[deliverable/linux.git] / virt / kvm / iommu.c

/*
 * Copyright (c) 2006, Intel Corporation.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms and conditions of the GNU General Public License,
 * version 2, as published by the Free Software Foundation.
 *
 * This program is distributed in the hope it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
 * more details.
 *
 * You should have received a copy of the GNU General Public License along with
 * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
 * Place - Suite 330, Boston, MA 02111-1307 USA.
 *
 * Copyright (C) 2006-2008 Intel Corporation
 * Copyright IBM Corporation, 2008
 * Copyright 2010 Red Hat, Inc. and/or its affiliates.
 *
 * Author: Allen M. Kay <allen.m.kay@intel.com>
 * Author: Weidong Han <weidong.han@intel.com>
 * Author: Ben-Ami Yassour <benami@il.ibm.com>
 */

#include <linux/list.h>
#include <linux/kvm_host.h>
#include <linux/pci.h>
#include <linux/dmar.h>
#include <linux/iommu.h>
#include <linux/intel-iommu.h>

static int kvm_iommu_unmap_memslots(struct kvm *kvm);
static void kvm_iommu_put_pages(struct kvm *kvm,
				gfn_t base_gfn, unsigned long npages);

static pfn_t kvm_pin_pages(struct kvm *kvm, struct kvm_memory_slot *slot,
			   gfn_t gfn, unsigned long size)
{
	gfn_t end_gfn;
	pfn_t pfn;

	pfn     = gfn_to_pfn_memslot(kvm, slot, gfn);
	end_gfn = gfn + (size >> PAGE_SHIFT);
	gfn    += 1;

	if (is_error_pfn(pfn))
		return pfn;

	while (gfn < end_gfn)
		gfn_to_pfn_memslot(kvm, slot, gfn++);

	return pfn;
}

int kvm_iommu_map_pages(struct kvm *kvm, struct kvm_memory_slot *slot)
{
	gfn_t gfn, end_gfn;
	pfn_t pfn;
	int r = 0;
	struct iommu_domain *domain = kvm->arch.iommu_domain;
	int flags;

	/* check if iommu exists and in use */
	if (!domain)
		return 0;

	gfn     = slot->base_gfn;
	end_gfn = gfn + slot->npages;

	flags = IOMMU_READ | IOMMU_WRITE;
	if (kvm->arch.iommu_flags & KVM_IOMMU_CACHE_COHERENCY)
		flags |= IOMMU_CACHE;


	while (gfn < end_gfn) {
		unsigned long page_size;

		/* Check if already mapped */
		if (iommu_iova_to_phys(domain, gfn_to_gpa(gfn))) {
			gfn += 1;
			continue;
		}

		/* Get the page size we could use to map */
		page_size = kvm_host_page_size(kvm, gfn);

		/* Make sure the page_size does not exceed the memslot */
		while ((gfn + (page_size >> PAGE_SHIFT)) > end_gfn)
			page_size >>= 1;

		/* Make sure gfn is aligned to the page size we want to map */
		while ((gfn << PAGE_SHIFT) & (page_size - 1))
			page_size >>= 1;

		/*
		 * Pin all pages we are about to map in memory. This is
		 * important because we unmap and unpin in 4kb steps later.
		 */
		pfn = kvm_pin_pages(kvm, slot, gfn, page_size);
		if (is_error_pfn(pfn)) {
			gfn += 1;
			continue;
		}

		/* Map into IO address space */
		r = iommu_map(domain, gfn_to_gpa(gfn), pfn_to_hpa(pfn),
			      get_order(page_size), flags);
		if (r) {
			printk(KERN_ERR "kvm_iommu_map_address:"
			       "iommu failed to map pfn=%lx\n", pfn);
			goto unmap_pages;
		}

		gfn += page_size >> PAGE_SHIFT;


	}

	return 0;

unmap_pages:
	kvm_iommu_put_pages(kvm, slot->base_gfn, gfn);
	return r;
}

static int kvm_iommu_map_memslots(struct kvm *kvm)
{
	int i, r = 0;
	struct kvm_memslots *slots;

	slots = kvm_memslots(kvm);

	for (i = 0; i < slots->nmemslots; i++) {
		r = kvm_iommu_map_pages(kvm, &slots->memslots[i]);
		if (r)
			break;
	}

	return r;
}

int kvm_assign_device(struct kvm *kvm,
		      struct kvm_assigned_dev_kernel *assigned_dev)
{
	struct pci_dev *pdev = NULL;
	struct iommu_domain *domain = kvm->arch.iommu_domain;
	int r, last_flags;

