[mirror_ubuntu-bionic-kernel.git] / drivers / virtio / virtio_pci_common.c

/*
 * Virtio PCI driver - common functionality for all device versions
 *
 * This module allows virtio devices to be used over a virtual PCI device.
 * This can be used with QEMU based VMMs like KVM or Xen.
 *
 * Copyright IBM Corp. 2007
 * Copyright Red Hat, Inc. 2014
 *
 * Authors:
 *  Anthony Liguori  <aliguori@us.ibm.com>
 *  Rusty Russell <rusty@rustcorp.com.au>
 *  Michael S. Tsirkin <mst@redhat.com>
 *
 * This work is licensed under the terms of the GNU GPL, version 2 or later.
 * See the COPYING file in the top-level directory.
 *
 */

#include "virtio_pci_common.h"

static bool force_legacy = false;

#if IS_ENABLED(CONFIG_VIRTIO_PCI_LEGACY)
module_param(force_legacy, bool, 0444);
MODULE_PARM_DESC(force_legacy,
		 "Force legacy mode for transitional virtio 1 devices");
#endif

/* wait for pending irq handlers */
void vp_synchronize_vectors(struct virtio_device *vdev)
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	int i;

	if (vp_dev->intx_enabled)
		synchronize_irq(vp_dev->pci_dev->irq);

	for (i = 0; i < vp_dev->msix_vectors; ++i)
		synchronize_irq(pci_irq_vector(vp_dev->pci_dev, i));
}

/* the notify function used when creating a virt queue */
bool vp_notify(struct virtqueue *vq)
{
	/* we write the queue's selector into the notification register to
	 * signal the other end */
	iowrite16(vq->index, (void __iomem *)vq->priv);
	return true;
}

/* Handle a configuration change: Tell driver if it wants to know. */
static irqreturn_t vp_config_changed(int irq, void *opaque)
{
	struct virtio_pci_device *vp_dev = opaque;

	virtio_config_changed(&vp_dev->vdev);
	return IRQ_HANDLED;
}

/* Notify all virtqueues on an interrupt. */
static irqreturn_t vp_vring_interrupt(int irq, void *opaque)
{
	struct virtio_pci_device *vp_dev = opaque;
	struct virtio_pci_vq_info *info;
	irqreturn_t ret = IRQ_NONE;
	unsigned long flags;

	spin_lock_irqsave(&vp_dev->lock, flags);
	list_for_each_entry(info, &vp_dev->virtqueues, node) {
		if (vring_interrupt(irq, info->vq) == IRQ_HANDLED)
			ret = IRQ_HANDLED;
	}
	spin_unlock_irqrestore(&vp_dev->lock, flags);

	return ret;
}

/* A small wrapper to also acknowledge the interrupt when it's handled.
 * I really need an EIO hook for the vring so I can ack the interrupt once we
 * know that we'll be handling the IRQ but before we invoke the callback since
 * the callback may notify the host which results in the host attempting to
 * raise an interrupt that we would then mask once we acknowledged the
 * interrupt. */
static irqreturn_t vp_interrupt(int irq, void *opaque)
{
	struct virtio_pci_device *vp_dev = opaque;
	u8 isr;

	/* reading the ISR has the effect of also clearing it so it's very
	 * important to save off the value. */
	isr = ioread8(vp_dev->isr);

	/* It's definitely not us if the ISR was not high */
	if (!isr)
		return IRQ_NONE;

	/* Configuration change?  Tell driver if it wants to know. */
	if (isr & VIRTIO_PCI_ISR_CONFIG)
		vp_config_changed(irq, opaque);

	return vp_vring_interrupt(irq, opaque);
}

static int vp_request_msix_vectors(struct virtio_device *vdev, int nvectors,
				   bool per_vq_vectors, struct irq_affinity *desc)
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	const char *name = dev_name(&vp_dev->vdev.dev);
	unsigned i, v;
	int err = -ENOMEM;

	vp_dev->msix_vectors = nvectors;

	vp_dev->msix_names = kmalloc(nvectors * sizeof *vp_dev->msix_names,
				     GFP_KERNEL);
	if (!vp_dev->msix_names)
		goto error;
	vp_dev->msix_affinity_masks
		= kzalloc(nvectors * sizeof *vp_dev->msix_affinity_masks,
			  GFP_KERNEL);
	if (!vp_dev->msix_affinity_masks)
		goto error;
	for (i = 0; i < nvectors; ++i)
		if (!alloc_cpumask_var(&vp_dev->msix_affinity_masks[i],
					GFP_KERNEL))
			goto error;

	err = pci_alloc_irq_vectors_affinity(vp_dev->pci_dev, nvectors,
					     nvectors, PCI_IRQ_MSIX |
					     (desc ? PCI_IRQ_AFFINITY : 0),
					     desc);
	if (err < 0)
		goto error;
	vp_dev->msix_enabled = 1;

	/* Set the vector used for configuration */
	v = vp_dev->msix_used_vectors;
	snprintf(vp_dev->msix_names[v], sizeof *vp_dev->msix_names,
		 "%s-config", name);
	err = request_irq(pci_irq_vector(vp_dev->pci_dev, v),
			  vp_config_changed, 0, vp_dev->msix_names[v],
			  vp_dev);
	if (err)
		goto error;
	++vp_dev->msix_used_vectors;

	v = vp_dev->config_vector(vp_dev, v);
	/* Verify we had enough resources to assign the vector */
	if (v == VIRTIO_MSI_NO_VECTOR) {
		err = -EBUSY;
		goto error;
	}

	if (!per_vq_vectors) {
		/* Shared vector for all VQs */
		v = vp_dev->msix_used_vectors;
		snprintf(vp_dev->msix_names[v], sizeof *vp_dev->msix_names,
			 "%s-virtqueues", name);
		err = request_irq(pci_irq_vector(vp_dev->pci_dev, v),
				  vp_vring_interrupt, 0, vp_dev->msix_names[v],
				  vp_dev);
		if (err)
			goto error;
		++vp_dev->msix_used_vectors;
	}
	return 0;
error:
	return err;
}

static struct virtqueue *vp_setup_vq(struct virtio_device *vdev, unsigned index,
				     void (*callback)(struct virtqueue *vq),
				     const char *name,
				     u16 msix_vec)
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	struct virtio_pci_vq_info *info = kmalloc(sizeof *info, GFP_KERNEL);
	struct virtqueue *vq;
	unsigned long flags;

