arm.c 35.2 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
#include <kvm/arm_psci.h>
35 36 37 38

#define CREATE_TRACE_POINTS
#include "trace.h"

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

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

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

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

77 78
DEFINE_STATIC_KEY_FALSE(userspace_irqchip_in_use);

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)
{
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)
{
298 299 300
	if (vcpu->arch.has_run_once && unlikely(!irqchip_in_kernel(vcpu->kvm)))
		static_branch_dec(&userspace_irqchip_in_use);

301
	kvm_mmu_free_memory_caches(vcpu);
302
	kvm_timer_vcpu_terminate(vcpu);
303
	kvm_pmu_vcpu_destroy(vcpu);
304
	kvm_vcpu_uninit(vcpu);
305
	kmem_cache_free(kvm_vcpu_cache, vcpu);
306 307 308 309 310 311 312 313 314
}

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

int kvm_cpu_has_pending_timer(struct kvm_vcpu *vcpu)
{
315
	return kvm_timer_is_pending(vcpu);
316 317
}

318 319 320
void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu)
{
	kvm_timer_schedule(vcpu);
321
	kvm_vgic_v4_enable_doorbell(vcpu);
322 323 324 325 326
}

void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu)
{
	kvm_timer_unschedule(vcpu);
327
	kvm_vgic_v4_disable_doorbell(vcpu);
328 329
}

330 331
int kvm_arch_vcpu_init(struct kvm_vcpu *vcpu)
{
332 333
	/* Force users to call KVM_ARM_VCPU_INIT */
	vcpu->arch.target = -1;
334
	bitmap_zero(vcpu->arch.features, KVM_VCPU_MAX_FEATURES);
335

336 337 338
	/* Set up the timer */
	kvm_timer_vcpu_init(vcpu);

339 340
	kvm_arm_reset_debug_ptr(vcpu);

341
	return kvm_vgic_vcpu_init(vcpu);
342 343 344 345
}

void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
{
346 347 348 349 350 351 352 353 354 355 356 357 358
	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;
	}

359
	vcpu->cpu = cpu;
360
	vcpu->arch.host_cpu_context = this_cpu_ptr(&kvm_host_cpu_state);
361

362
	kvm_arm_set_running_vcpu(vcpu);
363
	kvm_vgic_load(vcpu);
364
	kvm_timer_vcpu_load(vcpu);
365 366 367 368
}

void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu)
{
369
	kvm_timer_vcpu_put(vcpu);
370 371
	kvm_vgic_put(vcpu);

372 373
	vcpu->cpu = -1;

374
	kvm_arm_set_running_vcpu(NULL);
375 376
}

A
Andrew Jones 已提交
377 378 379
static void vcpu_power_off(struct kvm_vcpu *vcpu)
{
	vcpu->arch.power_off = true;
380
	kvm_make_request(KVM_REQ_SLEEP, vcpu);
A
Andrew Jones 已提交
381 382 383
	kvm_vcpu_kick(vcpu);
}

384 385 386
int kvm_arch_vcpu_ioctl_get_mpstate(struct kvm_vcpu *vcpu,
				    struct kvm_mp_state *mp_state)
{
387
	if (vcpu->arch.power_off)
388 389 390 391 392
		mp_state->mp_state = KVM_MP_STATE_STOPPED;
	else
		mp_state->mp_state = KVM_MP_STATE_RUNNABLE;

	return 0;
393 394 395 396 397
}

int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu *vcpu,
				    struct kvm_mp_state *mp_state)
{
398 399
	int ret = 0;

400 401
	switch (mp_state->mp_state) {
	case KVM_MP_STATE_RUNNABLE:
402
		vcpu->arch.power_off = false;
403 404
		break;
	case KVM_MP_STATE_STOPPED:
A
Andrew Jones 已提交
405
		vcpu_power_off(vcpu);
406 407
		break;
	default:
408
		ret = -EINVAL;
409 410
	}

411
	return ret;
412 413
}

414 415 416 417 418 419 420
/**
 * 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.
 */
421 422
int kvm_arch_vcpu_runnable(struct kvm_vcpu *v)
{
423
	return ((!!v->arch.irq_lines || kvm_vgic_vcpu_pending_irq(v))
424
		&& !v->arch.power_off && !v->arch.pause);
425 426
}

427 428
bool kvm_arch_vcpu_in_kernel(struct kvm_vcpu *vcpu)
{
429
	return vcpu_mode_priv(vcpu);
430 431
}

