forked from torvalds/linux
-
Notifications
You must be signed in to change notification settings - Fork 1
/
pnd2_edac.c
1552 lines (1330 loc) · 41.7 KB
/
pnd2_edac.c
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
/*
* Driver for Pondicherry2 memory controller.
*
* Copyright (c) 2016, Intel Corporation.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
* version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope 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.
*
* [Derived from sb_edac.c]
*
* Translation of system physical addresses to DIMM addresses
* is a two stage process:
*
* First the Pondicherry 2 memory controller handles slice and channel interleaving
* in "sys2pmi()". This is (almost) completley common between platforms.
*
* Then a platform specific dunit (DIMM unit) completes the process to provide DIMM,
* rank, bank, row and column using the appropriate "dunit_ops" functions/parameters.
*/
#include <linux/module.h>
#include <linux/init.h>
#include <linux/pci.h>
#include <linux/pci_ids.h>
#include <linux/slab.h>
#include <linux/delay.h>
#include <linux/edac.h>
#include <linux/mmzone.h>
#include <linux/smp.h>
#include <linux/bitmap.h>
#include <linux/math64.h>
#include <linux/mod_devicetable.h>
#include <asm/cpu_device_id.h>
#include <asm/intel-family.h>
#include <asm/processor.h>
#include <asm/mce.h>
#include "edac_mc.h"
#include "edac_module.h"
#include "pnd2_edac.h"
#define APL_NUM_CHANNELS 4
#define DNV_NUM_CHANNELS 2
#define DNV_MAX_DIMMS 2 /* Max DIMMs per channel */
enum type {
APL,
DNV, /* All requests go to PMI CH0 on each slice (CH1 disabled) */
};
struct dram_addr {
int chan;
int dimm;
int rank;
int bank;
int row;
int col;
};
struct pnd2_pvt {
int dimm_geom[APL_NUM_CHANNELS];
u64 tolm, tohm;
};
/*
* System address space is divided into multiple regions with
* different interleave rules in each. The as0/as1 regions
* have no interleaving at all. The as2 region is interleaved
* between two channels. The mot region is magic and may overlap
* other regions, with its interleave rules taking precedence.
* Addresses not in any of these regions are interleaved across
* all four channels.
*/
static struct region {
u64 base;
u64 limit;
u8 enabled;
} mot, as0, as1, as2;
static struct dunit_ops {
char *name;
enum type type;
int pmiaddr_shift;
int pmiidx_shift;
int channels;
int dimms_per_channel;
int (*rd_reg)(int port, int off, int op, void *data, size_t sz, char *name);
int (*get_registers)(void);
int (*check_ecc)(void);
void (*mk_region)(char *name, struct region *rp, void *asym);
void (*get_dimm_config)(struct mem_ctl_info *mci);
int (*pmi2mem)(struct mem_ctl_info *mci, u64 pmiaddr, u32 pmiidx,
struct dram_addr *daddr, char *msg);
} *ops;
static struct mem_ctl_info *pnd2_mci;
#define PND2_MSG_SIZE 256
/* Debug macros */
#define pnd2_printk(level, fmt, arg...) \
edac_printk(level, "pnd2", fmt, ##arg)
#define pnd2_mc_printk(mci, level, fmt, arg...) \
edac_mc_chipset_printk(mci, level, "pnd2", fmt, ##arg)
#define MOT_CHAN_INTLV_BIT_1SLC_2CH 12
#define MOT_CHAN_INTLV_BIT_2SLC_2CH 13
#define SELECTOR_DISABLED (-1)
#define _4GB (1ul << 32)
#define PMI_ADDRESS_WIDTH 31
#define PND_MAX_PHYS_BIT 39
#define APL_ASYMSHIFT 28
#define DNV_ASYMSHIFT 31
#define CH_HASH_MASK_LSB 6
#define SLICE_HASH_MASK_LSB 6
#define MOT_SLC_INTLV_BIT 12
#define LOG2_PMI_ADDR_GRANULARITY 5
#define MOT_SHIFT 24
#define GET_BITFIELD(v, lo, hi) (((v) & GENMASK_ULL(hi, lo)) >> (lo))
#define U64_LSHIFT(val, s) ((u64)(val) << (s))
#ifdef CONFIG_X86_INTEL_SBI_APL
#include "linux/platform_data/sbi_apl.h"
static int sbi_send(int port, int off, int op, u32 *data)