	/* fill out our structure that represents an active queue */
	if (!info)
		return ERR_PTR(-ENOMEM);

	vq = vp_dev->setup_vq(vp_dev, info, index, callback, name,
			      msix_vec);
	if (IS_ERR(vq))
		goto out_info;

	info->vq = vq;
	if (callback) {
		spin_lock_irqsave(&vp_dev->lock, flags);
		list_add(&info->node, &vp_dev->virtqueues);
		spin_unlock_irqrestore(&vp_dev->lock, flags);
	} else {
		INIT_LIST_HEAD(&info->node);
	}

	vp_dev->vqs[index] = info;
	return vq;

out_info:
	kfree(info);
	return vq;
}

static void vp_del_vq(struct virtqueue *vq)
{
	struct virtio_pci_device *vp_dev = to_vp_device(vq->vdev);
	struct virtio_pci_vq_info *info = vp_dev->vqs[vq->index];
	unsigned long flags;

	spin_lock_irqsave(&vp_dev->lock, flags);
	list_del(&info->node);
	spin_unlock_irqrestore(&vp_dev->lock, flags);

	vp_dev->del_vq(info);
	kfree(info);
}

/* the config->del_vqs() implementation */
void vp_del_vqs(struct virtio_device *vdev)
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	struct virtqueue *vq, *n;
	int i;

	list_for_each_entry_safe(vq, n, &vdev->vqs, list) {
		if (vp_dev->per_vq_vectors) {
			int v = vp_dev->vqs[vq->index]->msix_vector;

			if (v != VIRTIO_MSI_NO_VECTOR) {
				int irq = pci_irq_vector(vp_dev->pci_dev, v);

				irq_set_affinity_hint(irq, NULL);
				free_irq(irq, vq);
			}
		}
		vp_del_vq(vq);
	}
	vp_dev->per_vq_vectors = false;

	if (vp_dev->intx_enabled) {
		free_irq(vp_dev->pci_dev->irq, vp_dev);
		vp_dev->intx_enabled = 0;
	}

	for (i = 0; i < vp_dev->msix_used_vectors; ++i)
		free_irq(pci_irq_vector(vp_dev->pci_dev, i), vp_dev);

	for (i = 0; i < vp_dev->msix_vectors; i++)
		if (vp_dev->msix_affinity_masks[i])
			free_cpumask_var(vp_dev->msix_affinity_masks[i]);

	if (vp_dev->msix_enabled) {
		/* Disable the vector used for configuration */
		vp_dev->config_vector(vp_dev, VIRTIO_MSI_NO_VECTOR);

		pci_free_irq_vectors(vp_dev->pci_dev);
		vp_dev->msix_enabled = 0;
	}

	vp_dev->msix_vectors = 0;
	vp_dev->msix_used_vectors = 0;
	kfree(vp_dev->msix_names);
	vp_dev->msix_names = NULL;
	kfree(vp_dev->msix_affinity_masks);
	vp_dev->msix_affinity_masks = NULL;
	kfree(vp_dev->vqs);
	vp_dev->vqs = NULL;
}

static int vp_find_vqs_msix(struct virtio_device *vdev, unsigned nvqs,
		struct virtqueue *vqs[], vq_callback_t *callbacks[],
		const char * const names[], bool per_vq_vectors,
		struct irq_affinity *desc)
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	u16 msix_vec;
	int i, err, nvectors, allocated_vectors;

	vp_dev->vqs = kcalloc(nvqs, sizeof(*vp_dev->vqs), GFP_KERNEL);
	if (!vp_dev->vqs)
		return -ENOMEM;

	if (per_vq_vectors) {
		/* Best option: one for change interrupt, one per vq. */
		nvectors = 1;
		for (i = 0; i < nvqs; ++i)
			if (callbacks[i])
				++nvectors;
	} else {
		/* Second best: one for change, shared for all vqs. */
		nvectors = 2;
	}

	err = vp_request_msix_vectors(vdev, nvectors, per_vq_vectors,
				      per_vq_vectors ? desc : NULL);
	if (err)
		goto error_find;

	vp_dev->per_vq_vectors = per_vq_vectors;
	allocated_vectors = vp_dev->msix_used_vectors;
	for (i = 0; i < nvqs; ++i) {
		if (!names[i]) {
			vqs[i] = NULL;
			continue;
		}

		if (!callbacks[i])
			msix_vec = VIRTIO_MSI_NO_VECTOR;
		else if (vp_dev->per_vq_vectors)
			msix_vec = allocated_vectors++;
		else
			msix_vec = VP_MSIX_VQ_VECTOR;
		vqs[i] = vp_setup_vq(vdev, i, callbacks[i], names[i],
				     msix_vec);
		if (IS_ERR(vqs[i])) {
			err = PTR_ERR(vqs[i]);
			goto error_find;
		}

		if (!vp_dev->per_vq_vectors || msix_vec == VIRTIO_MSI_NO_VECTOR)
			continue;

