summaryrefslogtreecommitdiffstats
path: root/hw/phb4.c
blob: 823ed431cf8eb702779afdbae14a80bbe7ac73f7 (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
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
5322
5323
5324
5325
5326
5327
5328
5329
5330
5331
5332
5333
5334
5335
5336
5337
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
5349
5350
5351
5352
5353
5354
5355
5356
5357
5358
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
5377
5378
5379
5380
5381
5382
5383
5384
5385
5386
5387
5388
5389
5390
5391
5392
5393
5394
5395
5396
5397
5398
5399
5400
5401
5402
5403
5404
5405
5406
5407
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
/* Copyright 2013-2016 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.
 */
/*
 * PHB4 support
 *
 */

/*
 *
 * FIXME:
 *   More stuff for EEH support:
 *      - PBCQ error reporting interrupt
 *	- I2C-based power management (replacing SHPC)
 *	- Directly detect fenced PHB through one dedicated HW reg
 */

/*
 * This is a simplified view of the PHB4 reset and link training steps
 *
 * Step 1:
 * - Check for hotplug status:
 *  o PHB_PCIE_HOTPLUG_STATUS bit PHB_PCIE_HPSTAT_PRESENCE
 *  o If not set -> Bail out (Slot is empty)
 *
 * Step 2:
 * - Do complete PHB reset:
 *   o PHB/ETU reset procedure
 *
 * Step 3:
 * - Drive PERST active (skip if already asserted. ie. after cold reboot)
 * - Wait 250ms (for cards to reset)
 *   o powervm have used 250ms for a long time without any problems
 *
 * Step 4:
 * - Drive PERST inactive
 *
 * Step 5:
 * - Look for inband presence:
 *   o From PERST we have two stages to get inband presence detected
 *     1) Devices must enter Detect state within 20 ms of the end of
 *          Fundamental Reset
 *     2) Receiver detect pulse are every 12ms
 *      - Hence minimum wait time 20 + 12 = 32ms
 *   o Unfortunatey, we've seen cards take 440ms
 *   o Hence we are conservative and poll here for 1000ms (> 440ms)
 * - If no inband presence after 100ms -> Bail out (Slot is broken)
 *   o PHB_PCIE_DLP_TRAIN_CTL bit PHB_PCIE_DLP_INBAND_PRESENCE
 *
 * Step 6:
 * - Look for link training done:
 *   o PHB_PCIE_DLP_TRAIN_CTL bit PHB_PCIE_DLP_TL_LINKACT
 * - If not set after 2000ms, Retry (3 times) -> Goto Step 2
 *   o phy lockup could link training failure, hence going back to a
 *     complete PHB reset on retry
 *   o not expect to happen very often
 *
 * Step 7:
 * - Wait for 1 sec (before touching device config space):
 * -  From PCIe spec:
 *     Root Complex and/or system software must allow at least 1.0 s after
 *     a Conventional Reset of a device, before it may determine that a
 *     device which fails to return a Successful Completion status for a
 *     valid Configuration Request is a broken device.
 *
 * Step 8:
 * - Sanity check for fence and link still up:
 *   o If fenced or link down, Retry (3 times) -> Goto Step 2
 *   o This is not nessary but takes no time and can be useful
 *   o Once we leave here, much harder to recover from errors
 *
 * Step 9:
 * - Check for optimised link for directly attached devices:
 *   o Wait for CRS (so we can read device config space)
 *   o Check chip and device are in whitelist. if not, Goto Step 10
 *   o If trained link speed is degraded, retry ->  Goto Step 2
 *   o If trained link width is degraded, retry -> Goto Step 2
 *   o If still degraded after 3 retries. Give up, Goto Step 10.
 *
 * Step 10:
 *  - PHB good, start probing config space.
 *    o core/pci.c: pci_reset_phb() -> pci_scan_phb()
 */


#undef NO_ASB
#undef LOG_CFG

#include <skiboot.h>
#include <io.h>
#include <timebase.h>
#include <pci.h>
#include <pci-cfg.h>
#include <pci-slot.h>
#include <vpd.h>
#include <interrupts.h>
#include <opal.h>
#include <cpu.h>
#include <device.h>
#include <ccan/str/str.h>
#include <ccan/array_size/array_size.h>
#include <xscom.h>
#include <affinity.h>
#include <phb4.h>
#include <phb4-regs.h>
#include <phb4-capp.h>
#include <capp.h>
#include <fsp.h>
#include <chip.h>
#include <chiptod.h>
#include <xive.h>
#include <xscom-p9-regs.h>
#include <phys-map.h>
#include <nvram.h>

/* Enable this to disable error interrupts for debug purposes */
#define DISABLE_ERR_INTS

static void phb4_init_hw(struct phb4 *p, bool first_init);

#define PHBDBG(p, fmt, a...)	prlog(PR_DEBUG, "PHB#%04x[%d:%d]: " fmt, \
				      (p)->phb.opal_id, (p)->chip_id, \
				      (p)->index,  ## a)
#define PHBINF(p, fmt, a...)	prlog(PR_INFO, "PHB#%04x[%d:%d]: " fmt, \
				      (p)->phb.opal_id, (p)->chip_id, \
				      (p)->index,  ## a)
#define PHBERR(p, fmt, a...)	prlog(PR_ERR, "PHB#%04x[%d:%d]: " fmt, \
				      (p)->phb.opal_id, (p)->chip_id, \
				      (p)->index,  ## a)
#ifdef LOG_CFG
#define PHBLOGCFG(p, fmt, a...)	PHBDBG(p, fmt, ## a)
#else
#define PHBLOGCFG(p, fmt, a...) do {} while (0)
#endif

#define PHB4_CAN_STORE_EOI(p) \
	(XIVE_STORE_EOI_ENABLED && ((p)->rev >= PHB4_REV_NIMBUS_DD20))

static bool verbose_eeh;
static bool pci_tracing;
static bool pci_eeh_mmio;
static bool pci_retry_all;

/* Note: The "ASB" name is historical, practically this means access via
 * the XSCOM backdoor
 */
static inline uint64_t phb4_read_reg_asb(struct phb4 *p, uint32_t offset)
{
#ifdef NO_ASB
	return in_be64(p->regs + offset);
#else
	int64_t rc;
	uint64_t addr, val;

	/* Address register: must use 4 bytes for built-in config space.
	 *
	 * This path isn't usable for outbound configuration space
	 */
	if (((offset & 0xfffffffc) == PHB_CONFIG_DATA) && (offset & 3)) {
		PHBERR(p, "XSCOM unaligned access to CONFIG_DATA unsupported\n");
		return -1ull;
	}
	addr = XETU_HV_IND_ADDR_VALID | offset;
	if ((offset >= 0x1000 && offset < 0x1800) || (offset == PHB_CONFIG_DATA))
		addr |= XETU_HV_IND_ADDR_4B;
 	rc = xscom_write(p->chip_id, p->etu_xscom + XETU_HV_IND_ADDRESS, addr);
	if (rc != 0) {
		PHBERR(p, "XSCOM error addressing register 0x%x\n", offset);
		return -1ull;
	}
 	rc = xscom_read(p->chip_id, p->etu_xscom + XETU_HV_IND_DATA, &val);
	if (rc != 0) {
		PHBERR(p, "XSCOM error reading register 0x%x\n", offset);
		return -1ull;
	}
	return val;
#endif
}

static inline void phb4_write_reg_asb(struct phb4 *p,
				      uint32_t offset, uint64_t val)
{
#ifdef NO_ASB
	out_be64(p->regs + offset, val);
#else
	int64_t rc;
	uint64_t addr;

	/* Address register: must use 4 bytes for built-in config space.
	 *
	 * This path isn't usable for outbound configuration space
	 */
	if (((offset & 0xfffffffc) == PHB_CONFIG_DATA) && (offset & 3)) {
		PHBERR(p, "XSCOM access to CONFIG_DATA unsupported\n");
		return;
	}
	addr = XETU_HV_IND_ADDR_VALID | offset;
	if ((offset >= 0x1000 && offset < 0x1800) || (offset == PHB_CONFIG_DATA))
		addr |= XETU_HV_IND_ADDR_4B;
 	rc = xscom_write(p->chip_id, p->etu_xscom + XETU_HV_IND_ADDRESS, addr);
	if (rc != 0) {
		PHBERR(p, "XSCOM error addressing register 0x%x\n", offset);
		return;
	}
 	rc = xscom_write(p->chip_id, p->etu_xscom + XETU_HV_IND_DATA, val);
	if (rc != 0) {
		PHBERR(p, "XSCOM error writing register 0x%x\n", offset);
		return;
	}
#endif
}

static uint64_t phb4_read_reg(struct phb4 *p, uint32_t offset)
{
	if (p->flags & PHB4_CFG_USE_ASB)
		return phb4_read_reg_asb(p, offset);
	else
		return in_be64(p->regs + offset);
}

static void phb4_write_reg(struct phb4 *p, uint32_t offset, uint64_t val)
{
	if (p->flags & PHB4_CFG_USE_ASB)
		phb4_write_reg_asb(p, offset, val);
	else
		return out_be64(p->regs + offset, val);
}

/* Helper to select an IODA table entry */
static inline void phb4_ioda_sel(struct phb4 *p, uint32_t table,
				 uint32_t addr, bool autoinc)
{
	phb4_write_reg(p, PHB_IODA_ADDR,
		       (autoinc ? PHB_IODA_AD_AUTOINC : 0)	|
		       SETFIELD(PHB_IODA_AD_TSEL, 0ul, table)	|
		       SETFIELD(PHB_IODA_AD_TADR, 0ul, addr));
}

/*
 * Configuration space access
 *
 * The PHB lock is assumed to be already held
 */
static int64_t phb4_pcicfg_check(struct phb4 *p, uint32_t bdfn,
				 uint32_t offset, uint32_t size,
				 uint16_t *pe)
{
	uint32_t sm = size - 1;

	if (offset > 0xfff || bdfn > 0xffff)
		return OPAL_PARAMETER;
	if (offset & sm)
		return OPAL_PARAMETER;

	/* The root bus only has a device at 0 and we get into an
	 * error state if we try to probe beyond that, so let's
	 * avoid that and just return an error to Linux
	 */
	if ((bdfn >> 8) == 0 && (bdfn & 0xff))
		return OPAL_HARDWARE;

	/* Check PHB state */
	if (p->broken)
		return OPAL_HARDWARE;

	/* Fetch the PE# from cache */
	*pe = p->rte_cache[bdfn];

	return OPAL_SUCCESS;
}

static int64_t phb4_rc_read(struct phb4 *p, uint32_t offset, uint8_t sz,
			    void *data, bool use_asb)
{
	uint32_t reg = offset & ~3;
	uint32_t oval;

	/* Some registers are handled locally */
	switch (reg) {
		/* Bridge base/limit registers are cached here as HW
		 * doesn't implement them (it hard codes values that
		 * will confuse a proper PCI implementation).
		 */
	case PCI_CFG_MEM_BASE:		/* Includes PCI_CFG_MEM_LIMIT */
		oval = p->rc_cache[(reg - 0x20) >> 2] & 0xfff0fff0;
		break;
	case PCI_CFG_PREF_MEM_BASE:	/* Includes PCI_CFG_PREF_MEM_LIMIT */
		oval = p->rc_cache[(reg - 0x20) >> 2] & 0xfff0fff0;
		oval |= 0x00010001;
		break;
	case PCI_CFG_IO_BASE_U16:	/* Includes PCI_CFG_IO_LIMIT_U16 */
		oval = 0;
		break;
	case PCI_CFG_PREF_MEM_BASE_U32:
	case PCI_CFG_PREF_MEM_LIMIT_U32:
		oval = p->rc_cache[(reg - 0x20) >> 2];
		break;
	default:
		oval = 0xffffffff; /* default if offset too big */
		if (reg < PHB_RC_CONFIG_SIZE) {
			if (use_asb)
				oval = bswap_32(phb4_read_reg_asb(p, PHB_RC_CONFIG_BASE
								  + reg));
			else
				oval = in_le32(p->regs + PHB_RC_CONFIG_BASE + reg);
		}
	}
	switch (sz) {
	case 1:
		offset &= 3;
		*((uint8_t *)data) = (oval >> (offset << 3)) & 0xff;
		PHBLOGCFG(p, "000 CFG08 Rd %02x=%02x\n",
			  offset, *((uint8_t *)data));
		break;
	case 2:
		offset &= 2;
		*((uint16_t *)data) = (oval >> (offset << 3)) & 0xffff;
		PHBLOGCFG(p, "000 CFG16 Rd %02x=%04x\n",
			  offset, *((uint16_t *)data));
		break;
	case 4:
		*((uint32_t *)data) = oval;
		PHBLOGCFG(p, "000 CFG32 Rd %02x=%08x\n",
			  offset, *((uint32_t *)data));
		break;
	default:
		assert(false);
	}
	return OPAL_SUCCESS;
}

static int64_t phb4_rc_write(struct phb4 *p, uint32_t offset, uint8_t sz,
			     uint32_t val, bool use_asb)
{
	uint32_t reg = offset & ~3;
	uint32_t old, mask, shift, oldold;
	int64_t rc;

	if (reg > PHB_RC_CONFIG_SIZE)
		return OPAL_SUCCESS;

	/* If size isn't 4-bytes, do a RMW cycle */
	if (sz < 4) {
		rc = phb4_rc_read(p, reg, 4, &old, use_asb);
		if (rc != OPAL_SUCCESS)
			return rc;

		/*
		 * Since we have to Read-Modify-Write here, we need to filter
		 * out registers that have write-1-to-clear bits to prevent
		 * clearing stuff we shouldn't be.  So for any register this
		 * applies to, mask out those bits.
		 */
		oldold = old;
		switch(reg) {
		case 0x1C: /* Secondary status */
			old &= 0x00ffffff; /* mask out 24-31 */
			break;
		case 0x50: /* EC - Device status */
			old &= 0xfff0ffff; /* mask out 16-19 */
			break;
		case 0x58: /* EC - Link status */
			old &= 0x3fffffff; /* mask out 30-31 */
			break;
		case 0x78: /* EC - Link status 2 */
			old &= 0xf000ffff; /* mask out 16-27 */
			break;
		/* These registers *only* have write-1-to-clear bits */
		case 0x104: /* AER - Uncorr. error status */
		case 0x110: /* AER - Corr. error status */
		case 0x130: /* AER - Root error status */
		case 0x180: /* P16 - status */
		case 0x184: /* P16 - LDPM status */
		case 0x188: /* P16 - FRDPM status */
		case 0x18C: /* P16 - SRDPM status */
			old &= 0x00000000;
			break;
		}

		if (old != oldold) {
			PHBLOGCFG(p, "Rewrote %x to %x for reg %x for W1C\n",
				  oldold, old, reg);
		}

		if (sz == 1) {
			shift = (offset & 3) << 3;
			mask = 0xff << shift;
			val = (old & ~mask) | ((val & 0xff) << shift);
		} else {
			shift = (offset & 2) << 3;
			mask = 0xffff << shift;
			val = (old & ~mask) | ((val & 0xffff) << shift);
		}
	}

	/* Some registers are handled locally */
	switch (reg) {
		/* See comment in phb4_rc_read() */
	case PCI_CFG_MEM_BASE:		/* Includes PCI_CFG_MEM_LIMIT */
	case PCI_CFG_PREF_MEM_BASE:	/* Includes PCI_CFG_PREF_MEM_LIMIT */
	case PCI_CFG_PREF_MEM_BASE_U32:
	case PCI_CFG_PREF_MEM_LIMIT_U32:
		p->rc_cache[(reg - 0x20) >> 2] = val;
		break;
	case PCI_CFG_IO_BASE_U16:	/* Includes PCI_CFG_IO_LIMIT_U16 */
		break;
	default:
		/* Workaround PHB config space enable */
		if ((p->rev == PHB4_REV_NIMBUS_DD10) && (reg == PCI_CFG_CMD))
			val |= PCI_CFG_CMD_MEM_EN | PCI_CFG_CMD_BUS_MASTER_EN;
		PHBLOGCFG(p, "000 CFG%02d Wr %02x=%08x\n", 8 * sz, reg, val);
		if (use_asb)
			phb4_write_reg_asb(p, PHB_RC_CONFIG_BASE + reg, val);
		else
			out_le32(p->regs + PHB_RC_CONFIG_BASE + reg, val);
	}
	return OPAL_SUCCESS;
}

static int64_t phb4_pcicfg_read(struct phb4 *p, uint32_t bdfn,
				uint32_t offset, uint32_t size,
				void *data)
{
	uint64_t addr, val64;
	int64_t rc;
	uint16_t pe;
	bool use_asb = false;

	rc = phb4_pcicfg_check(p, bdfn, offset, size, &pe);
	if (rc)
		return rc;

	if (p->flags & PHB4_AIB_FENCED) {
		if (!(p->flags & PHB4_CFG_USE_ASB))
			return OPAL_HARDWARE;
		if (bdfn != 0)
			return OPAL_HARDWARE;
		use_asb = true;
	} else if ((p->flags & PHB4_CFG_BLOCKED) && bdfn != 0) {
		return OPAL_HARDWARE;
	}

	/* Handle per-device filters */
	rc = pci_handle_cfg_filters(&p->phb, bdfn, offset, size,
				    (uint32_t *)data, false);
	if (rc != OPAL_PARTIAL)
		return rc;

	/* Handle root complex MMIO based config space */
	if (bdfn == 0)
		return phb4_rc_read(p, offset, size, data, use_asb);

	addr = PHB_CA_ENABLE;
	addr = SETFIELD(PHB_CA_BDFN, addr, bdfn);
	addr = SETFIELD(PHB_CA_REG, addr, offset & ~3u);
	addr = SETFIELD(PHB_CA_PE, addr, pe);
	if (use_asb) {
		phb4_write_reg_asb(p, PHB_CONFIG_ADDRESS, addr);
		sync();
		val64 = bswap_64(phb4_read_reg_asb(p, PHB_CONFIG_DATA));
		switch(size) {
		case 1:
			*((uint8_t *)data) = val64 >> (8 * (offset & 3));
			break;
		case 2:
			*((uint16_t *)data) = val64 >> (8 * (offset & 2));
			break;
		case 4:
			*((uint32_t *)data) = val64;
			break;
		default:
			return OPAL_PARAMETER;
		}
	} else {
		out_be64(p->regs + PHB_CONFIG_ADDRESS, addr);
		switch(size) {
		case 1:
			*((uint8_t *)data) =
				in_8(p->regs + PHB_CONFIG_DATA + (offset & 3));
			PHBLOGCFG(p, "%03x CFG08 Rd %02x=%02x\n",
				  bdfn, offset, *((uint8_t *)data));
			break;
		case 2:
			*((uint16_t *)data) =
				in_le16(p->regs + PHB_CONFIG_DATA + (offset & 2));
			PHBLOGCFG(p, "%03x CFG16 Rd %02x=%04x\n",
				  bdfn, offset, *((uint16_t *)data));
			break;
		case 4:
			*((uint32_t *)data) = in_le32(p->regs + PHB_CONFIG_DATA);
			PHBLOGCFG(p, "%03x CFG32 Rd %02x=%08x\n",
				  bdfn, offset, *((uint32_t *)data));
			break;
		default:
			return OPAL_PARAMETER;
		}
	}
	return OPAL_SUCCESS;
}


#define PHB4_PCI_CFG_READ(size, type)					\
static int64_t phb4_pcicfg_read##size(struct phb *phb, uint32_t bdfn,	\
				      uint32_t offset, type *data)	\
{									\
	struct phb4 *p = phb_to_phb4(phb);				\
									\
	/* Initialize data in case of error */				\
	*data = (type)0xffffffff;					\
	return phb4_pcicfg_read(p, bdfn, offset, sizeof(type), data);	\
}

static int64_t phb4_pcicfg_write(struct phb4 *p, uint32_t bdfn,
				 uint32_t offset, uint32_t size,
				 uint32_t data)
{
	uint64_t addr;
	int64_t rc;
	uint16_t pe;
	bool use_asb = false;

	rc = phb4_pcicfg_check(p, bdfn, offset, size, &pe);
	if (rc)
		return rc;

	if (p->flags & PHB4_AIB_FENCED) {
		if (!(p->flags & PHB4_CFG_USE_ASB))
			return OPAL_HARDWARE;
		if (bdfn != 0)
			return OPAL_HARDWARE;
		use_asb = true;
	} else if ((p->flags & PHB4_CFG_BLOCKED) && bdfn != 0) {
		return OPAL_HARDWARE;
	}

	/* Handle per-device filters */
	rc = pci_handle_cfg_filters(&p->phb, bdfn, offset, size,
				    (uint32_t *)&data, true);
	if (rc != OPAL_PARTIAL)
		return rc;

	/* Handle root complex MMIO based config space */
	if (bdfn == 0)
		return phb4_rc_write(p, offset, size, data, use_asb);

	addr = PHB_CA_ENABLE;
	addr = SETFIELD(PHB_CA_BDFN, addr, bdfn);
	addr = SETFIELD(PHB_CA_REG, addr, offset & ~3u);
	addr = SETFIELD(PHB_CA_PE, addr, pe);
	if (use_asb) {
		/* We don't support ASB config space writes */
		return OPAL_UNSUPPORTED;
	} else {
		out_be64(p->regs + PHB_CONFIG_ADDRESS, addr);
		switch(size) {
		case 1:
			out_8(p->regs + PHB_CONFIG_DATA + (offset & 3), data);
			break;
		case 2:
			out_le16(p->regs + PHB_CONFIG_DATA + (offset & 2), data);
			break;
		case 4:
			out_le32(p->regs + PHB_CONFIG_DATA, data);
			break;
		default:
			return OPAL_PARAMETER;
		}
	}
	PHBLOGCFG(p, "%03x CFG%d Wr %02x=%08x\n", bdfn, 8 * size, offset, data);
	return OPAL_SUCCESS;
}

#define PHB4_PCI_CFG_WRITE(size, type)					\
static int64_t phb4_pcicfg_write##size(struct phb *phb, uint32_t bdfn,	\
				       uint32_t offset, type data)	\
{									\
	struct phb4 *p = phb_to_phb4(phb);				\
									\
	return phb4_pcicfg_write(p, bdfn, offset, sizeof(type), data);	\
}

PHB4_PCI_CFG_READ(8, u8)
PHB4_PCI_CFG_READ(16, u16)
PHB4_PCI_CFG_READ(32, u32)
PHB4_PCI_CFG_WRITE(8, u8)
PHB4_PCI_CFG_WRITE(16, u16)
PHB4_PCI_CFG_WRITE(32, u32)

static uint8_t phb4_choose_bus(struct phb *phb __unused,
			       struct pci_device *bridge __unused,
			       uint8_t candidate, uint8_t *max_bus __unused,
			       bool *use_max)
{
	/* Use standard bus number selection */
	*use_max = false;
	return candidate;
}

static int64_t phb4_get_reserved_pe_number(struct phb *phb)
{
	struct phb4 *p = phb_to_phb4(phb);

	return PHB4_RESERVED_PE_NUM(p);
}


static void phb4_root_port_init(struct phb *phb, struct pci_device *dev,
				int ecap, int aercap)
{
	uint16_t bdfn = dev->bdfn;
	uint16_t val16;
	uint32_t val32;

	/*
	 * Use the PHB's callback so that UTL events will be masked or
	 * unmasked when the link is down or up.
	 */
	if (dev->slot && dev->slot->ops.prepare_link_change &&
	    phb->slot && phb->slot->ops.prepare_link_change)
		dev->slot->ops.prepare_link_change =
			phb->slot->ops.prepare_link_change;

	// FIXME: check recommended init values for phb4

	/* Enable SERR and parity checking */
	pci_cfg_read16(phb, bdfn, PCI_CFG_CMD, &val16);
	val16 |= (PCI_CFG_CMD_SERR_EN | PCI_CFG_CMD_PERR_RESP |
		  PCI_CFG_CMD_MEM_EN);
	pci_cfg_write16(phb, bdfn, PCI_CFG_CMD, val16);

	/* Enable reporting various errors */
	if (!ecap) return;
	pci_cfg_read16(phb, bdfn, ecap + PCICAP_EXP_DEVCTL, &val16);
	val16 |= (PCICAP_EXP_DEVCTL_CE_REPORT |
		  PCICAP_EXP_DEVCTL_NFE_REPORT |
		  PCICAP_EXP_DEVCTL_FE_REPORT |
		  PCICAP_EXP_DEVCTL_UR_REPORT);
	pci_cfg_write16(phb, bdfn, ecap + PCICAP_EXP_DEVCTL, val16);

	if (!aercap) return;

