summaryrefslogtreecommitdiff
path: root/src/cpu/amd/model_10xxx/init_cpus.c
blob: defbce2571f2c719f0f4a81c2144d115cf24a7cd (plain)
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
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
/*
 * This file is part of the coreboot project.
 *
 * Copyright (C) 2007-2008 Advanced Micro Devices, Inc.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; version 2 of the License.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301 USA
 */

#include "defaults.h"

//it takes the CONFIG_ENABLE_APIC_EXT_ID and CONFIG_APIC_ID_OFFSET and CONFIG_LIFT_BSP_APIC_ID
#ifndef FAM10_SET_FIDVID
	#define FAM10_SET_FIDVID 1
#endif

#ifndef FAM10_SET_FIDVID_CORE0_ONLY
	/* MSR FIDVID_CTL and FIDVID_STATUS are shared by cores,
	   Need to do every AP to set common FID/VID*/
	#define FAM10_SET_FIDVID_CORE0_ONLY 0
#endif

static void print_initcpu8 (const char *strval, u8 val)
{
	printk_debug("%s%02x\n", strval, val);
}

static void print_initcpu8_nocr (const char *strval, u8 val)
{
	printk_debug("%s%02x", strval, val);
}


static void print_initcpu16 (const char *strval, u16 val)
{
	printk_debug("%s%04x\n", strval, val);
}


static void print_initcpu(const char *strval, u32 val)
{
	printk_debug("%s%08x\n", strval, val);
}


void update_microcode(u32 cpu_deviceid);
static void prep_fid_change(void);
static void init_fidvid_stage2(u32 apicid, u32 nodeid);
void cpuSetAMDMSR(void);

#if CONFIG_PCI_IO_CFG_EXT == 1
static void set_EnableCf8ExtCfg(void)
{
	// set the NB_CFG[46]=1;
	msr_t msr;
	msr = rdmsr(NB_CFG_MSR);
	// EnableCf8ExtCfg: We need that to access CONFIG_PCI_IO_CFG_EXT 4K range
	msr.hi |= (1<<(46-32));
	wrmsr(NB_CFG_MSR, msr);
}
#else
static void set_EnableCf8ExtCfg(void) { }
#endif


/*[39:8] */
#define PCI_MMIO_BASE 0xfe000000
/* because we will use gs to store hi, so need to make sure lo can start
   from 0, So PCI_MMIO_BASE & 0x00ffffff should be equal to 0*/

static void set_pci_mmio_conf_reg(void)
{
#if CONFIG_MMCONF_SUPPORT
	msr_t msr;
	msr = rdmsr(0xc0010058);
	msr.lo &= ~(0xfff00000 | (0xf << 2));
	// 256 bus per segment, MMIO reg will be 4G , enable MMIO Config space
	msr.lo |= ((8+CONFIG_PCI_BUS_SEGN_BITS) << 2) | (1 << 0);
	msr.hi &= ~(0x0000ffff);
	msr.hi |= (PCI_MMIO_BASE >> (32-8));
	wrmsr(0xc0010058, msr); // MMIO Config Base Address Reg

	//mtrr for that range?
//	set_var_mtrr_x(7, PCI_MMIO_BASE<<8, PCI_MMIO_BASE>>(32-8), 0x00000000, 0x01, MTRR_TYPE_UNCACHEABLE);

	set_wrap32dis();

	msr.hi = (PCI_MMIO_BASE >> (32-8));
	msr.lo = 0;
	wrmsr(0xc0000101, msr); //GS_Base Reg



#endif
}


typedef void (*process_ap_t)(u32 apicid, void *gp);

//core_range = 0 : all cores
//core range = 1 : core 0 only
//core range = 2 : cores other than core0

static void for_each_ap(u32 bsp_apicid, u32 core_range,
				process_ap_t process_ap, void *gp)
{
	// here assume the OS don't change our apicid
	u32 ap_apicid;

	u32 nodes;
	u32 siblings;
	u32 disable_siblings;
	u32 cores_found;
	u32 nb_cfg_54;
	int i,j;
	u32 ApicIdCoreIdSize;

	/* get_nodes define in ht_wrapper.c */
	nodes = get_nodes();

	disable_siblings = !CONFIG_LOGICAL_CPUS;