		/* allocate per-vq irq if available and necessary */
		snprintf(vp_dev->msix_names[msix_vec],
			 sizeof *vp_dev->msix_names,
			 "%s-%s",
			 dev_name(&vp_dev->vdev.dev), names[i]);
		err = request_irq(pci_irq_vector(vp_dev->pci_dev, msix_vec),
				  vring_interrupt, 0,
				  vp_dev->msix_names[msix_vec],
				  vqs[i]);
		if (err)
			goto error_find;
	}
	return 0;

error_find:
	vp_del_vqs(vdev);
	return err;
}

static int vp_find_vqs_intx(struct virtio_device *vdev, unsigned nvqs,
		struct virtqueue *vqs[], vq_callback_t *callbacks[],
		const char * const names[])
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	int i, err;

	vp_dev->vqs = kcalloc(nvqs, sizeof(*vp_dev->vqs), GFP_KERNEL);
	if (!vp_dev->vqs)
		return -ENOMEM;

	err = request_irq(vp_dev->pci_dev->irq, vp_interrupt, IRQF_SHARED,
			dev_name(&vdev->dev), vp_dev);
	if (err)
		goto out_del_vqs;

	vp_dev->intx_enabled = 1;
	vp_dev->per_vq_vectors = false;
	for (i = 0; i < nvqs; ++i) {
		if (!names[i]) {
			vqs[i] = NULL;
			continue;
		}
		vqs[i] = vp_setup_vq(vdev, i, callbacks[i], names[i],
				     VIRTIO_MSI_NO_VECTOR);
		if (IS_ERR(vqs[i])) {
			err = PTR_ERR(vqs[i]);
			goto out_del_vqs;
		}
	}

	return 0;
out_del_vqs:
	vp_del_vqs(vdev);
	return err;
}

/* the config->find_vqs() implementation */
int vp_find_vqs(struct virtio_device *vdev, unsigned nvqs,
		struct virtqueue *vqs[], vq_callback_t *callbacks[],
		const char * const names[], struct irq_affinity *desc)
{
	int err;

	/* Try MSI-X with one vector per queue. */
	err = vp_find_vqs_msix(vdev, nvqs, vqs, callbacks, names, true, desc);
	if (!err)
		return 0;
	/* Fallback: MSI-X with one vector for config, one shared for queues. */
	err = vp_find_vqs_msix(vdev, nvqs, vqs, callbacks, names, false, desc);
	if (!err)
		return 0;
	/* Finally fall back to regular interrupts. */
	return vp_find_vqs_intx(vdev, nvqs, vqs, callbacks, names);
}

const char *vp_bus_name(struct virtio_device *vdev)
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);

	return pci_name(vp_dev->pci_dev);
}

/* Setup the affinity for a virtqueue:
 * - force the affinity for per vq vector
 * - OR over all affinities for shared MSI
 * - ignore the affinity request if we're using INTX
 */
int vp_set_vq_affinity(struct virtqueue *vq, int cpu)
{
	struct virtio_device *vdev = vq->vdev;
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	struct virtio_pci_vq_info *info = vp_dev->vqs[vq->index];
	struct cpumask *mask;
	unsigned int irq;

	if (!vq->callback)
		return -EINVAL;

	if (vp_dev->msix_enabled) {
		mask = vp_dev->msix_affinity_masks[info->msix_vector];
		irq = pci_irq_vector(vp_dev->pci_dev, info->msix_vector);
		if (cpu == -1)
			irq_set_affinity_hint(irq, NULL);
		else {
			cpumask_clear(mask);
			cpumask_set_cpu(cpu, mask);
			irq_set_affinity_hint(irq, mask);
		}
	}
	return 0;
}

const struct cpumask *vp_get_vq_affinity(struct virtio_device *vdev, int index)
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);

	if (!vp_dev->per_vq_vectors ||
	    vp_dev->vqs[index]->msix_vector == VIRTIO_MSI_NO_VECTOR)
		return NULL;

	return pci_irq_get_affinity(vp_dev->pci_dev,
				    vp_dev->vqs[index]->msix_vector);
}

#ifdef CONFIG_PM_SLEEP
static int virtio_pci_freeze(struct device *dev)
{
	struct pci_dev *pci_dev = to_pci_dev(dev);
	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
	int ret;

	ret = virtio_device_freeze(&vp_dev->vdev);

	if (!ret)
		pci_disable_device(pci_dev);
	return ret;
}

static int virtio_pci_restore(struct device *dev)
{
	struct pci_dev *pci_dev = to_pci_dev(dev);
	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
	int ret;

	ret = pci_enable_device(pci_dev);
	if (ret)
		return ret;

	pci_set_master(pci_dev);
	return virtio_device_restore(&vp_dev->vdev);
}

static const struct dev_pm_ops virtio_pci_pm_ops = {
	SET_SYSTEM_SLEEP_PM_OPS(virtio_pci_freeze, virtio_pci_restore)
};
#endif


/* Qumranet donated their vendor ID for devices 0x1000 thru 0x10FF. */
static const struct pci_device_id virtio_pci_id_table[] = {
	{ PCI_DEVICE(PCI_VENDOR_ID_REDHAT_QUMRANET, PCI_ANY_ID) },
	{ 0 }
};

MODULE_DEVICE_TABLE(pci, virtio_pci_id_table);

static void virtio_pci_release_dev(struct device *_d)
{
	struct virtio_device *vdev = dev_to_virtio(_d);
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);

	/* As struct device is a kobject, it's not safe to
	 * free the memory (including the reference counter itself)
	 * until it's release callback. */
	kfree(vp_dev);
}

static int virtio_pci_probe(struct pci_dev *pci_dev,
			    const struct pci_device_id *id)
{
	struct virtio_pci_device *vp_dev;
	int rc;

	/* allocate our structure and fill it out */
	vp_dev = kzalloc(sizeof(struct virtio_pci_device), GFP_KERNEL);
	if (!vp_dev)
		return -ENOMEM;

	pci_set_drvdata(pci_dev, vp_dev);
	vp_dev->vdev.dev.parent = &pci_dev->dev;
	vp_dev->vdev.dev.release = virtio_pci_release_dev;
	vp_dev->pci_dev = pci_dev;
	INIT_LIST_HEAD(&vp_dev->virtqueues);
	spin_lock_init(&vp_dev->lock);

