msi.c 21.0 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8
/*
 * File:	msi.c
 * Purpose:	PCI Message Signaled Interrupt (MSI)
 *
 * Copyright (C) 2003-2004 Intel
 * Copyright (C) Tom Long Nguyen (tom.l.nguyen@intel.com)
 */

9
#include <linux/err.h>
L
Linus Torvalds 已提交
10 11 12 13 14 15 16
#include <linux/mm.h>
#include <linux/irq.h>
#include <linux/interrupt.h>
#include <linux/init.h>
#include <linux/ioport.h>
#include <linux/pci.h>
#include <linux/proc_fs.h>
17
#include <linux/msi.h>
D
Dan Williams 已提交
18
#include <linux/smp.h>
H
Hidetoshi Seto 已提交
19 20
#include <linux/errno.h>
#include <linux/io.h>
21
#include <linux/slab.h>
L
Linus Torvalds 已提交
22 23 24 25 26 27

#include "pci.h"
#include "msi.h"

static int pci_msi_enable = 1;

28 29
/* Arch hooks */

30 31
#ifndef arch_msi_check_device
int arch_msi_check_device(struct pci_dev *dev, int nvec, int type)
32 33 34
{
	return 0;
}
35
#endif
36

37 38
#ifndef arch_setup_msi_irqs
int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type)
39 40 41 42
{
	struct msi_desc *entry;
	int ret;

43 44 45 46 47 48 49
	/*
	 * If an architecture wants to support multiple MSI, it needs to
	 * override arch_setup_msi_irqs()
	 */
	if (type == PCI_CAP_ID_MSI && nvec > 1)
		return 1;

50 51
	list_for_each_entry(entry, &dev->msi_list, list) {
		ret = arch_setup_msi_irq(dev, entry);
52
		if (ret < 0)
53
			return ret;
54 55
		if (ret > 0)
			return -ENOSPC;
56 57 58 59
	}

	return 0;
}
60
#endif
61

62 63
#ifndef arch_teardown_msi_irqs
void arch_teardown_msi_irqs(struct pci_dev *dev)
64 65 66 67
{
	struct msi_desc *entry;

	list_for_each_entry(entry, &dev->msi_list, list) {
68 69 70 71 72 73
		int i, nvec;
		if (entry->irq == 0)
			continue;
		nvec = 1 << entry->msi_attrib.multiple;
		for (i = 0; i < nvec; i++)
			arch_teardown_msi_irq(entry->irq + i);
74 75
	}
}
76
#endif
77

78
static void msi_set_enable(struct pci_dev *dev, int pos, int enable)
79 80 81
{
	u16 control;

82
	BUG_ON(!pos);
83

84 85 86 87 88
	pci_read_config_word(dev, pos + PCI_MSI_FLAGS, &control);
	control &= ~PCI_MSI_FLAGS_ENABLE;
	if (enable)
		control |= PCI_MSI_FLAGS_ENABLE;
	pci_write_config_word(dev, pos + PCI_MSI_FLAGS, control);
89 90
}

91 92 93 94 95 96 97 98 99 100 101 102 103 104 105
static void msix_set_enable(struct pci_dev *dev, int enable)
{
	int pos;
	u16 control;

	pos = pci_find_capability(dev, PCI_CAP_ID_MSIX);
	if (pos) {
		pci_read_config_word(dev, pos + PCI_MSIX_FLAGS, &control);
		control &= ~PCI_MSIX_FLAGS_ENABLE;
		if (enable)
			control |= PCI_MSIX_FLAGS_ENABLE;
		pci_write_config_word(dev, pos + PCI_MSIX_FLAGS, control);
	}
}

106 107
static inline __attribute_const__ u32 msi_mask(unsigned x)
{
108 109 110 111
	/* Don't shift by >= width of type */
	if (x >= 5)
		return 0xffffffff;
	return (1 << (1 << x)) - 1;
112 113
}

114
static inline __attribute_const__ u32 msi_capable_mask(u16 control)
M
Mitch Williams 已提交
115
{
116 117
	return msi_mask((control >> 1) & 7);
}
M
Mitch Williams 已提交
118

119 120 121
static inline __attribute_const__ u32 msi_enabled_mask(u16 control)
{
	return msi_mask((control >> 4) & 7);
M
Mitch Williams 已提交
122 123
}

