kvm_main.c

来自「linux 内核源代码」· C语言 代码 · 共 2,825 行 · 第 1/5 页

	kvm_x86_ops->cache_regs(vcpu);
	function = vcpu->regs[VCPU_REGS_RAX];
	vcpu->regs[VCPU_REGS_RAX] = 0;
	vcpu->regs[VCPU_REGS_RBX] = 0;
	vcpu->regs[VCPU_REGS_RCX] = 0;
	vcpu->regs[VCPU_REGS_RDX] = 0;
	best = NULL;
	for (i = 0; i < vcpu->cpuid_nent; ++i) {
		e = &vcpu->cpuid_entries[i];
		if (e->function == function) {
			best = e;
			break;
		}
		/*
		 * Both basic or both extended?
		 */
		if (((e->function ^ function) & 0x80000000) == 0)
			if (!best || e->function > best->function)
				best = e;
	}
	if (best) {
		vcpu->regs[VCPU_REGS_RAX] = best->eax;
		vcpu->regs[VCPU_REGS_RBX] = best->ebx;
		vcpu->regs[VCPU_REGS_RCX] = best->ecx;
		vcpu->regs[VCPU_REGS_RDX] = best->edx;
	}
	kvm_x86_ops->decache_regs(vcpu);
	kvm_x86_ops->skip_emulated_instruction(vcpu);
}
EXPORT_SYMBOL_GPL(kvm_emulate_cpuid);

static int pio_copy_data(struct kvm_vcpu *vcpu)
{
	void *p = vcpu->pio_data;
	void *q;
	unsigned bytes;
	int nr_pages = vcpu->pio.guest_pages[1] ? 2 : 1;

	q = vmap(vcpu->pio.guest_pages, nr_pages, VM_READ|VM_WRITE,
		 PAGE_KERNEL);
	if (!q) {
		free_pio_guest_pages(vcpu);
		return -ENOMEM;
	}
	q += vcpu->pio.guest_page_offset;
	bytes = vcpu->pio.size * vcpu->pio.cur_count;
	if (vcpu->pio.in)
		memcpy(q, p, bytes);
	else
		memcpy(p, q, bytes);
	q -= vcpu->pio.guest_page_offset;
	vunmap(q);
	free_pio_guest_pages(vcpu);
	return 0;
}

static int complete_pio(struct kvm_vcpu *vcpu)
{
	struct kvm_pio_request *io = &vcpu->pio;
	long delta;
	int r;

	kvm_x86_ops->cache_regs(vcpu);

	if (!io->string) {
		if (io->in)
			memcpy(&vcpu->regs[VCPU_REGS_RAX], vcpu->pio_data,
			       io->size);
	} else {
		if (io->in) {
			r = pio_copy_data(vcpu);
			if (r) {
				kvm_x86_ops->cache_regs(vcpu);
				return r;
			}
		}

		delta = 1;
		if (io->rep) {
			delta *= io->cur_count;
			/*
			 * The size of the register should really depend on
			 * current address size.
			 */
			vcpu->regs[VCPU_REGS_RCX] -= delta;
		}
		if (io->down)
			delta = -delta;
		delta *= io->size;
		if (io->in)
			vcpu->regs[VCPU_REGS_RDI] += delta;
		else
			vcpu->regs[VCPU_REGS_RSI] += delta;
	}

	kvm_x86_ops->decache_regs(vcpu);

	io->count -= io->cur_count;
	io->cur_count = 0;

	return 0;
}

static void kernel_pio(struct kvm_io_device *pio_dev,
		       struct kvm_vcpu *vcpu,
		       void *pd)
{
	/* TODO: String I/O for in kernel device */

	mutex_lock(&vcpu->kvm->lock);
	if (vcpu->pio.in)
		kvm_iodevice_read(pio_dev, vcpu->pio.port,
				  vcpu->pio.size,
				  pd);
	else
		kvm_iodevice_write(pio_dev, vcpu->pio.port,
				   vcpu->pio.size,
				   pd);
	mutex_unlock(&vcpu->kvm->lock);
}

