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

/*
 * kvm asynchronous fault support
 *
 * Copyright 2010 Red Hat, Inc.
 *
 * Author:
 *      Gleb Natapov <gleb@redhat.com>
 *
 * This file is free software; you can redistribute it and/or modify
 * it under the terms of version 2 of the GNU General Public License
 * 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 St, Fifth Floor, Boston, MA 02110-1301, USA.
 */

#include <linux/kvm_host.h>
#include <linux/slab.h>
#include <linux/module.h>
#include <linux/mmu_context.h>

#include "async_pf.h"
#include <trace/events/kvm.h>

static inline void kvm_async_page_present_sync(struct kvm_vcpu *vcpu,
					       struct kvm_async_pf *work)
{
#ifdef CONFIG_KVM_ASYNC_PF_SYNC
	kvm_arch_async_page_present(vcpu, work);
#endif
}
static inline void kvm_async_page_present_async(struct kvm_vcpu *vcpu,
						struct kvm_async_pf *work)
{
#ifndef CONFIG_KVM_ASYNC_PF_SYNC
	kvm_arch_async_page_present(vcpu, work);
#endif
}

static struct kmem_cache *async_pf_cache;

int kvm_async_pf_init(void)
{
	async_pf_cache = KMEM_CACHE(kvm_async_pf, 0);

	if (!async_pf_cache)
		return -ENOMEM;

	return 0;
}

void kvm_async_pf_deinit(void)
{
	kmem_cache_destroy(async_pf_cache);
	async_pf_cache = NULL;
}

void kvm_async_pf_vcpu_init(struct kvm_vcpu *vcpu)
{
	INIT_LIST_HEAD(&vcpu->async_pf.done);
	INIT_LIST_HEAD(&vcpu->async_pf.queue);
	spin_lock_init(&vcpu->async_pf.lock);
}

static void async_pf_execute(struct work_struct *work)
{
	struct kvm_async_pf *apf =
		container_of(work, struct kvm_async_pf, work);
	struct mm_struct *mm = apf->mm;
	struct kvm_vcpu *vcpu = apf->vcpu;
	unsigned long addr = apf->addr;
	gva_t gva = apf->gva;
	int locked = 1;

	might_sleep();

	/*
	 * This work is run asynchromously to the task which owns
	 * mm and might be done in another context, so we must
	 * access remotely.
	 */
	down_read(&mm->mmap_sem);
	get_user_pages_remote(NULL, mm, addr, 1, FOLL_WRITE, NULL, NULL,
			&locked);
	if (locked)
		up_read(&mm->mmap_sem);

	kvm_async_page_present_sync(vcpu, apf);

	spin_lock(&vcpu->async_pf.lock);
	list_add_tail(&apf->link, &vcpu->async_pf.done);
	apf->vcpu = NULL;
	spin_unlock(&vcpu->async_pf.lock);

	/*
	 * apf may be freed by kvm_check_async_pf_completion() after
	 * this point
	 */

	trace_kvm_async_pf_completed(addr, gva);

	/*
	 * This memory barrier pairs with prepare_to_wait's set_current_state()
	 */
	smp_mb();
	if (swait_active(&vcpu->wq))
		swake_up(&vcpu->wq);

	mmput(mm);
	kvm_put_kvm(vcpu->kvm);
}

void kvm_clear_async_pf_completion_queue(struct kvm_vcpu *vcpu)
{
	spin_lock(&vcpu->async_pf.lock);

	/* cancel outstanding work queue item */
	while (!list_empty(&vcpu->async_pf.queue)) {
		struct kvm_async_pf *work =
			list_first_entry(&vcpu->async_pf.queue,
					 typeof(*work), queue);
		list_del(&work->queue);

		/*
		 * We know it's present in vcpu->async_pf.done, do
		 * nothing here.
		 */
		if (!work->vcpu)
			continue;

		spin_unlock(&vcpu->async_pf.lock);
#ifdef CONFIG_KVM_ASYNC_PF_SYNC
		flush_work(&work->work);
#else
		if (cancel_work_sync(&work->work)) {
			mmput(work->mm);
			kvm_put_kvm(vcpu->kvm); /* == work->vcpu->kvm */
			kmem_cache_free(async_pf_cache, work);
		}
#endif
		spin_lock(&vcpu->async_pf.lock);
	}

	while (!list_empty(&vcpu->async_pf.done)) {
		struct kvm_async_pf *work =
			list_first_entry(&vcpu->async_pf.done,
					 typeof(*work), link);
		list_del(&work->link);
		kmem_cache_free(async_pf_cache, work);
	}
	spin_unlock(&vcpu->async_pf.lock);