124 125 126 127 128 129
/*
 * PCI 2.3 does not specify mask bits for each MSI interrupt.  Attempting to
 * mask all MSI interrupts by clearing the MSI enable bit does not work
 * reliably as devices without an INTx disable bit will then generate a
 * level IRQ which will never be cleared.
 */
130
static u32 __msi_mask_irq(struct msi_desc *desc, u32 mask, u32 flag)
L
Linus Torvalds 已提交
131
{
132
	u32 mask_bits = desc->masked;
L
Linus Torvalds 已提交
133

134
	if (!desc->msi_attrib.maskbit)
135
		return 0;
136 137 138 139

	mask_bits &= ~mask;
	mask_bits |= flag;
	pci_write_config_dword(desc->dev, desc->mask_pos, mask_bits);
140 141 142 143 144 145 146

	return mask_bits;
}

static void msi_mask_irq(struct msi_desc *desc, u32 mask, u32 flag)
{
	desc->masked = __msi_mask_irq(desc, mask, flag);
147 148 149 150 151 152 153 154 155
}

/*
 * This internal function does not flush PCI writes to the device.
 * All users must ensure that they read from the device before either
 * assuming that the device state is up to date, or returning out of this
 * file.  This saves a few milliseconds when initialising devices with lots
 * of MSI-X interrupts.
 */
156
static u32 __msix_mask_irq(struct msi_desc *desc, u32 flag)
157 158 159
{
	u32 mask_bits = desc->masked;
	unsigned offset = desc->msi_attrib.entry_nr * PCI_MSIX_ENTRY_SIZE +
160
						PCI_MSIX_ENTRY_VECTOR_CTRL;
161 162 163
	mask_bits &= ~1;
	mask_bits |= flag;
	writel(mask_bits, desc->mask_base + offset);
164 165 166 167 168 169 170

	return mask_bits;
}

static void msix_mask_irq(struct msi_desc *desc, u32 flag)
{
	desc->masked = __msix_mask_irq(desc, flag);
171
}
172

173 174 175
static void msi_set_mask_bit(unsigned irq, u32 flag)
{
	struct msi_desc *desc = get_irq_msi(irq);
176

177 178 179 180
	if (desc->msi_attrib.is_msix) {
		msix_mask_irq(desc, flag);
		readl(desc->mask_base);		/* Flush write to device */
	} else {
181 182
		unsigned offset = irq - desc->dev->irq;
		msi_mask_irq(desc, 1 << offset, flag << offset);
L
Linus Torvalds 已提交
183
	}
184 185 186 187 188 189 190 191 192 193
}

void mask_msi_irq(unsigned int irq)
{
	msi_set_mask_bit(irq, 1);
}

void unmask_msi_irq(unsigned int irq)
{
	msi_set_mask_bit(irq, 0);
L
Linus Torvalds 已提交
194 195
}

Y
Yinghai Lu 已提交
196
void read_msi_msg_desc(struct irq_desc *desc, struct msi_msg *msg)
L
Linus Torvalds 已提交
197
{
Y
Yinghai Lu 已提交
198
	struct msi_desc *entry = get_irq_desc_msi(desc);
199

200 201 202 203 204 205
	/* We do not touch the hardware (which may not even be
	 * accessible at the moment) but return the last message
	 * written.  Assert that this is valid, assuming that
	 * valid messages are not all-zeroes. */
	BUG_ON(!(entry->msg.address_hi | entry->msg.address_lo |
		 entry->msg.data));
206

207
	*msg = entry->msg;
208
}
L
Linus Torvalds 已提交
209

Y
Yinghai Lu 已提交
210
void read_msi_msg(unsigned int irq, struct msi_msg *msg)
211
{
Y
Yinghai Lu 已提交
212 213 214 215 216 217 218 219
	struct irq_desc *desc = irq_to_desc(irq);

	read_msi_msg_desc(desc, msg);
}

void write_msi_msg_desc(struct irq_desc *desc, struct msi_msg *msg)
{
	struct msi_desc *entry = get_irq_desc_msi(desc);
220 221 222 223

	if (entry->dev->current_state != PCI_D0) {
		/* Don't touch the hardware now */
	} else if (entry->msi_attrib.is_msix) {
224 225 226 227
		void __iomem *base;
		base = entry->mask_base +
			entry->msi_attrib.entry_nr * PCI_MSIX_ENTRY_SIZE;

