arm.c 37.4 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2012 - Virtual Open Systems and Columbia University
 * Author: Christoffer Dall <c.dall@virtualopensystems.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.
 *
 * 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, 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
 */

19
#include <linux/bug.h>
20
#include <linux/cpu_pm.h>
21 22 23
#include <linux/errno.h>
#include <linux/err.h>
#include <linux/kvm_host.h>
24
#include <linux/list.h>
25 26 27 28 29
#include <linux/module.h>
#include <linux/vmalloc.h>
#include <linux/fs.h>
#include <linux/mman.h>
#include <linux/sched.h>
30
#include <linux/kvm.h>
31 32
#include <linux/kvm_irqfd.h>
#include <linux/irqbypass.h>
33
#include <linux/sched/stat.h>
34
#include <trace/events/kvm.h>
35
#include <kvm/arm_pmu.h>
36
#include <kvm/arm_psci.h>
37 38 39 40

#define CREATE_TRACE_POINTS
#include "trace.h"

41
#include <linux/uaccess.h>
42 43
#include <asm/ptrace.h>
#include <asm/mman.h>
44
#include <asm/tlbflush.h>
45
#include <asm/cacheflush.h>
46
#include <asm/cpufeature.h>
47 48 49 50
#include <asm/virt.h>
#include <asm/kvm_arm.h>
#include <asm/kvm_asm.h>
#include <asm/kvm_mmu.h>
51
#include <asm/kvm_emulate.h>
52
#include <asm/kvm_coproc.h>
53
#include <asm/sections.h>
54 55 56 57 58

#ifdef REQUIRES_VIRT
__asm__(".arch_extension	virt");
#endif

59
DEFINE_PER_CPU(kvm_cpu_context_t, kvm_host_cpu_state);
60 61
static DEFINE_PER_CPU(unsigned long, kvm_arm_hyp_stack_page);

62 63 64
/* Per-CPU variable containing the currently running vcpu. */
static DEFINE_PER_CPU(struct kvm_vcpu *, kvm_arm_running_vcpu);

65 66
/* The VMID used in the VTTBR */
static atomic64_t kvm_vmid_gen = ATOMIC64_INIT(1);
67
static u32 kvm_next_vmid;
68
static DEFINE_SPINLOCK(kvm_vmid_lock);
69

70 71
static bool vgic_present;

72 73
static DEFINE_PER_CPU(unsigned char, kvm_arm_hardware_enabled);

74 75
static void kvm_arm_set_running_vcpu(struct kvm_vcpu *vcpu)
{
76
	__this_cpu_write(kvm_arm_running_vcpu, vcpu);
77 78
}

79 80
DEFINE_STATIC_KEY_FALSE(userspace_irqchip_in_use);

81 82 83 84 85 86
/**
 * kvm_arm_get_running_vcpu - get the vcpu running on the current CPU.
 * Must be called from non-preemptible context
 */
struct kvm_vcpu *kvm_arm_get_running_vcpu(void)
{
87
	return __this_cpu_read(kvm_arm_running_vcpu);
88 89 90 91 92
}

/**
 * kvm_arm_get_running_vcpus - get the per-CPU array of currently running vcpus.
 */
93
struct kvm_vcpu * __percpu *kvm_get_running_vcpus(void)
94 95 96 97
{
	return &kvm_arm_running_vcpu;
}

98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113
int kvm_arch_vcpu_should_kick(struct kvm_vcpu *vcpu)
{
	return kvm_vcpu_exiting_guest_mode(vcpu) == IN_GUEST_MODE;
}

int kvm_arch_hardware_setup(void)
{
	return 0;
}

void kvm_arch_check_processor_compat(void *rtn)
{
	*(int *)rtn = 0;
}


114 115 116 117
/**
 * kvm_arch_init_vm - initializes a VM data structure
 * @kvm:	pointer to the KVM struct
 */
118 119
int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
{
120
	int ret, cpu;
121

122
	ret = kvm_arm_setup_stage2(kvm, type);
123 124
	if (ret)
		return ret;
125

126 127 128 129 130 131 132
	kvm->arch.last_vcpu_ran = alloc_percpu(typeof(*kvm->arch.last_vcpu_ran));
	if (!kvm->arch.last_vcpu_ran)
		return -ENOMEM;

	for_each_possible_cpu(cpu)
		*per_cpu_ptr(kvm->arch.last_vcpu_ran, cpu) = -1;

133 134 135 136
	ret = kvm_alloc_stage2_pgd(kvm);
	if (ret)
		goto out_fail_alloc;

137
	ret = create_hyp_mappings(kvm, kvm + 1, PAGE_HYP);
138 139 140
	if (ret)
		goto out_free_stage2_pgd;

141
	kvm_vgic_early_init(kvm);
142

143
	/* Mark the initial VMID generation invalid */
144
	kvm->arch.vmid.vmid_gen = 0;
145

146
	/* The maximum number of VCPUs is limited by the host's GIC model */
147 148
	kvm->arch.max_vcpus = vgic_present ?
				kvm_vgic_get_max_vcpus() : KVM_MAX_VCPUS;
149

150 151 152 153
	return ret;
out_free_stage2_pgd:
	kvm_free_stage2_pgd(kvm);
out_fail_alloc:
154 155
	free_percpu(kvm->arch.last_vcpu_ran);
	kvm->arch.last_vcpu_ran = NULL;
156
	return ret;
157 158
}

159 160 161 162 163 164 165 166 167 168
bool kvm_arch_has_vcpu_debugfs(void)
{
	return false;
}

int kvm_arch_create_vcpu_debugfs(struct kvm_vcpu *vcpu)
{
	return 0;
}

169
vm_fault_t kvm_arch_vcpu_fault(struct kvm_vcpu *vcpu, struct vm_fault *vmf)
170 171 172 173 174
{
	return VM_FAULT_SIGBUS;
}


175 176 177 178
/**
 * kvm_arch_destroy_vm - destroy the VM data structure
 * @kvm:	pointer to the KVM struct
 */
179 180 181 182
void kvm_arch_destroy_vm(struct kvm *kvm)
{
	int i;

183 184
	kvm_vgic_destroy(kvm);

185 186 187
	free_percpu(kvm->arch.last_vcpu_ran);
	kvm->arch.last_vcpu_ran = NULL;

188 189 190 191 192 193
	for (i = 0; i < KVM_MAX_VCPUS; ++i) {
		if (kvm->vcpus[i]) {
			kvm_arch_vcpu_free(kvm->vcpus[i]);
			kvm->vcpus[i] = NULL;
		}
	}
194
	atomic_set(&kvm->online_vcpus, 0);
195 196
}

