msi.c 20.6 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>
L
Linus Torvalds 已提交
19 20 21 22 23 24 25 26 27

#include <asm/errno.h>
#include <asm/io.h>

#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 void 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

134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159
	if (!desc->msi_attrib.maskbit)
		return;

	mask_bits &= ~mask;
	mask_bits |= flag;
	pci_write_config_dword(desc->dev, desc->mask_pos, mask_bits);
	desc->masked = mask_bits;
}

/*
 * 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.
 */
static void msix_mask_irq(struct msi_desc *desc, u32 flag)
{
	u32 mask_bits = desc->masked;
	unsigned offset = desc->msi_attrib.entry_nr * PCI_MSIX_ENTRY_SIZE +
					PCI_MSIX_ENTRY_VECTOR_CTRL_OFFSET;
	mask_bits &= ~1;
	mask_bits |= flag;
	writel(mask_bits, desc->mask_base + offset);
	desc->masked = mask_bits;
}
160

161 162 163
static void msi_set_mask_bit(unsigned irq, u32 flag)
{
	struct msi_desc *desc = get_irq_msi(irq);
164

165 166 167 168
	if (desc->msi_attrib.is_msix) {
		msix_mask_irq(desc, flag);
		readl(desc->mask_base);		/* Flush write to device */
	} else {
169 170
		unsigned offset = irq - desc->dev->irq;
		msi_mask_irq(desc, 1 << offset, flag << offset);
L
Linus Torvalds 已提交
171
	}
172 173 174 175 176 177 178 179 180 181
}

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
182 183
}

Y
Yinghai Lu 已提交
184
void read_msi_msg_desc(struct irq_desc *desc, struct msi_msg *msg)
L
Linus Torvalds 已提交
185
{
Y
Yinghai Lu 已提交
186
	struct msi_desc *entry = get_irq_desc_msi(desc);
187 188 189 190 191 192 193 194
	if (entry->msi_attrib.is_msix) {
		void __iomem *base = entry->mask_base +
			entry->msi_attrib.entry_nr * PCI_MSIX_ENTRY_SIZE;

		msg->address_lo = readl(base + PCI_MSIX_ENTRY_LOWER_ADDR_OFFSET);
		msg->address_hi = readl(base + PCI_MSIX_ENTRY_UPPER_ADDR_OFFSET);
		msg->data = readl(base + PCI_MSIX_ENTRY_DATA_OFFSET);
	} else {
195 196 197 198 199 200 201 202 203 204 205 206
		struct pci_dev *dev = entry->dev;
		int pos = entry->msi_attrib.pos;
		u16 data;

		pci_read_config_dword(dev, msi_lower_address_reg(pos),
					&msg->address_lo);
		if (entry->msi_attrib.is_64) {
			pci_read_config_dword(dev, msi_upper_address_reg(pos),
						&msg->address_hi);
			pci_read_config_word(dev, msi_data_reg(pos, 1), &data);
		} else {
			msg->address_hi = 0;
207
			pci_read_config_word(dev, msi_data_reg(pos, 0), &data);
208 209 210 211
		}
		msg->data = data;
	}
}
L
Linus Torvalds 已提交
212

Y
Yinghai Lu 已提交
213
void read_msi_msg(unsigned int irq, struct msi_msg *msg)
214
{
Y
Yinghai Lu 已提交
215 216 217 218 219 220 221 222
	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);
223 224 225 226 227 228 229 230 231 232 233
	if (entry->msi_attrib.is_msix) {
		void __iomem *base;
		base = entry->mask_base +
			entry->msi_attrib.entry_nr * PCI_MSIX_ENTRY_SIZE;

		writel(msg->address_lo,
			base + PCI_MSIX_ENTRY_LOWER_ADDR_OFFSET);
		writel(msg->address_hi,
			base + PCI_MSIX_ENTRY_UPPER_ADDR_OFFSET);
		writel(msg->data, base + PCI_MSIX_ENTRY_DATA_OFFSET);
	} else {
234 235
		struct pci_dev *dev = entry->dev;
		int pos = entry->msi_attrib.pos;
236 237 238 239 240 241
		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);
242 243 244 245 246 247 248 249 250 251 252 253