#if CONFIG_LOGICAL_CPUS == 1
	if(read_option(CMOS_VSTART_quad_core, CMOS_VLEN_quad_core, 0) != 0) { // 0 mean quad core
		disable_siblings = 1;
	}
#endif

	/* Assume that all node are same stepping, otherwise we can use use
	   nb_cfg_54 from bsp for all nodes */
	nb_cfg_54 = read_nb_cfg_54();

	ApicIdCoreIdSize = (cpuid_ecx(0x80000008) >> 12 & 0xf);
	if(ApicIdCoreIdSize) {
		siblings = ((1 << ApicIdCoreIdSize) - 1);
	} else {
		siblings = 3; //quad core
	}

	for (i = 0; i < nodes; i++) {
		cores_found = get_core_num_in_bsp(i);

		u32 jstart, jend;

		if (core_range == 2) {
			jstart = 1;
		} else {
			jstart = 0;
		}

		if (disable_siblings || (core_range==1)) {
			jend = 0;
		} else {
			jend = cores_found;
		}


		for (j = jstart; j <= jend; j++) {
			ap_apicid = i * (nb_cfg_54 ? (siblings + 1):1) + j * (nb_cfg_54 ? 1:64);

		#if (CONFIG_ENABLE_APIC_EXT_ID == 1) && (CONFIG_APIC_ID_OFFSET > 0)
			#if CONFIG_LIFT_BSP_APIC_ID == 0
			if( (i != 0) || (j != 0)) /* except bsp */
			#endif
				ap_apicid += CONFIG_APIC_ID_OFFSET;
		#endif

			if(ap_apicid == bsp_apicid) continue;

			process_ap(ap_apicid, gp);

		}
	}
}

/* FIXME: Duplicate of what is in lapic.h? */
static int lapic_remote_read(int apicid, int reg, u32 *pvalue)
{
	int timeout;
	u32 status;
	int result;
	lapic_wait_icr_idle();
	lapic_write(LAPIC_ICR2, SET_LAPIC_DEST_FIELD(apicid));
	lapic_write(LAPIC_ICR, LAPIC_DM_REMRD | (reg >> 4));
	timeout = 0;

	do {
		status = lapic_read(LAPIC_ICR) & LAPIC_ICR_BUSY;
	} while (status == LAPIC_ICR_BUSY && timeout++ < 1000);

	timeout = 0;
	do {
		status = lapic_read(LAPIC_ICR) & LAPIC_ICR_RR_MASK;
	} while (status == LAPIC_ICR_RR_INPROG && timeout++ < 1000);

	result = -1;

	if (status == LAPIC_ICR_RR_VALID) {
		*pvalue = lapic_read(LAPIC_RRR);
		result = 0;
	}
	return result;
}


/* Use the LAPIC timer count register to hold each cores init status */
#define LAPIC_MSG_REG 0x380


#if FAM10_SET_FIDVID == 1
static void init_fidvid_ap(u32 bsp_apicid, u32 apicid, u32 nodeid, u32 coreid);
#endif

static inline __attribute__((always_inline)) void print_apicid_nodeid_coreid(u32 apicid, struct node_core_id id, const char *str)
{
		printk_debug("%s --- {	 APICID = %02x NODEID = %02x COREID = %02x} ---\n", str, apicid, id.nodeid, id.coreid);
}


static unsigned wait_cpu_state(u32 apicid, u32 state)
{
	u32 readback = 0;
	u32 timeout = 1;
	int loop = 4000000;
	while (--loop > 0) {
		if (lapic_remote_read(apicid, LAPIC_MSG_REG, &readback) != 0) continue;
		if ((readback & 0x3f) == state) {
			timeout = 0;
			break; //target cpu is in stage started
		}
	}
	if (timeout) {
		if (readback) {
			timeout = readback;
		}
	}

	return timeout;
}


static void wait_ap_started(u32 ap_apicid, void *gp )
{
	u32 timeout;
	timeout = wait_cpu_state(ap_apicid, 0x13); // started
	if(timeout) {
		print_initcpu8_nocr("* AP ", ap_apicid);
		print_initcpu(" didn't start timeout:", timeout);
	}
	else {
		print_initcpu8_nocr("AP started: ", ap_apicid);
	}
}


static void wait_all_other_cores_started(u32 bsp_apicid)
{
	// all aps other than core0
	print_debug("started ap apicid: ");
	for_each_ap(bsp_apicid, 2 , wait_ap_started, (void *)0);
	print_debug("\n");
}


static void allow_all_aps_stop(u32 bsp_apicid)
{
	/* Called by the BSP to indicate AP can stop */

