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
|
// SPDX-License-Identifier: GPL-2.0+
/*
* Copyright 2014-2015 Freescale Semiconductor, Inc.
*/
#include <common.h>
#include <cpu_func.h>
#include <image.h>
#include <asm/cache.h>
#include <asm/io.h>
#include <asm/system.h>
#include <asm/arch/mp.h>
#include <asm/arch/soc.h>
#include <linux/delay.h>
#include "cpu.h"
#include <asm/arch-fsl-layerscape/soc.h>
DECLARE_GLOBAL_DATA_PTR;
void *get_spin_tbl_addr(void)
{
return &__spin_table;
}
phys_addr_t determine_mp_bootpg(void)
{
return (phys_addr_t)&secondary_boot_code;
}
void update_os_arch_secondary_cores(uint8_t os_arch)
{
u64 *table = get_spin_tbl_addr();
int i;
for (i = 1; i < CONFIG_MAX_CPUS; i++) {
if (os_arch == IH_ARCH_DEFAULT)
table[i * WORDS_PER_SPIN_TABLE_ENTRY +
SPIN_TABLE_ELEM_ARCH_COMP_IDX] = OS_ARCH_SAME;
else
table[i * WORDS_PER_SPIN_TABLE_ENTRY +
SPIN_TABLE_ELEM_ARCH_COMP_IDX] = OS_ARCH_DIFF;
}
}
#ifdef CONFIG_FSL_LSCH3
void wake_secondary_core_n(int cluster, int core, int cluster_cores)
{
struct ccsr_gur __iomem *gur = (void *)(CONFIG_SYS_FSL_GUTS_ADDR);
struct ccsr_reset __iomem *rst = (void *)(CONFIG_SYS_FSL_RST_ADDR);
u32 mpidr = 0;
mpidr = ((cluster << 8) | core);
/*
* mpidr_el1 register value of core which needs to be released
* is written to scratchrw[6] register
*/
gur_out32(&gur->scratchrw[6], mpidr);
asm volatile("dsb st" : : : "memory");
rst->brrl |= 1 << ((cluster * cluster_cores) + core);
asm volatile("dsb st" : : : "memory");
/*
* scratchrw[6] register value is polled
* when the value becomes zero, this means that this core is up
* and running, next core can be released now
*/
while (gur_in32(&gur->scratchrw[6]) != 0)
;
}
#endif
int fsl_layerscape_wake_seconday_cores(void)
{
struct ccsr_gur __iomem *gur = (void *)(CONFIG_SYS_FSL_GUTS_ADDR);
#ifdef CONFIG_FSL_LSCH3
struct ccsr_reset __iomem *rst = (void *)(CONFIG_SYS_FSL_RST_ADDR);
u32 svr, ver, cluster, type;
int j = 0, cluster_cores = 0;
#elif defined(CONFIG_FSL_LSCH2)
struct ccsr_scfg __iomem *scfg = (void *)(CONFIG_SYS_FSL_SCFG_ADDR);
#endif
u32 cores, cpu_up_mask = 1;
int i, timeout = 10;
u64 *table = get_spin_tbl_addr();
#ifdef COUNTER_FREQUENCY_REAL
/* update for secondary cores */
__real_cntfrq = COUNTER_FREQUENCY_REAL;
flush_dcache_range((unsigned long)&__real_cntfrq,
(unsigned long)&__real_cntfrq + 8);
#endif
cores = cpu_mask();
/* Clear spin table so that secondary processors
* observe the correct value after waking up from wfe.
*/
memset(table, 0, CONFIG_MAX_CPUS*SPIN_TABLE_ELEM_SIZE);
flush_dcache_range((unsigned long)table,
(unsigned long)table +
(CONFIG_MAX_CPUS*SPIN_TABLE_ELEM_SIZE));
printf("Waking secondary cores to start from %lx\n", gd->relocaddr);
#ifdef CONFIG_FSL_LSCH3
gur_out32(&gur->bootlocptrh, (u32)(gd->relocaddr >> 32));
gur_out32(&gur->bootlocptrl, (u32)gd->relocaddr);
svr = gur_in32(&gur->svr);
ver = SVR_SOC_VER(svr);
if (ver == SVR_LS2080A || ver == SVR_LS2085A) {
gur_out32(&gur->scratchrw[6], 1);
asm volatile("dsb st" : : : "memory");
rst->brrl = cores;
asm volatile("dsb st" : : : "memory");
} else {
/*
* Release the cores out of reset one-at-a-time to avoid
* power spikes
*/
i = 0;
cluster = in_le32(&gur->tp_cluster[i].lower);
for (j = 0; j < TP_INIT_PER_CLUSTER; j++) {
type = initiator_type(cluster, j);
if (type &&
TP_ITYP_TYPE(type) == TP_ITYP_TYPE_ARM)
cluster_cores++;
}
do {
cluster = in_le32(&gur->tp_cluster[i].lower);
for (j = 0; j < TP_INIT_PER_CLUSTER; j++) {
type = initiator_type(cluster, j);
if (type &&
TP_ITYP_TYPE(type) == TP_ITYP_TYPE_ARM)
wake_secondary_core_n(i, j,
cluster_cores);
}
i++;
} while ((cluster & TP_CLUSTER_EOC) != TP_CLUSTER_EOC);
}
#elif defined(CONFIG_FSL_LSCH2)
scfg_out32(&scfg->scratchrw[0], (u32)(gd->relocaddr >> 32));
scfg_out32(&scfg->scratchrw[1], (u32)gd->relocaddr);
asm volatile("dsb st" : : : "memory");
gur_out32(&gur->brrl, cores);
asm volatile("dsb st" : : : "memory");
/* Bootup online cores */
scfg_out32(&scfg->corebcr, cores);
#endif
/* This is needed as a precautionary measure.
