_iommu_lock, flags);
+	return ret;
+}
+
+static int msm_iommu_unmap(struct iommu_domain *domain, unsigned long va,
+			    int order)
+{
+	struct msm_priv *priv;
+	unsigned long flags;
+	unsigned long *fl_table;
+	unsigned long *fl_pte;
+	unsigned long fl_offset;
+	unsigned long *sl_table;
+	unsigned long *sl_pte;
+	unsigned long sl_offset;
+	size_t len = 0x1000UL << order;
+	int i, ret = 0;
+
+	spin_lock_irqsave(&msm_iommu_lock, flags);
+
+	priv = domain->priv;
+
+	if (!priv) {
+		ret = -ENODEV;
+		goto fail;
+	}
+
+	fl_table = priv->pgtable;
+
+	if (len != SZ_16M && len != SZ_1M &&
+	    len != SZ_64K && len != SZ_4K) {
+		pr_debug("Bad length: %d\n", len);
+		ret = -EINVAL;
+		goto fail;
+	}
+
+	if (!fl_table) {
+		pr_debug("Null page table\n");
+		ret = -EINVAL;
+		goto fail;
+	}
+
+	fl_offset = FL_OFFSET(va);	/* Upper 12 bits */
+	fl_pte = fl_table + fl_offset;	/* int pointers, 4 bytes */
+
+	if (*fl_pte == 0) {
+		pr_debug("First level PTE is 0\n");
+		ret = -ENODEV;
+		goto fail;
+	}
+
+	/* Unmap supersection */
+	if (len == SZ_16M)
+		for (i = 0; i < 16; i++)
+			*(fl_pte+i) = 0;
+
+	if (len == SZ_1M)
+		*fl_pte = 0;
+
+	sl_table = (unsigned long *) __va(((*fl_pte) & FL_BASE_MASK));
+	sl_offset = SL_OFFSET(va);
+	sl_pte = sl_table + sl_offset;
+
+	if (len == SZ_64K) {
+		for (i = 0; i < 16; i++)
+			*(sl_pte+i) = 0;
+	}
+
+	if (len == SZ_4K)
+		*sl_pte = 0;
+
+	if (len == SZ_4K || len == SZ_64K) {
+		int used = 0;
+
+		for (i = 0; i < NUM_SL_PTE; i++)
+			if (sl_table[i])
+				used = 1;
+		if (!used) {
+			free_page((unsigned long)sl_table);
+			*fl_pte = 0;
+		}
+	}
+
+	ret = __flush_iotlb(domain);
+fail:
+	spin_unlock_irqrestore(&msm_iommu_lock, flags);
+	return ret;
+}
+
+static phys_addr_t msm_iommu_iova_to_phys(struct iommu_domain *domain,
+					  unsigned long va)
+{
+	struct msm_priv *priv;
+	struct msm_iommu_drvdata *iommu_drvdata;
+	struct msm_iommu_ctx_drvdata *ctx_drvdata;
+	unsigned int par;
+	unsigned long flags;
+	void __iomem *base;
+	phys_addr_t ret = 0;
+	int ctx;
+
+	spin_lock_irqsave(&msm_iommu_lock, flags);
+
+	priv = domain->priv;
+	if (list_empty(&priv->list_attached))
+		goto fail;
+
+	ctx_drvdata = list_entry(priv->list_attached.next,
+				 struct msm_iommu_ctx_drvdata, attached_elm);
+	iommu_drvdata = dev_get_drvdata(ctx_drvdata->pdev->dev.parent);
+
+	base = iommu_drvdata->base;
+	ctx = ctx_drvdata->num;
+
+	ret = __enable_clocks(iommu_drvdata);
+	if (ret)
+		goto fail;
+
+	/* Invalidate context TLB */
+	SET_CTX_TLBIALL(base, ctx, 0);
+	SET_V2PPR(base, ctx, va & V2Pxx_VA);
+
+	par = GET_PAR(base, ctx);
+
+	/* We are dealing with a supersection */
+	if (GET_NOFAULT_SS(base, ctx))
+		ret = (par & 0xFF000000) | (va & 0x00FFFFFF);
+	else	/* Upper 20 bits from PAR, lower 12 from VA */
