arm.c 34.3 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/cpu_pm.h>
20 21 22
#include <linux/errno.h>
#include <linux/err.h>
#include <linux/kvm_host.h>
23
#include <linux/list.h>
24 25 26 27 28
#include <linux/module.h>
#include <linux/vmalloc.h>
#include <linux/fs.h>
#include <linux/mman.h>
#include <linux/sched.h>
29
#include <linux/kvm.h>
30 31
#include <linux/kvm_irqfd.h>
#include <linux/irqbypass.h>
32
#include <trace/events/kvm.h>
33
#include <kvm/arm_pmu.h>
34 35 36 37

#define CREATE_TRACE_POINTS
#include "trace.h"

38
#include <linux/uaccess.h>
39 40
#include <asm/ptrace.h>
#include <asm/mman.h>
41
#include <asm/tlbflush.h>
42
#include <asm/cacheflush.h>
43 44 45 46
#include <asm/virt.h>
#include <asm/kvm_arm.h>
#include <asm/kvm_asm.h>
#include <asm/kvm_mmu.h>
47
#include <asm/kvm_emulate.h>
48
#include <asm/kvm_coproc.h>
49
#include <asm/kvm_psci.h>
50
#include <asm/sections.h>
51 52 53 54 55

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

56
static DEFINE_PER_CPU(unsigned long, kvm_arm_hyp_stack_page);
57
static kvm_cpu_context_t __percpu *kvm_host_cpu_state;
58

59 60 61
/* Per-CPU variable containing the currently running vcpu. */
static DEFINE_PER_CPU(struct kvm_vcpu *, kvm_arm_running_vcpu);

62 63
/* The VMID used in the VTTBR */
static atomic64_t kvm_vmid_gen = ATOMIC64_INIT(1);
64 65
static u32 kvm_next_vmid;
static unsigned int kvm_vmid_bits __read_mostly;
66
static DEFINE_SPINLOCK(kvm_vmid_lock);
67

68 69
static bool vgic_present;

70 71
static DEFINE_PER_CPU(unsigned char, kvm_arm_hardware_enabled);

72 73 74
static void kvm_arm_set_running_vcpu(struct kvm_vcpu *vcpu)
{
	BUG_ON(preemptible());
75
	__this_cpu_write(kvm_arm_running_vcpu, vcpu);
76 77 78 79 80 81 82 83 84
}

/**
 * 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)
{
	BUG_ON(preemptible());
85
	return __this_cpu_read(kvm_arm_running_vcpu);
86 87 88 89 90
}

/**
 * kvm_arm_get_running_vcpus - get the per-CPU array of currently running vcpus.
 */
91
struct kvm_vcpu * __percpu *kvm_get_running_vcpus(void)
92 93 94 95
{
	return &kvm_arm_running_vcpu;
}

96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
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;
}


112 113 114 115
/**
 * kvm_arch_init_vm - initializes a VM data structure
 * @kvm:	pointer to the KVM struct
 */
116 117
int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
{
118
	int ret, cpu;
119

120 121 122
	if (type)
		return -EINVAL;

123 124 125 126 127 128 129
	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;

130 131 132 133
	ret = kvm_alloc_stage2_pgd(kvm);
	if (ret)
		goto out_fail_alloc;

134
	ret = create_hyp_mappings(kvm, kvm + 1, PAGE_HYP);
135 136 137
	if (ret)
		goto out_free_stage2_pgd;

138
	kvm_vgic_early_init(kvm);
139

140 141 142
	/* Mark the initial VMID generation invalid */
	kvm->arch.vmid_gen = 0;

143
	/* The maximum number of VCPUs is limited by the host's GIC model */
144 145
	kvm->arch.max_vcpus = vgic_present ?
				kvm_vgic_get_max_vcpus() : KVM_MAX_VCPUS;
146

147 148 149 150
	return ret;
out_free_stage2_pgd:
	kvm_free_stage2_pgd(kvm);
out_fail_alloc:
151 152
	free_percpu(kvm->arch.last_vcpu_ran);
	kvm->arch.last_vcpu_ran = NULL;
153
	return ret;
154 155
}

156 157 158 159 160 161 162 163 164 165
bool kvm_arch_has_vcpu_debugfs(void)
{
	return false;
}

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

166 167 168 169 170 171
int kvm_arch_vcpu_fault(struct kvm_vcpu *vcpu, struct vm_fault *vmf)
{
	return VM_FAULT_SIGBUS;
}


172 173 174 175
/**
 * kvm_arch_destroy_vm - destroy the VM data structure
 * @kvm:	pointer to the KVM struct
 */
176 177 178 179
void kvm_arch_destroy_vm(struct kvm *kvm)
{
	int i;

180 181
	kvm_vgic_destroy(kvm);

182 183 184
	free_percpu(kvm->arch.last_vcpu_ran);
	kvm->arch.last_vcpu_ran = NULL;