228 229 230
		writel(msg->address_lo, base + PCI_MSIX_ENTRY_LOWER_ADDR);
		writel(msg->address_hi, base + PCI_MSIX_ENTRY_UPPER_ADDR);
		writel(msg->data, base + PCI_MSIX_ENTRY_DATA);
231
	} else {
232 233
		struct pci_dev *dev = entry->dev;
		int pos = entry->msi_attrib.pos;
234 235 236 237 238 239
		u16 msgctl;

		pci_read_config_word(dev, msi_control_reg(pos), &msgctl);
		msgctl &= ~PCI_MSI_FLAGS_QSIZE;
		msgctl |= entry->msi_attrib.multiple << 4;
		pci_write_config_word(dev, msi_control_reg(pos), msgctl);
240 241 242 243 244 245 246 247 248 249 250 251

		pci_write_config_dword(dev, msi_lower_address_reg(pos),
					msg->address_lo);
		if (entry->msi_attrib.is_64) {
			pci_write_config_dword(dev, msi_upper_address_reg(pos),
						msg->address_hi);
			pci_write_config_word(dev, msi_data_reg(pos, 1),
						msg->data);
		} else {
			pci_write_config_word(dev, msi_data_reg(pos, 0),
						msg->data);
		}
L
Linus Torvalds 已提交
252
	}
253
	entry->msg = *msg;
L
Linus Torvalds 已提交
254
}
255

Y
Yinghai Lu 已提交
256 257 258 259 260 261 262
void write_msi_msg(unsigned int irq, struct msi_msg *msg)
{
	struct irq_desc *desc = irq_to_desc(irq);

	write_msi_msg_desc(desc, msg);
}

263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286
static void free_msi_irqs(struct pci_dev *dev)
{
	struct msi_desc *entry, *tmp;

	list_for_each_entry(entry, &dev->msi_list, list) {
		int i, nvec;
		if (!entry->irq)
			continue;
		nvec = 1 << entry->msi_attrib.multiple;
		for (i = 0; i < nvec; i++)
			BUG_ON(irq_has_action(entry->irq + i));
	}

	arch_teardown_msi_irqs(dev);

	list_for_each_entry_safe(entry, tmp, &dev->msi_list, list) {
		if (entry->msi_attrib.is_msix) {
			if (list_is_last(&entry->list, &dev->msi_list))
				iounmap(entry->mask_base);
		}
		list_del(&entry->list);
		kfree(entry);
	}
}
S
Satoru Takeuchi 已提交
287

288
static struct msi_desc *alloc_msi_entry(struct pci_dev *dev)
L
Linus Torvalds 已提交
289
{
290 291
	struct msi_desc *desc = kzalloc(sizeof(*desc), GFP_KERNEL);
	if (!desc)
L
Linus Torvalds 已提交
292 293
		return NULL;

294 295
	INIT_LIST_HEAD(&desc->list);
	desc->dev = dev;
L
Linus Torvalds 已提交
296

297
	return desc;
L
Linus Torvalds 已提交
298 299
}

300 301 302 303 304 305
static void pci_intx_for_msi(struct pci_dev *dev, int enable)
{
	if (!(dev->dev_flags & PCI_DEV_FLAGS_MSI_INTX_DISABLE_BUG))
		pci_intx(dev, enable);
}

306
static void __pci_restore_msi_state(struct pci_dev *dev)
307
{
308
	int pos;
309
	u16 control;
310
	struct msi_desc *entry;
311

312 313 314
	if (!dev->msi_enabled)
		return;

315 316
	entry = get_irq_msi(dev->irq);
	pos = entry->msi_attrib.pos;
317

318
	pci_intx_for_msi(dev, 0);
319
	msi_set_enable(dev, pos, 0);
320 321 322
	write_msi_msg(dev->irq, &entry->msg);

	pci_read_config_word(dev, pos + PCI_MSI_FLAGS, &control);
323
	msi_mask_irq(entry, msi_capable_mask(control), entry->masked);
324
	control &= ~PCI_MSI_FLAGS_QSIZE;
325
	control |= (entry->msi_attrib.multiple << 4) | PCI_MSI_FLAGS_ENABLE;
326
	pci_write_config_word(dev, pos + PCI_MSI_FLAGS, control);
327 328 329
}

static void __pci_restore_msix_state(struct pci_dev *dev)
330 331 332
{
	int pos;
	struct msi_desc *entry;
333
	u16 control;
334