	/* Mask various unrecoverable errors */
	pci_cfg_read32(phb, bdfn, aercap + PCIECAP_AER_UE_MASK, &val32);
	val32 |= (PCIECAP_AER_UE_MASK_POISON_TLP |
		  PCIECAP_AER_UE_MASK_COMPL_TIMEOUT |
		  PCIECAP_AER_UE_MASK_COMPL_ABORT |
		  PCIECAP_AER_UE_MASK_ECRC);
	pci_cfg_write32(phb, bdfn, aercap + PCIECAP_AER_UE_MASK, val32);

	/* Report various unrecoverable errors as fatal errors */
	pci_cfg_read32(phb, bdfn, aercap + PCIECAP_AER_UE_SEVERITY, &val32);
	val32 |= (PCIECAP_AER_UE_SEVERITY_DLLP |
		  PCIECAP_AER_UE_SEVERITY_SURPRISE_DOWN |
		  PCIECAP_AER_UE_SEVERITY_FLOW_CTL_PROT |
		  PCIECAP_AER_UE_SEVERITY_UNEXP_COMPL |
		  PCIECAP_AER_UE_SEVERITY_RECV_OVFLOW |
		  PCIECAP_AER_UE_SEVERITY_MALFORMED_TLP);
	pci_cfg_write32(phb, bdfn, aercap + PCIECAP_AER_UE_SEVERITY, val32);

	/* Mask various recoverable errors */
	pci_cfg_read32(phb, bdfn, aercap + PCIECAP_AER_CE_MASK, &val32);
	val32 |= PCIECAP_AER_CE_MASK_ADV_NONFATAL;
	pci_cfg_write32(phb, bdfn, aercap + PCIECAP_AER_CE_MASK, val32);

	/* Enable ECRC check */
	pci_cfg_read32(phb, bdfn, aercap + PCIECAP_AER_CAPCTL, &val32);
	val32 |= (PCIECAP_AER_CAPCTL_ECRCG_EN |
		  PCIECAP_AER_CAPCTL_ECRCC_EN);
	pci_cfg_write32(phb, bdfn, aercap + PCIECAP_AER_CAPCTL, val32);

	/* Enable all error reporting */
	pci_cfg_read32(phb, bdfn, aercap + PCIECAP_AER_RERR_CMD, &val32);
	val32 |= (PCIECAP_AER_RERR_CMD_FE |
		  PCIECAP_AER_RERR_CMD_NFE |
		  PCIECAP_AER_RERR_CMD_CE);
	pci_cfg_write32(phb, bdfn, aercap + PCIECAP_AER_RERR_CMD, val32);
}

static void phb4_switch_port_init(struct phb *phb,
				  struct pci_device *dev,
				  int ecap, int aercap)
{
	uint16_t bdfn = dev->bdfn;
	uint16_t val16;
	uint32_t val32;

	// FIXME: update AER settings for phb4

	/* Enable SERR and parity checking and disable INTx */
	pci_cfg_read16(phb, bdfn, PCI_CFG_CMD, &val16);
	val16 |= (PCI_CFG_CMD_PERR_RESP |
		  PCI_CFG_CMD_SERR_EN |
		  PCI_CFG_CMD_INTx_DIS);
	pci_cfg_write16(phb, bdfn, PCI_CFG_CMD, val16);

	/* Disable partity error and enable system error */
	pci_cfg_read16(phb, bdfn, PCI_CFG_BRCTL, &val16);
	val16 &= ~PCI_CFG_BRCTL_PERR_RESP_EN;
	val16 |= PCI_CFG_BRCTL_SERR_EN;
	pci_cfg_write16(phb, bdfn, PCI_CFG_BRCTL, val16);

	/* Enable reporting various errors */
	if (!ecap) return;
	pci_cfg_read16(phb, bdfn, ecap + PCICAP_EXP_DEVCTL, &val16);
	val16 |= (PCICAP_EXP_DEVCTL_CE_REPORT |
		  PCICAP_EXP_DEVCTL_NFE_REPORT |
		  PCICAP_EXP_DEVCTL_FE_REPORT);
	/* HW279570 - Disable reporting of correctable errors */
	val16 &= ~PCICAP_EXP_DEVCTL_CE_REPORT;
	pci_cfg_write16(phb, bdfn, ecap + PCICAP_EXP_DEVCTL, val16);

	/* Unmask all unrecoverable errors */
	if (!aercap) return;
	pci_cfg_write32(phb, bdfn, aercap + PCIECAP_AER_UE_MASK, 0x0);

	/* Severity of unrecoverable errors */
	if (dev->dev_type == PCIE_TYPE_SWITCH_UPPORT)
		val32 = (PCIECAP_AER_UE_SEVERITY_DLLP |
			 PCIECAP_AER_UE_SEVERITY_SURPRISE_DOWN |
			 PCIECAP_AER_UE_SEVERITY_FLOW_CTL_PROT |
			 PCIECAP_AER_UE_SEVERITY_RECV_OVFLOW |
			 PCIECAP_AER_UE_SEVERITY_MALFORMED_TLP |
			 PCIECAP_AER_UE_SEVERITY_INTERNAL);
	else
		val32 = (PCIECAP_AER_UE_SEVERITY_FLOW_CTL_PROT |
			 PCIECAP_AER_UE_SEVERITY_INTERNAL);
	pci_cfg_write32(phb, bdfn, aercap + PCIECAP_AER_UE_SEVERITY, val32);

	/*
	 * Mask various correctable errors
	 */
	val32 = PCIECAP_AER_CE_MASK_ADV_NONFATAL;
	pci_cfg_write32(phb, bdfn, aercap + PCIECAP_AER_CE_MASK, val32);

	/* Enable ECRC generation and disable ECRC check */
	pci_cfg_read32(phb, bdfn, aercap + PCIECAP_AER_CAPCTL, &val32);
	val32 |= PCIECAP_AER_CAPCTL_ECRCG_EN;
	val32 &= ~PCIECAP_AER_CAPCTL_ECRCC_EN;
	pci_cfg_write32(phb, bdfn, aercap + PCIECAP_AER_CAPCTL, val32);
}

static void phb4_endpoint_init(struct phb *phb,
			       struct pci_device *dev,
			       int ecap, int aercap)
{
	uint16_t bdfn = dev->bdfn;
	uint16_t val16;
	uint32_t val32;

	/* Enable SERR and parity checking */
	pci_cfg_read16(phb, bdfn, PCI_CFG_CMD, &val16);
	val16 |= (PCI_CFG_CMD_PERR_RESP |
		  PCI_CFG_CMD_SERR_EN);
	pci_cfg_write16(phb, bdfn, PCI_CFG_CMD, val16);

	/* Enable reporting various errors */
	if (!ecap) return;
	pci_cfg_read16(phb, bdfn, ecap + PCICAP_EXP_DEVCTL, &val16);
	val16 &= ~PCICAP_EXP_DEVCTL_CE_REPORT;
	val16 |= (PCICAP_EXP_DEVCTL_NFE_REPORT |
		  PCICAP_EXP_DEVCTL_FE_REPORT |
		  PCICAP_EXP_DEVCTL_UR_REPORT);

	/* Enable ECRC generation and check */
	pci_cfg_read32(phb, bdfn, aercap + PCIECAP_AER_CAPCTL, &val32);
	val32 |= (PCIECAP_AER_CAPCTL_ECRCG_EN |
		  PCIECAP_AER_CAPCTL_ECRCC_EN);
	pci_cfg_write32(phb, bdfn, aercap + PCIECAP_AER_CAPCTL, val32);
}

static int64_t phb4_pcicfg_no_dstate(void *dev,
				     struct pci_cfg_reg_filter *pcrf,
				     uint32_t offset, uint32_t len,
				     uint32_t *data,  bool write)
{
	uint32_t loff = offset - pcrf->start;

	/* Disable D-state change on children of the PHB. For now we
	 * simply block all writes to the PM control/status
	 */
	if (write && loff >= 4 && loff < 6)
		return OPAL_SUCCESS;

	return OPAL_PARTIAL;
}

static void phb4_check_device_quirks(struct phb *phb, struct pci_device *dev)
{
	/* Some special adapter tweaks for devices directly under the PHB */
	if (dev->primary_bus != 1)
		return;

	/* PM quirk */
	if (!pci_has_cap(dev, PCI_CFG_CAP_ID_PM, false))
		return;

	pci_add_cfg_reg_filter(dev,
			       pci_cap(dev, PCI_CFG_CAP_ID_PM, false), 8,
			       PCI_REG_FLAG_WRITE,
			       phb4_pcicfg_no_dstate);
}

static int phb4_device_init(struct phb *phb, struct pci_device *dev,
			    void *data __unused)
{
	int ecap, aercap;

	/* Setup special device quirks */
	phb4_check_device_quirks(phb, dev);

	/* Common initialization for the device */
	pci_device_init(phb, dev);

	ecap = pci_cap(dev, PCI_CFG_CAP_ID_EXP, false);
	aercap = pci_cap(dev, PCIECAP_ID_AER, true);
	if (dev->dev_type == PCIE_TYPE_ROOT_PORT)
		phb4_root_port_init(phb, dev, ecap, aercap);
	else if (dev->dev_type == PCIE_TYPE_SWITCH_UPPORT ||
		 dev->dev_type == PCIE_TYPE_SWITCH_DNPORT)
		phb4_switch_port_init(phb, dev, ecap, aercap);
	else
		phb4_endpoint_init(phb, dev, ecap, aercap);

	return 0;
}

static int64_t phb4_pci_reinit(struct phb *phb, uint64_t scope, uint64_t data)
{
	struct pci_device *pd;
	uint16_t bdfn = data;
	int ret;

	if (scope != OPAL_REINIT_PCI_DEV)
		return OPAL_PARAMETER;

	pd = pci_find_dev(phb, bdfn);
	if (!pd)
		return OPAL_PARAMETER;

	ret = phb4_device_init(phb, pd, NULL);
	if (ret)
		return OPAL_HARDWARE;

	return OPAL_SUCCESS;
}

/* Default value for MBT0, see comments in init_ioda_cache() */
static uint64_t phb4_default_mbt0(struct phb4 *p, unsigned int bar_idx)
{
	uint64_t mbt0;

	if (p->rev == PHB4_REV_NIMBUS_DD10) {
		mbt0 = SETFIELD(IODA3_MBT0_MODE, 0ull, IODA3_MBT0_MODE_MDT);
		if (bar_idx == 0)
			mbt0 |= SETFIELD(IODA3_MBT0_MDT_COLUMN, 0ull, 0);
		else
			mbt0 |= SETFIELD(IODA3_MBT0_MDT_COLUMN, 0ull, 1);
	} else {
		switch (p->mbt_size - bar_idx - 1) {
		case 0:
			mbt0 = SETFIELD(IODA3_MBT0_MODE, 0ull, IODA3_MBT0_MODE_MDT);
			mbt0 = SETFIELD(IODA3_MBT0_MDT_COLUMN, mbt0, 3);
			break;
		case 1:
			mbt0 = SETFIELD(IODA3_MBT0_MODE, 0ull, IODA3_MBT0_MODE_MDT);
			mbt0 = SETFIELD(IODA3_MBT0_MDT_COLUMN, mbt0, 2);
			break;
		case 2:
			mbt0 = SETFIELD(IODA3_MBT0_MODE, 0ull, IODA3_MBT0_MODE_MDT);
			mbt0 = SETFIELD(IODA3_MBT0_MDT_COLUMN, mbt0, 1);
			break;
		default:
			mbt0 = SETFIELD(IODA3_MBT0_MODE, 0ull, IODA3_MBT0_MODE_PE_SEG);
		}
	}
	return mbt0;
}

/* Clear IODA cache tables */
static void phb4_init_ioda_cache(struct phb4 *p)
{
	uint32_t i;

	/*
	 * RTT and PELTV. RTE should be 0xFF's to indicate
	 * invalid PE# for the corresponding RID.
	 *
	 * Note: Instead we set all RTE entries to 0x00 to
	 * work around a problem where PE lookups might be
	 * done before Linux has established valid PE's
	 * (during PCI probing). We can revisit that once/if
	 * Linux has been fixed to always setup valid PEs.
	 *
	 * The value 0x00 corresponds to the default PE# Linux
	 * uses to check for config space freezes before it
	 * has assigned PE# to busses.
	 *
	 * WARNING: Additionally, we need to be careful, there's
	 * a HW issue, if we get an MSI on an RTT entry that is
	 * FF, things will go bad. We need to ensure we don't
	 * ever let a live FF RTT even temporarily when resetting
	 * for EEH etc... (HW278969).
	 */
	for (i = 0; i < ARRAY_SIZE(p->rte_cache); i++)
		p->rte_cache[i] = PHB4_RESERVED_PE_NUM(p);
	memset(p->peltv_cache, 0x0,  sizeof(p->peltv_cache));
	memset(p->tve_cache, 0x0, sizeof(p->tve_cache));

	/* XXX Should we mask them ? */
	memset(p->mist_cache, 0x0, sizeof(p->mist_cache));

	/* Configure MBT entries 1...N */
	if (p->rev == PHB4_REV_NIMBUS_DD10) {
		/* Since we configure the DD1.0 PHB4 with half the PE's,
		 * we need to give the illusion that we support only
		 * 128/256 segments half the segments.
		 *
		 * To achieve that, we configure *all* the M64 windows to use
		 * column 1 of the MDT, which is itself set so that segment 0
		 * and 1 map to PE0, 2 and 3 to PE1 etc...
		 *
		 * Column 0, 2 and 3 are left all 0, column 0 will be used for
		 * M32 and configured by the OS.
		 */
		for (i = 0; i < p->max_num_pes; i++)
			p->mdt_cache[i] = SETFIELD(IODA3_MDT_PE_B, 0ull, i >> 1);

	} else {
		/* On DD2.0 we don't have the above problem. We still use MDT
		 * column 1..3 for the last 3 BARs however, thus allowing Linux
		 * to remap those, and setup all the other ones for now in mode 00
		 * (segment# == PE#). By default those columns are set to map
		 * the same way.
		 */
		for (i = 0; i < p->max_num_pes; i++) {
			p->mdt_cache[i]  = SETFIELD(IODA3_MDT_PE_B, 0ull, i);
			p->mdt_cache[i] |= SETFIELD(IODA3_MDT_PE_C, 0ull, i);
			p->mdt_cache[i] |= SETFIELD(IODA3_MDT_PE_D, 0ull, i);
		}

	}

	/* Initialize MBT entries for BARs 1...N */
	for (i = 1; i < p->mbt_size; i++) {
		p->mbt_cache[i][0] = phb4_default_mbt0(p, i);
		p->mbt_cache[i][1] = 0;
	}

	/* Initialize M32 bar using MBT entry 0, MDT colunm A */
	p->mbt_cache[0][0] = SETFIELD(IODA3_MBT0_MODE, 0ull, IODA3_MBT0_MODE_MDT);
	p->mbt_cache[0][0] |= SETFIELD(IODA3_MBT0_MDT_COLUMN, 0ull, 0);
	p->mbt_cache[0][0] |= IODA3_MBT0_TYPE_M32 | (p->mm1_base & IODA3_MBT0_BASE_ADDR);
	p->mbt_cache[0][1] = IODA3_MBT1_ENABLE | ((~(M32_PCI_SIZE - 1)) & IODA3_MBT1_MASK);
}

static int64_t phb4_wait_bit(struct phb4 *p, uint32_t reg,
			     uint64_t mask, uint64_t want_val)
{
	uint64_t val;

	/* Wait for all pending TCE kills to complete
	 *
	 * XXX Add timeout...
	 */
	/* XXX SIMICS is nasty... */
	if ((reg == PHB_TCE_KILL || reg == PHB_DMARD_SYNC) &&
	    chip_quirk(QUIRK_SIMICS))
		return OPAL_SUCCESS;

	for (;;) {
		val = in_be64(p->regs + reg);
		if (val == 0xffffffffffffffffull) {
			/* XXX Fenced ? */
			return OPAL_HARDWARE;
		}
		if ((val & mask) == want_val)
			break;

	}
	return OPAL_SUCCESS;
}

static int64_t phb4_tce_kill(struct phb *phb, uint32_t kill_type,
			     uint64_t pe_number, uint32_t tce_size,
			     uint64_t dma_addr, uint32_t npages)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t val;
	int64_t rc;

	sync();
	switch(kill_type) {
	case OPAL_PCI_TCE_KILL_PAGES:
		while (npages--) {
			/* Wait for a slot in the HW kill queue */
			rc = phb4_wait_bit(p, PHB_TCE_KILL,
					   PHB_TCE_KILL_ALL |
					   PHB_TCE_KILL_PE |
					   PHB_TCE_KILL_ONE, 0);
			if (rc)
				return rc;
			val = SETFIELD(PHB_TCE_KILL_PENUM, dma_addr, pe_number);

			/* Set appropriate page size */
			switch(tce_size) {
			case 0x1000:
				if (dma_addr & 0xf000000000000fffull)
					return OPAL_PARAMETER;
				break;
			case 0x10000:
				if (dma_addr & 0xf00000000000ffffull)
					return OPAL_PARAMETER;
				val |= PHB_TCE_KILL_PSEL | PHB_TCE_KILL_64K;
				break;
			case 0x200000:
				if (dma_addr & 0xf0000000001fffffull)
					return OPAL_PARAMETER;
				val |= PHB_TCE_KILL_PSEL | PHB_TCE_KILL_2M;
				break;
			case 0x40000000:
				if (dma_addr & 0xf00000003fffffffull)
					return OPAL_PARAMETER;
				val |= PHB_TCE_KILL_PSEL | PHB_TCE_KILL_1G;
				break;
			default:
				return OPAL_PARAMETER;
			}
			/* Perform kill */
			out_be64(p->regs + PHB_TCE_KILL, PHB_TCE_KILL_ONE | val);
			/* Next page */
			dma_addr += tce_size;
		}
		break;
	case OPAL_PCI_TCE_KILL_PE:
		/* Wait for a slot in the HW kill queue */
		rc = phb4_wait_bit(p, PHB_TCE_KILL,
				   PHB_TCE_KILL_ALL |
				   PHB_TCE_KILL_PE |
				   PHB_TCE_KILL_ONE, 0);
		if (rc)
			return rc;
		/* Perform kill */
		out_be64(p->regs + PHB_TCE_KILL, PHB_TCE_KILL_PE |
			 SETFIELD(PHB_TCE_KILL_PENUM, 0ull, pe_number));
		break;
	case OPAL_PCI_TCE_KILL_ALL:
		/* Wait for a slot in the HW kill queue */
		rc = phb4_wait_bit(p, PHB_TCE_KILL,
				   PHB_TCE_KILL_ALL |
				   PHB_TCE_KILL_PE |
				   PHB_TCE_KILL_ONE, 0);
		if (rc)
			return rc;
		/* Perform kill */
		out_be64(p->regs + PHB_TCE_KILL, PHB_TCE_KILL_ALL);
		break;
	default:
		return OPAL_PARAMETER;
	}

	/* Start DMA sync process */
	out_be64(p->regs + PHB_DMARD_SYNC, PHB_DMARD_SYNC_START);

	/* Wait for kill to complete */
	rc = phb4_wait_bit(p, PHB_Q_DMA_R, PHB_Q_DMA_R_TCE_KILL_STATUS, 0);
	if (rc)
		return rc;

	/* Wait for DMA sync to complete */
	return phb4_wait_bit(p, PHB_DMARD_SYNC,
			     PHB_DMARD_SYNC_COMPLETE,
			     PHB_DMARD_SYNC_COMPLETE);
}

/* phb4_ioda_reset - Reset the IODA tables
 *
 * @purge: If true, the cache is cleared and the cleared values
 *         are applied to HW. If false, the cached values are
 *         applied to HW
 *
 * This reset the IODA tables in the PHB. It is called at
 * initialization time, on PHB reset, and can be called
 * explicitly from OPAL
 */
static int64_t phb4_ioda_reset(struct phb *phb, bool purge)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint32_t i;
	uint64_t val;

	if (purge) {
		PHBDBG(p, "Purging all IODA tables...\n");
		if (phb->slot)
			phb->slot->link_retries = PHB4_LINK_LINK_RETRIES;
		phb4_init_ioda_cache(p);
	}

	/* Init_30..31 - Errata workaround, clear PESTA entry 0 */
	phb4_ioda_sel(p, IODA3_TBL_PESTA, 0, false);
	out_be64(p->regs + PHB_IODA_DATA0, 0);

	/* Init_32..33 - MIST  */
	phb4_ioda_sel(p, IODA3_TBL_MIST, 0, true);
	val = in_be64(p->regs + PHB_IODA_ADDR);
	val = SETFIELD(PHB_IODA_AD_MIST_PWV, val, 0xf);
	out_be64(p->regs + PHB_IODA_ADDR, val);
	for (i = 0; i < (p->num_irqs/4); i++)
		out_be64(p->regs + PHB_IODA_DATA0, p->mist_cache[i]);

	/* Init_34..35 - MRT */
	phb4_ioda_sel(p, IODA3_TBL_MRT, 0, true);
	for (i = 0; i < p->mrt_size; i++)
		out_be64(p->regs + PHB_IODA_DATA0, 0);

	/* Init_36..37 - TVT */
	phb4_ioda_sel(p, IODA3_TBL_TVT, 0, true);
	for (i = 0; i < p->tvt_size; i++)
		out_be64(p->regs + PHB_IODA_DATA0, p->tve_cache[i]);

	/* Init_38..39 - MBT */
	phb4_ioda_sel(p, IODA3_TBL_MBT, 0, true);
	for (i = 0; i < p->mbt_size; i++) {
		out_be64(p->regs + PHB_IODA_DATA0, p->mbt_cache[i][0]);
		out_be64(p->regs + PHB_IODA_DATA0, p->mbt_cache[i][1]);
	}

	/* Init_40..41 - MDT */
	phb4_ioda_sel(p, IODA3_TBL_MDT, 0, true);
	for (i = 0; i < p->max_num_pes; i++)
		out_be64(p->regs + PHB_IODA_DATA0, p->mdt_cache[i]);

	/* Additional OPAL specific inits */

	/* Clear RTT and PELTV and PEST */
	if (p->tbl_rtt)
		memcpy((void *)p->tbl_rtt, p->rte_cache, RTT_TABLE_SIZE);
	if (p->tbl_peltv)
		memcpy((void *)p->tbl_peltv, p->peltv_cache, p->tbl_peltv_size);

	/* Clear PEST & PEEV */
	for (i = 0; i < p->max_num_pes; i++) {
		phb4_ioda_sel(p, IODA3_TBL_PESTA, i, false);
		out_be64(p->regs + PHB_IODA_DATA0, 0);
		phb4_ioda_sel(p, IODA3_TBL_PESTB, i, false);
		out_be64(p->regs + PHB_IODA_DATA0, 0);
	}

	phb4_ioda_sel(p, IODA3_TBL_PEEV, 0, true);
	for (i = 0; i < p->max_num_pes/64; i++)
		out_be64(p->regs + PHB_IODA_DATA0, 0);

	/* Invalidate RTE, TCE cache */
	out_be64(p->regs + PHB_RTC_INVALIDATE, PHB_RTC_INVALIDATE_ALL);

	return phb4_tce_kill(&p->phb, OPAL_PCI_TCE_KILL_ALL, 0, 0, 0, 0);
}

/*
 * Clear anything we have in PAPR Error Injection registers. Though
 * the spec says the PAPR error injection should be one-shot without
 * the "sticky" bit. However, that's false according to the experiments
 * I had. So we have to clear it at appropriate point in kernel to
 * avoid endless frozen PE.
 */
static int64_t phb4_papr_errinjct_reset(struct phb *phb)
{
	struct phb4 *p = phb_to_phb4(phb);

	out_be64(p->regs + PHB_PAPR_ERR_INJ_CTL, 0x0ul);
	out_be64(p->regs + PHB_PAPR_ERR_INJ_ADDR, 0x0ul);
	out_be64(p->regs + PHB_PAPR_ERR_INJ_MASK, 0x0ul);

	return OPAL_SUCCESS;
}

static int64_t phb4_set_phb_mem_window(struct phb *phb,
				       uint16_t window_type,
				       uint16_t window_num,
				       uint64_t addr,
				       uint64_t pci_addr,
				       uint64_t size)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t mbt0, mbt1;