432 433 434 435 436 437 438
/* Just ensure a guest exit from a particular CPU */
static void exit_vm_noop(void *info)
{
}

void force_vm_exit(const cpumask_t *mask)
{
439
	preempt_disable();
440
	smp_call_function_many(mask, exit_vm_noop, NULL, true);
441
	preempt_enable();
442 443 444 445
}

/**
 * need_new_vmid_gen - check that the VMID is still valid
A
Andrea Gelmini 已提交
446
 * @kvm: The VM's VMID to check
447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510
 *
 * 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++;
511
	kvm_next_vmid &= (1 << kvm_vmid_bits) - 1;
512 513

	/* update vttbr to be used with the new vmid */
514
	pgd_phys = virt_to_phys(kvm->arch.pgd);
515
	BUG_ON(pgd_phys & ~VTTBR_BADDR_MASK);
516
	vmid = ((u64)(kvm->arch.vmid) << VTTBR_VMID_SHIFT) & VTTBR_VMID_MASK(kvm_vmid_bits);
517
	kvm->arch.vttbr = kvm_phys_to_vttbr(pgd_phys) | vmid;
518 519 520 521 522 523

	spin_unlock(&kvm_vmid_lock);
}

static int kvm_vcpu_first_run_init(struct kvm_vcpu *vcpu)
{
524
	struct kvm *kvm = vcpu->kvm;
525
	int ret = 0;
526

527 528 529 530
	if (likely(vcpu->arch.has_run_once))
		return 0;

	vcpu->arch.has_run_once = true;
531

532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547
	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);
548 549
	}

550
	ret = kvm_timer_enable(vcpu);
551 552 553 554
	if (ret)
		return ret;

	ret = kvm_arm_pmu_v3_enable(vcpu);
555

556
	return ret;
557 558
}

559 560 561 562 563
bool kvm_arch_intc_initialized(struct kvm *kvm)
{
	return vgic_initialized(kvm);
}

564
void kvm_arm_halt_guest(struct kvm *kvm)
565 566 567 568 569 570
{
	int i;
	struct kvm_vcpu *vcpu;

	kvm_for_each_vcpu(i, vcpu, kvm)
		vcpu->arch.pause = true;
571
	kvm_make_all_cpus_request(kvm, KVM_REQ_SLEEP);
572 573
}

574
void kvm_arm_resume_guest(struct kvm *kvm)
575 576 577 578
{
	int i;
	struct kvm_vcpu *vcpu;

579 580 581 582
	kvm_for_each_vcpu(i, vcpu, kvm) {
		vcpu->arch.pause = false;
		swake_up(kvm_arch_vcpu_wq(vcpu));
	}
583 584
}

585
static void vcpu_req_sleep(struct kvm_vcpu *vcpu)
586
{
587
	struct swait_queue_head *wq = kvm_arch_vcpu_wq(vcpu);
588

589
	swait_event_interruptible(*wq, ((!vcpu->arch.power_off) &&
590
				       (!vcpu->arch.pause)));
591

A
Andrew Jones 已提交
592
	if (vcpu->arch.power_off || vcpu->arch.pause) {
593
		/* Awaken to handle a signal, request we sleep again later. */
594
		kvm_make_request(KVM_REQ_SLEEP, vcpu);
595
	}
596 597
}

598 599 600 601 602
static int kvm_vcpu_initialized(struct kvm_vcpu *vcpu)
{
	return vcpu->arch.target >= 0;
}

603 604 605
static void check_vcpu_requests(struct kvm_vcpu *vcpu)
{
	if (kvm_request_pending(vcpu)) {
606 607
		if (kvm_check_request(KVM_REQ_SLEEP, vcpu))
			vcpu_req_sleep(vcpu);
608 609 610 611 612 613

		/*
		 * Clear IRQ_PENDING requests that were made to guarantee
		 * that a VCPU sees new virtual interrupts.
		 */
		kvm_check_request(KVM_REQ_IRQ_PENDING, vcpu);
614 615 616
	}
}

617 618 619 620 621 622 623 624 625 626 627
/**
 * 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.
 */
628 629
int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *run)
{
630 631
	int ret;