static void pio_string_write(struct kvm_io_device *pio_dev,
			     struct kvm_vcpu *vcpu)
{
	struct kvm_pio_request *io = &vcpu->pio;
	void *pd = vcpu->pio_data;
	int i;

	mutex_lock(&vcpu->kvm->lock);
	for (i = 0; i < io->cur_count; i++) {
		kvm_iodevice_write(pio_dev, io->port,
				   io->size,
				   pd);
		pd += io->size;
	}
	mutex_unlock(&vcpu->kvm->lock);
}

int kvm_emulate_pio (struct kvm_vcpu *vcpu, struct kvm_run *run, int in,
		  int size, unsigned port)
{
	struct kvm_io_device *pio_dev;

	vcpu->run->exit_reason = KVM_EXIT_IO;
	vcpu->run->io.direction = in ? KVM_EXIT_IO_IN : KVM_EXIT_IO_OUT;
	vcpu->run->io.size = vcpu->pio.size = size;
	vcpu->run->io.data_offset = KVM_PIO_PAGE_OFFSET * PAGE_SIZE;
	vcpu->run->io.count = vcpu->pio.count = vcpu->pio.cur_count = 1;
	vcpu->run->io.port = vcpu->pio.port = port;
	vcpu->pio.in = in;
	vcpu->pio.string = 0;
	vcpu->pio.down = 0;
	vcpu->pio.guest_page_offset = 0;
	vcpu->pio.rep = 0;

	kvm_x86_ops->cache_regs(vcpu);
	memcpy(vcpu->pio_data, &vcpu->regs[VCPU_REGS_RAX], 4);
	kvm_x86_ops->decache_regs(vcpu);

	kvm_x86_ops->skip_emulated_instruction(vcpu);

	pio_dev = vcpu_find_pio_dev(vcpu, port);
	if (pio_dev) {
		kernel_pio(pio_dev, vcpu, vcpu->pio_data);
		complete_pio(vcpu);
		return 1;
	}
	return 0;
}
EXPORT_SYMBOL_GPL(kvm_emulate_pio);

int kvm_emulate_pio_string(struct kvm_vcpu *vcpu, struct kvm_run *run, int in,
		  int size, unsigned long count, int down,
		  gva_t address, int rep, unsigned port)
{
	unsigned now, in_page;
	int i, ret = 0;
	int nr_pages = 1;
	struct page *page;
	struct kvm_io_device *pio_dev;

	vcpu->run->exit_reason = KVM_EXIT_IO;
	vcpu->run->io.direction = in ? KVM_EXIT_IO_IN : KVM_EXIT_IO_OUT;
	vcpu->run->io.size = vcpu->pio.size = size;
	vcpu->run->io.data_offset = KVM_PIO_PAGE_OFFSET * PAGE_SIZE;
	vcpu->run->io.count = vcpu->pio.count = vcpu->pio.cur_count = count;
	vcpu->run->io.port = vcpu->pio.port = port;
	vcpu->pio.in = in;
	vcpu->pio.string = 1;
	vcpu->pio.down = down;
	vcpu->pio.guest_page_offset = offset_in_page(address);
	vcpu->pio.rep = rep;

	if (!count) {
		kvm_x86_ops->skip_emulated_instruction(vcpu);
		return 1;
	}

	if (!down)
		in_page = PAGE_SIZE - offset_in_page(address);
	else
		in_page = offset_in_page(address) + size;
	now = min(count, (unsigned long)in_page / size);
	if (!now) {
		/*
		 * String I/O straddles page boundary.  Pin two guest pages
		 * so that we satisfy atomicity constraints.  Do just one
		 * transaction to avoid complexity.
		 */
		nr_pages = 2;
		now = 1;
	}
	if (down) {
		/*
		 * String I/O in reverse.  Yuck.  Kill the guest, fix later.
		 */
		pr_unimpl(vcpu, "guest string pio down\n");
		inject_gp(vcpu);
		return 1;
	}
	vcpu->run->io.count = now;
	vcpu->pio.cur_count = now;

	if (vcpu->pio.cur_count == vcpu->pio.count)
		kvm_x86_ops->skip_emulated_instruction(vcpu);

