[mirror_ubuntu-zesty-kernel.git] / virt / kvm / vfio.c

/*
 * VFIO-KVM bridge pseudo device
 *
 * Copyright (C) 2013 Red Hat, Inc.  All rights reserved.
 *     Author: Alex Williamson <alex.williamson@redhat.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.
 */

#include <linux/errno.h>
#include <linux/file.h>
#include <linux/kvm_host.h>
#include <linux/list.h>
#include <linux/module.h>
#include <linux/mutex.h>
#include <linux/slab.h>
#include <linux/uaccess.h>
#include <linux/vfio.h>
#include "vfio.h"

struct kvm_vfio_group {
	struct list_head node;
	struct vfio_group *vfio_group;
};

struct kvm_vfio {
	struct list_head group_list;
	struct mutex lock;
	bool noncoherent;
};

static struct vfio_group *kvm_vfio_group_get_external_user(struct file *filep)
{
	struct vfio_group *vfio_group;
	struct vfio_group *(*fn)(struct file *);

	fn = symbol_get(vfio_group_get_external_user);
	if (!fn)
		return ERR_PTR(-EINVAL);

	vfio_group = fn(filep);

	symbol_put(vfio_group_get_external_user);

	return vfio_group;
}

static void kvm_vfio_group_put_external_user(struct vfio_group *vfio_group)
{
	void (*fn)(struct vfio_group *);

	fn = symbol_get(vfio_group_put_external_user);
	if (!fn)
		return;

	fn(vfio_group);

	symbol_put(vfio_group_put_external_user);
}

static void kvm_vfio_group_set_kvm(struct vfio_group *group, struct kvm *kvm)
{
	void (*fn)(struct vfio_group *, struct kvm *);

	fn = symbol_get(vfio_group_set_kvm);
	if (!fn)
		return;

	fn(group, kvm);

	symbol_put(vfio_group_set_kvm);
}

static bool kvm_vfio_group_is_coherent(struct vfio_group *vfio_group)
{
	long (*fn)(struct vfio_group *, unsigned long);
	long ret;

	fn = symbol_get(vfio_external_check_extension);
	if (!fn)
		return false;

	ret = fn(vfio_group, VFIO_DMA_CC_IOMMU);

	symbol_put(vfio_external_check_extension);

	return ret > 0;
}

/*
 * Groups can use the same or different IOMMU domains.  If the same then
 * adding a new group may change the coherency of groups we've previously
 * been told about.  We don't want to care about any of that so we retest
 * each group and bail as soon as we find one that's noncoherent.  This
 * means we only ever [un]register_noncoherent_dma once for the whole device.
 */
static void kvm_vfio_update_coherency(struct kvm_device *dev)
{
	struct kvm_vfio *kv = dev->private;
	bool noncoherent = false;
	struct kvm_vfio_group *kvg;

	mutex_lock(&kv->lock);

	list_for_each_entry(kvg, &kv->group_list, node) {
		if (!kvm_vfio_group_is_coherent(kvg->vfio_group)) {
			noncoherent = true;
			break;
		}
	}

	if (noncoherent != kv->noncoherent) {
		kv->noncoherent = noncoherent;

		if (kv->noncoherent)
			kvm_arch_register_noncoherent_dma(dev->kvm);
		else
			kvm_arch_unregister_noncoherent_dma(dev->kvm);
	}

	mutex_unlock(&kv->lock);
}

static int kvm_vfio_set_group(struct kvm_device *dev, long attr, u64 arg)
{
	struct kvm_vfio *kv = dev->private;
	struct vfio_group *vfio_group;
	struct kvm_vfio_group *kvg;
	int32_t __user *argp = (int32_t __user *)(unsigned long)arg;
	struct fd f;
	int32_t fd;
	int ret;

	switch (attr) {
	case KVM_DEV_VFIO_GROUP_ADD:
		if (get_user(fd, argp))
			return -EFAULT;

		f = fdget(fd);
		if (!f.file)
			return -EBADF;

		vfio_group = kvm_vfio_group_get_external_user(f.file);
		fdput(f);

		if (IS_ERR(vfio_group))
			return PTR_ERR(vfio_group);