* If some code before this has accidentally released the secondary
* cores then the pre-bootloader code will trap them in a "wfe" unless
* the scratchrw[6] is set. In this case we need a sev here to get these
* cores moving again.
*/
asm volatile("sev");
while (timeout--) {
flush_dcache_range((unsigned long)table, (unsigned long)table +
CONFIG_MAX_CPUS * 64);
for (i = 1; i < CONFIG_MAX_CPUS; i++) {
if (table[i * WORDS_PER_SPIN_TABLE_ENTRY +
SPIN_TABLE_ELEM_STATUS_IDX])
cpu_up_mask |= 1 << i;
}
if (hweight32(cpu_up_mask) == hweight32(cores))
break;
udelay(10);
}
if (timeout <= 0) {
printf("Not all cores (0x%x) are up (0x%x)\n",
cores, cpu_up_mask);
return 1;
}
printf("All (%d) cores are up.\n", hweight32(cores));
return 0;
}
int is_core_valid(unsigned int core)
{
return !!((1 << core) & cpu_mask());
}
static int is_pos_valid(unsigned int pos)
{
return !!((1 << pos) & cpu_pos_mask());
}
int is_core_online(u64 cpu_id)
{
u64 *table;
int pos = id_to_core(cpu_id);
table = (u64 *)get_spin_tbl_addr() + pos * WORDS_PER_SPIN_TABLE_ENTRY;
return table[SPIN_TABLE_ELEM_STATUS_IDX] == 1;
}
int cpu_reset(u32 nr)
{
puts("Feature is not implemented.\n");
return 0;
}
int cpu_disable(u32 nr)
{
puts("Feature is not implemented.\n");
return 0;
}
static int core_to_pos(int nr)
{
u32 cores = cpu_pos_mask();
int i, count = 0;
if (nr == 0) {
return 0;
} else if (nr >= hweight32(cores)) {
puts("Not a valid core number.\n");
return -1;
}
for (i = 1; i < 32; i++) {
if (is_pos_valid(i)) {
count++;
if (count == nr)
break;
}
}
if (count != nr)
return -1;
return i;
}
int cpu_status(u32 nr)
{
u64 *table;
int pos;
if (nr == 0) {
table = (u64 *)get_spin_tbl_addr();
printf("table base @ 0x%p\n", table);
} else {
pos = core_to_pos(nr);
if (pos < 0)
return -1;
table = (u64 *)get_spin_tbl_addr() + pos *
WORDS_PER_SPIN_TABLE_ENTRY;
printf("table @ 0x%p\n", table);
printf(" addr - 0x%016llx\n",
table[SPIN_TABLE_ELEM_ENTRY_ADDR_IDX]);
printf(" status - 0x%016llx\n",
table[SPIN_TABLE_ELEM_STATUS_IDX]);
printf(" lpid - 0x%016llx\n",
table[SPIN_TABLE_ELEM_LPID_IDX]);
}
return 0;
}
int cpu_release(u32 nr, int argc, char *const argv[])
{
u64 boot_addr;
u64 *table = (u64 *)get_spin_tbl_addr();
int pos;
pos = core_to_pos(nr);
if (pos <= 0)
return -1;
table += pos * WORDS_PER_SPIN_TABLE_ENTRY;
boot_addr = simple_strtoull(argv[0], NULL, 16);
table[SPIN_TABLE_ELEM_ENTRY_ADDR_IDX] = boot_addr;
flush_dcache_range((unsigned long)table,
(unsigned long)table + SPIN_TABLE_ELEM_SIZE);
asm volatile("dsb st");
smp_kick_all_cpus(); /* only those with entry addr set will run */
/*
* When the first release command runs, all cores are set to go. Those
* without a valid entry address will be trapped by "wfe". "sev" kicks
* them off to check the address again. When set, they continue to run.
*/
asm volatile("sev");
return 0;
}
|