		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
254
	}
255
	entry->msg = *msg;
L
Linus Torvalds 已提交
256
}
257

Y
Yinghai Lu 已提交
258 259 260 261 262 263 264
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);
}

265
static int msi_free_irqs(struct pci_dev* dev);
S
Satoru Takeuchi 已提交
266

267
static struct msi_desc *alloc_msi_entry(struct pci_dev *dev)
L
Linus Torvalds 已提交
268
{
269 270
	struct msi_desc *desc = kzalloc(sizeof(*desc), GFP_KERNEL);
	if (!desc)
L
Linus Torvalds 已提交
271 272
		return NULL;

273 274
	INIT_LIST_HEAD(&desc->list);
	desc->dev = dev;
L
Linus Torvalds 已提交
275

276
	return desc;
L
Linus Torvalds 已提交
277 278
}

279 280 281 282 283 284
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);
}

285
static void __pci_restore_msi_state(struct pci_dev *dev)
286
{
287
	int pos;
288
	u16 control;
289
	struct msi_desc *entry;
290

291 292 293
	if (!dev->msi_enabled)
		return;

294 295
	entry = get_irq_msi(dev->irq);
	pos = entry->msi_attrib.pos;
296

297
	pci_intx_for_msi(dev, 0);
298
	msi_set_enable(dev, pos, 0);
299 300 301
	write_msi_msg(dev->irq, &entry->msg);

	pci_read_config_word(dev, pos + PCI_MSI_FLAGS, &control);
302
	msi_mask_irq(entry, msi_capable_mask(control), entry->masked);
303
	control &= ~PCI_MSI_FLAGS_QSIZE;
304
	control |= (entry->msi_attrib.multiple << 4) | PCI_MSI_FLAGS_ENABLE;
305
	pci_write_config_word(dev, pos + PCI_MSI_FLAGS, control);
306 307 308
}

static void __pci_restore_msix_state(struct pci_dev *dev)
309 310 311
{
	int pos;
	struct msi_desc *entry;
312
	u16 control;
313

E
Eric W. Biederman 已提交
314 315
	if (!dev->msix_enabled)
		return;
316 317 318 319
	BUG_ON(list_empty(&dev->msi_list));
	entry = list_entry(dev->msi_list.next, struct msi_desc, list);
	pos = entry->msi_attrib.pos;
	pci_read_config_word(dev, pos + PCI_MSIX_FLAGS, &control);
E
Eric W. Biederman 已提交
320

321
	/* route the table */
322
	pci_intx_for_msi(dev, 0);
323 324
	control |= PCI_MSIX_FLAGS_ENABLE | PCI_MSIX_FLAGS_MASKALL;
	pci_write_config_word(dev, pos + PCI_MSIX_FLAGS, control);
325

326 327
	list_for_each_entry(entry, &dev->msi_list, list) {
		write_msi_msg(entry->irq, &entry->msg);
328
		msix_mask_irq(entry, entry->masked);
329 330
	}

331 332
	control &= ~PCI_MSIX_FLAGS_MASKALL;
	pci_write_config_word(dev, pos + PCI_MSIX_FLAGS, control);
333
}
334 335 336 337 338 339

void pci_restore_msi_state(struct pci_dev *dev)
{
	__pci_restore_msi_state(dev);
	__pci_restore_msix_state(dev);
}
340
EXPORT_SYMBOL_GPL(pci_restore_msi_state);
341