185 186 187 188 189 190
	for (i = 0; i < KVM_MAX_VCPUS; ++i) {
		if (kvm->vcpus[i]) {
			kvm_arch_vcpu_free(kvm->vcpus[i]);
			kvm->vcpus[i] = NULL;
		}
	}
191
	atomic_set(&kvm->online_vcpus, 0);
192 193
}

194
int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
195 196 197
{
	int r;
	switch (ext) {
198
	case KVM_CAP_IRQCHIP:
199 200
		r = vgic_present;
		break;
201
	case KVM_CAP_IOEVENTFD:
202
	case KVM_CAP_DEVICE_CTRL:
203 204 205 206
	case KVM_CAP_USER_MEMORY:
	case KVM_CAP_SYNC_MMU:
	case KVM_CAP_DESTROY_MEMORY_REGION_WORKS:
	case KVM_CAP_ONE_REG:
207
	case KVM_CAP_ARM_PSCI:
208
	case KVM_CAP_ARM_PSCI_0_2:
209
	case KVM_CAP_READONLY_MEM:
210
	case KVM_CAP_MP_STATE:
211
	case KVM_CAP_IMMEDIATE_EXIT:
212 213
		r = 1;
		break;
214 215
	case KVM_CAP_ARM_SET_DEVICE_ADDR:
		r = 1;
216
		break;
217 218 219 220 221 222
	case KVM_CAP_NR_VCPUS:
		r = num_online_cpus();
		break;
	case KVM_CAP_MAX_VCPUS:
		r = KVM_MAX_VCPUS;
		break;
223 224 225
	case KVM_CAP_NR_MEMSLOTS:
		r = KVM_USER_MEM_SLOTS;
		break;
V
Vladimir Murzin 已提交
226 227 228 229 230 231
	case KVM_CAP_MSI_DEVID:
		if (!kvm)
			r = -EINVAL;
		else
			r = kvm->arch.vgic.msis_require_devid;
		break;
232 233 234 235 236 237 238
	case KVM_CAP_ARM_USER_IRQ:
		/*
		 * 1: EL1_VTIMER, EL1_PTIMER, and PMU.
		 * (bump this number if adding more devices)
		 */
		r = 1;
		break;
239
	default:
240
		r = kvm_arch_dev_ioctl_check_extension(kvm, ext);
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
		break;
	}
	return r;
}

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


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

258 259 260 261 262
	if (irqchip_in_kernel(kvm) && vgic_initialized(kvm)) {
		err = -EBUSY;
		goto out;
	}

263 264 265 266 267
	if (id >= kvm->arch.max_vcpus) {
		err = -EINVAL;
		goto out;
	}

268 269 270 271 272 273 274 275 276 277
	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;

278
	err = create_hyp_mappings(vcpu, vcpu + 1, PAGE_HYP);
279 280 281
	if (err)
		goto vcpu_uninit;

282
	return vcpu;
283 284
vcpu_uninit:
	kvm_vcpu_uninit(vcpu);
285 286 287 288 289 290
free_vcpu:
	kmem_cache_free(kvm_vcpu_cache, vcpu);
out:
	return ERR_PTR(err);
}

291
void kvm_arch_vcpu_postcreate(struct kvm_vcpu *vcpu)
292
{
293
	kvm_vgic_vcpu_early_init(vcpu);
294 295 296 297
}

void kvm_arch_vcpu_free(struct kvm_vcpu *vcpu)
{
298
	kvm_mmu_free_memory_caches(vcpu);
299
	kvm_timer_vcpu_terminate(vcpu);
300
	kvm_pmu_vcpu_destroy(vcpu);
301
	kvm_vcpu_uninit(vcpu);
302
	kmem_cache_free(kvm_vcpu_cache, vcpu);
303 304 305 306 307 308 309 310 311
}

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

int kvm_cpu_has_pending_timer(struct kvm_vcpu *vcpu)
{
312
	return kvm_timer_is_pending(vcpu);
313 314
}

315 316 317
void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu)
{
	kvm_timer_schedule(vcpu);
318
	kvm_vgic_v4_enable_doorbell(vcpu);
319 320 321 322 323
}

void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu)
{
	kvm_timer_unschedule(vcpu);
324
	kvm_vgic_v4_disable_doorbell(vcpu);
325 326
}

327 328
int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu)
{
329 330
	/* Force users to call KVM_ARM_VCPU_INIT */
	vcpu->arch.target = -1;
331
	bitmap_zero(vcpu->arch.features, KVM_VCPU_MAX_FEATURES);
332

333 334 335
	/* Set up the timer */
	kvm_timer_vcpu_init(vcpu);

336 337
	kvm_arm_reset_debug_ptr(vcpu);

338
	return kvm_vgic_vcpu_init(vcpu);
339 340 341 342
}

void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
{
343 344 345 346 347 348 349 350 351 352 353 354 355
	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;
	}

356
	vcpu->cpu = cpu;
357
	vcpu->arch.host_cpu_context = this_cpu_ptr(kvm_host_cpu_state);
358