		mutex_lock(&kv->lock);

		list_for_each_entry(kvg, &kv->group_list, node) {
			if (kvg->vfio_group == vfio_group) {
				mutex_unlock(&kv->lock);
				kvm_vfio_group_put_external_user(vfio_group);
				return -EEXIST;
			}
		}

		kvg = kzalloc(sizeof(*kvg), GFP_KERNEL);
		if (!kvg) {
			mutex_unlock(&kv->lock);
			kvm_vfio_group_put_external_user(vfio_group);
			return -ENOMEM;
		}

		list_add_tail(&kvg->node, &kv->group_list);
		kvg->vfio_group = vfio_group;

		kvm_arch_start_assignment(dev->kvm);

		mutex_unlock(&kv->lock);

		kvm_vfio_group_set_kvm(vfio_group, dev->kvm);

		kvm_vfio_update_coherency(dev);

		return 0;

	case KVM_DEV_VFIO_GROUP_DEL:
		if (get_user(fd, argp))
			return -EFAULT;

		f = fdget(fd);
		if (!f.file)
			return -EBADF;

		vfio_group = kvm_vfio_group_get_external_user(f.file);
		fdput(f);

		if (IS_ERR(vfio_group))
			return PTR_ERR(vfio_group);

		ret = -ENOENT;

		mutex_lock(&kv->lock);

		list_for_each_entry(kvg, &kv->group_list, node) {
			if (kvg->vfio_group != vfio_group)
				continue;

			list_del(&kvg->node);
			kvm_vfio_group_put_external_user(kvg->vfio_group);
			kfree(kvg);
			ret = 0;
			break;
		}

		kvm_arch_end_assignment(dev->kvm);

		mutex_unlock(&kv->lock);

		kvm_vfio_group_set_kvm(vfio_group, NULL);

		kvm_vfio_group_put_external_user(vfio_group);

		kvm_vfio_update_coherency(dev);

		return ret;
	}

	return -ENXIO;
}

static int kvm_vfio_set_attr(struct kvm_device *dev,
			     struct kvm_device_attr *attr)
{
	switch (attr->group) {
	case KVM_DEV_VFIO_GROUP:
		return kvm_vfio_set_group(dev, attr->attr, attr->addr);
	}

	return -ENXIO;
}

static int kvm_vfio_has_attr(struct kvm_device *dev,
			     struct kvm_device_attr *attr)
{
	switch (attr->group) {
	case KVM_DEV_VFIO_GROUP:
		switch (attr->attr) {
		case KVM_DEV_VFIO_GROUP_ADD:
		case KVM_DEV_VFIO_GROUP_DEL:
			return 0;
		}

		break;
	}

	return -ENXIO;
}

static void kvm_vfio_destroy(struct kvm_device *dev)
{
	struct kvm_vfio *kv = dev->private;
	struct kvm_vfio_group *kvg, *tmp;

	list_for_each_entry_safe(kvg, tmp, &kv->group_list, node) {
		kvm_vfio_group_set_kvm(kvg->vfio_group, NULL);
		kvm_vfio_group_put_external_user(kvg->vfio_group);
		list_del(&kvg->node);
		kfree(kvg);
		kvm_arch_end_assignment(dev->kvm);
	}

	kvm_vfio_update_coherency(dev);

	kfree(kv);
	kfree(dev); /* alloc by kvm_ioctl_create_device, free by .destroy */
}

static int kvm_vfio_create(struct kvm_device *dev, u32 type);

static struct kvm_device_ops kvm_vfio_ops = {
	.name = "kvm-vfio",
	.create = kvm_vfio_create,
	.destroy = kvm_vfio_destroy,
	.set_attr = kvm_vfio_set_attr,
	.has_attr = kvm_vfio_has_attr,
};

static int kvm_vfio_create(struct kvm_device *dev, u32 type)
{
	struct kvm_device *tmp;
	struct kvm_vfio *kv;

	/* Only one VFIO "device" per VM */
	list_for_each_entry(tmp, &dev->kvm->devices, vm_node)
		if (tmp->ops == &kvm_vfio_ops)
			return -EBUSY;

	kv = kzalloc(sizeof(*kv), GFP_KERNEL);
	if (!kv)
		return -ENOMEM;