	/*
	 * We have a unified MBT for all BARs on PHB4. However we
	 * also have a current limitation that only half of the PEs
	 * are available (in order to have 2 TVT entries per PE)
	 * on DD1.0
	 *
	 * So we use it as follow:
	 *
	 *  - M32 is hard wired to be MBT[0] and uses MDT column 0
	 *    for remapping.
	 *
	 *  - MBT[1..n] are available to the OS, currently only as
	 *    fully segmented or single PE (we don't yet expose the
	 *    new segmentation modes).
	 *
	 *  - [DD1.0] In order to deal with the above PE# limitations, since
	 *    the OS assumes the segmentation is done with as many
	 *    segments as PEs, we effectively fake it by mapping all
	 *    MBT[1..n] to NDT column 1 which has been configured to
	 *    give 2 adjacent segments the same PE# (see comment in
	 *    ioda cache init). We don't expose the other columns to
	 *    the OS.
	 *
	 *  - [DD2.0] We configure the 3 last BARs to columnt 1..3
	 *    initially set to segment# == PE#. We will need to provide some
	 *    extensions to the existing APIs to enable remapping of
	 *    segments on those BARs (and only those) as the current
	 *    API forces single segment mode.
	 */
	switch (window_type) {
	case OPAL_IO_WINDOW_TYPE:
	case OPAL_M32_WINDOW_TYPE:
		return OPAL_UNSUPPORTED;
	case OPAL_M64_WINDOW_TYPE:
		if (window_num == 0 || window_num >= p->mbt_size) {
			PHBERR(p, "%s: Invalid window %d\n",
			       __func__, window_num);
			return OPAL_PARAMETER;
		}

		mbt0 = p->mbt_cache[window_num][0];
		mbt1 = p->mbt_cache[window_num][1];

		/* XXX For now we assume the 4K minimum alignment,
		 * todo: check with the HW folks what the exact limits
		 * are based on the segmentation model.
		 */
		if ((addr & 0xFFFul) || (size & 0xFFFul)) {
			PHBERR(p, "%s: Bad addr/size alignment %llx/%llx\n",
			       __func__, addr, size);
			return OPAL_PARAMETER;
		}

		/* size should be 2^N */
		if (!size || size & (size-1)) {
			PHBERR(p, "%s: size not a power of 2: %llx\n",
			       __func__,  size);
			return OPAL_PARAMETER;
		}

		/* address should be size aligned */
		if (addr & (size - 1)) {
			PHBERR(p, "%s: addr not size aligned %llx/%llx\n",
			       __func__, addr, size);
			return OPAL_PARAMETER;
		}

		break;
	default:
		return OPAL_PARAMETER;
	}

	/* The BAR shouldn't be enabled yet */
	if (mbt0 & IODA3_MBT0_ENABLE)
		return OPAL_PARTIAL;

	/* Apply the settings */
	mbt0 = SETFIELD(IODA3_MBT0_BASE_ADDR, mbt0, addr >> 12);
	mbt1 = SETFIELD(IODA3_MBT1_MASK, mbt1, ~((size >> 12) -1));
	p->mbt_cache[window_num][0] = mbt0;
	p->mbt_cache[window_num][1] = mbt1;

	return OPAL_SUCCESS;
}

/*
 * For one specific M64 BAR, it can be shared by all PEs,
 * or owned by single PE exclusively.
 */
static int64_t phb4_phb_mmio_enable(struct phb __unused *phb,
				    uint16_t window_type,
				    uint16_t window_num,
				    uint16_t enable)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t mbt0, mbt1, base, mask;

	/*
	 * By design, PHB4 doesn't support IODT any more.
	 * Besides, we can't enable M32 BAR as well. So
	 * the function is used to do M64 mapping and each
	 * BAR is supposed to be shared by all PEs.
	 *
	 * TODO: Add support for some of the new PHB4 split modes
	 */
	switch (window_type) {
	case OPAL_IO_WINDOW_TYPE:
	case OPAL_M32_WINDOW_TYPE:
		return OPAL_UNSUPPORTED;
	case OPAL_M64_WINDOW_TYPE:
		/* Window 0 is reserved for M32 */
		if (window_num == 0 || window_num >= p->mbt_size ||
		    enable > OPAL_ENABLE_M64_NON_SPLIT) {
			PHBDBG(p,
			       "phb4_phb_mmio_enable wrong args (window %d enable %d)\n",
			       window_num, enable);
			return OPAL_PARAMETER;
		}
		break;
	default:
		return OPAL_PARAMETER;
	}

	/*
	 * We need check the base/mask while enabling
	 * the M64 BAR. Otherwise, invalid base/mask
	 * might cause fenced AIB unintentionally
	 */
	mbt0 = p->mbt_cache[window_num][0];
	mbt1 = p->mbt_cache[window_num][1];

	if (enable == OPAL_DISABLE_M64) {
		/* Reset the window to disabled & default mode */
		mbt0 = phb4_default_mbt0(p, window_num);
		mbt1 = 0;
	} else {
		/* Verify that the mode is valid and consistent */
		if (enable == OPAL_ENABLE_M64_SPLIT) {
			uint64_t mode = GETFIELD(IODA3_MBT0_MODE, mbt0);
			if (mode != IODA3_MBT0_MODE_PE_SEG &&
			    mode != IODA3_MBT0_MODE_MDT)
				return OPAL_PARAMETER;
		} else if (enable == OPAL_ENABLE_M64_NON_SPLIT) {
			if (GETFIELD(IODA3_MBT0_MODE, mbt0) !=
			    IODA3_MBT0_MODE_SINGLE_PE)
				return OPAL_PARAMETER;
		} else
			return OPAL_PARAMETER;

		base = GETFIELD(IODA3_MBT0_BASE_ADDR, mbt0);
		base = (base << 12);
		mask = GETFIELD(IODA3_MBT1_MASK, mbt1);
		if (base < p->mm0_base || !mask)
			return OPAL_PARTIAL;

		mbt0 |= IODA3_MBT0_ENABLE;
		mbt1 |= IODA3_MBT1_ENABLE;
	}

	/* Update HW and cache */
	p->mbt_cache[window_num][0] = mbt0;
	p->mbt_cache[window_num][1] = mbt1;
	phb4_ioda_sel(p, IODA3_TBL_MBT, window_num << 1, true);
	out_be64(p->regs + PHB_IODA_DATA0, mbt0);
	out_be64(p->regs + PHB_IODA_DATA0, mbt1);

	return OPAL_SUCCESS;
}

static int64_t phb4_map_pe_mmio_window(struct phb *phb,
				       uint64_t pe_number,
				       uint16_t window_type,
				       uint16_t window_num,
				       uint16_t segment_num)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t mbt0, mbt1, mdt0, mdt1;

	if (pe_number >= p->num_pes)
		return OPAL_PARAMETER;

	/*
	 * We support a combined MDT that has 4 columns. We let the OS
	 * use kernel 0 for M32.
	 *
	 * On DD1.0 we configure column1 ourselves to handle the "half PEs"
	 * problem and thus simulate having a smaller number of segments.
	 * columns 2 and 3 unused.
	 *
	 * On DD2.0 we configure the 3 last BARs to map column 3..1 which
	 * by default are set to map segment# == pe#, but can be remapped
	 * here if we extend this function.
	 *
	 * The problem is that the current API was "hijacked" so that an
	 * attempt at remapping any segment of an M64 has the effect of
	 * turning it into a single-PE mode BAR. So if we want to support
	 * remapping we'll have to play around this for example by creating
	 * a new API or a new window type...
	 */
	switch(window_type) {
	case OPAL_IO_WINDOW_TYPE:
		return OPAL_UNSUPPORTED;
	case OPAL_M32_WINDOW_TYPE:
		if (window_num != 0 || segment_num >= p->num_pes)
			return OPAL_PARAMETER;

		if (p->rev == PHB4_REV_NIMBUS_DD10) {
			mdt0 = p->mdt_cache[segment_num << 1];
			mdt1 = p->mdt_cache[(segment_num << 1) + 1];
			mdt0 = SETFIELD(IODA3_MDT_PE_A, mdt0, pe_number);
			mdt1 = SETFIELD(IODA3_MDT_PE_A, mdt1, pe_number);
			p->mdt_cache[segment_num << 1] = mdt0;
			p->mdt_cache[(segment_num << 1) + 1] = mdt1;
			phb4_ioda_sel(p, IODA3_TBL_MDT, segment_num << 1, true);
			out_be64(p->regs + PHB_IODA_DATA0, mdt0);
			out_be64(p->regs + PHB_IODA_DATA0, mdt1);
		} else {
			mdt0 = p->mdt_cache[segment_num];
			mdt0 = SETFIELD(IODA3_MDT_PE_A, mdt0, pe_number);
			phb4_ioda_sel(p, IODA3_TBL_MDT, segment_num, false);
			out_be64(p->regs + PHB_IODA_DATA0, mdt0);
		}
		break;
	case OPAL_M64_WINDOW_TYPE:
		if (window_num == 0 || window_num >= p->mbt_size)
			return OPAL_PARAMETER;

		mbt0 = p->mbt_cache[window_num][0];
		mbt1 = p->mbt_cache[window_num][1];

		/* The BAR shouldn't be enabled yet */
		if (mbt0 & IODA3_MBT0_ENABLE)
			return OPAL_PARTIAL;

		/* Set to single PE mode and configure the PE */
		mbt0 = SETFIELD(IODA3_MBT0_MODE, mbt0,
				IODA3_MBT0_MODE_SINGLE_PE);
		mbt1 = SETFIELD(IODA3_MBT1_SINGLE_PE_NUM, mbt1, pe_number);
		p->mbt_cache[window_num][0] = mbt0;
		p->mbt_cache[window_num][1] = mbt1;
		break;
	default:
		return OPAL_PARAMETER;
	}

	return OPAL_SUCCESS;
}

static int64_t phb4_map_pe_dma_window(struct phb *phb,
				      uint64_t pe_number,
				      uint16_t window_id,
				      uint16_t tce_levels,
				      uint64_t tce_table_addr,
				      uint64_t tce_table_size,
				      uint64_t tce_page_size)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t tts_encoded;
	uint64_t data64 = 0;

	/*
	 * We configure the PHB in 2 TVE per PE mode to match phb3.
	 * Current Linux implementation *requires* the two windows per
	 * PE.
	 *
	 * Note: On DD2.0 this is the normal mode of operation.
	 */

	/*
	 * Sanity check. We currently only support "2 window per PE" mode
	 * ie, only bit 59 of the PCI address is used to select the window
	 */
	if (pe_number >= p->num_pes || (window_id >> 1) != pe_number)
		return OPAL_PARAMETER;

	/*
	 * tce_table_size == 0 is used to disable an entry, in this case
	 * we ignore other arguments
	 */
	if (tce_table_size == 0) {
		phb4_ioda_sel(p, IODA3_TBL_TVT, window_id, false);
		out_be64(p->regs + PHB_IODA_DATA0, 0);
		p->tve_cache[window_id] = 0;
		return OPAL_SUCCESS;
	}

	/* Additional arguments validation */
	if (tce_levels < 1 || tce_levels > 5 ||
	    !is_pow2(tce_table_size) ||
	    tce_table_size < 0x1000)
		return OPAL_PARAMETER;

	/* Encode TCE table size */
	data64 = SETFIELD(IODA3_TVT_TABLE_ADDR, 0ul, tce_table_addr >> 12);
	tts_encoded = ilog2(tce_table_size) - 11;
	if (tts_encoded > 31)
		return OPAL_PARAMETER;
	data64 = SETFIELD(IODA3_TVT_TCE_TABLE_SIZE, data64, tts_encoded);

	/* Encode TCE page size */
	switch (tce_page_size) {
	case 0x1000:	/* 4K */
		data64 = SETFIELD(IODA3_TVT_IO_PSIZE, data64, 1);
		break;
	case 0x10000:	/* 64K */
		data64 = SETFIELD(IODA3_TVT_IO_PSIZE, data64, 5);
		break;
	case 0x200000:	/* 2M */
		data64 = SETFIELD(IODA3_TVT_IO_PSIZE, data64, 10);
		break;
	case 0x40000000: /* 1G */
		data64 = SETFIELD(IODA3_TVT_IO_PSIZE, data64, 19);
		break;
	default:
		return OPAL_PARAMETER;
	}

	/* Encode number of levels */
	data64 = SETFIELD(IODA3_TVT_NUM_LEVELS, data64, tce_levels - 1);

	phb4_ioda_sel(p, IODA3_TBL_TVT, window_id, false);
	out_be64(p->regs + PHB_IODA_DATA0, data64);
	p->tve_cache[window_id] = data64;

	return OPAL_SUCCESS;
}

static int64_t phb4_map_pe_dma_window_real(struct phb *phb,
					   uint64_t pe_number,
					   uint16_t window_id,
					   uint64_t pci_start_addr,
					   uint64_t pci_mem_size)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t end = pci_start_addr + pci_mem_size;
	uint64_t tve;

	if (pe_number >= p->num_pes ||
	    (window_id >> 1) != pe_number)
		return OPAL_PARAMETER;

	if (pci_mem_size) {
		/* Enable */

		/*
		 * Check that the start address has the right TVE index,
		 * we only support the 1 bit mode where each PE has 2
		 * TVEs
		 */
		if ((pci_start_addr >> 59) != (window_id & 1))
			return OPAL_PARAMETER;
		pci_start_addr &= ((1ull << 59) - 1);
		end = pci_start_addr + pci_mem_size;

		/* We have to be 16M aligned */
		if ((pci_start_addr & 0x00ffffff) ||
		    (pci_mem_size & 0x00ffffff))
			return OPAL_PARAMETER;

		/*
		 * It *looks* like this is the max we can support (we need
		 * to verify this. Also we are not checking for rollover,
		 * but then we aren't trying too hard to protect ourselves
		 * againt a completely broken OS.
		 */
		if (end > 0x0003ffffffffffffull)
			return OPAL_PARAMETER;

		/*
		 * Put start address bits 49:24 into TVE[52:53]||[0:23]
		 * and end address bits 49:24 into TVE[54:55]||[24:47]
		 * and set TVE[51]
		 */
		tve  = (pci_start_addr << 16) & (0xffffffull << 40);
		tve |= (pci_start_addr >> 38) & (3ull << 10);
		tve |= (end >>  8) & (0xfffffful << 16);
		tve |= (end >> 40) & (3ull << 8);
		tve |= PPC_BIT(51) | IODA3_TVT_NON_TRANSLATE_50;
	} else {
		/* Disable */
		tve = 0;
	}

	phb4_ioda_sel(p, IODA3_TBL_TVT, window_id, false);
	out_be64(p->regs + PHB_IODA_DATA0, tve);
	p->tve_cache[window_id] = tve;

	return OPAL_SUCCESS;
}

static int64_t phb4_set_ive_pe(struct phb *phb,
			       uint64_t pe_number,
			       uint32_t ive_num)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint32_t mist_idx;
	uint32_t mist_quad;
	uint32_t mist_shift;
	uint64_t val;

	if (pe_number >= p->num_pes || ive_num >= (p->num_irqs - 8))
		return OPAL_PARAMETER;

	mist_idx = ive_num >> 2;
	mist_quad = ive_num & 3;
	mist_shift = (3 - mist_quad) << 4;
	p->mist_cache[mist_idx] &= ~(0x0fffull << mist_shift);
	p->mist_cache[mist_idx] |=  ((uint64_t)pe_number) << mist_shift;

	/* Note: This has the side effect of clearing P/Q, so this
	 * shouldn't be called while the interrupt is "hot"
	 */

	phb4_ioda_sel(p, IODA3_TBL_MIST, mist_idx, false);

	/* We need to inject the appropriate MIST write enable bit
	 * in the IODA table address register
	 */
	val = in_be64(p->regs + PHB_IODA_ADDR);
	val = SETFIELD(PHB_IODA_AD_MIST_PWV, val, 8 >> mist_quad);
	out_be64(p->regs + PHB_IODA_ADDR, val);

	/* Write entry */
	out_be64(p->regs + PHB_IODA_DATA0, p->mist_cache[mist_idx]);

	return OPAL_SUCCESS;
}

static int64_t phb4_get_msi_32(struct phb *phb,
			       uint64_t pe_number,
			       uint32_t ive_num,
			       uint8_t msi_range,
			       uint32_t *msi_address,
			       uint32_t *message_data)
{
	struct phb4 *p = phb_to_phb4(phb);

	/*
	 * Sanity check. We needn't check on mve_number (PE#)
	 * on PHB3 since the interrupt source is purely determined
	 * by its DMA address and data, but the check isn't
	 * harmful.
	 */
	if (pe_number >= p->num_pes ||
	    ive_num >= (p->num_irqs - 8) ||
	    msi_range != 1 || !msi_address|| !message_data)
		return OPAL_PARAMETER;

	/*
	 * DMA address and data will form the IVE index.
	 * For more details, please refer to IODA2 spec.
	 */
	*msi_address = 0xFFFF0000 | ((ive_num << 4) & 0xFFFFFE0F);
	*message_data = ive_num & 0x1F;

	return OPAL_SUCCESS;
}

static int64_t phb4_get_msi_64(struct phb *phb,
			       uint64_t pe_number,
			       uint32_t ive_num,
			       uint8_t msi_range,
			       uint64_t *msi_address,
			       uint32_t *message_data)
{
	struct phb4 *p = phb_to_phb4(phb);

	/* Sanity check */
	if (pe_number >= p->num_pes ||
	    ive_num >= (p->num_irqs - 8) ||
	    msi_range != 1 || !msi_address || !message_data)
		return OPAL_PARAMETER;

	/*
	 * DMA address and data will form the IVE index.
	 * For more details, please refer to IODA2 spec.
	 */
	*msi_address = (0x1ul << 60) | ((ive_num << 4) & 0xFFFFFFFFFFFFFE0Ful);
	*message_data = ive_num & 0x1F;

	return OPAL_SUCCESS;
}

static void phb4_rc_err_clear(struct phb4 *p)
{
	/* Init_47 - Clear errors */
	phb4_pcicfg_write16(&p->phb, 0, PCI_CFG_SECONDARY_STATUS, 0xffff);

	if (p->ecap <= 0)
		return;

	phb4_pcicfg_write16(&p->phb, 0, p->ecap + PCICAP_EXP_DEVSTAT,
			     PCICAP_EXP_DEVSTAT_CE	|
			     PCICAP_EXP_DEVSTAT_NFE	|
			     PCICAP_EXP_DEVSTAT_FE	|
			     PCICAP_EXP_DEVSTAT_UE);

	if (p->aercap <= 0)
		return;

	/* Clear all UE status */
	phb4_pcicfg_write32(&p->phb, 0, p->aercap + PCIECAP_AER_UE_STATUS,
			     0xffffffff);
	/* Clear all CE status */
	phb4_pcicfg_write32(&p->phb, 0, p->aercap + PCIECAP_AER_CE_STATUS,
			     0xffffffff);
	/* Clear root error status */
	phb4_pcicfg_write32(&p->phb, 0, p->aercap + PCIECAP_AER_RERR_STA,
			     0xffffffff);
}

static void phb4_err_clear_regb(struct phb4 *p)
{
	uint64_t val64;

	val64 = phb4_read_reg(p, PHB_REGB_ERR_STATUS);
	phb4_write_reg(p, PHB_REGB_ERR_STATUS, val64);
	phb4_write_reg(p, PHB_REGB_ERR1_STATUS, 0x0ul);
	phb4_write_reg(p, PHB_REGB_ERR_LOG_0, 0x0ul);
	phb4_write_reg(p, PHB_REGB_ERR_LOG_1, 0x0ul);
}

/*
 * The function can be called during error recovery for all classes of
 * errors.  This is new to PHB4; previous revisions had separate
 * sequences for INF/ER/Fatal errors.
 *
 * "Rec #" in this function refer to "Recov_#" steps in the
 * PHB4 INF recovery sequence.
 */
static void phb4_err_clear(struct phb4 *p)
{
	uint64_t val64;
	uint64_t fir = phb4_read_reg(p, PHB_LEM_FIR_ACCUM);

	/* Rec 1: Acquire the PCI config lock (we don't need to do this) */

	/* Rec 2...15: Clear error status in RC config space */
	phb4_rc_err_clear(p);

	/* Rec 16...23: Clear PBL errors */
	val64 = phb4_read_reg(p, PHB_PBL_ERR_STATUS);
	phb4_write_reg(p, PHB_PBL_ERR_STATUS, val64);
	phb4_write_reg(p, PHB_PBL_ERR1_STATUS, 0x0ul);
	phb4_write_reg(p, PHB_PBL_ERR_LOG_0, 0x0ul);
	phb4_write_reg(p, PHB_PBL_ERR_LOG_1, 0x0ul);

	/* Rec 24...31: Clear REGB errors */
	phb4_err_clear_regb(p);

	/* Rec 32...59: Clear PHB error trap */
	val64 = phb4_read_reg(p, PHB_TXE_ERR_STATUS);
	phb4_write_reg(p, PHB_TXE_ERR_STATUS, val64);
	phb4_write_reg(p, PHB_TXE_ERR1_STATUS, 0x0ul);
	phb4_write_reg(p, PHB_TXE_ERR_LOG_0, 0x0ul);
	phb4_write_reg(p, PHB_TXE_ERR_LOG_1, 0x0ul);

	val64 = phb4_read_reg(p, PHB_RXE_ARB_ERR_STATUS);
	phb4_write_reg(p, PHB_RXE_ARB_ERR_STATUS, val64);
	phb4_write_reg(p, PHB_RXE_ARB_ERR1_STATUS, 0x0ul);
	phb4_write_reg(p, PHB_RXE_ARB_ERR_LOG_0, 0x0ul);
	phb4_write_reg(p, PHB_RXE_ARB_ERR_LOG_1, 0x0ul);

	val64 = phb4_read_reg(p, PHB_RXE_MRG_ERR_STATUS);
	phb4_write_reg(p, PHB_RXE_MRG_ERR_STATUS, val64);
	phb4_write_reg(p, PHB_RXE_MRG_ERR1_STATUS, 0x0ul);
	phb4_write_reg(p, PHB_RXE_MRG_ERR_LOG_0, 0x0ul);
	phb4_write_reg(p, PHB_RXE_MRG_ERR_LOG_1, 0x0ul);

	val64 = phb4_read_reg(p, PHB_RXE_TCE_ERR_STATUS);
	phb4_write_reg(p, PHB_RXE_TCE_ERR_STATUS, val64);
	phb4_write_reg(p, PHB_RXE_TCE_ERR1_STATUS, 0x0ul);
	phb4_write_reg(p, PHB_RXE_TCE_ERR_LOG_0, 0x0ul);
	phb4_write_reg(p, PHB_RXE_TCE_ERR_LOG_1, 0x0ul);

	val64 = phb4_read_reg(p, PHB_ERR_STATUS);
	phb4_write_reg(p, PHB_ERR_STATUS, val64);
	phb4_write_reg(p, PHB_ERR1_STATUS, 0x0ul);
	phb4_write_reg(p, PHB_ERR_LOG_0, 0x0ul);
	phb4_write_reg(p, PHB_ERR_LOG_1, 0x0ul);

	/* Rec 61/62: Clear FIR/WOF */
	phb4_write_reg(p, PHB_LEM_FIR_AND_MASK, ~fir);
	phb4_write_reg(p, PHB_LEM_WOF, 0x0ul);

	/* Rec 63: Update LEM mask to its initial value */
	phb4_write_reg(p, PHB_LEM_ERROR_MASK, 0x0ul);

	/* Rec 64: Clear the PCI config lock (we don't need to do this) */
}

static void phb4_read_phb_status(struct phb4 *p,
				 struct OpalIoPhb4ErrorData *stat)
{
	uint16_t val = 0;
	uint32_t i;
	uint64_t *pPEST;

	memset(stat, 0, sizeof(struct OpalIoPhb4ErrorData));

	/* Error data common part */
	stat->common.version = OPAL_PHB_ERROR_DATA_VERSION_1;
	stat->common.ioType  = OPAL_PHB_ERROR_DATA_TYPE_PHB4;
	stat->common.len     = sizeof(struct OpalIoPhb4ErrorData);

	/* Use ASB for config space if the PHB is fenced */
	if (p->flags & PHB4_AIB_FENCED)
		p->flags |= PHB4_CFG_USE_ASB;