359
	kvm_arm_set_running_vcpu(vcpu);
360
	kvm_vgic_load(vcpu);
361
	kvm_timer_vcpu_load(vcpu);
362 363 364 365
}

void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)
{
366
	kvm_timer_vcpu_put(vcpu);
367 368
	kvm_vgic_put(vcpu);

369 370
	vcpu->cpu = -1;

371
	kvm_arm_set_running_vcpu(NULL);
372 373
}

A
Andrew Jones 已提交
374 375 376
static void vcpu_power_off(struct kvm_vcpu *vcpu)
{
	vcpu->arch.power_off = true;
377
	kvm_make_request(KVM_REQ_SLEEP, vcpu);
A
Andrew Jones 已提交
378 379 380
	kvm_vcpu_kick(vcpu);
}

381 382 383
int kvm_arch_vcpu_ioctl_get_mpstate(struct kvm_vcpu *vcpu,
				    struct kvm_mp_state *mp_state)
{
384
	if (vcpu->arch.power_off)
385 386 387 388 389
		mp_state->mp_state = KVM_MP_STATE_STOPPED;
	else
		mp_state->mp_state = KVM_MP_STATE_RUNNABLE;

	return 0;
390 391 392 393 394
}

int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu *vcpu,
				    struct kvm_mp_state *mp_state)
{
395 396
	switch (mp_state->mp_state) {
	case KVM_MP_STATE_RUNNABLE:
397
		vcpu->arch.power_off = false;
398 399
		break;
	case KVM_MP_STATE_STOPPED:
A
Andrew Jones 已提交
400
		vcpu_power_off(vcpu);
401 402 403 404 405 406
		break;
	default:
		return -EINVAL;
	}

	return 0;
407 408
}

409 410 411 412 413 414 415
/**
 * 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.
 */
416 417
int kvm_arch_vcpu_runnable(struct kvm_vcpu *v)
{
418
	return ((!!v->arch.irq_lines || kvm_vgic_vcpu_pending_irq(v))
419
		&& !v->arch.power_off && !v->arch.pause);
420 421
}

422 423
bool kvm_arch_vcpu_in_kernel(struct kvm_vcpu *vcpu)
{
424
	return vcpu_mode_priv(vcpu);
425 426
}

427 428 429 430 431 432 433
/* Just ensure a guest exit from a particular CPU */
static void exit_vm_noop(void *info)
{
}

void force_vm_exit(const cpumask_t *mask)
{
434
	preempt_disable();
435
	smp_call_function_many(mask, exit_vm_noop, NULL, true);
436
	preempt_enable();
437 438 439 440
}

/**
 * need_new_vmid_gen - check that the VMID is still valid
A
Andrea Gelmini 已提交
441
 * @kvm: The VM's VMID to check
442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505
 *
 * return true if there is a new generation of VMIDs being used
 *
 * The hardware supports only 256 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.
 */
static bool need_new_vmid_gen(struct kvm *kvm)
{
	return unlikely(kvm->arch.vmid_gen != atomic64_read(&kvm_vmid_gen));
}

/**
 * update_vttbr - Update the VTTBR with a valid VMID before the guest runs
 * @kvm	The guest that we are about to run
 *
 * Called from kvm_arch_vcpu_ioctl_run before entering the guest to ensure the
 * VM has a valid VMID, otherwise assigns a new one and flushes corresponding
 * caches and TLBs.
 */
static void update_vttbr(struct kvm *kvm)
{
	phys_addr_t pgd_phys;
	u64 vmid;

	if (!need_new_vmid_gen(kvm))
		return;

	spin_lock(&kvm_vmid_lock);

	/*
	 * 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.
	 */
	if (!need_new_vmid_gen(kvm)) {
		spin_unlock(&kvm_vmid_lock);
		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);
	}

	kvm->arch.vmid_gen = atomic64_read(&kvm_vmid_gen);
	kvm->arch.vmid = kvm_next_vmid;
	kvm_next_vmid++;
506
	kvm_next_vmid &= (1 << kvm_vmid_bits) - 1;
507 508

	/* update vttbr to be used with the new vmid */
509
	pgd_phys = virt_to_phys(kvm->arch.pgd);
510
	BUG_ON(pgd_phys & ~VTTBR_BADDR_MASK);
511
	vmid = ((u64)(kvm->arch.vmid) << VTTBR_VMID_SHIFT) & VTTBR_VMID_MASK(kvm_vmid_bits);
512
	kvm->arch.vttbr = kvm_phys_to_vttbr(pgd_phys) | vmid;
513 514 515 516 517 518

	spin_unlock(&kvm_vmid_lock);
}

static int kvm_vcpu_first_run_init(struct kvm_vcpu *vcpu)
{
519
	struct kvm *kvm = vcpu->kvm;
520
	int ret = 0;
521