	for (i = 0; i < nr_pages; ++i) {
		mutex_lock(&vcpu->kvm->lock);
		page = gva_to_page(vcpu, address + i * PAGE_SIZE);
		if (page)
			get_page(page);
		vcpu->pio.guest_pages[i] = page;
		mutex_unlock(&vcpu->kvm->lock);
		if (!page) {
			inject_gp(vcpu);
			free_pio_guest_pages(vcpu);
			return 1;
		}
	}

	pio_dev = vcpu_find_pio_dev(vcpu, port);
	if (!vcpu->pio.in) {
		/* string PIO write */
		ret = pio_copy_data(vcpu);
		if (ret >= 0 && pio_dev) {
			pio_string_write(pio_dev, vcpu);
			complete_pio(vcpu);
			if (vcpu->pio.count == 0)
				ret = 1;
		}
	} else if (pio_dev)
		pr_unimpl(vcpu, "no string pio read support yet, "
		       "port %x size %d count %ld\n",
			port, size, count);

	return ret;
}
EXPORT_SYMBOL_GPL(kvm_emulate_pio_string);

/*
 * Check if userspace requested an interrupt window, and that the
 * interrupt window is open.
 *
 * No need to exit to userspace if we already have an interrupt queued.
 */
static int dm_request_for_irq_injection(struct kvm_vcpu *vcpu,
					  struct kvm_run *kvm_run)
{
	return (!vcpu->irq_summary &&
		kvm_run->request_interrupt_window &&
		vcpu->interrupt_window_open &&
		(kvm_x86_ops->get_rflags(vcpu) & X86_EFLAGS_IF));
}

static void post_kvm_run_save(struct kvm_vcpu *vcpu,
			      struct kvm_run *kvm_run)
{
	kvm_run->if_flag = (kvm_x86_ops->get_rflags(vcpu) & X86_EFLAGS_IF) != 0;
	kvm_run->cr8 = get_cr8(vcpu);
	kvm_run->apic_base = kvm_get_apic_base(vcpu);
	if (irqchip_in_kernel(vcpu->kvm))
		kvm_run->ready_for_interrupt_injection = 1;
	else
		kvm_run->ready_for_interrupt_injection =
					(vcpu->interrupt_window_open &&
					 vcpu->irq_summary == 0);
}

static int __vcpu_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
	int r;

	if (unlikely(vcpu->mp_state == VCPU_MP_STATE_SIPI_RECEIVED)) {
		printk("vcpu %d received sipi with vector # %x\n",
		       vcpu->vcpu_id, vcpu->sipi_vector);
		kvm_lapic_reset(vcpu);
		kvm_x86_ops->vcpu_reset(vcpu);
		vcpu->mp_state = VCPU_MP_STATE_RUNNABLE;
	}

preempted:
	if (vcpu->guest_debug.enabled)
		kvm_x86_ops->guest_debug_pre(vcpu);

again:
	r = kvm_mmu_reload(vcpu);
	if (unlikely(r))
		goto out;

	preempt_disable();

	kvm_x86_ops->prepare_guest_switch(vcpu);
	kvm_load_guest_fpu(vcpu);

	local_irq_disable();

	if (signal_pending(current)) {
		local_irq_enable();
		preempt_enable();
		r = -EINTR;
		kvm_run->exit_reason = KVM_EXIT_INTR;
		++vcpu->stat.signal_exits;
		goto out;
	}

	if (irqchip_in_kernel(vcpu->kvm))
		kvm_x86_ops->inject_pending_irq(vcpu);
	else if (!vcpu->mmio_read_completed)
		kvm_x86_ops->inject_pending_vectors(vcpu, kvm_run);

	vcpu->guest_mode = 1;
	kvm_guest_enter();

	if (vcpu->requests)
		if (test_and_clear_bit(KVM_TLB_FLUSH, &vcpu->requests))
			kvm_x86_ops->tlb_flush(vcpu);

	kvm_x86_ops->run(vcpu, kvm_run);

	vcpu->guest_mode = 0;
	local_irq_enable();

	++vcpu->stat.exits;

	/*
	 * We must have an instruction between local_irq_enable() and
	 * kvm_guest_exit(), so the timer interrupt isn't delayed by
	 * the interrupt shadow.  The stat.exits increment will do nicely.
	 * But we need to prevent reordering, hence this barrier():
	 */
	barrier();

	kvm_guest_exit();

	preempt_enable();