+		ret = (par & 0xFFFFF000) | (va & 0x00000FFF);
+
+	if (GET_FAULT(base, ctx))
+		ret = 0;
+
+	__disable_clocks(iommu_drvdata);
+fail:
+	spin_unlock_irqrestore(&msm_iommu_lock, flags);
+	return ret;
+}
+
+static int msm_iommu_domain_has_cap(struct iommu_domain *domain,
+				    unsigned long cap)
+{
+	return 0;
+}
+
+static void print_ctx_regs(void __iomem *base, int ctx)
+{
+	unsigned int fsr = GET_FSR(base, ctx);
+	pr_err("FAR    = %08x    PAR    = %08x\n",
+	       GET_FAR(base, ctx), GET_PAR(base, ctx));
+	pr_err("FSR    = %08x [%s%s%s%s%s%s%s%s%s%s]\n", fsr,
+			(fsr & 0x02) ? "TF " : "",
+			(fsr & 0x04) ? "AFF " : "",
+			(fsr & 0x08) ? "APF " : "",
+			(fsr & 0x10) ? "TLBMF " : "",
+			(fsr & 0x20) ? "HTWDEEF " : "",
+			(fsr & 0x40) ? "HTWSEEF " : "",
+			(fsr & 0x80) ? "MHF " : "",
+			(fsr & 0x10000) ? "SL " : "",
+			(fsr & 0x40000000) ? "SS " : "",
+			(fsr & 0x80000000) ? "MULTI " : "");
+
+	pr_err("FSYNR0 = %08x    FSYNR1 = %08x\n",
+	       GET_FSYNR0(base, ctx), GET_FSYNR1(base, ctx));
+	pr_err("TTBR0  = %08x    TTBR1  = %08x\n",
+	       GET_TTBR0(base, ctx), GET_TTBR1(base, ctx));
+	pr_err("SCTLR  = %08x    ACTLR  = %08x\n",
+	       GET_SCTLR(base, ctx), GET_ACTLR(base, ctx));
+	pr_err("PRRR   = %08x    NMRR   = %08x\n",
+	       GET_PRRR(base, ctx), GET_NMRR(base, ctx));
+}
+
+irqreturn_t msm_iommu_fault_handler(int irq, void *dev_id)
+{
+	struct msm_iommu_drvdata *drvdata = dev_id;
+	void __iomem *base;
+	unsigned int fsr;
+	int i, ret;
+
+	spin_lock(&msm_iommu_lock);
+
+	if (!drvdata) {
+		pr_err("Invalid device ID in context interrupt handler\n");
+		goto fail;
+	}
+
+	base = drvdata->base;
+
+	pr_err("Unexpected IOMMU page fault!\n");
+	pr_err("base = %08x\n", (unsigned int) base);
+
+	ret = __enable_clocks(drvdata);
+	if (ret)
+		goto fail;
+
+	for (i = 0; i < drvdata->ncb; i++) {
+		fsr = GET_FSR(base, i);
+		if (fsr) {
+			pr_err("Fault occurred in context %d.\n", i);
+			pr_err("Interesting registers:\n");
+			print_ctx_regs(base, i);
+			SET_FSR(base, i, 0x4000000F);
+		}
+	}
+	__disable_clocks(drvdata);
+fail:
+	spin_unlock(&msm_iommu_lock);
+	return 0;
+}
+
+static struct iommu_ops msm_iommu_ops = {
+	.domain_init = msm_iommu_domain_init,
+	.domain_destroy = msm_iommu_domain_destroy,
+	.attach_dev = msm_iommu_attach_dev,
+	.detach_dev = msm_iommu_detach_dev,
+	.map = msm_iommu_map,
+	.unmap = msm_iommu_unmap,
+	.iova_to_phys = msm_iommu_iova_to_phys,
+	.domain_has_cap = msm_iommu_domain_has_cap
+};
+
+static int __init get_tex_class(int icp, int ocp, int mt, int nos)
+{
+	int i = 0;
+	unsigned int prrr = 0;
+	unsigned int nmrr = 0;
+	int c_icp, c_ocp, c_mt, c_nos;
+
+	RCP15_PRRR(prrr);
+	RCP15_NMRR(nmrr);
+
+	for (i = 0; i < NUM_TEX_CLASS; i++) {