L
Linus Torvalds 已提交
342 343 344
/**
 * msi_capability_init - configure device's MSI capability structure
 * @dev: pointer to the pci_dev data structure of MSI device function
345
 * @nvec: number of interrupts to allocate
L
Linus Torvalds 已提交
346
 *
347 348 349 350 351 352 353
 * 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
354 355
{
	struct msi_desc *entry;
356
	int pos, ret;
L
Linus Torvalds 已提交
357
	u16 control;
358
	unsigned mask;
L
Linus Torvalds 已提交
359 360

   	pos = pci_find_capability(dev, PCI_CAP_ID_MSI);
361 362
	msi_set_enable(dev, pos, 0);	/* Disable MSI during set up */

L
Linus Torvalds 已提交
363 364
	pci_read_config_word(dev, msi_control_reg(pos), &control);
	/* MSI Entry Initialization */
365
	entry = alloc_msi_entry(dev);
366 367
	if (!entry)
		return -ENOMEM;
368

369
	entry->msi_attrib.is_msix = 0;
370
	entry->msi_attrib.is_64 = is_64bit_address(control);
L
Linus Torvalds 已提交
371 372
	entry->msi_attrib.entry_nr = 0;
	entry->msi_attrib.maskbit = is_mask_bit_support(control);
373
	entry->msi_attrib.default_irq = dev->irq;	/* Save IOAPIC IRQ */
374
	entry->msi_attrib.pos = pos;
375

376
	entry->mask_pos = msi_mask_reg(pos, entry->msi_attrib.is_64);
377 378 379 380 381 382
	/* 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);

383
	list_add_tail(&entry->list, &dev->msi_list);
384

L
Linus Torvalds 已提交
385
	/* Configure MSI capability structure */
386
	ret = arch_setup_msi_irqs(dev, nvec, PCI_CAP_ID_MSI);
387
	if (ret) {
388
		msi_free_irqs(dev);
389
		return ret;
390
	}
391

L
Linus Torvalds 已提交
392
	/* Set MSI enabled bits	 */
393
	pci_intx_for_msi(dev, 0);
394
	msi_set_enable(dev, pos, 1);
395
	dev->msi_enabled = 1;
L
Linus Torvalds 已提交
396

397
	dev->irq = entry->irq;
L
Linus Torvalds 已提交
398 399 400 401 402 403
	return 0;
}

/**
 * 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
404 405
 * @entries: pointer to an array of struct msix_entry entries
 * @nvec: number of @entries
L
Linus Torvalds 已提交
406
 *
407
 * Setup the MSI-X capability structure of device function with a
408 409
 * 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
410 411 412 413
 **/
static int msix_capability_init(struct pci_dev *dev,
				struct msix_entry *entries, int nvec)
{
414
	struct msi_desc *entry;
415
	int pos, i, j, nr_entries, ret;
416 417
	unsigned long phys_addr;
	u32 table_offset;
L
Linus Torvalds 已提交
418 419 420 421 422
 	u16 control;
	u8 bir;
	void __iomem *base;

   	pos = pci_find_capability(dev, PCI_CAP_ID_MSIX);
423 424 425 426 427 428
	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
429 430
	/* Request & Map MSI-X table region */
	nr_entries = multi_msix_capable(control);
431 432

 	pci_read_config_dword(dev, msix_table_offset_reg(pos), &table_offset);
L
Linus Torvalds 已提交
433
	bir = (u8)(table_offset & PCI_MSIX_FLAGS_BIRMASK);
434 435
	table_offset &= ~PCI_MSIX_FLAGS_BIRMASK;
	phys_addr = pci_resource_start (dev, bir) + table_offset;
L
Linus Torvalds 已提交
436 437 438 439 440
	base = ioremap_nocache(phys_addr, nr_entries * PCI_MSIX_ENTRY_SIZE);
	if (base == NULL)
		return -ENOMEM;

	for (i = 0; i < nvec; i++) {
441
		entry = alloc_msi_entry(dev);
442 443 444 445 446 447 448 449
		if (!entry) {
			if (!i)
				iounmap(base);
			else
				msi_free_irqs(dev);
			/* No enough memory. Don't try again */
			return -ENOMEM;
		}
L
Linus Torvalds 已提交
450 451

 		j = entries[i].entry;
452
		entry->msi_attrib.is_msix = 1;
453
		entry->msi_attrib.is_64 = 1;
L
Linus Torvalds 已提交
454
		entry->msi_attrib.entry_nr = j;
455
		entry->msi_attrib.default_irq = dev->irq;
456
		entry->msi_attrib.pos = pos;
L
Linus Torvalds 已提交
457
		entry->mask_base = base;
458