	/* FIXME Do APs use this?
	   Looks like wait_till_sysinfo_in_ram is used instead. */

	// allow aps to stop use 6 bits for state
	lapic_write(LAPIC_MSG_REG, (bsp_apicid << 24) | 0x14);
}

static void enable_apic_ext_id(u32 node)
{
	u32 val;

	val = pci_read_config32(NODE_HT(node), 0x68);
	val |= (HTTC_APIC_EXT_SPUR | HTTC_APIC_EXT_ID | HTTC_APIC_EXT_BRD_CST);
	pci_write_config32(NODE_HT(node), 0x68, val);
}


static void STOP_CAR_AND_CPU()
{
	msr_t msr;

	/* Disable L2 IC to L3 connection (Only for CAR) */
	msr = rdmsr(BU_CFG2);
	msr.lo &= ~(1 << ClLinesToNbDis);
	wrmsr(BU_CFG2, msr);

	disable_cache_as_ram(); // inline
	stop_this_cpu();
}

#if RAMINIT_SYSINFO == 1
static u32 init_cpus(u32 cpu_init_detectedx ,struct sys_info *sysinfo)
#else
static u32 init_cpus(u32 cpu_init_detectedx)
#endif
{
	u32 bsp_apicid = 0;
	u32 apicid;
	struct node_core_id id;

	/*
	 * already set early mtrr in cache_as_ram.inc
	 */

	/* enable access pci conf via mmio*/
	set_pci_mmio_conf_reg();

	/* that is from initial apicid, we need nodeid and coreid
	   later */
	id = get_node_core_id_x();

	/* NB_CFG MSR is shared between cores, so we need make sure
	  core0 is done at first --- use wait_all_core0_started  */
	if(id.coreid == 0) {
		set_apicid_cpuid_lo(); /* only set it on core0 */
		set_EnableCf8ExtCfg(); /* only set it on core0 */
		#if (CONFIG_ENABLE_APIC_EXT_ID == 1)
		enable_apic_ext_id(id.nodeid);
		#endif
	}

	enable_lapic();


#if (CONFIG_ENABLE_APIC_EXT_ID == 1) && (CONFIG_APIC_ID_OFFSET > 0)
	u32 initial_apicid = get_initial_apicid();

	#if CONFIG_LIFT_BSP_APIC_ID == 0
	if( initial_apicid != 0 ) // other than bsp
	#endif
	{
		/* use initial apic id to lift it */
		u32 dword = lapic_read(LAPIC_ID);
		dword &= ~(0xff << 24);
		dword |= (((initial_apicid + CONFIG_APIC_ID_OFFSET) & 0xff) << 24);

		lapic_write(LAPIC_ID, dword);
	}

	#if CONFIG_LIFT_BSP_APIC_ID == 1
	bsp_apicid += CONFIG_APIC_ID_OFFSET;
	#endif

#endif

	/* get the apicid, it may be lifted already */
	apicid = lapicid();

	// show our apicid, nodeid, and coreid
	if( id.coreid==0 ) {
		if (id.nodeid!=0) //all core0 except bsp
			print_apicid_nodeid_coreid(apicid, id, " core0: ");
	}
	else { //all other cores
		print_apicid_nodeid_coreid(apicid, id, " corex: ");
	}


	if (cpu_init_detectedx) {
		print_apicid_nodeid_coreid(apicid, id, "\n\n\nINIT detected from ");
		print_debug("\nIssuing SOFT_RESET...\n");
		soft_reset();
	}

	if(id.coreid == 0) {
		if(!(warm_reset_detect(id.nodeid))) //FIXME: INIT is checked above but check for more resets?
			distinguish_cpu_resets(id.nodeid); // Also indicates we are started
	}

