Linux Audio

Check our new training course

Embedded Linux Audio

Check our new training course
with Creative Commons CC-BY-SA
lecture materials

Bootlin logo

Elixir Cross Referencer

Loading...
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
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
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
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
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
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
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
// SPDX-License-Identifier: GPL-2.0-or-later
/*
 * CPU Microcode Update Driver for Linux
 *
 * Copyright (C) 2000-2006 Tigran Aivazian <aivazian.tigran@gmail.com>
 *	      2006	Shaohua Li <shaohua.li@intel.com>
 *	      2013-2016	Borislav Petkov <bp@alien8.de>
 *
 * X86 CPU microcode early update for Linux:
 *
 *	Copyright (C) 2012 Fenghua Yu <fenghua.yu@intel.com>
 *			   H Peter Anvin" <hpa@zytor.com>
 *		  (C) 2015 Borislav Petkov <bp@alien8.de>
 *
 * This driver allows to upgrade microcode on x86 processors.
 */

#define pr_fmt(fmt) "microcode: " fmt

#include <linux/platform_device.h>
#include <linux/stop_machine.h>
#include <linux/syscore_ops.h>
#include <linux/miscdevice.h>
#include <linux/capability.h>
#include <linux/firmware.h>
#include <linux/cpumask.h>
#include <linux/kernel.h>
#include <linux/delay.h>
#include <linux/mutex.h>
#include <linux/cpu.h>
#include <linux/nmi.h>
#include <linux/fs.h>
#include <linux/mm.h>

#include <asm/apic.h>
#include <asm/cpu_device_id.h>
#include <asm/perf_event.h>
#include <asm/processor.h>
#include <asm/cmdline.h>
#include <asm/setup.h>

#include "internal.h"

static struct microcode_ops	*microcode_ops;
bool dis_ucode_ldr = true;

bool force_minrev = IS_ENABLED(CONFIG_MICROCODE_LATE_FORCE_MINREV);
module_param(force_minrev, bool, S_IRUSR | S_IWUSR);

/*
 * Synchronization.
 *
 * All non cpu-hotplug-callback call sites use:
 *
 * - cpus_read_lock/unlock() to synchronize with
 *   the cpu-hotplug-callback call sites.
 *
 * We guarantee that only a single cpu is being
 * updated at any particular moment of time.
 */
struct ucode_cpu_info		ucode_cpu_info[NR_CPUS];

/*
 * Those patch levels cannot be updated to newer ones and thus should be final.
 */
static u32 final_levels[] = {
	0x01000098,
	0x0100009f,
	0x010000af,
	0, /* T-101 terminator */
};

struct early_load_data early_data;

/*
 * Check the current patch level on this CPU.
 *
 * Returns:
 *  - true: if update should stop
 *  - false: otherwise
 */
static bool amd_check_current_patch_level(void)
{
	u32 lvl, dummy, i;
	u32 *levels;

	native_rdmsr(MSR_AMD64_PATCH_LEVEL, lvl, dummy);

	levels = final_levels;

	for (i = 0; levels[i]; i++) {
		if (lvl == levels[i])
			return true;
	}
	return false;
}

static bool __init check_loader_disabled_bsp(void)
{
	static const char *__dis_opt_str = "dis_ucode_ldr";
	const char *cmdline = boot_command_line;
	const char *option  = __dis_opt_str;

	/*
	 * CPUID(1).ECX[31]: reserved for hypervisor use. This is still not
	 * completely accurate as xen pv guests don't see that CPUID bit set but
	 * that's good enough as they don't land on the BSP path anyway.
	 */
	if (native_cpuid_ecx(1) & BIT(31))
		return true;

	if (x86_cpuid_vendor() == X86_VENDOR_AMD) {
		if (amd_check_current_patch_level())
			return true;
	}

	if (cmdline_find_option_bool(cmdline, option) <= 0)
		dis_ucode_ldr = false;

