irq.c 14.3 KB
Newer Older
1 2 3 4 5 6
/*
 * Common interrupt code for 32 and 64 bit
 */
#include <linux/cpu.h>
#include <linux/interrupt.h>
#include <linux/kernel_stat.h>
7
#include <linux/of.h>
8
#include <linux/seq_file.h>
9
#include <linux/smp.h>
J
Jeremy Fitzhardinge 已提交
10
#include <linux/ftrace.h>
11
#include <linux/delay.h>
12
#include <linux/export.h>
13

I
Ingo Molnar 已提交
14
#include <asm/apic.h>
15
#include <asm/io_apic.h>
16
#include <asm/irq.h>
J
Jeremy Fitzhardinge 已提交
17
#include <asm/idle.h>
18
#include <asm/mce.h>
19
#include <asm/hw_irq.h>
20
#include <asm/desc.h>
21 22

#define CREATE_TRACE_POINTS
23
#include <asm/trace/irq_vectors.h>
24

B
Brian Gerst 已提交
25 26 27 28 29 30
DEFINE_PER_CPU_SHARED_ALIGNED(irq_cpustat_t, irq_stat);
EXPORT_PER_CPU_SYMBOL(irq_stat);

DEFINE_PER_CPU(struct pt_regs *, irq_regs);
EXPORT_PER_CPU_SYMBOL(irq_regs);

31 32
atomic_t irq_err_count;

33
/* Function pointer for generic interrupt vector handling */
34
void (*x86_platform_ipi_callback)(void) = NULL;
35

T
Thomas Gleixner 已提交
36 37 38 39 40 41
/*
 * 'what should we do if we get a hw irq event on an illegal vector'.
 * each architecture has to answer this themselves.
 */
void ack_bad_irq(unsigned int irq)
{
C
Cyrill Gorcunov 已提交
42 43
	if (printk_ratelimit())
		pr_err("unexpected IRQ trap at vector %02x\n", irq);
T
Thomas Gleixner 已提交
44 45 46 47 48 49 50 51 52 53

	/*
	 * Currently unexpected vectors happen only on SMP and APIC.
	 * We _must_ ack these because every local APIC has only N
	 * irq slots per priority level, and a 'hanging, unacked' IRQ
	 * holds up an irq slot - in excessive cases (when multiple
	 * unexpected vectors occur) that might lock up the APIC
	 * completely.
	 * But only ack when the APIC is enabled -AK
	 */
54
	ack_APIC_irq();
T
Thomas Gleixner 已提交
55 56
}

57
#define irq_stats(x)		(&per_cpu(irq_stat, x))
58
/*
59
 * /proc/interrupts printing for arch specific interrupts
60
 */
61
int arch_show_interrupts(struct seq_file *p, int prec)
62 63 64
{
	int j;

65
	seq_printf(p, "%*s: ", prec, "NMI");
66 67
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", irq_stats(j)->__nmi_count);
68
	seq_puts(p, "  Non-maskable interrupts\n");
69
#ifdef CONFIG_X86_LOCAL_APIC
70
	seq_printf(p, "%*s: ", prec, "LOC");
71 72
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", irq_stats(j)->apic_timer_irqs);
73
	seq_puts(p, "  Local timer interrupts\n");
74 75 76 77