{
struct sbi_apl_message sbi_arg;
int ret, read = 0;
memset(&sbi_arg, 0, sizeof(sbi_arg));
if (op == 0 || op == 4 || op == 6)
read = 1;
else
sbi_arg.data = *data;
sbi_arg.opcode = op;
sbi_arg.port_address = port;
sbi_arg.register_offset = off;
ret = sbi_apl_commit(&sbi_arg);
if (ret || sbi_arg.status)
edac_dbg(2, "sbi_send status=%d ret=%d data=%x\n",
sbi_arg.status, ret, sbi_arg.data);
if (ret == 0)
ret = sbi_arg.status;
if (ret == 0 && read)
*data = sbi_arg.data;
return ret;
}
#else
static int sbi_send(int port, int off, int op, u32 *data)
{
return -EUNATCH;
}
#endif
static int apl_rd_reg(int port, int off, int op, void *data, size_t sz, char *name)
{
int ret = 0;
edac_dbg(2, "Read %s port=%x off=%x op=%x\n", name, port, off, op);
switch (sz) {
case 8:
ret = sbi_send(port, off + 4, op, (u32 *)(data + 4));
/* fall through */
case 4:
ret |= sbi_send(port, off, op, (u32 *)data);
pnd2_printk(KERN_DEBUG, "%s=%x%08x ret=%d\n", name,
sz == 8 ? *((u32 *)(data + 4)) : 0, *((u32 *)data), ret);
break;
}
return ret;
}
static u64 get_mem_ctrl_hub_base_addr(void)
{
struct b_cr_mchbar_lo_pci lo;
struct b_cr_mchbar_hi_pci hi;
struct pci_dev *pdev;
pdev = pci_get_device(PCI_VENDOR_ID_INTEL, 0x1980, NULL);
if (pdev) {
pci_read_config_dword(pdev, 0x48, (u32 *)&lo);
pci_read_config_dword(pdev, 0x4c, (u32 *)&hi);
pci_dev_put(pdev);
} else {
return 0;
}
if (!lo.enable) {
edac_dbg(2, "MMIO via memory controller hub base address is disabled!\n");
return 0;
}
return U64_LSHIFT(hi.base, 32) | U64_LSHIFT(lo.base, 15);
}
static u64 get_sideband_reg_base_addr(void)
{
struct pci_dev *pdev;
u32 hi, lo;
pdev = pci_get_device(PCI_VENDOR_ID_INTEL, 0x19dd, NULL);
if (pdev) {
pci_read_config_dword(pdev, 0x10, &lo);
pci_read_config_dword(pdev, 0x14, &hi);
pci_dev_put(pdev);
return (U64_LSHIFT(hi, 32) | U64_LSHIFT(lo, 0));
} else {
return 0xfd000000;
}
}
static int dnv_rd_reg(int port, int off, int op, void *data, size_t sz, char *name)
{
struct pci_dev *pdev;
char *base;
u64 addr;
if (op == 4) {
pdev = pci_get_device(PCI_VENDOR_ID_INTEL, 0x1980, NULL);
if (!pdev)
return -ENODEV;
pci_read_config_dword(pdev, off, data);
pci_dev_put(pdev);
} else {
/* MMIO via memory controller hub base address */
if (op == 0 && port == 0x4c) {
addr = get_mem_ctrl_hub_base_addr();
if (!addr)
return -ENODEV;
} else {
/* MMIO via sideband register base address */
addr = get_sideband_reg_base_addr();
if (!addr)
return -ENODEV;
addr += (port << 16);
}
base = ioremap((resource_size_t)addr, 0x10000);
if (!base)
return -ENODEV;
if (sz == 8)
*(u32 *)(data + 4) = *(u32 *)(base + off + 4);
*(u32 *)data = *(u32 *)(base + off);
iounmap(base);
}
edac_dbg(2, "Read %s=%.8x_%.8x\n", name,
(sz == 8) ? *(u32 *)(data + 4) : 0, *(u32 *)data);
return 0;
}
#define RD_REGP(regp, regname, port) \
ops->rd_reg(port, \
regname##_offset, \
regname##_r_opcode, \
regp, sizeof(struct regname), \
#regname)
#define RD_REG(regp, regname) \
ops->rd_reg(regname ## _port, \
regname##_offset, \
regname##_r_opcode, \
regp, sizeof(struct regname), \
#regname)
static u64 top_lm, top_hm;
static bool two_slices;
static bool two_channels; /* Both PMI channels in one slice enabled */
static u8 sym_chan_mask;
static u8 asym_chan_mask;
static u8 chan_mask;
static int slice_selector = -1;
static int chan_selector = -1;
static u64 slice_hash_mask;
static u64 chan_hash_mask;
static void mk_region(char *name, struct region *rp, u64 base, u64 limit)
{
rp->enabled = 1;
rp->base = base;
rp->limit = limit;
edac_dbg(2, "Region:%s [%llx, %llx]\n", name, base, limit);
}
static void mk_region_mask(char *name, struct region *rp, u64 base, u64 mask)