+		c_nos = PRRR_NOS(prrr, i);
+		c_mt = PRRR_MT(prrr, i);
+		c_icp = NMRR_ICP(nmrr, i);
+		c_ocp = NMRR_OCP(nmrr, i);
+
+		if (icp == c_icp && ocp == c_ocp && c_mt == mt && c_nos == nos)
+			return i;
+	}
+
+	return -ENODEV;
+}
+
+static void __init setup_iommu_tex_classes(void)
+{
+	msm_iommu_tex_class[MSM_IOMMU_ATTR_NONCACHED] =
+			get_tex_class(CP_NONCACHED, CP_NONCACHED, MT_NORMAL, 1);
+
+	msm_iommu_tex_class[MSM_IOMMU_ATTR_CACHED_WB_WA] =
+			get_tex_class(CP_WB_WA, CP_WB_WA, MT_NORMAL, 1);
+
+	msm_iommu_tex_class[MSM_IOMMU_ATTR_CACHED_WB_NWA] =
+			get_tex_class(CP_WB_NWA, CP_WB_NWA, MT_NORMAL, 1);
+
+	msm_iommu_tex_class[MSM_IOMMU_ATTR_CACHED_WT] =
+			get_tex_class(CP_WT, CP_WT, MT_NORMAL, 1);
+}
+
+static int __init msm_iommu_init(void)
+{
+	setup_iommu_tex_classes();
+	register_iommu(&msm_iommu_ops);
+	return 0;
+}
+
+subsys_initcall(msm_iommu_init);
+
+MODULE_LICENSE("GPL v2");
+MODULE_AUTHOR("Stepan Moskovchenko <stepanm@codeaurora.org>");
diff --git a/drivers/iommu/msm_iommu_dev.c b/drivers/iommu/msm_iommu_dev.c
new file mode 100644
index 00000000000..8e8fb079852
--- /dev/null
+++ b/drivers/iommu/msm_iommu_dev.c
@@ -0,0 +1,422 @@
+/* Copyright (c) 2010-2011, Code Aurora Forum. 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.
+ *
+ * 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 Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#define pr_fmt(fmt)	KBUILD_MODNAME ": " fmt
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/platform_device.h>
+#include <linux/io.h>
+#include <linux/clk.h>
+#include <linux/iommu.h>
+#include <linux/interrupt.h>
+#include <linux/err.h>
+#include <linux/slab.h>
+
+#include <mach/iommu_hw-8xxx.h>
+#include <mach/iommu.h>
+#include <mach/clk.h>
+
+struct iommu_ctx_iter_data {
+	/* input */
+	const char *name;
+
+	/* output */
+	struct device *dev;
+};
+
+static struct platform_device *msm_iommu_root_dev;
+
+static int each_iommu_ctx(struct device *dev, void *data)
+{
+	struct iommu_ctx_iter_data *res = data;
+	struct msm_iommu_ctx_dev *c = dev->platform_data;
+
+	if (!res || !c || !c->name || !res->name)
+		return -EINVAL;
+
+	if (!strcmp(res->name, c->name)) {
+		res->dev = dev;
+		return 1;
+	}
+	return 0;
+}
+
+static int each_iommu(struct device *dev, void *data)
+{
+	return device_for_each_child(dev, data, each_iommu_ctx);
+}
+
+struct device *msm_iommu_get_ctx(const char *ctx_name)
+{
+	struct iommu_ctx_iter_data r;
+	int found;
+
+	if (!msm_iommu_root_dev) {
+		pr_err("No root IOMMU device.\n");
+		goto fail;
+	}
+
+	r.name = ctx_name;
+	found = device_for_each_child(&msm_iommu_root_dev->dev, &r, each_iommu);
+
+	if (!found) {
+		pr_err("Could not find context <%s>\n", ctx_name);
+		goto fail;
+	}
+
+	return r.dev;
+fail:
+	return NULL;
+}