	seq_printf(p, "%*s: ", prec, "SPU");
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", irq_stats(j)->irq_spurious_count);
78
	seq_puts(p, "  Spurious interrupts\n");
79
	seq_printf(p, "%*s: ", prec, "PMI");
I
Ingo Molnar 已提交
80 81
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", irq_stats(j)->apic_perf_irqs);
82
	seq_puts(p, "  Performance monitoring interrupts\n");
83
	seq_printf(p, "%*s: ", prec, "IWI");
84
	for_each_online_cpu(j)
85
		seq_printf(p, "%10u ", irq_stats(j)->apic_irq_work_irqs);
86
	seq_puts(p, "  IRQ work interrupts\n");
87 88
	seq_printf(p, "%*s: ", prec, "RTR");
	for_each_online_cpu(j)
89
		seq_printf(p, "%10u ", irq_stats(j)->icr_read_retry_count);
90
	seq_puts(p, "  APIC ICR read retries\n");
91
#endif
92
	if (x86_platform_ipi_callback) {
93
		seq_printf(p, "%*s: ", prec, "PLT");
94
		for_each_online_cpu(j)
95
			seq_printf(p, "%10u ", irq_stats(j)->x86_platform_ipis);
96
		seq_puts(p, "  Platform interrupts\n");
97
	}
98
#ifdef CONFIG_SMP
99
	seq_printf(p, "%*s: ", prec, "RES");
100 101
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", irq_stats(j)->irq_resched_count);
102
	seq_puts(p, "  Rescheduling interrupts\n");
103
	seq_printf(p, "%*s: ", prec, "CAL");
104
	for_each_online_cpu(j)
105 106
		seq_printf(p, "%10u ", irq_stats(j)->irq_call_count -
					irq_stats(j)->irq_tlb_count);
107
	seq_puts(p, "  Function call interrupts\n");
108
	seq_printf(p, "%*s: ", prec, "TLB");
109 110
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", irq_stats(j)->irq_tlb_count);
111
	seq_puts(p, "  TLB shootdowns\n");
112
#endif
113
#ifdef CONFIG_X86_THERMAL_VECTOR
114
	seq_printf(p, "%*s: ", prec, "TRM");
115 116
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", irq_stats(j)->irq_thermal_count);
117
	seq_puts(p, "  Thermal event interrupts\n");
118 119
#endif
#ifdef CONFIG_X86_MCE_THRESHOLD
120
	seq_printf(p, "%*s: ", prec, "THR");
121 122
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", irq_stats(j)->irq_threshold_count);
123
	seq_puts(p, "  Threshold APIC interrupts\n");
124
#endif
125 126 127 128 129 130
#ifdef CONFIG_X86_MCE_AMD
	seq_printf(p, "%*s: ", prec, "DFR");
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", irq_stats(j)->irq_deferred_error_count);
	seq_puts(p, "  Deferred Error APIC interrupts\n");
#endif
131
#ifdef CONFIG_X86_MCE
132 133 134
	seq_printf(p, "%*s: ", prec, "MCE");
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", per_cpu(mce_exception_count, j));
135
	seq_puts(p, "  Machine check exceptions\n");
136 137 138
	seq_printf(p, "%*s: ", prec, "MCP");
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", per_cpu(mce_poll_count, j));
139
	seq_puts(p, "  Machine check polls\n");
140
#endif
141
#if IS_ENABLED(CONFIG_HYPERV) || defined(CONFIG_XEN)
142
	seq_printf(p, "%*s: ", prec, "HYP");
143 144
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", irq_stats(j)->irq_hv_callback_count);
145
	seq_puts(p, "  Hypervisor callback interrupts\n");
146
#endif
147
	seq_printf(p, "%*s: %10u\n", prec, "ERR", atomic_read(&irq_err_count));
148
#if defined(CONFIG_X86_IO_APIC)
149
	seq_printf(p, "%*s: %10u\n", prec, "MIS", atomic_read(&irq_mis_count));
150 151 152 153 154 155 156 157 158 159 160 161
#endif
#ifdef CONFIG_HAVE_KVM
	seq_printf(p, "%*s: ", prec, "PIN");
	for_each_online_cpu(j)
		seq_printf(p, "%10u ", irq_stats(j)->kvm_posted_intr_ipis);
	seq_puts(p, "  Posted-interrupt notification event\n");

	seq_printf(p, "%*s: ", prec, "PIW");
	for_each_online_cpu(j)
		seq_printf(p, "%10u ",
			   irq_stats(j)->kvm_posted_intr_wakeup_ipis);
	seq_puts(p, "  Posted-interrupt wakeup event\n");
162 163 164 165 166 167 168 169 170 171 172 173 174
#endif
	return 0;
}

/*
 * /proc/stat helpers
 */
u64 arch_irq_stat_cpu(unsigned int cpu)
{
	u64 sum = irq_stats(cpu)->__nmi_count;