	/*
	 * Profile KVM exit RIPs:
	 */
	if (unlikely(prof_on == KVM_PROFILING)) {
		kvm_x86_ops->cache_regs(vcpu);
		profile_hit(KVM_PROFILING, (void *)vcpu->rip);
	}

	r = kvm_x86_ops->handle_exit(kvm_run, vcpu);

	if (r > 0) {
		if (dm_request_for_irq_injection(vcpu, kvm_run)) {
			r = -EINTR;
			kvm_run->exit_reason = KVM_EXIT_INTR;
			++vcpu->stat.request_irq_exits;
			goto out;
		}
		if (!need_resched()) {
			++vcpu->stat.light_exits;
			goto again;
		}
	}

out:
	if (r > 0) {
		kvm_resched(vcpu);
		goto preempted;
	}

	post_kvm_run_save(vcpu, kvm_run);

	return r;
}


static int kvm_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
{
	int r;
	sigset_t sigsaved;

	vcpu_load(vcpu);

	if (unlikely(vcpu->mp_state == VCPU_MP_STATE_UNINITIALIZED)) {
		kvm_vcpu_block(vcpu);
		vcpu_put(vcpu);
		return -EAGAIN;
	}

	if (vcpu->sigset_active)
		sigprocmask(SIG_SETMASK, &vcpu->sigset, &sigsaved);

	/* re-sync apic's tpr */
	if (!irqchip_in_kernel(vcpu->kvm))
		set_cr8(vcpu, kvm_run->cr8);

	if (vcpu->pio.cur_count) {
		r = complete_pio(vcpu);
		if (r)
			goto out;
	}

	if (vcpu->mmio_needed) {
		memcpy(vcpu->mmio_data, kvm_run->mmio.data, 8);
		vcpu->mmio_read_completed = 1;
		vcpu->mmio_needed = 0;
		r = emulate_instruction(vcpu, kvm_run,
					vcpu->mmio_fault_cr2, 0);
		if (r == EMULATE_DO_MMIO) {
			/*
			 * Read-modify-write.  Back to userspace.
			 */
			r = 0;
			goto out;
		}
	}

	if (kvm_run->exit_reason == KVM_EXIT_HYPERCALL) {
		kvm_x86_ops->cache_regs(vcpu);
		vcpu->regs[VCPU_REGS_RAX] = kvm_run->hypercall.ret;
		kvm_x86_ops->decache_regs(vcpu);
	}

	r = __vcpu_run(vcpu, kvm_run);

out:
	if (vcpu->sigset_active)
		sigprocmask(SIG_SETMASK, &sigsaved, NULL);

	vcpu_put(vcpu);
	return r;
}

static int kvm_vcpu_ioctl_get_regs(struct kvm_vcpu *vcpu,
				   struct kvm_regs *regs)
{
	vcpu_load(vcpu);

	kvm_x86_ops->cache_regs(vcpu);

	regs->rax = vcpu->regs[VCPU_REGS_RAX];
	regs->rbx = vcpu->regs[VCPU_REGS_RBX];
	regs->rcx = vcpu->regs[VCPU_REGS_RCX];
	regs->rdx = vcpu->regs[VCPU_REGS_RDX];
	regs->rsi = vcpu->regs[VCPU_REGS_RSI];
	regs->rdi = vcpu->regs[VCPU_REGS_RDI];
	regs->rsp = vcpu->regs[VCPU_REGS_RSP];
	regs->rbp = vcpu->regs[VCPU_REGS_RBP];
#ifdef CONFIG_X86_64
	regs->r8 = vcpu->regs[VCPU_REGS_R8];
	regs->r9 = vcpu->regs[VCPU_REGS_R9];
	regs->r10 = vcpu->regs[VCPU_REGS_R10];
	regs->r11 = vcpu->regs[VCPU_REGS_R11];
	regs->r12 = vcpu->regs[VCPU_REGS_R12];
	regs->r13 = vcpu->regs[VCPU_REGS_R13];
	regs->r14 = vcpu->regs[VCPU_REGS_R14];
	regs->r15 = vcpu->regs[VCPU_REGS_R15];
#endif

	regs->rip = vcpu->rip;
	regs->rflags = kvm_x86_ops->get_rflags(vcpu);