632
	if (unlikely(!kvm_vcpu_initialized(vcpu)))
633 634 635 636
		return -ENOEXEC;

	ret = kvm_vcpu_first_run_init(vcpu);
	if (ret)
637
		return ret;
638

C
Christoffer Dall 已提交
639 640 641
	if (run->exit_reason == KVM_EXIT_MMIO) {
		ret = kvm_handle_mmio_return(vcpu, vcpu->run);
		if (ret)
642 643 644
			return ret;
		if (kvm_arm_handle_step_debug(vcpu, vcpu->run))
			return 0;
C
Christoffer Dall 已提交
645 646
	}

647 648 649 650
	if (run->immediate_exit)
		return -EINTR;

	vcpu_load(vcpu);
651

652
	kvm_sigset_activate(vcpu);
653 654 655 656 657 658 659 660 661 662 663

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

		update_vttbr(vcpu->kvm);

664 665
		check_vcpu_requests(vcpu);

666 667 668 669 670
		/*
		 * Preparing the interrupts to be injected also
		 * involves poking the GIC, which must be done in a
		 * non-preemptible context.
		 */
671
		preempt_disable();
672

673 674 675
		/* Flush FP/SIMD state that can't survive guest entry/exit */
		kvm_fpsimd_flush_cpu_state();

676
		kvm_pmu_flush_hwstate(vcpu);
677

678 679
		local_irq_disable();

680 681
		kvm_vgic_flush_hwstate(vcpu);

682
		/*
683 684
		 * Exit if we have a signal pending so that we can deliver the
		 * signal to user space.
685
		 */
686
		if (signal_pending(current)) {
687 688 689 690
			ret = -EINTR;
			run->exit_reason = KVM_EXIT_INTR;
		}

691 692 693 694 695 696 697 698 699 700 701 702 703 704 705
		/*
		 * 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;
			}
		}

706 707 708 709 710 711 712 713
		/*
		 * 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);

714
		if (ret <= 0 || need_new_vmid_gen(vcpu->kvm) ||
A
Andrew Jones 已提交
715
		    kvm_request_pending(vcpu)) {
716
			vcpu->mode = OUTSIDE_GUEST_MODE;
717
			kvm_pmu_sync_hwstate(vcpu);
718 719
			if (static_branch_unlikely(&userspace_irqchip_in_use))
				kvm_timer_sync_hwstate(vcpu);
720
			kvm_vgic_sync_hwstate(vcpu);
721
			local_irq_enable();
722
			preempt_enable();
723 724 725
			continue;
		}

726 727
		kvm_arm_setup_debug(vcpu);

728 729 730 731
		/**************************************************************
		 * Enter the guest
		 */
		trace_kvm_entry(*vcpu_pc(vcpu));
732
		guest_enter_irqoff();
733 734
		if (has_vhe())
			kvm_arm_vhe_guest_enter();
735 736 737

		ret = kvm_call_hyp(__kvm_vcpu_run, vcpu);

738 739
		if (has_vhe())
			kvm_arm_vhe_guest_exit();
740
		vcpu->mode = OUTSIDE_GUEST_MODE;
741
		vcpu->stat.exits++;
742 743 744 745
		/*
		 * Back from guest
		 *************************************************************/

746 747
		kvm_arm_clear_debug(vcpu);

748
		/*
749
		 * We must sync the PMU state before the vgic state so
750 751 752 753 754
		 * that the vgic can properly sample the updated state of the
		 * interrupt line.
		 */
		kvm_pmu_sync_hwstate(vcpu);

755 756 757 758 759
		/*
		 * Sync the vgic state before syncing the timer state because
		 * the timer code needs to know if the virtual timer
		 * interrupts are active.
		 */
760 761
		kvm_vgic_sync_hwstate(vcpu);

762 763 764 765 766
		/*
		 * Sync the timer hardware state before enabling interrupts as
		 * we don't want vtimer interrupts to race with syncing the
		 * timer virtual interrupt state.
		 */
767 768
		if (static_branch_unlikely(&userspace_irqchip_in_use))
			kvm_timer_sync_hwstate(vcpu);
769

770 771 772 773 774 775 776 777 778 779 780 781 782
		/*
		 * 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();

		/*
783
		 * We do local_irq_enable() before calling guest_exit() so
784 785
		 * that if a timer interrupt hits while running the guest we
		 * account that tick as being spent in the guest.  We enable
786
		 * preemption after calling guest_exit() so that if we get
787 788 789
		 * preempted we make sure ticks after that is not counted as
		 * guest time.
		 */
790
		guest_exit();
791
		trace_kvm_exit(ret, kvm_vcpu_trap_get_class(vcpu), *vcpu_pc(vcpu));
792

