[mirror_ubuntu-artful-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 <linux/sched/mm.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;
	mmget(work->mm);
	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>
6e84f315	27	#include <linux/sched/mm.h>
af585b92 GN	28
	29	#include "async_pf.h"
	30	#include <trace/events/kvm.h>
	31
e0ead41a DD	32	static inline void kvm_async_page_present_sync(struct kvm_vcpu *vcpu,
	33	struct kvm_async_pf *work)
	34	{
	35	#ifdef CONFIG_KVM_ASYNC_PF_SYNC
	36	kvm_arch_async_page_present(vcpu, work);
	37	#endif
	38	}
	39	static inline void kvm_async_page_present_async(struct kvm_vcpu *vcpu,
	40	struct kvm_async_pf *work)
	41	{
	42	#ifndef CONFIG_KVM_ASYNC_PF_SYNC
	43	kvm_arch_async_page_present(vcpu, work);
	44	#endif
	45	}
	46
af585b92 GN	47	static struct kmem_cache *async_pf_cache;
	48
	49	int kvm_async_pf_init(void)
	50	{
	51	async_pf_cache = KMEM_CACHE(kvm_async_pf, 0);
	52
	53	if (!async_pf_cache)
	54	return -ENOMEM;
	55
	56	return 0;
	57	}
	58
	59	void kvm_async_pf_deinit(void)
	60	{
4f52696a	61	kmem_cache_destroy(async_pf_cache);
af585b92 GN	62	async_pf_cache = NULL;
	63	}
	64
	65	void kvm_async_pf_vcpu_init(struct kvm_vcpu *vcpu)
	66	{
	67	INIT_LIST_HEAD(&vcpu->async_pf.done);
	68	INIT_LIST_HEAD(&vcpu->async_pf.queue);
	69	spin_lock_init(&vcpu->async_pf.lock);
	70	}
	71
	72	static void async_pf_execute(struct work_struct *work)
	73	{
af585b92 GN	74	struct kvm_async_pf *apf =
	75	container_of(work, struct kvm_async_pf, work);
	76	struct mm_struct *mm = apf->mm;
	77	struct kvm_vcpu *vcpu = apf->vcpu;
	78	unsigned long addr = apf->addr;
	79	gva_t gva = apf->gva;
8b7457ef	80	int locked = 1;
af585b92 GN	81
	82	might_sleep();
	83
1e987790 DH	84	/*
	85	* This work is run asynchromously to the task which owns
	86	* mm and might be done in another context, so we must
8b7457ef	87	* access remotely.
1e987790	88	*/
8b7457ef LS	89	down_read(&mm->mmap_sem);
	90	get_user_pages_remote(NULL, mm, addr, 1, FOLL_WRITE, NULL, NULL,
	91	&locked);
	92	if (locked)
	93	up_read(&mm->mmap_sem);
1e987790	94
e0ead41a	95	kvm_async_page_present_sync(vcpu, apf);
af585b92 GN	96
	97	spin_lock(&vcpu->async_pf.lock);
	98	list_add_tail(&apf->link, &vcpu->async_pf.done);
22583f0d	99	apf->vcpu = NULL;
af585b92 GN	100	spin_unlock(&vcpu->async_pf.lock);
	101
	102	/*
	103	* apf may be freed by kvm_check_async_pf_completion() after
	104	* this point
	105	*/
	106
f2e10669	107	trace_kvm_async_pf_completed(addr, gva);
af585b92	108
6003a420 KT	109	/*
	110	* This memory barrier pairs with prepare_to_wait's set_current_state()
	111	*/
	112	smp_mb();
8577370f MT	113	if (swait_active(&vcpu->wq))
8577370f MT	114	swake_up(&vcpu->wq);
af585b92	115
41c22f62	116	mmput(mm);
af585b92 GN	117	kvm_put_kvm(vcpu->kvm);
	118	}
	119
	120	void kvm_clear_async_pf_completion_queue(struct kvm_vcpu *vcpu)
	121	{
22583f0d PB	122	spin_lock(&vcpu->async_pf.lock);
22583f0d PB	123
af585b92 GN	124	/* cancel outstanding work queue item */
	125	while (!list_empty(&vcpu->async_pf.queue)) {
	126	struct kvm_async_pf *work =
433da860 GT	127	list_first_entry(&vcpu->async_pf.queue,
433da860 GT	128	typeof(*work), queue);
af585b92	129	list_del(&work->queue);
9f2ceda4	130
22583f0d PB	131	/*
	132	* We know it's present in vcpu->async_pf.done, do
	133	* nothing here.