{
if (mask == 0) {
pr_info(FW_BUG "MOT mask cannot be zero\n");
return;
}
if (mask != GENMASK_ULL(PND_MAX_PHYS_BIT, __ffs(mask))) {
pr_info(FW_BUG "MOT mask not power of two\n");
return;
}
if (base & ~mask) {
pr_info(FW_BUG "MOT region base/mask alignment error\n");
return;
}
rp->base = base;
rp->limit = (base | ~mask) & GENMASK_ULL(PND_MAX_PHYS_BIT, 0);
rp->enabled = 1;
edac_dbg(2, "Region:%s [%llx, %llx]\n", name, base, rp->limit);
}
static bool in_region(struct region *rp, u64 addr)
{
if (!rp->enabled)
return false;
return rp->base <= addr && addr <= rp->limit;
}
static int gen_sym_mask(struct b_cr_slice_channel_hash *p)
{
int mask = 0;
if (!p->slice_0_mem_disabled)
mask |= p->sym_slice0_channel_enabled;
if (!p->slice_1_disabled)
mask |= p->sym_slice1_channel_enabled << 2;
if (p->ch_1_disabled || p->enable_pmi_dual_data_mode)
mask &= 0x5;
return mask;
}
static int gen_asym_mask(struct b_cr_slice_channel_hash *p,
struct b_cr_asym_mem_region0_mchbar *as0,
struct b_cr_asym_mem_region1_mchbar *as1,
struct b_cr_asym_2way_mem_region_mchbar *as2way)
{
const int intlv[] = { 0x5, 0xA, 0x3, 0xC };
int mask = 0;
if (as2way->asym_2way_interleave_enable)
mask = intlv[as2way->asym_2way_intlv_mode];
if (as0->slice0_asym_enable)
mask |= (1 << as0->slice0_asym_channel_select);
if (as1->slice1_asym_enable)
mask |= (4 << as1->slice1_asym_channel_select);
if (p->slice_0_mem_disabled)
mask &= 0xc;
if (p->slice_1_disabled)
mask &= 0x3;
if (p->ch_1_disabled || p->enable_pmi_dual_data_mode)
mask &= 0x5;
return mask;
}
static struct b_cr_tolud_pci tolud;
static struct b_cr_touud_lo_pci touud_lo;
static struct b_cr_touud_hi_pci touud_hi;
static struct b_cr_asym_mem_region0_mchbar asym0;
static struct b_cr_asym_mem_region1_mchbar asym1;
static struct b_cr_asym_2way_mem_region_mchbar asym_2way;
static struct b_cr_mot_out_base_mchbar mot_base;
static struct b_cr_mot_out_mask_mchbar mot_mask;
static struct b_cr_slice_channel_hash chash;
/* Apollo Lake dunit */
/*
* Validated on board with just two DIMMs in the [0] and [2] positions
* in this array. Other port number matches documentation, but caution
* advised.
*/
static const int apl_dports[APL_NUM_CHANNELS] = { 0x18, 0x10, 0x11, 0x19 };
static struct d_cr_drp0 drp0[APL_NUM_CHANNELS];
/* Denverton dunit */
static const int dnv_dports[DNV_NUM_CHANNELS] = { 0x10, 0x12 };
static struct d_cr_dsch dsch;
static struct d_cr_ecc_ctrl ecc_ctrl[DNV_NUM_CHANNELS];
static struct d_cr_drp drp[DNV_NUM_CHANNELS];
static struct d_cr_dmap dmap[DNV_NUM_CHANNELS];
static struct d_cr_dmap1 dmap1[DNV_NUM_CHANNELS];
static struct d_cr_dmap2 dmap2[DNV_NUM_CHANNELS];
static struct d_cr_dmap3 dmap3[DNV_NUM_CHANNELS];
static struct d_cr_dmap4 dmap4[DNV_NUM_CHANNELS];
static struct d_cr_dmap5 dmap5[DNV_NUM_CHANNELS];
static void apl_mk_region(char *name, struct region *rp, void *asym)
{
struct b_cr_asym_mem_region0_mchbar *a = asym;
mk_region(name, rp,
U64_LSHIFT(a->slice0_asym_base, APL_ASYMSHIFT),
U64_LSHIFT(a->slice0_asym_limit, APL_ASYMSHIFT) +
GENMASK_ULL(APL_ASYMSHIFT - 1, 0));
}
static void dnv_mk_region(char *name, struct region *rp, void *asym)
{
struct b_cr_asym_mem_region_denverton *a = asym;
mk_region(name, rp,
U64_LSHIFT(a->slice_asym_base, DNV_ASYMSHIFT),
U64_LSHIFT(a->slice_asym_limit, DNV_ASYMSHIFT) +
GENMASK_ULL(DNV_ASYMSHIFT - 1, 0));
}
static int apl_get_registers(void)
{
int ret = -ENODEV;
int i;
if (RD_REG(&asym_2way, b_cr_asym_2way_mem_region_mchbar))
return -ENODEV;
/*
* RD_REGP() will fail for unpopulated or non-existent
* DIMM slots. Return success if we find at least one DIMM.