E
Eric W. Biederman 已提交
335 336
	if (!dev->msix_enabled)
		return;
337
	BUG_ON(list_empty(&dev->msi_list));
H
Hidetoshi Seto 已提交
338
	entry = list_first_entry(&dev->msi_list, struct msi_desc, list);
339 340
	pos = entry->msi_attrib.pos;
	pci_read_config_word(dev, pos + PCI_MSIX_FLAGS, &control);
E
Eric W. Biederman 已提交
341

342
	/* route the table */
343
	pci_intx_for_msi(dev, 0);
344 345
	control |= PCI_MSIX_FLAGS_ENABLE | PCI_MSIX_FLAGS_MASKALL;
	pci_write_config_word(dev, pos + PCI_MSIX_FLAGS, control);
346

347 348
	list_for_each_entry(entry, &dev->msi_list, list) {
		write_msi_msg(entry->irq, &entry->msg);
349
		msix_mask_irq(entry, entry->masked);
350 351
	}

352 353
	control &= ~PCI_MSIX_FLAGS_MASKALL;
	pci_write_config_word(dev, pos + PCI_MSIX_FLAGS, control);
354
}
355 356 357 358 359 360

void pci_restore_msi_state(struct pci_dev *dev)
{
	__pci_restore_msi_state(dev);
	__pci_restore_msix_state(dev);
}
361
EXPORT_SYMBOL_GPL(pci_restore_msi_state);
362

L
Linus Torvalds 已提交
363 364 365
/**
 * msi_capability_init - configure device's MSI capability structure
 * @dev: pointer to the pci_dev data structure of MSI device function
366
 * @nvec: number of interrupts to allocate
L
Linus Torvalds 已提交
367
 *
368 369 370 371 372 373 374
 * Setup the MSI capability structure of the device with the requested
 * number of interrupts.  A return value of zero indicates the successful
 * setup of an entry with the new MSI irq.  A negative return value indicates
 * an error, and a positive return value indicates the number of interrupts
 * which could have been allocated.
 */
static int msi_capability_init(struct pci_dev *dev, int nvec)
L
Linus Torvalds 已提交
375 376
{
	struct msi_desc *entry;
377
	int pos, ret;
L
Linus Torvalds 已提交
378
	u16 control;
379
	unsigned mask;
L
Linus Torvalds 已提交
380

H
Hidetoshi Seto 已提交
381
	pos = pci_find_capability(dev, PCI_CAP_ID_MSI);
382 383
	msi_set_enable(dev, pos, 0);	/* Disable MSI during set up */

L
Linus Torvalds 已提交
384 385
	pci_read_config_word(dev, msi_control_reg(pos), &control);
	/* MSI Entry Initialization */
386
	entry = alloc_msi_entry(dev);
387 388
	if (!entry)
		return -ENOMEM;
389

H
Hidetoshi Seto 已提交
390 391 392 393 394 395
	entry->msi_attrib.is_msix	= 0;
	entry->msi_attrib.is_64		= is_64bit_address(control);
	entry->msi_attrib.entry_nr	= 0;
	entry->msi_attrib.maskbit	= is_mask_bit_support(control);
	entry->msi_attrib.default_irq	= dev->irq;	/* Save IOAPIC IRQ */
	entry->msi_attrib.pos		= pos;
396

397
	entry->mask_pos = msi_mask_reg(pos, entry->msi_attrib.is_64);
398 399 400 401 402 403
	/* All MSIs are unmasked by default, Mask them all */
	if (entry->msi_attrib.maskbit)
		pci_read_config_dword(dev, entry->mask_pos, &entry->masked);
	mask = msi_capable_mask(control);
	msi_mask_irq(entry, mask, mask);

404
	list_add_tail(&entry->list, &dev->msi_list);
405

L
Linus Torvalds 已提交
406
	/* Configure MSI capability structure */
407
	ret = arch_setup_msi_irqs(dev, nvec, PCI_CAP_ID_MSI);
408
	if (ret) {
409
		msi_mask_irq(entry, mask, ~mask);
410
		free_msi_irqs(dev);
411
		return ret;
412
	}
413

L
Linus Torvalds 已提交
414
	/* Set MSI enabled bits	 */
415
	pci_intx_for_msi(dev, 0);
416
	msi_set_enable(dev, pos, 1);
417
	dev->msi_enabled = 1;
L
Linus Torvalds 已提交
418