	/* check if iommu exists and in use */
	if (!domain)
		return 0;

	pdev = assigned_dev->dev;
	if (pdev == NULL)
		return -ENODEV;

	r = iommu_attach_device(domain, &pdev->dev);
	if (r) {
		printk(KERN_ERR "assign device %x:%x:%x.%x failed",
			pci_domain_nr(pdev->bus),
			pdev->bus->number,
			PCI_SLOT(pdev->devfn),
			PCI_FUNC(pdev->devfn));
		return r;
	}

	last_flags = kvm->arch.iommu_flags;
	if (iommu_domain_has_cap(kvm->arch.iommu_domain,
				 IOMMU_CAP_CACHE_COHERENCY))
		kvm->arch.iommu_flags |= KVM_IOMMU_CACHE_COHERENCY;

	/* Check if need to update IOMMU page table for guest memory */
	if ((last_flags ^ kvm->arch.iommu_flags) ==
			KVM_IOMMU_CACHE_COHERENCY) {
		kvm_iommu_unmap_memslots(kvm);
		r = kvm_iommu_map_memslots(kvm);
		if (r)
			goto out_unmap;
	}

	printk(KERN_DEBUG "assign device %x:%x:%x.%x\n",
		assigned_dev->host_segnr,
		assigned_dev->host_busnr,
		PCI_SLOT(assigned_dev->host_devfn),
		PCI_FUNC(assigned_dev->host_devfn));

	return 0;
out_unmap:
	kvm_iommu_unmap_memslots(kvm);
	return r;
}

int kvm_deassign_device(struct kvm *kvm,
			struct kvm_assigned_dev_kernel *assigned_dev)
{
	struct iommu_domain *domain = kvm->arch.iommu_domain;
	struct pci_dev *pdev = NULL;

	/* check if iommu exists and in use */
	if (!domain)
		return 0;

	pdev = assigned_dev->dev;
	if (pdev == NULL)
		return -ENODEV;

	iommu_detach_device(domain, &pdev->dev);

	printk(KERN_DEBUG "deassign device %x:%x:%x.%x\n",
		assigned_dev->host_segnr,
		assigned_dev->host_busnr,
		PCI_SLOT(assigned_dev->host_devfn),
		PCI_FUNC(assigned_dev->host_devfn));

	return 0;
}

int kvm_iommu_map_guest(struct kvm *kvm)
{
	int r;

	if (!iommu_found()) {
		printk(KERN_ERR "%s: iommu not found\n", __func__);
		return -ENODEV;
	}

	kvm->arch.iommu_domain = iommu_domain_alloc();
	if (!kvm->arch.iommu_domain)
		return -ENOMEM;

	r = kvm_iommu_map_memslots(kvm);
	if (r)
		goto out_unmap;

	return 0;

out_unmap:
	kvm_iommu_unmap_memslots(kvm);
	return r;
}

static void kvm_unpin_pages(struct kvm *kvm, pfn_t pfn, unsigned long npages)
{
	unsigned long i;

	for (i = 0; i < npages; ++i)
		kvm_release_pfn_clean(pfn + i);
}

static void kvm_iommu_put_pages(struct kvm *kvm,
				gfn_t base_gfn, unsigned long npages)
{
	struct iommu_domain *domain;
	gfn_t end_gfn, gfn;
	pfn_t pfn;
	u64 phys;

	domain  = kvm->arch.iommu_domain;
	end_gfn = base_gfn + npages;
	gfn     = base_gfn;

	/* check if iommu exists and in use */
	if (!domain)
		return;

	while (gfn < end_gfn) {
		unsigned long unmap_pages;
		int order;

		/* Get physical address */
		phys = iommu_iova_to_phys(domain, gfn_to_gpa(gfn));
		pfn  = phys >> PAGE_SHIFT;

		/* Unmap address from IO address space */
		order       = iommu_unmap(domain, gfn_to_gpa(gfn), 0);
		unmap_pages = 1ULL << order;