	/* Grab RC bridge control, make it 32-bit */
	phb4_pcicfg_read16(&p->phb, 0, PCI_CFG_BRCTL, &val);
	stat->brdgCtl = val;

	/*
	 * Grab various RC PCIe capability registers. All device, slot
	 * and link status are 16-bit, so we grab the pair control+status
	 * for each of them
	 */
	phb4_pcicfg_read32(&p->phb, 0, p->ecap + PCICAP_EXP_DEVCTL,
			   &stat->deviceStatus);
	phb4_pcicfg_read32(&p->phb, 0, p->ecap + PCICAP_EXP_SLOTCTL,
			   &stat->slotStatus);
	phb4_pcicfg_read32(&p->phb, 0, p->ecap + PCICAP_EXP_LCTL,
			   &stat->linkStatus);

	 /*
	 * I assume those are the standard config space header, cmd & status
	 * together makes 32-bit. Secondary status is 16-bit so I'll clear
	 * the top on that one
	 */
	phb4_pcicfg_read32(&p->phb, 0, PCI_CFG_CMD, &stat->devCmdStatus);
	phb4_pcicfg_read16(&p->phb, 0, PCI_CFG_SECONDARY_STATUS, &val);
	stat->devSecStatus = val;

	/* Grab a bunch of AER regs */
	phb4_pcicfg_read32(&p->phb, 0, p->aercap + PCIECAP_AER_RERR_STA,
			   &stat->rootErrorStatus);
	phb4_pcicfg_read32(&p->phb, 0, p->aercap + PCIECAP_AER_UE_STATUS,
			   &stat->uncorrErrorStatus);
	phb4_pcicfg_read32(&p->phb, 0, p->aercap + PCIECAP_AER_CE_STATUS,
			   &stat->corrErrorStatus);
	phb4_pcicfg_read32(&p->phb, 0, p->aercap + PCIECAP_AER_HDR_LOG0,
			   &stat->tlpHdr1);
	phb4_pcicfg_read32(&p->phb, 0, p->aercap + PCIECAP_AER_HDR_LOG1,
			   &stat->tlpHdr2);
	phb4_pcicfg_read32(&p->phb, 0, p->aercap + PCIECAP_AER_HDR_LOG2,
			   &stat->tlpHdr3);
	phb4_pcicfg_read32(&p->phb, 0, p->aercap + PCIECAP_AER_HDR_LOG3,
			   &stat->tlpHdr4);
	phb4_pcicfg_read32(&p->phb, 0, p->aercap + PCIECAP_AER_SRCID,
			   &stat->sourceId);

	/* PEC NFIR, same as P8/PHB3 */
	xscom_read(p->chip_id, p->pe_stk_xscom + 0x0, &stat->nFir);
	xscom_read(p->chip_id, p->pe_stk_xscom + 0x3, &stat->nFirMask);
	xscom_read(p->chip_id, p->pe_stk_xscom + 0x8, &stat->nFirWOF);

	/* PHB4 inbound and outbound error Regs */
	stat->phbPlssr = phb4_read_reg_asb(p, PHB_CPU_LOADSTORE_STATUS);
	stat->phbCsr = phb4_read_reg_asb(p, PHB_DMA_CHAN_STATUS);
	stat->lemFir = phb4_read_reg_asb(p, PHB_LEM_FIR_ACCUM);
	stat->lemErrorMask = phb4_read_reg_asb(p, PHB_LEM_ERROR_MASK);
	stat->lemWOF = phb4_read_reg_asb(p, PHB_LEM_WOF);
	stat->phbErrorStatus = phb4_read_reg_asb(p, PHB_ERR_STATUS);
	stat->phbFirstErrorStatus = phb4_read_reg_asb(p, PHB_ERR1_STATUS);
	stat->phbErrorLog0 = phb4_read_reg_asb(p, PHB_ERR_LOG_0);
	stat->phbErrorLog1 = phb4_read_reg_asb(p, PHB_ERR_LOG_1);
	stat->phbTxeErrorStatus = phb4_read_reg_asb(p, PHB_TXE_ERR_STATUS);
	stat->phbTxeFirstErrorStatus = phb4_read_reg_asb(p, PHB_TXE_ERR1_STATUS);
	stat->phbTxeErrorLog0 = phb4_read_reg_asb(p, PHB_TXE_ERR_LOG_0);
	stat->phbTxeErrorLog1 = phb4_read_reg_asb(p, PHB_TXE_ERR_LOG_1);
	stat->phbRxeArbErrorStatus = phb4_read_reg_asb(p, PHB_RXE_ARB_ERR_STATUS);
	stat->phbRxeArbFirstErrorStatus = phb4_read_reg_asb(p, PHB_RXE_ARB_ERR1_STATUS);
	stat->phbRxeArbErrorLog0 = phb4_read_reg_asb(p, PHB_RXE_ARB_ERR_LOG_0);
	stat->phbRxeArbErrorLog1 = phb4_read_reg_asb(p, PHB_RXE_ARB_ERR_LOG_1);
	stat->phbRxeMrgErrorStatus = phb4_read_reg_asb(p, PHB_RXE_MRG_ERR_STATUS);
	stat->phbRxeMrgFirstErrorStatus = phb4_read_reg_asb(p, PHB_RXE_MRG_ERR1_STATUS);
	stat->phbRxeMrgErrorLog0 = phb4_read_reg_asb(p, PHB_RXE_MRG_ERR_LOG_0);
	stat->phbRxeMrgErrorLog1 = phb4_read_reg_asb(p, PHB_RXE_MRG_ERR_LOG_1);
	stat->phbRxeTceErrorStatus = phb4_read_reg_asb(p, PHB_RXE_TCE_ERR_STATUS);
	stat->phbRxeTceFirstErrorStatus = phb4_read_reg_asb(p, PHB_RXE_TCE_ERR1_STATUS);
	stat->phbRxeTceErrorLog0 = phb4_read_reg_asb(p, PHB_RXE_TCE_ERR_LOG_0);
	stat->phbRxeTceErrorLog1 = phb4_read_reg_asb(p, PHB_RXE_TCE_ERR_LOG_1);

	/* PHB4 REGB error registers */
	stat->phbPblErrorStatus = phb4_read_reg_asb(p, PHB_PBL_ERR_STATUS);
	stat->phbPblFirstErrorStatus = phb4_read_reg_asb(p, PHB_PBL_ERR1_STATUS);
	stat->phbPblErrorLog0 = phb4_read_reg_asb(p, PHB_PBL_ERR_LOG_0);
	stat->phbPblErrorLog1 = phb4_read_reg_asb(p, PHB_PBL_ERR_LOG_1);

	stat->phbPcieDlpErrorStatus = phb4_read_reg_asb(p, PHB_PCIE_DLP_ERR_STATUS);
	stat->phbPcieDlpErrorLog1 = phb4_read_reg_asb(p, PHB_PCIE_DLP_ERRLOG1);
	stat->phbPcieDlpErrorLog2 = phb4_read_reg_asb(p, PHB_PCIE_DLP_ERRLOG2);

	stat->phbRegbErrorStatus = phb4_read_reg_asb(p, PHB_REGB_ERR_STATUS);
	stat->phbRegbFirstErrorStatus = phb4_read_reg_asb(p, PHB_REGB_ERR1_STATUS);
	stat->phbRegbErrorLog0 = phb4_read_reg_asb(p, PHB_REGB_ERR_LOG_0);
	stat->phbRegbErrorLog1 = phb4_read_reg_asb(p, PHB_REGB_ERR_LOG_1);

	/*
	 * Grab PESTA & B content. The error bit (bit#0) should
	 * be fetched from IODA and the left content from memory
	 * resident tables.
	 */
	 pPEST = (uint64_t *)p->tbl_pest;
	 phb4_ioda_sel(p, IODA3_TBL_PESTA, 0, true);
	 for (i = 0; i < OPAL_PHB4_NUM_PEST_REGS; i++) {
		 stat->pestA[i] = phb4_read_reg_asb(p, PHB_IODA_DATA0);
		 stat->pestA[i] |= pPEST[2 * i];
	 }

	 phb4_ioda_sel(p, IODA3_TBL_PESTB, 0, true);
	 for (i = 0; i < OPAL_PHB4_NUM_PEST_REGS; i++) {
		 stat->pestB[i] = phb4_read_reg_asb(p, PHB_IODA_DATA0);
		 stat->pestB[i] |= pPEST[2 * i + 1];
	 }
}

static void phb4_eeh_dump_regs(struct phb4 *p)
{
	struct OpalIoPhb4ErrorData *s;
	uint16_t reg;
	unsigned int i;

	if (!verbose_eeh)
		return;

	s = zalloc(sizeof(struct OpalIoPhb4ErrorData));
	phb4_read_phb_status(p, s);


	PHBERR(p, "brdgCtl        = %08x\n", s->brdgCtl);

	/* PHB4 cfg regs */
	PHBERR(p, "            deviceStatus = %08x\n", s->deviceStatus);
	PHBERR(p, "              slotStatus = %08x\n", s->slotStatus);
	PHBERR(p, "              linkStatus = %08x\n", s->linkStatus);
	PHBERR(p, "            devCmdStatus = %08x\n", s->devCmdStatus);
	PHBERR(p, "            devSecStatus = %08x\n", s->devSecStatus);
	PHBERR(p, "         rootErrorStatus = %08x\n", s->rootErrorStatus);
	PHBERR(p, "         corrErrorStatus = %08x\n", s->corrErrorStatus);
	PHBERR(p, "       uncorrErrorStatus = %08x\n", s->uncorrErrorStatus);

	/* Two non OPAL API registers that are useful */
	phb4_pcicfg_read16(&p->phb, 0, p->ecap + PCICAP_EXP_DEVCTL, &reg);
	PHBERR(p, "                  devctl = %08x\n", reg);
	phb4_pcicfg_read16(&p->phb, 0, p->ecap + PCICAP_EXP_DEVSTAT,
			   &reg);
	PHBERR(p, "                 devStat = %08x\n", reg);

	/* Byte swap TLP headers so they are the same as the PCIe spec */
	PHBERR(p, "                 tlpHdr1 = %08x\n", bswap_32(s->tlpHdr1));
	PHBERR(p, "                 tlpHdr2 = %08x\n", bswap_32(s->tlpHdr2));
	PHBERR(p, "                 tlpHdr3 = %08x\n", bswap_32(s->tlpHdr3));
	PHBERR(p, "                 tlpHdr4 = %08x\n", bswap_32(s->tlpHdr4));
	PHBERR(p, "                sourceId = %08x\n", s->sourceId);
	PHBERR(p, "                    nFir = %016llx\n", s->nFir);
	PHBERR(p, "                nFirMask = %016llx\n", s->nFirMask);
	PHBERR(p, "                 nFirWOF = %016llx\n", s->nFirWOF);
	PHBERR(p, "                phbPlssr = %016llx\n", s->phbPlssr);
	PHBERR(p, "                  phbCsr = %016llx\n", s->phbCsr);
	PHBERR(p, "                  lemFir = %016llx\n", s->lemFir);
	PHBERR(p, "            lemErrorMask = %016llx\n", s->lemErrorMask);
	PHBERR(p, "                  lemWOF = %016llx\n", s->lemWOF);
	PHBERR(p, "          phbErrorStatus = %016llx\n", s->phbErrorStatus);
	PHBERR(p, "     phbFirstErrorStatus = %016llx\n", s->phbFirstErrorStatus);
	PHBERR(p, "            phbErrorLog0 = %016llx\n", s->phbErrorLog0);
	PHBERR(p, "            phbErrorLog1 = %016llx\n", s->phbErrorLog1);
	PHBERR(p, "       phbTxeErrorStatus = %016llx\n", s->phbTxeErrorStatus);
	PHBERR(p, "  phbTxeFirstErrorStatus = %016llx\n", s->phbTxeFirstErrorStatus);
	PHBERR(p, "         phbTxeErrorLog0 = %016llx\n", s->phbTxeErrorLog0);
	PHBERR(p, "         phbTxeErrorLog1 = %016llx\n", s->phbTxeErrorLog1);
	PHBERR(p, "    phbRxeArbErrorStatus = %016llx\n", s->phbRxeArbErrorStatus);
	PHBERR(p, "phbRxeArbFrstErrorStatus = %016llx\n", s->phbRxeArbFirstErrorStatus);
	PHBERR(p, "      phbRxeArbErrorLog0 = %016llx\n", s->phbRxeArbErrorLog0);
	PHBERR(p, "      phbRxeArbErrorLog1 = %016llx\n", s->phbRxeArbErrorLog1);
	PHBERR(p, "    phbRxeMrgErrorStatus = %016llx\n", s->phbRxeMrgErrorStatus);
	PHBERR(p, "phbRxeMrgFrstErrorStatus = %016llx\n", s->phbRxeMrgFirstErrorStatus);
	PHBERR(p, "      phbRxeMrgErrorLog0 = %016llx\n", s->phbRxeMrgErrorLog0);
	PHBERR(p, "      phbRxeMrgErrorLog1 = %016llx\n", s->phbRxeMrgErrorLog1);
	PHBERR(p, "    phbRxeTceErrorStatus = %016llx\n", s->phbRxeTceErrorStatus);
	PHBERR(p, "phbRxeTceFrstErrorStatus = %016llx\n", s->phbRxeTceFirstErrorStatus);
	PHBERR(p, "      phbRxeTceErrorLog0 = %016llx\n", s->phbRxeTceErrorLog0);
	PHBERR(p, "      phbRxeTceErrorLog1 = %016llx\n", s->phbRxeTceErrorLog1);
	PHBERR(p, "       phbPblErrorStatus = %016llx\n", s->phbPblErrorStatus);
	PHBERR(p, "  phbPblFirstErrorStatus = %016llx\n", s->phbPblFirstErrorStatus);
	PHBERR(p, "         phbPblErrorLog0 = %016llx\n", s->phbPblErrorLog0);
	PHBERR(p, "         phbPblErrorLog1 = %016llx\n", s->phbPblErrorLog1);
	PHBERR(p, "     phbPcieDlpErrorLog1 = %016llx\n", s->phbPcieDlpErrorLog1);
	PHBERR(p, "     phbPcieDlpErrorLog2 = %016llx\n", s->phbPcieDlpErrorLog2);
	PHBERR(p, "   phbPcieDlpErrorStatus = %016llx\n", s->phbPcieDlpErrorStatus);

	PHBERR(p, "      phbRegbErrorStatus = %016llx\n", s->phbRegbErrorStatus);
	PHBERR(p, " phbRegbFirstErrorStatus = %016llx\n", s->phbRegbFirstErrorStatus);
	PHBERR(p, "        phbRegbErrorLog0 = %016llx\n", s->phbRegbErrorLog0);
	PHBERR(p, "        phbRegbErrorLog1 = %016llx\n", s->phbRegbErrorLog1);

	for (i = 0; i < OPAL_PHB4_NUM_PEST_REGS; i++) {
		if (!s->pestA[i] && !s->pestB[i])
			continue;
		PHBERR(p, "               PEST[%03d] = %016llx %016llx\n",
		       i, s->pestA[i], s->pestB[i]);
	}
	free(s);
}

static int64_t phb4_set_pe(struct phb *phb,
			   uint64_t pe_number,
			   uint64_t bdfn,
			   uint8_t bcompare,
			   uint8_t dcompare,
			   uint8_t fcompare,
			   uint8_t action)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t mask, val, tmp, idx;
	int32_t all = 0;
	uint16_t *rte;

	/* Sanity check */
	if (!p->tbl_rtt)
		return OPAL_HARDWARE;
	if (action != OPAL_MAP_PE && action != OPAL_UNMAP_PE)
		return OPAL_PARAMETER;
	if (pe_number >= p->num_pes || bdfn > 0xffff ||
	    bcompare > OpalPciBusAll ||
	    dcompare > OPAL_COMPARE_RID_DEVICE_NUMBER ||
	    fcompare > OPAL_COMPARE_RID_FUNCTION_NUMBER)
		return OPAL_PARAMETER;

	/* Figure out the RID range */
	if (bcompare == OpalPciBusAny) {
		mask = 0x0;
		val  = 0x0;
		all  = 0x1;
	} else {
		tmp  = ((0x1 << (bcompare + 1)) - 1) << (15 - bcompare);
		mask = tmp;
		val  = bdfn & tmp;
	}

	if (dcompare == OPAL_IGNORE_RID_DEVICE_NUMBER)
		all = (all << 1) | 0x1;
	else {
		mask |= 0xf8;
		val  |= (bdfn & 0xf8);
	}

	if (fcompare == OPAL_IGNORE_RID_FUNCTION_NUMBER)
		all = (all << 1) | 0x1;
	else {
		mask |= 0x7;
		val  |= (bdfn & 0x7);
	}

	/* Map or unmap the RTT range */
	if (all == 0x7) {
		if (action == OPAL_MAP_PE) {
			for (idx = 0; idx < RTT_TABLE_ENTRIES; idx++)
				p->rte_cache[idx] = pe_number;
		} else {
			for (idx = 0; idx < ARRAY_SIZE(p->rte_cache); idx++)
				p->rte_cache[idx] = PHB4_RESERVED_PE_NUM(p);
		}
		memcpy((void *)p->tbl_rtt, p->rte_cache, RTT_TABLE_SIZE);
	} else {
		rte = (uint16_t *)p->tbl_rtt;
		for (idx = 0; idx < RTT_TABLE_ENTRIES; idx++, rte++) {
			if ((idx & mask) != val)
				continue;
			if (action == OPAL_MAP_PE)
				p->rte_cache[idx] = pe_number;
			else
				p->rte_cache[idx] = PHB4_RESERVED_PE_NUM(p);
			*rte = p->rte_cache[idx];
		}
	}

	/* Invalidate the entire RTC */
	out_be64(p->regs + PHB_RTC_INVALIDATE, PHB_RTC_INVALIDATE_ALL);

	return OPAL_SUCCESS;
}

static int64_t phb4_set_peltv(struct phb *phb,
			      uint32_t parent_pe,
			      uint32_t child_pe,
			      uint8_t state)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint8_t *peltv;
	uint32_t idx, mask;

	/* Sanity check */
	if (!p->tbl_peltv)
		return OPAL_HARDWARE;
	if (parent_pe >= p->num_pes || child_pe >= p->num_pes)
		return OPAL_PARAMETER;

	/* Find index for parent PE */
	idx = parent_pe * (p->max_num_pes / 8);
	idx += (child_pe / 8);
	mask = 0x1 << (7 - (child_pe % 8));

	peltv = (uint8_t *)p->tbl_peltv;
	peltv += idx;
	if (state) {
		*peltv |= mask;
		p->peltv_cache[idx] |= mask;
	} else {
		*peltv &= ~mask;
		p->peltv_cache[idx] &= ~mask;
	}

	return OPAL_SUCCESS;
}

static void phb4_prepare_link_change(struct pci_slot *slot, bool is_up)
{
	struct phb4 *p = phb_to_phb4(slot->phb);
	uint32_t reg32;

	p->has_link = is_up;

	if (is_up) {
		/* Clear AER receiver error status */
		phb4_pcicfg_write32(&p->phb, 0, p->aercap +
				    PCIECAP_AER_CE_STATUS,
				    PCIECAP_AER_CE_RECVR_ERR);
		/* Unmask receiver error status in AER */
		phb4_pcicfg_read32(&p->phb, 0, p->aercap +
				   PCIECAP_AER_CE_MASK, &reg32);
		reg32 &= ~PCIECAP_AER_CE_RECVR_ERR;
		phb4_pcicfg_write32(&p->phb, 0, p->aercap +
				    PCIECAP_AER_CE_MASK, reg32);

		/* Don't block PCI-CFG */
		p->flags &= ~PHB4_CFG_BLOCKED;

		/* Re-enable link down errors */
		out_be64(p->regs + PHB_PCIE_MISC_STRAP,
			 0x0000060000000000ull);

		/* Re-enable error status indicators that trigger irqs */
		out_be64(p->regs + PHB_REGB_ERR_INF_ENABLE,
			 0x2130006efca8bc00ull);
		out_be64(p->regs + PHB_REGB_ERR_ERC_ENABLE,
			 0x0080000000000000ull);
		out_be64(p->regs + PHB_REGB_ERR_FAT_ENABLE,
			 0xde0fff91035743ffull);

		/*
		 * We might lose the bus numbers during the reset operation
		 * and we need to restore them. Otherwise, some adapters (e.g.
		 * IPR) can't be probed properly by the kernel. We don't need
		 * to restore bus numbers for every kind of reset, however,
		 * it's not harmful to always restore the bus numbers, which
		 * simplifies the logic.
		 */
		pci_restore_bridge_buses(slot->phb, slot->pd);
		if (slot->phb->ops->device_init)
			pci_walk_dev(slot->phb, slot->pd,
				     slot->phb->ops->device_init, NULL);
	} else {
		/* Mask AER receiver error */
		phb4_pcicfg_read32(&p->phb, 0, p->aercap +
				   PCIECAP_AER_CE_MASK, &reg32);
		reg32 |= PCIECAP_AER_CE_RECVR_ERR;
		phb4_pcicfg_write32(&p->phb, 0, p->aercap +
				    PCIECAP_AER_CE_MASK, reg32);

		/* Clear error link enable & error link down kill enable */
		out_be64(p->regs + PHB_PCIE_MISC_STRAP, 0);

		/* Disable all error status indicators that trigger irqs */
		out_be64(p->regs + PHB_REGB_ERR_INF_ENABLE, 0);
		out_be64(p->regs + PHB_REGB_ERR_ERC_ENABLE, 0);
		out_be64(p->regs + PHB_REGB_ERR_FAT_ENABLE, 0);

		/* Block PCI-CFG access */
		p->flags |= PHB4_CFG_BLOCKED;
	}
}

static int64_t phb4_get_presence_state(struct pci_slot *slot, uint8_t *val)
{
	struct phb4 *p = phb_to_phb4(slot->phb);
	uint64_t hps, dtctl;

	/* Test for PHB in error state ? */
	if (p->broken)
		return OPAL_HARDWARE;

	/* Check hotplug status */
	hps = in_be64(p->regs + PHB_PCIE_HOTPLUG_STATUS);
	if (!(hps & PHB_PCIE_HPSTAT_PRESENCE)) {
		*val = OPAL_PCI_SLOT_PRESENT;
	} else {
		/*
		 * If it says not present but link is up, then we assume
		 * we are on a broken simulation environment and still
		 * return a valid presence. Otherwise, not present.
		 */
		dtctl = in_be64(p->regs + PHB_PCIE_DLP_TRAIN_CTL);
		if (dtctl & PHB_PCIE_DLP_TL_LINKACT) {
			PHBERR(p, "Presence detect 0 but link set !\n");
			*val = OPAL_PCI_SLOT_PRESENT;
		} else {
			*val = OPAL_PCI_SLOT_EMPTY;
		}
	}

	return OPAL_SUCCESS;
}

static int64_t phb4_get_link_info(struct pci_slot *slot, uint8_t *speed,
				   uint8_t *width)
{
	struct phb4 *p = phb_to_phb4(slot->phb);
	uint64_t reg;
	uint16_t state;
	int64_t rc;
	uint8_t s;

	/* Link is up, let's find the actual speed */
	reg = in_be64(p->regs + PHB_PCIE_DLP_TRAIN_CTL);
	if (!(reg & PHB_PCIE_DLP_TL_LINKACT)) {
		*width = 0;
		if (speed)
			*speed = 0;
		return OPAL_SUCCESS;
	}

	rc = phb4_pcicfg_read16(&p->phb, 0,
				p->ecap + PCICAP_EXP_LSTAT, &state);
	if (rc != OPAL_SUCCESS) {
		PHBERR(p, "%s: Error %lld getting link state\n", __func__, rc);
		return OPAL_HARDWARE;
	}

	if (state & PCICAP_EXP_LSTAT_DLLL_ACT) {
		*width = ((state & PCICAP_EXP_LSTAT_WIDTH) >> 4);
		s =  state & PCICAP_EXP_LSTAT_SPEED;
	} else {
		*width = 0;
		s = 0;
	}

	if (speed)
		*speed = s;

	return OPAL_SUCCESS;
}

static int64_t phb4_get_link_state(struct pci_slot *slot, uint8_t *val)
{
	return phb4_get_link_info(slot, NULL, val);
}

static int64_t phb4_retry_state(struct pci_slot *slot)
{
	struct phb4 *p = phb_to_phb4(slot->phb);