419
	dev->irq = entry->irq;
L
Linus Torvalds 已提交
420 421 422
	return 0;
}

423 424 425
static void __iomem *msix_map_region(struct pci_dev *dev, unsigned pos,
							unsigned nr_entries)
{
426
	resource_size_t phys_addr;
427 428 429 430 431 432 433 434 435 436 437
	u32 table_offset;
	u8 bir;

	pci_read_config_dword(dev, msix_table_offset_reg(pos), &table_offset);
	bir = (u8)(table_offset & PCI_MSIX_FLAGS_BIRMASK);
	table_offset &= ~PCI_MSIX_FLAGS_BIRMASK;
	phys_addr = pci_resource_start(dev, bir) + table_offset;

	return ioremap_nocache(phys_addr, nr_entries * PCI_MSIX_ENTRY_SIZE);
}

438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468
static int msix_setup_entries(struct pci_dev *dev, unsigned pos,
				void __iomem *base, struct msix_entry *entries,
				int nvec)
{
	struct msi_desc *entry;
	int i;

	for (i = 0; i < nvec; i++) {
		entry = alloc_msi_entry(dev);
		if (!entry) {
			if (!i)
				iounmap(base);
			else
				free_msi_irqs(dev);
			/* No enough memory. Don't try again */
			return -ENOMEM;
		}

		entry->msi_attrib.is_msix	= 1;
		entry->msi_attrib.is_64		= 1;
		entry->msi_attrib.entry_nr	= entries[i].entry;
		entry->msi_attrib.default_irq	= dev->irq;
		entry->msi_attrib.pos		= pos;
		entry->mask_base		= base;

		list_add_tail(&entry->list, &dev->msi_list);
	}

	return 0;
}

469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486
static void msix_program_entries(struct pci_dev *dev,
					struct msix_entry *entries)
{
	struct msi_desc *entry;
	int i = 0;

	list_for_each_entry(entry, &dev->msi_list, list) {
		int offset = entries[i].entry * PCI_MSIX_ENTRY_SIZE +
						PCI_MSIX_ENTRY_VECTOR_CTRL;

		entries[i].vector = entry->irq;
		set_irq_msi(entry->irq, entry);
		entry->masked = readl(entry->mask_base + offset);
		msix_mask_irq(entry, 1);
		i++;
	}
}

L
Linus Torvalds 已提交
487 488 489
/**
 * msix_capability_init - configure device's MSI-X capability
 * @dev: pointer to the pci_dev data structure of MSI-X device function
R
Randy Dunlap 已提交
490 491
 * @entries: pointer to an array of struct msix_entry entries
 * @nvec: number of @entries
L
Linus Torvalds 已提交
492
 *
493
 * Setup the MSI-X capability structure of device function with a
494 495
 * single MSI-X irq. A return of zero indicates the successful setup of
 * requested MSI-X entries with allocated irqs or non-zero for otherwise.
L
Linus Torvalds 已提交
496 497 498 499
 **/
static int msix_capability_init(struct pci_dev *dev,
				struct msix_entry *entries, int nvec)
{
500
	int pos, ret;
501
	u16 control;
L
Linus Torvalds 已提交
502 503
	void __iomem *base;

H
Hidetoshi Seto 已提交
504
	pos = pci_find_capability(dev, PCI_CAP_ID_MSIX);
505 506 507 508 509 510
	pci_read_config_word(dev, pos + PCI_MSIX_FLAGS, &control);

	/* Ensure MSI-X is disabled while it is set up */
	control &= ~PCI_MSIX_FLAGS_ENABLE;
	pci_write_config_word(dev, pos + PCI_MSIX_FLAGS, control);

L
Linus Torvalds 已提交
511
	/* Request & Map MSI-X table region */
512 513
	base = msix_map_region(dev, pos, multi_msix_capable(control));
	if (!base)
L
Linus Torvalds 已提交
514 515
		return -ENOMEM;

516 517 518
	ret = msix_setup_entries(dev, pos, base, entries, nvec);
	if (ret)
		return ret;
519 520

	ret = arch_setup_msi_irqs(dev, nvec, PCI_CAP_ID_MSIX);
521 522
	if (ret)
		goto error;
523

524 525 526 527 528 529 530 531
	/*
	 * Some devices require MSI-X to be enabled before we can touch the
	 * MSI-X registers.  We need to mask all the vectors to prevent
	 * interrupts coming in before they're fully set up.
	 */
	control |= PCI_MSIX_FLAGS_MASKALL | PCI_MSIX_FLAGS_ENABLE;
	pci_write_config_word(dev, pos + PCI_MSIX_FLAGS, control);