197
int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
198 199 200
{
	int r;
	switch (ext) {
201
	case KVM_CAP_IRQCHIP:
202 203
		r = vgic_present;
		break;
204
	case KVM_CAP_IOEVENTFD:
205
	case KVM_CAP_DEVICE_CTRL:
206 207 208 209
	case KVM_CAP_USER_MEMORY:
	case KVM_CAP_SYNC_MMU:
	case KVM_CAP_DESTROY_MEMORY_REGION_WORKS:
	case KVM_CAP_ONE_REG:
210
	case KVM_CAP_ARM_PSCI:
211
	case KVM_CAP_ARM_PSCI_0_2:
212
	case KVM_CAP_READONLY_MEM:
213
	case KVM_CAP_MP_STATE:
214
	case KVM_CAP_IMMEDIATE_EXIT:
215
	case KVM_CAP_VCPU_EVENTS:
216 217
		r = 1;
		break;
218 219
	case KVM_CAP_ARM_SET_DEVICE_ADDR:
		r = 1;
220
		break;
221 222 223 224 225 226
	case KVM_CAP_NR_VCPUS:
		r = num_online_cpus();
		break;
	case KVM_CAP_MAX_VCPUS:
		r = KVM_MAX_VCPUS;
		break;
227 228 229
	case KVM_CAP_NR_MEMSLOTS:
		r = KVM_USER_MEM_SLOTS;
		break;
V
Vladimir Murzin 已提交
230 231 232 233 234 235
	case KVM_CAP_MSI_DEVID:
		if (!kvm)
			r = -EINVAL;
		else
			r = kvm->arch.vgic.msis_require_devid;
		break;
236 237 238 239 240 241 242
	case KVM_CAP_ARM_USER_IRQ:
		/*
		 * 1: EL1_VTIMER, EL1_PTIMER, and PMU.
		 * (bump this number if adding more devices)
		 */
		r = 1;
		break;
243
	default:
244
		r = kvm_arch_vm_ioctl_check_extension(kvm, ext);
245 246 247 248 249 250 251 252 253 254 255
		break;
	}
	return r;
}

long kvm_arch_dev_ioctl(struct file *filp,
			unsigned int ioctl, unsigned long arg)
{
	return -EINVAL;
}

256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
struct kvm *kvm_arch_alloc_vm(void)
{
	if (!has_vhe())
		return kzalloc(sizeof(struct kvm), GFP_KERNEL);

	return vzalloc(sizeof(struct kvm));
}

void kvm_arch_free_vm(struct kvm *kvm)
{
	if (!has_vhe())
		kfree(kvm);
	else
		vfree(kvm);
}
271 272 273 274 275 276

struct kvm_vcpu *kvm_arch_vcpu_create(struct kvm *kvm, unsigned int id)
{
	int err;
	struct kvm_vcpu *vcpu;

277 278 279 280 281
	if (irqchip_in_kernel(kvm) && vgic_initialized(kvm)) {
		err = -EBUSY;
		goto out;
	}

282 283 284 285 286
	if (id >= kvm->arch.max_vcpus) {
		err = -EINVAL;
		goto out;
	}

287 288 289 290 291 292 293 294 295 296
	vcpu = kmem_cache_zalloc(kvm_vcpu_cache, GFP_KERNEL);
	if (!vcpu) {
		err = -ENOMEM;
		goto out;
	}

	err = kvm_vcpu_init(vcpu, kvm, id);
	if (err)
		goto free_vcpu;

297
	err = create_hyp_mappings(vcpu, vcpu + 1, PAGE_HYP);
298 299 300
	if (err)
		goto vcpu_uninit;

301
	return vcpu;
302 303
vcpu_uninit:
	kvm_vcpu_uninit(vcpu);
304 305 306 307 308 309
free_vcpu:
	kmem_cache_free(kvm_vcpu_cache, vcpu);
out:
	return ERR_PTR(err);
}

310
void kvm_arch_vcpu_postcreate(struct kvm_vcpu *vcpu)
311 312 313 314 315
{
}

void kvm_arch_vcpu_free(struct kvm_vcpu *vcpu)
{
316 317 318
	if (vcpu->arch.has_run_once && unlikely(!irqchip_in_kernel(vcpu->kvm)))
		static_branch_dec(&userspace_irqchip_in_use);

319
	kvm_mmu_free_memory_caches(vcpu);
320
	kvm_timer_vcpu_terminate(vcpu);
321
	kvm_pmu_vcpu_destroy(vcpu);
322
	kvm_vcpu_uninit(vcpu);
323
	kmem_cache_free(kvm_vcpu_cache, vcpu);
324 325 326 327 328 329 330 331 332
}

void kvm_arch_vcpu_destroy(struct kvm_vcpu *vcpu)
{
	kvm_arch_vcpu_free(vcpu);
}

int kvm_cpu_has_pending_timer(struct kvm_vcpu *vcpu)
{
333
	return kvm_timer_is_pending(vcpu);
334 335
}

336 337
void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu)
{
338
	kvm_vgic_v4_enable_doorbell(vcpu);
339 340 341 342
}

void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu)
{
343
	kvm_vgic_v4_disable_doorbell(vcpu);
344 345
}

346 347
int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu)
{
348 349
	/* Force users to call KVM_ARM_VCPU_INIT */
	vcpu->arch.target = -1;
350
	bitmap_zero(vcpu->arch.features, KVM_VCPU_MAX_FEATURES);
351

352 353 354
	/* Set up the timer */
	kvm_timer_vcpu_init(vcpu);

355 356
	kvm_arm_reset_debug_ptr(vcpu);

357
	return kvm_vgic_vcpu_init(vcpu);
358 359 360 361
}

void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
{
362 363 364 365 366 367 368 369 370 371 372 373 374
	int *last_ran;

	last_ran = this_cpu_ptr(vcpu->kvm->arch.last_vcpu_ran);

	/*
	 * We might get preempted before the vCPU actually runs, but
	 * over-invalidation doesn't affect correctness.
	 */
	if (*last_ran != vcpu->vcpu_id) {
		kvm_call_hyp(__kvm_tlb_flush_local_vmid, vcpu);
		*last_ran = vcpu->vcpu_id;
	}

375
	vcpu->cpu = cpu;
376
	vcpu->arch.host_cpu_context = this_cpu_ptr(&kvm_host_cpu_state);
377

378
	kvm_arm_set_running_vcpu(vcpu);
379
	kvm_vgic_load(vcpu);
380
	kvm_timer_vcpu_load(vcpu);
381
	kvm_vcpu_load_sysregs(vcpu);
382
	kvm_arch_vcpu_load_fp(vcpu);
383 384 385 386 387

	if (single_task_running())
		vcpu_clear_wfe_traps(vcpu);
	else
		vcpu_set_wfe_traps(vcpu);
388 389 390 391
}

void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)
{
392
	kvm_arch_vcpu_put_fp(vcpu);
393
	kvm_vcpu_put_sysregs(vcpu);
394
	kvm_timer_vcpu_put(vcpu);
395 396
	kvm_vgic_put(vcpu);

397 398
	vcpu->cpu = -1;

399
	kvm_arm_set_running_vcpu(NULL);
400 401
}

A
Andrew Jones 已提交
402 403 404
static void vcpu_power_off(struct kvm_vcpu *vcpu)
{
	vcpu->arch.power_off = true;
405
	kvm_make_request(KVM_REQ_SLEEP, vcpu);
A
Andrew Jones 已提交
406 407 408
	kvm_vcpu_kick(vcpu);
}

409 410 411
int kvm_arch_vcpu_ioctl_get_mpstate(struct kvm_vcpu *vcpu,
				    struct kvm_mp_state *mp_state)
{
412
	if (vcpu->arch.power_off)
413 414 415 416 417
		mp_state->mp_state = KVM_MP_STATE_STOPPED;
	else
		mp_state->mp_state = KVM_MP_STATE_RUNNABLE;

	return 0;
418 419 420 421 422
}

int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu *vcpu,
				    struct kvm_mp_state *mp_state)
{
423 424
	int ret = 0;

425 426
	switch (mp_state->mp_state) {
	case KVM_MP_STATE_RUNNABLE:
427
		vcpu->arch.power_off = false;
428 429
		break;
	case KVM_MP_STATE_STOPPED:
A
Andrew Jones 已提交
430
		vcpu_power_off(vcpu);
431 432
		break;
	default:
433
		ret = -EINVAL;
434 435
	}