	/* enable the device */
	rc = pci_enable_device(pci_dev);
	if (rc)
		goto err_enable_device;

	if (force_legacy) {
		rc = virtio_pci_legacy_probe(vp_dev);
		/* Also try modern mode if we can't map BAR0 (no IO space). */
		if (rc == -ENODEV || rc == -ENOMEM)
			rc = virtio_pci_modern_probe(vp_dev);
		if (rc)
			goto err_probe;
	} else {
		rc = virtio_pci_modern_probe(vp_dev);
		if (rc == -ENODEV)
			rc = virtio_pci_legacy_probe(vp_dev);
		if (rc)
			goto err_probe;
	}

	pci_set_master(pci_dev);

	rc = register_virtio_device(&vp_dev->vdev);
	if (rc)
		goto err_register;

	return 0;

err_register:
	if (vp_dev->ioaddr)
	     virtio_pci_legacy_remove(vp_dev);
	else
	     virtio_pci_modern_remove(vp_dev);
err_probe:
	pci_disable_device(pci_dev);
err_enable_device:
	kfree(vp_dev);
	return rc;
}

static void virtio_pci_remove(struct pci_dev *pci_dev)
{
	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
	struct device *dev = get_device(&vp_dev->vdev.dev);

	unregister_virtio_device(&vp_dev->vdev);

	if (vp_dev->ioaddr)
		virtio_pci_legacy_remove(vp_dev);
	else
		virtio_pci_modern_remove(vp_dev);

	pci_disable_device(pci_dev);
	put_device(dev);
}

static struct pci_driver virtio_pci_driver = {
	.name		= "virtio-pci",
	.id_table	= virtio_pci_id_table,
	.probe		= virtio_pci_probe,
	.remove		= virtio_pci_remove,
#ifdef CONFIG_PM_SLEEP
	.driver.pm	= &virtio_pci_pm_ops,
#endif
};

module_pci_driver(virtio_pci_driver);

