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
|
/* Copyright 2013-2014 IBM Corp.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
* implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
/*
* TODO: Index array by PIR to be able to catch them easily
* from assembly such as machine checks etc...
*/
#include <skiboot.h>
#include <cpu.h>
#include <device.h>
#include <opal.h>
#include <stack.h>
#include <trace.h>
#include <affinity.h>
#include <chip.h>
#include <timebase.h>
#include <ccan/str/str.h>
#include <ccan/container_of/container_of.h>
/* The cpu_threads array is static and indexed by PIR in
* order to speed up lookup from asm entry points
*/
struct cpu_stack {
union {
uint8_t stack[STACK_SIZE];
struct cpu_thread cpu;
};
} __align(STACK_SIZE);
static struct cpu_stack *cpu_stacks = (struct cpu_stack *)CPU_STACKS_BASE;
unsigned int cpu_thread_count;
unsigned int cpu_max_pir;
struct cpu_thread *boot_cpu;
static struct lock reinit_lock = LOCK_UNLOCKED;
unsigned long cpu_secondary_start __force_data = 0;
struct cpu_job {
struct list_node link;
void (*func)(void *data);
void *data;
bool complete;
bool no_return;
};
/* attribute const as cpu_stacks is constant. */
void __attrconst *cpu_stack_bottom(unsigned int pir)
{
return (void *)&cpu_stacks[pir] + sizeof(struct cpu_thread);
}
void __attrconst *cpu_stack_top(unsigned int pir)
{
/* This is the top of the MC stack which is above the normal
* stack, which means a SP between cpu_stack_bottom() and
* cpu_stack_top() can either be a normal stack pointer or
* a Machine Check stack pointer
*/
return (void *)&cpu_stacks[pir] + STACK_SIZE - STACK_TOP_GAP;
}
void __nomcount cpu_relax(void)
{
/* Relax a bit to give sibling threads some breathing space */
smt_low();
smt_very_low();
asm volatile("nop; nop; nop; nop\n");
asm volatile("nop; nop; nop; nop\n");
asm volatile("nop; nop; nop; nop\n");
asm volatile("nop; nop; nop; nop\n");
smt_medium();
}
struct cpu_job *__cpu_queue_job(struct cpu_thread *cpu,
void (*func)(void *data), void *data,
bool no_return)
{
struct cpu_job *job;
if (!cpu_is_available(cpu)) {
prerror("CPU: Tried to queue job on unavailable CPU 0x%04x\n",
cpu->pir);
return NULL;
}
job = zalloc(sizeof(struct cpu_job));
if (!job)
return NULL;
job->func = func;
job->data = data;
job->complete = false;
job->no_return = no_return;
if (cpu != this_cpu()) {
lock(&cpu->job_lock);
list_add_tail(&cpu->job_queue, &job->link);
unlock(&cpu->job_lock);
} else {
func(data);
job->complete = true;
}
/* XXX Add poking of CPU with interrupt */
return job;
}
bool cpu_poll_job(struct cpu_job *job)
{
lwsync();
return job->complete;
}
void cpu_wait_job(struct cpu_job *job, bool free_it)
{
unsigned long ticks = usecs_to_tb(5);
if (!job)
return;
while(!job->complete) {
time_wait(ticks);
lwsync();
}
lwsync();
smt_medium();
if (free_it)
free(job);
}
void cpu_free_job(struct cpu_job *job)
{
if (!job)
return;
assert(job->complete);
free(job);
}
void cpu_process_jobs(void)
{
struct cpu_thread *cpu = this_cpu();
struct cpu_job *job;
void (*func)(void *);
void *data;
sync();
if (list_empty(&cpu->job_queue))
return;
lock(&cpu->job_lock);
while (true) {
bool no_return;
if (list_empty(&cpu->job_queue))
break;
smt_medium();
job = list_pop(&cpu->job_queue, struct cpu_job, link);
if (!job)
break;
func = job->func;
data = job->data;
no_return = job->no_return;
unlock(&cpu->job_lock);
if (no_return)
free(job);
func(data);
lock(&cpu->job_lock);
if (!no_return) {
lwsync();
job->complete = true;
}
}
unlock(&cpu->job_lock);
}
struct dt_node *get_cpu_node(u32 pir)
{
struct cpu_thread *t = find_cpu_by_pir(pir);