	// Mark the core as started.
	lapic_write(LAPIC_MSG_REG, (apicid << 24) | 0x13);


	if(apicid != bsp_apicid) {
		/* Setup each AP's cores MSRs.
		 * This happens after HTinit.
		 * The BSP runs this code in it's own path.
		 */
		update_microcode(cpuid_eax(1));
		cpuSetAMDMSR();


#if FAM10_SET_FIDVID == 1
	#if (CONFIG_LOGICAL_CPUS == 1)  && (FAM10_SET_FIDVID_CORE0_ONLY == 1)
		// Run on all AP for proper FID/VID setup.
		if(id.coreid == 0 ) // only need set fid for core0
	#endif
		{
		// check warm(bios) reset to call stage2 otherwise do stage1
			if (warm_reset_detect(id.nodeid)) {
				printk_debug("init_fidvid_stage2 apicid: %02x\n", apicid);
				init_fidvid_stage2(apicid, id.nodeid);
			} else {
				printk_debug("init_fidvid_ap(stage1) apicid: %02x\n", apicid);
				init_fidvid_ap(bsp_apicid, apicid, id.nodeid, id.coreid);
			}
		}
#endif

		/* AP is ready, Wait for the BSP to get memory configured */
		/* FIXME: many cores spinning on node0 pci register seems to be bad.
		 * Why do we need to wait? These APs are just going to go sit in a hlt.
		 */
		//wait_till_sysinfo_in_ram();

		set_init_ram_access();

		STOP_CAR_AND_CPU();
		printk_debug("\nAP %02x should be halted but you are reading this....\n", apicid);
	}

	return bsp_apicid;
}


static u32 is_core0_started(u32 nodeid)
{
	u32 htic;
	device_t device;
	device = NODE_PCI(nodeid, 0);
	htic = pci_read_config32(device, HT_INIT_CONTROL);
	htic &= HTIC_ColdR_Detect;
	return htic;
}


static void wait_all_core0_started(void)
{
	/* When core0 is started, it will distingush_cpu_resets
	  . So wait for that to finish */
	u32 i;
	u32 nodes = get_nodes();

	printk_debug("Wait all core0s started \n");
	for(i=1;i<nodes;i++) { // skip bsp, because it is running on bsp
		while(!is_core0_started(i)) {}
		print_initcpu8("  Core0 started on node: ", i);
	}
	printk_debug("Wait all core0s started done\n");
}
#if CONFIG_MAX_PHYSICAL_CPUS > 1
/**
 * void start_node(u32 node)
 *
 *  start the core0 in node, so it can generate HT packet to feature code.
 *
 * This function starts the AP nodes core0s. wait_all_core0_started() in
 * cache_as_ram_auto.c waits for all the AP to be finished before continuing
 * system init.
 */
static void start_node(u8 node)
{
	u32 val;

	/* Enable routing table */
	printk_debug("Start node %02x", node);

#if CONFIG_NORTHBRIDGE_AMD_AMDFAM10
	/* For FAM10 support, we need to set Dram base/limit for the new node */
	pci_write_config32(NODE_MP(node), 0x44, 0);
	pci_write_config32(NODE_MP(node), 0x40, 3);
#endif

	/* Allow APs to make requests (ROM fetch) */
	val=pci_read_config32(NODE_HT(node), 0x6c);
	val &= ~(1 << 1);
	pci_write_config32(NODE_HT(node), 0x6c, val);