	return dis_ucode_ldr;
}

void __init load_ucode_bsp(void)
{
	unsigned int cpuid_1_eax;
	bool intel = true;

	if (!have_cpuid_p())
		return;

	cpuid_1_eax = native_cpuid_eax(1);

	switch (x86_cpuid_vendor()) {
	case X86_VENDOR_INTEL:
		if (x86_family(cpuid_1_eax) < 6)
			return;
		break;

	case X86_VENDOR_AMD:
		if (x86_family(cpuid_1_eax) < 0x10)
			return;
		intel = false;
		break;

	default:
		return;
	}

	if (check_loader_disabled_bsp())
		return;

	if (intel)
		load_ucode_intel_bsp(&early_data);
	else
		load_ucode_amd_bsp(&early_data, cpuid_1_eax);
}

void load_ucode_ap(void)
{
	unsigned int cpuid_1_eax;

	if (dis_ucode_ldr)
		return;

	cpuid_1_eax = native_cpuid_eax(1);

	switch (x86_cpuid_vendor()) {
	case X86_VENDOR_INTEL:
		if (x86_family(cpuid_1_eax) >= 6)
			load_ucode_intel_ap();
		break;
	case X86_VENDOR_AMD:
		if (x86_family(cpuid_1_eax) >= 0x10)
			load_ucode_amd_ap(cpuid_1_eax);
		break;
	default:
		break;
	}
}

struct cpio_data __init find_microcode_in_initrd(const char *path)
{
#ifdef CONFIG_BLK_DEV_INITRD
	unsigned long start = 0;
	size_t size;

#ifdef CONFIG_X86_32
	size = boot_params.hdr.ramdisk_size;
	/* Early load on BSP has a temporary mapping. */
	if (size)
		start = initrd_start_early;

#else /* CONFIG_X86_64 */
	size  = (unsigned long)boot_params.ext_ramdisk_size << 32;
	size |= boot_params.hdr.ramdisk_size;

	if (size) {
		start  = (unsigned long)boot_params.ext_ramdisk_image << 32;
		start |= boot_params.hdr.ramdisk_image;
		start += PAGE_OFFSET;
	}
#endif

	/*
	 * Fixup the start address: after reserve_initrd() runs, initrd_start
	 * has the virtual address of the beginning of the initrd. It also
	 * possibly relocates the ramdisk. In either case, initrd_start contains
	 * the updated address so use that instead.
	 */
	if (initrd_start)
		start = initrd_start;

	return find_cpio_data(path, (void *)start, size, NULL);
#else /* !CONFIG_BLK_DEV_INITRD */
	return (struct cpio_data){ NULL, 0, "" };
#endif
}

static void reload_early_microcode(unsigned int cpu)
{
	int vendor, family;

	vendor = x86_cpuid_vendor();
	family = x86_cpuid_family();

	switch (vendor) {
	case X86_VENDOR_INTEL:
		if (family >= 6)
			reload_ucode_intel();
		break;
	case X86_VENDOR_AMD:
		if (family >= 0x10)
			reload_ucode_amd(cpu);
		break;
	default:
		break;
	}
}

/* fake device for request_firmware */
static struct platform_device	*microcode_pdev;

#ifdef CONFIG_MICROCODE_LATE_LOADING
/*
 * Late loading dance. Why the heavy-handed stomp_machine effort?
 *
 * - HT siblings must be idle and not execute other code while the other sibling
 *   is loading microcode in order to avoid any negative interactions caused by
 *   the loading.
 *
 * - In addition, microcode update on the cores must be serialized until this
 *   requirement can be relaxed in the future. Right now, this is conservative
 *   and good.
 */
enum sibling_ctrl {
	/* Spinwait with timeout */
	SCTRL_WAIT,
	/* Invoke the microcode_apply() callback */
	SCTRL_APPLY,
	/* Proceed without invoking the microcode_apply() callback */
	SCTRL_DONE,
};

struct microcode_ctrl {
	enum sibling_ctrl	ctrl;
	enum ucode_state	result;
	unsigned int		ctrl_cpu;
	bool			nmi_enabled;
};