#ifdef CONFIG_X86_LOCAL_APIC
	sum += irq_stats(cpu)->apic_timer_irqs;
175
	sum += irq_stats(cpu)->irq_spurious_count;
I
Ingo Molnar 已提交
176
	sum += irq_stats(cpu)->apic_perf_irqs;
177
	sum += irq_stats(cpu)->apic_irq_work_irqs;
178
	sum += irq_stats(cpu)->icr_read_retry_count;
179
#endif
180 181
	if (x86_platform_ipi_callback)
		sum += irq_stats(cpu)->x86_platform_ipis;
182 183 184 185
#ifdef CONFIG_SMP
	sum += irq_stats(cpu)->irq_resched_count;
	sum += irq_stats(cpu)->irq_call_count;
#endif
186
#ifdef CONFIG_X86_THERMAL_VECTOR
187
	sum += irq_stats(cpu)->irq_thermal_count;
188 189
#endif
#ifdef CONFIG_X86_MCE_THRESHOLD
190
	sum += irq_stats(cpu)->irq_threshold_count;
H
Hidetoshi Seto 已提交
191
#endif
192
#ifdef CONFIG_X86_MCE
H
Hidetoshi Seto 已提交
193 194
	sum += per_cpu(mce_exception_count, cpu);
	sum += per_cpu(mce_poll_count, cpu);
195 196 197 198 199 200 201 202 203
#endif
	return sum;
}

u64 arch_irq_stat(void)
{
	u64 sum = atomic_read(&irq_err_count);
	return sum;
}
204

J
Jeremy Fitzhardinge 已提交
205 206 207 208 209 210

/*
 * do_IRQ handles all normal device IRQ's (the special
 * SMP cross-CPU interrupts have their own specific
 * handlers).
 */
211
__visible unsigned int __irq_entry do_IRQ(struct pt_regs *regs)
J
Jeremy Fitzhardinge 已提交
212 213 214 215 216 217 218
{
	struct pt_regs *old_regs = set_irq_regs(regs);

	/* high bit used in ret_from_ code  */
	unsigned vector = ~regs->orig_ax;
	unsigned irq;

219
	entering_irq();
J
Jeremy Fitzhardinge 已提交
220

T
Tejun Heo 已提交
221
	irq = __this_cpu_read(vector_irq[vector]);
J
Jeremy Fitzhardinge 已提交
222 223

	if (!handle_irq(irq, regs)) {
224
		ack_APIC_irq();
J
Jeremy Fitzhardinge 已提交
225

226 227 228 229 230 231 232
		if (irq != VECTOR_RETRIGGERED) {
			pr_emerg_ratelimited("%s: %d.%d No irq handler for vector (irq %d)\n",
					     __func__, smp_processor_id(),
					     vector, irq);
		} else {
			__this_cpu_write(vector_irq[vector], VECTOR_UNDEFINED);
		}
J
Jeremy Fitzhardinge 已提交
233 234
	}

235
	exiting_irq();
J
Jeremy Fitzhardinge 已提交
236 237 238 239 240

	set_irq_regs(old_regs);
	return 1;
}

241
/*
242
 * Handler for X86_PLATFORM_IPI_VECTOR.
243
 */
244
void __smp_x86_platform_ipi(void)
245
{
246
	inc_irq_stat(x86_platform_ipis);
247

248 249
	if (x86_platform_ipi_callback)
		x86_platform_ipi_callback();
250
}
251

252
__visible void smp_x86_platform_ipi(struct pt_regs *regs)
253 254
{
	struct pt_regs *old_regs = set_irq_regs(regs);
255

256 257 258
	entering_ack_irq();
	__smp_x86_platform_ipi();
	exiting_irq();
259 260 261
	set_irq_regs(old_regs);
}

262
#ifdef CONFIG_HAVE_KVM
263 264 265 266 267 268 269 270 271 272 273 274
static void dummy_handler(void) {}
static void (*kvm_posted_intr_wakeup_handler)(void) = dummy_handler;

void kvm_set_posted_intr_wakeup_handler(void (*handler)(void))
{
	if (handler)
		kvm_posted_intr_wakeup_handler = handler;
	else
		kvm_posted_intr_wakeup_handler = dummy_handler;
}
EXPORT_SYMBOL_GPL(kvm_set_posted_intr_wakeup_handler);

275 276 277
/*
 * Handler for POSTED_INTERRUPT_VECTOR.
 */
278
__visible void smp_kvm_posted_intr_ipi(struct pt_regs *regs)
279 280 281
{
	struct pt_regs *old_regs = set_irq_regs(regs);

282
	entering_ack_irq();
283
	inc_irq_stat(kvm_posted_intr_ipis);
284 285 286 287 288 289 290 291 292 293 294 295 296 297
	exiting_irq();
	set_irq_regs(old_regs);
}

