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
|
/* Needed so the AMD K8 runs correctly. */
/* this should be done by Eric
* 2004.11 yhlu add d0 e0 support
* 2004.12 yhlu add dual core support
* 2005.02 yhlu add e0 memory hole support
*/
#include <console/console.h>
#include <cpu/x86/msr.h>
#include <cpu/amd/mtrr.h>
#include <device/device.h>
#include <device/device.h>
#include <device/pci.h>
#include <string.h>
#include <cpu/x86/msr.h>
#include <cpu/x86/pae.h>
#include <pc80/mc146818rtc.h>
#include <cpu/x86/lapic.h>
#include "../../../northbridge/amd/amdk8/amdk8.h"
#include "../../../northbridge/amd/amdk8/cpu_rev.c"
#include <cpu/cpu.h>
#include <cpu/x86/cache.h>
#include <cpu/x86/mtrr.h>
#include <cpu/x86/mem.h>
#include <cpu/amd/dualcore.h>
#include "model_fxx_msr.h"
#define MCI_STATUS 0x401
static inline msr_t rdmsr_amd(unsigned index)
{
msr_t result;
__asm__ __volatile__ (
"rdmsr"
: "=a" (result.lo), "=d" (result.hi)
: "c" (index), "D" (0x9c5a203a)
);
return result;
}
static inline void wrmsr_amd(unsigned index, msr_t msr)
{
__asm__ __volatile__ (
"wrmsr"
: /* No outputs */
: "c" (index), "a" (msr.lo), "d" (msr.hi), "D" (0x9c5a203a)
);
}
#define MTRR_COUNT 8
#define ZERO_CHUNK_KB 0x800UL /* 2M */
#define TOLM_KB 0x400000UL
struct mtrr {
msr_t base;
msr_t mask;
};
struct mtrr_state {
struct mtrr mtrrs[MTRR_COUNT];
msr_t top_mem, top_mem2;
msr_t def_type;
};
static void save_mtrr_state(struct mtrr_state *state)
{
int i;
for(i = 0; i < MTRR_COUNT; i++) {
state->mtrrs[i].base = rdmsr(MTRRphysBase_MSR(i));
state->mtrrs[i].mask = rdmsr(MTRRphysMask_MSR(i));
}
state->top_mem = rdmsr(TOP_MEM);
state->top_mem2 = rdmsr(TOP_MEM2);
state->def_type = rdmsr(MTRRdefType_MSR);
}
static void restore_mtrr_state(struct mtrr_state *state)
{
int i;
disable_cache();
for(i = 0; i < MTRR_COUNT; i++) {
wrmsr(MTRRphysBase_MSR(i), state->mtrrs[i].base);
wrmsr(MTRRphysMask_MSR(i), state->mtrrs[i].mask);
}
wrmsr(TOP_MEM, state->top_mem);
wrmsr(TOP_MEM2, state->top_mem2);
wrmsr(MTRRdefType_MSR, state->def_type);
enable_cache();
}
#if 0
static void print_mtrr_state(struct mtrr_state *state)
{
int i;
for(i = 0; i < MTRR_COUNT; i++) {
printk_debug("var mtrr %d: %08x%08x mask: %08x%08x\n",
i,
state->mtrrs[i].base.hi, state->mtrrs[i].base.lo,
state->mtrrs[i].mask.hi, state->mtrrs[i].mask.lo);
}
printk_debug("top_mem: %08x%08x\n",
state->top_mem.hi, state->top_mem.lo);
printk_debug("top_mem2: %08x%08x\n",
state->top_mem2.hi, state->top_mem2.lo);
printk_debug("def_type: %08x%08x\n",
state->def_type.hi, state->def_type.lo);
}
#endif
static void set_init_ecc_mtrrs(void)
{
msr_t msr;
int i;
disable_cache();
/* First clear all of the msrs to be safe */
for(i = 0; i < MTRR_COUNT; i++) {
msr_t zero;
zero.lo = zero.hi = 0;
wrmsr(MTRRphysBase_MSR(i), zero);
wrmsr(MTRRphysMask_MSR(i), zero);
}
/* Write back cache the first 1MB */
msr.hi = 0x00000000;
msr.lo = 0x00000000 | MTRR_TYPE_WRBACK;