	/*
	 * Don't leak debug flags in case they were set for guest debugging
	 */
	if (vcpu->guest_debug.enabled && vcpu->guest_debug.singlestep)
		regs->rflags &= ~(X86_EFLAGS_TF | X86_EFLAGS_RF);

	vcpu_put(vcpu);

	return 0;
}

static int kvm_vcpu_ioctl_set_regs(struct kvm_vcpu *vcpu,
				   struct kvm_regs *regs)
{
	vcpu_load(vcpu);

	vcpu->regs[VCPU_REGS_RAX] = regs->rax;
	vcpu->regs[VCPU_REGS_RBX] = regs->rbx;
	vcpu->regs[VCPU_REGS_RCX] = regs->rcx;
	vcpu->regs[VCPU_REGS_RDX] = regs->rdx;
	vcpu->regs[VCPU_REGS_RSI] = regs->rsi;
	vcpu->regs[VCPU_REGS_RDI] = regs->rdi;
	vcpu->regs[VCPU_REGS_RSP] = regs->rsp;
	vcpu->regs[VCPU_REGS_RBP] = regs->rbp;
#ifdef CONFIG_X86_64
	vcpu->regs[VCPU_REGS_R8] = regs->r8;
	vcpu->regs[VCPU_REGS_R9] = regs->r9;
	vcpu->regs[VCPU_REGS_R10] = regs->r10;
	vcpu->regs[VCPU_REGS_R11] = regs->r11;
	vcpu->regs[VCPU_REGS_R12] = regs->r12;
	vcpu->regs[VCPU_REGS_R13] = regs->r13;
	vcpu->regs[VCPU_REGS_R14] = regs->r14;
	vcpu->regs[VCPU_REGS_R15] = regs->r15;
#endif

	vcpu->rip = regs->rip;
	kvm_x86_ops->set_rflags(vcpu, regs->rflags);

	kvm_x86_ops->decache_regs(vcpu);

	vcpu_put(vcpu);

	return 0;
}

static void get_segment(struct kvm_vcpu *vcpu,
			struct kvm_segment *var, int seg)
{
	return kvm_x86_ops->get_segment(vcpu, var, seg);
}

static int kvm_vcpu_ioctl_get_sregs(struct kvm_vcpu *vcpu,
				    struct kvm_sregs *sregs)
{
	struct descriptor_table dt;
	int pending_vec;

	vcpu_load(vcpu);

	get_segment(vcpu, &sregs->cs, VCPU_SREG_CS);
	get_segment(vcpu, &sregs->ds, VCPU_SREG_DS);
	get_segment(vcpu, &sregs->es, VCPU_SREG_ES);
	get_segment(vcpu, &sregs->fs, VCPU_SREG_FS);
	get_segment(vcpu, &sregs->gs, VCPU_SREG_GS);
	get_segment(vcpu, &sregs->ss, VCPU_SREG_SS);

	get_segment(vcpu, &sregs->tr, VCPU_SREG_TR);
	get_segment(vcpu, &sregs->ldt, VCPU_SREG_LDTR);

	kvm_x86_ops->get_idt(vcpu, &dt);
	sregs->idt.limit = dt.limit;
	sregs->idt.base = dt.base;
	kvm_x86_ops->get_gdt(vcpu, &dt);
	sregs->gdt.limit = dt.limit;
	sregs->gdt.base = dt.base;

	kvm_x86_ops->decache_cr4_guest_bits(vcpu);
	sregs->cr0 = vcpu->cr0;
	sregs->cr2 = vcpu->cr2;
	sregs->cr3 = vcpu->cr3;
	sregs->cr4 = vcpu->cr4;
	sregs->cr8 = get_cr8(vcpu);
	sregs->efer = vcpu->shadow_efer;
	sregs->apic_base = kvm_get_apic_base(vcpu);

	if (irqchip_in_kernel(vcpu->kvm)) {
		memset(sregs->interrupt_bitmap, 0,