436
	return ret;
437 438
}

439 440 441 442 443 444 445
/**
 * kvm_arch_vcpu_runnable - determine if the vcpu can be scheduled
 * @v:		The VCPU pointer
 *
 * If the guest CPU is not waiting for interrupts or an interrupt line is
 * asserted, the CPU is by definition runnable.
 */
446 447
int kvm_arch_vcpu_runnable(struct kvm_vcpu *v)
{
448 449
	bool irq_lines = *vcpu_hcr(v) & (HCR_VI | HCR_VF);
	return ((irq_lines || kvm_vgic_vcpu_pending_irq(v))
450
		&& !v->arch.power_off && !v->arch.pause);
451 452
}

453 454
bool kvm_arch_vcpu_in_kernel(struct kvm_vcpu *vcpu)
{
455
	return vcpu_mode_priv(vcpu);
456 457
}

458 459 460 461 462 463 464
/* Just ensure a guest exit from a particular CPU */
static void exit_vm_noop(void *info)
{
}

void force_vm_exit(const cpumask_t *mask)
{
465
	preempt_disable();
466
	smp_call_function_many(mask, exit_vm_noop, NULL, true);
467
	preempt_enable();
468 469 470 471
}

/**
 * need_new_vmid_gen - check that the VMID is still valid
472
 * @vmid: The VMID to check
473 474 475
 *
 * return true if there is a new generation of VMIDs being used
 *
476 477 478 479 480
 * The hardware supports a limited set of values with the value zero reserved
 * for the host, so we check if an assigned value belongs to a previous
 * generation, which which requires us to assign a new value. If we're the
 * first to use a VMID for the new generation, we must flush necessary caches
 * and TLBs on all CPUs.
481
 */
482
static bool need_new_vmid_gen(struct kvm_vmid *vmid)
483
{
484 485
	u64 current_vmid_gen = atomic64_read(&kvm_vmid_gen);
	smp_rmb(); /* Orders read of kvm_vmid_gen and kvm->arch.vmid */
486
	return unlikely(READ_ONCE(vmid->vmid_gen) != current_vmid_gen);
487 488 489
}

/**
490 491 492
 * update_vmid - Update the vmid with a valid VMID for the current generation
 * @kvm: The guest that struct vmid belongs to
 * @vmid: The stage-2 VMID information struct
493
 */
494
static void update_vmid(struct kvm_vmid *vmid)
495
{
496
	if (!need_new_vmid_gen(vmid))
497 498
		return;

499
	spin_lock(&kvm_vmid_lock);
500 501 502 503 504 505

	/*
	 * We need to re-check the vmid_gen here to ensure that if another vcpu
	 * already allocated a valid vmid for this vm, then this vcpu should
	 * use the same vmid.
	 */
506
	if (!need_new_vmid_gen(vmid)) {
507
		spin_unlock(&kvm_vmid_lock);
508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529
		return;
	}

	/* First user of a new VMID generation? */
	if (unlikely(kvm_next_vmid == 0)) {
		atomic64_inc(&kvm_vmid_gen);
		kvm_next_vmid = 1;

		/*
		 * On SMP we know no other CPUs can use this CPU's or each
		 * other's VMID after force_vm_exit returns since the
		 * kvm_vmid_lock blocks them from reentry to the guest.
		 */
		force_vm_exit(cpu_all_mask);
		/*
		 * Now broadcast TLB + ICACHE invalidation over the inner
		 * shareable domain to make sure all data structures are
		 * clean.
		 */
		kvm_call_hyp(__kvm_flush_vm_context);
	}

530
	vmid->vmid = kvm_next_vmid;
531
	kvm_next_vmid++;
532
	kvm_next_vmid &= (1 << kvm_get_vmid_bits()) - 1;
533

534
	smp_wmb();
535
	WRITE_ONCE(vmid->vmid_gen, atomic64_read(&kvm_vmid_gen));
536 537

	spin_unlock(&kvm_vmid_lock);
538 539 540 541
}

static int kvm_vcpu_first_run_init(struct kvm_vcpu *vcpu)
{
542
	struct kvm *kvm = vcpu->kvm;
543
	int ret = 0;
544

545 546 547
	if (likely(vcpu->arch.has_run_once))
		return 0;

548 549 550
	if (!kvm_arm_vcpu_is_finalized(vcpu))
		return -EPERM;

551
	vcpu->arch.has_run_once = true;
552

553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568
	if (likely(irqchip_in_kernel(kvm))) {
		/*
		 * Map the VGIC hardware resources before running a vcpu the
		 * first time on this VM.
		 */
		if (unlikely(!vgic_ready(kvm))) {
			ret = kvm_vgic_map_resources(kvm);
			if (ret)
				return ret;
		}
	} else {
		/*
		 * Tell the rest of the code that there are userspace irqchip
		 * VMs in the wild.
		 */
		static_branch_inc(&userspace_irqchip_in_use);
569 570
	}

571
	ret = kvm_timer_enable(vcpu);
572 573 574 575
	if (ret)
		return ret;

	ret = kvm_arm_pmu_v3_enable(vcpu);
576

577
	return ret;
578 579
}

580 581 582 583 584
bool kvm_arch_intc_initialized(struct kvm *kvm)
{
	return vgic_initialized(kvm);
}

585
void kvm_arm_halt_guest(struct kvm *kvm)
586 587 588 589 590 591
{
	int i;
	struct kvm_vcpu *vcpu;

	kvm_for_each_vcpu(i, vcpu, kvm)
		vcpu->arch.pause = true;
592
	kvm_make_all_cpus_request(kvm, KVM_REQ_SLEEP);
593 594
}

595
void kvm_arm_resume_guest(struct kvm *kvm)
596 597 598 599
{
	int i;
	struct kvm_vcpu *vcpu;

600 601
	kvm_for_each_vcpu(i, vcpu, kvm) {
		vcpu->arch.pause = false;
602
		swake_up_one(kvm_arch_vcpu_wq(vcpu));
603
	}
604 605
}

606
static void vcpu_req_sleep(struct kvm_vcpu *vcpu)
607
{
608
	struct swait_queue_head *wq = kvm_arch_vcpu_wq(vcpu);
609

610
	swait_event_interruptible_exclusive(*wq, ((!vcpu->arch.power_off) &&
611
				       (!vcpu->arch.pause)));
612

A
Andrew Jones 已提交
613
	if (vcpu->arch.power_off || vcpu->arch.pause) {
614
		/* Awaken to handle a signal, request we sleep again later. */
615
		kvm_make_request(KVM_REQ_SLEEP, vcpu);
616
	}
617 618 619 620 621 622 623

	/*
	 * Make sure we will observe a potential reset request if we've
	 * observed a change to the power state. Pairs with the smp_wmb() in
	 * kvm_psci_vcpu_on().
	 */
	smp_rmb();
624 625
}

626 627 628 629 630
static int kvm_vcpu_initialized(struct kvm_vcpu *vcpu)
{
	return vcpu->arch.target >= 0;
}

631 632 633
static void check_vcpu_requests(struct kvm_vcpu *vcpu)
{
	if (kvm_request_pending(vcpu)) {
634 635
		if (kvm_check_request(KVM_REQ_SLEEP, vcpu))
			vcpu_req_sleep(vcpu);
636

637 638 639
		if (kvm_check_request(KVM_REQ_VCPU_RESET, vcpu))
			kvm_reset_vcpu(vcpu);

640 641 642 643 644
		/*
		 * Clear IRQ_PENDING requests that were made to guarantee
		 * that a VCPU sees new virtual interrupts.
		 */
		kvm_check_request(KVM_REQ_IRQ_PENDING, vcpu);
645 646 647
	}
}