*/
for (i = 0; i < APL_NUM_CHANNELS; i++)
if (!RD_REGP(&drp0[i], d_cr_drp0, apl_dports[i]))
ret = 0;
return ret;
}
static int dnv_get_registers(void)
{
int i;
if (RD_REG(&dsch, d_cr_dsch))
return -ENODEV;
for (i = 0; i < DNV_NUM_CHANNELS; i++)
if (RD_REGP(&ecc_ctrl[i], d_cr_ecc_ctrl, dnv_dports[i]) ||
RD_REGP(&drp[i], d_cr_drp, dnv_dports[i]) ||
RD_REGP(&dmap[i], d_cr_dmap, dnv_dports[i]) ||
RD_REGP(&dmap1[i], d_cr_dmap1, dnv_dports[i]) ||
RD_REGP(&dmap2[i], d_cr_dmap2, dnv_dports[i]) ||
RD_REGP(&dmap3[i], d_cr_dmap3, dnv_dports[i]) ||
RD_REGP(&dmap4[i], d_cr_dmap4, dnv_dports[i]) ||
RD_REGP(&dmap5[i], d_cr_dmap5, dnv_dports[i]))
return -ENODEV;
return 0;
}
/*
* Read all the h/w config registers once here (they don't
* change at run time. Figure out which address ranges have
* which interleave characteristics.
*/
static int get_registers(void)
{
const int intlv[] = { 10, 11, 12, 12 };
if (RD_REG(&tolud, b_cr_tolud_pci) ||
RD_REG(&touud_lo, b_cr_touud_lo_pci) ||
RD_REG(&touud_hi, b_cr_touud_hi_pci) ||
RD_REG(&asym0, b_cr_asym_mem_region0_mchbar) ||
RD_REG(&asym1, b_cr_asym_mem_region1_mchbar) ||
RD_REG(&mot_base, b_cr_mot_out_base_mchbar) ||
RD_REG(&mot_mask, b_cr_mot_out_mask_mchbar) ||
RD_REG(&chash, b_cr_slice_channel_hash))
return -ENODEV;
if (ops->get_registers())
return -ENODEV;
if (ops->type == DNV) {
/* PMI channel idx (always 0) for asymmetric region */
asym0.slice0_asym_channel_select = 0;
asym1.slice1_asym_channel_select = 0;
/* PMI channel bitmap (always 1) for symmetric region */
chash.sym_slice0_channel_enabled = 0x1;
chash.sym_slice1_channel_enabled = 0x1;
}
if (asym0.slice0_asym_enable)
ops->mk_region("as0", &as0, &asym0);
if (asym1.slice1_asym_enable)
ops->mk_region("as1", &as1, &asym1);
if (asym_2way.asym_2way_interleave_enable) {
mk_region("as2way", &as2,
U64_LSHIFT(asym_2way.asym_2way_base, APL_ASYMSHIFT),
U64_LSHIFT(asym_2way.asym_2way_limit, APL_ASYMSHIFT) +
GENMASK_ULL(APL_ASYMSHIFT - 1, 0));
}
if (mot_base.imr_en) {
mk_region_mask("mot", &mot,
U64_LSHIFT(mot_base.mot_out_base, MOT_SHIFT),
U64_LSHIFT(mot_mask.mot_out_mask, MOT_SHIFT));
}
top_lm = U64_LSHIFT(tolud.tolud, 20);
top_hm = U64_LSHIFT(touud_hi.touud, 32) | U64_LSHIFT(touud_lo.touud, 20);
two_slices = !chash.slice_1_disabled &&
!chash.slice_0_mem_disabled &&
(chash.sym_slice0_channel_enabled != 0) &&
(chash.sym_slice1_channel_enabled != 0);
two_channels = !chash.ch_1_disabled &&
!chash.enable_pmi_dual_data_mode &&
((chash.sym_slice0_channel_enabled == 3) ||
(chash.sym_slice1_channel_enabled == 3));
sym_chan_mask = gen_sym_mask(&chash);
asym_chan_mask = gen_asym_mask(&chash, &asym0, &asym1, &asym_2way);
chan_mask = sym_chan_mask | asym_chan_mask;
if (two_slices && !two_channels) {
if (chash.hvm_mode)
slice_selector = 29;
else
slice_selector = intlv[chash.interleave_mode];
} else if (!two_slices && two_channels) {
if (chash.hvm_mode)
chan_selector = 29;
else
chan_selector = intlv[chash.interleave_mode];
} else if (two_slices && two_channels) {
if (chash.hvm_mode) {
slice_selector = 29;
chan_selector = 30;
} else {
slice_selector = intlv[chash.interleave_mode];
chan_selector = intlv[chash.interleave_mode] + 1;
}
}
if (two_slices) {
if (!chash.hvm_mode)