DEFINE_STATIC_KEY_FALSE(microcode_nmi_handler_enable);
static DEFINE_PER_CPU(struct microcode_ctrl, ucode_ctrl);
static atomic_t late_cpus_in, offline_in_nmi;
static unsigned int loops_per_usec;
static cpumask_t cpu_offline_mask;

static noinstr bool wait_for_cpus(atomic_t *cnt)
{
	unsigned int timeout, loops;

	WARN_ON_ONCE(raw_atomic_dec_return(cnt) < 0);

	for (timeout = 0; timeout < USEC_PER_SEC; timeout++) {
		if (!raw_atomic_read(cnt))
			return true;

		for (loops = 0; loops < loops_per_usec; loops++)
			cpu_relax();

		/* If invoked directly, tickle the NMI watchdog */
		if (!microcode_ops->use_nmi && !(timeout % USEC_PER_MSEC)) {
			instrumentation_begin();
			touch_nmi_watchdog();
			instrumentation_end();
		}
	}
	/* Prevent the late comers from making progress and let them time out */
	raw_atomic_inc(cnt);
	return false;
}

static noinstr bool wait_for_ctrl(void)
{
	unsigned int timeout, loops;

	for (timeout = 0; timeout < USEC_PER_SEC; timeout++) {
		if (raw_cpu_read(ucode_ctrl.ctrl) != SCTRL_WAIT)
			return true;

		for (loops = 0; loops < loops_per_usec; loops++)
			cpu_relax();

		/* If invoked directly, tickle the NMI watchdog */
		if (!microcode_ops->use_nmi && !(timeout % USEC_PER_MSEC)) {
			instrumentation_begin();
			touch_nmi_watchdog();
			instrumentation_end();
		}
	}
	return false;
}

/*
 * Protected against instrumentation up to the point where the primary
 * thread completed the update. See microcode_nmi_handler() for details.
 */
static noinstr bool load_secondary_wait(unsigned int ctrl_cpu)
{
	/* Initial rendezvous to ensure that all CPUs have arrived */
	if (!wait_for_cpus(&late_cpus_in)) {
		raw_cpu_write(ucode_ctrl.result, UCODE_TIMEOUT);
		return false;
	}

	/*
	 * Wait for primary threads to complete. If one of them hangs due
	 * to the update, there is no way out. This is non-recoverable
	 * because the CPU might hold locks or resources and confuse the
	 * scheduler, watchdogs etc. There is no way to safely evacuate the
	 * machine.
	 */
	if (wait_for_ctrl())
		return true;

	instrumentation_begin();
	panic("Microcode load: Primary CPU %d timed out\n", ctrl_cpu);
	instrumentation_end();
}

/*
 * Protected against instrumentation up to the point where the primary
 * thread completed the update. See microcode_nmi_handler() for details.
 */
static noinstr void load_secondary(unsigned int cpu)
{
	unsigned int ctrl_cpu = raw_cpu_read(ucode_ctrl.ctrl_cpu);
	enum ucode_state ret;

	if (!load_secondary_wait(ctrl_cpu)) {
		instrumentation_begin();
		pr_err_once("load: %d CPUs timed out\n",
			    atomic_read(&late_cpus_in) - 1);
		instrumentation_end();
		return;
	}

	/* Primary thread completed. Allow to invoke instrumentable code */
	instrumentation_begin();
	/*
	 * If the primary succeeded then invoke the apply() callback,
	 * otherwise copy the state from the primary thread.
	 */
	if (this_cpu_read(ucode_ctrl.ctrl) == SCTRL_APPLY)
		ret = microcode_ops->apply_microcode(cpu);
	else
		ret = per_cpu(ucode_ctrl.result, ctrl_cpu);

	this_cpu_write(ucode_ctrl.result, ret);
	this_cpu_write(ucode_ctrl.ctrl, SCTRL_DONE);
	instrumentation_end();
}

static void __load_primary(unsigned int cpu)
{
	struct cpumask *secondaries = topology_sibling_cpumask(cpu);
	enum sibling_ctrl ctrl;
	enum ucode_state ret;
	unsigned int sibling;