648 649 650 651 652 653 654 655 656 657 658
/**
 * kvm_arch_vcpu_ioctl_run - the main VCPU run function to execute guest code
 * @vcpu:	The VCPU pointer
 * @run:	The kvm_run structure pointer used for userspace state exchange
 *
 * This function is called through the VCPU_RUN ioctl called from user space. It
 * will execute VM code in a loop until the time slice for the process is used
 * or some emulation is needed from user space in which case the function will
 * return with return value 0 and with the kvm_run structure filled in with the
 * required data for the requested emulation.
 */
659 660
int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *run)
{
661 662
	int ret;

663
	if (unlikely(!kvm_vcpu_initialized(vcpu)))
664 665 666 667
		return -ENOEXEC;

	ret = kvm_vcpu_first_run_init(vcpu);
	if (ret)
668
		return ret;
669

C
Christoffer Dall 已提交
670 671 672
	if (run->exit_reason == KVM_EXIT_MMIO) {
		ret = kvm_handle_mmio_return(vcpu, vcpu->run);
		if (ret)
673
			return ret;
C
Christoffer Dall 已提交
674 675
	}

676 677 678 679
	if (run->immediate_exit)
		return -EINTR;

	vcpu_load(vcpu);
680

681
	kvm_sigset_activate(vcpu);
682 683 684 685 686 687 688 689 690

	ret = 1;
	run->exit_reason = KVM_EXIT_UNKNOWN;
	while (ret > 0) {
		/*
		 * Check conditions before entering the guest
		 */
		cond_resched();

691
		update_vmid(&vcpu->kvm->arch.vmid);
692

693 694
		check_vcpu_requests(vcpu);

695 696 697 698 699
		/*
		 * Preparing the interrupts to be injected also
		 * involves poking the GIC, which must be done in a
		 * non-preemptible context.
		 */
700
		preempt_disable();
701

702
		kvm_pmu_flush_hwstate(vcpu);
703

704 705
		local_irq_disable();

706 707
		kvm_vgic_flush_hwstate(vcpu);

708
		/*
709 710
		 * Exit if we have a signal pending so that we can deliver the
		 * signal to user space.
711
		 */
712
		if (signal_pending(current)) {
713 714 715 716
			ret = -EINTR;
			run->exit_reason = KVM_EXIT_INTR;
		}

717 718 719 720 721 722 723 724 725 726 727 728 729 730 731
		/*
		 * If we're using a userspace irqchip, then check if we need
		 * to tell a userspace irqchip about timer or PMU level
		 * changes and if so, exit to userspace (the actual level
		 * state gets updated in kvm_timer_update_run and
		 * kvm_pmu_update_run below).
		 */
		if (static_branch_unlikely(&userspace_irqchip_in_use)) {
			if (kvm_timer_should_notify_user(vcpu) ||
			    kvm_pmu_should_notify_user(vcpu)) {
				ret = -EINTR;
				run->exit_reason = KVM_EXIT_INTR;
			}
		}

732 733 734 735 736 737 738 739
		/*
		 * Ensure we set mode to IN_GUEST_MODE after we disable
		 * interrupts and before the final VCPU requests check.
		 * See the comment in kvm_vcpu_exiting_guest_mode() and
		 * Documentation/virtual/kvm/vcpu-requests.rst
		 */
		smp_store_mb(vcpu->mode, IN_GUEST_MODE);

740
		if (ret <= 0 || need_new_vmid_gen(&vcpu->kvm->arch.vmid) ||
A
Andrew Jones 已提交
741
		    kvm_request_pending(vcpu)) {
742
			vcpu->mode = OUTSIDE_GUEST_MODE;
743
			isb(); /* Ensure work in x_flush_hwstate is committed */
744
			kvm_pmu_sync_hwstate(vcpu);
745 746
			if (static_branch_unlikely(&userspace_irqchip_in_use))
				kvm_timer_sync_hwstate(vcpu);
747
			kvm_vgic_sync_hwstate(vcpu);
748
			local_irq_enable();
749
			preempt_enable();
750 751 752
			continue;
		}

753 754
		kvm_arm_setup_debug(vcpu);

755 756 757 758
		/**************************************************************
		 * Enter the guest
		 */
		trace_kvm_entry(*vcpu_pc(vcpu));
759
		guest_enter_irqoff();
760

761 762 763
		if (has_vhe()) {
			kvm_arm_vhe_guest_enter();
			ret = kvm_vcpu_run_vhe(vcpu);
764
			kvm_arm_vhe_guest_exit();
765
		} else {
766
			ret = kvm_call_hyp_ret(__kvm_vcpu_run_nvhe, vcpu);
767 768
		}

769
		vcpu->mode = OUTSIDE_GUEST_MODE;
770
		vcpu->stat.exits++;
771 772 773 774
		/*
		 * Back from guest
		 *************************************************************/

775 776
		kvm_arm_clear_debug(vcpu);

777
		/*
778
		 * We must sync the PMU state before the vgic state so
779 780 781 782 783
		 * that the vgic can properly sample the updated state of the
		 * interrupt line.
		 */
		kvm_pmu_sync_hwstate(vcpu);

784 785 786 787 788
		/*
		 * Sync the vgic state before syncing the timer state because
		 * the timer code needs to know if the virtual timer
		 * interrupts are active.
		 */
789 790
		kvm_vgic_sync_hwstate(vcpu);

791 792 793 794 795
		/*
		 * Sync the timer hardware state before enabling interrupts as
		 * we don't want vtimer interrupts to race with syncing the
		 * timer virtual interrupt state.
		 */
796 797
		if (static_branch_unlikely(&userspace_irqchip_in_use))
			kvm_timer_sync_hwstate(vcpu);
798

799 800
		kvm_arch_vcpu_ctxsync_fp(vcpu);

801 802 803 804 805 806 807 808 809 810 811 812 813
		/*
		 * We may have taken a host interrupt in HYP mode (ie
		 * while executing the guest). This interrupt is still
		 * pending, as we haven't serviced it yet!
		 *
		 * We're now back in SVC mode, with interrupts
		 * disabled.  Enabling the interrupts now will have
		 * the effect of taking the interrupt again, in SVC
		 * mode this time.
		 */
		local_irq_enable();

		/*
814
		 * We do local_irq_enable() before calling guest_exit() so
815 816
		 * that if a timer interrupt hits while running the guest we
		 * account that tick as being spent in the guest.  We enable
817
		 * preemption after calling guest_exit() so that if we get
818 819 820
		 * preempted we make sure ticks after that is not counted as
		 * guest time.
		 */
821
		guest_exit();
822
		trace_kvm_exit(ret, kvm_vcpu_trap_get_class(vcpu), *vcpu_pc(vcpu));
823

824 825 826
		/* Exit types that need handling before we can be preempted */
		handle_exit_early(vcpu, run, ret);

827 828
		preempt_enable();

829 830 831
		ret = handle_exit(vcpu, run, ret);
	}

832
	/* Tell userspace about in-kernel device output levels */
833 834 835 836
	if (unlikely(!irqchip_in_kernel(vcpu->kvm))) {
		kvm_timer_update_run(vcpu);
		kvm_pmu_update_run(vcpu);
	}
837

838 839
	kvm_sigset_deactivate(vcpu);

840
	vcpu_put(vcpu);
841
	return ret;
842 843
}

844 845 846 847
static int vcpu_interrupt_line(struct kvm_vcpu *vcpu, int number, bool level)
{
	int bit_index;
	bool set;
848
	unsigned long *hcr;
849 850 851 852 853 854

	if (number == KVM_ARM_IRQ_CPU_IRQ)
		bit_index = __ffs(HCR_VI);
	else /* KVM_ARM_IRQ_CPU_FIQ */
		bit_index = __ffs(HCR_VF);