532
	msix_program_entries(dev, entries);
533 534

	/* Set MSI-X enabled bits and unmask the function */
535
	pci_intx_for_msi(dev, 0);
536
	dev->msix_enabled = 1;
L
Linus Torvalds 已提交
537

538 539
	control &= ~PCI_MSIX_FLAGS_MASKALL;
	pci_write_config_word(dev, pos + PCI_MSIX_FLAGS, control);
540

L
Linus Torvalds 已提交
541
	return 0;
542 543 544 545 546 547 548

error:
	if (ret < 0) {
		/*
		 * If we had some success, report the number of irqs
		 * we succeeded in setting up.
		 */
549
		struct msi_desc *entry;
550 551 552 553 554 555 556 557 558 559 560 561 562
		int avail = 0;

		list_for_each_entry(entry, &dev->msi_list, list) {
			if (entry->irq != 0)
				avail++;
		}
		if (avail != 0)
			ret = avail;
	}

	free_msi_irqs(dev);

	return ret;
L
Linus Torvalds 已提交
563 564
}

565
/**
566
 * pci_msi_check_device - check whether MSI may be enabled on a device
567
 * @dev: pointer to the pci_dev data structure of MSI device function
568
 * @nvec: how many MSIs have been requested ?
569
 * @type: are we checking for MSI or MSI-X ?
570
 *
571
 * Look at global flags, the device itself, and its parent busses
572 573
 * to determine if MSI/-X are supported for the device. If MSI/-X is
 * supported return 0, else return an error code.
574
 **/
H
Hidetoshi Seto 已提交
575
static int pci_msi_check_device(struct pci_dev *dev, int nvec, int type)
576 577
{
	struct pci_bus *bus;
578
	int ret;
579

580
	/* MSI must be globally enabled and supported by the device */
581 582 583
	if (!pci_msi_enable || !dev || dev->no_msi)
		return -EINVAL;

584 585 586 587 588 589 590 591
	/*
	 * You can't ask to have 0 or less MSIs configured.
	 *  a) it's stupid ..
	 *  b) the list manipulation code assumes nvec >= 1.
	 */
	if (nvec < 1)
		return -ERANGE;

H
Hidetoshi Seto 已提交
592 593 594
	/*
	 * Any bridge which does NOT route MSI transactions from its
	 * secondary bus to its primary bus must set NO_MSI flag on
595 596 597 598
	 * the secondary pci_bus.
	 * We expect only arch-specific PCI host bus controller driver
	 * or quirks for specific PCI bridges to be setting NO_MSI.
	 */
599 600 601 602
	for (bus = dev->bus; bus; bus = bus->parent)
		if (bus->bus_flags & PCI_BUS_FLAGS_NO_MSI)
			return -EINVAL;

603 604 605 606
	ret = arch_msi_check_device(dev, nvec, type);
	if (ret)
		return ret;

607 608 609
	if (!pci_find_capability(dev, type))
		return -EINVAL;

610 611 612
	return 0;
}

L
Linus Torvalds 已提交
613
/**
614 615 616
 * pci_enable_msi_block - configure device's MSI capability structure
 * @dev: device to configure
 * @nvec: number of interrupts to configure
L
Linus Torvalds 已提交
617
 *
618 619 620 621 622 623 624 625 626
 * Allocate IRQs for a device with the MSI capability.
 * This function returns a negative errno if an error occurs.  If it
 * is unable to allocate the number of interrupts requested, it returns
 * the number of interrupts it might be able to allocate.  If it successfully
 * allocates at least the number of interrupts requested, it returns 0 and
 * updates the @dev's irq member to the lowest new interrupt number; the
 * other interrupt numbers allocated to this device are consecutive.
 */
int pci_enable_msi_block(struct pci_dev *dev, unsigned int nvec)
L
Linus Torvalds 已提交
627
{
628 629 630 631 632 633 634 635 636 637
	int status, pos, maxvec;
	u16 msgctl;

	pos = pci_find_capability(dev, PCI_CAP_ID_MSI);
	if (!pos)
		return -EINVAL;
	pci_read_config_word(dev, pos + PCI_MSI_FLAGS, &msgctl);
	maxvec = 1 << ((msgctl & PCI_MSI_FLAGS_QMASK) >> 1);
	if (nvec > maxvec)
		return maxvec;
L
Linus Torvalds 已提交
638