slice_hash_mask = chash.slice_hash_mask << SLICE_HASH_MASK_LSB;
if (!two_channels)
slice_hash_mask |= BIT_ULL(slice_selector);
}
if (two_channels) {
if (!chash.hvm_mode)
chan_hash_mask = chash.ch_hash_mask << CH_HASH_MASK_LSB;
if (!two_slices)
chan_hash_mask |= BIT_ULL(chan_selector);
}
return 0;
}
/* Get a contiguous memory address (remove the MMIO gap) */
static u64 remove_mmio_gap(u64 sys)
{
return (sys < _4GB) ? sys : sys - (_4GB - top_lm);
}
/* Squeeze out one address bit, shift upper part down to fill gap */
static void remove_addr_bit(u64 *addr, int bitidx)
{
u64 mask;
if (bitidx == -1)
return;
mask = (1ull << bitidx) - 1;
*addr = ((*addr >> 1) & ~mask) | (*addr & mask);
}
/* XOR all the bits from addr specified in mask */
static int hash_by_mask(u64 addr, u64 mask)
{
u64 result = addr & mask;
result = (result >> 32) ^ result;
result = (result >> 16) ^ result;
result = (result >> 8) ^ result;
result = (result >> 4) ^ result;
result = (result >> 2) ^ result;
result = (result >> 1) ^ result;
return (int)result & 1;
}
/*
* First stage decode. Take the system address and figure out which
* second stage will deal with it based on interleave modes.
*/
static int sys2pmi(const u64 addr, u32 *pmiidx, u64 *pmiaddr, char *msg)
{
u64 contig_addr, contig_base, contig_offset, contig_base_adj;
int mot_intlv_bit = two_slices ? MOT_CHAN_INTLV_BIT_2SLC_2CH :
MOT_CHAN_INTLV_BIT_1SLC_2CH;
int slice_intlv_bit_rm = SELECTOR_DISABLED;
int chan_intlv_bit_rm = SELECTOR_DISABLED;
/* Determine if address is in the MOT region. */
bool mot_hit = in_region(&mot, addr);
/* Calculate the number of symmetric regions enabled. */
int sym_channels = hweight8(sym_chan_mask);
/*
* The amount we need to shift the asym base can be determined by the
* number of enabled symmetric channels.
* NOTE: This can only work because symmetric memory is not supposed
* to do a 3-way interleave.
*/
int sym_chan_shift = sym_channels >> 1;
/* Give up if address is out of range, or in MMIO gap */
if (addr >= (1ul << PND_MAX_PHYS_BIT) ||
(addr >= top_lm && addr < _4GB) || addr >= top_hm) {
snprintf(msg, PND2_MSG_SIZE, "Error address 0x%llx is not DRAM", addr);
return -EINVAL;
}
/* Get a contiguous memory address (remove the MMIO gap) */
contig_addr = remove_mmio_gap(addr);
if (in_region(&as0, addr)) {
*pmiidx = asym0.slice0_asym_channel_select;
contig_base = remove_mmio_gap(as0.base);
contig_offset = contig_addr - contig_base;
contig_base_adj = (contig_base >> sym_chan_shift) *
((chash.sym_slice0_channel_enabled >> (*pmiidx & 1)) & 1);
contig_addr = contig_offset + ((sym_channels > 0) ? contig_base_adj : 0ull);
} else if (in_region(&as1, addr)) {
*pmiidx = 2u + asym1.slice1_asym_channel_select;
contig_base = remove_mmio_gap(as1.base);
contig_offset = contig_addr - contig_base;
contig_base_adj = (contig_base >> sym_chan_shift) *
((chash.sym_slice1_channel_enabled >> (*pmiidx & 1)) & 1);
contig_addr = contig_offset + ((sym_channels > 0) ? contig_base_adj : 0ull);
} else if (in_region(&as2, addr) && (asym_2way.asym_2way_intlv_mode == 0x3ul)) {
bool channel1;
mot_intlv_bit = MOT_CHAN_INTLV_BIT_1SLC_2CH;
*pmiidx = (asym_2way.asym_2way_intlv_mode & 1) << 1;
channel1 = mot_hit ? ((bool)((addr >> mot_intlv_bit) & 1)) :
