2 * Copyright (c) 2006, Intel Corporation.
4 * This program is free software; you can redistribute it and/or modify it
5 * under the terms and conditions of the GNU General Public License,
6 * version 2, as published by the Free Software Foundation.
8 * This program is distributed in the hope it will be useful, but WITHOUT
9 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
10 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
13 * You should have received a copy of the GNU General Public License along with
14 * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
15 * Place - Suite 330, Boston, MA 02111-1307 USA.
17 * Copyright (C) 2006-2008 Intel Corporation
18 * Copyright IBM Corporation, 2008
19 * Copyright 2010 Red Hat, Inc. and/or its affiliates.
21 * Author: Allen M. Kay <allen.m.kay@intel.com>
22 * Author: Weidong Han <weidong.han@intel.com>
23 * Author: Ben-Ami Yassour <benami@il.ibm.com>
26 #include <linux/list.h>
27 #include <linux/kvm_host.h>
28 #include <linux/module.h>
29 #include <linux/pci.h>
30 #include <linux/stat.h>
31 #include <linux/dmar.h>
32 #include <linux/iommu.h>
33 #include <linux/intel-iommu.h>
35 static bool allow_unsafe_assigned_interrupts;
36 module_param_named(allow_unsafe_assigned_interrupts,
37 allow_unsafe_assigned_interrupts, bool, S_IRUGO | S_IWUSR);
38 MODULE_PARM_DESC(allow_unsafe_assigned_interrupts,
39 "Enable device assignment on platforms without interrupt remapping support.");
41 static int kvm_iommu_unmap_memslots(struct kvm *kvm);
42 static void kvm_iommu_put_pages(struct kvm *kvm,
43 gfn_t base_gfn, unsigned long npages);
45 static pfn_t kvm_pin_pages(struct kvm_memory_slot *slot, gfn_t gfn,
51 pfn = gfn_to_pfn_memslot(slot, gfn);
52 end_gfn = gfn + (size >> PAGE_SHIFT);
55 if (is_error_pfn(pfn))
59 gfn_to_pfn_memslot(slot, gfn++);
64 int kvm_iommu_map_pages(struct kvm *kvm, struct kvm_memory_slot *slot)
69 struct iommu_domain *domain = kvm->arch.iommu_domain;
72 /* check if iommu exists and in use */
77 end_gfn = gfn + slot->npages;
79 flags = IOMMU_READ | IOMMU_WRITE;
80 if (kvm->arch.iommu_flags & KVM_IOMMU_CACHE_COHERENCY)
84 while (gfn < end_gfn) {
85 unsigned long page_size;
87 /* Check if already mapped */
88 if (iommu_iova_to_phys(domain, gfn_to_gpa(gfn))) {
93 /* Get the page size we could use to map */
94 page_size = kvm_host_page_size(kvm, gfn);
96 /* Make sure the page_size does not exceed the memslot */
97 while ((gfn + (page_size >> PAGE_SHIFT)) > end_gfn)
100 /* Make sure gfn is aligned to the page size we want to map */
101 while ((gfn << PAGE_SHIFT) & (page_size - 1))
105 * Pin all pages we are about to map in memory. This is
106 * important because we unmap and unpin in 4kb steps later.
108 pfn = kvm_pin_pages(slot, gfn, page_size);
109 if (is_error_pfn(pfn)) {
110 kvm_release_pfn_clean(pfn);
115 /* Map into IO address space */
116 r = iommu_map(domain, gfn_to_gpa(gfn), pfn_to_hpa(pfn),
119 printk(KERN_ERR "kvm_iommu_map_address:"
120 "iommu failed to map pfn=%llx\n", pfn);
124 gfn += page_size >> PAGE_SHIFT;
132 kvm_iommu_put_pages(kvm, slot->base_gfn, gfn);
136 static int kvm_iommu_map_memslots(struct kvm *kvm)
139 struct kvm_memslots *slots;
140 struct kvm_memory_slot *memslot;
142 idx = srcu_read_lock(&kvm->srcu);
143 slots = kvm_memslots(kvm);
145 kvm_for_each_memslot(memslot, slots) {
146 r = kvm_iommu_map_pages(kvm, memslot);
150 srcu_read_unlock(&kvm->srcu, idx);
155 int kvm_assign_device(struct kvm *kvm,
156 struct kvm_assigned_dev_kernel *assigned_dev)
158 struct pci_dev *pdev = NULL;
159 struct iommu_domain *domain = kvm->arch.iommu_domain;
162 /* check if iommu exists and in use */
166 pdev = assigned_dev->dev;
170 r = iommu_attach_device(domain, &pdev->dev);