return t ? t->node : NULL;
}
/* This only covers primary, active cpus */
struct cpu_thread *find_cpu_by_chip_id(u32 chip_id)
{
struct cpu_thread *t;
for_each_available_cpu(t) {
if (t->is_secondary)
continue;
if (t->chip_id == chip_id)
return t;
}
return NULL;
}
struct cpu_thread *find_cpu_by_node(struct dt_node *cpu)
{
struct cpu_thread *t;
for_each_available_cpu(t) {
if (t->node == cpu)
return t;
}
return NULL;
}
struct cpu_thread *find_cpu_by_pir(u32 pir)
{
if (pir > cpu_max_pir)
return NULL;
return &cpu_stacks[pir].cpu;
}
struct cpu_thread *find_cpu_by_server(u32 server_no)
{
struct cpu_thread *t;
for_each_cpu(t) {
if (t->server_no == server_no)
return t;
}
return NULL;
}
struct cpu_thread *next_cpu(struct cpu_thread *cpu)
{
struct cpu_stack *s = container_of(cpu, struct cpu_stack, cpu);
unsigned int index;
if (cpu == NULL)
index = 0;
else
index = s - cpu_stacks + 1;
for (; index <= cpu_max_pir; index++) {
cpu = &cpu_stacks[index].cpu;
if (cpu->state != cpu_state_no_cpu)
return cpu;
}
return NULL;
}
struct cpu_thread *first_cpu(void)
{
return next_cpu(NULL);
}
struct cpu_thread *next_available_cpu(struct cpu_thread *cpu)
{
do {
cpu = next_cpu(cpu);
} while(cpu && !cpu_is_available(cpu));
return cpu;
}
struct cpu_thread *first_available_cpu(void)
{
return next_available_cpu(NULL);
}
struct cpu_thread *next_available_core_in_chip(struct cpu_thread *core,
u32 chip_id)
{
do {
core = next_cpu(core);
} while(core && (!cpu_is_available(core) ||
core->chip_id != chip_id ||
core->is_secondary));
return core;
}
struct cpu_thread *first_available_core_in_chip(u32 chip_id)
{
return next_available_core_in_chip(NULL, chip_id);
}
uint32_t cpu_get_core_index(struct cpu_thread *cpu)
{
return pir_to_core_id(cpu->pir);
}
void cpu_remove_node(const struct cpu_thread *t)
{
struct dt_node *i;
/* Find this cpu node */
dt_for_each_node(dt_root, i) {
const struct dt_property *p;
if (!dt_has_node_property(i, "device_type", "cpu"))
continue;
p = dt_find_property(i, "ibm,pir");
if (dt_property_get_cell(p, 0) == t->pir) {
dt_free(i);
return;
}
}
prerror("CPU: Could not find cpu node %i to remove!\n", t->pir);
abort();
}
void cpu_disable_all_threads(struct cpu_thread *cpu)
{
unsigned int i;
for (i = 0; i <= cpu_max_pir; i++) {
struct cpu_thread *t = &cpu_stacks[i].cpu;
if (t->primary == cpu->primary)
t->state = cpu_state_disabled;
}
/* XXX Do something to actually stop the core */
}
static void init_cpu_thread(struct cpu_thread *t,
enum cpu_thread_state state,
unsigned int pir)
{
init_lock(&t->job_lock);
list_head_init(&t->job_queue);
t->state = state;
t->pir = pir;
t->stack_bot_mark = LONG_MAX;
assert(pir == container_of(t, struct cpu_stack, cpu) - cpu_stacks);
}
void pre_init_boot_cpu(void)
{
struct cpu_thread *cpu = this_cpu();
memset(cpu, 0, sizeof(struct cpu_thread));
}
void init_boot_cpu(void)
{
unsigned int i, pir, pvr;
pir = mfspr(SPR_PIR);
pvr = mfspr(SPR_PVR);
/* Get a CPU thread count and an initial max PIR based on PVR */
switch(PVR_TYPE(pvr)) {
case PVR_TYPE_P7:
case PVR_TYPE_P7P:
cpu_thread_count = 4;
cpu_max_pir = SPR_PIR_P7_MASK;
proc_gen = proc_gen_p7;
prlog(PR_INFO, "CPU: P7 generation processor"
"(max %d threads/core)\n", cpu_thread_count);
break;
case PVR_TYPE_P8E:
case PVR_TYPE_P8:
cpu_thread_count = 8;
cpu_max_pir = SPR_PIR_P8_MASK;
proc_gen = proc_gen_p8;
prlog(PR_INFO, "CPU: P8 generation processor"
"(max %d threads/core)\n", cpu_thread_count);
break;
default:
prerror("CPU: Unknown PVR, assuming 1 thread\n");
cpu_thread_count = 1;
cpu_max_pir = mfspr(SPR_PIR);
proc_gen = proc_gen_unknown;
}
