virtio_pci_common.c 13.1 KB
Newer Older
Anthony Liguori's avatar
Anthony Liguori committed
1
/*
2
 * Virtio PCI driver - common functionality for all device versions
Anthony Liguori's avatar
Anthony Liguori committed
3 4 5 6 7
 *
 * 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
8
 * Copyright Red Hat, Inc. 2014
Anthony Liguori's avatar
Anthony Liguori committed
9 10 11
 *
 * Authors:
 *  Anthony Liguori  <aliguori@us.ibm.com>
12 13
 *  Rusty Russell <rusty@rustcorp.com.au>
 *  Michael S. Tsirkin <mst@redhat.com>
Anthony Liguori's avatar
Anthony Liguori committed
14 15 16 17 18 19
 *
 * This work is licensed under the terms of the GNU GPL, version 2 or later.
 * See the COPYING file in the top-level directory.
 *
 */

20
#include "virtio_pci_common.h"
Anthony Liguori's avatar
Anthony Liguori committed
21

22 23 24 25 26 27 28 29
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

30
/* wait for pending irq handlers */
31
void vp_synchronize_vectors(struct virtio_device *vdev)
32 33 34 35
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	int i;

36 37
	synchronize_irq(pci_irq_vector(vp_dev->pci_dev, 0));
	for (i = 1; i < vp_dev->msix_vectors; i++)
38
		synchronize_irq(pci_irq_vector(vp_dev->pci_dev, i));
39 40
}

Anthony Liguori's avatar
Anthony Liguori committed
41
/* the notify function used when creating a virt queue */
42
bool vp_notify(struct virtqueue *vq)
Anthony Liguori's avatar
Anthony Liguori committed
43 44 45
{
	/* we write the queue's selector into the notification register to
	 * signal the other end */
46
	iowrite16(vq->index, (void __iomem *)vq->priv);
47
	return true;
Anthony Liguori's avatar
Anthony Liguori committed
48 49
}

50 51 52 53 54
/* 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;

55
	virtio_config_changed(&vp_dev->vdev);
56 57 58 59 60 61 62 63
	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;
	irqreturn_t ret = IRQ_NONE;
64
	struct virtqueue *vq;
65

66 67
	list_for_each_entry(vq, &vp_dev->vdev.vqs, list) {
		if (vq->callback && vring_interrupt(irq, vq) == IRQ_HANDLED)
68 69 70 71 72 73
			ret = IRQ_HANDLED;
	}

	return ret;
}

Anthony Liguori's avatar
Anthony Liguori committed
74 75 76 77 78 79 80 81 82 83 84 85 86
/* 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. */
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
87
	isr = ioread8(vp_dev->isr);
Anthony Liguori's avatar
Anthony Liguori committed
88 89 90 91 92 93

	/* 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. */
94 95
	if (isr & VIRTIO_PCI_ISR_CONFIG)
		vp_config_changed(irq, opaque);
Anthony Liguori's avatar
Anthony Liguori committed
96

97
	return vp_vring_interrupt(irq, opaque);
Anthony Liguori's avatar
Anthony Liguori committed
98 99
}

100
static void vp_remove_vqs(struct virtio_device *vdev)
101
{
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
102
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
103 104
	struct virtqueue *vq, *n;

Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
105
	list_for_each_entry_safe(vq, n, &vdev->vqs, list) {
106 107
		if (vp_dev->msix_vector_map) {
			int v = vp_dev->msix_vector_map[vq->index];
108 109 110 111 112

			if (v != VIRTIO_MSI_NO_VECTOR)
				free_irq(pci_irq_vector(vp_dev->pci_dev, v),
					vq);
		}
113
		vp_dev->del_vq(vq);
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
114
	}
115
}
116

117 118 119 120 121
/* the config->del_vqs() implementation */
void vp_del_vqs(struct virtio_device *vdev)
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	int i;
122

123 124
	if (WARN_ON_ONCE(list_empty_careful(&vdev->vqs)))
		return;
125

126
	vp_remove_vqs(vdev);
127

128
	if (vp_dev->msix_enabled) {
129 130 131
		for (i = 0; i < vp_dev->msix_vectors; i++)
			free_cpumask_var(vp_dev->msix_affinity_masks[i]);