MODULE_AUTHOR("Anthony Liguori <aliguori@us.ibm.com>");
MODULE_DESCRIPTION("virtio-pci");
MODULE_LICENSE("GPL");
MODULE_VERSION("1");
Commit	Line	Data
3343660d	1	/*
a90fdce9	2	* Virtio PCI driver - common functionality for all device versions
3343660d AL	3	*
	4	* This module allows virtio devices to be used over a virtual PCI device.
	5	* This can be used with QEMU based VMMs like KVM or Xen.
	6	*
	7	* Copyright IBM Corp. 2007
a90fdce9	8	* Copyright Red Hat, Inc. 2014
3343660d AL	9	*
	10	* Authors:
	11	* Anthony Liguori <aliguori@us.ibm.com>
a90fdce9 MT	12	* Rusty Russell <rusty@rustcorp.com.au>
a90fdce9 MT	13	* Michael S. Tsirkin <mst@redhat.com>
3343660d AL	14	*
	15	* This work is licensed under the terms of the GNU GPL, version 2 or later.
	16	* See the COPYING file in the top-level directory.
	17	*
	18	*/
	19
5f4c9760	20	#include "virtio_pci_common.h"
3343660d	21
ac399d8f MT	22	static bool force_legacy = false;
	23
	24	#if IS_ENABLED(CONFIG_VIRTIO_PCI_LEGACY)
	25	module_param(force_legacy, bool, 0444);
	26	MODULE_PARM_DESC(force_legacy,
	27	"Force legacy mode for transitional virtio 1 devices");
	28	#endif
	29
e6af578c	30	/* wait for pending irq handlers */
38eb4a29	31	void vp_synchronize_vectors(struct virtio_device *vdev)
e6af578c MT	32	{
	33	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	34	int i;
	35
0b0f9dc5 MT	36	if (vp_dev->intx_enabled)
	37	synchronize_irq(vp_dev->pci_dev->irq);
	38
	39	for (i = 0; i < vp_dev->msix_vectors; ++i)
fa3a3279	40	synchronize_irq(pci_irq_vector(vp_dev->pci_dev, i));
e6af578c MT	41	}
e6af578c MT	42
3343660d	43	/* the notify function used when creating a virt queue */
38eb4a29	44	bool vp_notify(struct virtqueue *vq)
3343660d	45	{
3343660d AL	46	/* we write the queue's selector into the notification register to
3343660d AL	47	* signal the other end */
f30eaf4a	48	iowrite16(vq->index, (void __iomem *)vq->priv);
46f9c2b9	49	return true;
3343660d AL	50	}
3343660d AL	51
77cf5246 MT	52	/* Handle a configuration change: Tell driver if it wants to know. */
	53	static irqreturn_t vp_config_changed(int irq, void *opaque)
	54	{
	55	struct virtio_pci_device *vp_dev = opaque;
77cf5246	56
016c98c6	57	virtio_config_changed(&vp_dev->vdev);
77cf5246 MT	58	return IRQ_HANDLED;
	59	}
	60
	61	/* Notify all virtqueues on an interrupt. */
	62	static irqreturn_t vp_vring_interrupt(int irq, void *opaque)
	63	{
	64	struct virtio_pci_device *vp_dev = opaque;
0a9b3f47	65	struct virtio_pci_vq_info *info;
77cf5246	66	irqreturn_t ret = IRQ_NONE;
0a9b3f47	67	unsigned long flags;
77cf5246	68
0a9b3f47 MT	69	spin_lock_irqsave(&vp_dev->lock, flags);
	70	list_for_each_entry(info, &vp_dev->virtqueues, node) {
	71	if (vring_interrupt(irq, info->vq) == IRQ_HANDLED)
77cf5246 MT	72	ret = IRQ_HANDLED;
77cf5246 MT	73	}
0a9b3f47	74	spin_unlock_irqrestore(&vp_dev->lock, flags);
77cf5246 MT	75
	76	return ret;
	77	}
	78
3343660d AL	79	/* A small wrapper to also acknowledge the interrupt when it's handled.
	80	* I really need an EIO hook for the vring so I can ack the interrupt once we
	81	* know that we'll be handling the IRQ but before we invoke the callback since
	82	* the callback may notify the host which results in the host attempting to
	83	* raise an interrupt that we would then mask once we acknowledged the
	84	* interrupt. */
	85	static irqreturn_t vp_interrupt(int irq, void *opaque)
	86	{
	87	struct virtio_pci_device *vp_dev = opaque;
3343660d AL	88	u8 isr;
	89
	90	/* reading the ISR has the effect of also clearing it so it's very
	91	* important to save off the value. */
af535722	92	isr = ioread8(vp_dev->isr);
3343660d AL	93
	94	/* It's definitely not us if the ISR was not high */
	95	if (!isr)
	96	return IRQ_NONE;
	97
	98	/* Configuration change? Tell driver if it wants to know. */
77cf5246 MT	99	if (isr & VIRTIO_PCI_ISR_CONFIG)
77cf5246 MT	100	vp_config_changed(irq, opaque);
3343660d	101
77cf5246	102	return vp_vring_interrupt(irq, opaque);
3343660d AL	103	}
3343660d AL	104
0b0f9dc5 MT	105	static int vp_request_msix_vectors(struct virtio_device *vdev, int nvectors,
	106	bool per_vq_vectors, struct irq_affinity *desc)
	107	{
	108	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	109	const char *name = dev_name(&vp_dev->vdev.dev);
	110	unsigned i, v;
	111	int err = -ENOMEM;
	112
	113	vp_dev->msix_vectors = nvectors;
	114
	115	vp_dev->msix_names = kmalloc(nvectors * sizeof *vp_dev->msix_names,
	116	GFP_KERNEL);
	117	if (!vp_dev->msix_names)
	118	goto error;
	119	vp_dev->msix_affinity_masks
	120	= kzalloc(nvectors * sizeof *vp_dev->msix_affinity_masks,
	121	GFP_KERNEL);
	122	if (!vp_dev->msix_affinity_masks)
	123	goto error;
	124	for (i = 0; i < nvectors; ++i)
	125	if (!alloc_cpumask_var(&vp_dev->msix_affinity_masks[i],
	126	GFP_KERNEL))
	127	goto error;
	128
	129	err = pci_alloc_irq_vectors_affinity(vp_dev->pci_dev, nvectors,
	130	nvectors, PCI_IRQ_MSIX \|
	131	(desc ? PCI_IRQ_AFFINITY : 0),
	132	desc);
	133	if (err < 0)
	134	goto error;