855
	hcr = vcpu_hcr(vcpu);
856
	if (level)
857
		set = test_and_set_bit(bit_index, hcr);
858
	else
859
		set = test_and_clear_bit(bit_index, hcr);
860 861 862 863 864 865 866 867 868 869 870 871

	/*
	 * If we didn't change anything, no need to wake up or kick other CPUs
	 */
	if (set == level)
		return 0;

	/*
	 * The vcpu irq_lines field was updated, wake up sleeping VCPUs and
	 * trigger a world-switch round on the running physical CPU to set the
	 * virtual IRQ/FIQ fields in the HCR appropriately.
	 */
872
	kvm_make_request(KVM_REQ_IRQ_PENDING, vcpu);
873 874 875 876 877
	kvm_vcpu_kick(vcpu);

	return 0;
}

878 879
int kvm_vm_ioctl_irq_line(struct kvm *kvm, struct kvm_irq_level *irq_level,
			  bool line_status)
880 881 882 883 884 885 886 887 888 889 890 891 892
{
	u32 irq = irq_level->irq;
	unsigned int irq_type, vcpu_idx, irq_num;
	int nrcpus = atomic_read(&kvm->online_vcpus);
	struct kvm_vcpu *vcpu = NULL;
	bool level = irq_level->level;

	irq_type = (irq >> KVM_ARM_IRQ_TYPE_SHIFT) & KVM_ARM_IRQ_TYPE_MASK;
	vcpu_idx = (irq >> KVM_ARM_IRQ_VCPU_SHIFT) & KVM_ARM_IRQ_VCPU_MASK;
	irq_num = (irq >> KVM_ARM_IRQ_NUM_SHIFT) & KVM_ARM_IRQ_NUM_MASK;

	trace_kvm_irq_line(irq_type, vcpu_idx, irq_num, irq_level->level);

893 894 895 896
	switch (irq_type) {
	case KVM_ARM_IRQ_TYPE_CPU:
		if (irqchip_in_kernel(kvm))
			return -ENXIO;
897

898 899
		if (vcpu_idx >= nrcpus)
			return -EINVAL;
900

901 902 903
		vcpu = kvm_get_vcpu(kvm, vcpu_idx);
		if (!vcpu)
			return -EINVAL;
904

905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921
		if (irq_num > KVM_ARM_IRQ_CPU_FIQ)
			return -EINVAL;

		return vcpu_interrupt_line(vcpu, irq_num, level);
	case KVM_ARM_IRQ_TYPE_PPI:
		if (!irqchip_in_kernel(kvm))
			return -ENXIO;

		if (vcpu_idx >= nrcpus)
			return -EINVAL;

		vcpu = kvm_get_vcpu(kvm, vcpu_idx);
		if (!vcpu)
			return -EINVAL;

		if (irq_num < VGIC_NR_SGIS || irq_num >= VGIC_NR_PRIVATE_IRQS)
			return -EINVAL;
922

923
		return kvm_vgic_inject_irq(kvm, vcpu->vcpu_id, irq_num, level, NULL);
924 925 926 927
	case KVM_ARM_IRQ_TYPE_SPI:
		if (!irqchip_in_kernel(kvm))
			return -ENXIO;

928
		if (irq_num < VGIC_NR_PRIVATE_IRQS)
929 930
			return -EINVAL;

931
		return kvm_vgic_inject_irq(kvm, 0, irq_num, level, NULL);
932 933 934
	}

	return -EINVAL;
935 936
}

937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978
static int kvm_vcpu_set_target(struct kvm_vcpu *vcpu,
			       const struct kvm_vcpu_init *init)
{
	unsigned int i;
	int phys_target = kvm_target_cpu();

	if (init->target != phys_target)
		return -EINVAL;

	/*
	 * Secondary and subsequent calls to KVM_ARM_VCPU_INIT must
	 * use the same target.
	 */
	if (vcpu->arch.target != -1 && vcpu->arch.target != init->target)
		return -EINVAL;

	/* -ENOENT for unknown features, -EINVAL for invalid combinations. */
	for (i = 0; i < sizeof(init->features) * 8; i++) {
		bool set = (init->features[i / 32] & (1 << (i % 32)));

		if (set && i >= KVM_VCPU_MAX_FEATURES)
			return -ENOENT;

		/*
		 * Secondary and subsequent calls to KVM_ARM_VCPU_INIT must
		 * use the same feature set.
		 */
		if (vcpu->arch.target != -1 && i < KVM_VCPU_MAX_FEATURES &&
		    test_bit(i, vcpu->arch.features) != set)
			return -EINVAL;

		if (set)
			set_bit(i, vcpu->arch.features);
	}

	vcpu->arch.target = phys_target;

	/* Now we know what it is, we can reset it. */
	return kvm_reset_vcpu(vcpu);
}


979 980 981 982 983 984 985 986 987
static int kvm_arch_vcpu_ioctl_vcpu_init(struct kvm_vcpu *vcpu,
					 struct kvm_vcpu_init *init)
{
	int ret;

	ret = kvm_vcpu_set_target(vcpu, init);
	if (ret)
		return ret;

988 989 990 991 992 993 994
	/*
	 * Ensure a rebooted VM will fault in RAM pages and detect if the
	 * guest MMU is turned off and flush the caches as needed.
	 */
	if (vcpu->arch.has_run_once)
		stage2_unmap_vm(vcpu->kvm);

995 996
	vcpu_reset_hcr(vcpu);

997
	/*
998
	 * Handle the "start in power-off" case.
999
	 */
1000
	if (test_bit(KVM_ARM_VCPU_POWER_OFF, vcpu->arch.features))
A
Andrew Jones 已提交
1001
		vcpu_power_off(vcpu);
1002
	else
1003
		vcpu->arch.power_off = false;
1004 1005 1006 1007

	return 0;
}

1008 1009 1010 1011 1012 1013 1014
static int kvm_arm_vcpu_set_attr(struct kvm_vcpu *vcpu,
				 struct kvm_device_attr *attr)
{
	int ret = -ENXIO;

	switch (attr->group) {
	default:
1015
		ret = kvm_arm_vcpu_arch_set_attr(vcpu, attr);
1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028
		break;
	}

	return ret;
}

static int kvm_arm_vcpu_get_attr(struct kvm_vcpu *vcpu,
				 struct kvm_device_attr *attr)
{
	int ret = -ENXIO;

	switch (attr->group) {
	default:
1029
		ret = kvm_arm_vcpu_arch_get_attr(vcpu, attr);
1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042
		break;
	}

	return ret;
}

static int kvm_arm_vcpu_has_attr(struct kvm_vcpu *vcpu,
				 struct kvm_device_attr *attr)
{
	int ret = -ENXIO;

	switch (attr->group) {
	default:
1043
		ret = kvm_arm_vcpu_arch_has_attr(vcpu, attr);
1044 1045 1046 1047 1048 1049
		break;
	}

	return ret;
}

1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075
static int kvm_arm_vcpu_get_events(struct kvm_vcpu *vcpu,
				   struct kvm_vcpu_events *events)
{
	memset(events, 0, sizeof(*events));

	return __kvm_arm_vcpu_get_events(vcpu, events);
}

static int kvm_arm_vcpu_set_events(struct kvm_vcpu *vcpu,
				   struct kvm_vcpu_events *events)
{
	int i;

	/* check whether the reserved field is zero */
	for (i = 0; i < ARRAY_SIZE(events->reserved); i++)
		if (events->reserved[i])
			return -EINVAL;