639
	status = pci_msi_check_device(dev, nvec, PCI_CAP_ID_MSI);
640 641
	if (status)
		return status;
L
Linus Torvalds 已提交
642

E
Eric W. Biederman 已提交
643
	WARN_ON(!!dev->msi_enabled);
L
Linus Torvalds 已提交
644

645
	/* Check whether driver already requested MSI-X irqs */
646
	if (dev->msix_enabled) {
647 648
		dev_info(&dev->dev, "can't enable MSI "
			 "(MSI-X already enabled)\n");
649
		return -EINVAL;
L
Linus Torvalds 已提交
650
	}
651 652

	status = msi_capability_init(dev, nvec);
L
Linus Torvalds 已提交
653 654
	return status;
}
655
EXPORT_SYMBOL(pci_enable_msi_block);
L
Linus Torvalds 已提交
656

657
void pci_msi_shutdown(struct pci_dev *dev)
L
Linus Torvalds 已提交
658
{
659 660 661
	struct msi_desc *desc;
	u32 mask;
	u16 ctrl;
662
	unsigned pos;
L
Linus Torvalds 已提交
663

664
	if (!pci_msi_enable || !dev || !dev->msi_enabled)
E
Eric W. Biederman 已提交
665 666
		return;

667 668 669 670 671
	BUG_ON(list_empty(&dev->msi_list));
	desc = list_first_entry(&dev->msi_list, struct msi_desc, list);
	pos = desc->msi_attrib.pos;

	msi_set_enable(dev, pos, 0);
672
	pci_intx_for_msi(dev, 1);
673
	dev->msi_enabled = 0;
674

675
	/* Return the device with MSI unmasked as initial states */
676
	pci_read_config_word(dev, pos + PCI_MSI_FLAGS, &ctrl);
677
	mask = msi_capable_mask(ctrl);
678 679
	/* Keep cached state to be restored */
	__msi_mask_irq(desc, mask, ~mask);
680 681

	/* Restore dev->irq to its default pin-assertion irq */
682
	dev->irq = desc->msi_attrib.default_irq;
683
}
684

H
Hidetoshi Seto 已提交
685
void pci_disable_msi(struct pci_dev *dev)
686 687 688 689 690
{
	if (!pci_msi_enable || !dev || !dev->msi_enabled)
		return;

	pci_msi_shutdown(dev);
691
	free_msi_irqs(dev);
L
Linus Torvalds 已提交
692
}
693
EXPORT_SYMBOL(pci_disable_msi);
L
Linus Torvalds 已提交
694

695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711
/**
 * pci_msix_table_size - return the number of device's MSI-X table entries
 * @dev: pointer to the pci_dev data structure of MSI-X device function
 */
int pci_msix_table_size(struct pci_dev *dev)
{
	int pos;
	u16 control;

	pos = pci_find_capability(dev, PCI_CAP_ID_MSIX);
	if (!pos)
		return 0;

	pci_read_config_word(dev, msi_control_reg(pos), &control);
	return multi_msix_capable(control);
}

L
Linus Torvalds 已提交
712 713 714
/**
 * pci_enable_msix - configure device's MSI-X capability structure
 * @dev: pointer to the pci_dev data structure of MSI-X device function
715
 * @entries: pointer to an array of MSI-X entries
716
 * @nvec: number of MSI-X irqs requested for allocation by device driver
L
Linus Torvalds 已提交
717 718
 *
 * Setup the MSI-X capability structure of device function with the number
719
 * of requested irqs upon its software driver call to request for
L
Linus Torvalds 已提交
720 721
 * MSI-X mode enabled on its hardware device function. A return of zero
 * indicates the successful configuration of MSI-X capability structure
722
 * with new allocated MSI-X irqs. A return of < 0 indicates a failure.
L
Linus Torvalds 已提交
723
 * Or a return of > 0 indicates that driver request is exceeding the number
724 725
 * of irqs or MSI-X vectors available. Driver should use the returned value to
 * re-send its request.
L
Linus Torvalds 已提交
726
 **/
H
Hidetoshi Seto 已提交
727
int pci_enable_msix(struct pci_dev *dev, struct msix_entry *entries, int nvec)
L
Linus Torvalds 已提交
728
{
729
	int status, nr_entries;
E
Eric W. Biederman 已提交
730
	int i, j;
L
Linus Torvalds 已提交
731