	vcpu->async_pf.queued = 0;
}

void kvm_check_async_pf_completion(struct kvm_vcpu *vcpu)
{
	struct kvm_async_pf *work;

	while (!list_empty_careful(&vcpu->async_pf.done) &&
	      kvm_arch_can_inject_async_page_present(vcpu)) {
		spin_lock(&vcpu->async_pf.lock);
		work = list_first_entry(&vcpu->async_pf.done, typeof(*work),
					      link);
		list_del(&work->link);
		spin_unlock(&vcpu->async_pf.lock);

		kvm_arch_async_page_ready(vcpu, work);
		kvm_async_page_present_async(vcpu, work);

		list_del(&work->queue);
		vcpu->async_pf.queued--;
		kmem_cache_free(async_pf_cache, work);
	}
}

int kvm_setup_async_pf(struct kvm_vcpu *vcpu, gva_t gva, unsigned long hva,
		       struct kvm_arch_async_pf *arch)
{
	struct kvm_async_pf *work;

	if (vcpu->async_pf.queued >= ASYNC_PF_PER_VCPU)
		return 0;

	/* setup delayed work */

	/*
	 * do alloc nowait since if we are going to sleep anyway we
	 * may as well sleep faulting in page
	 */
	work = kmem_cache_zalloc(async_pf_cache, GFP_NOWAIT | __GFP_NOWARN);
	if (!work)
		return 0;

	work->wakeup_all = false;
	work->vcpu = vcpu;
	work->gva = gva;
	work->addr = hva;
	work->arch = *arch;
	work->mm = current->mm;
	atomic_inc(&work->mm->mm_users);
	kvm_get_kvm(work->vcpu->kvm);

	/* this can't really happen otherwise gfn_to_pfn_async
	   would succeed */
	if (unlikely(kvm_is_error_hva(work->addr)))
		goto retry_sync;

	INIT_WORK(&work->work, async_pf_execute);
	if (!schedule_work(&work->work))
		goto retry_sync;

	list_add_tail(&work->queue, &vcpu->async_pf.queue);
	vcpu->async_pf.queued++;
	kvm_arch_async_page_not_present(vcpu, work);
	return 1;
retry_sync:
	kvm_put_kvm(work->vcpu->kvm);
	mmput(work->mm);
	kmem_cache_free(async_pf_cache, work);
	return 0;
}

int kvm_async_pf_wakeup_all(struct kvm_vcpu *vcpu)
{
	struct kvm_async_pf *work;

	if (!list_empty_careful(&vcpu->async_pf.done))
		return 0;

	work = kmem_cache_zalloc(async_pf_cache, GFP_ATOMIC);
	if (!work)
		return -ENOMEM;

	work->wakeup_all = true;
	INIT_LIST_HEAD(&work->queue); /* for list_del to work */

	spin_lock(&vcpu->async_pf.lock);
	list_add_tail(&work->link, &vcpu->async_pf.done);
	spin_unlock(&vcpu->async_pf.lock);

