[net-next-2.6.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
 * 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
	19	* Author: Allen M. Kay <allen.m.kay@intel.com>
	20	* Author: Weidong Han <weidong.han@intel.com>
	21	* Author: Ben-Ami Yassour <benami@il.ibm.com>
	22	*/
	23
	24	#include <linux/list.h>
	25	#include <linux/kvm_host.h>
	26	#include <linux/pci.h>
	27	#include <linux/dmar.h>
19de40a8	28	#include <linux/iommu.h>
62c476c7 BAY	29	#include <linux/intel-iommu.h>
	30
	31	static int kvm_iommu_unmap_memslots(struct kvm *kvm);
	32	static void kvm_iommu_put_pages(struct kvm *kvm,
	33	gfn_t base_gfn, unsigned long npages);
	34
fcd95807 JR	35	static pfn_t kvm_pin_pages(struct kvm kvm, struct kvm_memory_slot slot,
	36	gfn_t gfn, unsigned long size)
	37	{
	38	gfn_t end_gfn;
	39	pfn_t pfn;
	40
	41	pfn = gfn_to_pfn_memslot(kvm, slot, gfn);
	42	end_gfn = gfn + (size >> PAGE_SHIFT);
	43	gfn += 1;
	44
	45	if (is_error_pfn(pfn))
	46	return pfn;
	47
	48	while (gfn < end_gfn)
	49	gfn_to_pfn_memslot(kvm, slot, gfn++);
	50
	51	return pfn;
	52	}
	53
3ad26d81	54	int kvm_iommu_map_pages(struct kvm kvm, struct kvm_memory_slot slot)
62c476c7	55	{
fcd95807	56	gfn_t gfn, end_gfn;
62c476c7	57	pfn_t pfn;
fcd95807	58	int r = 0;
19de40a8	59	struct iommu_domain *domain = kvm->arch.iommu_domain;
522c68c4	60	int flags;
62c476c7 BAY	61
	62	/* check if iommu exists and in use */
	63	if (!domain)
	64	return 0;
	65
fcd95807 JR	66	gfn = slot->base_gfn;
	67	end_gfn = gfn + slot->npages;
	68
522c68c4 SY	69	flags = IOMMU_READ \| IOMMU_WRITE;
	70	if (kvm->arch.iommu_flags & KVM_IOMMU_CACHE_COHERENCY)
	71	flags \|= IOMMU_CACHE;
	72
fcd95807 JR	73
	74	while (gfn < end_gfn) {
	75	unsigned long page_size;
	76
	77	/* Check if already mapped */
	78	if (iommu_iova_to_phys(domain, gfn_to_gpa(gfn))) {
	79	gfn += 1;
	80	continue;
	81	}
	82
	83	/* Get the page size we could use to map */
	84	page_size = kvm_host_page_size(kvm, gfn);
	85
	86	/* Make sure the page_size does not exceed the memslot */
	87	while ((gfn + (page_size >> PAGE_SHIFT)) > end_gfn)
	88	page_size >>= 1;
	89
	90	/* Make sure gfn is aligned to the page size we want to map */
	91	while ((gfn << PAGE_SHIFT) & (page_size - 1))
	92	page_size >>= 1;
	93
	94	/*
	95	* Pin all pages we are about to map in memory. This is
	96	* important because we unmap and unpin in 4kb steps later.