793 794 795
		/* Exit types that need handling before we can be preempted */
		handle_exit_early(vcpu, run, ret);

796 797
		preempt_enable();

798 799 800
		ret = handle_exit(vcpu, run, ret);
	}

801
	/* Tell userspace about in-kernel device output levels */
802 803 804 805
	if (unlikely(!irqchip_in_kernel(vcpu->kvm))) {
		kvm_timer_update_run(vcpu);
		kvm_pmu_update_run(vcpu);
	}
806

807 808
	kvm_sigset_deactivate(vcpu);

809
	vcpu_put(vcpu);
810
	return ret;
811 812
}

813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840
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.
	 */
841
	kvm_make_request(KVM_REQ_IRQ_PENDING, vcpu);
842 843 844 845 846
	kvm_vcpu_kick(vcpu);

	return 0;
}

847 848
int kvm_vm_ioctl_irq_line(struct kvm *kvm, struct kvm_irq_level *irq_level,
			  bool line_status)
849 850 851 852 853 854 855 856 857 858 859 860 861
{
	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);

862 863 864 865
	switch (irq_type) {
	case KVM_ARM_IRQ_TYPE_CPU:
		if (irqchip_in_kernel(kvm))
			return -ENXIO;
866

867 868
		if (vcpu_idx >= nrcpus)
			return -EINVAL;
869

870 871 872
		vcpu = kvm_get_vcpu(kvm, vcpu_idx);
		if (!vcpu)
			return -EINVAL;
873

874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890
		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;
891

892
		return kvm_vgic_inject_irq(kvm, vcpu->vcpu_id, irq_num, level, NULL);
893 894 895 896
	case KVM_ARM_IRQ_TYPE_SPI:
		if (!irqchip_in_kernel(kvm))
			return -ENXIO;

897
		if (irq_num < VGIC_NR_PRIVATE_IRQS)
898 899
			return -EINVAL;

900
		return kvm_vgic_inject_irq(kvm, 0, irq_num, level, NULL);
901 902 903
	}

	return -EINVAL;
904 905
}

906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947
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);
}


948 949 950 951 952 953 954 955 956
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;

957 958 959 960 961 962 963
	/*
	 * 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);

964 965
	vcpu_reset_hcr(vcpu);

966
	/*
967
	 * Handle the "start in power-off" case.
968
	 */
969
	if (test_bit(KVM_ARM_VCPU_POWER_OFF, vcpu->arch.features))
A
Andrew Jones 已提交
970
		vcpu_power_off(vcpu);
971
	else
972
		vcpu->arch.power_off = false;
973 974 975 976

	return 0;
}

977 978 979 980 981 982 983
static int kvm_arm_vcpu_set_attr(struct kvm_vcpu *vcpu,
				 struct kvm_device_attr *attr)
{
	int ret = -ENXIO;

	switch (attr->group) {
	default:
984
		ret = kvm_arm_vcpu_arch_set_attr(vcpu, attr);
985 986 987 988 989 990 991 992 993 994 995 996 997
		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:
998
		ret = kvm_arm_vcpu_arch_get_attr(vcpu, attr);
999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011
		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:
1012
		ret = kvm_arm_vcpu_arch_has_attr(vcpu, attr);
1013 1014 1015 1016 1017 1018
		break;
	}

	return ret;
}

1019 1020 1021 1022 1023
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;
1024
	struct kvm_device_attr attr;
1025 1026
	long r;

1027 1028 1029 1030
	switch (ioctl) {
	case KVM_ARM_VCPU_INIT: {
		struct kvm_vcpu_init init;

1031
		r = -EFAULT;
1032
		if (copy_from_user(&init, argp, sizeof(init)))
1033
			break;
1034

1035 1036
		r = kvm_arch_vcpu_ioctl_vcpu_init(vcpu, &init);
		break;
1037 1038 1039 1040
	}
	case KVM_SET_ONE_REG:
	case KVM_GET_ONE_REG: {
		struct kvm_one_reg reg;
1041

1042
		r = -ENOEXEC;
1043
		if (unlikely(!kvm_vcpu_initialized(vcpu)))
1044
			break;
1045

1046
		r = -EFAULT;
1047
		if (copy_from_user(&reg, argp, sizeof(reg)))
1048 1049
			break;