	/* check whether the pad field is zero */
	for (i = 0; i < ARRAY_SIZE(events->exception.pad); i++)
		if (events->exception.pad[i])
			return -EINVAL;

	return __kvm_arm_vcpu_set_events(vcpu, events);
}

1076 1077 1078 1079 1080
long kvm_arch_vcpu_ioctl(struct file *filp,
			 unsigned int ioctl, unsigned long arg)
{
	struct kvm_vcpu *vcpu = filp->private_data;
	void __user *argp = (void __user *)arg;
1081
	struct kvm_device_attr attr;
1082 1083
	long r;

1084 1085 1086 1087
	switch (ioctl) {
	case KVM_ARM_VCPU_INIT: {
		struct kvm_vcpu_init init;

1088
		r = -EFAULT;
1089
		if (copy_from_user(&init, argp, sizeof(init)))
1090
			break;
1091

1092 1093
		r = kvm_arch_vcpu_ioctl_vcpu_init(vcpu, &init);
		break;
1094 1095 1096 1097
	}
	case KVM_SET_ONE_REG:
	case KVM_GET_ONE_REG: {
		struct kvm_one_reg reg;
1098

1099
		r = -ENOEXEC;
1100
		if (unlikely(!kvm_vcpu_initialized(vcpu)))
1101
			break;
1102

1103
		r = -EFAULT;
1104
		if (copy_from_user(&reg, argp, sizeof(reg)))
1105 1106
			break;

1107
		if (ioctl == KVM_SET_ONE_REG)
1108
			r = kvm_arm_set_reg(vcpu, &reg);
1109
		else
1110 1111
			r = kvm_arm_get_reg(vcpu, &reg);
		break;
1112 1113 1114 1115 1116 1117
	}
	case KVM_GET_REG_LIST: {
		struct kvm_reg_list __user *user_list = argp;
		struct kvm_reg_list reg_list;
		unsigned n;

1118
		r = -ENOEXEC;
1119
		if (unlikely(!kvm_vcpu_initialized(vcpu)))
1120
			break;
1121

1122 1123 1124 1125
		r = -EPERM;
		if (!kvm_arm_vcpu_is_finalized(vcpu))
			break;

1126
		r = -EFAULT;
1127
		if (copy_from_user(&reg_list, user_list, sizeof(reg_list)))
1128
			break;
1129 1130 1131
		n = reg_list.n;
		reg_list.n = kvm_arm_num_regs(vcpu);
		if (copy_to_user(user_list, &reg_list, sizeof(reg_list)))
1132 1133
			break;
		r = -E2BIG;
1134
		if (n < reg_list.n)
1135 1136 1137
			break;
		r = kvm_arm_copy_reg_indices(vcpu, user_list->reg);
		break;
1138
	}
1139
	case KVM_SET_DEVICE_ATTR: {
1140
		r = -EFAULT;
1141
		if (copy_from_user(&attr, argp, sizeof(attr)))
1142 1143 1144
			break;
		r = kvm_arm_vcpu_set_attr(vcpu, &attr);
		break;
1145 1146
	}
	case KVM_GET_DEVICE_ATTR: {
1147
		r = -EFAULT;
1148
		if (copy_from_user(&attr, argp, sizeof(attr)))
1149 1150 1151
			break;
		r = kvm_arm_vcpu_get_attr(vcpu, &attr);
		break;
1152 1153
	}
	case KVM_HAS_DEVICE_ATTR: {
1154
		r = -EFAULT;
1155
		if (copy_from_user(&attr, argp, sizeof(attr)))
1156 1157 1158
			break;
		r = kvm_arm_vcpu_has_attr(vcpu, &attr);
		break;
1159
	}
1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178
	case KVM_GET_VCPU_EVENTS: {
		struct kvm_vcpu_events events;

		if (kvm_arm_vcpu_get_events(vcpu, &events))
			return -EINVAL;

		if (copy_to_user(argp, &events, sizeof(events)))
			return -EFAULT;

		return 0;
	}
	case KVM_SET_VCPU_EVENTS: {
		struct kvm_vcpu_events events;

		if (copy_from_user(&events, argp, sizeof(events)))
			return -EFAULT;

		return kvm_arm_vcpu_set_events(vcpu, &events);
	}
1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189
	case KVM_ARM_VCPU_FINALIZE: {
		int what;

		if (!kvm_vcpu_initialized(vcpu))
			return -ENOEXEC;

		if (get_user(what, (const int __user *)argp))
			return -EFAULT;

		return kvm_arm_vcpu_finalize(vcpu, what);
	}
1190
	default:
1191
		r = -EINVAL;
1192
	}
1193 1194

	return r;
1195 1196
}

1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215
/**
 * kvm_vm_ioctl_get_dirty_log - get and clear the log of dirty pages in a slot
 * @kvm: kvm instance
 * @log: slot id and address to which we copy the log
 *
 * Steps 1-4 below provide general overview of dirty page logging. See
 * kvm_get_dirty_log_protect() function description for additional details.
 *
 * We call kvm_get_dirty_log_protect() to handle steps 1-3, upon return we
 * always flush the TLB (step 4) even if previous step failed  and the dirty
 * bitmap may be corrupt. Regardless of previous outcome the KVM logging API
 * does not preclude user space subsequent dirty log read. Flushing TLB ensures
 * writes will be marked dirty for next log read.
 *
 *   1. Take a snapshot of the bit and clear it if needed.
 *   2. Write protect the corresponding page.
 *   3. Copy the snapshot to the userspace.
 *   4. Flush TLB's if needed.
 */
1216 1217
int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm, struct kvm_dirty_log *log)
{
1218
	bool flush = false;
1219 1220 1221 1222
	int r;

	mutex_lock(&kvm->slots_lock);

1223
	r = kvm_get_dirty_log_protect(kvm, log, &flush);
1224

1225
	if (flush)
1226 1227 1228 1229
		kvm_flush_remote_tlbs(kvm);

	mutex_unlock(&kvm->slots_lock);
	return r;
1230 1231
}

1232 1233 1234
int kvm_vm_ioctl_clear_dirty_log(struct kvm *kvm, struct kvm_clear_dirty_log *log)
{
	bool flush = false;
1235 1236 1237 1238
	int r;

	mutex_lock(&kvm->slots_lock);

1239
	r = kvm_clear_dirty_log_protect(kvm, log, &flush);
1240

1241
	if (flush)
1242 1243 1244 1245
		kvm_flush_remote_tlbs(kvm);

	mutex_unlock(&kvm->slots_lock);
	return r;
1246 1247
}

1248 1249 1250
static int kvm_vm_ioctl_set_device_addr(struct kvm *kvm,
					struct kvm_arm_device_addr *dev_addr)
{
1251 1252 1253 1254 1255 1256 1257 1258 1259
	unsigned long dev_id, type;

	dev_id = (dev_addr->id & KVM_ARM_DEVICE_ID_MASK) >>
		KVM_ARM_DEVICE_ID_SHIFT;
	type = (dev_addr->id & KVM_ARM_DEVICE_TYPE_MASK) >>
		KVM_ARM_DEVICE_TYPE_SHIFT;

	switch (dev_id) {
	case KVM_ARM_DEVICE_VGIC_V2:
1260 1261
		if (!vgic_present)
			return -ENXIO;
1262
		return kvm_vgic_addr(kvm, type, &dev_addr->addr, true);
1263 1264 1265
	default:
		return -ENODEV;
	}
1266 1267
}