459
		list_add_tail(&entry->list, &dev->msi_list);
L
Linus Torvalds 已提交
460
	}
461 462

	ret = arch_setup_msi_irqs(dev, nvec, PCI_CAP_ID_MSIX);
463 464 465
	if (ret < 0) {
		/* If we had some success report the number of irqs
		 * we succeeded in setting up. */
466 467 468 469 470
		int avail = 0;
		list_for_each_entry(entry, &dev->msi_list, list) {
			if (entry->irq != 0) {
				avail++;
			}
L
Linus Torvalds 已提交
471
		}
472

473 474 475
		if (avail != 0)
			ret = avail;
	}
476

477 478 479
	if (ret) {
		msi_free_irqs(dev);
		return ret;
L
Linus Torvalds 已提交
480
	}
481

482 483 484 485 486 487 488 489
	/*
	 * 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);

490 491 492 493
	i = 0;
	list_for_each_entry(entry, &dev->msi_list, list) {
		entries[i].vector = entry->irq;
		set_irq_msi(entry->irq, entry);
494 495 496 497
		j = entries[i].entry;
		entry->masked = readl(base + j * PCI_MSIX_ENTRY_SIZE +
					PCI_MSIX_ENTRY_VECTOR_CTRL_OFFSET);
		msix_mask_irq(entry, 1);
498 499
		i++;
	}
500 501

	/* Set MSI-X enabled bits and unmask the function */
502
	pci_intx_for_msi(dev, 0);
503
	dev->msix_enabled = 1;
L
Linus Torvalds 已提交
504

505 506
	control &= ~PCI_MSIX_FLAGS_MASKALL;
	pci_write_config_word(dev, pos + PCI_MSIX_FLAGS, control);
507

L
Linus Torvalds 已提交
508 509 510
	return 0;
}

511
/**
512
 * pci_msi_check_device - check whether MSI may be enabled on a device
513
 * @dev: pointer to the pci_dev data structure of MSI device function
514
 * @nvec: how many MSIs have been requested ?
515
 * @type: are we checking for MSI or MSI-X ?
516
 *
517
 * Look at global flags, the device itself, and its parent busses
518 519
 * to determine if MSI/-X are supported for the device. If MSI/-X is
 * supported return 0, else return an error code.
520
 **/
521
static int pci_msi_check_device(struct pci_dev* dev, int nvec, int type)
522 523
{
	struct pci_bus *bus;
524
	int ret;
525

526
	/* MSI must be globally enabled and supported by the device */
527 528 529
	if (!pci_msi_enable || !dev || dev->no_msi)
		return -EINVAL;

530 531 532 533 534 535 536 537
	/*
	 * 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;

538 539 540 541 542 543
	/* Any bridge which does NOT route MSI transactions from it's
	 * secondary bus to it's primary bus must set NO_MSI flag on
	 * 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.
	 */
544 545 546 547
	for (bus = dev->bus; bus; bus = bus->parent)
		if (bus->bus_flags & PCI_BUS_FLAGS_NO_MSI)
			return -EINVAL;

548 549 550 551
	ret = arch_msi_check_device(dev, nvec, type);
	if (ret)
		return ret;

552 553 554
	if (!pci_find_capability(dev, type))
		return -EINVAL;

555 556 557
	return 0;
}

L
Linus Torvalds 已提交
558
/**
559 560 561
 * 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
562
 *
563 564 565 566 567 568 569 570 571
 * 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
572
{
573 574 575 576 577 578 579 580 581 582
	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
583