prlog(PR_DEBUG, "CPU: Boot CPU PIR is 0x%04x PVR is 0x%08x\n",
pir, pvr);
prlog(PR_DEBUG, "CPU: Initial max PIR set to 0x%x\n", cpu_max_pir);
/* Clear the CPU structs */
for (i = 0; i <= cpu_max_pir; i++)
memset(&cpu_stacks[i].cpu, 0, sizeof(struct cpu_thread));
/* Setup boot CPU state */
boot_cpu = &cpu_stacks[pir].cpu;
init_cpu_thread(boot_cpu, cpu_state_active, pir);
init_boot_tracebuf(boot_cpu);
assert(this_cpu() == boot_cpu);
}
void init_all_cpus(void)
{
struct dt_node *cpus, *cpu;
unsigned int thread, new_max_pir = 0;
cpus = dt_find_by_path(dt_root, "/cpus");
assert(cpus);
/* Iterate all CPUs in the device-tree */
dt_for_each_child(cpus, cpu) {
unsigned int pir, server_no, chip_id;
enum cpu_thread_state state;
const struct dt_property *p;
struct cpu_thread *t, *pt;
/* Skip cache nodes */
if (strcmp(dt_prop_get(cpu, "device_type"), "cpu"))
continue;
server_no = dt_prop_get_u32(cpu, "reg");
/* If PIR property is absent, assume it's the same as the
* server number
*/
pir = dt_prop_get_u32_def(cpu, "ibm,pir", server_no);
/* We should always have an ibm,chip-id property */
chip_id = dt_get_chip_id(cpu);
/* Only use operational CPUs */
if (!strcmp(dt_prop_get(cpu, "status"), "okay"))
state = cpu_state_present;
else
state = cpu_state_unavailable;
prlog(PR_INFO, "CPU: CPU from DT PIR=0x%04x Server#=0x%x"
" State=%d\n", pir, server_no, state);
/* Setup thread 0 */
t = pt = &cpu_stacks[pir].cpu;
if (t != boot_cpu) {
init_cpu_thread(t, state, pir);
/* Each cpu gets its own later in init_trace_buffers */
t->trace = boot_cpu->trace;
}
t->server_no = server_no;
t->primary = t;
t->node = cpu;
t->chip_id = chip_id;
t->icp_regs = 0; /* Will be set later */
/* Add associativity properties */
add_core_associativity(t);
/* Adjust max PIR */
if (new_max_pir < (pir + cpu_thread_count - 1))
new_max_pir = pir + cpu_thread_count - 1;
/* Iterate threads */
p = dt_find_property(cpu, "ibm,ppc-interrupt-server#s");
if (!p)
continue;
for (thread = 1; thread < (p->len / 4); thread++) {
prlog(PR_TRACE, "CPU: secondary thread %d found\n",
thread);
t = &cpu_stacks[pir + thread].cpu;
init_cpu_thread(t, state, pir + thread);
t->trace = boot_cpu->trace;
t->server_no = ((const u32 *)p->prop)[thread];
t->is_secondary = true;
t->primary = pt;
t->node = cpu;
t->chip_id = chip_id;
}
prlog(PR_INFO, "CPU: %d secondary threads\n", thread);
}
cpu_max_pir = new_max_pir;
prlog(PR_DEBUG, "CPU: New max PIR set to 0x%x\n", new_max_pir);
}
void cpu_bringup(void)
{
struct cpu_thread *t;
prlog(PR_INFO, "CPU: Setting up secondary CPU state\n");
op_display(OP_LOG, OP_MOD_CPU, 0x0000);
/* Tell everybody to chime in ! */
prlog(PR_INFO, "CPU: Calling in all processors...\n");
cpu_secondary_start = 1;
sync();
op_display(OP_LOG, OP_MOD_CPU, 0x0002);
for_each_cpu(t) {
if (t->state != cpu_state_present &&
t->state != cpu_state_active)
continue;
/* Add a callin timeout ? If so, call cpu_remove_node(t). */
while (t->state != cpu_state_active) {
smt_very_low();
sync();
}
smt_medium();
}
prlog(PR_INFO, "CPU: All processors called in...\n");
op_display(OP_LOG, OP_MOD_CPU, 0x0003);
}
void cpu_callin(struct cpu_thread *cpu)
{
cpu->state = cpu_state_active;
}
static void opal_start_thread_job(void *data)
{
cpu_give_self_os();
/* We do not return, so let's mark the job as
* complete
*/
start_kernel_secondary((uint64_t)data);
}
static int64_t opal_start_cpu_thread(uint64_t server_no, uint64_t start_address)
{
struct cpu_thread *cpu;
struct cpu_job *job;
cpu = find_cpu_by_server(server_no);
if (!cpu) {
prerror("OPAL: Start invalid CPU 0x%04llx !\n", server_no);
return OPAL_PARAMETER;
}