		/* Unpin all pages we just unmapped to not leak any memory */
		kvm_unpin_pages(kvm, pfn, unmap_pages);

		gfn += unmap_pages;
	}
}

static int kvm_iommu_unmap_memslots(struct kvm *kvm)
{
	int i;
	struct kvm_memslots *slots;

	slots = kvm_memslots(kvm);

	for (i = 0; i < slots->nmemslots; i++) {
		kvm_iommu_put_pages(kvm, slots->memslots[i].base_gfn,
				    slots->memslots[i].npages);
	}

	return 0;
}

int kvm_iommu_unmap_guest(struct kvm *kvm)
{
	struct iommu_domain *domain = kvm->arch.iommu_domain;

	/* check if iommu exists and in use */
	if (!domain)
		return 0;

	kvm_iommu_unmap_memslots(kvm);
	iommu_domain_free(domain);
	return 0;
}
Commit	Line	Data
62c476c7 BAY	1	/*
	2	* Copyright (c) 2006, Intel Corporation.
	3	*
	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.
	7	*
	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
	11	* more details.
	12	*
	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.
	16	*
	17	* Copyright (C) 2006-2008 Intel Corporation
	18	* Copyright IBM Corporation, 2008
221d059d AK	19	* Copyright 2010 Red Hat, Inc. and/or its affiliates.
221d059d AK	20	*
62c476c7 BAY	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>
	24	*/
	25
	26	#include <linux/list.h>
	27	#include <linux/kvm_host.h>
	28	#include <linux/pci.h>
	29	#include <linux/dmar.h>
19de40a8	30	#include <linux/iommu.h>
62c476c7 BAY	31	#include <linux/intel-iommu.h>
	32
	33	static int kvm_iommu_unmap_memslots(struct kvm *kvm);
	34	static void kvm_iommu_put_pages(struct kvm *kvm,
	35	gfn_t base_gfn, unsigned long npages);
	36
fcd95807 JR	37	static pfn_t kvm_pin_pages(struct kvm kvm, struct kvm_memory_slot slot,
	38	gfn_t gfn, unsigned long size)
	39	{
	40	gfn_t end_gfn;
	41	pfn_t pfn;
	42
	43	pfn = gfn_to_pfn_memslot(kvm, slot, gfn);
	44	end_gfn = gfn + (size >> PAGE_SHIFT);
	45	gfn += 1;
	46
	47	if (is_error_pfn(pfn))
	48	return pfn;
	49
	50	while (gfn < end_gfn)
	51	gfn_to_pfn_memslot(kvm, slot, gfn++);
	52
	53	return pfn;
	54	}
	55
3ad26d81	56	int kvm_iommu_map_pages(struct kvm kvm, struct kvm_memory_slot slot)
62c476c7	57	{
fcd95807	58	gfn_t gfn, end_gfn;
62c476c7	59	pfn_t pfn;
fcd95807	60	int r = 0;
19de40a8	61	struct iommu_domain *domain = kvm->arch.iommu_domain;
522c68c4	62	int flags;
62c476c7 BAY	63
	64	/* check if iommu exists and in use */
	65	if (!domain)
	66	return 0;
	67
fcd95807 JR	68	gfn = slot->base_gfn;
	69	end_gfn = gfn + slot->npages;
	70
522c68c4 SY	71	flags = IOMMU_READ \| IOMMU_WRITE;
	72	if (kvm->arch.iommu_flags & KVM_IOMMU_CACHE_COHERENCY)
	73	flags \|= IOMMU_CACHE;
	74
fcd95807 JR	75
	76	while (gfn < end_gfn) {
	77	unsigned long page_size;
	78
	79	/* Check if already mapped */
	80	if (iommu_iova_to_phys(domain, gfn_to_gpa(gfn))) {
	81	gfn += 1;
	82	continue;
	83	}
	84
	85	/* Get the page size we could use to map */
	86	page_size = kvm_host_page_size(kvm, gfn);
	87
	88	/* Make sure the page_size does not exceed the memslot */
	89	while ((gfn + (page_size >> PAGE_SHIFT)) > end_gfn)
	90	page_size >>= 1;
	91
	92	/* Make sure gfn is aligned to the page size we want to map */
	93	while ((gfn << PAGE_SHIFT) & (page_size - 1))
	94	page_size >>= 1;
	95