	135	vp_dev->msix_enabled = 1;
	136
	137	/* Set the vector used for configuration */
	138	v = vp_dev->msix_used_vectors;
	139	snprintf(vp_dev->msix_names[v], sizeof *vp_dev->msix_names,
	140	"%s-config", name);
	141	err = request_irq(pci_irq_vector(vp_dev->pci_dev, v),
	142	vp_config_changed, 0, vp_dev->msix_names[v],
	143	vp_dev);
	144	if (err)
	145	goto error;
	146	++vp_dev->msix_used_vectors;
	147
	148	v = vp_dev->config_vector(vp_dev, v);
	149	/* Verify we had enough resources to assign the vector */
	150	if (v == VIRTIO_MSI_NO_VECTOR) {
	151	err = -EBUSY;
	152	goto error;
	153	}
	154
	155	if (!per_vq_vectors) {
	156	/* Shared vector for all VQs */
	157	v = vp_dev->msix_used_vectors;
	158	snprintf(vp_dev->msix_names[v], sizeof *vp_dev->msix_names,
	159	"%s-virtqueues", name);
	160	err = request_irq(pci_irq_vector(vp_dev->pci_dev, v),
	161	vp_vring_interrupt, 0, vp_dev->msix_names[v],
	162	vp_dev);
	163	if (err)
	164	goto error;
	165	++vp_dev->msix_used_vectors;
	166	}
	167	return 0;
	168	error:
169	return err;
170	}
171
0a9b3f47 MT	172	static struct virtqueue vp_setup_vq(struct virtio_device vdev, unsigned index,
	173	void (callback)(struct virtqueue vq),
	174	const char *name,
	175	u16 msix_vec)
	176	{
	177	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	178	struct virtio_pci_vq_info info = kmalloc(sizeof info, GFP_KERNEL);
	179	struct virtqueue *vq;
	180	unsigned long flags;
	181
	182	/* fill out our structure that represents an active queue */
	183	if (!info)
	184	return ERR_PTR(-ENOMEM);
	185
	186	vq = vp_dev->setup_vq(vp_dev, info, index, callback, name,
	187	msix_vec);
	188	if (IS_ERR(vq))
	189	goto out_info;
	190
	191	info->vq = vq;
	192	if (callback) {
	193	spin_lock_irqsave(&vp_dev->lock, flags);
	194	list_add(&info->node, &vp_dev->virtqueues);
	195	spin_unlock_irqrestore(&vp_dev->lock, flags);
	196	} else {
	197	INIT_LIST_HEAD(&info->node);
	198	}
	199
	200	vp_dev->vqs[index] = info;
	201	return vq;
	202
	203	out_info:
	204	kfree(info);
	205	return vq;
	206	}
	207
	208	static void vp_del_vq(struct virtqueue *vq)
	209	{
	210	struct virtio_pci_device *vp_dev = to_vp_device(vq->vdev);
	211	struct virtio_pci_vq_info *info = vp_dev->vqs[vq->index];
	212	unsigned long flags;
	213
	214	spin_lock_irqsave(&vp_dev->lock, flags);
	215	list_del(&info->node);
	216	spin_unlock_irqrestore(&vp_dev->lock, flags);
	217
	218	vp_dev->del_vq(info);
	219	kfree(info);
	220	}
	221
0b0f9dc5 MT	222	/* the config->del_vqs() implementation */
0b0f9dc5 MT	223	void vp_del_vqs(struct virtio_device *vdev)
d2a7ddda	224	{
e969fed5	225	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
d2a7ddda	226	struct virtqueue vq, n;
0b0f9dc5	227	int i;
d2a7ddda	228
e969fed5	229	list_for_each_entry_safe(vq, n, &vdev->vqs, list) {
0a9b3f47 MT	230	if (vp_dev->per_vq_vectors) {
0a9b3f47 MT	231	int v = vp_dev->vqs[vq->index]->msix_vector;
fa3a3279	232
2f8dc3a0 MZ	233	if (v != VIRTIO_MSI_NO_VECTOR) {
	234	int irq = pci_irq_vector(vp_dev->pci_dev, v);
	235
	236	irq_set_affinity_hint(irq, NULL);
	237	free_irq(irq, vq);
	238	}
fa3a3279	239	}
0a9b3f47	240	vp_del_vq(vq);
e969fed5	241	}
0a9b3f47	242	vp_dev->per_vq_vectors = false;
82af8ce8	243
0b0f9dc5 MT	244	if (vp_dev->intx_enabled) {
	245	free_irq(vp_dev->pci_dev->irq, vp_dev);
	246	vp_dev->intx_enabled = 0;
	247	}
66f2f555	248
0b0f9dc5 MT	249	for (i = 0; i < vp_dev->msix_used_vectors; ++i)
0b0f9dc5 MT	250	free_irq(pci_irq_vector(vp_dev->pci_dev, i), vp_dev);
66f2f555	251
0b0f9dc5 MT	252	for (i = 0; i < vp_dev->msix_vectors; i++)
0b0f9dc5 MT	253	if (vp_dev->msix_affinity_masks[i])
07ec5148 CH	254	free_cpumask_var(vp_dev->msix_affinity_masks[i]);
07ec5148 CH	255
0b0f9dc5	256	if (vp_dev->msix_enabled) {
66f2f555 CH	257	/* Disable the vector used for configuration */
	258	vp_dev->config_vector(vp_dev, VIRTIO_MSI_NO_VECTOR);
	259
0b0f9dc5 MT	260	pci_free_irq_vectors(vp_dev->pci_dev);
0b0f9dc5 MT	261	vp_dev->msix_enabled = 0;
66f2f555 CH	262	}
66f2f555 CH	263
0b0f9dc5 MT	264	vp_dev->msix_vectors = 0;
	265	vp_dev->msix_used_vectors = 0;
	266	kfree(vp_dev->msix_names);
	267	vp_dev->msix_names = NULL;
	268	kfree(vp_dev->msix_affinity_masks);
	269	vp_dev->msix_affinity_masks = NULL;
0a9b3f47 MT	270	kfree(vp_dev->vqs);
0a9b3f47 MT	271	vp_dev->vqs = NULL;
d2a7ddda MT	272	}
d2a7ddda MT	273
a3cbec69	274	static int vp_find_vqs_msix(struct virtio_device *vdev, unsigned nvqs,
52a61516	275	struct virtqueue vqs[], vq_callback_t callbacks[],
bf951b10 MT	276	const char * const names[], bool per_vq_vectors,
bf951b10 MT	277	struct irq_affinity *desc)
d2a7ddda	278	{
e969fed5	279	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
f68d2408	280	u16 msix_vec;
0b0f9dc5	281	int i, err, nvectors, allocated_vectors;
82af8ce8	282
0a9b3f47 MT	283	vp_dev->vqs = kcalloc(nvqs, sizeof(*vp_dev->vqs), GFP_KERNEL);
	284	if (!vp_dev->vqs)
	285	return -ENOMEM;
	286
bf951b10	287	if (per_vq_vectors) {
0b0f9dc5 MT	288	/* Best option: one for change interrupt, one per vq. */