prlog(PR_DEBUG, "OPAL: Start CPU 0x%04llx (PIR 0x%04x) -> 0x%016llx\n",
server_no, cpu->pir, start_address);
lock(&reinit_lock);
if (!cpu_is_available(cpu)) {
unlock(&reinit_lock);
prerror("OPAL: CPU not active in OPAL !\n");
return OPAL_WRONG_STATE;
}
job = __cpu_queue_job(cpu, opal_start_thread_job, (void *)start_address,
true);
unlock(&reinit_lock);
if (!job) {
prerror("OPAL: Failed to create CPU start job !\n");
return OPAL_INTERNAL_ERROR;
}
return OPAL_SUCCESS;
}
opal_call(OPAL_START_CPU, opal_start_cpu_thread, 2);
static int64_t opal_query_cpu_status(uint64_t server_no, uint8_t *thread_status)
{
struct cpu_thread *cpu;
cpu = find_cpu_by_server(server_no);
if (!cpu) {
prerror("OPAL: Query invalid CPU 0x%04llx !\n", server_no);
return OPAL_PARAMETER;
}
if (!cpu_is_available(cpu) && cpu->state != cpu_state_os) {
prerror("OPAL: CPU not active in OPAL nor OS !\n");
return OPAL_PARAMETER;
}
switch(cpu->state) {
case cpu_state_os:
*thread_status = OPAL_THREAD_STARTED;
break;
case cpu_state_active:
/* Active in skiboot -> inactive in OS */
*thread_status = OPAL_THREAD_INACTIVE;
break;
default:
*thread_status = OPAL_THREAD_UNAVAILABLE;
}
return OPAL_SUCCESS;
}
opal_call(OPAL_QUERY_CPU_STATUS, opal_query_cpu_status, 2);
static int64_t opal_return_cpu(void)
{
prlog(PR_DEBUG, "OPAL: Returning CPU 0x%04x\n", this_cpu()->pir);
__secondary_cpu_entry();
return OPAL_HARDWARE; /* Should not happen */
}
opal_call(OPAL_RETURN_CPU, opal_return_cpu, 0);
static void cpu_change_hile(void *hilep)
{
bool hile = *(bool *)hilep;
unsigned long hid0;
hid0 = mfspr(SPR_HID0);
if (hile)
hid0 |= SPR_HID0_HILE;
else
hid0 &= ~SPR_HID0_HILE;
prlog(PR_DEBUG, "CPU: [%08x] HID0 set to 0x%016lx\n",
this_cpu()->pir, hid0);
set_hid0(hid0);
this_cpu()->current_hile = hile;
}
static int64_t cpu_change_all_hile(bool hile)
{
struct cpu_thread *cpu;
prlog(PR_INFO, "CPU: Switching HILE on all CPUs to %d\n", hile);
for_each_available_cpu(cpu) {
if (cpu->current_hile == hile)
continue;
if (cpu == this_cpu()) {
cpu_change_hile(&hile);
continue;
}
cpu_wait_job(cpu_queue_job(cpu, cpu_change_hile, &hile), true);
}
return OPAL_SUCCESS;
}
static int64_t opal_reinit_cpus(uint64_t flags)
{
struct cpu_thread *cpu;
int64_t rc = OPAL_SUCCESS;
int i;
lock(&reinit_lock);
prerror("OPAL: Trying a CPU re-init with flags: 0x%llx\n", flags);
for (cpu = first_cpu(); cpu; cpu = next_cpu(cpu)) {
if (cpu == this_cpu())
continue;
if (cpu->state == cpu_state_os) {
/*
* That might be a race with return CPU during kexec
* where we are still, wait a bit and try again
*/
for (i = 0; (i < 1000) &&
(cpu->state == cpu_state_os); i++)
time_wait_ms(1);
if (cpu->state == cpu_state_os) {
prerror("OPAL: CPU 0x%x not in OPAL !\n", cpu->pir);
rc = OPAL_WRONG_STATE;
goto bail;
}
}
}
/*
* Now we need to mark ourselves "active" or we'll be skipped
* by the various "for_each_active_..." calls done by slw_reinit()
*/
this_cpu()->state = cpu_state_active;
/*
* If the flags affect endianness and we are on P8 DD2 or later, then
* use the HID bit. We use the PVR (we could use the EC level in
* the chip but the PVR is more readily available).
*/
if (proc_gen == proc_gen_p8 && PVR_VERS_MAJ(mfspr(SPR_PVR)) >= 2 &&
(flags & (OPAL_REINIT_CPUS_HILE_BE | OPAL_REINIT_CPUS_HILE_LE))) {
bool hile = !!(flags & OPAL_REINIT_CPUS_HILE_LE);
flags &= ~(OPAL_REINIT_CPUS_HILE_BE | OPAL_REINIT_CPUS_HILE_LE);
rc = cpu_change_all_hile(hile);
}
/* Any flags left ? */
if (flags != 0)
rc = slw_reinit(flags);
/* And undo the above */
this_cpu()->state = cpu_state_os;
bail:
unlock(&reinit_lock);
return rc;
}
opal_call(OPAL_REINIT_CPUS, opal_reinit_cpus, 1);
|