	/* Initial rendezvous to ensure that all CPUs have arrived */
	if (!wait_for_cpus(&late_cpus_in)) {
		this_cpu_write(ucode_ctrl.result, UCODE_TIMEOUT);
		pr_err_once("load: %d CPUs timed out\n", atomic_read(&late_cpus_in) - 1);
		return;
	}

	ret = microcode_ops->apply_microcode(cpu);
	this_cpu_write(ucode_ctrl.result, ret);
	this_cpu_write(ucode_ctrl.ctrl, SCTRL_DONE);

	/*
	 * If the update was successful, let the siblings run the apply()
	 * callback. If not, tell them it's done. This also covers the
	 * case where the CPU has uniform loading at package or system
	 * scope implemented but does not advertise it.
	 */
	if (ret == UCODE_UPDATED || ret == UCODE_OK)
		ctrl = SCTRL_APPLY;
	else
		ctrl = SCTRL_DONE;

	for_each_cpu(sibling, secondaries) {
		if (sibling != cpu)
			per_cpu(ucode_ctrl.ctrl, sibling) = ctrl;
	}
}

static bool kick_offline_cpus(unsigned int nr_offl)
{
	unsigned int cpu, timeout;

	for_each_cpu(cpu, &cpu_offline_mask) {
		/* Enable the rendezvous handler and send NMI */
		per_cpu(ucode_ctrl.nmi_enabled, cpu) = true;
		apic_send_nmi_to_offline_cpu(cpu);
	}

	/* Wait for them to arrive */
	for (timeout = 0; timeout < (USEC_PER_SEC / 2); timeout++) {
		if (atomic_read(&offline_in_nmi) == nr_offl)
			return true;
		udelay(1);
	}
	/* Let the others time out */
	return false;
}

static void release_offline_cpus(void)
{
	unsigned int cpu;

	for_each_cpu(cpu, &cpu_offline_mask)
		per_cpu(ucode_ctrl.ctrl, cpu) = SCTRL_DONE;
}

static void load_primary(unsigned int cpu)
{
	unsigned int nr_offl = cpumask_weight(&cpu_offline_mask);
	bool proceed = true;

	/* Kick soft-offlined SMT siblings if required */
	if (!cpu && nr_offl)
		proceed = kick_offline_cpus(nr_offl);

	/* If the soft-offlined CPUs did not respond, abort */
	if (proceed)
		__load_primary(cpu);

	/* Unconditionally release soft-offlined SMT siblings if required */
	if (!cpu && nr_offl)
		release_offline_cpus();
}

/*
 * Minimal stub rendezvous handler for soft-offlined CPUs which participate
 * in the NMI rendezvous to protect against a concurrent NMI on affected
 * CPUs.
 */
void noinstr microcode_offline_nmi_handler(void)
{
	if (!raw_cpu_read(ucode_ctrl.nmi_enabled))
		return;
	raw_cpu_write(ucode_ctrl.nmi_enabled, false);
	raw_cpu_write(ucode_ctrl.result, UCODE_OFFLINE);
	raw_atomic_inc(&offline_in_nmi);
	wait_for_ctrl();
}

static noinstr bool microcode_update_handler(void)
{
	unsigned int cpu = raw_smp_processor_id();

	if (raw_cpu_read(ucode_ctrl.ctrl_cpu) == cpu) {
		instrumentation_begin();
		load_primary(cpu);
		instrumentation_end();
	} else {
		load_secondary(cpu);
	}

	instrumentation_begin();
	touch_nmi_watchdog();
	instrumentation_end();