/*
 * Handler for POSTED_INTERRUPT_WAKEUP_VECTOR.
 */
__visible void smp_kvm_posted_intr_wakeup_ipi(struct pt_regs *regs)
{
	struct pt_regs *old_regs = set_irq_regs(regs);

	entering_ack_irq();
	inc_irq_stat(kvm_posted_intr_wakeup_ipis);
	kvm_posted_intr_wakeup_handler();
298
	exiting_irq();
299 300 301 302
	set_irq_regs(old_regs);
}
#endif

303
__visible void smp_trace_x86_platform_ipi(struct pt_regs *regs)
304 305 306 307 308 309 310 311 312 313 314
{
	struct pt_regs *old_regs = set_irq_regs(regs);

	entering_ack_irq();
	trace_x86_platform_ipi_entry(X86_PLATFORM_IPI_VECTOR);
	__smp_x86_platform_ipi();
	trace_x86_platform_ipi_exit(X86_PLATFORM_IPI_VECTOR);
	exiting_irq();
	set_irq_regs(old_regs);
}

315
EXPORT_SYMBOL_GPL(vector_used_by_percpu_irq);
316 317

#ifdef CONFIG_HOTPLUG_CPU
318 319 320 321 322 323 324 325

/* These two declarations are only used in check_irq_vectors_for_cpu_disable()
 * below, which is protected by stop_machine().  Putting them on the stack
 * results in a stack frame overflow.  Dynamically allocating could result in a
 * failure so declare these two cpumasks as global.
 */
static struct cpumask affinity_new, online_new;

326 327 328 329 330 331 332 333 334 335 336 337 338 339
/*
 * This cpu is going to be removed and its vectors migrated to the remaining
 * online cpus.  Check to see if there are enough vectors in the remaining cpus.
 * This function is protected by stop_machine().
 */
int check_irq_vectors_for_cpu_disable(void)
{
	int irq, cpu;
	unsigned int this_cpu, vector, this_count, count;
	struct irq_desc *desc;
	struct irq_data *data;

	this_cpu = smp_processor_id();
	cpumask_copy(&online_new, cpu_online_mask);
340
	cpumask_clear_cpu(this_cpu, &online_new);
341 342 343 344 345 346

	this_count = 0;
	for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) {
		irq = __this_cpu_read(vector_irq[vector]);
		if (irq >= 0) {
			desc = irq_to_desc(irq);
347 348 349
			if (!desc)
				continue;

350 351 352 353 354
			/*
			 * Protect against concurrent action removal,
			 * affinity changes etc.
			 */
			raw_spin_lock(&desc->lock);
355 356
			data = irq_desc_get_irq_data(desc);
			cpumask_copy(&affinity_new, data->affinity);
357
			cpumask_clear_cpu(this_cpu, &affinity_new);
358 359

			/* Do not count inactive or per-cpu irqs. */
360 361
			if (!irq_has_action(irq) || irqd_is_per_cpu(data)) {
				raw_spin_unlock(&desc->lock);
362
				continue;
363
			}
364

365
			raw_spin_unlock(&desc->lock);
366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390
			/*
			 * A single irq may be mapped to multiple
			 * cpu's vector_irq[] (for example IOAPIC cluster
			 * mode).  In this case we have two
			 * possibilities:
			 *
			 * 1) the resulting affinity mask is empty; that is
			 * this the down'd cpu is the last cpu in the irq's
			 * affinity mask, or
			 *
			 * 2) the resulting affinity mask is no longer
			 * a subset of the online cpus but the affinity
			 * mask is not zero; that is the down'd cpu is the
			 * last online cpu in a user set affinity mask.
			 */
			if (cpumask_empty(&affinity_new) ||
			    !cpumask_subset(&affinity_new, &online_new))
				this_count++;
		}
	}

	count = 0;
	for_each_online_cpu(cpu) {
		if (cpu == this_cpu)
			continue;
391 392 393 394 395
		/*
		 * We scan from FIRST_EXTERNAL_VECTOR to first system
		 * vector. If the vector is marked in the used vectors
		 * bitmap or an irq is assigned to it, we don't count
		 * it as available.
396 397 398
		 *
		 * As this is an inaccurate snapshot anyway, we can do
		 * this w/o holding vector_lock.
399 400 401 402 403 404
		 */
		for (vector = FIRST_EXTERNAL_VECTOR;
		     vector < first_system_vector; vector++) {
			if (!test_bit(vector, used_vectors) &&
			    per_cpu(vector_irq, cpu)[vector] < 0)
					count++;
405 406 407 408 409 410 411 412 413 414 415
		}
	}

	if (count < this_count) {
		pr_warn("CPU %d disable failed: CPU has %u vectors assigned and there are only %u available.\n",
			this_cpu, this_count, count);
		return -ERANGE;
	}
	return 0;
}