	printk_debug(" done.\n");
}


/**
 * static void setup_remote_node(u32 node)
 *
 * Copy the BSP Adress Map to each AP.
 */
static void setup_remote_node(u8 node)
{
	/* There registers can be used with F1x114_x Address Map at the
	   same time, So must set them even 32 node */
	static const u16 pci_reg[] = {
		/* DRAM Base/Limits Registers */
		0x44, 0x4c, 0x54, 0x5c, 0x64, 0x6c, 0x74, 0x7c,
		0x40, 0x48, 0x50, 0x58, 0x60, 0x68, 0x70, 0x78,
		0x144, 0x14c, 0x154, 0x15c, 0x164, 0x16c, 0x174, 0x17c,
		0x140, 0x148, 0x150, 0x158, 0x160, 0x168, 0x170, 0x178,
		/* MMIO Base/Limits Registers */
		0x84, 0x8c, 0x94, 0x9c, 0xa4, 0xac, 0xb4, 0xbc,
		0x80, 0x88, 0x90, 0x98, 0xa0, 0xa8, 0xb0, 0xb8,
		/* IO Base/Limits Registers */
		0xc4, 0xcc, 0xd4, 0xdc,
		0xc0, 0xc8, 0xd0, 0xd8,
		/* Configuration Map Registers */
		0xe0, 0xe4, 0xe8, 0xec,
	};
	u16 i;

	printk_debug("setup_remote_node: %02x", node);

	/* copy the default resource map from node 0 */
	for(i = 0; i < ARRAY_SIZE(pci_reg); i++) {
		u32 value;
		u16 reg;
		reg = pci_reg[i];
		value = pci_read_config32(NODE_MP(0), reg);
		pci_write_config32(NODE_MP(node), reg, value);

	}
	printk_debug(" done\n");
}
#endif	/* CONFIG_MAX_PHYSICAL_CPUS > 1 */

void AMD_Errata281(u8 node, u32 revision, u32 platform)
{
	/* Workaround for Transaction Scheduling Conflict in
	 * Northbridge Cross Bar.  Implement XCS Token adjustment
	 * for ganged links.  Also, perform fix up for the mixed
	 * revision case.
	 */

	u32 reg, val;
	u8 i;
	u8 mixed = 0;
	u8 nodes = get_nodes();

	if (platform & AMD_PTYPE_SVR) {
		/* For each node we need to check for a "broken" node */
		if (!(revision & (AMD_DR_B0 | AMD_DR_B1))) {
			for (i = 0; i < nodes; i++) {
				if (mctGetLogicalCPUID(i) & (AMD_DR_B0 | AMD_DR_B1)) {
					mixed = 1;
					break;
				}
			}
		}

		if ((revision & (AMD_DR_B0 | AMD_DR_B1)) || mixed) {

			/* F0X68[22:21] DsNpReqLmt0 = 01b */
			val = pci_read_config32(NODE_PCI(node, 0), 0x68);
			val &= ~0x00600000;
			val |= 0x00200000;
			pci_write_config32(NODE_PCI(node, 0), 0x68, val);

			/* F3X6C */
			val = pci_read_config32(NODE_PCI(node, 3), 0x6C);
			val &= ~0x700780F7;
			val |= 0x00010094;
			pci_write_config32(NODE_PCI(node, 3), 0x6C, val);

			/* F3X7C */
			val = pci_read_config32(NODE_PCI(node, 3), 0x7C);
			val &= ~0x707FFF1F;
			val |= 0x00144514;
			pci_write_config32(NODE_PCI(node, 3), 0x7C, val);

			/* F3X144[3:0] RspTok = 0001b */
			val = pci_read_config32(NODE_PCI(node, 3), 0x144);
			val &= ~0x0000000F;
			val |= 0x00000001;
			pci_write_config32(NODE_PCI(node, 3), 0x144, val);

			for (i = 0; i < 3; i++) {
				reg = 0x148 + (i * 4);
				val = pci_read_config32(NODE_PCI(node, 3), reg);
				val &= ~0x000000FF;
				val |= 0x000000DB;
				pci_write_config32(NODE_PCI(node, 3), reg, val);
			}
		}
	}
}


void AMD_Errata298(void)
{
	/* Workaround for L2 Eviction May Occur during operation to
	 * set Accessed or dirty bit.
	 */

	msr_t msr;
	u8 i;
	u8 affectedRev = 0;
	u8 nodes = get_nodes();