522 523 524 525
	if (likely(vcpu->arch.has_run_once))
		return 0;

	vcpu->arch.has_run_once = true;
526

527
	/*
528 529
	 * Map the VGIC hardware resources before running a vcpu the first
	 * time on this VM.
530
	 */
531
	if (unlikely(irqchip_in_kernel(kvm) && !vgic_ready(kvm))) {
532
		ret = kvm_vgic_map_resources(kvm);
533 534 535 536
		if (ret)
			return ret;
	}

537
	ret = kvm_timer_enable(vcpu);
538 539 540 541
	if (ret)
		return ret;

	ret = kvm_arm_pmu_v3_enable(vcpu);
542

543
	return ret;
544 545
}

546 547 548 549 550
bool kvm_arch_intc_initialized(struct kvm *kvm)
{
	return vgic_initialized(kvm);
}

551
void kvm_arm_halt_guest(struct kvm *kvm)
552 553 554 555 556 557
{
	int i;
	struct kvm_vcpu *vcpu;

	kvm_for_each_vcpu(i, vcpu, kvm)
		vcpu->arch.pause = true;
558
	kvm_make_all_cpus_request(kvm, KVM_REQ_SLEEP);
559 560
}

561
void kvm_arm_resume_guest(struct kvm *kvm)
562 563 564 565
{
	int i;
	struct kvm_vcpu *vcpu;

566 567 568 569
	kvm_for_each_vcpu(i, vcpu, kvm) {
		vcpu->arch.pause = false;
		swake_up(kvm_arch_vcpu_wq(vcpu));
	}
570 571
}

572
static void vcpu_req_sleep(struct kvm_vcpu *vcpu)
573
{
574
	struct swait_queue_head *wq = kvm_arch_vcpu_wq(vcpu);
575

576
	swait_event_interruptible(*wq, ((!vcpu->arch.power_off) &&
577
				       (!vcpu->arch.pause)));
578

A
Andrew Jones 已提交
579
	if (vcpu->arch.power_off || vcpu->arch.pause) {
580
		/* Awaken to handle a signal, request we sleep again later. */
581
		kvm_make_request(KVM_REQ_SLEEP, vcpu);
582
	}
583 584
}

585 586 587 588 589
static int kvm_vcpu_initialized(struct kvm_vcpu *vcpu)
{
	return vcpu->arch.target >= 0;
}

590 591 592
static void check_vcpu_requests(struct kvm_vcpu *vcpu)
{
	if (kvm_request_pending(vcpu)) {
593 594
		if (kvm_check_request(KVM_REQ_SLEEP, vcpu))
			vcpu_req_sleep(vcpu);
595 596 597 598 599 600

		/*
		 * Clear IRQ_PENDING requests that were made to guarantee
		 * that a VCPU sees new virtual interrupts.
		 */
		kvm_check_request(KVM_REQ_IRQ_PENDING, vcpu);
601 602 603
	}
}

604 605 606 607 608 609 610 611 612 613 614
/**
 * 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.
 */
615 616
int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *run)
{
617 618
	int ret;

619
	if (unlikely(!kvm_vcpu_initialized(vcpu)))
620 621 622 623 624 625
		return -ENOEXEC;

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

C
Christoffer Dall 已提交
626 627 628 629
	if (run->exit_reason == KVM_EXIT_MMIO) {
		ret = kvm_handle_mmio_return(vcpu, vcpu->run);
		if (ret)
			return ret;
630 631 632
		if (kvm_arm_handle_step_debug(vcpu, vcpu->run))
			return 0;

C
Christoffer Dall 已提交
633 634
	}

635 636 637
	if (run->immediate_exit)
		return -EINTR;

638
	kvm_sigset_activate(vcpu);
639 640 641 642 643 644 645 646 647 648 649

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

		update_vttbr(vcpu->kvm);

650 651
		check_vcpu_requests(vcpu);

652 653 654 655 656
		/*
		 * Preparing the interrupts to be injected also
		 * involves poking the GIC, which must be done in a
		 * non-preemptible context.
		 */
657
		preempt_disable();
658

659 660 661
		/* Flush FP/SIMD state that can't survive guest entry/exit */
		kvm_fpsimd_flush_cpu_state();

662
		kvm_pmu_flush_hwstate(vcpu);
663

664 665
		local_irq_disable();

666 667
		kvm_vgic_flush_hwstate(vcpu);

668
		/*
669
		 * If we have a singal pending, or need to notify a userspace
670 671 672
		 * irqchip about timer or PMU level changes, then we exit (and
		 * update the timer level state in kvm_timer_update_run
		 * below).
673
		 */
674
		if (signal_pending(current) ||
675 676
		    kvm_timer_should_notify_user(vcpu) ||
		    kvm_pmu_should_notify_user(vcpu)) {
677 678 679 680
			ret = -EINTR;
			run->exit_reason = KVM_EXIT_INTR;
		}

681 682 683 684 685 686 687 688
		/*
		 * 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);