+EXPORT_SYMBOL(msm_iommu_get_ctx);
+
+static void msm_iommu_reset(void __iomem *base, int ncb)
+{
+	int ctx;
+
+	SET_RPUE(base, 0);
+	SET_RPUEIE(base, 0);
+	SET_ESRRESTORE(base, 0);
+	SET_TBE(base, 0);
+	SET_CR(base, 0);
+	SET_SPDMBE(base, 0);
+	SET_TESTBUSCR(base, 0);
+	SET_TLBRSW(base, 0);
+	SET_GLOBAL_TLBIALL(base, 0);
+	SET_RPU_ACR(base, 0);
+	SET_TLBLKCRWE(base, 1);
+
+	for (ctx = 0; ctx < ncb; ctx++) {
+		SET_BPRCOSH(base, ctx, 0);
+		SET_BPRCISH(base, ctx, 0);
+		SET_BPRCNSH(base, ctx, 0);
+		SET_BPSHCFG(base, ctx, 0);
+		SET_BPMTCFG(base, ctx, 0);
+		SET_ACTLR(base, ctx, 0);
+		SET_SCTLR(base, ctx, 0);
+		SET_FSRRESTORE(base, ctx, 0);
+		SET_TTBR0(base, ctx, 0);
+		SET_TTBR1(base, ctx, 0);
+		SET_TTBCR(base, ctx, 0);
+		SET_BFBCR(base, ctx, 0);
+		SET_PAR(base, ctx, 0);
+		SET_FAR(base, ctx, 0);
+		SET_CTX_TLBIALL(base, ctx, 0);
+		SET_TLBFLPTER(base, ctx, 0);
+		SET_TLBSLPTER(base, ctx, 0);
+		SET_TLBLKCR(base, ctx, 0);
+		SET_PRRR(base, ctx, 0);
+		SET_NMRR(base, ctx, 0);
+		SET_CONTEXTIDR(base, ctx, 0);
+	}
+}
+
+static int msm_iommu_probe(struct platform_device *pdev)
+{
+	struct resource *r, *r2;
+	struct clk *iommu_clk;
+	struct clk *iommu_pclk;
+	struct msm_iommu_drvdata *drvdata;
+	struct msm_iommu_dev *iommu_dev = pdev->dev.platform_data;
+	void __iomem *regs_base;
+	resource_size_t	len;
+	int ret, irq, par;
+
+	if (pdev->id == -1) {
+		msm_iommu_root_dev = pdev;
+		return 0;
+	}
+
+	drvdata = kzalloc(sizeof(*drvdata), GFP_KERNEL);
+
+	if (!drvdata) {
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	if (!iommu_dev) {
+		ret = -ENODEV;
+		goto fail;
+	}
+
+	iommu_pclk = clk_get(NULL, "smmu_pclk");
+	if (IS_ERR(iommu_pclk)) {
+		ret = -ENODEV;
+		goto fail;
+	}
+
+	ret = clk_enable(iommu_pclk);
+	if (ret)
+		goto fail_enable;
+
+	iommu_clk = clk_get(&pdev->dev, "iommu_clk");
+
+	if (!IS_ERR(iommu_clk))	{
+		if (clk_get_rate(iommu_clk) == 0)
+			clk_set_min_rate(iommu_clk, 1);
+
+		ret = clk_enable(iommu_clk);
+		if (ret) {
+			clk_put(iommu_clk);
+			goto fail_pclk;
+		}
+	} else
+		iommu_clk = NULL;
+
+	r = platform_get_resource_byname(pdev, IORESOURCE_MEM, "physbase");
+
+	if (!r) {
+		ret = -ENODEV;
+		goto fail_clk;
+	}
+
+	len = resource_size(r);
+
+	r2 = request_mem_region(r->start, len, r->name);
+	if (!r2) {
+		pr_err("Could not request memory region: start=%p, len=%d\n",
+							(void *) r->start, len);
+		ret = -EBUSY;
+		goto fail_clk;
+	}
+
+	regs_base = ioremap(r2->start, len);
+
+	if (!regs_base) {
+		pr_err("Could not ioremap: start=%p, len=%d\n",
+			 (void *) r2->start, len);
+		ret = -EBUSY;
+		goto fail_mem;
+	}
+
+	irq = platform_get_irq_byname(pdev, "secure_irq");
+	if (irq < 0) {
+		ret = -ENODEV;
+		goto fail_io;
+	}
+
+	msm_iommu_reset(regs_base, iommu_dev->ncb);
+