	INIT_LIST_HEAD(&kv->group_list);
	mutex_init(&kv->lock);

	dev->private = kv;

	return 0;
}

int kvm_vfio_ops_init(void)
{
	return kvm_register_device_ops(&kvm_vfio_ops, KVM_DEV_TYPE_VFIO);
}

void kvm_vfio_ops_exit(void)
{
	kvm_unregister_device_ops(KVM_DEV_TYPE_VFIO);
}
Commit	Line	Data
ec53500f AW	1	/*
	2	* VFIO-KVM bridge pseudo device
	3	*
	4	* Copyright (C) 2013 Red Hat, Inc. All rights reserved.
	5	* Author: Alex Williamson <alex.williamson@redhat.com>
	6	*
	7	* This program is free software; you can redistribute it and/or modify
	8	* it under the terms of the GNU General Public License version 2 as
	9	* published by the Free Software Foundation.
	10	*/
	11
	12	#include <linux/errno.h>
	13	#include <linux/file.h>
	14	#include <linux/kvm_host.h>
	15	#include <linux/list.h>
	16	#include <linux/module.h>
	17	#include <linux/mutex.h>
	18	#include <linux/slab.h>
	19	#include <linux/uaccess.h>
	20	#include <linux/vfio.h>
3c3c29fd	21	#include "vfio.h"
ec53500f AW	22
	23	struct kvm_vfio_group {
	24	struct list_head node;
	25	struct vfio_group *vfio_group;
	26	};
	27
	28	struct kvm_vfio {
	29	struct list_head group_list;
	30	struct mutex lock;
e0f0bbc5	31	bool noncoherent;
ec53500f AW	32	};
	33
	34	static struct vfio_group kvm_vfio_group_get_external_user(struct file filep)
	35	{
	36	struct vfio_group *vfio_group;
	37	struct vfio_group (fn)(struct file *);
	38
	39	fn = symbol_get(vfio_group_get_external_user);
	40	if (!fn)
	41	return ERR_PTR(-EINVAL);
	42
	43	vfio_group = fn(filep);
	44
	45	symbol_put(vfio_group_get_external_user);
	46
	47	return vfio_group;
	48	}
	49
	50	static void kvm_vfio_group_put_external_user(struct vfio_group *vfio_group)
	51	{
	52	void (fn)(struct vfio_group );
	53
	54	fn = symbol_get(vfio_group_put_external_user);
	55	if (!fn)
	56	return;
	57
	58	fn(vfio_group);
	59
	60	symbol_put(vfio_group_put_external_user);
	61	}
	62
2fc1bec1 JS	63	static void kvm_vfio_group_set_kvm(struct vfio_group group, struct kvm kvm)
	64	{
	65	void (fn)(struct vfio_group , struct kvm *);
	66
	67	fn = symbol_get(vfio_group_set_kvm);
	68	if (!fn)
	69	return;
	70
	71	fn(group, kvm);
	72
	73	symbol_put(vfio_group_set_kvm);
	74	}
	75
9d830d47 AW	76	static bool kvm_vfio_group_is_coherent(struct vfio_group *vfio_group)
	77	{
	78	long (fn)(struct vfio_group , unsigned long);
	79	long ret;
	80
	81	fn = symbol_get(vfio_external_check_extension);
	82	if (!fn)
	83	return false;
	84
	85	ret = fn(vfio_group, VFIO_DMA_CC_IOMMU);
	86
	87	symbol_put(vfio_external_check_extension);
	88
	89	return ret > 0;
	90	}
	91
e0f0bbc5 AW	92	/*
	93	* Groups can use the same or different IOMMU domains. If the same then
	94	* adding a new group may change the coherency of groups we've previously
	95	* been told about. We don't want to care about any of that so we retest
	96	* each group and bail as soon as we find one that's noncoherent. This