	97	*/
	98	pfn = kvm_pin_pages(kvm, slot, gfn, page_size);
	99	if (is_error_pfn(pfn)) {
	100	gfn += 1;
62c476c7	101	continue;
fcd95807	102	}
62c476c7	103
fcd95807 JR	104	/* Map into IO address space */
	105	r = iommu_map(domain, gfn_to_gpa(gfn), pfn_to_hpa(pfn),
	106	get_order(page_size), flags);
e5fcfc82	107	if (r) {
260782bc	108	printk(KERN_ERR "kvm_iommu_map_address:"
e5fcfc82	109	"iommu failed to map pfn=%lx\n", pfn);
62c476c7 BAY	110	goto unmap_pages;
62c476c7 BAY	111	}
fcd95807 JR	112
	113	gfn += page_size >> PAGE_SHIFT;
	114
	115
62c476c7	116	}
fcd95807	117
62c476c7 BAY	118	return 0;
	119
	120	unmap_pages:
fcd95807	121	kvm_iommu_put_pages(kvm, slot->base_gfn, gfn);
62c476c7 BAY	122	return r;
	123	}
	124
	125	static int kvm_iommu_map_memslots(struct kvm *kvm)
	126	{
7398ca79	127	int i, r = 0;
46a26bf5	128	struct kvm_memslots *slots;
62c476c7	129
90d83dc3	130	slots = kvm_memslots(kvm);
46a26bf5 MT	131
46a26bf5 MT	132	for (i = 0; i < slots->nmemslots; i++) {
3ad26d81	133	r = kvm_iommu_map_pages(kvm, &slots->memslots[i]);
62c476c7 BAY	134	if (r)
	135	break;
	136	}
682edb4c	137
62c476c7 BAY	138	return r;
	139	}
	140
260782bc WH	141	int kvm_assign_device(struct kvm *kvm,
260782bc WH	142	struct kvm_assigned_dev_kernel *assigned_dev)
62c476c7 BAY	143	{
62c476c7 BAY	144	struct pci_dev *pdev = NULL;
19de40a8	145	struct iommu_domain *domain = kvm->arch.iommu_domain;
522c68c4	146	int r, last_flags;
62c476c7	147
260782bc WH	148	/* check if iommu exists and in use */
	149	if (!domain)
	150	return 0;
	151
	152	pdev = assigned_dev->dev;
	153	if (pdev == NULL)
62c476c7	154	return -ENODEV;
260782bc	155
19de40a8	156	r = iommu_attach_device(domain, &pdev->dev);
260782bc	157	if (r) {
ab9f4ecb ZE	158	printk(KERN_ERR "assign device %x:%x:%x.%x failed",
ab9f4ecb ZE	159	pci_domain_nr(pdev->bus),
260782bc WH	160	pdev->bus->number,
	161	PCI_SLOT(pdev->devfn),
	162	PCI_FUNC(pdev->devfn));
	163	return r;
62c476c7 BAY	164	}
62c476c7 BAY	165
522c68c4 SY	166	last_flags = kvm->arch.iommu_flags;
	167	if (iommu_domain_has_cap(kvm->arch.iommu_domain,
	168	IOMMU_CAP_CACHE_COHERENCY))
	169	kvm->arch.iommu_flags \|= KVM_IOMMU_CACHE_COHERENCY;
	170
	171	/* Check if need to update IOMMU page table for guest memory */
	172	if ((last_flags ^ kvm->arch.iommu_flags) ==
	173	KVM_IOMMU_CACHE_COHERENCY) {
	174	kvm_iommu_unmap_memslots(kvm);
	175	r = kvm_iommu_map_memslots(kvm);
	176	if (r)
	177	goto out_unmap;
	178	}
	179
ab9f4ecb ZE	180	printk(KERN_DEBUG "assign device %x:%x:%x.%x\n",
ab9f4ecb ZE	181	assigned_dev->host_segnr,
260782bc WH	182	assigned_dev->host_busnr,
	183	PCI_SLOT(assigned_dev->host_devfn),
	184	PCI_FUNC(assigned_dev->host_devfn));
62c476c7	185
260782bc	186	return 0;
522c68c4 SY	187	out_unmap:
	188	kvm_iommu_unmap_memslots(kvm);
	189	return r;
260782bc	190	}
62c476c7	191
0a920356 WH	192	int kvm_deassign_device(struct kvm *kvm,
	193	struct kvm_assigned_dev_kernel *assigned_dev)
	194	{
19de40a8	195	struct iommu_domain *domain = kvm->arch.iommu_domain;
0a920356 WH	196	struct pci_dev *pdev = NULL;
	197
	198	/* check if iommu exists and in use */