689
		if (ret <= 0 || need_new_vmid_gen(vcpu->kvm) ||
A
Andrew Jones 已提交
690
		    kvm_request_pending(vcpu)) {
691
			vcpu->mode = OUTSIDE_GUEST_MODE;
692
			kvm_pmu_sync_hwstate(vcpu);
693
			kvm_timer_sync_hwstate(vcpu);
694
			kvm_vgic_sync_hwstate(vcpu);
695
			local_irq_enable();
696
			preempt_enable();
697 698 699
			continue;
		}

700 701
		kvm_arm_setup_debug(vcpu);

702 703 704 705
		/**************************************************************
		 * Enter the guest
		 */
		trace_kvm_entry(*vcpu_pc(vcpu));
706
		guest_enter_irqoff();
707 708 709 710

		ret = kvm_call_hyp(__kvm_vcpu_run, vcpu);

		vcpu->mode = OUTSIDE_GUEST_MODE;
711
		vcpu->stat.exits++;
712 713 714 715
		/*
		 * Back from guest
		 *************************************************************/

716 717
		kvm_arm_clear_debug(vcpu);

718
		/*
719
		 * We must sync the PMU state before the vgic state so
720 721 722 723 724
		 * that the vgic can properly sample the updated state of the
		 * interrupt line.
		 */
		kvm_pmu_sync_hwstate(vcpu);

725 726 727 728 729
		/*
		 * Sync the vgic state before syncing the timer state because
		 * the timer code needs to know if the virtual timer
		 * interrupts are active.
		 */
730 731
		kvm_vgic_sync_hwstate(vcpu);

732 733 734 735 736 737 738
		/*
		 * Sync the timer hardware state before enabling interrupts as
		 * we don't want vtimer interrupts to race with syncing the
		 * timer virtual interrupt state.
		 */
		kvm_timer_sync_hwstate(vcpu);

739 740 741 742 743 744 745 746 747 748 749 750 751
		/*
		 * 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();

		/*
752
		 * We do local_irq_enable() before calling guest_exit() so
753 754
		 * that if a timer interrupt hits while running the guest we
		 * account that tick as being spent in the guest.  We enable
755
		 * preemption after calling guest_exit() so that if we get
756 757 758
		 * preempted we make sure ticks after that is not counted as
		 * guest time.
		 */
759
		guest_exit();
760
		trace_kvm_exit(ret, kvm_vcpu_trap_get_class(vcpu), *vcpu_pc(vcpu));
761

762 763
		preempt_enable();

764 765 766
		ret = handle_exit(vcpu, run, ret);
	}

767
	/* Tell userspace about in-kernel device output levels */
768 769 770 771
	if (unlikely(!irqchip_in_kernel(vcpu->kvm))) {
		kvm_timer_update_run(vcpu);
		kvm_pmu_update_run(vcpu);
	}
772

773 774
	kvm_sigset_deactivate(vcpu);

775
	return ret;
776 777
}

778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805
static int vcpu_interrupt_line(struct kvm_vcpu *vcpu, int number, bool level)
{
	int bit_index;
	bool set;
	unsigned long *ptr;

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

	ptr = (unsigned long *)&vcpu->arch.irq_lines;
	if (level)
		set = test_and_set_bit(bit_index, ptr);
	else
		set = test_and_clear_bit(bit_index, ptr);

	/*
	 * 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.
	 */
806
	kvm_make_request(KVM_REQ_IRQ_PENDING, vcpu);
807 808 809 810 811
	kvm_vcpu_kick(vcpu);

	return 0;
}

812 813
int kvm_vm_ioctl_irq_line(struct kvm *kvm, struct kvm_irq_level *irq_level,
			  bool line_status)
814 815 816 817 818 819 820 821 822 823 824 825 826
{
	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);

827 828 829 830
	switch (irq_type) {
	case KVM_ARM_IRQ_TYPE_CPU:
		if (irqchip_in_kernel(kvm))
			return -ENXIO;
831

832 833
		if (vcpu_idx >= nrcpus)
			return -EINVAL;
834

835 836 837
		vcpu = kvm_get_vcpu(kvm, vcpu_idx);
		if (!vcpu)
			return -EINVAL;
838

839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855
		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;
856

857
		return kvm_vgic_inject_irq(kvm, vcpu->vcpu_id, irq_num, level, NULL);
858 859 860 861
	case KVM_ARM_IRQ_TYPE_SPI:
		if (!irqchip_in_kernel(kvm))
			return -ENXIO;

862
		if (irq_num < VGIC_NR_PRIVATE_IRQS)
863 864
			return -EINVAL;

865
		return kvm_vgic_inject_irq(kvm, 0, irq_num, level, NULL);
866 867 868
	}

	return -EINVAL;
869 870
}

871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912
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);
}


913 914 915 916 917 918 919 920 921
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;

922 923 924 925 926 927 928
	/*
	 * 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);

929 930
	vcpu_reset_hcr(vcpu);