	289	nvectors = 1;
	290	for (i = 0; i < nvqs; ++i)
	291	if (callbacks[i])
	292	++nvectors;
bf951b10	293	} else {
0b0f9dc5 MT	294	/* Second best: one for change, shared for all vqs. */
0b0f9dc5 MT	295	nvectors = 2;
07ec5148 CH	296	}
07ec5148 CH	297
0b0f9dc5 MT	298	err = vp_request_msix_vectors(vdev, nvectors, per_vq_vectors,
0b0f9dc5 MT	299	per_vq_vectors ? desc : NULL);
a3cbec69	300	if (err)
0b0f9dc5	301	goto error_find;
a3cbec69	302
0a9b3f47	303	vp_dev->per_vq_vectors = per_vq_vectors;
0b0f9dc5	304	allocated_vectors = vp_dev->msix_used_vectors;
d2a7ddda	305	for (i = 0; i < nvqs; ++i) {
6457f126 MT	306	if (!names[i]) {
	307	vqs[i] = NULL;
	308	continue;
a3cbec69 CH	309	}
a3cbec69 CH	310
0b0f9dc5	311	if (!callbacks[i])
07ec5148	312	msix_vec = VIRTIO_MSI_NO_VECTOR;
0a9b3f47	313	else if (vp_dev->per_vq_vectors)
0b0f9dc5 MT	314	msix_vec = allocated_vectors++;
	315	else
	316	msix_vec = VP_MSIX_VQ_VECTOR;
0a9b3f47 MT	317	vqs[i] = vp_setup_vq(vdev, i, callbacks[i], names[i],
0a9b3f47 MT	318	msix_vec);
e969fed5 MT	319	if (IS_ERR(vqs[i])) {
e969fed5 MT	320	err = PTR_ERR(vqs[i]);
0b0f9dc5	321	goto error_find;
e969fed5	322	}
0b22bd0b	323
0a9b3f47	324	if (!vp_dev->per_vq_vectors \|\| msix_vec == VIRTIO_MSI_NO_VECTOR)
5c34d002	325	continue;
5c34d002	326
0b0f9dc5 MT	327	/* allocate per-vq irq if available and necessary */
	328	snprintf(vp_dev->msix_names[msix_vec],
	329	sizeof *vp_dev->msix_names,
	330	"%s-%s",
0b22bd0b	331	dev_name(&vp_dev->vdev.dev), names[i]);
fa3a3279	332	err = request_irq(pci_irq_vector(vp_dev->pci_dev, msix_vec),
0b0f9dc5 MT	333	vring_interrupt, 0,
	334	vp_dev->msix_names[msix_vec],
	335	vqs[i]);
0a9b3f47	336	if (err)
0b0f9dc5	337	goto error_find;
d2a7ddda MT	338	}
	339	return 0;
	340
0b0f9dc5 MT	341	error_find:
0b0f9dc5 MT	342	vp_del_vqs(vdev);
e969fed5 MT	343	return err;
	344	}
	345
a3cbec69 CH	346	static int vp_find_vqs_intx(struct virtio_device *vdev, unsigned nvqs,
	347	struct virtqueue vqs[], vq_callback_t callbacks[],
	348	const char * const names[])
	349	{
	350	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	351	int i, err;
	352
0a9b3f47 MT	353	vp_dev->vqs = kcalloc(nvqs, sizeof(*vp_dev->vqs), GFP_KERNEL);
	354	if (!vp_dev->vqs)
	355	return -ENOMEM;
	356
a3cbec69 CH	357	err = request_irq(vp_dev->pci_dev->irq, vp_interrupt, IRQF_SHARED,
	358	dev_name(&vdev->dev), vp_dev);
	359	if (err)
0b0f9dc5	360	goto out_del_vqs;
a3cbec69	361
0b0f9dc5	362	vp_dev->intx_enabled = 1;
0a9b3f47	363	vp_dev->per_vq_vectors = false;
a3cbec69 CH	364	for (i = 0; i < nvqs; ++i) {
	365	if (!names[i]) {
	366	vqs[i] = NULL;
	367	continue;
	368	}
0a9b3f47 MT	369	vqs[i] = vp_setup_vq(vdev, i, callbacks[i], names[i],
0a9b3f47 MT	370	VIRTIO_MSI_NO_VECTOR);
a3cbec69 CH	371	if (IS_ERR(vqs[i])) {
a3cbec69 CH	372	err = PTR_ERR(vqs[i]);
0b0f9dc5	373	goto out_del_vqs;
a3cbec69 CH	374	}
	375	}
	376
	377	return 0;
0b0f9dc5 MT	378	out_del_vqs:
0b0f9dc5 MT	379	vp_del_vqs(vdev);
a3cbec69 CH	380	return err;
	381	}
	382
e969fed5	383	/* the config->find_vqs() implementation */
38eb4a29	384	int vp_find_vqs(struct virtio_device *vdev, unsigned nvqs,
fb5e31d9 CH	385	struct virtqueue vqs[], vq_callback_t callbacks[],
fb5e31d9 CH	386	const char * const names[], struct irq_affinity *desc)
e969fed5	387	{
f68d2408	388	int err;
e969fed5	389
bf951b10 MT	390	/* Try MSI-X with one vector per queue. */
	391	err = vp_find_vqs_msix(vdev, nvqs, vqs, callbacks, names, true, desc);
	392	if (!err)
	393	return 0;
	394	/* Fallback: MSI-X with one vector for config, one shared for queues. */
	395	err = vp_find_vqs_msix(vdev, nvqs, vqs, callbacks, names, false, desc);
e969fed5 MT	396	if (!err)
e969fed5 MT	397	return 0;
bf951b10	398	/* Finally fall back to regular interrupts. */
a3cbec69	399	return vp_find_vqs_intx(vdev, nvqs, vqs, callbacks, names);
d2a7ddda MT	400	}
d2a7ddda MT	401
38eb4a29	402	const char vp_bus_name(struct virtio_device vdev)
66846048 RJ	403	{
	404	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	405
	406	return pci_name(vp_dev->pci_dev);
	407	}
	408
75a0a52b JW	409	/* Setup the affinity for a virtqueue:
	410	* - force the affinity for per vq vector
	411	* - OR over all affinities for shared MSI
	412	* - ignore the affinity request if we're using INTX
	413	*/
38eb4a29	414	int vp_set_vq_affinity(struct virtqueue *vq, int cpu)
75a0a52b JW	415	{
	416	struct virtio_device *vdev = vq->vdev;
	417	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
0a9b3f47 MT	418	struct virtio_pci_vq_info *info = vp_dev->vqs[vq->index];
	419	struct cpumask *mask;
	420	unsigned int irq;
75a0a52b JW	421
	422	if (!vq->callback)
	423	return -EINVAL;
	424
2008c154	425	if (vp_dev->msix_enabled) {
0a9b3f47 MT	426	mask = vp_dev->msix_affinity_masks[info->msix_vector];
0a9b3f47 MT	427	irq = pci_irq_vector(vp_dev->pci_dev, info->msix_vector);
75a0a52b JW	428	if (cpu == -1)
	429	irq_set_affinity_hint(irq, NULL);
	430	else {
210d150e	431	cpumask_clear(mask);
75a0a52b JW	432	cpumask_set_cpu(cpu, mask);