hash_by_mask(contig_addr, chan_hash_mask);
*pmiidx |= (u32)channel1;
contig_base = remove_mmio_gap(as2.base);
chan_intlv_bit_rm = mot_hit ? mot_intlv_bit : chan_selector;
contig_offset = contig_addr - contig_base;
remove_addr_bit(&contig_offset, chan_intlv_bit_rm);
contig_addr = (contig_base >> sym_chan_shift) + contig_offset;
} else {
/* Otherwise we're in normal, boring symmetric mode. */
*pmiidx = 0u;
if (two_slices) {
bool slice1;
if (mot_hit) {
slice_intlv_bit_rm = MOT_SLC_INTLV_BIT;
slice1 = (addr >> MOT_SLC_INTLV_BIT) & 1;
} else {
slice_intlv_bit_rm = slice_selector;
slice1 = hash_by_mask(addr, slice_hash_mask);
}
*pmiidx = (u32)slice1 << 1;
}
if (two_channels) {
bool channel1;
mot_intlv_bit = two_slices ? MOT_CHAN_INTLV_BIT_2SLC_2CH :
MOT_CHAN_INTLV_BIT_1SLC_2CH;
if (mot_hit) {
chan_intlv_bit_rm = mot_intlv_bit;
channel1 = (addr >> mot_intlv_bit) & 1;
} else {
chan_intlv_bit_rm = chan_selector;
channel1 = hash_by_mask(contig_addr, chan_hash_mask);
}
*pmiidx |= (u32)channel1;
}
}
/* Remove the chan_selector bit first */
remove_addr_bit(&contig_addr, chan_intlv_bit_rm);
/* Remove the slice bit (we remove it second because it must be lower */
remove_addr_bit(&contig_addr, slice_intlv_bit_rm);
*pmiaddr = contig_addr;
return 0;
}
/* Translate PMI address to memory (rank, row, bank, column) */
#define C(n) (0x10 | (n)) /* column */
#define B(n) (0x20 | (n)) /* bank */
#define R(n) (0x40 | (n)) /* row */
#define RS (0x80) /* rank */
/* addrdec values */
#define AMAP_1KB 0
#define AMAP_2KB 1
#define AMAP_4KB 2
#define AMAP_RSVD 3
/* dden values */
#define DEN_4Gb 0
#define DEN_8Gb 2
/* dwid values */
#define X8 0
#define X16 1
static struct dimm_geometry {
u8 addrdec;
u8 dden;
u8 dwid;
u8 rowbits, colbits;
u16 bits[PMI_ADDRESS_WIDTH];
} dimms[] = {
{
.addrdec = AMAP_1KB, .dden = DEN_4Gb, .dwid = X16,
.rowbits = 15, .colbits = 10,
.bits = {
C(2), C(3), C(4), C(5), C(6), B(0), B(1), B(2), R(0),
R(1), R(2), R(3), R(4), R(5), R(6), R(7), R(8), R(9),
R(10), C(7), C(8), C(9), R(11), RS, R(12), R(13), R(14),
0, 0, 0, 0
}
},
{
.addrdec = AMAP_1KB, .dden = DEN_4Gb, .dwid = X8,
.rowbits = 16, .colbits = 10,
.bits = {
C(2), C(3), C(4), C(5), C(6), B(0), B(1), B(2), R(0),
R(1), R(2), R(3), R(4), R(5), R(6), R(7), R(8), R(9),
R(10), C(7), C(8), C(9), R(11), RS, R(12), R(13), R(14),
R(15), 0, 0, 0
}
},
{
.addrdec = AMAP_1KB, .dden = DEN_8Gb, .dwid = X16,
.rowbits = 16, .colbits = 10,
.bits = {
C(2), C(3), C(4), C(5), C(6), B(0), B(1), B(2), R(0),
R(1), R(2), R(3), R(4), R(5), R(6), R(7), R(8), R(9),
R(10), C(7), C(8), C(9), R(11), RS, R(12), R(13), R(14),
R(15), 0, 0, 0
}
},
{
.addrdec = AMAP_1KB, .dden = DEN_8Gb, .dwid = X8,
.rowbits = 16, .colbits = 11,
.bits = {
C(2), C(3), C(4), C(5), C(6), B(0), B(1), B(2), R(0),
R(1), R(2), R(3), R(4), R(5), R(6), R(7), R(8), R(9),
R(10), C(7), C(8), C(9), R(11), RS, C(11), R(12), R(13),
R(14), R(15), 0, 0
}
},
{
.addrdec = AMAP_2KB, .dden = DEN_4Gb, .dwid = X16,
.rowbits = 15, .colbits = 10,
.bits = {
C(2), C(3), C(4), C(5), C(6), C(7), B(0), B(1), B(2),
R(0), R(1), R(2), R(3), R(4), R(5), R(6), R(7), R(8),
R(9), R(10), C(8), C(9), R(11), RS, R(12), R(13), R(14),
0, 0, 0, 0
}
},
{