	/* Mark link as down */
	phb4_prepare_link_change(slot, false);

	if (!slot->link_retries--) {
		switch (slot->state) {
		case PHB4_SLOT_LINK_WAIT_ELECTRICAL:
			PHBERR(p, "Presence detected but no electrical link\n");
			break;
		case PHB4_SLOT_LINK_WAIT:
			PHBERR(p, "Electrical link detected but won't train\n");
			break;
		case PHB4_SLOT_LINK_STABLE:
			PHBERR(p, "Linked trained but was degraded or unstable\n");
			break;
		default:
			PHBERR(p, "Unknown link issue\n");
		}
		return OPAL_HARDWARE;
	}

	pci_slot_set_state(slot, PHB4_SLOT_CRESET_START);
	return pci_slot_set_sm_timeout(slot, msecs_to_tb(1));
}

static void phb4_train_info(struct phb4 *p, uint64_t reg, unsigned long time)
{
	char s[80];

	snprintf(s, sizeof(s), "TRACE:0x%016llx % 2lims",
		 reg, tb_to_msecs(time));

	if (reg & PHB_PCIE_DLP_TL_LINKACT)
		snprintf(s, sizeof(s), "%s trained ", s);
	else if (reg & PHB_PCIE_DLP_TRAINING)
		snprintf(s, sizeof(s), "%s training", s);
	else if (reg & PHB_PCIE_DLP_INBAND_PRESENCE)
		snprintf(s, sizeof(s), "%s presence", s);
	else
		snprintf(s, sizeof(s), "%s         ", s);

	snprintf(s, sizeof(s), "%s GEN%lli:x%02lli:", s,
		 GETFIELD(PHB_PCIE_DLP_LINK_SPEED, reg),
		 GETFIELD(PHB_PCIE_DLP_LINK_WIDTH, reg));

	switch (GETFIELD(PHB_PCIE_DLP_LTSSM_TRC, reg)) {
	case PHB_PCIE_DLP_LTSSM_RESET:
		snprintf(s, sizeof(s), "%sreset", s);
		break;
	case PHB_PCIE_DLP_LTSSM_DETECT:
		snprintf(s, sizeof(s), "%sdetect", s);
		break;
	case PHB_PCIE_DLP_LTSSM_POLLING:
		snprintf(s, sizeof(s), "%spolling", s);
		break;
	case PHB_PCIE_DLP_LTSSM_CONFIG:
		snprintf(s, sizeof(s), "%sconfig", s);
		break;
	case PHB_PCIE_DLP_LTSSM_L0:
		snprintf(s, sizeof(s), "%sL0", s);
		break;
	case PHB_PCIE_DLP_LTSSM_REC:
		snprintf(s, sizeof(s), "%srecovery", s);
		break;
	case PHB_PCIE_DLP_LTSSM_L1:
		snprintf(s, sizeof(s), "%sL1", s);
		break;
	case PHB_PCIE_DLP_LTSSM_L2:
		snprintf(s, sizeof(s), "%sL2", s);
		break;
	case PHB_PCIE_DLP_LTSSM_HOTRESET:
		snprintf(s, sizeof(s), "%shotreset", s);
		break;
	default:
		snprintf(s, sizeof(s), "%sunvalid", s);
	}
	PHBERR(p, "%s\n", s);
}

static void phb4_dump_capp_err_regs(struct phb4 *p)
{
	uint64_t fir, apc_master_err, snoop_err, transport_err;
	uint64_t tlbi_err, capp_err_status;
	uint64_t offset = PHB4_CAPP_REG_OFFSET(p);

	xscom_read(p->chip_id, CAPP_FIR + offset, &fir);
	xscom_read(p->chip_id, CAPP_APC_MASTER_ERR_RPT + offset,
		   &apc_master_err);
	xscom_read(p->chip_id, CAPP_SNOOP_ERR_RTP + offset, &snoop_err);
	xscom_read(p->chip_id, CAPP_TRANSPORT_ERR_RPT + offset, &transport_err);
	xscom_read(p->chip_id, CAPP_TLBI_ERR_RPT + offset, &tlbi_err);
	xscom_read(p->chip_id, CAPP_ERR_STATUS_CTRL + offset, &capp_err_status);

	PHBERR(p, "           CAPP FIR=%016llx\n", fir);
	PHBERR(p, "CAPP APC MASTER ERR=%016llx\n", apc_master_err);
	PHBERR(p, "     CAPP SNOOP ERR=%016llx\n", snoop_err);
	PHBERR(p, " CAPP TRANSPORT ERR=%016llx\n", transport_err);
	PHBERR(p, "      CAPP TLBI ERR=%016llx\n", tlbi_err);
	PHBERR(p, "    CAPP ERR STATUS=%016llx\n", capp_err_status);
}

/* Check if AIB is fenced via PBCQ NFIR */
static bool phb4_fenced(struct phb4 *p)
{
	uint64_t nfir_p, nfir_n, err_aib;
	uint64_t err_rpt0, err_rpt1;

	/* Already fenced ? */
	if (p->flags & PHB4_AIB_FENCED)
		return true;

	/*
	 * An all 1's from the PHB indicates a PHB freeze/fence. We
	 * don't really differenciate them at this point.
	 */
	if (in_be64(p->regs + PHB_CPU_LOADSTORE_STATUS)!= 0xfffffffffffffffful)
		return false;

	PHBERR(p, "PHB Freeze/Fence detected !\n");

	/* We read the PCI and NEST FIRs and dump them */
	xscom_read(p->chip_id,
		   p->pci_stk_xscom + XPEC_PCI_STK_PCI_FIR, &nfir_p);
	xscom_read(p->chip_id,
		   p->pe_stk_xscom + XPEC_NEST_STK_PCI_NFIR, &nfir_n);
	xscom_read(p->chip_id,
		   p->pe_stk_xscom + XPEC_NEST_STK_ERR_RPT0, &err_rpt0);
	xscom_read(p->chip_id,
		   p->pe_stk_xscom + XPEC_NEST_STK_ERR_RPT1, &err_rpt1);
	xscom_read(p->chip_id,
		   p->pci_stk_xscom + XPEC_PCI_STK_PBAIB_ERR_REPORT, &err_aib);

	PHBERR(p, "            PCI FIR=%016llx\n", nfir_p);
	PHBERR(p, "           NEST FIR=%016llx\n", nfir_n);
	PHBERR(p, "           ERR RPT0=%016llx\n", err_rpt0);
	PHBERR(p, "           ERR RPT1=%016llx\n", err_rpt1);
	PHBERR(p, "            AIB ERR=%016llx\n", err_aib);

	/* Mark ourselves fenced */
	p->flags |= PHB4_AIB_FENCED;

	/* dump capp error registers in case phb was fenced due to capp */
	if (nfir_n & XPEC_NEST_STK_PCI_NFIR_CXA_PE_CAPP)
		phb4_dump_capp_err_regs(p);

	phb4_eeh_dump_regs(p);

	return true;
}

static bool phb4_check_reg(struct phb4 *p, uint64_t reg)
{
	if (reg == 0xffffffffffffffffUL)
		return !phb4_fenced(p);
	return true;
}

static void phb4_get_info(struct phb *phb, uint16_t bdfn, uint8_t *speed,
			  uint8_t *width)
{
	int32_t ecap;
	uint32_t cap;

	ecap = pci_find_cap(phb, bdfn, PCI_CFG_CAP_ID_EXP);
	pci_cfg_read32(phb, bdfn, ecap + PCICAP_EXP_LCAP, &cap);
	*width = (cap & PCICAP_EXP_LCAP_MAXWDTH) >> 4;
	*speed = cap & PCICAP_EXP_LCAP_MAXSPD;
}

#define PVR_POWER9_CUMULUS		0x00002000

static bool phb4_chip_retry_workaround(void)
{
	unsigned int pvr;

	if (pci_retry_all)
		return true;

	/* Chips that need this retry are:
	 *  - CUMULUS DD1.0
	 *  - NIMBUS DD2.0 and below
	 */
	pvr = mfspr(SPR_PVR);
	if (pvr & PVR_POWER9_CUMULUS) {
		if ((PVR_VERS_MAJ(pvr) == 1) && (PVR_VERS_MIN(pvr) == 0))
			return true;
	} else { /* NIMBUS */
		if (PVR_VERS_MAJ(pvr) == 1)
			return true;
		if ((PVR_VERS_MAJ(pvr) == 2) && (PVR_VERS_MIN(pvr) == 0))
			return true;
	}
	return false;
}

struct pci_card_id {
	uint16_t vendor;
	uint16_t device;
};

static struct pci_card_id retry_whitelist[] = {
	{ 0x1000, 0x005d }, /* LSI Logic MegaRAID SAS-3 3108 */
	{ 0x1000, 0x00c9 }, /* LSI MPT SAS-3 */
	{ 0x104c, 0x8241 }, /* TI xHCI USB */
	{ 0x1077, 0x2261 }, /* QLogic ISP2722-based 16/32Gb FC */
	{ 0x10b5, 0x8725 }, /* PLX Switch: p9dsu, witherspoon */
	{ 0x10b5, 0x8748 }, /* PLX Switch: ZZ */
	{ 0x11f8, 0xf117 }, /* PMC-Sierra/MicroSemi NV1604 */
	{ 0x15b3, 0x1013 }, /* Mellanox ConnectX-4 */
	{ 0x15b3, 0x1017 }, /* Mellanox ConnectX-5 */
	{ 0x15b3, 0x1019 }, /* Mellanox ConnectX-5 Ex */
	{ 0x1a03, 0x1150 }, /* ASPEED AST2500 Switch */
	{ 0x8086, 0x10fb }, /* Intel x520 10G Eth */
	{ 0x9005, 0x028d }, /* MicroSemi PM8069 */
};

#define VENDOR(vdid) ((vdid) & 0xffff)
#define DEVICE(vdid) (((vdid) >> 16) & 0xffff)

static bool phb4_adapter_in_whitelist(uint32_t vdid)
{
	int i;

	if (pci_retry_all)
		return true;

	for (i = 0; i < ARRAY_SIZE(retry_whitelist); i++)
		if ((retry_whitelist[i].vendor == VENDOR(vdid)) &&
		    (retry_whitelist[i].device == DEVICE(vdid)))
			return true;

	return false;
}

static struct pci_card_id lane_eq_disable[] = {
	{ 0x10de, 0x17fd }, /* Nvidia GM200GL [Tesla M40] */
	{ 0x10de, 0x1db4 }, /* Nvidia GV100 */
};

static bool phb4_lane_eq_retry_whitelist(uint32_t vdid)
{
	int i;

	for (i = 0; i < ARRAY_SIZE(lane_eq_disable); i++)
		if ((lane_eq_disable[i].vendor == VENDOR(vdid)) &&
		    (lane_eq_disable[i].device == DEVICE(vdid)))
			return true;
	return false;
}

static void phb4_lane_eq_change(struct phb4 *p, uint32_t vdid)
{
	p->lane_eq_en = !phb4_lane_eq_retry_whitelist(vdid);
}

#define min(x,y) ((x) < (y) ? x : y)

static bool phb4_link_optimal(struct pci_slot *slot, uint32_t *vdid)
{
	struct phb4 *p = phb_to_phb4(slot->phb);
	uint32_t id;
	uint16_t bdfn;
	uint8_t trained_speed, phb_speed, dev_speed, target_speed;
	uint8_t trained_width, phb_width, dev_width, target_width;
	bool optimal_speed, optimal_width, optimal, retry_enabled;


	/* Current trained state */
	phb4_get_link_info(slot, &trained_speed, &trained_width);

	/* Get PHB capability */
	/* NOTE: phb_speed will account for the software speed limit */
	phb4_get_info(slot->phb, 0, &phb_speed, &phb_width);

	/* Get device capability */
	bdfn = 0x0100; /* bus=1 dev=0 device=0 */
	/* Since this is the first access, we need to wait for CRS */
	if (!pci_wait_crs(slot->phb, bdfn , &id))
		return true;
	phb4_get_info(slot->phb, bdfn, &dev_speed, &dev_width);

	/* Work out if we are optimally trained */
	target_speed = min(phb_speed, dev_speed);
	optimal_speed = (trained_speed >= target_speed);
	target_width = min(phb_width, dev_width);
	optimal_width = (trained_width >= target_width);
	optimal = optimal_width && optimal_speed;
	retry_enabled = (phb4_chip_retry_workaround() &&
			 phb4_adapter_in_whitelist(id)) ||
		phb4_lane_eq_retry_whitelist(id);

	PHBDBG(p, "LINK: Card [%04x:%04x] %s Retry:%s\n", VENDOR(id),
	       DEVICE(id), optimal ? "Optimal" : "Degraded",
	       retry_enabled ? "enabled" : "disabled");
	PHBDBG(p, "LINK: Speed Train:GEN%i PHB:GEN%i DEV:GEN%i%s\n",
	       trained_speed, phb_speed, dev_speed, optimal_speed ? "" : " *");
	PHBDBG(p, "LINK: Width Train:x%02i PHB:x%02i DEV:x%02i%s\n",
	       trained_width, phb_width, dev_width, optimal_width ? "" : " *");

	if (vdid)
		*vdid = id;

	if (!retry_enabled)
		return true;

	return optimal;
}

/*
 * This is a trace function to watch what's happening duing pcie link
 * training.  If any errors are detected it simply returns so the
 * normal code can deal with it.
 */
static void phb4_training_trace(struct phb4 *p)
{
	uint64_t reg, reglast = -1;
	unsigned long now, start = mftb();

	if (!pci_tracing)
		return;

	while(1) {
		now = mftb();
		reg = in_be64(p->regs + PHB_PCIE_DLP_TRAIN_CTL);
		if (reg != reglast)
			phb4_train_info(p, reg, now - start);
		reglast = reg;

		if (!phb4_check_reg(p, reg)) {
			PHBERR(p, "TRACE: PHB fence waiting link.\n");
			break;
		}
		if (reg & PHB_PCIE_DLP_TL_LINKACT) {
			PHBERR(p, "TRACE: Link trained.\n");
			break;
		}
		if ((now - start) > secs_to_tb(3)) {
			PHBERR(p, "TRACE: Timeout waiting for link up.\n");
			break;
		}
	}
}

static int64_t phb4_poll_link(struct pci_slot *slot)
{
	struct phb4 *p = phb_to_phb4(slot->phb);
	uint64_t reg;
	uint32_t vdid;

	switch (slot->state) {
	case PHB4_SLOT_NORMAL:
	case PHB4_SLOT_LINK_START:
		PHBDBG(p, "LINK: Start polling\n");
		slot->retries = PHB4_LINK_ELECTRICAL_RETRIES;
		pci_slot_set_state(slot, PHB4_SLOT_LINK_WAIT_ELECTRICAL);
		/* Polling early here has no chance of a false positive */
		return pci_slot_set_sm_timeout(slot, msecs_to_tb(1));
	case PHB4_SLOT_LINK_WAIT_ELECTRICAL:
		/*
		 * Wait for the link electrical connection to be
		 * established (shorter timeout). This allows us to
		 * workaround spurrious presence detect on some machines
		 * without waiting 10s each time
		 *
		 * Note: We *also* check for the full link up bit here
		 * because simics doesn't seem to implement the electrical
		 * link bit at all
		 */
		reg = in_be64(p->regs + PHB_PCIE_DLP_TRAIN_CTL);
		if (!phb4_check_reg(p, reg)) {
			PHBERR(p, "PHB fence waiting for electrical link\n");
			return phb4_retry_state(slot);
		}

		if (reg & (PHB_PCIE_DLP_INBAND_PRESENCE |
			   PHB_PCIE_DLP_TL_LINKACT)) {
			PHBDBG(p, "LINK: Electrical link detected\n");
			pci_slot_set_state(slot, PHB4_SLOT_LINK_WAIT);
			slot->retries = PHB4_LINK_WAIT_RETRIES;
			/* No wait here since already have an elec link */
			return pci_slot_set_sm_timeout(slot, msecs_to_tb(1));
		}

		if (slot->retries-- == 0) {
			PHBDBG(p, "LINK: No in-band presence\n");
			return OPAL_SUCCESS;
		}
		/* Retry */
		return pci_slot_set_sm_timeout(slot, msecs_to_tb(10));
	case PHB4_SLOT_LINK_WAIT:
		reg = in_be64(p->regs + PHB_PCIE_DLP_TRAIN_CTL);
		if (!phb4_check_reg(p, reg)) {
			PHBERR(p, "LINK: PHB fence waiting for link training\n");
			return phb4_retry_state(slot);
		}
		if (reg & PHB_PCIE_DLP_TL_LINKACT) {
			PHBDBG(p, "LINK: Link is up\n");
			phb4_prepare_link_change(slot, true);
			pci_slot_set_state(slot, PHB4_SLOT_LINK_STABLE);
			return pci_slot_set_sm_timeout(slot, secs_to_tb(1));
		}

		if (slot->retries-- == 0) {
			PHBERR(p, "LINK: Timeout waiting for link up\n");
			PHBDBG(p, "LINK: DLP train control: 0x%016llx\n", reg);
			return phb4_retry_state(slot);
		}
		/* Retry */
		return pci_slot_set_sm_timeout(slot, msecs_to_tb(10));
	case PHB4_SLOT_LINK_STABLE:
		/* Sanity check link */
		if (phb4_fenced(p)) {
			PHBERR(p, "LINK: PHB fenced waiting for stabilty\n");
			return phb4_retry_state(slot);
		}
		reg = in_be64(p->regs + PHB_PCIE_DLP_TRAIN_CTL);
		if (!phb4_check_reg(p, reg)) {
			PHBERR(p, "LINK: PHB fence reading training control\n");
			return phb4_retry_state(slot);
		}
		if (reg & PHB_PCIE_DLP_TL_LINKACT) {
			PHBDBG(p, "LINK: Link is stable\n");
			if (!phb4_link_optimal(slot, &vdid)) {
				PHBDBG(p, "LINK: Link degraded\n");
				if (slot->link_retries) {
					phb4_lane_eq_change(p, vdid);
					return phb4_retry_state(slot);
				}
				/*
				 * Link is degraded but no more retries, so
				 * settle for what we have :-(
				 */
				PHBERR(p, "LINK: Degraded but no more retries\n");
			}
			pci_slot_set_state(slot, PHB4_SLOT_NORMAL);
			return OPAL_SUCCESS;
		}
		PHBERR(p, "LINK: Went down waiting for stabilty\n");
		PHBDBG(p, "LINK: DLP train control: 0x%016llx\n", reg);
		return phb4_retry_state(slot);
	default:
		PHBERR(p, "LINK: Unexpected slot state %08x\n",
		       slot->state);
	}

	pci_slot_set_state(slot, PHB4_SLOT_NORMAL);
	return OPAL_HARDWARE;
}

static int64_t phb4_hreset(struct pci_slot *slot)
{
	struct phb4 *p = phb_to_phb4(slot->phb);
	uint16_t brctl;
	uint8_t presence = 1;

	switch (slot->state) {
	case PHB4_SLOT_NORMAL:
		PHBDBG(p, "HRESET: Starts\n");
		if (slot->ops.get_presence_state)
			slot->ops.get_presence_state(slot, &presence);
		if (!presence) {
			PHBDBG(p, "HRESET: No device\n");
			return OPAL_SUCCESS;
		}

		PHBDBG(p, "HRESET: Prepare for link down\n");
		phb4_prepare_link_change(slot, false);
		/* fall through */
	case PHB4_SLOT_HRESET_START:
		PHBDBG(p, "HRESET: Assert\n");

		phb4_pcicfg_read16(&p->phb, 0, PCI_CFG_BRCTL, &brctl);
		brctl |= PCI_CFG_BRCTL_SECONDARY_RESET;
		phb4_pcicfg_write16(&p->phb, 0, PCI_CFG_BRCTL, brctl);
		pci_slot_set_state(slot, PHB4_SLOT_HRESET_DELAY);

		return pci_slot_set_sm_timeout(slot, secs_to_tb(1));
	case PHB4_SLOT_HRESET_DELAY:
		PHBDBG(p, "HRESET: Deassert\n");

		/* Clear link errors before we deassert reset */
		phb4_err_clear_regb(p);

		phb4_pcicfg_read16(&p->phb, 0, PCI_CFG_BRCTL, &brctl);
		brctl &= ~PCI_CFG_BRCTL_SECONDARY_RESET;
		phb4_pcicfg_write16(&p->phb, 0, PCI_CFG_BRCTL, brctl);

		/*
		 * Due to some oddball adapters bouncing the link
		 * training a couple of times, we wait for a full second
		 * before we start checking the link status, otherwise
		 * we can get a spurrious link down interrupt which
		 * causes us to EEH immediately.
		 */
		pci_slot_set_state(slot, PHB4_SLOT_HRESET_DELAY2);
		return pci_slot_set_sm_timeout(slot, secs_to_tb(1));
	case PHB4_SLOT_HRESET_DELAY2:
		pci_slot_set_state(slot, PHB4_SLOT_LINK_START);
		return slot->ops.poll_link(slot);
	default:
		PHBERR(p, "Unexpected slot state %08x\n", slot->state);
	}

	pci_slot_set_state(slot, PHB4_SLOT_NORMAL);
	return OPAL_HARDWARE;
}

static int64_t phb4_freset(struct pci_slot *slot)
{
	struct phb4 *p = phb_to_phb4(slot->phb);
	uint8_t presence = 1;
	uint64_t reg;

	switch(slot->state) {
	case PHB4_SLOT_NORMAL:
		PHBDBG(p, "FRESET: Starts\n");

		/* Nothing to do without adapter connected */
		if (slot->ops.get_presence_state)
			slot->ops.get_presence_state(slot, &presence);
		if (!presence) {
			PHBDBG(p, "FRESET: No device\n");
			return OPAL_SUCCESS;
		}

		PHBDBG(p, "FRESET: Prepare for link down\n");

		phb4_prepare_link_change(slot, false);
		/* fall through */
	case PHB4_SLOT_FRESET_START:
		if (!p->skip_perst) {
			PHBDBG(p, "FRESET: Assert\n");
			reg = in_be64(p->regs + PHB_PCIE_CRESET);
			reg &= ~PHB_PCIE_CRESET_PERST_N;
			out_be64(p->regs + PHB_PCIE_CRESET, reg);
			pci_slot_set_state(slot,
				PHB4_SLOT_FRESET_ASSERT_DELAY);
			/* 250ms assert time aligns with powernv */
			return pci_slot_set_sm_timeout(slot, msecs_to_tb(250));
		}

		/* To skip the assert during boot time */
		PHBDBG(p, "FRESET: Assert skipped\n");
		pci_slot_set_state(slot, PHB4_SLOT_FRESET_ASSERT_DELAY);
		p->skip_perst = false;
		/* fall through */
	case PHB4_SLOT_FRESET_ASSERT_DELAY:
		/* Clear link errors before we deassert PERST */
		phb4_err_clear_regb(p);

		if (pci_tracing) {
			/* Enable tracing */
			reg = in_be64(p->regs + PHB_PCIE_DLP_TRWCTL);
			out_be64(p->regs + PHB_PCIE_DLP_TRWCTL,
				 reg | PHB_PCIE_DLP_TRWCTL_EN);
		}

		PHBDBG(p, "FRESET: Deassert\n");
		reg = in_be64(p->regs + PHB_PCIE_CRESET);
		reg |= PHB_PCIE_CRESET_PERST_N;
		out_be64(p->regs + PHB_PCIE_CRESET, reg);
		pci_slot_set_state(slot,
			PHB4_SLOT_FRESET_DEASSERT_DELAY);

		phb4_training_trace(p);

		/* Move on to link poll right away */
		return pci_slot_set_sm_timeout(slot, msecs_to_tb(1));
	case PHB4_SLOT_FRESET_DEASSERT_DELAY:
		pci_slot_set_state(slot, PHB4_SLOT_LINK_START);
		return slot->ops.poll_link(slot);
	default:
		PHBERR(p, "Unexpected slot state %08x\n", slot->state);
	}

	pci_slot_set_state(slot, PHB4_SLOT_NORMAL);
	return OPAL_HARDWARE;
}

static int64_t load_capp_ucode(struct phb4 *p)
{
	int64_t rc;

	if (p->index != CAPP0_PHB_INDEX && p->index != CAPP1_PHB_INDEX)
		return OPAL_HARDWARE;