	134	*/
	135	if (!work->vcpu)
	136	continue;
	137
	138	spin_unlock(&vcpu->async_pf.lock);
9f2ceda4 DD	139	#ifdef CONFIG_KVM_ASYNC_PF_SYNC
	140	flush_work(&work->work);
	141	#else
98fda169	142	if (cancel_work_sync(&work->work)) {
41c22f62	143	mmput(work->mm);
28b441e2	144	kvm_put_kvm(vcpu->kvm); /* == work->vcpu->kvm */
af585b92	145	kmem_cache_free(async_pf_cache, work);
28b441e2	146	}
9f2ceda4	147	#endif
22583f0d	148	spin_lock(&vcpu->async_pf.lock);
af585b92 GN	149	}
af585b92 GN	150
af585b92 GN	151	while (!list_empty(&vcpu->async_pf.done)) {
af585b92 GN	152	struct kvm_async_pf *work =
433da860 GT	153	list_first_entry(&vcpu->async_pf.done,
433da860 GT	154	typeof(*work), link);
af585b92	155	list_del(&work->link);
af585b92 GN	156	kmem_cache_free(async_pf_cache, work);
	157	}
	158	spin_unlock(&vcpu->async_pf.lock);
	159
	160	vcpu->async_pf.queued = 0;
	161	}
	162
	163	void kvm_check_async_pf_completion(struct kvm_vcpu *vcpu)
	164	{
	165	struct kvm_async_pf *work;
	166
15096ffc XG	167	while (!list_empty_careful(&vcpu->async_pf.done) &&
	168	kvm_arch_can_inject_async_page_present(vcpu)) {
	169	spin_lock(&vcpu->async_pf.lock);
	170	work = list_first_entry(&vcpu->async_pf.done, typeof(*work),
	171	link);
	172	list_del(&work->link);
	173	spin_unlock(&vcpu->async_pf.lock);
af585b92	174
f2e10669	175	kvm_arch_async_page_ready(vcpu, work);
1179ba53	176	kvm_async_page_present_async(vcpu, work);
af585b92	177
15096ffc XG	178	list_del(&work->queue);
15096ffc XG	179	vcpu->async_pf.queued--;
15096ffc XG	180	kmem_cache_free(async_pf_cache, work);
15096ffc XG	181	}
af585b92 GN	182	}
af585b92 GN	183
e0ead41a	184	int kvm_setup_async_pf(struct kvm_vcpu *vcpu, gva_t gva, unsigned long hva,
af585b92 GN	185	struct kvm_arch_async_pf *arch)
	186	{
	187	struct kvm_async_pf *work;
	188
	189	if (vcpu->async_pf.queued >= ASYNC_PF_PER_VCPU)
	190	return 0;
	191
	192	/* setup delayed work */
	193
	194	/*
	195	* do alloc nowait since if we are going to sleep anyway we
	196	* may as well sleep faulting in page
	197	*/
d7444794	198	work = kmem_cache_zalloc(async_pf_cache, GFP_NOWAIT \| __GFP_NOWARN);
af585b92 GN	199	if (!work)
	200	return 0;
	201
f2e10669	202	work->wakeup_all = false;
af585b92 GN	203	work->vcpu = vcpu;
af585b92 GN	204	work->gva = gva;
e0ead41a	205	work->addr = hva;
af585b92 GN	206	work->arch = *arch;
af585b92 GN	207	work->mm = current->mm;
3fce371b	208	mmget(work->mm);
af585b92 GN	209	kvm_get_kvm(work->vcpu->kvm);
	210
	211	/* this can't really happen otherwise gfn_to_pfn_async
	212	would succeed */
	213	if (unlikely(kvm_is_error_hva(work->addr)))
	214	goto retry_sync;
	215
	216	INIT_WORK(&work->work, async_pf_execute);
	217	if (!schedule_work(&work->work))
	218	goto retry_sync;
	219
	220	list_add_tail(&work->queue, &vcpu->async_pf.queue);
	221	vcpu->async_pf.queued++;
	222	kvm_arch_async_page_not_present(vcpu, work);
	223	return 1;
	224	retry_sync:
	225	kvm_put_kvm(work->vcpu->kvm);
41c22f62	226	mmput(work->mm);
af585b92 GN	227	kmem_cache_free(async_pf_cache, work);
	228	return 0;
	229	}
344d9588 GN	230
	231	int kvm_async_pf_wakeup_all(struct kvm_vcpu *vcpu)
	232	{
	233	struct kvm_async_pf *work;
	234
64f638c7	235	if (!list_empty_careful(&vcpu->async_pf.done))
344d9588 GN	236	return 0;
	237
	238	work = kmem_cache_zalloc(async_pf_cache, GFP_ATOMIC);
	239	if (!work)
	240	return -ENOMEM;
	241
f2e10669	242	work->wakeup_all = true;
344d9588 GN	243	INIT_LIST_HEAD(&work->queue); /* for list_del to work */
344d9588 GN	244
64f638c7	245	spin_lock(&vcpu->async_pf.lock);
344d9588	246	list_add_tail(&work->link, &vcpu->async_pf.done);
64f638c7 XG	247	spin_unlock(&vcpu->async_pf.lock);
64f638c7 XG	248
344d9588 GN	249	vcpu->async_pf.queued++;
	250	return 0;
	251	}