172 printk(KERN_ERR "assign device %x:%x:%x.%x failed",
173 pci_domain_nr(pdev->bus),
175 PCI_SLOT(pdev->devfn),
176 PCI_FUNC(pdev->devfn));
180 last_flags = kvm->arch.iommu_flags;
181 if (iommu_domain_has_cap(kvm->arch.iommu_domain,
182 IOMMU_CAP_CACHE_COHERENCY))
183 kvm->arch.iommu_flags |= KVM_IOMMU_CACHE_COHERENCY;
185 /* Check if need to update IOMMU page table for guest memory */
186 if ((last_flags ^ kvm->arch.iommu_flags) ==
187 KVM_IOMMU_CACHE_COHERENCY) {
188 kvm_iommu_unmap_memslots(kvm);
189 r = kvm_iommu_map_memslots(kvm);
194 pdev->dev_flags |= PCI_DEV_FLAGS_ASSIGNED;
196 printk(KERN_DEBUG "assign device %x:%x:%x.%x\n",
197 assigned_dev->host_segnr,
198 assigned_dev->host_busnr,
199 PCI_SLOT(assigned_dev->host_devfn),
200 PCI_FUNC(assigned_dev->host_devfn));
204 kvm_iommu_unmap_memslots(kvm);
208 int kvm_deassign_device(struct kvm *kvm,
209 struct kvm_assigned_dev_kernel *assigned_dev)
211 struct iommu_domain *domain = kvm->arch.iommu_domain;
212 struct pci_dev *pdev = NULL;
214 /* check if iommu exists and in use */
218 pdev = assigned_dev->dev;
222 iommu_detach_device(domain, &pdev->dev);
224 pdev->dev_flags &= ~PCI_DEV_FLAGS_ASSIGNED;
226 printk(KERN_DEBUG "deassign device %x:%x:%x.%x\n",
227 assigned_dev->host_segnr,
228 assigned_dev->host_busnr,
229 PCI_SLOT(assigned_dev->host_devfn),
230 PCI_FUNC(assigned_dev->host_devfn));
235 int kvm_iommu_map_guest(struct kvm *kvm)
239 if (!iommu_present(&pci_bus_type)) {
240 printk(KERN_ERR "%s: iommu not found\n", __func__);
244 mutex_lock(&kvm->slots_lock);
246 kvm->arch.iommu_domain = iommu_domain_alloc(&pci_bus_type);
247 if (!kvm->arch.iommu_domain) {
252 if (!allow_unsafe_assigned_interrupts &&
253 !iommu_domain_has_cap(kvm->arch.iommu_domain,
254 IOMMU_CAP_INTR_REMAP)) {
255 printk(KERN_WARNING "%s: No interrupt remapping support,"
256 " disallowing device assignment."
257 " Re-enble with \"allow_unsafe_assigned_interrupts=1\""
258 " module option.\n", __func__);
259 iommu_domain_free(kvm->arch.iommu_domain);
260 kvm->arch.iommu_domain = NULL;
265 r = kvm_iommu_map_memslots(kvm);
267 kvm_iommu_unmap_memslots(kvm);
270 mutex_unlock(&kvm->slots_lock);
274 static void kvm_unpin_pages(struct kvm *kvm, pfn_t pfn, unsigned long npages)
278 for (i = 0; i < npages; ++i)
279 kvm_release_pfn_clean(pfn + i);
282 static void kvm_iommu_put_pages(struct kvm *kvm,
283 gfn_t base_gfn, unsigned long npages)
285 struct iommu_domain *domain;
290 domain = kvm->arch.iommu_domain;
291 end_gfn = base_gfn + npages;
294 /* check if iommu exists and in use */
298 while (gfn < end_gfn) {
299 unsigned long unmap_pages;
302 /* Get physical address */
303 phys = iommu_iova_to_phys(domain, gfn_to_gpa(gfn));
310 pfn = phys >> PAGE_SHIFT;
312 /* Unmap address from IO address space */
313 size = iommu_unmap(domain, gfn_to_gpa(gfn), PAGE_SIZE);
314 unmap_pages = 1ULL << get_order(size);
316 /* Unpin all pages we just unmapped to not leak any memory */
317 kvm_unpin_pages(kvm, pfn, unmap_pages);
323 void kvm_iommu_unmap_pages(struct kvm *kvm, struct kvm_memory_slot *slot)
325 kvm_iommu_put_pages(kvm, slot->base_gfn, slot->npages);
328 static int kvm_iommu_unmap_memslots(struct kvm *kvm)
331 struct kvm_memslots *slots;
332 struct kvm_memory_slot *memslot;
334 idx = srcu_read_lock(&kvm->srcu);
335 slots = kvm_memslots(kvm);
337 kvm_for_each_memslot(memslot, slots)
338 kvm_iommu_unmap_pages(kvm, memslot);
340 srcu_read_unlock(&kvm->srcu, idx);
345 int kvm_iommu_unmap_guest(struct kvm *kvm)
347 struct iommu_domain *domain = kvm->arch.iommu_domain;
349 /* check if iommu exists and in use */
353 mutex_lock(&kvm->slots_lock);
354 kvm_iommu_unmap_memslots(kvm);
355 kvm->arch.iommu_domain = NULL;
356 mutex_unlock(&kvm->slots_lock);
358 iommu_domain_free(domain);