1268 1269 1270
long kvm_arch_vm_ioctl(struct file *filp,
		       unsigned int ioctl, unsigned long arg)
{
1271 1272 1273 1274
	struct kvm *kvm = filp->private_data;
	void __user *argp = (void __user *)arg;

	switch (ioctl) {
1275
	case KVM_CREATE_IRQCHIP: {
1276
		int ret;
1277 1278
		if (!vgic_present)
			return -ENXIO;
1279 1280 1281 1282
		mutex_lock(&kvm->lock);
		ret = kvm_vgic_create(kvm, KVM_DEV_TYPE_ARM_VGIC_V2);
		mutex_unlock(&kvm->lock);
		return ret;
1283
	}
1284 1285 1286 1287 1288 1289 1290
	case KVM_ARM_SET_DEVICE_ADDR: {
		struct kvm_arm_device_addr dev_addr;

		if (copy_from_user(&dev_addr, argp, sizeof(dev_addr)))
			return -EFAULT;
		return kvm_vm_ioctl_set_device_addr(kvm, &dev_addr);
	}
1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303
	case KVM_ARM_PREFERRED_TARGET: {
		int err;
		struct kvm_vcpu_init init;

		err = kvm_vcpu_preferred_target(&init);
		if (err)
			return err;

		if (copy_to_user(argp, &init, sizeof(init)))
			return -EFAULT;

		return 0;
	}
1304 1305 1306
	default:
		return -EINVAL;
	}
1307 1308
}

1309
static void cpu_init_hyp_mode(void *dummy)
1310
{
1311
	phys_addr_t pgd_ptr;
1312 1313 1314 1315 1316
	unsigned long hyp_stack_ptr;
	unsigned long stack_page;
	unsigned long vector_ptr;

	/* Switch from the HYP stub to our own HYP init vector */
1317
	__hyp_set_vectors(kvm_get_idmap_vector());
1318

1319
	pgd_ptr = kvm_mmu_get_httbr();
1320
	stack_page = __this_cpu_read(kvm_arm_hyp_stack_page);
1321
	hyp_stack_ptr = stack_page + PAGE_SIZE;
1322
	vector_ptr = (unsigned long)kvm_get_hyp_vector();
1323

M
Marc Zyngier 已提交
1324
	__cpu_init_hyp_mode(pgd_ptr, hyp_stack_ptr, vector_ptr);
1325
	__cpu_init_stage2();
1326 1327
}

1328 1329 1330 1331 1332 1333
static void cpu_hyp_reset(void)
{
	if (!is_kernel_in_hyp_mode())
		__hyp_reset_vectors();
}

1334 1335
static void cpu_hyp_reinit(void)
{
1336 1337
	cpu_hyp_reset();

1338
	if (is_kernel_in_hyp_mode())
1339
		kvm_timer_init_vhe();
1340
	else
1341
		cpu_init_hyp_mode(NULL);
1342

1343
	kvm_arm_init_debug();
1344 1345 1346

	if (vgic_present)
		kvm_vgic_init_cpu_hardware();
1347 1348
}

1349 1350 1351
static void _kvm_arch_hardware_enable(void *discard)
{
	if (!__this_cpu_read(kvm_arm_hardware_enabled)) {
1352
		cpu_hyp_reinit();
1353
		__this_cpu_write(kvm_arm_hardware_enabled, 1);
1354
	}
1355
}
1356

1357 1358 1359 1360
int kvm_arch_hardware_enable(void)
{
	_kvm_arch_hardware_enable(NULL);
	return 0;
1361 1362
}

1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374
static void _kvm_arch_hardware_disable(void *discard)
{
	if (__this_cpu_read(kvm_arm_hardware_enabled)) {
		cpu_hyp_reset();
		__this_cpu_write(kvm_arm_hardware_enabled, 0);
	}
}

void kvm_arch_hardware_disable(void)
{
	_kvm_arch_hardware_disable(NULL);
}
1375

1376 1377 1378 1379 1380
#ifdef CONFIG_CPU_PM
static int hyp_init_cpu_pm_notifier(struct notifier_block *self,
				    unsigned long cmd,
				    void *v)
{
1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395
	/*
	 * kvm_arm_hardware_enabled is left with its old value over
	 * PM_ENTER->PM_EXIT. It is used to indicate PM_EXIT should
	 * re-enable hyp.
	 */
	switch (cmd) {
	case CPU_PM_ENTER:
		if (__this_cpu_read(kvm_arm_hardware_enabled))
			/*
			 * don't update kvm_arm_hardware_enabled here
			 * so that the hardware will be re-enabled
			 * when we resume. See below.
			 */
			cpu_hyp_reset();

1396
		return NOTIFY_OK;
1397
	case CPU_PM_ENTER_FAILED:
1398 1399 1400 1401
	case CPU_PM_EXIT:
		if (__this_cpu_read(kvm_arm_hardware_enabled))
			/* The hardware was enabled before suspend. */
			cpu_hyp_reinit();
1402

1403 1404 1405 1406 1407
		return NOTIFY_OK;

	default:
		return NOTIFY_DONE;
	}
1408 1409 1410 1411 1412 1413 1414 1415 1416 1417
}

static struct notifier_block hyp_init_cpu_pm_nb = {
	.notifier_call = hyp_init_cpu_pm_notifier,
};

static void __init hyp_cpu_pm_init(void)
{
	cpu_pm_register_notifier(&hyp_init_cpu_pm_nb);
}
1418 1419 1420 1421
static void __init hyp_cpu_pm_exit(void)
{
	cpu_pm_unregister_notifier(&hyp_init_cpu_pm_nb);
}
1422 1423 1424 1425
#else
static inline void hyp_cpu_pm_init(void)
{
}
1426 1427 1428
static inline void hyp_cpu_pm_exit(void)
{
}
1429 1430
#endif

1431 1432
static int init_common_resources(void)
{
1433 1434
	kvm_set_ipa_limit();

1435 1436 1437 1438 1439
	return 0;
}

static int init_subsystems(void)
{
1440
	int err = 0;
1441

1442
	/*
1443
	 * Enable hardware so that subsystem initialisation can access EL2.
1444
	 */
1445
	on_each_cpu(_kvm_arch_hardware_enable, NULL, 1);
1446 1447 1448 1449 1450 1451

	/*
	 * Register CPU lower-power notifier
	 */
	hyp_cpu_pm_init();

1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462
	/*
	 * Init HYP view of VGIC
	 */
	err = kvm_vgic_hyp_init();
	switch (err) {
	case 0:
		vgic_present = true;
		break;
	case -ENODEV:
	case -ENXIO:
		vgic_present = false;
1463
		err = 0;
1464 1465
		break;
	default:
1466
		goto out;
1467 1468 1469 1470 1471
	}

	/*
	 * Init HYP architected timer support
	 */
1472
	err = kvm_timer_hyp_init(vgic_present);
1473
	if (err)
1474
		goto out;
1475 1476 1477 1478

	kvm_perf_init();
	kvm_coproc_table_init();

1479 1480 1481 1482
out:
	on_each_cpu(_kvm_arch_hardware_disable, NULL, 1);

	return err;
1483 1484 1485 1486 1487 1488 1489 1490 1491
}

static void teardown_hyp_mode(void)
{
	int cpu;

	free_hyp_pgds();
	for_each_possible_cpu(cpu)
		free_page(per_cpu(kvm_arm_hyp_stack_page, cpu));
1492
	hyp_cpu_pm_exit();
1493 1494
}

1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518
/**
 * Inits Hyp-mode on all online CPUs
 */
static int init_hyp_mode(void)
{
	int cpu;
	int err = 0;

	/*
	 * Allocate Hyp PGD and setup Hyp identity mapping
	 */
	err = kvm_mmu_init();
	if (err)
		goto out_err;