132 133 134
		/* Disable the vector used for configuration */
		vp_dev->config_vector(vp_dev, VIRTIO_MSI_NO_VECTOR);

135 136 137
		kfree(vp_dev->msix_affinity_masks);
		kfree(vp_dev->msix_names);
		kfree(vp_dev->msix_vector_map);
138 139
	}

140 141
	free_irq(pci_irq_vector(vp_dev->pci_dev, 0), vp_dev);
	pci_free_irq_vectors(vp_dev->pci_dev);
142 143
}

144
static int vp_find_vqs_msix(struct virtio_device *vdev, unsigned nvqs,
145
		struct virtqueue *vqs[], vq_callback_t *callbacks[],
146 147
		const char * const names[], bool per_vq_vectors,
		struct irq_affinity *desc)
148
{
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
149
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
150
	const char *name = dev_name(&vp_dev->vdev.dev);
151
	int i, err = -ENOMEM, allocated_vectors, nvectors;
152
	unsigned flags = PCI_IRQ_MSIX;
153
	u16 msix_vec;
154

155 156 157 158 159
	if (desc) {
		flags |= PCI_IRQ_AFFINITY;
		desc->pre_vectors++; /* virtio config vector */
	}

160 161 162 163
	nvectors = 1;
	for (i = 0; i < nvqs; i++)
		if (callbacks[i])
			nvectors++;
164

165 166 167 168
	if (per_vq_vectors) {
		err = pci_alloc_irq_vectors_affinity(vp_dev->pci_dev, nvectors,
				nvectors, flags, desc);
	} else {
169 170
		err = pci_alloc_irq_vectors(vp_dev->pci_dev, 2, 2,
				PCI_IRQ_MSIX);
171
	}
172 173
	if (err < 0)
		return err;
174

175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196
	vp_dev->msix_vectors = nvectors;
	vp_dev->msix_names = kmalloc_array(nvectors,
			sizeof(*vp_dev->msix_names), GFP_KERNEL);
	if (!vp_dev->msix_names)
		goto out_free_irq_vectors;

	vp_dev->msix_affinity_masks = kcalloc(nvectors,
			sizeof(*vp_dev->msix_affinity_masks), GFP_KERNEL);
	if (!vp_dev->msix_affinity_masks)
		goto out_free_msix_names;

	for (i = 0; i < nvectors; ++i) {
		if (!alloc_cpumask_var(&vp_dev->msix_affinity_masks[i],
				GFP_KERNEL))
			goto out_free_msix_affinity_masks;
	}

	/* Set the vector used for configuration */
	snprintf(vp_dev->msix_names[0], sizeof(*vp_dev->msix_names),
		 "%s-config", name);
	err = request_irq(pci_irq_vector(vp_dev->pci_dev, 0), vp_config_changed,
			0, vp_dev->msix_names[0], vp_dev);
197
	if (err)
198
		goto out_free_irq_vectors;
199

200 201 202 203
	/* Verify we had enough resources to assign the vector */
	if (vp_dev->config_vector(vp_dev, 0) == VIRTIO_MSI_NO_VECTOR) {
		err = -EBUSY;
		goto out_free_config_irq;
204 205
	}

206 207 208 209 210
	vp_dev->msix_vector_map = kmalloc_array(nvqs,
			sizeof(*vp_dev->msix_vector_map), GFP_KERNEL);
	if (!vp_dev->msix_vector_map)
		goto out_disable_config_irq;

211
	allocated_vectors = 1; /* vector 0 is the config interrupt */
212
	for (i = 0; i < nvqs; ++i) {
213 214 215
		if (!names[i]) {
			vqs[i] = NULL;
			continue;
216 217
		}

218 219
		if (callbacks[i])
			msix_vec = allocated_vectors;
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
220
		else
221 222
			msix_vec = VIRTIO_MSI_NO_VECTOR;