416 417 418
/* A cpu has been removed from cpu_online_mask.  Reset irq affinities. */
void fixup_irqs(void)
{
419
	unsigned int irq, vector;
420 421
	static int warned;
	struct irq_desc *desc;
422
	struct irq_data *data;
423
	struct irq_chip *chip;
424
	int ret;
425 426 427 428 429 430 431 432 433 434 435 436

	for_each_irq_desc(irq, desc) {
		int break_affinity = 0;
		int set_affinity = 1;
		const struct cpumask *affinity;

		if (!desc)
			continue;
		if (irq == 2)
			continue;

		/* interrupt's are disabled at this point */
437
		raw_spin_lock(&desc->lock);
438

439
		data = irq_desc_get_irq_data(desc);
440
		affinity = data->affinity;
441
		if (!irq_has_action(irq) || irqd_is_per_cpu(data) ||
442
		    cpumask_subset(affinity, cpu_online_mask)) {
443
			raw_spin_unlock(&desc->lock);
444 445 446
			continue;
		}

447 448 449 450 451 452 453
		/*
		 * Complete the irq move. This cpu is going down and for
		 * non intr-remapping case, we can't wait till this interrupt
		 * arrives at this cpu before completing the irq move.
		 */
		irq_force_complete_move(irq);

454 455
		if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) {
			break_affinity = 1;
456
			affinity = cpu_online_mask;
457 458
		}

459 460 461
		chip = irq_data_get_irq_chip(data);
		if (!irqd_can_move_in_process_context(data) && chip->irq_mask)
			chip->irq_mask(data);
462

463 464 465 466 467 468 469 470
		if (chip->irq_set_affinity) {
			ret = chip->irq_set_affinity(data, affinity, true);
			if (ret == -ENOSPC)
				pr_crit("IRQ %d set affinity failed because there are no available vectors.  The device assigned to this IRQ is unstable.\n", irq);
		} else {
			if (!(warned++))
				set_affinity = 0;
		}
471

472 473 474 475 476
		/*
		 * We unmask if the irq was not marked masked by the
		 * core code. That respects the lazy irq disable
		 * behaviour.
		 */
477
		if (!irqd_can_move_in_process_context(data) &&
478
		    !irqd_irq_masked(data) && chip->irq_unmask)
479
			chip->irq_unmask(data);
480

481
		raw_spin_unlock(&desc->lock);
482 483

		if (break_affinity && set_affinity)
484
			pr_notice("Broke affinity for irq %i\n", irq);
485
		else if (!set_affinity)
486
			pr_notice("Cannot set affinity for irq %i\n", irq);
487 488
	}

489 490 491 492 493 494 495 496 497
	/*
	 * We can remove mdelay() and then send spuriuous interrupts to
	 * new cpu targets for all the irqs that were handled previously by
	 * this cpu. While it works, I have seen spurious interrupt messages
	 * (nothing wrong but still...).
	 *
	 * So for now, retain mdelay(1) and check the IRR and then send those
	 * interrupts to new targets as this cpu is already offlined...
	 */
498
	mdelay(1);
499 500 501 502

	for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) {
		unsigned int irr;

503
		if (__this_cpu_read(vector_irq[vector]) <= VECTOR_UNDEFINED)
504 505 506 507
			continue;

		irr = apic_read(APIC_IRR + (vector / 32 * 0x10));
		if (irr  & (1 << (vector % 32))) {
T
Tejun Heo 已提交
508
			irq = __this_cpu_read(vector_irq[vector]);
509

510
			desc = irq_to_desc(irq);
511 512
			data = irq_desc_get_irq_data(desc);
			chip = irq_data_get_irq_chip(data);
513
			raw_spin_lock(&desc->lock);
514
			if (chip->irq_retrigger) {
515
				chip->irq_retrigger(data);
516 517
				__this_cpu_write(vector_irq[vector], VECTOR_RETRIGGERED);
			}
518
			raw_spin_unlock(&desc->lock);
519
		}
520 521
		if (__this_cpu_read(vector_irq[vector]) != VECTOR_RETRIGGERED)
			__this_cpu_write(vector_irq[vector], VECTOR_UNDEFINED);
522
	}
523 524
}
#endif