	/* For each core we need to check for a "broken" node */
	for (i = 0; i < nodes; i++) {
		if (mctGetLogicalCPUID(i) & (AMD_DR_B0 | AMD_DR_B1 | AMD_DR_B2)) {
			affectedRev = 1;
			break;
		}
	}

	if (affectedRev) {
		msr = rdmsr(HWCR);
		msr.lo |= 0x08;		/* Set TlbCacheDis bit[3] */
		wrmsr(HWCR, msr);

		msr = rdmsr(BU_CFG);
		msr.lo |= 0x02;		/* Set TlbForceMemTypeUc bit[1] */
		wrmsr(BU_CFG, msr);

		msr = rdmsr(OSVW_ID_Length);
		msr.lo |= 0x01;		/* OS Visible Workaround - MSR */
		wrmsr(OSVW_ID_Length, msr);

		msr = rdmsr(OSVW_Status);
		msr.lo |= 0x01;		/* OS Visible Workaround - MSR */
		wrmsr(OSVW_Status, msr);
	}

	if (!affectedRev && (mctGetLogicalCPUID(0xFF) & AMD_DR_B3)) {
		msr = rdmsr(OSVW_ID_Length);
		msr.lo |= 0x01;		/* OS Visible Workaround - MSR */
		wrmsr(OSVW_ID_Length, msr);

	}
}


u32 get_platform_type(void)
{
	u32 ret = 0;

	switch(SYSTEM_TYPE) {
	case 1:
		ret |= AMD_PTYPE_DSK;
		break;
	case 2:
		ret |= AMD_PTYPE_MOB;
		break;
	case 0:
		ret |= AMD_PTYPE_SVR;
		break;
	default:
		break;
	}

	/* FIXME: add UMA support. */

	/* All Fam10 are multi core */
	ret |= AMD_PTYPE_MC;

	return ret;
}


void AMD_SetupPSIVID_d (u32 platform_type, u8 node)
{
	u32 dword;
	int i;
	msr_t msr;

	if (platform_type & (AMD_PTYPE_MOB | AMD_PTYPE_DSK)) {

	/* The following code sets the PSIVID to the lowest support P state
	 * assuming that the VID for the lowest power state is below
	 * the VDD voltage regulator threshold. (This also assumes that there
	 * is a Pstate lower than P0)
	 */

		for( i = 4; i >= 0; i--) {
			msr = rdmsr(PS_REG_BASE + i);
			/*  Pstate valid? */
			if (msr.hi & PS_EN_MASK) {
				dword = pci_read_config32(NODE_PCI(i,3), 0xA0);
				dword &= ~0x7F;
				dword |= (msr.lo >> 9) & 0x7F;
				pci_write_config32(NODE_PCI(i,3), 0xA0, dword);
				break;
			}
		}
	}
}


/**
 * AMD_CpuFindCapability - Traverse PCI capability list to find host HT links.
 *  HT Phy operations are not valid on links that aren't present, so this
 *  prevents invalid accesses.
 *
 * Returns the offset of the link register.
 */
BOOL AMD_CpuFindCapability (u8 node, u8 cap_count, u8 *offset)
{
	u32 val;

	/* get start of CPU HT Host Capabilities */
	val = pci_read_config32(NODE_PCI(node, 0), 0x34);
	val &= 0xFF;

	cap_count++;

	/* Traverse through the capabilities. */
	do {
		val = pci_read_config32(NODE_PCI(node, 0), val);
		/* Is the capability block a HyperTransport capability block? */
		if ((val & 0xFF) == 0x08) {
			/* Is the HT capability block an HT Host Capability? */
			if ((val & 0xE0000000) == (1 << 29))
				cap_count--;
		}
		if (cap_count)
			val = (val >> 8) & 0xFF;
	} while (cap_count && val);

	*offset = (u8) val;

	/* If requested capability found val != 0 */
	if (!cap_count)
		return TRUE;
	else
		return FALSE;
}


/**
 * AMD_checkLinkType - Compare desired link characteristics using a logical
 *     link type mask.
 *
 * Returns the link characteristic mask.
 */
u32 AMD_checkLinkType (u8 node, u8 link, u8 regoff)
{
	u32 val;
	u32 linktype = 0;

	/* Check connect, init and coherency */
	val = pci_read_config32(NODE_PCI(node, 0), regoff + 0x18);
	val &= 0x1F;

	if (val == 3)
		linktype |= HTPHY_LINKTYPE_COHERENT;

	if (val == 7)
		linktype |= HTPHY_LINKTYPE_NONCOHERENT;

	if (linktype) {
		/* Check gen3 */
		val = pci_read_config32(NODE_PCI(node, 0), regoff + 0x08);

		if (((val >> 8) & 0x0F) > 6)
			linktype |= HTPHY_LINKTYPE_HT3;
		else
			linktype |= HTPHY_LINKTYPE_HT1;