	return true;
}

/*
 * Protection against instrumentation is required for CPUs which are not
 * safe against an NMI which is delivered to the secondary SMT sibling
 * while the primary thread updates the microcode. Instrumentation can end
 * up in #INT3, #DB and #PF. The IRET from those exceptions reenables NMI
 * which is the opposite of what the NMI rendezvous is trying to achieve.
 *
 * The primary thread is safe versus instrumentation as the actual
 * microcode update handles this correctly. It's only the sibling code
 * path which must be NMI safe until the primary thread completed the
 * update.
 */
bool noinstr microcode_nmi_handler(void)
{
	if (!raw_cpu_read(ucode_ctrl.nmi_enabled))
		return false;

	raw_cpu_write(ucode_ctrl.nmi_enabled, false);
	return microcode_update_handler();
}

static int load_cpus_stopped(void *unused)
{
	if (microcode_ops->use_nmi) {
		/* Enable the NMI handler and raise NMI */
		this_cpu_write(ucode_ctrl.nmi_enabled, true);
		apic->send_IPI(smp_processor_id(), NMI_VECTOR);
	} else {
		/* Just invoke the handler directly */
		microcode_update_handler();
	}
	return 0;
}

static int load_late_stop_cpus(bool is_safe)
{
	unsigned int cpu, updated = 0, failed = 0, timedout = 0, siblings = 0;
	unsigned int nr_offl, offline = 0;
	int old_rev = boot_cpu_data.microcode;
	struct cpuinfo_x86 prev_info;

	if (!is_safe) {
		pr_err("Late microcode loading without minimal revision check.\n");
		pr_err("You should switch to early loading, if possible.\n");
	}

	atomic_set(&late_cpus_in, num_online_cpus());
	atomic_set(&offline_in_nmi, 0);
	loops_per_usec = loops_per_jiffy / (TICK_NSEC / 1000);

	/*
	 * Take a snapshot before the microcode update in order to compare and
	 * check whether any bits changed after an update.
	 */
	store_cpu_caps(&prev_info);

	if (microcode_ops->use_nmi)
		static_branch_enable_cpuslocked(&microcode_nmi_handler_enable);

	stop_machine_cpuslocked(load_cpus_stopped, NULL, cpu_online_mask);

	if (microcode_ops->use_nmi)
		static_branch_disable_cpuslocked(&microcode_nmi_handler_enable);

	/* Analyze the results */
	for_each_cpu_and(cpu, cpu_present_mask, &cpus_booted_once_mask) {
		switch (per_cpu(ucode_ctrl.result, cpu)) {
		case UCODE_UPDATED:	updated++; break;
		case UCODE_TIMEOUT:	timedout++; break;
		case UCODE_OK:		siblings++; break;
		case UCODE_OFFLINE:	offline++; break;
		default:		failed++; break;
		}
	}

	if (microcode_ops->finalize_late_load)
		microcode_ops->finalize_late_load(!updated);

	if (!updated) {
		/* Nothing changed. */
		if (!failed && !timedout)
			return 0;

		nr_offl = cpumask_weight(&cpu_offline_mask);
		if (offline < nr_offl) {
			pr_warn("%u offline siblings did not respond.\n",
				nr_offl - atomic_read(&offline_in_nmi));
			return -EIO;
		}
		pr_err("update failed: %u CPUs failed %u CPUs timed out\n",
		       failed, timedout);
		return -EIO;
	}

	if (!is_safe || failed || timedout)
		add_taint(TAINT_CPU_OUT_OF_SPEC, LOCKDEP_STILL_OK);

	pr_info("load: updated on %u primary CPUs with %u siblings\n", updated, siblings);
	if (failed || timedout) {
		pr_err("load incomplete. %u CPUs timed out or failed\n",
		       num_online_cpus() - (updated + siblings));
	}
	pr_info("revision: 0x%x -> 0x%x\n", old_rev, boot_cpu_data.microcode);
	microcode_check(&prev_info);