	199	if (!domain)
	200	return 0;
	201
	202	pdev = assigned_dev->dev;
	203	if (pdev == NULL)
	204	return -ENODEV;
	205
19de40a8	206	iommu_detach_device(domain, &pdev->dev);
0a920356	207
ab9f4ecb ZE	208	printk(KERN_DEBUG "deassign device %x:%x:%x.%x\n",
ab9f4ecb ZE	209	assigned_dev->host_segnr,
0a920356 WH	210	assigned_dev->host_busnr,
	211	PCI_SLOT(assigned_dev->host_devfn),
	212	PCI_FUNC(assigned_dev->host_devfn));
	213
	214	return 0;
	215	}
	216
260782bc WH	217	int kvm_iommu_map_guest(struct kvm *kvm)
	218	{
	219	int r;
	220
19de40a8 JR	221	if (!iommu_found()) {
19de40a8 JR	222	printk(KERN_ERR "%s: iommu not found\n", __func__);
62c476c7 BAY	223	return -ENODEV;
	224	}
	225
19de40a8 JR	226	kvm->arch.iommu_domain = iommu_domain_alloc();
19de40a8 JR	227	if (!kvm->arch.iommu_domain)
260782bc	228	return -ENOMEM;
62c476c7 BAY	229
	230	r = kvm_iommu_map_memslots(kvm);
	231	if (r)
	232	goto out_unmap;
	233
62c476c7 BAY	234	return 0;
	235
	236	out_unmap:
	237	kvm_iommu_unmap_memslots(kvm);
	238	return r;
	239	}
	240
fcd95807 JR	241	static void kvm_unpin_pages(struct kvm *kvm, pfn_t pfn, unsigned long npages)
	242	{
	243	unsigned long i;
	244
	245	for (i = 0; i < npages; ++i)
	246	kvm_release_pfn_clean(pfn + i);
	247	}
	248
62c476c7	249	static void kvm_iommu_put_pages(struct kvm *kvm,
260782bc	250	gfn_t base_gfn, unsigned long npages)
62c476c7	251	{
fcd95807 JR	252	struct iommu_domain *domain;
fcd95807 JR	253	gfn_t end_gfn, gfn;
62c476c7	254	pfn_t pfn;
260782bc WH	255	u64 phys;
260782bc WH	256
fcd95807 JR	257	domain = kvm->arch.iommu_domain;
	258	end_gfn = base_gfn + npages;
	259	gfn = base_gfn;
	260
260782bc WH	261	/* check if iommu exists and in use */
	262	if (!domain)
	263	return;
62c476c7	264
fcd95807 JR	265	while (gfn < end_gfn) {
	266	unsigned long unmap_pages;
	267	int order;
	268
	269	/* Get physical address */
19de40a8	270	phys = iommu_iova_to_phys(domain, gfn_to_gpa(gfn));
fcd95807 JR	271	pfn = phys >> PAGE_SHIFT;
	272
	273	/* Unmap address from IO address space */
05b782ab	274	order = iommu_unmap(domain, gfn_to_gpa(gfn), 0);
fcd95807	275	unmap_pages = 1ULL << order;
260782bc	276
fcd95807 JR	277	/* Unpin all pages we just unmapped to not leak any memory */
	278	kvm_unpin_pages(kvm, pfn, unmap_pages);
	279
	280	gfn += unmap_pages;
	281	}
62c476c7 BAY	282	}
	283
	284	static int kvm_iommu_unmap_memslots(struct kvm *kvm)
	285	{
	286	int i;
46a26bf5 MT	287	struct kvm_memslots *slots;
46a26bf5 MT	288
90d83dc3	289	slots = kvm_memslots(kvm);
682edb4c	290
46a26bf5 MT	291	for (i = 0; i < slots->nmemslots; i++) {
	292	kvm_iommu_put_pages(kvm, slots->memslots[i].base_gfn,
	293	slots->memslots[i].npages);
62c476c7	294	}
62c476c7 BAY	295
	296	return 0;
	297	}
	298
	299	int kvm_iommu_unmap_guest(struct kvm *kvm)
	300	{
19de40a8	301	struct iommu_domain *domain = kvm->arch.iommu_domain;
62c476c7 BAY	302
	303	/* check if iommu exists and in use */
	304	if (!domain)
	305	return 0;
	306
62c476c7	307	kvm_iommu_unmap_memslots(kvm);
19de40a8	308	iommu_domain_free(domain);
62c476c7 BAY	309	return 0;
62c476c7 BAY	310	}