+	SET_M(regs_base, 0, 1);
+	SET_PAR(regs_base, 0, 0);
+	SET_V2PCFG(regs_base, 0, 1);
+	SET_V2PPR(regs_base, 0, 0);
+	par = GET_PAR(regs_base, 0);
+	SET_V2PCFG(regs_base, 0, 0);
+	SET_M(regs_base, 0, 0);
+
+	if (!par) {
+		pr_err("%s: Invalid PAR value detected\n", iommu_dev->name);
+		ret = -ENODEV;
+		goto fail_io;
+	}
+
+	ret = request_irq(irq, msm_iommu_fault_handler, 0,
+			"msm_iommu_secure_irpt_handler", drvdata);
+	if (ret) {
+		pr_err("Request IRQ %d failed with ret=%d\n", irq, ret);
+		goto fail_io;
+	}
+
+
+	drvdata->pclk = iommu_pclk;
+	drvdata->clk = iommu_clk;
+	drvdata->base = regs_base;
+	drvdata->irq = irq;
+	drvdata->ncb = iommu_dev->ncb;
+
+	pr_info("device %s mapped at %p, irq %d with %d ctx banks\n",
+		iommu_dev->name, regs_base, irq, iommu_dev->ncb);
+
+	platform_set_drvdata(pdev, drvdata);
+
+	if (iommu_clk)
+		clk_disable(iommu_clk);
+
+	clk_disable(iommu_pclk);
+
+	return 0;
+fail_io:
+	iounmap(regs_base);
+fail_mem:
+	release_mem_region(r->start, len);
+fail_clk:
+	if (iommu_clk) {
+		clk_disable(iommu_clk);
+		clk_put(iommu_clk);
+	}
+fail_pclk:
+	clk_disable(iommu_pclk);
+fail_enable:
+	clk_put(iommu_pclk);
+fail:
+	kfree(drvdata);
+	return ret;
+}
+
+static int msm_iommu_remove(struct platform_device *pdev)
+{
+	struct msm_iommu_drvdata *drv = NULL;
+
+	drv = platform_get_drvdata(pdev);
+	if (drv) {
+		if (drv->clk)
+			clk_put(drv->clk);
+		clk_put(drv->pclk);
+		memset(drv, 0, sizeof(*drv));
+		kfree(drv);
+		platform_set_drvdata(pdev, NULL);
+	}
+	return 0;
+}
+
+static int msm_iommu_ctx_probe(struct platform_device *pdev)
+{
+	struct msm_iommu_ctx_dev *c = pdev->dev.platform_data;
+	struct msm_iommu_drvdata *drvdata;
+	struct msm_iommu_ctx_drvdata *ctx_drvdata = NULL;
+	int i, ret;
+	if (!c || !pdev->dev.parent) {
+		ret = -EINVAL;
+		goto fail;
+	}
+
+	drvdata = dev_get_drvdata(pdev->dev.parent);
+
+	if (!drvdata) {
+		ret = -ENODEV;
+		goto fail;
+	}
+
+	ctx_drvdata = kzalloc(sizeof(*ctx_drvdata), GFP_KERNEL);
+	if (!ctx_drvdata) {
+		ret = -ENOMEM;
+		goto fail;
+	}
+	ctx_drvdata->num = c->num;
+	ctx_drvdata->pdev = pdev;
+
+	INIT_LIST_HEAD(&ctx_drvdata->attached_elm);
+	platform_set_drvdata(pdev, ctx_drvdata);
+
+	ret = clk_enable(drvdata->pclk);
+	if (ret)
+		goto fail;
+
+	if (drvdata->clk) {
+		ret = clk_enable(drvdata->clk);
+		if (ret) {
+			clk_disable(drvdata->pclk);
+			goto fail;
+		}
+	}
+
+	/* Program the M2V tables for this context */
+	for (i = 0; i < MAX_NUM_MIDS; i++) {
+		int mid = c->mids[i];
+		if (mid == -1)
+			break;
+
+		SET_M2VCBR_N(drvdata->base, mid, 0);
+		SET_CBACR_N(drvdata->base, c->num, 0);
+
+		/* Set VMID = 0 */
+		SET_VMID(drvdata->base, mid, 0);
+
+		/* Set the context number for that MID to this context */
+		SET_CBNDX(drvdata->base, mid, c->num);