	96	/*
	97	* Pin all pages we are about to map in memory. This is
	98	* important because we unmap and unpin in 4kb steps later.
	99	*/
	100	pfn = kvm_pin_pages(kvm, slot, gfn, page_size);
	101	if (is_error_pfn(pfn)) {
	102	gfn += 1;
62c476c7	103	continue;
fcd95807	104	}
62c476c7	105
fcd95807 JR	106	/* Map into IO address space */
	107	r = iommu_map(domain, gfn_to_gpa(gfn), pfn_to_hpa(pfn),
	108	get_order(page_size), flags);
e5fcfc82	109	if (r) {
260782bc	110	printk(KERN_ERR "kvm_iommu_map_address:"
e5fcfc82	111	"iommu failed to map pfn=%lx\n", pfn);
62c476c7 BAY	112	goto unmap_pages;
62c476c7 BAY	113	}
fcd95807 JR	114
	115	gfn += page_size >> PAGE_SHIFT;
	116
	117
62c476c7	118	}
fcd95807	119
62c476c7 BAY	120	return 0;
	121
	122	unmap_pages:
fcd95807	123	kvm_iommu_put_pages(kvm, slot->base_gfn, gfn);
62c476c7 BAY	124	return r;
	125	}
	126
	127	static int kvm_iommu_map_memslots(struct kvm *kvm)
	128	{
7398ca79	129	int i, r = 0;
46a26bf5	130	struct kvm_memslots *slots;
62c476c7	131
90d83dc3	132	slots = kvm_memslots(kvm);
46a26bf5 MT	133
46a26bf5 MT	134	for (i = 0; i < slots->nmemslots; i++) {
3ad26d81	135	r = kvm_iommu_map_pages(kvm, &slots->memslots[i]);
62c476c7 BAY	136	if (r)
	137	break;
	138	}
682edb4c	139
62c476c7 BAY	140	return r;
	141	}
	142
260782bc WH	143	int kvm_assign_device(struct kvm *kvm,
260782bc WH	144	struct kvm_assigned_dev_kernel *assigned_dev)
62c476c7 BAY	145	{
62c476c7 BAY	146	struct pci_dev *pdev = NULL;
19de40a8	147	struct iommu_domain *domain = kvm->arch.iommu_domain;
522c68c4	148	int r, last_flags;
62c476c7	149
260782bc WH	150	/* check if iommu exists and in use */
	151	if (!domain)
	152	return 0;
	153
	154	pdev = assigned_dev->dev;
	155	if (pdev == NULL)
62c476c7	156	return -ENODEV;
260782bc	157
19de40a8	158	r = iommu_attach_device(domain, &pdev->dev);
260782bc	159	if (r) {
ab9f4ecb ZE	160	printk(KERN_ERR "assign device %x:%x:%x.%x failed",
ab9f4ecb ZE	161	pci_domain_nr(pdev->bus),
260782bc WH	162	pdev->bus->number,
	163	PCI_SLOT(pdev->devfn),
	164	PCI_FUNC(pdev->devfn));
	165	return r;
62c476c7 BAY	166	}
62c476c7 BAY	167
522c68c4 SY	168	last_flags = kvm->arch.iommu_flags;
	169	if (iommu_domain_has_cap(kvm->arch.iommu_domain,
	170	IOMMU_CAP_CACHE_COHERENCY))
	171	kvm->arch.iommu_flags \|= KVM_IOMMU_CACHE_COHERENCY;
	172
	173	/* Check if need to update IOMMU page table for guest memory */
	174	if ((last_flags ^ kvm->arch.iommu_flags) ==
	175	KVM_IOMMU_CACHE_COHERENCY) {
	176	kvm_iommu_unmap_memslots(kvm);
	177	r = kvm_iommu_map_memslots(kvm);
	178	if (r)
	179	goto out_unmap;
	180	}
	181
ab9f4ecb ZE	182	printk(KERN_DEBUG "assign device %x:%x:%x.%x\n",
ab9f4ecb ZE	183	assigned_dev->host_segnr,
260782bc WH	184	assigned_dev->host_busnr,
	185	PCI_SLOT(assigned_dev->host_devfn),
	186	PCI_FUNC(assigned_dev->host_devfn));
62c476c7	187
260782bc	188	return 0;
522c68c4 SY	189	out_unmap:
	190	kvm_iommu_unmap_memslots(kvm);
	191	return r;
260782bc	192	}
62c476c7	193
0a920356 WH	194	int kvm_deassign_device(struct kvm *kvm,
	195	struct kvm_assigned_dev_kernel *assigned_dev)
	196	{
19de40a8	197	struct iommu_domain *domain = kvm->arch.iommu_domain;