223 224
		vqs[i] = vp_dev->setup_vq(vp_dev, i, callbacks[i], names[i],
				msix_vec);
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
225 226
		if (IS_ERR(vqs[i])) {
			err = PTR_ERR(vqs[i]);
227
			goto out_remove_vqs;
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
228
		}
229

230 231 232 233 234
		if (msix_vec == VIRTIO_MSI_NO_VECTOR) {
			vp_dev->msix_vector_map[i] = VIRTIO_MSI_NO_VECTOR;
			continue;
		}

235
		snprintf(vp_dev->msix_names[i + 1],
236
			 sizeof(*vp_dev->msix_names), "%s-%s",
237
			 dev_name(&vp_dev->vdev.dev), names[i]);
238
		err = request_irq(pci_irq_vector(vp_dev->pci_dev, msix_vec),
239
				  vring_interrupt, IRQF_SHARED,
240
				  vp_dev->msix_names[i + 1], vqs[i]);
241 242 243
		if (err) {
			/* don't free this irq on error */
			vp_dev->msix_vector_map[i] = VIRTIO_MSI_NO_VECTOR;
244
			goto out_remove_vqs;
245 246
		}
		vp_dev->msix_vector_map[i] = msix_vec;
247

248
		if (per_vq_vectors)
249
			allocated_vectors++;
250
	}
251

252
	vp_dev->msix_enabled = 1;
253 254
	return 0;

255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271
out_remove_vqs:
	vp_remove_vqs(vdev);
	kfree(vp_dev->msix_vector_map);
out_disable_config_irq:
	vp_dev->config_vector(vp_dev, VIRTIO_MSI_NO_VECTOR);
out_free_config_irq:
	free_irq(pci_irq_vector(vp_dev->pci_dev, 0), vp_dev);
out_free_msix_affinity_masks:
	for (i = 0; i < nvectors; i++) {
		if (vp_dev->msix_affinity_masks[i])
			free_cpumask_var(vp_dev->msix_affinity_masks[i]);
	}
	kfree(vp_dev->msix_affinity_masks);
out_free_msix_names:
	kfree(vp_dev->msix_names);
out_free_irq_vectors:
	pci_free_irq_vectors(vp_dev->pci_dev);
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
272 273 274
	return err;
}

275 276 277 278 279 280 281 282 283 284
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;

	err = request_irq(vp_dev->pci_dev->irq, vp_interrupt, IRQF_SHARED,
			dev_name(&vdev->dev), vp_dev);
	if (err)
285
		return err;
286 287 288 289 290 291

	for (i = 0; i < nvqs; ++i) {
		if (!names[i]) {
			vqs[i] = NULL;
			continue;
		}
292
		vqs[i] = vp_dev->setup_vq(vp_dev, i, callbacks[i], names[i],
293 294 295
				VIRTIO_MSI_NO_VECTOR);
		if (IS_ERR(vqs[i])) {
			err = PTR_ERR(vqs[i]);
296
			goto out_remove_vqs;
297 298 299 300
		}
	}

	return 0;
301 302 303 304

out_remove_vqs:
	vp_remove_vqs(vdev);
	free_irq(pci_irq_vector(vp_dev->pci_dev, 0), vp_dev);
305 306 307
	return err;
}

Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
308
/* the config->find_vqs() implementation */
309
int vp_find_vqs(struct virtio_device *vdev, unsigned nvqs,
310 311
		struct virtqueue *vqs[], vq_callback_t *callbacks[],
		const char * const names[], struct irq_affinity *desc)
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
312
{
313
	int err;
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
314

315 316 317 318 319 320
	/* 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);
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
321 322
	if (!err)
		return 0;
323
	/* Finally fall back to regular interrupts. */
324
	return vp_find_vqs_intx(vdev, nvqs, vqs, callbacks, names);
325 326
}

327
const char *vp_bus_name(struct virtio_device *vdev)
328 329 330 331 332 333
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);

	return pci_name(vp_dev->pci_dev);
}

334 335 336 337 338
/* 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
 */
339
int vp_set_vq_affinity(struct virtqueue *vq, int cpu)
340 341 342 343 344 345 346
{
	struct virtio_device *vdev = vq->vdev;
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);