931
	/*
932
	 * Handle the "start in power-off" case.
933
	 */
934
	if (test_bit(KVM_ARM_VCPU_POWER_OFF, vcpu->arch.features))
A
Andrew Jones 已提交
935
		vcpu_power_off(vcpu);
936
	else
937
		vcpu->arch.power_off = false;
938 939 940 941

	return 0;
}

942 943 944 945 946 947 948
static int kvm_arm_vcpu_set_attr(struct kvm_vcpu *vcpu,
				 struct kvm_device_attr *attr)
{
	int ret = -ENXIO;

	switch (attr->group) {
	default:
949
		ret = kvm_arm_vcpu_arch_set_attr(vcpu, attr);
950 951 952 953 954 955 956 957 958 959 960 961 962
		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:
963
		ret = kvm_arm_vcpu_arch_get_attr(vcpu, attr);
964 965 966 967 968 969 970 971 972 973 974 975 976
		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:
977
		ret = kvm_arm_vcpu_arch_has_attr(vcpu, attr);
978 979 980 981 982 983
		break;
	}

	return ret;
}

984 985 986 987 988
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;
989
	struct kvm_device_attr attr;
990 991 992 993 994 995 996 997

	switch (ioctl) {
	case KVM_ARM_VCPU_INIT: {
		struct kvm_vcpu_init init;

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

998
		return kvm_arch_vcpu_ioctl_vcpu_init(vcpu, &init);
999 1000 1001 1002
	}
	case KVM_SET_ONE_REG:
	case KVM_GET_ONE_REG: {
		struct kvm_one_reg reg;
1003 1004 1005 1006

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

1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018
		if (copy_from_user(&reg, argp, sizeof(reg)))
			return -EFAULT;
		if (ioctl == KVM_SET_ONE_REG)
			return kvm_arm_set_reg(vcpu, &reg);
		else
			return kvm_arm_get_reg(vcpu, &reg);
	}
	case KVM_GET_REG_LIST: {
		struct kvm_reg_list __user *user_list = argp;
		struct kvm_reg_list reg_list;
		unsigned n;

1019 1020 1021
		if (unlikely(!kvm_vcpu_initialized(vcpu)))
			return -ENOEXEC;

1022 1023 1024 1025 1026 1027 1028 1029 1030 1031
		if (copy_from_user(&reg_list, user_list, sizeof(reg_list)))
			return -EFAULT;
		n = reg_list.n;
		reg_list.n = kvm_arm_num_regs(vcpu);
		if (copy_to_user(user_list, &reg_list, sizeof(reg_list)))
			return -EFAULT;
		if (n < reg_list.n)
			return -E2BIG;
		return kvm_arm_copy_reg_indices(vcpu, user_list->reg);
	}
1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046
	case KVM_SET_DEVICE_ATTR: {
		if (copy_from_user(&attr, argp, sizeof(attr)))
			return -EFAULT;
		return kvm_arm_vcpu_set_attr(vcpu, &attr);
	}
	case KVM_GET_DEVICE_ATTR: {
		if (copy_from_user(&attr, argp, sizeof(attr)))
			return -EFAULT;
		return kvm_arm_vcpu_get_attr(vcpu, &attr);
	}
	case KVM_HAS_DEVICE_ATTR: {
		if (copy_from_user(&attr, argp, sizeof(attr)))
			return -EFAULT;
		return kvm_arm_vcpu_has_attr(vcpu, &attr);
	}
1047 1048 1049 1050 1051
	default:
		return -EINVAL;
	}
}

1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070
/**
 * 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.
 */
1071 1072
int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm, struct kvm_dirty_log *log)
{
1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084
	bool is_dirty = false;
	int r;

	mutex_lock(&kvm->slots_lock);

	r = kvm_get_dirty_log_protect(kvm, log, &is_dirty);

	if (is_dirty)
		kvm_flush_remote_tlbs(kvm);

	mutex_unlock(&kvm->slots_lock);
	return r;
1085 1086
}

1087 1088 1089
static int kvm_vm_ioctl_set_device_addr(struct kvm *kvm,
					struct kvm_arm_device_addr *dev_addr)
{
1090 1091 1092 1093 1094 1095 1096 1097 1098
	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:
1099 1100
		if (!vgic_present)
			return -ENXIO;
1101
		return kvm_vgic_addr(kvm, type, &dev_addr->addr, true);
1102 1103 1104
	default:
		return -ENODEV;
	}
1105 1106
}

1107 1108 1109
long kvm_arch_vm_ioctl(struct file *filp,
		       unsigned int ioctl, unsigned long arg)
{
1110 1111 1112 1113
	struct kvm *kvm = filp->private_data;
	void __user *argp = (void __user *)arg;

	switch (ioctl) {
1114
	case KVM_CREATE_IRQCHIP: {
1115
		int ret;
1116 1117
		if (!vgic_present)
			return -ENXIO;
1118 1119 1120 1121
		mutex_lock(&kvm->lock);
		ret = kvm_vgic_create(kvm, KVM_DEV_TYPE_ARM_VGIC_V2);
		mutex_unlock(&kvm->lock);
		return ret;
1122
	}
1123 1124 1125 1126 1127 1128 1129
	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);
	}
1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142
	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;
	}
1143 1144 1145
	default:
		return -EINVAL;
	}
1146 1147
}