	/* 0x4341505050534C4C = 'CAPPPSLL' in ASCII */
	rc = capp_load_ucode(p->chip_id, p->phb.opal_id, p->index,
			0x4341505050534C4C, PHB4_CAPP_REG_OFFSET(p),
			CAPP_APC_MASTER_ARRAY_ADDR_REG,
			CAPP_APC_MASTER_ARRAY_WRITE_REG,
			CAPP_SNP_ARRAY_ADDR_REG,
			CAPP_SNP_ARRAY_WRITE_REG);
	return rc;
}

static int do_capp_recovery_scoms(struct phb4 *p)
{
	uint64_t rc, reg, end;
	uint64_t offset = PHB4_CAPP_REG_OFFSET(p);


	/* Get the status of CAPP recovery */
	xscom_read(p->chip_id, CAPP_ERR_STATUS_CTRL + offset, &reg);

	/* No recovery in progress ignore */
	if ((reg & PPC_BIT(0)) == 0) {
		PHBDBG(p, "CAPP: No recovery in progress\n");
		return OPAL_SUCCESS;
	}

	PHBDBG(p, "CAPP: Waiting for recovery to complete\n");
	/* recovery timer failure period 168ms */
	end = mftb() + msecs_to_tb(168);
	while ((reg & (PPC_BIT(1) | PPC_BIT(5) | PPC_BIT(9))) == 0) {

		time_wait_ms(5);
		xscom_read(p->chip_id, CAPP_ERR_STATUS_CTRL + offset, &reg);

		if (end && tb_compare(mftb(), end) != TB_AAFTERB) {
			PHBERR(p, "CAPP: Capp recovery Timed-out.\n");
			end = 0;
			break;
		}
	}

	/* Check if the recovery failed or passed */
	if (reg & PPC_BIT(1)) {
		PHBDBG(p, "Doing CAPP recovery scoms\n");
		/* disable snoops */
		xscom_write(p->chip_id, SNOOP_CAPI_CONFIG + offset, 0);
		load_capp_ucode(p);

		/* clear err rpt reg*/
		xscom_write(p->chip_id, CAPP_ERR_RPT_CLR + offset, 0);

		/* clear capp fir */
		xscom_write(p->chip_id, CAPP_FIR + offset, 0);

		/* Just reset Bit-0,1 and dont touch any other bit */
		xscom_read(p->chip_id, CAPP_ERR_STATUS_CTRL + offset, &reg);
		reg &= ~(PPC_BIT(0) | PPC_BIT(1));
		xscom_write(p->chip_id, CAPP_ERR_STATUS_CTRL + offset, reg);

		PHBDBG(p, "CAPP recovery complete\n");
		rc = OPAL_SUCCESS;

	} else {
		/* Most likely will checkstop here due to FIR ACTION for
		 * failed recovery. So this message would never be logged.
		 * But if we still enter here then return an error forcing a
		 * fence of the PHB.
		 */
		if (reg  & PPC_BIT(5))
			PHBERR(p, "CAPP: Capp recovery Failed\n");
		else if (reg  & PPC_BIT(9))
			PHBERR(p, "CAPP: Capp recovery hang detected\n");
		else if (end != 0)
			PHBERR(p, "CAPP: Unknown recovery failure\n");

		PHBDBG(p, "CAPP: Err/Status-reg=0x%016llx\n", reg);
		rc = OPAL_HARDWARE;
	}

	return rc;
}

static int64_t phb4_creset(struct pci_slot *slot)
{
	struct phb4 *p = phb_to_phb4(slot->phb);
	uint64_t pbcq_status, reg;

	/* Don't even try fixing a broken PHB */
	if (p->broken)
		return OPAL_HARDWARE;

	switch (slot->state) {
	case PHB4_SLOT_NORMAL:
	case PHB4_SLOT_CRESET_START:
		PHBDBG(p, "CRESET: Starts\n");

		phb4_prepare_link_change(slot, false);
		/* Clear error inject register, preventing recursive errors */
		xscom_write(p->chip_id, p->pe_xscom + 0x2, 0x0);

		/* Force fence on the PHB to work around a non-existent PE */
		if (!phb4_fenced(p))
			xscom_write(p->chip_id, p->pe_stk_xscom + 0x2,
				    0x0000002000000000);

		/*
		 * Force use of ASB for register access until the PHB has
		 * been fully reset.
		 */
		p->flags |= PHB4_CFG_USE_ASB | PHB4_AIB_FENCED;

		/* Assert PREST before clearing errors */
		reg = phb4_read_reg(p, PHB_PCIE_CRESET);
		reg &= ~PHB_PCIE_CRESET_PERST_N;
		phb4_write_reg(p, PHB_PCIE_CRESET, reg);

		/* Clear errors, following the proper sequence */
		phb4_err_clear(p);

		/* Actual reset */
		xscom_write(p->chip_id, p->pci_stk_xscom + XPEC_PCI_STK_ETU_RESET,
			    0x8000000000000000);

		/* Read errors in PFIR and NFIR */
		xscom_read(p->chip_id, p->pci_stk_xscom + 0x0, &p->pfir_cache);
		xscom_read(p->chip_id, p->pe_stk_xscom + 0x0, &p->nfir_cache);

		pci_slot_set_state(slot, PHB4_SLOT_CRESET_WAIT_CQ);
		slot->retries = 500;
		return pci_slot_set_sm_timeout(slot, msecs_to_tb(10));
	case PHB4_SLOT_CRESET_WAIT_CQ:

		// Wait until operations are complete
		xscom_read(p->chip_id, p->pe_stk_xscom + 0xc, &pbcq_status);
		if (!(pbcq_status & 0xC000000000000000)) {
			PHBDBG(p, "CRESET: No pending transactions\n");

			/* capp recovery */
			if ((p->flags & PHB4_CAPP_RECOVERY) &&
			    (do_capp_recovery_scoms(p) != OPAL_SUCCESS))
				goto error;

			/* Clear errors in PFIR and NFIR */
			xscom_write(p->chip_id, p->pci_stk_xscom + 0x1,
				    ~p->pfir_cache);
			xscom_write(p->chip_id, p->pe_stk_xscom + 0x1,
				    ~p->nfir_cache);

			/* Clear PHB from reset */
			xscom_write(p->chip_id,
				    p->pci_stk_xscom + XPEC_PCI_STK_ETU_RESET, 0x0);

			/* DD1 errata: write to PEST to force update */
			phb4_ioda_sel(p, IODA3_TBL_PESTA, PHB4_RESERVED_PE_NUM(p),
				      false);
			phb4_write_reg(p, PHB_IODA_DATA0, 0);

			pci_slot_set_state(slot, PHB4_SLOT_CRESET_REINIT);
			/* After lifting PHB reset, wait while logic settles */
			return pci_slot_set_sm_timeout(slot, msecs_to_tb(10));
		}

		if (slot->retries-- == 0) {
			PHBERR(p, "Timeout waiting for pending transaction\n");
			goto error;
		}
		return pci_slot_set_sm_timeout(slot, msecs_to_tb(100));
	case PHB4_SLOT_CRESET_REINIT:
		PHBDBG(p, "CRESET: Reinitialization\n");
		p->flags &= ~PHB4_AIB_FENCED;
		p->flags &= ~PHB4_CAPP_RECOVERY;
		p->flags &= ~PHB4_CFG_USE_ASB;
		phb4_init_hw(p, false);
		pci_slot_set_state(slot, PHB4_SLOT_CRESET_FRESET);
		return pci_slot_set_sm_timeout(slot, msecs_to_tb(100));
	case PHB4_SLOT_CRESET_FRESET:
		pci_slot_set_state(slot, PHB4_SLOT_NORMAL);
		return slot->ops.freset(slot);
	default:
		PHBERR(p, "CRESET: Unexpected slot state %08x, resetting...\n",
		       slot->state);
		pci_slot_set_state(slot, PHB4_SLOT_NORMAL);
		return slot->ops.creset(slot);

	}

error:
	/* Mark the PHB as dead and expect it to be removed */
	p->broken = true;
	return OPAL_HARDWARE;
}

/*
 * Initialize root complex slot, which is mainly used to
 * do fundamental reset before PCI enumeration in PCI core.
 * When probing root complex and building its real slot,
 * the operations will be copied over.
 */
static struct pci_slot *phb4_slot_create(struct phb *phb)
{
	struct pci_slot *slot;

	slot = pci_slot_alloc(phb, NULL);
	if (!slot)
		return slot;

	/* Elementary functions */
	slot->ops.get_presence_state  = phb4_get_presence_state;
	slot->ops.get_link_state      = phb4_get_link_state;
	slot->ops.get_power_state     = NULL;
	slot->ops.get_attention_state = NULL;
	slot->ops.get_latch_state     = NULL;
	slot->ops.set_power_state     = NULL;
	slot->ops.set_attention_state = NULL;

	/*
	 * For PHB slots, we have to split the fundamental reset
	 * into 2 steps. We might not have the first step which
	 * is to power off/on the slot, or it's controlled by
	 * individual platforms.
	 */
	slot->ops.prepare_link_change	= phb4_prepare_link_change;
	slot->ops.poll_link		= phb4_poll_link;
	slot->ops.hreset		= phb4_hreset;
	slot->ops.freset		= phb4_freset;
	slot->ops.creset		= phb4_creset;
	slot->link_retries		= PHB4_LINK_LINK_RETRIES;

	return slot;
}

static uint64_t phb4_get_pesta(struct phb4 *p, uint64_t pe_number)
{
	uint64_t pesta, *pPEST;

	pPEST = (uint64_t *)p->tbl_pest;

	phb4_ioda_sel(p, IODA3_TBL_PESTA, pe_number, false);
	pesta = phb4_read_reg(p, PHB_IODA_DATA0);
	if (pesta & IODA3_PESTA_MMIO_FROZEN)
		pesta |= pPEST[2*pe_number];

	return pesta;
}

/* Check if the chip requires escalating a freeze to fence on MMIO loads */
static bool phb4_escalation_required(void)
{
	uint64_t pvr = mfspr(SPR_PVR);

	/*
	 * Escalation is required on the following chip versions:
	 * - Cumulus DD1.0
	 * - Nimbus DD1, DD2.0, DD2.1
	 */
	if (pvr & PVR_POWER9_CUMULUS) {
		if (PVR_VERS_MAJ(pvr) == 1 && PVR_VERS_MIN(pvr) == 0)
			return true;
	} else { /* Nimbus */
		if (PVR_VERS_MAJ(pvr) == 1)
			return true;
		if (PVR_VERS_MAJ(pvr) == 2 && PVR_VERS_MIN(pvr) < 2)
			return true;
	}

	return false;
}

static bool phb4_freeze_escalate(uint64_t pesta)
{
	if ((GETFIELD(IODA3_PESTA_TRANS_TYPE, pesta) ==
	     IODA3_PESTA_TRANS_TYPE_MMIOLOAD) &&
	    (pesta & (IODA3_PESTA_CA_CMPLT_TMT | IODA3_PESTA_UR)))
		return true;
	return false;
}

static int64_t phb4_eeh_freeze_status(struct phb *phb, uint64_t pe_number,
				      uint8_t *freeze_state,
				      uint16_t *pci_error_type,
				      uint16_t *severity,
				      uint64_t *phb_status)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t peev_bit = PPC_BIT(pe_number & 0x3f);
	uint64_t peev, pesta, pestb;

	/* Defaults: not frozen */
	*freeze_state = OPAL_EEH_STOPPED_NOT_FROZEN;
	*pci_error_type = OPAL_EEH_NO_ERROR;

	/* Check dead */
	if (p->broken) {
		*freeze_state = OPAL_EEH_STOPPED_MMIO_DMA_FREEZE;
		*pci_error_type = OPAL_EEH_PHB_ERROR;
		if (severity)
			*severity = OPAL_EEH_SEV_PHB_DEAD;
		return OPAL_HARDWARE;
	}

	/* Check fence and CAPP recovery */
	if (phb4_fenced(p) || (p->flags & PHB4_CAPP_RECOVERY)) {
		*freeze_state = OPAL_EEH_STOPPED_MMIO_DMA_FREEZE;
		*pci_error_type = OPAL_EEH_PHB_ERROR;
		if (severity)
			*severity = OPAL_EEH_SEV_PHB_FENCED;
		goto bail;
	}

	/* Check the PEEV */
	phb4_ioda_sel(p, IODA3_TBL_PEEV, pe_number / 64, false);
	peev = in_be64(p->regs + PHB_IODA_DATA0);
	if (!(peev & peev_bit))
		return OPAL_SUCCESS;

	/* Indicate that we have an ER pending */
	phb4_set_err_pending(p, true);
	if (severity)
		*severity = OPAL_EEH_SEV_PE_ER;

	/* Read the full PESTA */
	pesta = phb4_get_pesta(p, pe_number);
	/* Check if we need to escalate to fence */
	if (phb4_escalation_required() && phb4_freeze_escalate(pesta)) {
		PHBERR(p, "Escalating freeze to fence PESTA[%lli]=%016llx\n",
		       pe_number, pesta);
		*severity = OPAL_EEH_SEV_PHB_FENCED;
		*pci_error_type = OPAL_EEH_PHB_ERROR;
	}

	/* Read the PESTB in the PHB */
	phb4_ioda_sel(p, IODA3_TBL_PESTB, pe_number, false);
	pestb = phb4_read_reg(p, PHB_IODA_DATA0);

	/* Convert PESTA/B to freeze_state */
	if (pesta & IODA3_PESTA_MMIO_FROZEN)
		*freeze_state |= OPAL_EEH_STOPPED_MMIO_FREEZE;
	if (pestb & IODA3_PESTB_DMA_STOPPED)
		*freeze_state |= OPAL_EEH_STOPPED_DMA_FREEZE;

bail:
	if (phb_status)
		PHBERR(p, "%s: deprecated PHB status\n", __func__);

	return OPAL_SUCCESS;
}

static int64_t phb4_eeh_freeze_clear(struct phb *phb, uint64_t pe_number,
				     uint64_t eeh_action_token)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t err, peev;
	int32_t i;
	bool frozen_pe = false;

	if (p->broken)
		return OPAL_HARDWARE;

	/* Summary. If nothing, move to clearing the PESTs which can
	 * contain a freeze state from a previous error or simply set
	 * explicitely by the user
	 */
	err = in_be64(p->regs + PHB_ETU_ERR_SUMMARY);
	if (err == 0xffffffffffffffff) {
		if (phb4_fenced(p)) {
			PHBERR(p, "eeh_freeze_clear on fenced PHB\n");
			return OPAL_HARDWARE;
		}
	}
	if (err != 0)
		phb4_err_clear(p);

	/*
	 * We have PEEV in system memory. It would give more performance
	 * to access that directly.
	 */
	if (eeh_action_token & OPAL_EEH_ACTION_CLEAR_FREEZE_MMIO) {
		phb4_ioda_sel(p, IODA3_TBL_PESTA, pe_number, false);
		out_be64(p->regs + PHB_IODA_DATA0, 0);
	}
	if (eeh_action_token & OPAL_EEH_ACTION_CLEAR_FREEZE_DMA) {
		phb4_ioda_sel(p, IODA3_TBL_PESTB, pe_number, false);
		out_be64(p->regs + PHB_IODA_DATA0, 0);
	}


	/* Update ER pending indication */
	phb4_ioda_sel(p, IODA3_TBL_PEEV, 0, true);
	for (i = 0; i < p->num_pes/64; i++) {
		peev = in_be64(p->regs + PHB_IODA_DATA0);
		if (peev) {
			frozen_pe = true;
			break;
		}
	}
	if (frozen_pe) {
		p->err.err_src	 = PHB4_ERR_SRC_PHB;
		p->err.err_class = PHB4_ERR_CLASS_ER;
		p->err.err_bit   = -1;
		phb4_set_err_pending(p, true);
	} else
		phb4_set_err_pending(p, false);

	return OPAL_SUCCESS;
}

static int64_t phb4_eeh_freeze_set(struct phb *phb, uint64_t pe_number,
				   uint64_t eeh_action_token)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t data;

	if (p->broken)
		return OPAL_HARDWARE;

	if (pe_number >= p->num_pes)
		return OPAL_PARAMETER;

	if (eeh_action_token != OPAL_EEH_ACTION_SET_FREEZE_MMIO &&
	    eeh_action_token != OPAL_EEH_ACTION_SET_FREEZE_DMA &&
	    eeh_action_token != OPAL_EEH_ACTION_SET_FREEZE_ALL)
		return OPAL_PARAMETER;

	if (eeh_action_token & OPAL_EEH_ACTION_SET_FREEZE_MMIO) {
		phb4_ioda_sel(p, IODA3_TBL_PESTA, pe_number, false);
		data = in_be64(p->regs + PHB_IODA_DATA0);
		data |= IODA3_PESTA_MMIO_FROZEN;
		out_be64(p->regs + PHB_IODA_DATA0, data);
	}

	if (eeh_action_token & OPAL_EEH_ACTION_SET_FREEZE_DMA) {
		phb4_ioda_sel(p, IODA3_TBL_PESTB, pe_number, false);
		data = in_be64(p->regs + PHB_IODA_DATA0);
		data |= IODA3_PESTB_DMA_STOPPED;
		out_be64(p->regs + PHB_IODA_DATA0, data);
	}

	return OPAL_SUCCESS;
}

static int64_t phb4_eeh_next_error(struct phb *phb,
				   uint64_t *first_frozen_pe,
				   uint16_t *pci_error_type,
				   uint16_t *severity)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t peev, pesta;
	uint32_t peev_size = p->num_pes/64;
	int32_t i, j;

	/* If the PHB is broken, we needn't go forward */
	if (p->broken) {
		*pci_error_type = OPAL_EEH_PHB_ERROR;
		*severity = OPAL_EEH_SEV_PHB_DEAD;
		return OPAL_SUCCESS;
	}

	if ((p->flags & PHB4_CAPP_RECOVERY)) {
		*pci_error_type = OPAL_EEH_PHB_ERROR;
		*severity = OPAL_EEH_SEV_PHB_FENCED;
		return OPAL_SUCCESS;
	}

	/*
	 * Check if we already have pending errors. If that's
	 * the case, then to get more information about the
	 * pending errors. Here we try PBCQ prior to PHB.
	 */
	if (phb4_err_pending(p) /*&&
	    !phb4_err_check_pbcq(p) &&
	    !phb4_err_check_lem(p) */)
		phb4_set_err_pending(p, false);

	/* Clear result */
	*pci_error_type  = OPAL_EEH_NO_ERROR;
	*severity	 = OPAL_EEH_SEV_NO_ERROR;
	*first_frozen_pe = (uint64_t)-1;

	/* Check frozen PEs */
	if (!phb4_err_pending(p)) {
		phb4_ioda_sel(p, IODA3_TBL_PEEV, 0, true);
		for (i = 0; i < peev_size; i++) {
			peev = in_be64(p->regs + PHB_IODA_DATA0);
			if (peev) {
				p->err.err_src	 = PHB4_ERR_SRC_PHB;
				p->err.err_class = PHB4_ERR_CLASS_ER;
				p->err.err_bit	 = -1;
				phb4_set_err_pending(p, true);
				break;
			}
		}
	}

	if (!phb4_err_pending(p))
		return OPAL_SUCCESS;
	/*
	 * If the frozen PE is caused by a malfunctioning TLP, we
	 * need reset the PHB. So convert ER to PHB-fatal error
	 * for the case.
	 */
	if (p->err.err_class == PHB4_ERR_CLASS_ER) {
		for (i = peev_size - 1; i >= 0; i--) {
			phb4_ioda_sel(p, IODA3_TBL_PEEV, i, false);
			peev = in_be64(p->regs + PHB_IODA_DATA0);
			for (j = 0; j < 64; j++) {
				if (peev & PPC_BIT(j)) {
					*first_frozen_pe = i * 64 + j;
					break;
				}
			}
			if (*first_frozen_pe != (uint64_t)(-1))
				break;
		}
	}

	if (*first_frozen_pe != (uint64_t)(-1)) {
		pesta = phb4_get_pesta(p, *first_frozen_pe);
		if (phb4_freeze_escalate(pesta)) {
			PHBINF(p, "Escalating freeze to fence. PESTA[%lli]=%016llx\n",
			       *first_frozen_pe, pesta);
			p->err.err_class = PHB4_ERR_CLASS_FENCED;
		}
	}

	switch (p->err.err_class) {
	case PHB4_ERR_CLASS_DEAD:
		*pci_error_type = OPAL_EEH_PHB_ERROR;
		*severity = OPAL_EEH_SEV_PHB_DEAD;
		break;
	case PHB4_ERR_CLASS_FENCED:
		*pci_error_type = OPAL_EEH_PHB_ERROR;
		*severity = OPAL_EEH_SEV_PHB_FENCED;
		break;
	case PHB4_ERR_CLASS_ER:
		*pci_error_type = OPAL_EEH_PE_ERROR;
		*severity = OPAL_EEH_SEV_PE_ER;

		/* No frozen PE ? */
		if (*first_frozen_pe == (uint64_t)-1) {
			*pci_error_type = OPAL_EEH_NO_ERROR;
			*severity = OPAL_EEH_SEV_NO_ERROR;
			phb4_set_err_pending(p, false);
		}

		break;
	case PHB4_ERR_CLASS_INF:
		*pci_error_type = OPAL_EEH_PHB_ERROR;
		*severity = OPAL_EEH_SEV_INF;
		break;
	default:
		*pci_error_type = OPAL_EEH_NO_ERROR;
		*severity = OPAL_EEH_SEV_NO_ERROR;
		phb4_set_err_pending(p, false);
	}
	return OPAL_SUCCESS;
}

static int64_t phb4_err_inject_finalize(struct phb4 *phb, uint64_t addr,
					uint64_t mask, uint64_t ctrl,
					bool is_write)
{
	if (is_write)
		ctrl |= PHB_PAPR_ERR_INJ_CTL_WR;
	else
		ctrl |= PHB_PAPR_ERR_INJ_CTL_RD;

	out_be64(phb->regs + PHB_PAPR_ERR_INJ_ADDR, addr);
	out_be64(phb->regs + PHB_PAPR_ERR_INJ_MASK, mask);
	out_be64(phb->regs + PHB_PAPR_ERR_INJ_CTL, ctrl);

	return OPAL_SUCCESS;
}

static int64_t phb4_err_inject_mem32(struct phb4 *phb, uint64_t pe_number,
				     uint64_t addr, uint64_t mask,
				     bool is_write)
{
	return OPAL_UNSUPPORTED;
}

static int64_t phb4_err_inject_mem64(struct phb4 *phb, uint64_t pe_number,
				     uint64_t addr, uint64_t mask,
				     bool is_write)
{
	return OPAL_UNSUPPORTED;
}

static int64_t phb4_err_inject_cfg(struct phb4 *phb, uint64_t pe_number,
				   uint64_t addr, uint64_t mask,
				   bool is_write)
{
	uint64_t a, m, prefer, ctrl;
	int bdfn;
	bool is_bus_pe = false;

	a = 0xffffull;
	prefer = 0xffffull;
	m = PHB_PAPR_ERR_INJ_MASK_CFG_ALL;
	ctrl = PHB_PAPR_ERR_INJ_CTL_CFG;

	for (bdfn = 0; bdfn < RTT_TABLE_ENTRIES; bdfn++) {
		if (phb->rte_cache[bdfn] != pe_number)
			continue;

		/* The PE can be associated with PCI bus or device */
		is_bus_pe = false;
		if ((bdfn + 8) < RTT_TABLE_ENTRIES &&
		    phb->rte_cache[bdfn + 8] == pe_number)
			is_bus_pe = true;

		/* Figure out the PCI config address */
		if (prefer == 0xffffull) {
			if (is_bus_pe) {
				m = PHB_PAPR_ERR_INJ_MASK_CFG;
				prefer = SETFIELD(m, 0x0ull, (bdfn >> 8));
			} else {
				m = PHB_PAPR_ERR_INJ_MASK_CFG_ALL;
				prefer = SETFIELD(m, 0x0ull, bdfn);
			}
		}

		/* Check the input address is valid or not */
		if (!is_bus_pe &&
		    GETFIELD(PHB_PAPR_ERR_INJ_MASK_CFG_ALL, addr) == bdfn) {
			a = addr;
			break;
		}

		if (is_bus_pe &&
		    GETFIELD(PHB_PAPR_ERR_INJ_MASK_CFG, addr) == (bdfn >> 8)) {
			a = addr;
			break;
		}
	}

	/* Invalid PE number */
	if (prefer == 0xffffull)
		return OPAL_PARAMETER;