+
+		/* Set MID associated with this context bank to 0*/
+		SET_CBVMID(drvdata->base, c->num, 0);
+
+		/* Set the ASID for TLB tagging for this context */
+		SET_CONTEXTIDR_ASID(drvdata->base, c->num, c->num);
+
+		/* Set security bit override to be Non-secure */
+		SET_NSCFG(drvdata->base, mid, 3);
+	}
+
+	if (drvdata->clk)
+		clk_disable(drvdata->clk);
+	clk_disable(drvdata->pclk);
+
+	dev_info(&pdev->dev, "context %s using bank %d\n", c->name, c->num);
+	return 0;
+fail:
+	kfree(ctx_drvdata);
+	return ret;
+}
+
+static int msm_iommu_ctx_remove(struct platform_device *pdev)
+{
+	struct msm_iommu_ctx_drvdata *drv = NULL;
+	drv = platform_get_drvdata(pdev);
+	if (drv) {
+		memset(drv, 0, sizeof(struct msm_iommu_ctx_drvdata));
+		kfree(drv);
+		platform_set_drvdata(pdev, NULL);
+	}
+	return 0;
+}
+
+static struct platform_driver msm_iommu_driver = {
+	.driver = {
+		.name	= "msm_iommu",
+	},
+	.probe		= msm_iommu_probe,
+	.remove		= msm_iommu_remove,
+};
+
+static struct platform_driver msm_iommu_ctx_driver = {
+	.driver = {
+		.name	= "msm_iommu_ctx",
+	},
+	.probe		= msm_iommu_ctx_probe,
+	.remove		= msm_iommu_ctx_remove,
+};
+
+static int __init msm_iommu_driver_init(void)
+{
+	int ret;
+	ret = platform_driver_register(&msm_iommu_driver);
+	if (ret != 0) {
+		pr_err("Failed to register IOMMU driver\n");
+		goto error;
+	}
+
+	ret = platform_driver_register(&msm_iommu_ctx_driver);
+	if (ret != 0) {
+		pr_err("Failed to register IOMMU context driver\n");
+		goto error;
+	}
+
+error:
+	return ret;
+}
+
+static void __exit msm_iommu_driver_exit(void)
+{
+	platform_driver_unregister(&msm_iommu_ctx_driver);
+	platform_driver_unregister(&msm_iommu_driver);
+}
+
+subsys_initcall(msm_iommu_driver_init);
+module_exit(msm_iommu_driver_exit);
+
+MODULE_LICENSE("GPL v2");
+MODULE_AUTHOR("Stepan Moskovchenko <stepanm@codeaurora.org>");
diff --git a/drivers/pci/Makefile b/drivers/pci/Makefile
index 631f7302760..6fadae3ad13 100644
--- a/drivers/pci/Makefile
+++ b/drivers/pci/Makefile
@@ -29,11 +29,6 @@ obj-$(CONFIG_PCI_MSI) += msi.o
 # Build the Hypertransport interrupt support
 obj-$(CONFIG_HT_IRQ) += htirq.o
 
-# Build Intel IOMMU support
-obj-$(CONFIG_DMAR) += dmar.o iova.o intel-iommu.o
-
-obj-$(CONFIG_INTR_REMAP) += dmar.o intr_remapping.o
-
 obj-$(CONFIG_PCI_IOV) += iov.o
 
 #
diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
index 3a39bf1f1e2..c8cee764b0d 100644
--- a/drivers/pci/pci.h
+++ b/drivers/pci/pci.h
@@ -186,8 +186,6 @@ pci_match_one_device(const struct pci_device_id *id, const struct pci_dev *dev)
 	return NULL;
 }
 
-struct pci_dev *pci_find_upstream_pcie_bridge(struct pci_dev *pdev);
-
 /* PCI slot sysfs helper code */
 #define to_pci_slot(s) container_of(s, struct pci_slot, kobj)