	vcpu->async_pf.queued++;
	return 0;
}
Commit	Line	Data
af585b92 GN	1	/*
	2	* kvm asynchronous fault support
	3	*
	4	* Copyright 2010 Red Hat, Inc.
	5	*
	6	* Author:
	7	* Gleb Natapov <gleb@redhat.com>
	8	*
	9	* This file is free software; you can redistribute it and/or modify
	10	* it under the terms of version 2 of the GNU General Public License
	11	* as published by the Free Software Foundation.
	12	*
	13	* This program is distributed in the hope that it will be useful,
	14	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	15	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	16	* GNU General Public License for more details.
	17	*
	18	* You should have received a copy of the GNU General Public License
	19	* along with this program; if not, write to the Free Software Foundation,
	20	* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA.
	21	*/
	22
	23	#include <linux/kvm_host.h>
	24	#include <linux/slab.h>
	25	#include <linux/module.h>
	26	#include <linux/mmu_context.h>
	27
	28	#include "async_pf.h"
	29	#include <trace/events/kvm.h>
	30
e0ead41a DD	31	static inline void kvm_async_page_present_sync(struct kvm_vcpu *vcpu,
	32	struct kvm_async_pf *work)
	33	{
	34	#ifdef CONFIG_KVM_ASYNC_PF_SYNC
	35	kvm_arch_async_page_present(vcpu, work);
	36	#endif
	37	}
	38	static inline void kvm_async_page_present_async(struct kvm_vcpu *vcpu,
	39	struct kvm_async_pf *work)
	40	{
	41	#ifndef CONFIG_KVM_ASYNC_PF_SYNC
	42	kvm_arch_async_page_present(vcpu, work);
	43	#endif
	44	}
	45
af585b92 GN	46	static struct kmem_cache *async_pf_cache;
	47
	48	int kvm_async_pf_init(void)
	49	{
	50	async_pf_cache = KMEM_CACHE(kvm_async_pf, 0);
	51
	52	if (!async_pf_cache)
	53	return -ENOMEM;
	54
	55	return 0;
	56	}
	57
	58	void kvm_async_pf_deinit(void)
	59	{
4f52696a	60	kmem_cache_destroy(async_pf_cache);
af585b92 GN	61	async_pf_cache = NULL;
	62	}
	63
	64	void kvm_async_pf_vcpu_init(struct kvm_vcpu *vcpu)
	65	{
	66	INIT_LIST_HEAD(&vcpu->async_pf.done);
	67	INIT_LIST_HEAD(&vcpu->async_pf.queue);
	68	spin_lock_init(&vcpu->async_pf.lock);
	69	}
	70
	71	static void async_pf_execute(struct work_struct *work)
	72	{
af585b92 GN	73	struct kvm_async_pf *apf =
	74	container_of(work, struct kvm_async_pf, work);
	75	struct mm_struct *mm = apf->mm;
	76	struct kvm_vcpu *vcpu = apf->vcpu;
	77	unsigned long addr = apf->addr;
	78	gva_t gva = apf->gva;
8b7457ef	79	int locked = 1;
af585b92 GN	80
	81	might_sleep();
	82
1e987790 DH	83	/*
	84	* This work is run asynchromously to the task which owns
	85	* mm and might be done in another context, so we must
8b7457ef	86	* access remotely.
1e987790	87	*/
8b7457ef LS	88	down_read(&mm->mmap_sem);
	89	get_user_pages_remote(NULL, mm, addr, 1, FOLL_WRITE, NULL, NULL,
	90	&locked);
	91	if (locked)
	92	up_read(&mm->mmap_sem);
1e987790	93
e0ead41a	94	kvm_async_page_present_sync(vcpu, apf);
af585b92 GN	95
	96	spin_lock(&vcpu->async_pf.lock);
	97	list_add_tail(&apf->link, &vcpu->async_pf.done);
22583f0d	98	apf->vcpu = NULL;
af585b92 GN	99	spin_unlock(&vcpu->async_pf.lock);
	100
	101	/*
	102	* apf may be freed by kvm_check_async_pf_completion() after
	103	* this point
	104	*/
	105
f2e10669	106	trace_kvm_async_pf_completed(addr, gva);
af585b92	107
6003a420 KT	108	/*
	109	* This memory barrier pairs with prepare_to_wait's set_current_state()
	110	*/
	111	smp_mb();
8577370f MT	112	if (swait_active(&vcpu->wq))
8577370f MT	113	swake_up(&vcpu->wq);
af585b92	114
41c22f62	115	mmput(mm);
af585b92 GN	116	kvm_put_kvm(vcpu->kvm);
	117	}
	118
	119	void kvm_clear_async_pf_completion_queue(struct kvm_vcpu *vcpu)
	120	{
22583f0d PB	121	spin_lock(&vcpu->async_pf.lock);
22583f0d PB	122
af585b92 GN	123	/* cancel outstanding work queue item */
	124	while (!list_empty(&vcpu->async_pf.queue)) {
	125	struct kvm_async_pf *work =
433da860 GT	126	list_first_entry(&vcpu->async_pf.queue,
433da860 GT	127	typeof(*work), queue);