		/* Check ganged */
		val = pci_read_config32(NODE_PCI(node, 0), (link << 2) + 0x170);

		if ( val & 1)
			linktype |= HTPHY_LINKTYPE_GANGED;
		else
			linktype |= HTPHY_LINKTYPE_UNGANGED;
	}
	return linktype;
}


/**
 * AMD_SetHtPhyRegister - Use the HT link's HT Phy portal registers to update
 *   a phy setting for that link.
 */
void AMD_SetHtPhyRegister (u8 node, u8 link, u8 entry)
{
	u32 phyReg;
	u32 phyBase;
	u32 val;

	/* Determine this link's portal */
	if (link > 3)
		link -= 4;

	phyBase = ((u32)link << 3) | 0x180;


	/* Get the portal control register's initial value
	 * and update it to access the desired phy register
	 */
	phyReg = pci_read_config32(NODE_PCI(node, 4), phyBase);

	if (fam10_htphy_default[entry].htreg > 0x1FF) {
		phyReg &= ~HTPHY_DIRECT_OFFSET_MASK;
		phyReg |= HTPHY_DIRECT_MAP;
	} else {
		phyReg &= ~HTPHY_OFFSET_MASK;
	}

	/* Now get the current phy register data
	 * LinkPhyDone = 0, LinkPhyWrite = 0 is a read
	 */
	phyReg |= fam10_htphy_default[entry].htreg;
	pci_write_config32(NODE_PCI(node, 4), phyBase, phyReg);

	do {
		val = pci_read_config32(NODE_PCI(node, 4), phyBase);
	} while (!(val & HTPHY_IS_COMPLETE_MASK));

	/* Now we have the phy register data, apply the change */
	val = pci_read_config32(NODE_PCI(node, 4), phyBase + 4);
	val &= ~fam10_htphy_default[entry].mask;
	val |= fam10_htphy_default[entry].data;
	pci_write_config32(NODE_PCI(node, 4), phyBase + 4, val);

	/* write it through the portal to the phy
	 * LinkPhyDone = 0, LinkPhyWrite = 1 is a write
	 */
	phyReg |= HTPHY_WRITE_CMD;
	pci_write_config32(NODE_PCI(node, 4), phyBase, phyReg);

	do {
		val = pci_read_config32(NODE_PCI(node, 4), phyBase);
	} while (!(val & HTPHY_IS_COMPLETE_MASK));
}


void cpuSetAMDMSR(void)
{
	/* This routine loads the CPU with default settings in fam10_msr_default
	 * table . It must be run after Cache-As-RAM has been enabled, and
	 * Hypertransport initialization has taken place.  Also note
	 * that it is run on the current processor only, and only for the current
	 * processor core.
	 */
	msr_t msr;
	u8 i;
	u32 revision, platform;

	printk_debug("cpuSetAMDMSR ");

	revision = mctGetLogicalCPUID(0xFF);
	platform = get_platform_type();

	for(i = 0; i < ARRAY_SIZE(fam10_msr_default); i++) {
		if ((fam10_msr_default[i].revision & revision) &&
		    (fam10_msr_default[i].platform & platform)) {
			msr = rdmsr(fam10_msr_default[i].msr);
			msr.hi &= ~fam10_msr_default[i].mask_hi;
			msr.hi |= fam10_msr_default[i].data_hi;
			msr.lo &= ~fam10_msr_default[i].mask_lo;
			msr.lo |= fam10_msr_default[i].data_lo;
			wrmsr(fam10_msr_default[i].msr, msr);
		}
	}
	AMD_Errata298();

	printk_debug(" done\n");
}


void cpuSetAMDPCI(u8 node)
{
	/* This routine loads the CPU with default settings in fam10_pci_default
	 * table . It must be run after Cache-As-RAM has been enabled, and
	 * Hypertransport initialization has taken place.  Also note
	 * that it is run for the first core on each node
	 */
	u8 i, j;
	u32 revision, platform;
	u32 val;
	u8 offset;

	printk_debug("cpuSetAMDPCI %02d", node);


	revision = mctGetLogicalCPUID(node);
	platform = get_platform_type();