1148
static void cpu_init_hyp_mode(void *dummy)
1149
{
1150
	phys_addr_t pgd_ptr;
1151 1152 1153 1154 1155
	unsigned long hyp_stack_ptr;
	unsigned long stack_page;
	unsigned long vector_ptr;

	/* Switch from the HYP stub to our own HYP init vector */
1156
	__hyp_set_vectors(kvm_get_idmap_vector());
1157

1158
	pgd_ptr = kvm_mmu_get_httbr();
1159
	stack_page = __this_cpu_read(kvm_arm_hyp_stack_page);
1160
	hyp_stack_ptr = stack_page + PAGE_SIZE;
1161
	vector_ptr = (unsigned long)kvm_ksym_ref(__kvm_hyp_vector);
1162

M
Marc Zyngier 已提交
1163
	__cpu_init_hyp_mode(pgd_ptr, hyp_stack_ptr, vector_ptr);
1164
	__cpu_init_stage2();
1165 1166

	kvm_arm_init_debug();
1167 1168
}

1169 1170 1171 1172 1173 1174
static void cpu_hyp_reset(void)
{
	if (!is_kernel_in_hyp_mode())
		__hyp_reset_vectors();
}

1175 1176
static void cpu_hyp_reinit(void)
{
1177 1178
	cpu_hyp_reset();

1179 1180
	if (is_kernel_in_hyp_mode()) {
		/*
1181
		 * __cpu_init_stage2() is safe to call even if the PM
1182 1183
		 * event was cancelled before the CPU was reset.
		 */
1184
		__cpu_init_stage2();
1185
		kvm_timer_init_vhe();
1186
	} else {
1187
		cpu_init_hyp_mode(NULL);
1188
	}
1189 1190 1191

	if (vgic_present)
		kvm_vgic_init_cpu_hardware();
1192 1193
}

1194 1195 1196
static void _kvm_arch_hardware_enable(void *discard)
{
	if (!__this_cpu_read(kvm_arm_hardware_enabled)) {
1197
		cpu_hyp_reinit();
1198
		__this_cpu_write(kvm_arm_hardware_enabled, 1);
1199
	}
1200
}
1201

1202 1203 1204 1205
int kvm_arch_hardware_enable(void)
{
	_kvm_arch_hardware_enable(NULL);
	return 0;
1206 1207
}

1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219
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);
}
1220

1221 1222 1223 1224 1225
#ifdef CONFIG_CPU_PM
static int hyp_init_cpu_pm_notifier(struct notifier_block *self,
				    unsigned long cmd,
				    void *v)
{
1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240
	/*
	 * 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();

1241
		return NOTIFY_OK;
1242 1243 1244 1245
	case CPU_PM_EXIT:
		if (__this_cpu_read(kvm_arm_hardware_enabled))
			/* The hardware was enabled before suspend. */
			cpu_hyp_reinit();
1246

1247 1248 1249 1250 1251
		return NOTIFY_OK;

	default:
		return NOTIFY_DONE;
	}
1252 1253 1254 1255 1256 1257 1258 1259 1260 1261
}

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);
}
1262 1263 1264 1265
static void __init hyp_cpu_pm_exit(void)
{
	cpu_pm_unregister_notifier(&hyp_init_cpu_pm_nb);
}
1266 1267 1268 1269
#else
static inline void hyp_cpu_pm_init(void)
{
}
1270 1271 1272
static inline void hyp_cpu_pm_exit(void)
{
}
1273 1274
#endif

1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287
static void teardown_common_resources(void)
{
	free_percpu(kvm_host_cpu_state);
}

static int init_common_resources(void)
{
	kvm_host_cpu_state = alloc_percpu(kvm_cpu_context_t);
	if (!kvm_host_cpu_state) {
		kvm_err("Cannot allocate host CPU state\n");
		return -ENOMEM;
	}

1288 1289 1290 1291
	/* set size of VMID supported by CPU */
	kvm_vmid_bits = kvm_get_vmid_bits();
	kvm_info("%d-bit VMID\n", kvm_vmid_bits);

1292 1293 1294 1295 1296
	return 0;
}

static int init_subsystems(void)
{
1297
	int err = 0;
1298

1299
	/*
1300
	 * Enable hardware so that subsystem initialisation can access EL2.
1301
	 */
1302
	on_each_cpu(_kvm_arch_hardware_enable, NULL, 1);
1303 1304 1305 1306 1307 1308

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

1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319
	/*
	 * 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;
1320
		err = 0;
1321 1322
		break;
	default:
1323
		goto out;
1324 1325 1326 1327 1328 1329 1330
	}

