kvm-cpu.c

#include "kvm/kvm-cpu.h"

#include "kvm/symbol.h"
#include "kvm/util.h"
#include "kvm/kvm.h"
#include "kvm/virtio.h"
#include "kvm/mutex.h"
#include "kvm/barrier.h"

#include <sys/ioctl.h>
#include <sys/mman.h>
#include <sys/eventfd.h>
#include <signal.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>
#include <stdio.h>

extern __thread struct kvm_cpu *current_kvm_cpu;

int __attribute__((weak)) kvm_cpu__get_endianness(struct kvm_cpu *vcpu)
{
	return VIRTIO_ENDIAN_HOST;
}

void kvm_cpu__enable_singlestep(struct kvm_cpu *vcpu)
{
	struct kvm_guest_debug debug = {
		.control	= KVM_GUESTDBG_ENABLE | KVM_GUESTDBG_SINGLESTEP,
	};

	if (ioctl(vcpu->vcpu_fd, KVM_SET_GUEST_DEBUG, &debug) < 0)
		pr_warning("KVM_SET_GUEST_DEBUG failed");
}

void kvm_cpu__run(struct kvm_cpu *vcpu)
{
	int err;

	if (!vcpu->is_running)
		return;

	err = ioctl(vcpu->vcpu_fd, KVM_RUN, 0);
	if (err < 0 && (errno != EINTR && errno != EAGAIN))
		die_perror("KVM_RUN failed");
}

static void kvm_cpu_signal_handler(int signum)
{
	if (signum == SIGKVMEXIT) {
		if (current_kvm_cpu && current_kvm_cpu->is_running)
			current_kvm_cpu->is_running = false;
	} else if (signum == SIGKVMPAUSE) {
		if (current_kvm_cpu->paused)
			die("Pause signaled for already paused CPU\n");

		/* pause_lock is held by kvm__pause() */
		current_kvm_cpu->paused = 1;

		/*
		 * This is a blocking function and uses locks. It is safe
		 * to call it for this signal as a second pause event should
		 * not be send to this thread until it acquires and releases
		 * the pause_lock.
		 */
		kvm__notify_paused();
	}

	/* For SIGKVMTASK cpu->task is already set */
}

static void kvm_cpu__handle_coalesced_mmio(struct kvm_cpu *cpu)
{
	if (cpu->ring) {
		while (cpu->ring->first != cpu->ring->last) {
			struct kvm_coalesced_mmio *m;
			m = &cpu->ring->coalesced_mmio[cpu->ring->first];
			kvm_cpu__emulate_mmio(cpu,
					      m->phys_addr,
					      m->data,
					      m->len,
					      1);
			cpu->ring->first = (cpu->ring->first + 1) % KVM_COALESCED_MMIO_MAX;
		}
	}
}

static DEFINE_MUTEX(task_lock);
static int task_eventfd;

static void kvm_cpu__run_task(struct kvm_cpu *cpu)
{
	u64 inc = 1;

	pr_debug("Running task %p on cpu %lu", cpu->task, cpu->cpu_id);

	/* Make sure we see the store to cpu->task */
	rmb();
	cpu->task->func(cpu, cpu->task->data);

	/* Clear task before we signal completion */
	cpu->task = NULL;
	wmb();

	if (write(task_eventfd, &inc, sizeof(inc)) < 0)
		die("Failed notifying of completed task.");
}

void kvm_cpu__run_on_all_cpus(struct kvm *kvm, struct kvm_cpu_task *task)
{
	int i, done = 0;

	pr_debug("Running task %p on all cpus", task);

	mutex_lock(&task_lock);

	for (i = 0; i < kvm->nrcpus; i++) {
		if (kvm->cpus[i]->task) {
			/* Should never happen */
			die("CPU %d already has a task pending!", i);
		}

		kvm->cpus[i]->task = task;
		wmb();

		if (kvm->cpus[i] == current_kvm_cpu)
			kvm_cpu__run_task(current_kvm_cpu);
		else
			pthread_kill(kvm->cpus[i]->thread, SIGKVMTASK);
	}

	while (done < kvm->nrcpus) {
		u64 count;

		if (read(task_eventfd, &count, sizeof(count)) < 0)
			die("Failed reading task eventfd");

		done += count;
	}

	mutex_unlock(&task_lock);
}

int kvm_cpu__start(struct kvm_cpu *cpu)
{
	sigset_t sigset;

	sigemptyset(&sigset);
	sigaddset(&sigset, SIGALRM);

	pthread_sigmask(SIG_BLOCK, &sigset, NULL);

	signal(SIGKVMEXIT, kvm_cpu_signal_handler);
	signal(SIGKVMPAUSE, kvm_cpu_signal_handler);
	signal(SIGKVMTASK, kvm_cpu_signal_handler);

	if (cpu->kvm->cfg.single_step)
		kvm_cpu__enable_singlestep(cpu);

	while (cpu->is_running) {
		if (cpu->needs_nmi) {
			kvm_cpu__arch_nmi(cpu);
			cpu->needs_nmi = 0;
		}