wrmsr(MTRRphysBase_MSR(0), msr);
msr.hi = 0x000000ff;
msr.lo = ~((CONFIG_LB_MEM_TOPK << 10) - 1) | 0x800;
wrmsr(MTRRphysMask_MSR(0), msr);
/* Set the default type to write combining */
msr.hi = 0x00000000;
msr.lo = 0xc00 | MTRR_TYPE_WRCOMB;
wrmsr(MTRRdefType_MSR, msr);
/* Set TOP_MEM to 4G */
msr.hi = 0x00000001;
msr.lo = 0x00000000;
wrmsr(TOP_MEM, msr);
enable_cache();
}
static void init_ecc_memory(unsigned node_id)
{
unsigned long startk, begink, endk;
unsigned long basek;
struct mtrr_state mtrr_state;
device_t f1_dev, f2_dev, f3_dev;
int enable_scrubbing;
uint32_t dcl;
f1_dev = dev_find_slot(0, PCI_DEVFN(0x18 + node_id, 1));
if (!f1_dev) {
die("Cannot find cpu function 1\n");
}
f2_dev = dev_find_slot(0, PCI_DEVFN(0x18 + node_id, 2));
if (!f2_dev) {
die("Cannot find cpu function 2\n");
}
f3_dev = dev_find_slot(0, PCI_DEVFN(0x18 + node_id, 3));
if (!f3_dev) {
die("Cannot find cpu function 3\n");
}
/* See if we scrubbing should be enabled */
enable_scrubbing = 1;
get_option(&enable_scrubbing, "hw_scrubber");
/* Enable cache scrubbing at the lowest possible rate */
if (enable_scrubbing) {
pci_write_config32(f3_dev, SCRUB_CONTROL,
(SCRUB_84ms << 16) | (SCRUB_84ms << 8) | (SCRUB_NONE << 0));
} else {
pci_write_config32(f3_dev, SCRUB_CONTROL,
(SCRUB_NONE << 16) | (SCRUB_NONE << 8) | (SCRUB_NONE << 0));
printk_debug("Scrubbing Disabled\n");
}
/* If ecc support is not enabled don't touch memory */
dcl = pci_read_config32(f2_dev, DRAM_CONFIG_LOW);
if (!(dcl & DCL_DimmEccEn)) {
printk_debug("ECC Disabled\n");
return;
}
startk = (pci_read_config32(f1_dev, 0x40 + (node_id*8)) & 0xffff0000) >> 2;
endk = ((pci_read_config32(f1_dev, 0x44 + (node_id*8)) & 0xffff0000) >> 2) + 0x4000;
/* Don't start too early */
begink = startk;
if (begink < CONFIG_LB_MEM_TOPK) {
begink = CONFIG_LB_MEM_TOPK;
}
printk_debug("Clearing memory %uK - %uK: ", begink, endk);
/* Save the normal state */
save_mtrr_state(&mtrr_state);
/* Switch to the init ecc state */
set_init_ecc_mtrrs();
disable_lapic();
/* Walk through 2M chunks and zero them */
for(basek = begink; basek < endk;
basek = ((basek + ZERO_CHUNK_KB) & ~(ZERO_CHUNK_KB - 1)))
{
unsigned long limitk;
unsigned long size;
void *addr;
/* Report every 64M */
if ((basek % (64*1024)) == 0) {
/* Restore the normal state */
map_2M_page(0);
restore_mtrr_state(&mtrr_state);
enable_lapic();
/* Print a status message */
printk_debug("%c", (basek >= TOLM_KB)?'+':'-');
/* Return to the initialization state */
set_init_ecc_mtrrs();
disable_lapic();
}
limitk = (basek + ZERO_CHUNK_KB) & ~(ZERO_CHUNK_KB - 1);
if (limitk > endk) {
limitk = endk;
}
size = (limitk - basek) << 10;
addr = map_2M_page(basek >> 11);
if (addr == MAPPING_ERROR) {
printk_err("Cannot map page: %x\n", basek >> 11);
continue;
}
/* clear memory 2M (limitk - basek) */
addr = (void *)(((uint32_t)addr) | ((basek & 0x7ff) << 10));
clear_memory(addr, size);
}
/* Restore the normal state */
map_2M_page(0);