0a920356 WH	198	struct pci_dev *pdev = NULL;
	199
	200	/* check if iommu exists and in use */
	201	if (!domain)
	202	return 0;
	203
	204	pdev = assigned_dev->dev;
	205	if (pdev == NULL)
	206	return -ENODEV;
	207
19de40a8	208	iommu_detach_device(domain, &pdev->dev);
0a920356	209
ab9f4ecb ZE	210	printk(KERN_DEBUG "deassign device %x:%x:%x.%x\n",
ab9f4ecb ZE	211	assigned_dev->host_segnr,
0a920356 WH	212	assigned_dev->host_busnr,
	213	PCI_SLOT(assigned_dev->host_devfn),
	214	PCI_FUNC(assigned_dev->host_devfn));
	215
	216	return 0;
	217	}
	218
260782bc WH	219	int kvm_iommu_map_guest(struct kvm *kvm)
	220	{
	221	int r;
	222
19de40a8 JR	223	if (!iommu_found()) {
19de40a8 JR	224	printk(KERN_ERR "%s: iommu not found\n", __func__);
62c476c7 BAY	225	return -ENODEV;
	226	}
	227
19de40a8 JR	228	kvm->arch.iommu_domain = iommu_domain_alloc();
19de40a8 JR	229	if (!kvm->arch.iommu_domain)
260782bc	230	return -ENOMEM;
62c476c7 BAY	231
	232	r = kvm_iommu_map_memslots(kvm);
	233	if (r)
	234	goto out_unmap;
	235
62c476c7 BAY	236	return 0;
	237
	238	out_unmap:
	239	kvm_iommu_unmap_memslots(kvm);
	240	return r;
	241	}
	242
fcd95807 JR	243	static void kvm_unpin_pages(struct kvm *kvm, pfn_t pfn, unsigned long npages)
	244	{
	245	unsigned long i;
	246
	247	for (i = 0; i < npages; ++i)
	248	kvm_release_pfn_clean(pfn + i);
	249	}
	250
62c476c7	251	static void kvm_iommu_put_pages(struct kvm *kvm,
260782bc	252	gfn_t base_gfn, unsigned long npages)
62c476c7	253	{
fcd95807 JR	254	struct iommu_domain *domain;
fcd95807 JR	255	gfn_t end_gfn, gfn;
62c476c7	256	pfn_t pfn;
260782bc WH	257	u64 phys;
260782bc WH	258
fcd95807 JR	259	domain = kvm->arch.iommu_domain;
	260	end_gfn = base_gfn + npages;
	261	gfn = base_gfn;
	262
260782bc WH	263	/* check if iommu exists and in use */
	264	if (!domain)
	265	return;
62c476c7	266
fcd95807 JR	267	while (gfn < end_gfn) {
	268	unsigned long unmap_pages;
	269	int order;
	270
	271	/* Get physical address */
19de40a8	272	phys = iommu_iova_to_phys(domain, gfn_to_gpa(gfn));
fcd95807 JR	273	pfn = phys >> PAGE_SHIFT;
	274
	275	/* Unmap address from IO address space */
05b782ab	276	order = iommu_unmap(domain, gfn_to_gpa(gfn), 0);
fcd95807	277	unmap_pages = 1ULL << order;
260782bc	278
fcd95807 JR	279	/* Unpin all pages we just unmapped to not leak any memory */
	280	kvm_unpin_pages(kvm, pfn, unmap_pages);
	281
	282	gfn += unmap_pages;
	283	}
62c476c7 BAY	284	}
	285
	286	static int kvm_iommu_unmap_memslots(struct kvm *kvm)
	287	{
	288	int i;
46a26bf5 MT	289	struct kvm_memslots *slots;
46a26bf5 MT	290
90d83dc3	291	slots = kvm_memslots(kvm);
682edb4c	292
46a26bf5 MT	293	for (i = 0; i < slots->nmemslots; i++) {
	294	kvm_iommu_put_pages(kvm, slots->memslots[i].base_gfn,
	295	slots->memslots[i].npages);
62c476c7	296	}
62c476c7 BAY	297
	298	return 0;
	299	}
	300
	301	int kvm_iommu_unmap_guest(struct kvm *kvm)
	302	{
19de40a8	303	struct iommu_domain *domain = kvm->arch.iommu_domain;
62c476c7 BAY	304
	305	/* check if iommu exists and in use */
	306	if (!domain)
	307	return 0;
	308
62c476c7	309	kvm_iommu_unmap_memslots(kvm);
19de40a8	310	iommu_domain_free(domain);
62c476c7 BAY	311	return 0;
62c476c7 BAY	312	}