	/* Specified address is out of range */
	if (a == 0xffffull)
		a = prefer;
	else
		m = mask;

	return phb4_err_inject_finalize(phb, a, m, ctrl, is_write);
}

static int64_t phb4_err_inject_dma(struct phb4 *phb, uint64_t pe_number,
				   uint64_t addr, uint64_t mask,
				   bool is_write, bool is_64bits)
{
	return OPAL_UNSUPPORTED;
}

static int64_t phb4_err_inject_dma32(struct phb4 *phb, uint64_t pe_number,
				     uint64_t addr, uint64_t mask,
				     bool is_write)
{
	return phb4_err_inject_dma(phb, pe_number, addr, mask, is_write, false);
}

static int64_t phb4_err_inject_dma64(struct phb4 *phb, uint64_t pe_number,
				     uint64_t addr, uint64_t mask,
				     bool is_write)
{
	return phb4_err_inject_dma(phb, pe_number, addr, mask, is_write, true);
}


static int64_t phb4_err_inject(struct phb *phb, uint64_t pe_number,
			       uint32_t type, uint32_t func,
			       uint64_t addr, uint64_t mask)
{
	struct phb4 *p = phb_to_phb4(phb);
	int64_t (*handler)(struct phb4 *p, uint64_t pe_number,
			   uint64_t addr, uint64_t mask, bool is_write);
	bool is_write;

	/* How could we get here without valid RTT? */
	if (!p->tbl_rtt)
		return OPAL_HARDWARE;

	/* We can't inject error to the reserved PE */
	if (pe_number == PHB4_RESERVED_PE_NUM(p) || pe_number >= p->num_pes)
		return OPAL_PARAMETER;

	/* Clear leftover from last time */
	out_be64(p->regs + PHB_PAPR_ERR_INJ_CTL, 0x0ul);

	switch (func) {
	case OPAL_ERR_INJECT_FUNC_IOA_LD_MEM_ADDR:
	case OPAL_ERR_INJECT_FUNC_IOA_LD_MEM_DATA:
		is_write = false;
		if (type == OPAL_ERR_INJECT_TYPE_IOA_BUS_ERR64)
			handler = phb4_err_inject_mem64;
		else
			handler = phb4_err_inject_mem32;
		break;
	case OPAL_ERR_INJECT_FUNC_IOA_ST_MEM_ADDR:
	case OPAL_ERR_INJECT_FUNC_IOA_ST_MEM_DATA:
		is_write = true;
		if (type == OPAL_ERR_INJECT_TYPE_IOA_BUS_ERR64)
			handler = phb4_err_inject_mem64;
		else
			handler = phb4_err_inject_mem32;
		break;
	case OPAL_ERR_INJECT_FUNC_IOA_LD_CFG_ADDR:
	case OPAL_ERR_INJECT_FUNC_IOA_LD_CFG_DATA:
		is_write = false;
		handler = phb4_err_inject_cfg;
		break;
	case OPAL_ERR_INJECT_FUNC_IOA_ST_CFG_ADDR:
	case OPAL_ERR_INJECT_FUNC_IOA_ST_CFG_DATA:
		is_write = true;
		handler = phb4_err_inject_cfg;
		break;
	case OPAL_ERR_INJECT_FUNC_IOA_DMA_RD_ADDR:
	case OPAL_ERR_INJECT_FUNC_IOA_DMA_RD_DATA:
	case OPAL_ERR_INJECT_FUNC_IOA_DMA_RD_MASTER:
	case OPAL_ERR_INJECT_FUNC_IOA_DMA_RD_TARGET:
		is_write = false;
		if (type == OPAL_ERR_INJECT_TYPE_IOA_BUS_ERR64)
			handler = phb4_err_inject_dma64;
		else
			handler = phb4_err_inject_dma32;
		break;
	case OPAL_ERR_INJECT_FUNC_IOA_DMA_WR_ADDR:
	case OPAL_ERR_INJECT_FUNC_IOA_DMA_WR_DATA:
	case OPAL_ERR_INJECT_FUNC_IOA_DMA_WR_MASTER:
	case OPAL_ERR_INJECT_FUNC_IOA_DMA_WR_TARGET:
		is_write = true;
		if (type == OPAL_ERR_INJECT_TYPE_IOA_BUS_ERR64)
			handler = phb4_err_inject_dma64;
		else
			handler = phb4_err_inject_dma32;
		break;
	default:
		return OPAL_PARAMETER;
	}

	return handler(p, pe_number, addr, mask, is_write);
}

static int64_t phb4_get_diag_data(struct phb *phb,
				  void *diag_buffer,
				  uint64_t diag_buffer_len)
{
	struct phb4 *p = phb_to_phb4(phb);
	struct OpalIoPhb4ErrorData *data = diag_buffer;

	if (diag_buffer_len < sizeof(struct OpalIoPhb4ErrorData))
		return OPAL_PARAMETER;
	if (p->broken)
		return OPAL_HARDWARE;

	/*
	 * Dummy check for fence so that phb4_read_phb_status knows
	 * whether to use ASB or AIB
	 */
	phb4_fenced(p);
	phb4_read_phb_status(p, data);

	if (!(p->flags & PHB4_AIB_FENCED))
		phb4_eeh_dump_regs(p);

	/*
	 * We're running to here probably because of errors
	 * (INF class). For that case, we need clear the error
	 * explicitly.
	 */
	if (phb4_err_pending(p) &&
	    p->err.err_class == PHB4_ERR_CLASS_INF &&
	    p->err.err_src == PHB4_ERR_SRC_PHB) {
		phb4_err_clear(p);
		phb4_set_err_pending(p, false);
	}

	return OPAL_SUCCESS;
}

static uint64_t tve_encode_50b_noxlate(uint64_t start_addr, uint64_t end_addr)
{
	uint64_t tve;

	/*
	 * Put start address bits 49:24 into TVE[52:53]||[0:23]
	 * and end address bits 49:24 into TVE[54:55]||[24:47]
	 * and set TVE[51]
	 */
	tve  = (start_addr << 16) & (0xffffffull << 40);
	tve |= (start_addr >> 38) & (3ull << 10);
	tve |= (end_addr >>  8) & (0xfffffful << 16);
	tve |= (end_addr >> 40) & (3ull << 8);
	tve |= PPC_BIT(51) | IODA3_TVT_NON_TRANSLATE_50;
	return tve;
}

static int64_t phb4_get_capp_info(int chip_id, struct phb *phb,
				  struct capp_info *info)
{
	struct phb4 *p = phb_to_phb4(phb);
	struct proc_chip *chip = get_chip(p->chip_id);
	uint32_t offset;

	if (chip_id != p->chip_id)
		return OPAL_PARAMETER;

	if (!((1 << p->index) & chip->capp_phb4_attached_mask))
		return OPAL_PARAMETER;

	offset = PHB4_CAPP_REG_OFFSET(p);

	if (p->index == CAPP0_PHB_INDEX)
		info->capp_index = 0;
	if (p->index == CAPP1_PHB_INDEX)
		info->capp_index = 1;
	info->phb_index = p->index;
	info->capp_fir_reg = CAPP_FIR + offset;
	info->capp_fir_mask_reg = CAPP_FIR_MASK + offset;
	info->capp_fir_action0_reg = CAPP_FIR_ACTION0 + offset;
	info->capp_fir_action1_reg = CAPP_FIR_ACTION1 + offset;
	info->capp_err_status_ctrl_reg = CAPP_ERR_STATUS_CTRL + offset;

	return OPAL_SUCCESS;
}

static void phb4_init_capp_regs(struct phb4 *p, uint32_t capp_eng)
{
	uint64_t reg;
	uint32_t offset;

	offset = PHB4_CAPP_REG_OFFSET(p);

	/* APC Master PowerBus Control Register */
	xscom_read(p->chip_id, APC_MASTER_PB_CTRL + offset, &reg);
	reg |= PPC_BIT(0); /* enable cResp exam */
	reg |= PPC_BIT(3); /* disable vg not sys */
	reg |= PPC_BIT(12);/* HW417025: disable capp virtual machines */
	if (p->rev == PHB4_REV_NIMBUS_DD10) {
		reg |= PPC_BIT(1);
	} else {
		reg |= PPC_BIT(2); /* disable nn rn */
		reg |= PPC_BIT(4); /* disable g */
		reg |= PPC_BIT(5); /* disable ln */
	}
	xscom_write(p->chip_id, APC_MASTER_PB_CTRL + offset, reg);

	/* Set PHB mode, HPC Dir State and P9 mode */
	xscom_write(p->chip_id, APC_MASTER_CAPI_CTRL + offset,
		    0x1772000000000000);
	PHBINF(p, "CAPP: port attached\n");

	/* Set snoop ttype decoding , dir size to 512K */
	xscom_write(p->chip_id, SNOOP_CAPI_CONFIG + offset, 0x9000000000000000);

	/* Use Read Epsilon Tier2 for all scopes.
	 * Set Tier2 Read Epsilon.
	 */
	xscom_read(p->chip_id, SNOOP_CONTROL + offset, &reg);
	reg |= PPC_BIT(0);
	reg |= PPC_BIT(35);
	reg |= PPC_BIT(45);
	reg |= PPC_BIT(46);
	reg |= PPC_BIT(47);
	reg |= PPC_BIT(50);
	xscom_write(p->chip_id, SNOOP_CONTROL + offset, reg);

	/* Transport Control Register */
	xscom_read(p->chip_id, TRANSPORT_CONTROL + offset, &reg);
	if (p->index == CAPP0_PHB_INDEX) {
		reg |= PPC_BIT(1); /* Send Packet Timer Value */
		reg |= PPC_BITMASK(10, 13); /* Send Packet Timer Value */
		reg &= ~PPC_BITMASK(14, 17); /* Set Max LPC CI store buffer to zeros */
		reg &= ~PPC_BITMASK(18, 21); /* Set Max tlbi divider */
		if (capp_eng & CAPP_MIN_STQ_ENGINES) {
			/* 2 CAPP msg engines */
			reg |= PPC_BIT(58);
			reg |= PPC_BIT(59);
			reg |= PPC_BIT(60);
		}
		if (capp_eng & CAPP_MAX_STQ_ENGINES) {
			/* 14 CAPP msg engines */
			reg |= PPC_BIT(60);
		}
		reg |= PPC_BIT(62);
	}
	if (p->index == CAPP1_PHB_INDEX) {
		reg |= PPC_BIT(4); /* Send Packet Timer Value */
		reg &= ~PPC_BIT(10); /* Set CI Store Buffer Threshold=5 */
		reg |= PPC_BIT(11);  /* Set CI Store Buffer Threshold=5 */
		reg &= ~PPC_BIT(12); /* Set CI Store Buffer Threshold=5 */
		reg |= PPC_BIT(13);  /* Set CI Store Buffer Threshold=5 */
		reg &= ~PPC_BITMASK(14, 17); /* Set Max LPC CI store buffer to zeros */
		reg &= ~PPC_BITMASK(18, 21); /* Set Max tlbi divider */
		if (capp_eng & CAPP_MIN_STQ_ENGINES) {
			/* 2 CAPP msg engines */
			reg |= PPC_BIT(59);
			reg |= PPC_BIT(60);
		}
		if (capp_eng & CAPP_MAX_STQ_ENGINES) {
			/* 6 CAPP msg engines */
			reg |= PPC_BIT(60);
		}
	}
	xscom_write(p->chip_id, TRANSPORT_CONTROL + offset, reg);

	/* Initialize CI Store Buffers */
	xscom_read(p->chip_id, TRANSPORT_CONTROL + offset, &reg);
	reg |= PPC_BIT(63);
	xscom_write(p->chip_id, TRANSPORT_CONTROL + offset, reg);

	/* Enable epoch timer */
	xscom_write(p->chip_id, EPOCH_RECOVERY_TIMERS_CTRL + offset,
		    0xC0000000FFF8FFE0);

	/* Flush SUE State Map Register */
	xscom_write(p->chip_id, FLUSH_SUE_STATE_MAP + offset,
		    0x08020A0000000000);

	if (!(p->rev == PHB4_REV_NIMBUS_DD10)) {
		/* Flush SUE uOP1 Register */
		xscom_write(p->chip_id, FLUSH_SUE_UOP1 + offset,
			    0xDCE0280428000000);
	}

	/* capp owns PHB read buffers */
	if (p->index == CAPP0_PHB_INDEX) {
		/* max PHB read buffers 0-47 */
		reg = 0xFFFFFFFFFFFF0000;
		if (capp_eng & CAPP_MAX_DMA_READ_ENGINES)
			reg = 0xFF00000000000000;
		xscom_write(p->chip_id, APC_FSM_READ_MASK + offset, reg);
		xscom_write(p->chip_id, XPT_FSM_RMM + offset, reg);
	}
	if (p->index == CAPP1_PHB_INDEX) {
		/* Set 30 Read machines for CAPP Minus 20-27 for DMA */
		reg = 0xFFFFF00E00000000;
		if (capp_eng & CAPP_MAX_DMA_READ_ENGINES)
			reg = 0xFF00000000000000;
		xscom_write(p->chip_id, APC_FSM_READ_MASK + offset, reg);
		xscom_write(p->chip_id, XPT_FSM_RMM + offset, reg);
	}

	/* CAPP FIR Action 0 */
	xscom_write(p->chip_id, CAPP_FIR_ACTION0 + offset, 0x0b1c000104060000);

	/* CAPP FIR Action 1 */
	xscom_write(p->chip_id, CAPP_FIR_ACTION1 + offset, 0x2b9c0001240E0000);

	/* CAPP FIR MASK */
	xscom_write(p->chip_id, CAPP_FIR_MASK + offset, 0x80031f98d8717000);

	/* Mask the CAPP PSL Credit Timeout Register error */
	xscom_write_mask(p->chip_id, CAPP_FIR_MASK + offset,
			 PPC_BIT(46), PPC_BIT(46));

	/* Deassert TLBI_FENCED and tlbi_psl_is_dead */
	xscom_write(p->chip_id, CAPP_ERR_STATUS_CTRL + offset, 0);
}

/* override some inits with CAPI defaults */
static void phb4_init_capp_errors(struct phb4 *p)
{
	/* Init_77: TXE Error AIB Fence Enable Register */
	out_be64(p->regs + 0x0d30,	0xdff7bf0bf7ddfff0ull);

	/* Init_86: RXE_ARB Error AIB Fence Enable Register */
	out_be64(p->regs + 0x0db0,	0xfbffd7bbfb7fbfefull);

	/* Init_95: RXE_MRG Error AIB Fence Enable Register */
	out_be64(p->regs + 0x0e30,	0xfffffeffff7fff57ull);

	/* Init_104: RXE_TCE Error AIB Fence Enable Register */
	out_be64(p->regs + 0x0eb0,	0xffaeffafffffffffull);

	/* Init_113: PHB Error AIB Fence Enable Register */
	out_be64(p->regs + 0x0cb0,	0x35777073ff000000ull);
}

 /*
 * The capi indicator is over the 8 most significant bits on p9 (and
 * not 16). We stay away from bits 59 (TVE select), 60 and 61 (MSI)
 *
 * For the mask, we keep bit 59 in, as capi messages must hit TVE#0.
 * Bit 56 is not part of the mask, so that a NBW message (see below)
 * is also considered a capi message.
 */
#define CAPIIND		0x0200
#define CAPIMASK	0xFE00

/*
 * Non-Blocking Write messages are a subset of capi messages, so the
 * indicator is the same as capi + an extra bit (56) to differentiate.
 * Mask is the same as capi + the extra bit
 */
#define NBWIND		0x0300
#define NBWMASK		0xFF00

/*
 * The ASN indicator is used for tunneled operations (as_notify and
 * atomics).  Tunneled operation messages can be sent in PCI mode as
 * well as CAPI mode.
 *
 * The format of those messages is specific and, for as_notify
 * messages, the address field is hijacked to encode the LPID/PID/TID
 * of the target thread, so those messages should not go through
 * translation. They must hit TVE#1. Therefore bit 59 is part of the
 * indicator.
 */
#define ASNIND		0x0C00
#define ASNMASK		0xFF00

/* Power Bus Common Queue Registers
 * All PBCQ and PBAIB registers are accessed via SCOM
 * NestBase = 4010C00 for PEC0
 *            4011000 for PEC1
 *            4011400 for PEC2
 * PCIBase  = D010800 for PE0
 *            E010800 for PE1
 *            F010800 for PE2
 *
 * Some registers are shared amongst all of the stacks and will only
 * have 1 copy. Other registers are implemented one per stack.
 * Registers that are duplicated will have an additional offset
 * of “StackBase” so that they have a unique address.
 * Stackoffset = 00000040 for Stack0
 *             = 00000080 for Stack1
 *             = 000000C0 for Stack2
 */
static int64_t enable_capi_mode(struct phb4 *p, uint64_t pe_number,
				uint32_t capp_eng)
{
	uint64_t reg, start_addr, end_addr, stq_eng, dma_eng;
	int i;

	/* CAPP Control Register */
	xscom_read(p->chip_id, p->pe_xscom + XPEC_NEST_CAPP_CNTL, &reg);
	if (reg & PPC_BIT(0)) {
		PHBDBG(p, "Already in CAPP mode\n");
	}

	for (i = 0; i < 500000; i++) {
		/* PBCQ General Status Register */
		xscom_read(p->chip_id,
			   p->pe_stk_xscom + XPEC_NEST_STK_PBCQ_STAT,
			   &reg);
		if (!(reg & 0xC000000000000000))
			break;
		time_wait_us(10);
	}
	if (reg & 0xC000000000000000) {
		PHBERR(p, "CAPP: Timeout waiting for pending transaction\n");
		return OPAL_HARDWARE;
	}

	/* CAPP Control Register. Enable CAPP Mode */
	reg = 0x8000000000000000ULL; /* PEC works in CAPP Mode */
	if (p->index == CAPP0_PHB_INDEX) {
		/* PBCQ is operating as a x16 stack
		 * - The maximum number of engines give to CAPP will be
		 * 14 and will be assigned in the order of STQ 15 to 2.
		 * - 0-47 (Read machines) are available for capp use.
		 */
		stq_eng = 0x000E000000000000ULL; /* 14 CAPP msg engines */
		dma_eng = 0x0000FFFFFFFFFFFFULL; /* 48 CAPP Read machines */
	}
	if (p->index == CAPP1_PHB_INDEX) {
		/* PBCQ is operating as a x8 stack
		 * - The maximum number of engines given to CAPP should
		 * be 6 and will be assigned in the order of 7 to 2.
		 * - 0-30 (Read machines) are available for capp use.
		 */
		stq_eng = 0x0006000000000000ULL; /* 6 CAPP msg engines */
		dma_eng = 0x0000FFFFF00E0000ULL; /* 30 Read machines for CAPP Minus 20-27 for DMA */
	}
	if (capp_eng & CAPP_MIN_STQ_ENGINES)
		stq_eng = 0x0002000000000000ULL; /* 2 capp msg engines */
	reg |= stq_eng;
	if (capp_eng & CAPP_MAX_DMA_READ_ENGINES)
		dma_eng = 0x0000FF0000000000ULL; /* 16 CAPP Read machines */
	reg |= dma_eng;
	xscom_write(p->chip_id, p->pe_xscom + XPEC_NEST_CAPP_CNTL, reg);

	/* PCI to PB data movement ignores the PB init signal. */
	xscom_write_mask(p->chip_id, p->pe_xscom + XPEC_NEST_PBCQ_HW_CONFIG,
			 XPEC_NEST_PBCQ_HW_CONFIG_PBINIT,
			 XPEC_NEST_PBCQ_HW_CONFIG_PBINIT);

	/* PEC Phase 4 (PHB) registers adjustment
	 * Inbound CAPP traffic: The CAPI can send both CAPP packets and
	 * I/O packets. A PCIe packet is indentified as a CAPP packet in
	 * the PHB if the PCIe address matches either the CAPI
	 * Compare/Mask register or its NBW Compare/Mask register.
	 */

	/*
	 * Bit [0:7] XSL_DSNCTL[capiind]
	 * Init_26 - CAPI Compare/Mask
	 */
	out_be64(p->regs + PHB_CAPI_CMPM,
		 ((u64)CAPIIND << 48) |
		 ((u64)CAPIMASK << 32) | PHB_CAPI_CMPM_ENABLE);

	if (!(p->rev == PHB4_REV_NIMBUS_DD10)) {
		/* PB AIB Hardware Control Register
		 * Wait 32 PCI clocks for a credit to become available
		 * before rejecting.
		 */
		xscom_read(p->chip_id,
			   p->pci_xscom + XPEC_PCI_PBAIB_HW_CONFIG, &reg);
		reg |= PPC_BITMASK(40, 42);
		if (p->index == CAPP1_PHB_INDEX)
			reg |= PPC_BIT(30);
		xscom_write(p->chip_id,
			    p->pci_xscom + XPEC_PCI_PBAIB_HW_CONFIG,
			    reg);
	}

	/* non-translate/50-bit mode */
	out_be64(p->regs + PHB_NXLATE_PREFIX, 0x0000000000000000Ull);

	/* set tve no translate mode allow mmio window */
	memset(p->tve_cache, 0x0, sizeof(p->tve_cache));

	/*
	 * In 50-bit non-translate mode, the fields of the TVE are
	 * used to perform an address range check. In this mode TCE
	 * Table Size(0) must be a '1' (TVE[51] = 1)
	 *      PCI Addr(49:24) >= TVE[52:53]+TVE[0:23] and
	 *      PCI Addr(49:24) < TVE[54:55]+TVE[24:47]
	 *
	 * TVE[51] = 1
	 * TVE[56] = 1: 50-bit Non-Translate Mode Enable
	 * TVE[0:23] = 0x000000
	 * TVE[24:47] = 0xFFFFFF
	 *
	 * capi dma mode: CAPP DMA mode needs access to all of memory
	 * capi mode: Allow address range (bit 14 = 1)
	 *            0x0002000000000000: 0x0002FFFFFFFFFFFF
	 *            TVE[52:53] = '10' and TVE[54:55] = '10'
	 */

	/* TVT#0: CAPI window + DMA, all memory */
	start_addr = 0ull;
	end_addr   = 0x0003ffffffffffffull;
	p->tve_cache[pe_number * 2] =
		tve_encode_50b_noxlate(start_addr, end_addr);

	/* TVT#1: CAPI window + DMA, all memory, in bypass mode */
	start_addr = (1ull << 59);
	end_addr   = start_addr + 0x0003ffffffffffffull;
	p->tve_cache[pe_number * 2 + 1] =
		tve_encode_50b_noxlate(start_addr, end_addr);

	phb4_ioda_sel(p, IODA3_TBL_TVT, 0, true);
	for (i = 0; i < p->tvt_size; i++)
		out_be64(p->regs + PHB_IODA_DATA0, p->tve_cache[i]);

	/* set mbt bar to pass capi mmio window. First applied cleared
	 * values to HW
	 */
	for (i = 0; i < p->mbt_size; i++) {
		p->mbt_cache[i][0] = 0;
		p->mbt_cache[i][1] = 0;
	}
	phb4_ioda_sel(p, IODA3_TBL_MBT, 0, true);
	for (i = 0; i < p->mbt_size; i++) {
		out_be64(p->regs + PHB_IODA_DATA0, p->mbt_cache[i][0]);
		out_be64(p->regs + PHB_IODA_DATA0, p->mbt_cache[i][1]);
	}

	p->mbt_cache[0][0] = IODA3_MBT0_ENABLE |
			     IODA3_MBT0_TYPE_M64 |
		SETFIELD(IODA3_MBT0_MODE, 0ull, IODA3_MBT0_MODE_SINGLE_PE) |
		SETFIELD(IODA3_MBT0_MDT_COLUMN, 0ull, 0) |
		(p->mm0_base & IODA3_MBT0_BASE_ADDR);
	p->mbt_cache[0][1] = IODA3_MBT1_ENABLE |
		((~(p->mm0_size - 1)) & IODA3_MBT1_MASK) |
		SETFIELD(IODA3_MBT1_SINGLE_PE_NUM, 0ull, pe_number);

	p->mbt_cache[1][0] = IODA3_MBT0_ENABLE |
			     IODA3_MBT0_TYPE_M64 |
		SETFIELD(IODA3_MBT0_MODE, 0ull, IODA3_MBT0_MODE_SINGLE_PE) |
		SETFIELD(IODA3_MBT0_MDT_COLUMN, 0ull, 0) |
		(0x0002000000000000ULL & IODA3_MBT0_BASE_ADDR);
	p->mbt_cache[1][1] = IODA3_MBT1_ENABLE |
		(0x00ff000000000000ULL & IODA3_MBT1_MASK) |
		SETFIELD(IODA3_MBT1_SINGLE_PE_NUM, 0ull, pe_number);

	phb4_ioda_sel(p, IODA3_TBL_MBT, 0, true);
	for (i = 0; i < p->mbt_size; i++) {
		out_be64(p->regs + PHB_IODA_DATA0, p->mbt_cache[i][0]);
		out_be64(p->regs + PHB_IODA_DATA0, p->mbt_cache[i][1]);
	}

	phb4_init_capp_errors(p);

	phb4_init_capp_regs(p, capp_eng);

	if (!chiptod_capp_timebase_sync(p->chip_id, CAPP_TFMR,
					CAPP_TB,
					PHB4_CAPP_REG_OFFSET(p)))
		PHBERR(p, "CAPP: Failed to sync timebase\n");