732
	if (!entries)
H
Hidetoshi Seto 已提交
733
		return -EINVAL;
L
Linus Torvalds 已提交
734

735 736 737 738
	status = pci_msi_check_device(dev, nvec, PCI_CAP_ID_MSIX);
	if (status)
		return status;

739
	nr_entries = pci_msix_table_size(dev);
L
Linus Torvalds 已提交
740
	if (nvec > nr_entries)
741
		return nr_entries;
L
Linus Torvalds 已提交
742 743 744 745 746 747 748 749 750 751

	/* Check for any invalid entries */
	for (i = 0; i < nvec; i++) {
		if (entries[i].entry >= nr_entries)
			return -EINVAL;		/* invalid entry */
		for (j = i + 1; j < nvec; j++) {
			if (entries[i].entry == entries[j].entry)
				return -EINVAL;	/* duplicate entry */
		}
	}
E
Eric W. Biederman 已提交
752
	WARN_ON(!!dev->msix_enabled);
753

754
	/* Check whether driver already requested for MSI irq */
H
Hidetoshi Seto 已提交
755
	if (dev->msi_enabled) {
756 757
		dev_info(&dev->dev, "can't enable MSI-X "
		       "(MSI IRQ already assigned)\n");
L
Linus Torvalds 已提交
758 759 760 761 762
		return -EINVAL;
	}
	status = msix_capability_init(dev, entries, nvec);
	return status;
}
763
EXPORT_SYMBOL(pci_enable_msix);
L
Linus Torvalds 已提交
764

H
Hidetoshi Seto 已提交
765
void pci_msix_shutdown(struct pci_dev *dev)
766
{
767 768
	struct msi_desc *entry;

769
	if (!pci_msi_enable || !dev || !dev->msix_enabled)
E
Eric W. Biederman 已提交
770 771
		return;

772 773 774 775 776 777
	/* Return the device with MSI-X masked as initial states */
	list_for_each_entry(entry, &dev->msi_list, list) {
		/* Keep cached states to be restored */
		__msix_mask_irq(entry, 1);
	}

778
	msix_set_enable(dev, 0);
779
	pci_intx_for_msi(dev, 1);
780
	dev->msix_enabled = 0;
781
}
782

H
Hidetoshi Seto 已提交
783
void pci_disable_msix(struct pci_dev *dev)
784 785 786 787 788
{
	if (!pci_msi_enable || !dev || !dev->msix_enabled)
		return;

	pci_msix_shutdown(dev);
789
	free_msi_irqs(dev);
L
Linus Torvalds 已提交
790
}
791
EXPORT_SYMBOL(pci_disable_msix);
L
Linus Torvalds 已提交
792 793

/**
794
 * msi_remove_pci_irq_vectors - reclaim MSI(X) irqs to unused state
L
Linus Torvalds 已提交
795 796
 * @dev: pointer to the pci_dev data structure of MSI(X) device function
 *
797
 * Being called during hotplug remove, from which the device function
798
 * is hot-removed. All previous assigned MSI/MSI-X irqs, if
L
Linus Torvalds 已提交
799 800 801
 * allocated for this device function, are reclaimed to unused state,
 * which may be used later on.
 **/
H
Hidetoshi Seto 已提交
802
void msi_remove_pci_irq_vectors(struct pci_dev *dev)
L
Linus Torvalds 已提交
803 804
{
	if (!pci_msi_enable || !dev)
H
Hidetoshi Seto 已提交
805
		return;
L
Linus Torvalds 已提交
806

807 808
	if (dev->msi_enabled || dev->msix_enabled)
		free_msi_irqs(dev);
L
Linus Torvalds 已提交
809 810
}

811 812 813 814
void pci_no_msi(void)
{
	pci_msi_enable = 0;
}
815

816 817 818 819 820 821 822
/**
 * pci_msi_enabled - is MSI enabled?
 *
 * Returns true if MSI has not been disabled by the command-line option
 * pci=nomsi.
 **/
int pci_msi_enabled(void)
823
{
824
	return pci_msi_enable;
825
}
826
EXPORT_SYMBOL(pci_msi_enabled);
827

828
void pci_msi_init_pci_dev(struct pci_dev *dev)
829
{
830
	INIT_LIST_HEAD(&dev->msi_list);
831
}