	AMD_SetupPSIVID_d(platform, node);	/* Set PSIVID offset which is not table driven */

	for(i = 0; i < ARRAY_SIZE(fam10_pci_default); i++) {
		if ((fam10_pci_default[i].revision & revision) &&
		    (fam10_pci_default[i].platform & platform)) {
			val = pci_read_config32(NODE_PCI(node,
				fam10_pci_default[i].function),
				fam10_pci_default[i].offset);
			val &= ~fam10_pci_default[i].mask;
			val |= fam10_pci_default[i].data;
			pci_write_config32(NODE_PCI(node,
				fam10_pci_default[i].function),
				fam10_pci_default[i].offset, val);
		}
	}

	for(i = 0; i < ARRAY_SIZE(fam10_htphy_default); i++) {
		if ((fam10_htphy_default[i].revision & revision) &&
		    (fam10_htphy_default[i].platform & platform)) {
			/* HT Phy settings either apply to both sublinks or have
			 * separate registers for sublink zero and one, so there
			 * will be two table entries. So, here we only loop
			 cd t	* through the sublink zeros in function zero.
			 */
			for (j = 0; j < 4; j++) {
				if (AMD_CpuFindCapability(node, j, &offset)) {
					if (AMD_checkLinkType(node, j, offset)
					    & fam10_htphy_default[i].linktype) {
						AMD_SetHtPhyRegister(node, j, i);
					}
				} else {
					/* No more capabilities,
					 * link not present
					 */
					break;
				}
			}
		}
	}

	/* FIXME: add UMA support and programXbarToSriReg(); */

	AMD_Errata281(node, revision, platform);

	/* FIXME: if the dct phy doesn't init correct it needs to reset.
	if (revision & (AMD_DR_B2 | AMD_DR_B3))
		dctPhyDiag(); */

	printk_debug(" done\n");
}


void cpuInitializeMCA(void)
{
	/* Clears Machine Check Architecture (MCA) registers, which power on
	 * containing unknown data, on currently running processor.
	 * This routine should only be executed on initial power on (cold boot),
	 * not across a warm reset because valid data is present at that time.
	 */

	msr_t msr;
	u32 reg;
	u8 i;

	if (cpuid_edx(1) & 0x4080) { /* MCE and MCA (edx[7] and edx[14]) */
		msr = rdmsr(MCG_CAP);
		if (msr.lo & MCG_CTL_P){	/* MCG_CTL_P bit is set? */
			msr.lo &= 0xFF;
			msr.lo--;
			msr.lo <<= 2;		/* multiply the count by 4 */
			reg = MC0_STA + msr.lo;
			msr.lo = msr.hi = 0;
			for (i=0; i < 4; i++) {
				wrmsr (reg, msr);
				reg -=4;	/* Touch status regs for each bank */
			}
		}
	}
}


/**
 * finalize_node_setup()
 *
 * Do any additional post HT init
 *
 */
void finalize_node_setup(struct sys_info *sysinfo)
{
	u8 i;
	u8 nodes = get_nodes();
	u32 reg;

#if RAMINIT_SYSINFO == 1
	/* read Node0 F0_0x64 bit [8:10] to find out SbLink # */
	reg = pci_read_config32(NODE_HT(0), 0x64);
	sysinfo->sblk = (reg>>8) & 7;
	sysinfo->sbbusn = 0;
	sysinfo->nodes = nodes;
	sysinfo->sbdn = get_sbdn(sysinfo->sbbusn);
#endif


	for (i = 0; i < nodes; i++) {
		cpuSetAMDPCI(i);
	}

#if FAM10_SET_FIDVID == 1
	// Prep each node for FID/VID setup.
	prep_fid_change();
#endif

#if CONFIG_MAX_PHYSICAL_CPUS > 1
	/* Skip the BSP, start at node 1 */
	for(i=1; i<nodes; i++) {
		setup_remote_node(i);
		start_node(i);
	}
#endif
}