restore_mtrr_state(&mtrr_state);
enable_lapic();
/* Set the scrub base address registers */
pci_write_config32(f3_dev, SCRUB_ADDR_LOW, startk << 10);
pci_write_config32(f3_dev, SCRUB_ADDR_HIGH, startk >> 22);
/* Enable the scrubber? */
if (enable_scrubbing) {
/* Enable scrubbing at the lowest possible rate */
pci_write_config32(f3_dev, SCRUB_CONTROL,
(SCRUB_84ms << 16) | (SCRUB_84ms << 8) | (SCRUB_84ms << 0));
}
printk_debug(" done\n");
}
static inline void k8_errata(void)
{
msr_t msr;
if (is_cpu_pre_c0()) {
/* Erratum 63... */
msr = rdmsr(HWCR_MSR);
msr.lo |= (1 << 6);
wrmsr(HWCR_MSR, msr);
/* Erratum 69... */
msr = rdmsr_amd(BU_CFG_MSR);
msr.hi |= (1 << (45 - 32));
wrmsr_amd(BU_CFG_MSR, msr);
/* Erratum 81... */
msr = rdmsr_amd(DC_CFG_MSR);
msr.lo |= (1 << 10);
wrmsr_amd(DC_CFG_MSR, msr);
}
/* I can't touch this msr on early buggy cpus */
if (!is_cpu_pre_b3()) {
/* Erratum 89 ... */
msr = rdmsr(NB_CFG_MSR);
msr.lo |= 1 << 3;
if (!is_cpu_pre_c0() && is_cpu_pre_d0()) {
/* D0 later don't need it */
/* Erratum 86 Disable data masking on C0 and
* later processor revs.
* FIXME this is only needed if ECC is enabled.
*/
msr.hi |= 1 << (36 - 32);
}
wrmsr(NB_CFG_MSR, msr);
}
/* Erratum 97 ... */
if (!is_cpu_pre_c0() && is_cpu_pre_d0()) {
msr = rdmsr_amd(DC_CFG_MSR);
msr.lo |= 1 << 3;
wrmsr_amd(DC_CFG_MSR, msr);
}
/* Erratum 94 ... */
if (is_cpu_pre_d0()) {
msr = rdmsr_amd(IC_CFG_MSR);
msr.lo |= 1 << 11;
wrmsr_amd(IC_CFG_MSR, msr);
}
/* Erratum 91 prefetch miss is handled in the kernel */
/* Erratum 106 ... */
msr = rdmsr_amd(LS_CFG_MSR);
msr.lo |= 1 << 25;
wrmsr_amd(LS_CFG_MSR, msr);
/* Erratum 107 ... */
msr = rdmsr_amd(BU_CFG_MSR);
msr.hi |= 1 << (43 - 32);
wrmsr_amd(BU_CFG_MSR, msr);
if (is_cpu_pre_e0() && !is_cpu_pre_d0()) {
/* Erratum 110 ...*/
msr = rdmsr_amd(CPU_ID_HYPER_EXT_FEATURES);
msr.hi |=1;
wrmsr_amd(CPU_ID_HYPER_EXT_FEATURES, msr);
}
if (!is_cpu_pre_e0()) {
/* Erratum 110 ... */
msr = rdmsr_amd(CPU_ID_EXT_FEATURES_MSR);
msr.hi |=1;
wrmsr_amd(CPU_ID_EXT_FEATURES_MSR, msr);
/* Erratum 113 ... */
msr = rdmsr_amd(BU_CFG_MSR);
msr.hi |= (1 << 16);
wrmsr_amd(BU_CFG_MSR, msr);
}
/* Erratum 122 */
if (!is_cpu_pre_c0()) {
msr = rdmsr(HWCR_MSR);
msr.lo |= 1 << 6;
wrmsr(HWCR_MSR, msr);
}
/* Erratum 123? dual core deadlock? */
/* Erratum 131 */
msr = rdmsr(NB_CFG_MSR);
msr.lo |= 1 << 20;
wrmsr(NB_CFG_MSR, msr);
}
void model_fxx_init(device_t cpu)
{
unsigned long i;
msr_t msr;
struct node_core_id id;
/* Turn on caching if we haven't already */
x86_enable_cache();
amd_setup_mtrrs();
x86_mtrr_check();
disable_cache();
/* zero the machine check error status registers */
msr.lo = 0;
msr.hi = 0;
for(i=0; i<5; i++) {
wrmsr(MCI_STATUS + (i*4),msr);
}
k8_errata();
enable_cache();
/* Enable the local cpu apics */
setup_lapic();
/* Find our node and core */
id = get_node_core_id();
/* Is this a bad location? In particular can another node prefetch
* data from this node before we have initialized it?