	97	* means we only ever [un]register_noncoherent_dma once for the whole device.
	98	*/
	99	static void kvm_vfio_update_coherency(struct kvm_device *dev)
	100	{
	101	struct kvm_vfio *kv = dev->private;
	102	bool noncoherent = false;
	103	struct kvm_vfio_group *kvg;
	104
	105	mutex_lock(&kv->lock);
	106
	107	list_for_each_entry(kvg, &kv->group_list, node) {
9d830d47 AW	108	if (!kvm_vfio_group_is_coherent(kvg->vfio_group)) {
	109	noncoherent = true;
	110	break;
	111	}
e0f0bbc5 AW	112	}
	113
	114	if (noncoherent != kv->noncoherent) {
	115	kv->noncoherent = noncoherent;
	116
	117	if (kv->noncoherent)
	118	kvm_arch_register_noncoherent_dma(dev->kvm);
	119	else
	120	kvm_arch_unregister_noncoherent_dma(dev->kvm);
	121	}
	122
	123	mutex_unlock(&kv->lock);
	124	}
	125
ec53500f AW	126	static int kvm_vfio_set_group(struct kvm_device *dev, long attr, u64 arg)
	127	{
	128	struct kvm_vfio *kv = dev->private;
	129	struct vfio_group *vfio_group;
	130	struct kvm_vfio_group *kvg;
e81d1ad3	131	int32_t __user argp = (int32_t __user )(unsigned long)arg;
ec53500f AW	132	struct fd f;
	133	int32_t fd;
	134	int ret;
	135
	136	switch (attr) {
	137	case KVM_DEV_VFIO_GROUP_ADD:
e81d1ad3	138	if (get_user(fd, argp))
ec53500f AW	139	return -EFAULT;
	140
	141	f = fdget(fd);
	142	if (!f.file)
	143	return -EBADF;
	144
	145	vfio_group = kvm_vfio_group_get_external_user(f.file);
	146	fdput(f);
	147
	148	if (IS_ERR(vfio_group))
	149	return PTR_ERR(vfio_group);
	150
	151	mutex_lock(&kv->lock);
	152
	153	list_for_each_entry(kvg, &kv->group_list, node) {
	154	if (kvg->vfio_group == vfio_group) {
	155	mutex_unlock(&kv->lock);
	156	kvm_vfio_group_put_external_user(vfio_group);
	157	return -EEXIST;
	158	}
	159	}
	160
	161	kvg = kzalloc(sizeof(*kvg), GFP_KERNEL);
	162	if (!kvg) {
	163	mutex_unlock(&kv->lock);
	164	kvm_vfio_group_put_external_user(vfio_group);
	165	return -ENOMEM;
	166	}
	167
	168	list_add_tail(&kvg->node, &kv->group_list);
	169	kvg->vfio_group = vfio_group;
	170
5544eb9b PB	171	kvm_arch_start_assignment(dev->kvm);
5544eb9b PB	172
ec53500f AW	173	mutex_unlock(&kv->lock);
ec53500f AW	174
2fc1bec1 JS	175	kvm_vfio_group_set_kvm(vfio_group, dev->kvm);
2fc1bec1 JS	176
e0f0bbc5 AW	177	kvm_vfio_update_coherency(dev);
e0f0bbc5 AW	178
ec53500f AW	179	return 0;
	180
	181	case KVM_DEV_VFIO_GROUP_DEL:
e81d1ad3	182	if (get_user(fd, argp))
ec53500f AW	183	return -EFAULT;
	184
	185	f = fdget(fd);
	186	if (!f.file)
	187	return -EBADF;
	188
	189	vfio_group = kvm_vfio_group_get_external_user(f.file);
	190	fdput(f);
	191
	192	if (IS_ERR(vfio_group))
	193	return PTR_ERR(vfio_group);
	194
	195	ret = -ENOENT;
	196
	197	mutex_lock(&kv->lock);
	198
	199	list_for_each_entry(kvg, &kv->group_list, node) {
	200	if (kvg->vfio_group != vfio_group)
	201	continue;
	202
	203	list_del(&kvg->node);