af585b92	128	list_del(&work->queue);
9f2ceda4	129
22583f0d PB	130	/*
	131	* We know it's present in vcpu->async_pf.done, do
	132	* nothing here.
	133	*/
	134	if (!work->vcpu)
	135	continue;
	136
	137	spin_unlock(&vcpu->async_pf.lock);
9f2ceda4 DD	138	#ifdef CONFIG_KVM_ASYNC_PF_SYNC
	139	flush_work(&work->work);
	140	#else
98fda169	141	if (cancel_work_sync(&work->work)) {
41c22f62	142	mmput(work->mm);
28b441e2	143	kvm_put_kvm(vcpu->kvm); /* == work->vcpu->kvm */
af585b92	144	kmem_cache_free(async_pf_cache, work);
28b441e2	145	}
9f2ceda4	146	#endif
22583f0d	147	spin_lock(&vcpu->async_pf.lock);
af585b92 GN	148	}
af585b92 GN	149
af585b92 GN	150	while (!list_empty(&vcpu->async_pf.done)) {
af585b92 GN	151	struct kvm_async_pf *work =
433da860 GT	152	list_first_entry(&vcpu->async_pf.done,
433da860 GT	153	typeof(*work), link);
af585b92	154	list_del(&work->link);
af585b92 GN	155	kmem_cache_free(async_pf_cache, work);
	156	}
	157	spin_unlock(&vcpu->async_pf.lock);
	158
	159	vcpu->async_pf.queued = 0;
	160	}
	161
	162	void kvm_check_async_pf_completion(struct kvm_vcpu *vcpu)
	163	{
	164	struct kvm_async_pf *work;
	165
15096ffc XG	166	while (!list_empty_careful(&vcpu->async_pf.done) &&
	167	kvm_arch_can_inject_async_page_present(vcpu)) {
	168	spin_lock(&vcpu->async_pf.lock);
	169	work = list_first_entry(&vcpu->async_pf.done, typeof(*work),
	170	link);
	171	list_del(&work->link);
	172	spin_unlock(&vcpu->async_pf.lock);
af585b92	173
f2e10669	174	kvm_arch_async_page_ready(vcpu, work);
1179ba53	175	kvm_async_page_present_async(vcpu, work);
af585b92	176
15096ffc XG	177	list_del(&work->queue);
15096ffc XG	178	vcpu->async_pf.queued--;
15096ffc XG	179	kmem_cache_free(async_pf_cache, work);
15096ffc XG	180	}
af585b92 GN	181	}
af585b92 GN	182
e0ead41a	183	int kvm_setup_async_pf(struct kvm_vcpu *vcpu, gva_t gva, unsigned long hva,
af585b92 GN	184	struct kvm_arch_async_pf *arch)
	185	{
	186	struct kvm_async_pf *work;
	187
	188	if (vcpu->async_pf.queued >= ASYNC_PF_PER_VCPU)
	189	return 0;
	190
	191	/* setup delayed work */
	192
	193	/*
	194	* do alloc nowait since if we are going to sleep anyway we
	195	* may as well sleep faulting in page
	196	*/
d7444794	197	work = kmem_cache_zalloc(async_pf_cache, GFP_NOWAIT \| __GFP_NOWARN);
af585b92 GN	198	if (!work)
	199	return 0;
	200
f2e10669	201	work->wakeup_all = false;
af585b92 GN	202	work->vcpu = vcpu;
af585b92 GN	203	work->gva = gva;
e0ead41a	204	work->addr = hva;
af585b92 GN	205	work->arch = *arch;
af585b92 GN	206	work->mm = current->mm;
41c22f62	207	atomic_inc(&work->mm->mm_users);
af585b92 GN	208	kvm_get_kvm(work->vcpu->kvm);
	209
	210	/* this can't really happen otherwise gfn_to_pfn_async
	211	would succeed */
	212	if (unlikely(kvm_is_error_hva(work->addr)))
	213	goto retry_sync;
	214
	215	INIT_WORK(&work->work, async_pf_execute);
	216	if (!schedule_work(&work->work))
	217	goto retry_sync;
	218
	219	list_add_tail(&work->queue, &vcpu->async_pf.queue);
	220	vcpu->async_pf.queued++;
	221	kvm_arch_async_page_not_present(vcpu, work);
	222	return 1;
	223	retry_sync:
	224	kvm_put_kvm(work->vcpu->kvm);
41c22f62	225	mmput(work->mm);
af585b92 GN	226	kmem_cache_free(async_pf_cache, work);
	227	return 0;
	228	}
344d9588 GN	229
	230	int kvm_async_pf_wakeup_all(struct kvm_vcpu *vcpu)
	231	{
	232	struct kvm_async_pf *work;
	233
64f638c7	234	if (!list_empty_careful(&vcpu->async_pf.done))
344d9588 GN	235	return 0;
	236
	237	work = kmem_cache_zalloc(async_pf_cache, GFP_ATOMIC);
	238	if (!work)
	239	return -ENOMEM;
	240
f2e10669	241	work->wakeup_all = true;
344d9588 GN	242	INIT_LIST_HEAD(&work->queue); /* for list_del to work */
344d9588 GN	243
64f638c7	244	spin_lock(&vcpu->async_pf.lock);
344d9588	245	list_add_tail(&work->link, &vcpu->async_pf.done);
64f638c7 XG	246	spin_unlock(&vcpu->async_pf.lock);
64f638c7 XG	247
344d9588 GN	248	vcpu->async_pf.queued++;
	249	return 0;
	250	}