.addrdec = AMAP_2KB, .dden = DEN_4Gb, .dwid = X8,
.rowbits = 16, .colbits = 10,
.bits = {
C(2), C(3), C(4), C(5), C(6), C(7), B(0), B(1), B(2),
R(0), R(1), R(2), R(3), R(4), R(5), R(6), R(7), R(8),
R(9), R(10), C(8), C(9), R(11), RS, R(12), R(13), R(14),
R(15), 0, 0, 0
}
},
{
.addrdec = AMAP_2KB, .dden = DEN_8Gb, .dwid = X16,
.rowbits = 16, .colbits = 10,
.bits = {
C(2), C(3), C(4), C(5), C(6), C(7), B(0), B(1), B(2),
R(0), R(1), R(2), R(3), R(4), R(5), R(6), R(7), R(8),
R(9), R(10), C(8), C(9), R(11), RS, R(12), R(13), R(14),
R(15), 0, 0, 0
}
},
{
.addrdec = AMAP_2KB, .dden = DEN_8Gb, .dwid = X8,
.rowbits = 16, .colbits = 11,
.bits = {
C(2), C(3), C(4), C(5), C(6), C(7), B(0), B(1), B(2),
R(0), R(1), R(2), R(3), R(4), R(5), R(6), R(7), R(8),
R(9), R(10), C(8), C(9), R(11), RS, C(11), R(12), R(13),
R(14), R(15), 0, 0
}
},
{
.addrdec = AMAP_4KB, .dden = DEN_4Gb, .dwid = X16,
.rowbits = 15, .colbits = 10,
.bits = {
C(2), C(3), C(4), C(5), C(6), C(7), C(8), B(0), B(1),
B(2), R(0), R(1), R(2), R(3), R(4), R(5), R(6), R(7),
R(8), R(9), R(10), C(9), R(11), RS, R(12), R(13), R(14),
0, 0, 0, 0
}
},
{
.addrdec = AMAP_4KB, .dden = DEN_4Gb, .dwid = X8,
.rowbits = 16, .colbits = 10,
.bits = {
C(2), C(3), C(4), C(5), C(6), C(7), C(8), B(0), B(1),
B(2), R(0), R(1), R(2), R(3), R(4), R(5), R(6), R(7),
R(8), R(9), R(10), C(9), R(11), RS, R(12), R(13), R(14),
R(15), 0, 0, 0
}
},
{
.addrdec = AMAP_4KB, .dden = DEN_8Gb, .dwid = X16,
.rowbits = 16, .colbits = 10,
.bits = {
C(2), C(3), C(4), C(5), C(6), C(7), C(8), B(0), B(1),
B(2), R(0), R(1), R(2), R(3), R(4), R(5), R(6), R(7),
R(8), R(9), R(10), C(9), R(11), RS, R(12), R(13), R(14),
R(15), 0, 0, 0
}
},
{
.addrdec = AMAP_4KB, .dden = DEN_8Gb, .dwid = X8,
.rowbits = 16, .colbits = 11,
.bits = {
C(2), C(3), C(4), C(5), C(6), C(7), C(8), B(0), B(1),
B(2), R(0), R(1), R(2), R(3), R(4), R(5), R(6), R(7),
R(8), R(9), R(10), C(9), R(11), RS, C(11), R(12), R(13),
R(14), R(15), 0, 0
}
}
};
static int bank_hash(u64 pmiaddr, int idx, int shft)
{
int bhash = 0;
switch (idx) {
case 0:
bhash ^= ((pmiaddr >> (12 + shft)) ^ (pmiaddr >> (9 + shft))) & 1;
break;
case 1:
bhash ^= (((pmiaddr >> (10 + shft)) ^ (pmiaddr >> (8 + shft))) & 1) << 1;
bhash ^= ((pmiaddr >> 22) & 1) << 1;
break;
case 2:
bhash ^= (((pmiaddr >> (13 + shft)) ^ (pmiaddr >> (11 + shft))) & 1) << 2;
break;
}
return bhash;
}
static int rank_hash(u64 pmiaddr)
{
return ((pmiaddr >> 16) ^ (pmiaddr >> 10)) & 1;
}
/* Second stage decode. Compute rank, bank, row & column. */
static int apl_pmi2mem(struct mem_ctl_info *mci, u64 pmiaddr, u32 pmiidx,
struct dram_addr *daddr, char *msg)
{
struct d_cr_drp0 *cr_drp0 = &drp0[pmiidx];
struct pnd2_pvt *pvt = mci->pvt_info;
int g = pvt->dimm_geom[pmiidx];
struct dimm_geometry *d = &dimms[g];
int column = 0, bank = 0, row = 0, rank = 0;
int i, idx, type, skiprs = 0;
for (i = 0; i < PMI_ADDRESS_WIDTH; i++) {
int bit = (pmiaddr >> i) & 1;
if (i + skiprs >= PMI_ADDRESS_WIDTH) {
snprintf(msg, PND2_MSG_SIZE, "Bad dimm_geometry[] table\n");
return -EINVAL;
}
type = d->bits[i + skiprs] & ~0xf;
idx = d->bits[i + skiprs] & 0xf;
/*
* On single rank DIMMs ignore the rank select bit
* and shift remainder of "bits[]" down one place.