	204	kvm_vfio_group_put_external_user(kvg->vfio_group);
	205	kfree(kvg);
	206	ret = 0;
	207	break;
	208	}
	209
5544eb9b PB	210	kvm_arch_end_assignment(dev->kvm);
5544eb9b PB	211
ec53500f AW	212	mutex_unlock(&kv->lock);
ec53500f AW	213
2fc1bec1 JS	214	kvm_vfio_group_set_kvm(vfio_group, NULL);
2fc1bec1 JS	215
ec53500f AW	216	kvm_vfio_group_put_external_user(vfio_group);
ec53500f AW	217
e0f0bbc5 AW	218	kvm_vfio_update_coherency(dev);
e0f0bbc5 AW	219
ec53500f AW	220	return ret;
	221	}
	222
	223	return -ENXIO;
	224	}
	225
	226	static int kvm_vfio_set_attr(struct kvm_device *dev,
	227	struct kvm_device_attr *attr)
	228	{
	229	switch (attr->group) {
	230	case KVM_DEV_VFIO_GROUP:
	231	return kvm_vfio_set_group(dev, attr->attr, attr->addr);
	232	}
	233
	234	return -ENXIO;
	235	}
	236
	237	static int kvm_vfio_has_attr(struct kvm_device *dev,
	238	struct kvm_device_attr *attr)
	239	{
	240	switch (attr->group) {
	241	case KVM_DEV_VFIO_GROUP:
	242	switch (attr->attr) {
	243	case KVM_DEV_VFIO_GROUP_ADD:
	244	case KVM_DEV_VFIO_GROUP_DEL:
	245	return 0;
	246	}
	247
	248	break;
	249	}
	250
	251	return -ENXIO;
	252	}
	253
	254	static void kvm_vfio_destroy(struct kvm_device *dev)
	255	{
	256	struct kvm_vfio *kv = dev->private;
	257	struct kvm_vfio_group kvg, tmp;
	258
	259	list_for_each_entry_safe(kvg, tmp, &kv->group_list, node) {
2fc1bec1	260	kvm_vfio_group_set_kvm(kvg->vfio_group, NULL);
ec53500f AW	261	kvm_vfio_group_put_external_user(kvg->vfio_group);
	262	list_del(&kvg->node);
	263	kfree(kvg);
5544eb9b	264	kvm_arch_end_assignment(dev->kvm);
ec53500f AW	265	}
ec53500f AW	266
e0f0bbc5 AW	267	kvm_vfio_update_coherency(dev);
e0f0bbc5 AW	268
ec53500f AW	269	kfree(kv);
	270	kfree(dev); /* alloc by kvm_ioctl_create_device, free by .destroy */
	271	}
	272
80ce1639 WD	273	static int kvm_vfio_create(struct kvm_device *dev, u32 type);
	274
	275	static struct kvm_device_ops kvm_vfio_ops = {
	276	.name = "kvm-vfio",
	277	.create = kvm_vfio_create,
	278	.destroy = kvm_vfio_destroy,
	279	.set_attr = kvm_vfio_set_attr,
	280	.has_attr = kvm_vfio_has_attr,
	281	};
	282
ec53500f AW	283	static int kvm_vfio_create(struct kvm_device *dev, u32 type)
	284	{
	285	struct kvm_device *tmp;
	286	struct kvm_vfio *kv;
	287
	288	/* Only one VFIO "device" per VM */
	289	list_for_each_entry(tmp, &dev->kvm->devices, vm_node)
	290	if (tmp->ops == &kvm_vfio_ops)
	291	return -EBUSY;
	292
	293	kv = kzalloc(sizeof(*kv), GFP_KERNEL);
	294	if (!kv)
	295	return -ENOMEM;
	296
	297	INIT_LIST_HEAD(&kv->group_list);
	298	mutex_init(&kv->lock);
	299
	300	dev->private = kv;
	301
	302	return 0;
	303	}
	304
3c3c29fd	305	int kvm_vfio_ops_init(void)
80ce1639 WD	306	{
	307	return kvm_register_device_ops(&kvm_vfio_ops, KVM_DEV_TYPE_VFIO);
	308	}
571ee1b6 WL	309
	310	void kvm_vfio_ops_exit(void)
	311	{
	312	kvm_unregister_device_ops(KVM_DEV_TYPE_VFIO);
	313	}