	433	irq_set_affinity_hint(irq, mask);
	434	}
	435	}
	436	return 0;
	437	}
	438
bbaba479 CH	439	const struct cpumask vp_get_vq_affinity(struct virtio_device vdev, int index)
	440	{
	441	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
bbaba479	442
0a9b3f47 MT	443	if (!vp_dev->per_vq_vectors \|\|
0a9b3f47 MT	444	vp_dev->vqs[index]->msix_vector == VIRTIO_MSI_NO_VECTOR)
bbaba479 CH	445	return NULL;
bbaba479 CH	446
0a9b3f47 MT	447	return pci_irq_get_affinity(vp_dev->pci_dev,
0a9b3f47 MT	448	vp_dev->vqs[index]->msix_vector);
bbaba479 CH	449	}
bbaba479 CH	450
9e266ece	451	#ifdef CONFIG_PM_SLEEP
f0fe6f11 AS	452	static int virtio_pci_freeze(struct device *dev)
	453	{
	454	struct pci_dev *pci_dev = to_pci_dev(dev);
	455	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
f0fe6f11 AS	456	int ret;
f0fe6f11 AS	457
c6716bae	458	ret = virtio_device_freeze(&vp_dev->vdev);
f0fe6f11 AS	459
	460	if (!ret)
	461	pci_disable_device(pci_dev);
	462	return ret;
	463	}
	464
0517fdd1	465	static int virtio_pci_restore(struct device *dev)
f0fe6f11 AS	466	{
	467	struct pci_dev *pci_dev = to_pci_dev(dev);
	468	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
	469	int ret;
	470
	471	ret = pci_enable_device(pci_dev);
	472	if (ret)
	473	return ret;
0517fdd1	474
f0fe6f11	475	pci_set_master(pci_dev);
c6716bae	476	return virtio_device_restore(&vp_dev->vdev);
f0fe6f11 AS	477	}
f0fe6f11 AS	478
9a4253db	479	static const struct dev_pm_ops virtio_pci_pm_ops = {
f878d0be	480	SET_SYSTEM_SLEEP_PM_OPS(virtio_pci_freeze, virtio_pci_restore)
d0775363	481	};
3343660d	482	#endif
9a4253db MT	483
	484
	485	/* Qumranet donated their vendor ID for devices 0x1000 thru 0x10FF. */
	486	static const struct pci_device_id virtio_pci_id_table[] = {
caf02abf	487	{ PCI_DEVICE(PCI_VENDOR_ID_REDHAT_QUMRANET, PCI_ANY_ID) },
9a4253db MT	488	{ 0 }
	489	};
	490
	491	MODULE_DEVICE_TABLE(pci, virtio_pci_id_table);
	492
ff31d2e2 MT	493	static void virtio_pci_release_dev(struct device *_d)
	494	{
	495	struct virtio_device *vdev = dev_to_virtio(_d);
	496	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	497
	498	/* As struct device is a kobject, it's not safe to
	499	* free the memory (including the reference counter itself)
	500	* until it's release callback. */
	501	kfree(vp_dev);
	502	}
	503
9a4253db MT	504	static int virtio_pci_probe(struct pci_dev *pci_dev,
	505	const struct pci_device_id *id)
	506	{
ff31d2e2 MT	507	struct virtio_pci_device *vp_dev;
	508	int rc;
	509
	510	/* allocate our structure and fill it out */
	511	vp_dev = kzalloc(sizeof(struct virtio_pci_device), GFP_KERNEL);
	512	if (!vp_dev)
	513	return -ENOMEM;
	514
	515	pci_set_drvdata(pci_dev, vp_dev);
	516	vp_dev->vdev.dev.parent = &pci_dev->dev;
	517	vp_dev->vdev.dev.release = virtio_pci_release_dev;
	518	vp_dev->pci_dev = pci_dev;
0a9b3f47 MT	519	INIT_LIST_HEAD(&vp_dev->virtqueues);
0a9b3f47 MT	520	spin_lock_init(&vp_dev->lock);
ff31d2e2	521
ff31d2e2 MT	522	/* enable the device */
	523	rc = pci_enable_device(pci_dev);
	524	if (rc)
	525	goto err_enable_device;
	526
ac399d8f	527	if (force_legacy) {
1fcf0512	528	rc = virtio_pci_legacy_probe(vp_dev);
ac399d8f MT	529	/* Also try modern mode if we can't map BAR0 (no IO space). */
	530	if (rc == -ENODEV \|\| rc == -ENOMEM)
	531	rc = virtio_pci_modern_probe(vp_dev);
	532	if (rc)
	533	goto err_probe;
	534	} else {
	535	rc = virtio_pci_modern_probe(vp_dev);
	536	if (rc == -ENODEV)
	537	rc = virtio_pci_legacy_probe(vp_dev);
	538	if (rc)
	539	goto err_probe;
	540	}
ff31d2e2 MT	541
	542	pci_set_master(pci_dev);
	543
	544	rc = register_virtio_device(&vp_dev->vdev);
	545	if (rc)
	546	goto err_register;
	547
	548	return 0;
	549
	550	err_register:
1fcf0512 MT	551	if (vp_dev->ioaddr)
	552	virtio_pci_legacy_remove(vp_dev);
	553	else
	554	virtio_pci_modern_remove(vp_dev);
ff31d2e2	555	err_probe:
ff31d2e2 MT	556	pci_disable_device(pci_dev);
	557	err_enable_device:
	558	kfree(vp_dev);
	559	return rc;
9a4253db MT	560	}
	561
	562	static void virtio_pci_remove(struct pci_dev *pci_dev)
	563	{
ff31d2e2	564	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
2989be09	565	struct device *dev = get_device(&vp_dev->vdev.dev);
ff31d2e2 MT	566
	567	unregister_virtio_device(&vp_dev->vdev);
	568
1fcf0512 MT	569	if (vp_dev->ioaddr)
	570	virtio_pci_legacy_remove(vp_dev);
	571	else
	572	virtio_pci_modern_remove(vp_dev);
ff31d2e2	573
ff31d2e2	574	pci_disable_device(pci_dev);
2989be09	575	put_device(dev);
9a4253db MT	576	}
	577
	578	static struct pci_driver virtio_pci_driver = {
	579	.name = "virtio-pci",
	580	.id_table = virtio_pci_id_table,
	581	.probe = virtio_pci_probe,
	582	.remove = virtio_pci_remove,
	583	#ifdef CONFIG_PM_SLEEP
	584	.driver.pm = &virtio_pci_pm_ops,
	585	#endif
	586	};
	587
	588	module_pci_driver(virtio_pci_driver);
5ff16110 HX	589
	590	MODULE_AUTHOR("Anthony Liguori <aliguori@us.ibm.com>");
	591	MODULE_DESCRIPTION("virtio-pci");
	592	MODULE_LICENSE("GPL");
	593	MODULE_VERSION("1");