584
	status = pci_msi_check_device(dev, nvec, PCI_CAP_ID_MSI);
585 586
	if (status)
		return status;
L
Linus Torvalds 已提交
587

E
Eric W. Biederman 已提交
588
	WARN_ON(!!dev->msi_enabled);
L
Linus Torvalds 已提交
589

590
	/* Check whether driver already requested MSI-X irqs */
591
	if (dev->msix_enabled) {
592 593
		dev_info(&dev->dev, "can't enable MSI "
			 "(MSI-X already enabled)\n");
594
		return -EINVAL;
L
Linus Torvalds 已提交
595
	}
596 597

	status = msi_capability_init(dev, nvec);
L
Linus Torvalds 已提交
598 599
	return status;
}
600
EXPORT_SYMBOL(pci_enable_msi_block);
L
Linus Torvalds 已提交
601

602
void pci_msi_shutdown(struct pci_dev *dev)
L
Linus Torvalds 已提交
603
{
604 605 606
	struct msi_desc *desc;
	u32 mask;
	u16 ctrl;
607
	unsigned pos;
L
Linus Torvalds 已提交
608

609
	if (!pci_msi_enable || !dev || !dev->msi_enabled)
E
Eric W. Biederman 已提交
610 611
		return;

612 613 614 615 616
	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);
617
	pci_intx_for_msi(dev, 1);
618
	dev->msi_enabled = 0;
619

620
	pci_read_config_word(dev, pos + PCI_MSI_FLAGS, &ctrl);
621 622
	mask = msi_capable_mask(ctrl);
	msi_mask_irq(desc, mask, ~mask);
623 624

	/* Restore dev->irq to its default pin-assertion irq */
625
	dev->irq = desc->msi_attrib.default_irq;
626
}
627

628 629 630 631 632 633 634 635 636 637
void pci_disable_msi(struct pci_dev* dev)
{
	struct msi_desc *entry;

	if (!pci_msi_enable || !dev || !dev->msi_enabled)
		return;

	pci_msi_shutdown(dev);

	entry = list_entry(dev->msi_list.next, struct msi_desc, list);
638
	if (entry->msi_attrib.is_msix)
639 640 641
		return;

	msi_free_irqs(dev);
L
Linus Torvalds 已提交
642
}
643
EXPORT_SYMBOL(pci_disable_msi);
L
Linus Torvalds 已提交
644

645
static int msi_free_irqs(struct pci_dev* dev)
L
Linus Torvalds 已提交
646
{
647
	struct msi_desc *entry, *tmp;
M
Michael Ellerman 已提交
648

649
	list_for_each_entry(entry, &dev->msi_list, list) {
650 651 652 653 654 655
		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));
656
	}
L
Linus Torvalds 已提交
657

658
	arch_teardown_msi_irqs(dev);
L
Linus Torvalds 已提交
659

660
	list_for_each_entry_safe(entry, tmp, &dev->msi_list, list) {
661
		if (entry->msi_attrib.is_msix) {
662
			msix_mask_irq(entry, 1);
663 664
			if (list_is_last(&entry->list, &dev->msi_list))
				iounmap(entry->mask_base);
665 666 667
		}
		list_del(&entry->list);
		kfree(entry);
L
Linus Torvalds 已提交
668 669 670 671 672
	}

	return 0;
}

673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689
/**
 * 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
690 691 692
/**
 * pci_enable_msix - configure device's MSI-X capability structure
 * @dev: pointer to the pci_dev data structure of MSI-X device function
693
 * @entries: pointer to an array of MSI-X entries
694
 * @nvec: number of MSI-X irqs requested for allocation by device driver
L
Linus Torvalds 已提交
695 696
 *
 * Setup the MSI-X capability structure of device function with the number
697
 * of requested irqs upon its software driver call to request for
L
Linus Torvalds 已提交
698 699
 * MSI-X mode enabled on its hardware device function. A return of zero
 * indicates the successful configuration of MSI-X capability structure
700
 * with new allocated MSI-X irqs. A return of < 0 indicates a failure.
L
Linus Torvalds 已提交
701
 * Or a return of > 0 indicates that driver request is exceeding the number
702 703
 * of irqs or MSI-X vectors available. Driver should use the returned value to
 * re-send its request.
L
Linus Torvalds 已提交
704 705 706
 **/
int pci_enable_msix(struct pci_dev* dev, struct msix_entry *entries, int nvec)
{
707
	int status, nr_entries;
E
Eric W. Biederman 已提交
708
	int i, j;
L
Linus Torvalds 已提交
709