1050
		if (ioctl == KVM_SET_ONE_REG)
1051
			r = kvm_arm_set_reg(vcpu, &reg);
1052
		else
1053 1054
			r = kvm_arm_get_reg(vcpu, &reg);
		break;
1055 1056 1057 1058 1059 1060
	}
	case KVM_GET_REG_LIST: {
		struct kvm_reg_list __user *user_list = argp;
		struct kvm_reg_list reg_list;
		unsigned n;

1061
		r = -ENOEXEC;
1062
		if (unlikely(!kvm_vcpu_initialized(vcpu)))
1063
			break;
1064

1065
		r = -EFAULT;
1066
		if (copy_from_user(&reg_list, user_list, sizeof(reg_list)))
1067
			break;
1068 1069 1070
		n = reg_list.n;
		reg_list.n = kvm_arm_num_regs(vcpu);
		if (copy_to_user(user_list, &reg_list, sizeof(reg_list)))
1071 1072
			break;
		r = -E2BIG;
1073
		if (n < reg_list.n)
1074 1075 1076
			break;
		r = kvm_arm_copy_reg_indices(vcpu, user_list->reg);
		break;
1077
	}
1078
	case KVM_SET_DEVICE_ATTR: {
1079
		r = -EFAULT;
1080
		if (copy_from_user(&attr, argp, sizeof(attr)))
1081 1082 1083
			break;
		r = kvm_arm_vcpu_set_attr(vcpu, &attr);
		break;
1084 1085
	}
	case KVM_GET_DEVICE_ATTR: {
1086
		r = -EFAULT;
1087
		if (copy_from_user(&attr, argp, sizeof(attr)))
1088 1089 1090
			break;
		r = kvm_arm_vcpu_get_attr(vcpu, &attr);
		break;
1091 1092
	}
	case KVM_HAS_DEVICE_ATTR: {
1093
		r = -EFAULT;
1094
		if (copy_from_user(&attr, argp, sizeof(attr)))
1095 1096 1097
			break;
		r = kvm_arm_vcpu_has_attr(vcpu, &attr);
		break;
1098
	}
1099
	default:
1100
		r = -EINVAL;
1101
	}
1102 1103

	return r;
1104 1105
}

1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
/**
 * 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.
 */
1125 1126
int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm, struct kvm_dirty_log *log)
{
1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138
	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;
1139 1140
}

1141 1142 1143
static int kvm_vm_ioctl_set_device_addr(struct kvm *kvm,
					struct kvm_arm_device_addr *dev_addr)
{
1144 1145 1146 1147 1148 1149 1150 1151 1152
	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:
1153 1154
		if (!vgic_present)
			return -ENXIO;
1155
		return kvm_vgic_addr(kvm, type, &dev_addr->addr, true);
1156 1157 1158
	default:
		return -ENODEV;
	}
1159 1160
}

1161 1162 1163
long kvm_arch_vm_ioctl(struct file *filp,
		       unsigned int ioctl, unsigned long arg)
{
1164 1165 1166 1167
	struct kvm *kvm = filp->private_data;
	void __user *argp = (void __user *)arg;

	switch (ioctl) {
1168
	case KVM_CREATE_IRQCHIP: {
1169
		int ret;
1170 1171
		if (!vgic_present)
			return -ENXIO;
1172 1173 1174 1175
		mutex_lock(&kvm->lock);
		ret = kvm_vgic_create(kvm, KVM_DEV_TYPE_ARM_VGIC_V2);
		mutex_unlock(&kvm->lock);
		return ret;
1176
	}
1177 1178 1179 1180 1181 1182 1183
	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);
	}
1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196
	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;
	}
1197 1198 1199
	default:
		return -EINVAL;
	}
1200 1201
}

1202
static void cpu_init_hyp_mode(void *dummy)
1203
{
1204
	phys_addr_t pgd_ptr;
1205 1206 1207 1208 1209
	unsigned long hyp_stack_ptr;
	unsigned long stack_page;
	unsigned long vector_ptr;

	/* Switch from the HYP stub to our own HYP init vector */
1210
	__hyp_set_vectors(kvm_get_idmap_vector());
1211

1212
	pgd_ptr = kvm_mmu_get_httbr();
1213
	stack_page = __this_cpu_read(kvm_arm_hyp_stack_page);
1214
	hyp_stack_ptr = stack_page + PAGE_SIZE;
1215
	vector_ptr = (unsigned long)kvm_get_hyp_vector();
1216