	return updated + siblings == num_online_cpus() ? 0 : -EIO;
}

/*
 * This function does two things:
 *
 * 1) Ensure that all required CPUs which are present and have been booted
 *    once are online.
 *
 *    To pass this check, all primary threads must be online.
 *
 *    If the microcode load is not safe against NMI then all SMT threads
 *    must be online as well because they still react to NMIs when they are
 *    soft-offlined and parked in one of the play_dead() variants. So if a
 *    NMI hits while the primary thread updates the microcode the resulting
 *    behaviour is undefined. The default play_dead() implementation on
 *    modern CPUs uses MWAIT, which is also not guaranteed to be safe
 *    against a microcode update which affects MWAIT.
 *
 *    As soft-offlined CPUs still react on NMIs, the SMT sibling
 *    restriction can be lifted when the vendor driver signals to use NMI
 *    for rendezvous and the APIC provides a mechanism to send an NMI to a
 *    soft-offlined CPU. The soft-offlined CPUs are then able to
 *    participate in the rendezvous in a trivial stub handler.
 *
 * 2) Initialize the per CPU control structure and create a cpumask
 *    which contains "offline"; secondary threads, so they can be handled
 *    correctly by a control CPU.
 */
static bool setup_cpus(void)
{
	struct microcode_ctrl ctrl = { .ctrl = SCTRL_WAIT, .result = -1, };
	bool allow_smt_offline;
	unsigned int cpu;

	allow_smt_offline = microcode_ops->nmi_safe ||
		(microcode_ops->use_nmi && apic->nmi_to_offline_cpu);

	cpumask_clear(&cpu_offline_mask);

	for_each_cpu_and(cpu, cpu_present_mask, &cpus_booted_once_mask) {
		/*
		 * Offline CPUs sit in one of the play_dead() functions
		 * with interrupts disabled, but they still react on NMIs
		 * and execute arbitrary code. Also MWAIT being updated
		 * while the offline CPU sits there is not necessarily safe
		 * on all CPU variants.
		 *
		 * Mark them in the offline_cpus mask which will be handled
		 * by CPU0 later in the update process.
		 *
		 * Ensure that the primary thread is online so that it is
		 * guaranteed that all cores are updated.
		 */
		if (!cpu_online(cpu)) {
			if (topology_is_primary_thread(cpu) || !allow_smt_offline) {
				pr_err("CPU %u not online, loading aborted\n", cpu);
				return false;
			}
			cpumask_set_cpu(cpu, &cpu_offline_mask);
			per_cpu(ucode_ctrl, cpu) = ctrl;
			continue;
		}

		/*
		 * Initialize the per CPU state. This is core scope for now,
		 * but prepared to take package or system scope into account.
		 */
		ctrl.ctrl_cpu = cpumask_first(topology_sibling_cpumask(cpu));
		per_cpu(ucode_ctrl, cpu) = ctrl;
	}
	return true;
}

static int load_late_locked(void)
{
	if (!setup_cpus())
		return -EBUSY;

	switch (microcode_ops->request_microcode_fw(0, &microcode_pdev->dev)) {
	case UCODE_NEW:
		return load_late_stop_cpus(false);
	case UCODE_NEW_SAFE:
		return load_late_stop_cpus(true);
	case UCODE_NFOUND:
		return -ENOENT;
	default:
		return -EBADFD;
	}
}

static ssize_t reload_store(struct device *dev,
			    struct device_attribute *attr,
			    const char *buf, size_t size)
{
	unsigned long val;
	ssize_t ret;

	ret = kstrtoul(buf, 0, &val);
	if (ret || val != 1)
		return -EINVAL;

	cpus_read_lock();
	ret = load_late_locked();
	cpus_read_unlock();

	return ret ? : size;
}

static DEVICE_ATTR_WO(reload);
#endif

static ssize_t version_show(struct device *dev,
			struct device_attribute *attr, char *buf)
{
	struct ucode_cpu_info *uci = ucode_cpu_info + dev->id;

	return sprintf(buf, "0x%x\n", uci->cpu_sig.rev);
}

static ssize_t processor_flags_show(struct device *dev,
			struct device_attribute *attr, char *buf)
{
	struct ucode_cpu_info *uci = ucode_cpu_info + dev->id;