*/
if (type == RS && (cr_drp0->rken0 + cr_drp0->rken1) == 1) {
skiprs = 1;
type = d->bits[i + skiprs] & ~0xf;
idx = d->bits[i + skiprs] & 0xf;
}
switch (type) {
case C(0):
column |= (bit << idx);
break;
case B(0):
bank |= (bit << idx);
if (cr_drp0->bahen)
bank ^= bank_hash(pmiaddr, idx, d->addrdec);
break;
case R(0):
row |= (bit << idx);
break;
case RS:
rank = bit;
if (cr_drp0->rsien)
rank ^= rank_hash(pmiaddr);
break;
default:
if (bit) {
snprintf(msg, PND2_MSG_SIZE, "Bad translation\n");
return -EINVAL;
}
goto done;
}
}
done:
daddr->col = column;
daddr->bank = bank;
daddr->row = row;
daddr->rank = rank;
daddr->dimm = 0;
return 0;
}
/* Pluck bit "in" from pmiaddr and return value shifted to bit "out" */
#define dnv_get_bit(pmi, in, out) ((int)(((pmi) >> (in)) & 1u) << (out))
static int dnv_pmi2mem(struct mem_ctl_info *mci, u64 pmiaddr, u32 pmiidx,
struct dram_addr *daddr, char *msg)
{
/* Rank 0 or 1 */
daddr->rank = dnv_get_bit(pmiaddr, dmap[pmiidx].rs0 + 13, 0);
/* Rank 2 or 3 */
daddr->rank |= dnv_get_bit(pmiaddr, dmap[pmiidx].rs1 + 13, 1);
/*
* Normally ranks 0,1 are DIMM0, and 2,3 are DIMM1, but we
* flip them if DIMM1 is larger than DIMM0.
*/
daddr->dimm = (daddr->rank >= 2) ^ drp[pmiidx].dimmflip;
daddr->bank = dnv_get_bit(pmiaddr, dmap[pmiidx].ba0 + 6, 0);
daddr->bank |= dnv_get_bit(pmiaddr, dmap[pmiidx].ba1 + 6, 1);
daddr->bank |= dnv_get_bit(pmiaddr, dmap[pmiidx].bg0 + 6, 2);
if (dsch.ddr4en)
daddr->bank |= dnv_get_bit(pmiaddr, dmap[pmiidx].bg1 + 6, 3);
if (dmap1[pmiidx].bxor) {
if (dsch.ddr4en) {
daddr->bank ^= dnv_get_bit(pmiaddr, dmap3[pmiidx].row6 + 6, 0);
daddr->bank ^= dnv_get_bit(pmiaddr, dmap3[pmiidx].row7 + 6, 1);
if (dsch.chan_width == 0)
/* 64/72 bit dram channel width */
daddr->bank ^= dnv_get_bit(pmiaddr, dmap5[pmiidx].ca3 + 6, 2);
else
/* 32/40 bit dram channel width */
daddr->bank ^= dnv_get_bit(pmiaddr, dmap5[pmiidx].ca4 + 6, 2);
daddr->bank ^= dnv_get_bit(pmiaddr, dmap2[pmiidx].row2 + 6, 3);
} else {
daddr->bank ^= dnv_get_bit(pmiaddr, dmap2[pmiidx].row2 + 6, 0);
daddr->bank ^= dnv_get_bit(pmiaddr, dmap3[pmiidx].row6 + 6, 1);
if (dsch.chan_width == 0)
daddr->bank ^= dnv_get_bit(pmiaddr, dmap5[pmiidx].ca3 + 6, 2);
else
daddr->bank ^= dnv_get_bit(pmiaddr, dmap5[pmiidx].ca4 + 6, 2);
}
}
daddr->row = dnv_get_bit(pmiaddr, dmap2[pmiidx].row0 + 6, 0);
daddr->row |= dnv_get_bit(pmiaddr, dmap2[pmiidx].row1 + 6, 1);
daddr->row |= dnv_get_bit(pmiaddr, dmap2[pmiidx].row2 + 6, 2);
daddr->row |= dnv_get_bit(pmiaddr, dmap2[pmiidx].row3 + 6, 3);
daddr->row |= dnv_get_bit(pmiaddr, dmap2[pmiidx].row4 + 6, 4);
daddr->row |= dnv_get_bit(pmiaddr, dmap2[pmiidx].row5 + 6, 5);