M
Marc Zyngier 已提交
1217
	__cpu_init_hyp_mode(pgd_ptr, hyp_stack_ptr, vector_ptr);
1218
	__cpu_init_stage2();
1219 1220

	kvm_arm_init_debug();
1221 1222
}

1223 1224 1225 1226 1227 1228
static void cpu_hyp_reset(void)
{
	if (!is_kernel_in_hyp_mode())
		__hyp_reset_vectors();
}

1229 1230
static void cpu_hyp_reinit(void)
{
1231 1232
	cpu_hyp_reset();

1233 1234
	if (is_kernel_in_hyp_mode()) {
		/*
1235
		 * __cpu_init_stage2() is safe to call even if the PM
1236 1237
		 * event was cancelled before the CPU was reset.
		 */
1238
		__cpu_init_stage2();
1239
		kvm_timer_init_vhe();
1240
	} else {
1241
		cpu_init_hyp_mode(NULL);
1242
	}
1243 1244 1245

	if (vgic_present)
		kvm_vgic_init_cpu_hardware();
1246 1247
}

1248 1249 1250
static void _kvm_arch_hardware_enable(void *discard)
{
	if (!__this_cpu_read(kvm_arm_hardware_enabled)) {
1251
		cpu_hyp_reinit();
1252
		__this_cpu_write(kvm_arm_hardware_enabled, 1);
1253
	}
1254
}
1255

1256 1257 1258 1259
int kvm_arch_hardware_enable(void)
{
	_kvm_arch_hardware_enable(NULL);
	return 0;
1260 1261
}

1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273
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);
}
1274

1275 1276 1277 1278 1279
#ifdef CONFIG_CPU_PM
static int hyp_init_cpu_pm_notifier(struct notifier_block *self,
				    unsigned long cmd,
				    void *v)
{
1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294
	/*
	 * 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();

1295
		return NOTIFY_OK;
1296
	case CPU_PM_ENTER_FAILED:
1297 1298 1299 1300
	case CPU_PM_EXIT:
		if (__this_cpu_read(kvm_arm_hardware_enabled))
			/* The hardware was enabled before suspend. */
			cpu_hyp_reinit();
1301

1302 1303 1304 1305 1306
		return NOTIFY_OK;

	default:
		return NOTIFY_DONE;
	}
1307 1308 1309 1310 1311 1312 1313 1314 1315 1316
}

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);
}
1317 1318 1319 1320
static void __init hyp_cpu_pm_exit(void)
{
	cpu_pm_unregister_notifier(&hyp_init_cpu_pm_nb);
}
1321 1322 1323 1324
#else
static inline void hyp_cpu_pm_init(void)
{
}
1325 1326 1327
static inline void hyp_cpu_pm_exit(void)
{
}
1328 1329
#endif

1330 1331
static int init_common_resources(void)
{
1332 1333 1334 1335
	/* set size of VMID supported by CPU */
	kvm_vmid_bits = kvm_get_vmid_bits();
	kvm_info("%d-bit VMID\n", kvm_vmid_bits);

1336 1337 1338 1339 1340
	return 0;
}

static int init_subsystems(void)
{
1341
	int err = 0;
1342

1343
	/*
1344
	 * Enable hardware so that subsystem initialisation can access EL2.
1345
	 */
1346
	on_each_cpu(_kvm_arch_hardware_enable, NULL, 1);
1347 1348 1349 1350 1351 1352

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

1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363
	/*
	 * 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;
1364
		err = 0;
1365 1366
		break;
	default:
1367
		goto out;
1368 1369 1370 1371 1372
	}

	/*
	 * Init HYP architected timer support
	 */
1373
	err = kvm_timer_hyp_init(vgic_present);
1374
	if (err)
1375
		goto out;
1376 1377 1378 1379

	kvm_perf_init();
	kvm_coproc_table_init();

1380 1381 1382 1383
out:
	on_each_cpu(_kvm_arch_hardware_disable, NULL, 1);

	return err;
1384 1385 1386 1387 1388 1389 1390 1391 1392
}

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));
1393
	hyp_cpu_pm_exit();
1394 1395
}