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

347
	if (vp_dev->msix_enabled) {
348 349 350 351
		int vec = vp_dev->msix_vector_map[vq->index];
		struct cpumask *mask = vp_dev->msix_affinity_masks[vec];
		unsigned int irq = pci_irq_vector(vp_dev->pci_dev, vec);

352 353 354
		if (cpu == -1)
			irq_set_affinity_hint(irq, NULL);
		else {
355
			cpumask_clear(mask);
356 357 358 359 360 361 362
			cpumask_set_cpu(cpu, mask);
			irq_set_affinity_hint(irq, mask);
		}
	}
	return 0;
}

363 364 365 366 367 368 369 370 371 372 373
const struct cpumask *vp_get_vq_affinity(struct virtio_device *vdev, int index)
{
	struct virtio_pci_device *vp_dev = to_vp_device(vdev);
	unsigned int *map = vp_dev->msix_vector_map;

	if (!map || map[index] == VIRTIO_MSI_NO_VECTOR)
		return NULL;

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

374
#ifdef CONFIG_PM_SLEEP
375 376 377 378 379 380
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;

381
	ret = virtio_device_freeze(&vp_dev->vdev);
382 383 384 385 386 387

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

Amit Shah's avatar
Amit Shah committed
388
static int virtio_pci_restore(struct device *dev)
389 390 391 392 393 394 395 396
{
	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;
Amit Shah's avatar
Amit Shah committed
397

398
	pci_set_master(pci_dev);
399
	return virtio_device_restore(&vp_dev->vdev);
400 401
}

402
static const struct dev_pm_ops virtio_pci_pm_ops = {
403
	SET_SYSTEM_SLEEP_PM_OPS(virtio_pci_freeze, virtio_pci_restore)
Amit Shah's avatar
Amit Shah committed
404
};
Anthony Liguori's avatar
Anthony Liguori committed
405
#endif
406 407 408 409


/* Qumranet donated their vendor ID for devices 0x1000 thru 0x10FF. */
static const struct pci_device_id virtio_pci_id_table[] = {
410
	{ PCI_DEVICE(PCI_VENDOR_ID_REDHAT_QUMRANET, PCI_ANY_ID) },
411 412 413 414 415
	{ 0 }
};

MODULE_DEVICE_TABLE(pci, virtio_pci_id_table);

416 417 418 419 420 421 422 423 424 425 426
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);
}

427 428 429
static int virtio_pci_probe(struct pci_dev *pci_dev,
			    const struct pci_device_id *id)
{
430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447
	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;

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

448
	if (force_legacy) {
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
449
		rc = virtio_pci_legacy_probe(vp_dev);
450 451 452 453 454 455 456 457 458 459 460 461
		/* 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;
	}
462 463 464 465 466 467 468 469 470 471

	pci_set_master(pci_dev);

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

	return 0;

err_register:
Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
472 473 474 475
	if (vp_dev->ioaddr)
	     virtio_pci_legacy_remove(vp_dev);
	else
	     virtio_pci_modern_remove(vp_dev);
476 477 478 479 480
err_probe:
	pci_disable_device(pci_dev);
err_enable_device:
	kfree(vp_dev);
	return rc;
481 482 483 484
}

static void virtio_pci_remove(struct pci_dev *pci_dev)
{
485
	struct virtio_pci_device *vp_dev = pci_get_drvdata(pci_dev);
486
	struct device *dev = get_device(&vp_dev->vdev.dev);
487 488 489

	unregister_virtio_device(&vp_dev->vdev);

Michael S. Tsirkin's avatar
Michael S. Tsirkin committed
490 491 492 493
	if (vp_dev->ioaddr)
		virtio_pci_legacy_remove(vp_dev);
	else
		virtio_pci_modern_remove(vp_dev);
494 495

	pci_disable_device(pci_dev);
496
	put_device(dev);
497 498 499 500 501 502 503 504 505 506 507 508 509
}

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);
510 511 512 513 514

MODULE_AUTHOR("Anthony Liguori <aliguori@us.ibm.com>");
MODULE_DESCRIPTION("virtio-pci");
MODULE_LICENSE("GPL");
MODULE_VERSION("1");