	/*
	 * Allocate stack pages for Hypervisor-mode
	 */
	for_each_possible_cpu(cpu) {
		unsigned long stack_page;

		stack_page = __get_free_page(GFP_KERNEL);
		if (!stack_page) {
			err = -ENOMEM;
1519
			goto out_err;
1520 1521 1522 1523 1524 1525 1526 1527
		}

		per_cpu(kvm_arm_hyp_stack_page, cpu) = stack_page;
	}

	/*
	 * Map the Hyp-code called directly from the host
	 */
1528
	err = create_hyp_mappings(kvm_ksym_ref(__hyp_text_start),
1529
				  kvm_ksym_ref(__hyp_text_end), PAGE_HYP_EXEC);
1530 1531
	if (err) {
		kvm_err("Cannot map world-switch code\n");
1532
		goto out_err;
1533 1534
	}

1535
	err = create_hyp_mappings(kvm_ksym_ref(__start_rodata),
1536
				  kvm_ksym_ref(__end_rodata), PAGE_HYP_RO);
1537 1538
	if (err) {
		kvm_err("Cannot map rodata section\n");
M
Marc Zyngier 已提交
1539 1540 1541 1542 1543 1544 1545
		goto out_err;
	}

	err = create_hyp_mappings(kvm_ksym_ref(__bss_start),
				  kvm_ksym_ref(__bss_stop), PAGE_HYP_RO);
	if (err) {
		kvm_err("Cannot map bss section\n");
1546
		goto out_err;
1547 1548
	}

1549 1550 1551 1552 1553 1554
	err = kvm_map_vectors();
	if (err) {
		kvm_err("Cannot map vectors\n");
		goto out_err;
	}

1555 1556 1557 1558 1559
	/*
	 * Map the Hyp stack pages
	 */
	for_each_possible_cpu(cpu) {
		char *stack_page = (char *)per_cpu(kvm_arm_hyp_stack_page, cpu);
1560 1561
		err = create_hyp_mappings(stack_page, stack_page + PAGE_SIZE,
					  PAGE_HYP);
1562 1563 1564

		if (err) {
			kvm_err("Cannot map hyp stack\n");
1565
			goto out_err;
1566 1567 1568 1569
		}
	}

	for_each_possible_cpu(cpu) {
1570
		kvm_cpu_context_t *cpu_ctxt;
1571

1572
		cpu_ctxt = per_cpu_ptr(&kvm_host_cpu_state, cpu);
1573
		kvm_init_host_cpu_context(cpu_ctxt, cpu);
1574
		err = create_hyp_mappings(cpu_ctxt, cpu_ctxt + 1, PAGE_HYP);
1575 1576

		if (err) {
1577
			kvm_err("Cannot map host CPU state: %d\n", err);
1578
			goto out_err;
1579 1580 1581
		}
	}

1582 1583
	err = hyp_map_aux_data();
	if (err)
1584
		kvm_err("Cannot map host auxiliary data: %d\n", err);
1585

1586
	return 0;
1587

1588
out_err:
1589
	teardown_hyp_mode();
1590 1591 1592 1593
	kvm_err("error initializing Hyp mode: %d\n", err);
	return err;
}

1594 1595 1596 1597 1598
static void check_kvm_target_cpu(void *ret)
{
	*(int *)ret = kvm_target_cpu();
}

1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611
struct kvm_vcpu *kvm_mpidr_to_vcpu(struct kvm *kvm, unsigned long mpidr)
{
	struct kvm_vcpu *vcpu;
	int i;

	mpidr &= MPIDR_HWID_BITMASK;
	kvm_for_each_vcpu(i, vcpu, kvm) {
		if (mpidr == kvm_vcpu_get_mpidr_aff(vcpu))
			return vcpu;
	}
	return NULL;
}

1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622
bool kvm_arch_has_irq_bypass(void)
{
	return true;
}

int kvm_arch_irq_bypass_add_producer(struct irq_bypass_consumer *cons,
				      struct irq_bypass_producer *prod)
{
	struct kvm_kernel_irqfd *irqfd =
		container_of(cons, struct kvm_kernel_irqfd, consumer);

1623 1624
	return kvm_vgic_v4_set_forwarding(irqfd->kvm, prod->irq,
					  &irqfd->irq_entry);
1625 1626 1627 1628 1629 1630 1631
}
void kvm_arch_irq_bypass_del_producer(struct irq_bypass_consumer *cons,
				      struct irq_bypass_producer *prod)
{
	struct kvm_kernel_irqfd *irqfd =
		container_of(cons, struct kvm_kernel_irqfd, consumer);

1632 1633
	kvm_vgic_v4_unset_forwarding(irqfd->kvm, prod->irq,
				     &irqfd->irq_entry);
1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651
}

void kvm_arch_irq_bypass_stop(struct irq_bypass_consumer *cons)
{
	struct kvm_kernel_irqfd *irqfd =
		container_of(cons, struct kvm_kernel_irqfd, consumer);

	kvm_arm_halt_guest(irqfd->kvm);
}

void kvm_arch_irq_bypass_start(struct irq_bypass_consumer *cons)
{
	struct kvm_kernel_irqfd *irqfd =
		container_of(cons, struct kvm_kernel_irqfd, consumer);

	kvm_arm_resume_guest(irqfd->kvm);
}

1652 1653 1654
/**
 * Initialize Hyp-mode and memory mappings on all CPUs.
 */
1655 1656
int kvm_arch_init(void *opaque)
{
1657
	int err;
1658
	int ret, cpu;
1659
	bool in_hyp_mode;
1660 1661

	if (!is_hyp_mode_available()) {
1662
		kvm_info("HYP mode not available\n");
1663 1664 1665
		return -ENODEV;
	}

1666 1667 1668 1669
	in_hyp_mode = is_kernel_in_hyp_mode();

	if (!in_hyp_mode && kvm_arch_requires_vhe()) {
		kvm_pr_unimpl("CPU unsupported in non-VHE mode, not initializing\n");
1670 1671 1672
		return -ENODEV;
	}

1673 1674 1675 1676 1677 1678
	for_each_online_cpu(cpu) {
		smp_call_function_single(cpu, check_kvm_target_cpu, &ret, 1);
		if (ret < 0) {
			kvm_err("Error, CPU %d not supported!\n", cpu);
			return -ENODEV;
		}
1679 1680
	}

1681
	err = init_common_resources();
1682
	if (err)
1683
		return err;
1684

1685
	err = kvm_arm_init_sve();
1686 1687 1688
	if (err)
		return err;

1689
	if (!in_hyp_mode) {
1690
		err = init_hyp_mode();
1691 1692 1693
		if (err)
			goto out_err;
	}
1694

1695 1696 1697
	err = init_subsystems();
	if (err)
		goto out_hyp;
1698

1699 1700 1701 1702 1703
	if (in_hyp_mode)
		kvm_info("VHE mode initialized successfully\n");
	else
		kvm_info("Hyp mode initialized successfully\n");

1704
	return 0;
1705 1706

out_hyp:
1707 1708
	if (!in_hyp_mode)
		teardown_hyp_mode();
1709 1710
out_err:
	return err;
1711 1712 1713 1714 1715
}

/* NOP: Compiling as a module not supported */
void kvm_arch_exit(void)
{
1716
	kvm_perf_teardown();
1717 1718 1719 1720 1721 1722 1723 1724 1725
}

static int arm_init(void)
{
	int rc = kvm_init(NULL, sizeof(struct kvm_vcpu), 0, THIS_MODULE);
	return rc;
}

module_init(arm_init);