		if (cpu->task)
			kvm_cpu__run_task(cpu);

		kvm_cpu__run(cpu);

		switch (cpu->kvm_run->exit_reason) {
		case KVM_EXIT_UNKNOWN:
			break;
		case KVM_EXIT_DEBUG:
			kvm_cpu__show_registers(cpu);
			kvm_cpu__show_code(cpu);
			break;
		case KVM_EXIT_IO: {
			bool ret;

			ret = kvm_cpu__emulate_io(cpu,
						  cpu->kvm_run->io.port,
						  (u8 *)cpu->kvm_run +
						  cpu->kvm_run->io.data_offset,
						  cpu->kvm_run->io.direction,
						  cpu->kvm_run->io.size,
						  cpu->kvm_run->io.count);

			if (!ret)
				goto panic_kvm;
			break;
		}
		case KVM_EXIT_MMIO: {
			bool ret;

			/*
			 * If we had MMIO exit, coalesced ring should be processed
			 * *before* processing the exit itself
			 */
			kvm_cpu__handle_coalesced_mmio(cpu);

			ret = kvm_cpu__emulate_mmio(cpu,
						    cpu->kvm_run->mmio.phys_addr,
						    cpu->kvm_run->mmio.data,
						    cpu->kvm_run->mmio.len,
						    cpu->kvm_run->mmio.is_write);

			if (!ret)
				goto panic_kvm;
			break;
		}
		case KVM_EXIT_INTR:
			if (cpu->is_running)
				break;
			goto exit_kvm;
		case KVM_EXIT_SHUTDOWN:
			goto exit_kvm;
		case KVM_EXIT_SYSTEM_EVENT:
			/*
			 * Print the type of system event and
			 * treat all system events as shutdown request.
			 */
			switch (cpu->kvm_run->system_event.type) {
			default:
				pr_warning("unknown system event type %d",
					   cpu->kvm_run->system_event.type);
				/* fall through for now */
			case KVM_SYSTEM_EVENT_RESET:
				/* Fall through for now */
			case KVM_SYSTEM_EVENT_SHUTDOWN:
				/*
				 * Ensure that all VCPUs are torn down,
				 * regardless of which CPU generated the event.
				 */
				kvm__reboot(cpu->kvm);
				goto exit_kvm;
			};
			break;
		default: {
			bool ret;

			ret = kvm_cpu__handle_exit(cpu);
			if (!ret)
				goto panic_kvm;
			break;
		}
		}
		kvm_cpu__handle_coalesced_mmio(cpu);
	}

exit_kvm:
	return 0;

panic_kvm:
	return 1;
}

int kvm_cpu__init(struct kvm *kvm)
{
	int max_cpus, recommended_cpus, i;

	max_cpus = kvm__max_cpus(kvm);
	recommended_cpus = kvm__recommended_cpus(kvm);

	if (kvm->cfg.nrcpus > max_cpus) {
		pr_warning("Limiting the number of CPUs to %d", max_cpus);
		kvm->cfg.nrcpus = max_cpus;
	} else if (kvm->cfg.nrcpus > recommended_cpus) {
		pr_warning("The maximum recommended amount of VCPUs is %d",
			   recommended_cpus);
	}

	kvm->nrcpus = kvm->cfg.nrcpus;

	task_eventfd = eventfd(0, 0);
	if (task_eventfd < 0) {
		pr_err("Couldn't create task_eventfd");
		return task_eventfd;
	}

	/* Alloc one pointer too many, so array ends up 0-terminated */
	kvm->cpus = calloc(kvm->nrcpus + 1, sizeof(void *));
	if (!kvm->cpus) {
		pr_err("Couldn't allocate array for %d CPUs", kvm->nrcpus);
		return -ENOMEM;
	}

	for (i = 0; i < kvm->nrcpus; i++) {
		kvm->cpus[i] = kvm_cpu__arch_init(kvm, i);
		if (!kvm->cpus[i]) {
			pr_err("unable to initialize KVM VCPU");
			goto fail_alloc;
		}
		kvm_cpu__reset_vcpu(kvm->cpus[i]);
	}

	return 0;

fail_alloc:
	for (i = 0; i < kvm->nrcpus; i++)
		free(kvm->cpus[i]);
	return -ENOMEM;
}
base_init(kvm_cpu__init);

int kvm_cpu__exit(struct kvm *kvm)
{
	int i, r;
	void *ret = NULL;

	kvm_cpu__delete(kvm->cpus[0]);
	kvm->cpus[0] = NULL;

	kvm__pause(kvm);
	for (i = 1; i < kvm->nrcpus; i++) {
		if (kvm->cpus[i]->is_running) {
			pthread_kill(kvm->cpus[i]->thread, SIGKVMEXIT);
			if (pthread_join(kvm->cpus[i]->thread, &ret) != 0)
				die("pthread_join");
			kvm_cpu__delete(kvm->cpus[i]);
		}
		if (ret == NULL)
			r = 0;
	}
	kvm__continue(kvm);

	free(kvm->cpus);

	kvm->nrcpus = 0;

	close(task_eventfd);

	return r;
}