710
	if (!entries)
L
Linus Torvalds 已提交
711 712
 		return -EINVAL;

713 714 715 716
	status = pci_msi_check_device(dev, nvec, PCI_CAP_ID_MSIX);
	if (status)
		return status;

717
	nr_entries = pci_msix_table_size(dev);
L
Linus Torvalds 已提交
718
	if (nvec > nr_entries)
719
		return nr_entries;
L
Linus Torvalds 已提交
720 721 722 723 724 725 726 727 728 729

	/* 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
730
	WARN_ON(!!dev->msix_enabled);
731

732
	/* Check whether driver already requested for MSI irq */
733
   	if (dev->msi_enabled) {
734 735
		dev_info(&dev->dev, "can't enable MSI-X "
		       "(MSI IRQ already assigned)\n");
L
Linus Torvalds 已提交
736 737 738 739 740
		return -EINVAL;
	}
	status = msix_capability_init(dev, entries, nvec);
	return status;
}
741
EXPORT_SYMBOL(pci_enable_msix);
L
Linus Torvalds 已提交
742

743
static void msix_free_all_irqs(struct pci_dev *dev)
L
Linus Torvalds 已提交
744
{
745
	msi_free_irqs(dev);
746 747
}

748
void pci_msix_shutdown(struct pci_dev* dev)
749
{
750
	if (!pci_msi_enable || !dev || !dev->msix_enabled)
E
Eric W. Biederman 已提交
751 752
		return;

753
	msix_set_enable(dev, 0);
754
	pci_intx_for_msi(dev, 1);
755
	dev->msix_enabled = 0;
756 757 758 759 760 761 762
}
void pci_disable_msix(struct pci_dev* dev)
{
	if (!pci_msi_enable || !dev || !dev->msix_enabled)
		return;

	pci_msix_shutdown(dev);
763

764
	msix_free_all_irqs(dev);
L
Linus Torvalds 已提交
765
}
766
EXPORT_SYMBOL(pci_disable_msix);
L
Linus Torvalds 已提交
767 768

/**
769
 * msi_remove_pci_irq_vectors - reclaim MSI(X) irqs to unused state
L
Linus Torvalds 已提交
770 771
 * @dev: pointer to the pci_dev data structure of MSI(X) device function
 *
772
 * Being called during hotplug remove, from which the device function
773
 * is hot-removed. All previous assigned MSI/MSI-X irqs, if
L
Linus Torvalds 已提交
774 775 776 777 778 779 780 781
 * allocated for this device function, are reclaimed to unused state,
 * which may be used later on.
 **/
void msi_remove_pci_irq_vectors(struct pci_dev* dev)
{
	if (!pci_msi_enable || !dev)
 		return;

782 783
	if (dev->msi_enabled)
		msi_free_irqs(dev);
L
Linus Torvalds 已提交
784

785 786
	if (dev->msix_enabled)
		msix_free_all_irqs(dev);
L
Linus Torvalds 已提交
787 788
}

789 790 791 792
void pci_no_msi(void)
{
	pci_msi_enable = 0;
}
793

794 795 796 797 798 799 800
/**
 * 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)
801
{
802
	return pci_msi_enable;
803
}
804
EXPORT_SYMBOL(pci_msi_enabled);
805

806
void pci_msi_init_pci_dev(struct pci_dev *dev)
807
{
808
	INIT_LIST_HEAD(&dev->msi_list);
809
}