	/*
	 * Init HYP architected timer support
	 */
	err = kvm_timer_hyp_init();
	if (err)
1331
		goto out;
1332 1333 1334 1335

	kvm_perf_init();
	kvm_coproc_table_init();

1336 1337 1338 1339
out:
	on_each_cpu(_kvm_arch_hardware_disable, NULL, 1);

	return err;
1340 1341 1342 1343 1344 1345 1346 1347 1348
}

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));
1349
	hyp_cpu_pm_exit();
1350 1351
}

1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375
/**
 * 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;
1376
			goto out_err;
1377 1378 1379 1380 1381 1382 1383 1384
		}

		per_cpu(kvm_arm_hyp_stack_page, cpu) = stack_page;
	}

	/*
	 * Map the Hyp-code called directly from the host
	 */
1385
	err = create_hyp_mappings(kvm_ksym_ref(__hyp_text_start),
1386
				  kvm_ksym_ref(__hyp_text_end), PAGE_HYP_EXEC);
1387 1388
	if (err) {
		kvm_err("Cannot map world-switch code\n");
1389
		goto out_err;
1390 1391
	}

1392
	err = create_hyp_mappings(kvm_ksym_ref(__start_rodata),
1393
				  kvm_ksym_ref(__end_rodata), PAGE_HYP_RO);
1394 1395
	if (err) {
		kvm_err("Cannot map rodata section\n");
M
Marc Zyngier 已提交
1396 1397 1398 1399 1400 1401 1402
		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");
1403
		goto out_err;
1404 1405
	}

1406 1407 1408 1409 1410
	/*
	 * Map the Hyp stack pages
	 */
	for_each_possible_cpu(cpu) {
		char *stack_page = (char *)per_cpu(kvm_arm_hyp_stack_page, cpu);
1411 1412
		err = create_hyp_mappings(stack_page, stack_page + PAGE_SIZE,
					  PAGE_HYP);
1413 1414 1415

		if (err) {
			kvm_err("Cannot map hyp stack\n");
1416
			goto out_err;
1417 1418 1419 1420
		}
	}

	for_each_possible_cpu(cpu) {
1421
		kvm_cpu_context_t *cpu_ctxt;
1422

1423
		cpu_ctxt = per_cpu_ptr(kvm_host_cpu_state, cpu);
1424
		err = create_hyp_mappings(cpu_ctxt, cpu_ctxt + 1, PAGE_HYP);
1425 1426

		if (err) {
1427
			kvm_err("Cannot map host CPU state: %d\n", err);
1428
			goto out_err;
1429 1430 1431 1432
		}
	}

	return 0;
1433

1434
out_err:
1435
	teardown_hyp_mode();
1436 1437 1438 1439
	kvm_err("error initializing Hyp mode: %d\n", err);
	return err;
}

1440 1441 1442 1443 1444
static void check_kvm_target_cpu(void *ret)
{
	*(int *)ret = kvm_target_cpu();
}

1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457
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;
}

1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468
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);

1469 1470
	return kvm_vgic_v4_set_forwarding(irqfd->kvm, prod->irq,
					  &irqfd->irq_entry);
1471 1472 1473 1474 1475 1476 1477
}
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);

1478 1479
	kvm_vgic_v4_unset_forwarding(irqfd->kvm, prod->irq,
				     &irqfd->irq_entry);
1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497
}

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);
}

1498 1499 1500
/**
 * Initialize Hyp-mode and memory mappings on all CPUs.
 */
1501 1502
int kvm_arch_init(void *opaque)
{
1503
	int err;
1504
	int ret, cpu;
1505
	bool in_hyp_mode;
1506 1507

	if (!is_hyp_mode_available()) {
1508
		kvm_info("HYP mode not available\n");
1509 1510 1511
		return -ENODEV;
	}

1512 1513 1514 1515 1516 1517
	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;
		}
1518 1519
	}

1520
	err = init_common_resources();
1521
	if (err)
1522
		return err;
1523

1524 1525 1526
	in_hyp_mode = is_kernel_in_hyp_mode();

	if (!in_hyp_mode) {
1527
		err = init_hyp_mode();
1528 1529 1530
		if (err)
			goto out_err;
	}
1531

1532 1533 1534
	err = init_subsystems();
	if (err)
		goto out_hyp;
1535

1536 1537 1538 1539 1540
	if (in_hyp_mode)
		kvm_info("VHE mode initialized successfully\n");
	else
		kvm_info("Hyp mode initialized successfully\n");

1541
	return 0;
1542 1543

out_hyp:
1544 1545
	if (!in_hyp_mode)
		teardown_hyp_mode();
1546
out_err:
1547
	teardown_common_resources();
1548
	return err;
1549 1550 1551 1552 1553
}

/* NOP: Compiling as a module not supported */
void kvm_arch_exit(void)
{
1554
	kvm_perf_teardown();
1555 1556 1557 1558 1559 1560 1561 1562 1563
}

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

module_init(arm_init);