*/
if (id.coreid == 0) {
init_ecc_memory(id.nodeid); // only do it for core 0
}
/* Deal with sibling cpus */
amd_sibling_init(cpu, id);
}
static struct device_operations cpu_dev_ops = {
.init = model_fxx_init,
};
static struct cpu_device_id cpu_table[] = {
{ X86_VENDOR_AMD, 0xf50 }, /* B3 */
{ X86_VENDOR_AMD, 0xf51 }, /* SH7-B3 */
{ X86_VENDOR_AMD, 0xf58 }, /* SH7-C0 */
{ X86_VENDOR_AMD, 0xf48 },
{ X86_VENDOR_AMD, 0xf5A }, /* SH7-CG */
{ X86_VENDOR_AMD, 0xf4A },
{ X86_VENDOR_AMD, 0xf7A },
{ X86_VENDOR_AMD, 0xfc0 }, /* DH7-CG */
{ X86_VENDOR_AMD, 0xfe0 },
{ X86_VENDOR_AMD, 0xff0 },
{ X86_VENDOR_AMD, 0xf82 }, /* CH7-CG */
{ X86_VENDOR_AMD, 0xfb2 },
//AMD_D0_SUPPORT
{ X86_VENDOR_AMD, 0x10f50 }, /* SH7-D0 */
{ X86_VENDOR_AMD, 0x10f40 },
{ X86_VENDOR_AMD, 0x10f70 },
{ X86_VENDOR_AMD, 0x10fc0 }, /* DH7-D0 */
{ X86_VENDOR_AMD, 0x10ff0 },
{ X86_VENDOR_AMD, 0x10f80 }, /* CH7-D0 */
{ X86_VENDOR_AMD, 0x10fb0 },
//AMD_E0_SUPPORT
{ X86_VENDOR_AMD, 0x20f50 }, /* SH8-E0*/
{ X86_VENDOR_AMD, 0x20f40 },
{ X86_VENDOR_AMD, 0x20f70 },
{ X86_VENDOR_AMD, 0x20fc0 }, /* DH8-E0 */ /* DH-E3 */
{ X86_VENDOR_AMD, 0x20ff0 },
{ X86_VENDOR_AMD, 0x20f10 }, /* JH8-E1 */
{ X86_VENDOR_AMD, 0x20f30 },
{ X86_VENDOR_AMD, 0x20f51 }, /* SH-E4 */
{ X86_VENDOR_AMD, 0x20f71 },
{ X86_VENDOR_AMD, 0x20f42 }, /* SH-E5 */
{ X86_VENDOR_AMD, 0x20ff2 }, /* DH-E6 */
{ X86_VENDOR_AMD, 0x20fc2 },
{ X86_VENDOR_AMD, 0x20f12 }, /* JH-E6 */
{ X86_VENDOR_AMD, 0x20f32 },
{ 0, 0 },
};
static struct cpu_driver model_fxx __cpu_driver = {
.ops = &cpu_dev_ops,
.id_table = cpu_table,
};
|