1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419
/**
 * 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;
1420
			goto out_err;
1421 1422 1423 1424 1425 1426 1427 1428
		}

		per_cpu(kvm_arm_hyp_stack_page, cpu) = stack_page;
	}

	/*
	 * Map the Hyp-code called directly from the host
	 */
1429
	err = create_hyp_mappings(kvm_ksym_ref(__hyp_text_start),
1430
				  kvm_ksym_ref(__hyp_text_end), PAGE_HYP_EXEC);
1431 1432
	if (err) {
		kvm_err("Cannot map world-switch code\n");
1433
		goto out_err;
1434 1435
	}

1436
	err = create_hyp_mappings(kvm_ksym_ref(__start_rodata),
1437
				  kvm_ksym_ref(__end_rodata), PAGE_HYP_RO);
1438 1439
	if (err) {
		kvm_err("Cannot map rodata section\n");
M
Marc Zyngier 已提交
1440 1441 1442 1443 1444 1445 1446
		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");
1447
		goto out_err;
1448 1449
	}

1450 1451 1452 1453 1454 1455
	err = kvm_map_vectors();
	if (err) {
		kvm_err("Cannot map vectors\n");
		goto out_err;
	}

1456 1457 1458 1459 1460
	/*
	 * Map the Hyp stack pages
	 */
	for_each_possible_cpu(cpu) {
		char *stack_page = (char *)per_cpu(kvm_arm_hyp_stack_page, cpu);
1461 1462
		err = create_hyp_mappings(stack_page, stack_page + PAGE_SIZE,
					  PAGE_HYP);
1463 1464 1465

		if (err) {
			kvm_err("Cannot map hyp stack\n");
1466
			goto out_err;
1467 1468 1469 1470
		}
	}

	for_each_possible_cpu(cpu) {
1471
		kvm_cpu_context_t *cpu_ctxt;
1472

1473
		cpu_ctxt = per_cpu_ptr(&kvm_host_cpu_state, cpu);
1474
		err = create_hyp_mappings(cpu_ctxt, cpu_ctxt + 1, PAGE_HYP);
1475 1476

		if (err) {
1477
			kvm_err("Cannot map host CPU state: %d\n", err);
1478
			goto out_err;
1479 1480 1481 1482
		}
	}

	return 0;
1483

1484
out_err:
1485
	teardown_hyp_mode();
1486 1487 1488 1489
	kvm_err("error initializing Hyp mode: %d\n", err);
	return err;
}

1490 1491 1492 1493 1494
static void check_kvm_target_cpu(void *ret)
{
	*(int *)ret = kvm_target_cpu();
}

1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507
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;
}

1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518
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);

1519 1520
	return kvm_vgic_v4_set_forwarding(irqfd->kvm, prod->irq,
					  &irqfd->irq_entry);
1521 1522 1523 1524 1525 1526 1527
}
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);

1528 1529
	kvm_vgic_v4_unset_forwarding(irqfd->kvm, prod->irq,
				     &irqfd->irq_entry);
1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547
}

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);
}

1548 1549 1550
/**
 * Initialize Hyp-mode and memory mappings on all CPUs.
 */
1551 1552
int kvm_arch_init(void *opaque)
{
1553
	int err;
1554
	int ret, cpu;
1555
	bool in_hyp_mode;
1556 1557

	if (!is_hyp_mode_available()) {
1558
		kvm_info("HYP mode not available\n");
1559 1560 1561
		return -ENODEV;
	}

1562 1563 1564 1565 1566 1567
	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;
		}
1568 1569
	}

1570
	err = init_common_resources();
1571
	if (err)
1572
		return err;
1573

1574 1575 1576
	in_hyp_mode = is_kernel_in_hyp_mode();

	if (!in_hyp_mode) {
1577
		err = init_hyp_mode();
1578 1579 1580
		if (err)
			goto out_err;
	}
1581

1582 1583 1584
	err = init_subsystems();
	if (err)
		goto out_hyp;
1585

1586 1587 1588 1589 1590
	if (in_hyp_mode)
		kvm_info("VHE mode initialized successfully\n");
	else
		kvm_info("Hyp mode initialized successfully\n");

1591
	return 0;
1592 1593

out_hyp:
1594 1595
	if (!in_hyp_mode)
		teardown_hyp_mode();
1596 1597
out_err:
	return err;
1598 1599 1600 1601 1602
}

/* NOP: Compiling as a module not supported */
void kvm_arch_exit(void)
{
1603
	kvm_perf_teardown();
1604 1605 1606 1607 1608 1609 1610 1611 1612
}

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

module_init(arm_init);