	return sprintf(buf, "0x%x\n", uci->cpu_sig.pf);
}

static DEVICE_ATTR_RO(version);
static DEVICE_ATTR_RO(processor_flags);

static struct attribute *mc_default_attrs[] = {
	&dev_attr_version.attr,
	&dev_attr_processor_flags.attr,
	NULL
};

static const struct attribute_group mc_attr_group = {
	.attrs			= mc_default_attrs,
	.name			= "microcode",
};

static void microcode_fini_cpu(int cpu)
{
	if (microcode_ops->microcode_fini_cpu)
		microcode_ops->microcode_fini_cpu(cpu);
}

/**
 * microcode_bsp_resume - Update boot CPU microcode during resume.
 */
void microcode_bsp_resume(void)
{
	int cpu = smp_processor_id();
	struct ucode_cpu_info *uci = ucode_cpu_info + cpu;

	if (uci->mc)
		microcode_ops->apply_microcode(cpu);
	else
		reload_early_microcode(cpu);
}

static struct syscore_ops mc_syscore_ops = {
	.resume	= microcode_bsp_resume,
};

static int mc_cpu_online(unsigned int cpu)
{
	struct ucode_cpu_info *uci = ucode_cpu_info + cpu;
	struct device *dev = get_cpu_device(cpu);

	memset(uci, 0, sizeof(*uci));

	microcode_ops->collect_cpu_info(cpu, &uci->cpu_sig);
	cpu_data(cpu).microcode = uci->cpu_sig.rev;
	if (!cpu)
		boot_cpu_data.microcode = uci->cpu_sig.rev;

	if (sysfs_create_group(&dev->kobj, &mc_attr_group))
		pr_err("Failed to create group for CPU%d\n", cpu);
	return 0;
}

static int mc_cpu_down_prep(unsigned int cpu)
{
	struct device *dev = get_cpu_device(cpu);

	microcode_fini_cpu(cpu);
	sysfs_remove_group(&dev->kobj, &mc_attr_group);
	return 0;
}

static struct attribute *cpu_root_microcode_attrs[] = {
#ifdef CONFIG_MICROCODE_LATE_LOADING
	&dev_attr_reload.attr,
#endif
	NULL
};

static const struct attribute_group cpu_root_microcode_group = {
	.name  = "microcode",
	.attrs = cpu_root_microcode_attrs,
};

static int __init microcode_init(void)
{
	struct device *dev_root;
	struct cpuinfo_x86 *c = &boot_cpu_data;
	int error;

	if (dis_ucode_ldr)
		return -EINVAL;

	if (c->x86_vendor == X86_VENDOR_INTEL)
		microcode_ops = init_intel_microcode();
	else if (c->x86_vendor == X86_VENDOR_AMD)
		microcode_ops = init_amd_microcode();
	else
		pr_err("no support for this CPU vendor\n");

	if (!microcode_ops)
		return -ENODEV;

	pr_info_once("Current revision: 0x%08x\n", (early_data.new_rev ?: early_data.old_rev));

	if (early_data.new_rev)
		pr_info_once("Updated early from: 0x%08x\n", early_data.old_rev);

	microcode_pdev = platform_device_register_simple("microcode", -1, NULL, 0);
	if (IS_ERR(microcode_pdev))
		return PTR_ERR(microcode_pdev);

	dev_root = bus_get_dev_root(&cpu_subsys);
	if (dev_root) {
		error = sysfs_create_group(&dev_root->kobj, &cpu_root_microcode_group);
		put_device(dev_root);
		if (error) {
			pr_err("Error creating microcode group!\n");
			goto out_pdev;
		}
	}

	register_syscore_ops(&mc_syscore_ops);
	cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "x86/microcode:online",
			  mc_cpu_online, mc_cpu_down_prep);

	return 0;

 out_pdev:
	platform_device_unregister(microcode_pdev);
	return error;

}
late_initcall(microcode_init);