	/* set callbacks to handle HMI events */
	capi_ops.get_capp_info = &phb4_get_capp_info;

	return OPAL_SUCCESS;
}

static int64_t phb4_set_capi_mode(struct phb *phb, uint64_t mode,
				  uint64_t pe_number)
{
	struct phb4 *p = phb_to_phb4(phb);
	struct proc_chip *chip = get_chip(p->chip_id);
	uint64_t reg, ret;
	uint32_t offset;


	if (!capp_ucode_loaded(chip, p->index)) {
		PHBERR(p, "CAPP: ucode not loaded\n");
		return OPAL_RESOURCE;
	}

	lock(&capi_lock);
	chip->capp_phb4_attached_mask |= 1 << p->index;
	unlock(&capi_lock);

	offset = PHB4_CAPP_REG_OFFSET(p);
	xscom_read(p->chip_id, CAPP_ERR_STATUS_CTRL + offset, &reg);
	if ((reg & PPC_BIT(5))) {
		PHBERR(p, "CAPP: recovery failed (%016llx)\n", reg);
		return OPAL_HARDWARE;
	} else if ((reg & PPC_BIT(0)) && (!(reg & PPC_BIT(1)))) {
		PHBDBG(p, "CAPP: recovery in progress\n");
		return OPAL_BUSY;
	}

	switch (mode) {
	case OPAL_PHB_CAPI_MODE_CAPI:
		ret = enable_capi_mode(p, pe_number,
					CAPP_MAX_STQ_ENGINES |
					CAPP_MIN_DMA_READ_ENGINES);
		disable_fast_reboot("CAPP being enabled");
		break;
	case OPAL_PHB_CAPI_MODE_DMA_TVT1:
		ret = enable_capi_mode(p, pe_number,
					CAPP_MIN_STQ_ENGINES |
					CAPP_MAX_DMA_READ_ENGINES);
		disable_fast_reboot("CAPP being enabled");
		break;
	case OPAL_PHB_CAPI_MODE_SNOOP_ON:
		/* nothing to do P9 if CAPP is alreay enabled */
		ret = OPAL_SUCCESS;
		break;

	case OPAL_PHB_CAPI_MODE_PCIE: /* shouldn't be called on p9*/
	case OPAL_PHB_CAPI_MODE_DMA: /* Enabled by default on p9 */
	case OPAL_PHB_CAPI_MODE_SNOOP_OFF: /* shouldn't be called on p9*/
	default:
		ret = OPAL_UNSUPPORTED;
	}

	return ret;
}

static void phb4_p2p_set_initiator(struct phb4 *p, uint16_t pe_number)
{
	uint64_t tve;
	uint16_t window_id = (pe_number << 1) + 1;

	/*
	 * Initiator needs access to the MMIO space of the target,
	 * which is well beyond the 'normal' memory area. Set its TVE
	 * with no range checking.
	 */
	PHBDBG(p, "Setting TVE#1 for peer-to-peer for pe %d\n", pe_number);
	tve = PPC_BIT(51);
	phb4_ioda_sel(p, IODA3_TBL_TVT, window_id, false);
	out_be64(p->regs + PHB_IODA_DATA0, tve);
	p->tve_cache[window_id] = tve;
}

static void phb4_p2p_set_target(struct phb4 *p, bool enable)
{
	uint64_t val;

	/*
	 * Enabling p2p on a target PHB reserves an outbound (as seen
	 * from the CPU) store queue for p2p
	 */
	PHBDBG(p, "%s peer-to-peer\n", (enable ? "Enabling" : "Disabling"));
	xscom_read(p->chip_id,
		p->pe_stk_xscom + XPEC_NEST_STK_PBCQ_MODE, &val);
	if (enable)
		val |= XPEC_NEST_STK_PBCQ_MODE_P2P;
	else
		val &= ~XPEC_NEST_STK_PBCQ_MODE_P2P;
	xscom_write(p->chip_id,
		p->pe_stk_xscom + XPEC_NEST_STK_PBCQ_MODE, val);
}

static void phb4_set_p2p(struct phb *phb, uint64_t mode, uint64_t flags,
			uint16_t pe_number)
{
	struct phb4 *p = phb_to_phb4(phb);

	switch (mode) {
	case OPAL_PCI_P2P_INITIATOR:
		if (flags & OPAL_PCI_P2P_ENABLE)
			phb4_p2p_set_initiator(p, pe_number);
		/*
		 * When disabling p2p on the initiator, we should
		 * reset the TVE to its default bypass setting, but it
		 * is more easily done from the OS, as it knows the
		 * the start and end address and there's already an
		 * opal call for it, so let linux handle it.
		 */
		break;
	case OPAL_PCI_P2P_TARGET:
		phb4_p2p_set_target(p, !!(flags & OPAL_PCI_P2P_ENABLE));
		break;
	default:
		assert(0);
	}
}

static int64_t phb4_set_capp_recovery(struct phb *phb)
{
	struct phb4 *p = phb_to_phb4(phb);

	if (p->flags & PHB4_CAPP_RECOVERY)
		return 0;

	/* set opal event flag to indicate eeh condition */
	opal_update_pending_evt(OPAL_EVENT_PCI_ERROR,
				OPAL_EVENT_PCI_ERROR);

	p->flags |= PHB4_CAPP_RECOVERY;

	return 0;
}

/*
 * Return the address out of a PBCQ Tunnel Bar register.
 */
static void phb4_get_tunnel_bar(struct phb *phb, uint64_t *addr)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t val;

	xscom_read(p->chip_id, p->pe_stk_xscom + XPEC_NEST_STK_TUNNEL_BAR,
		   &val);
	*addr = val >> 8;
}

/*
 * Set PBCQ Tunnel Bar register.
 * Store addr bits [8:50] in PBCQ Tunnel Bar register bits [0:42].
 * Note that addr bits [8:50] must also match PSL_TNR_ADDR[8:50].
 * Reset register if val == 0.
 *
 * This interface is required to let device drivers set the Tunnel Bar
 * value of their choice.
 *
 * Compatibility with older versions of linux, that do not set the
 * Tunnel Bar with phb4_set_tunnel_bar(), is ensured by enable_capi_mode(),
 * that will set the default value that used to be assumed.
 */
static int64_t phb4_set_tunnel_bar(struct phb *phb, uint64_t addr)
{
	struct phb4 *p = phb_to_phb4(phb);
	uint64_t mask = 0x00FFFFFFFFFFE000ULL;

	if (!addr) {
		/* Reset register */
		xscom_write(p->chip_id,
			    p->pe_stk_xscom + XPEC_NEST_STK_TUNNEL_BAR, addr);
		return OPAL_SUCCESS;
	}
	if ((addr & ~mask))
		return OPAL_PARAMETER;
	if (!(addr & mask))
		return OPAL_PARAMETER;

	xscom_write(p->chip_id, p->pe_stk_xscom + XPEC_NEST_STK_TUNNEL_BAR,
		    (addr & mask) << 8);
	return OPAL_SUCCESS;
}

static const struct phb_ops phb4_ops = {
	.cfg_read8		= phb4_pcicfg_read8,
	.cfg_read16		= phb4_pcicfg_read16,
	.cfg_read32		= phb4_pcicfg_read32,
	.cfg_write8		= phb4_pcicfg_write8,
	.cfg_write16		= phb4_pcicfg_write16,
	.cfg_write32		= phb4_pcicfg_write32,
	.choose_bus		= phb4_choose_bus,
	.get_reserved_pe_number	= phb4_get_reserved_pe_number,
	.device_init		= phb4_device_init,
	.device_remove		= NULL,
	.ioda_reset		= phb4_ioda_reset,
	.papr_errinjct_reset	= phb4_papr_errinjct_reset,
	.pci_reinit		= phb4_pci_reinit,
	.set_phb_mem_window	= phb4_set_phb_mem_window,
	.phb_mmio_enable	= phb4_phb_mmio_enable,
	.map_pe_mmio_window	= phb4_map_pe_mmio_window,
	.map_pe_dma_window	= phb4_map_pe_dma_window,
	.map_pe_dma_window_real = phb4_map_pe_dma_window_real,
	.set_xive_pe		= phb4_set_ive_pe,
	.get_msi_32		= phb4_get_msi_32,
	.get_msi_64		= phb4_get_msi_64,
	.set_pe			= phb4_set_pe,
	.set_peltv		= phb4_set_peltv,
	.eeh_freeze_status	= phb4_eeh_freeze_status,
	.eeh_freeze_clear	= phb4_eeh_freeze_clear,
	.eeh_freeze_set		= phb4_eeh_freeze_set,
	.next_error		= phb4_eeh_next_error,
	.err_inject		= phb4_err_inject,
	.get_diag_data		= NULL,
	.get_diag_data2		= phb4_get_diag_data,
	.tce_kill		= phb4_tce_kill,
	.set_capi_mode		= phb4_set_capi_mode,
	.set_p2p		= phb4_set_p2p,
	.set_capp_recovery	= phb4_set_capp_recovery,
	.get_tunnel_bar         = phb4_get_tunnel_bar,
	.set_tunnel_bar         = phb4_set_tunnel_bar,
};

static void phb4_init_ioda3(struct phb4 *p)
{
	/* Init_18 - Interrupt Notify Base Address */
	out_be64(p->regs + PHB_INT_NOTIFY_ADDR, p->irq_port);

	/* Init_19 - Interrupt Notify Base Index */
	out_be64(p->regs + PHB_INT_NOTIFY_INDEX,
		 xive_get_notify_base(p->base_msi));

	/* Init_19x - Not in spec: Initialize source ID */
	PHBDBG(p, "Reset state SRC_ID: %016llx\n",
	       in_be64(p->regs + PHB_LSI_SOURCE_ID));
	out_be64(p->regs + PHB_LSI_SOURCE_ID,
		 SETFIELD(PHB_LSI_SRC_ID, 0ull, (p->num_irqs - 1) >> 3));

	/* Init_20 - RTT BAR */
	out_be64(p->regs + PHB_RTT_BAR, p->tbl_rtt | PHB_RTT_BAR_ENABLE);

	/* Init_21 - PELT-V BAR */
	out_be64(p->regs + PHB_PELTV_BAR, p->tbl_peltv | PHB_PELTV_BAR_ENABLE);

	/* Init_22 - Setup M32 starting address */
	out_be64(p->regs + PHB_M32_START_ADDR, M32_PCI_START);

	/* Init_23 - Setup PEST BAR */
	out_be64(p->regs + PHB_PEST_BAR,
		 p->tbl_pest | PHB_PEST_BAR_ENABLE);

	/* Init_24 - CRW Base Address Reg */
	/* See enable_capi_mode() */

	/* Init_25 - ASN Compare/Mask */
	out_be64(p->regs + PHB_ASN_CMPM, ((u64)ASNIND << 48) |
		 ((u64)ASNMASK << 32) | PHB_ASN_CMPM_ENABLE);

	/* Init_26 - CAPI Compare/Mask */
	/* See enable_capi_mode() */

	/* Init_27 - PCIE Outbound upper address */
	out_be64(p->regs + PHB_M64_UPPER_BITS, 0);

	/* Init_28 - PHB4 Configuration */
	out_be64(p->regs + PHB_PHB4_CONFIG,
		 PHB_PHB4C_32BIT_MSI_EN |
		 PHB_PHB4C_64BIT_MSI_EN);

	/* Init_29 - At least 256ns delay according to spec. Do a dummy
	 * read first to flush posted writes
	 */
	in_be64(p->regs + PHB_PHB4_CONFIG);
	time_wait_us(2);

	/* Init_30..41 - On-chip IODA tables init */
	phb4_ioda_reset(&p->phb, false);
}

/* phb4_init_rc - Initialize the Root Complex config space
 */
static bool phb4_init_rc_cfg(struct phb4 *p)
{
	int64_t ecap, aercap;

	/* XXX Handle errors ? */

	/* Init_46:
	 *
	 * Set primary bus to 0, secondary to 1 and subordinate to 0xff
	 */
	phb4_pcicfg_write32(&p->phb, 0, PCI_CFG_PRIMARY_BUS, 0x00ff0100);

	/* Init_47 - Clear errors */
	/* see phb4_rc_err_clear() called below */

	/* Init_48
	 *
	 * PCIE Device control/status, enable error reporting, disable relaxed
	 * ordering, set MPS to 128 (see note), clear errors.
	 *
	 * Note: The doc recommends to set MPS to 512. This has proved to have
	 * some issues as it requires specific clamping of MRSS on devices and
	 * we've found devices in the field that misbehave when doing that.
	 *
	 * We currently leave it all to 128 bytes (minimum setting) at init
	 * time. The generic PCIe probing later on might apply a different
	 * value, or the kernel will, but we play it safe at early init
	 */
	if (p->ecap <= 0) {
		ecap = pci_find_cap(&p->phb, 0, PCI_CFG_CAP_ID_EXP);
		if (ecap < 0) {
			PHBERR(p, "Can't locate PCI-E capability\n");
			return false;
		}
		p->ecap = ecap;
	} else {
		ecap = p->ecap;
	}

	phb4_pcicfg_write16(&p->phb, 0, ecap + PCICAP_EXP_DEVCTL,
			     PCICAP_EXP_DEVCTL_CE_REPORT	|
			     PCICAP_EXP_DEVCTL_NFE_REPORT	|
			     PCICAP_EXP_DEVCTL_FE_REPORT	|
			     PCICAP_EXP_DEVCTL_UR_REPORT	|
			     SETFIELD(PCICAP_EXP_DEVCTL_MPS, 0, PCIE_MPS_128B));

	/* Init_49 - Device Control/Status 2 */
	phb4_pcicfg_write16(&p->phb, 0, ecap + PCICAP_EXP_DCTL2,
			     SETFIELD(PCICAP_EXP_DCTL2_CMPTOUT, 0, 0x5) |
			     PCICAP_EXP_DCTL2_ARI_FWD);

	/* Init_50..54
	 *
	 * AER inits
	 */
	if (p->aercap <= 0) {
		aercap = pci_find_ecap(&p->phb, 0, PCIECAP_ID_AER, NULL);
		if (aercap < 0) {
			PHBERR(p, "Can't locate AER capability\n");
			return false;
		}
		p->aercap = aercap;
	} else {
		aercap = p->aercap;
	}

	/* Disable some error reporting as per the PHB4 spec */
	phb4_pcicfg_write32(&p->phb, 0, aercap + PCIECAP_AER_UE_MASK,
			     PCIECAP_AER_UE_POISON_TLP		|
			     PCIECAP_AER_UE_COMPL_TIMEOUT	|
			     PCIECAP_AER_UE_COMPL_ABORT);

	/* Enable ECRC generation & checking */
	phb4_pcicfg_write32(&p->phb, 0, aercap + PCIECAP_AER_CAPCTL,
			     PCIECAP_AER_CAPCTL_ECRCG_EN	|
			     PCIECAP_AER_CAPCTL_ECRCC_EN);

	phb4_rc_err_clear(p);

	return true;
}

static void phb4_init_errors(struct phb4 *p)
{
	/* Init_55..63 - PBL errors */
	out_be64(p->regs + 0x1900,	0xffffffffffffffffull);
	out_be64(p->regs + 0x1908,	0x0000000000000000ull);
	out_be64(p->regs + 0x1920,	0x000000004d1780f8ull);
	out_be64(p->regs + 0x1928,	0x0000000000000000ull);
	if (p->rev == PHB4_REV_NIMBUS_DD10)
		out_be64(p->regs + 0x1930,	0xffffffffb2e87f07ull);
	else
		out_be64(p->regs + 0x1930,	0xffffffffb2f87f07ull);
	out_be64(p->regs + 0x1940,	0x0000000000000000ull);
	out_be64(p->regs + 0x1948,	0x0000000000000000ull);
	out_be64(p->regs + 0x1950,	0x0000000000000000ull);
	out_be64(p->regs + 0x1958,	0x0000000000000000ull);

	/* Init_64..72 - REGB errors */
	out_be64(p->regs + 0x1c00,	0xffffffffffffffffull);
	out_be64(p->regs + 0x1c08,	0x0000000000000000ull);
	/* Enable/disable error status indicators that trigger irqs */
	if (p->has_link) {
		out_be64(p->regs + 0x1c20,	0x2130006efca8bc00ull);
		out_be64(p->regs + 0x1c30,	0xde1fff91035743ffull);
	} else {
		out_be64(p->regs + 0x1c20,	0x0000000000000000ull);
		out_be64(p->regs + 0x1c30,	0x0000000000000000ull);
	}
	out_be64(p->regs + 0x1c28,	0x0080000000000000ull);
	out_be64(p->regs + 0x1c40,	0x0000000000000000ull);
	out_be64(p->regs + 0x1c48,	0x0000000000000000ull);
	out_be64(p->regs + 0x1c50,	0x0000000000000000ull);
	out_be64(p->regs + 0x1c58,	0x0040000000000000ull);

	/* Init_73..81 - TXE errors */
	out_be64(p->regs + 0x0d00,	0xffffffffffffffffull);
	out_be64(p->regs + 0x0d08,	0x0000000000000000ull);
	out_be64(p->regs + 0x0d18,	0xffffff0fffffffffull);
	out_be64(p->regs + 0x0d28,	0x0000400a00000000ull);
	out_be64(p->regs + 0x0d30,	0xdff7bd01f7ddfff0ull); /* XXX CAPI has diff. value */
	out_be64(p->regs + 0x0d40,	0x0000000000000000ull);
	out_be64(p->regs + 0x0d48,	0x0000000000000000ull);
	out_be64(p->regs + 0x0d50,	0x0000000000000000ull);
	out_be64(p->regs + 0x0d58,	0x0000000000000000ull);

	/* Init_82..90 - RXE_ARB errors */
	out_be64(p->regs + 0x0d80,	0xffffffffffffffffull);
	out_be64(p->regs + 0x0d88,	0x0000000000000000ull);
	out_be64(p->regs + 0x0d98,	0xfffffffffbffffffull);
	if (p->rev == PHB4_REV_NIMBUS_DD10)
		out_be64(p->regs + 0x0da8,	0xc00000b801000060ull);
	else
		out_be64(p->regs + 0x0da8,	0xc00018b801000060ull);
	/*
	 * Errata ER20161123 says we should set the top two bits in
	 * 0x0db0 but this causes config space accesses which don't
	 * get a response to fence the PHB. This breaks probing,
	 * hence we don't set them here.
	 */
	out_be64(p->regs + 0x0db0,	0x3bffd703fa7fbf8full); /* XXX CAPI has diff. value */
	out_be64(p->regs + 0x0dc0,	0x0000000000000000ull);
	out_be64(p->regs + 0x0dc8,	0x0000000000000000ull);
	out_be64(p->regs + 0x0dd0,	0x0000000000000000ull);
	out_be64(p->regs + 0x0dd8,	0x0000000004000000ull);

	/* Init_91..99 - RXE_MRG errors */
	out_be64(p->regs + 0x0e00,	0xffffffffffffffffull);
	out_be64(p->regs + 0x0e08,	0x0000000000000000ull);
	out_be64(p->regs + 0x0e18,	0xffffffffffffffffull);
	out_be64(p->regs + 0x0e28,	0x0000600000000000ull);
	if (p->rev == PHB4_REV_NIMBUS_DD10) /* XXX CAPI has diff. value */
		out_be64(p->regs + 0x0e30,	0xffff9effff7fff57ull);
	else
		out_be64(p->regs + 0x0e30,	0xfffffeffff7fff57ull);
	out_be64(p->regs + 0x0e40,	0x0000000000000000ull);
	out_be64(p->regs + 0x0e48,	0x0000000000000000ull);
	out_be64(p->regs + 0x0e50,	0x0000000000000000ull);
	out_be64(p->regs + 0x0e58,	0x0000000000000000ull);

	/* Init_100..108 - RXE_TCE errors */
	out_be64(p->regs + 0x0e80,	0xffffffffffffffffull);
	out_be64(p->regs + 0x0e88,	0x0000000000000000ull);
	out_be64(p->regs + 0x0e98,	0xffffffffffffffffull);
	if (p->rev == PHB4_REV_NIMBUS_DD10)
		out_be64(p->regs + 0x0ea8,	0x6000000000000000ull);
	else
		out_be64(p->regs + 0x0ea8,	0x60000000c0000000ull);
	out_be64(p->regs + 0x0eb0,	0x9faeffaf3fffffffull); /* XXX CAPI has diff. value */
	out_be64(p->regs + 0x0ec0,	0x0000000000000000ull);
	out_be64(p->regs + 0x0ec8,	0x0000000000000000ull);
	out_be64(p->regs + 0x0ed0,	0x0000000000000000ull);
	out_be64(p->regs + 0x0ed8,	0x0000000000000000ull);

	/* Init_109..117 - RXPHB errors */
	out_be64(p->regs + 0x0c80,	0xffffffffffffffffull);
	out_be64(p->regs + 0x0c88,	0x0000000000000000ull);
	out_be64(p->regs + 0x0c98,	0xffffffffffffffffull);
	out_be64(p->regs + 0x0ca8,	0x0000004000000000ull);
	out_be64(p->regs + 0x0cb0,	0x35777033ff000000ull); /* XXX CAPI has diff. value */
	out_be64(p->regs + 0x0cc0,	0x0000000000000000ull);
	out_be64(p->regs + 0x0cc8,	0x0000000000000000ull);
	out_be64(p->regs + 0x0cd0,	0x0000000000000000ull);
	out_be64(p->regs + 0x0cd8,	0x0000000000000000ull);

	/* Init_118..121 - LEM */
	out_be64(p->regs + 0x0c00,	0x0000000000000000ull);
	out_be64(p->regs + 0x0c30,	0xffffffffffffffffull);
	out_be64(p->regs + 0x0c38,	0xffffffffffffffffull);
	out_be64(p->regs + 0x0c40,	0x0000000000000000ull);
}


static bool phb4_wait_dlp_reset(struct phb4 *p)
{
	unsigned int i;
	uint64_t val;

	/*
	 * Firmware cannot access the UTL core regs or PCI config space
	 * until the cores are out of DL_PGRESET.
	 * DL_PGRESET should be polled until it is inactive with a value
	 * of '0'. The recommended polling frequency is once every 1ms.
	 * Firmware should poll at least 200 attempts before giving up.
	 * MMIO Stores to the link are silently dropped by the UTL core if
	 * the link is down.
	 * MMIO Loads to the link will be dropped by the UTL core and will
	 * eventually time-out and will return an all ones response if the
	 * link is down.
	 */
#define DLP_RESET_ATTEMPTS	200

	PHBDBG(p, "Waiting for DLP PG reset to complete...\n");
	for (i = 0; i < DLP_RESET_ATTEMPTS; i++) {
		val = in_be64(p->regs + PHB_PCIE_DLP_TRAIN_CTL);
		if (!(val & PHB_PCIE_DLP_DL_PGRESET))
			break;
		time_wait_ms(1);
	}
	if (val & PHB_PCIE_DLP_DL_PGRESET) {
		PHBERR(p, "Timeout waiting for DLP PG reset !\n");
		return false;
	}
	return true;
}
static void phb4_init_hw(struct phb4 *p, bool first_init)
{
	uint64_t val, creset;

	PHBDBG(p, "Initializing PHB4...\n");

	/* Init_1 - Sync reset
	 *
	 * At this point we assume the PHB has already been reset.
	 */

	/* Init_2 - Mask FIRs */
	out_be64(p->regs + PHB_LEM_ERROR_MASK,			0xffffffffffffffffull);

	/* Init_3 - TCE tag enable */
	out_be64(p->regs + PHB_TCE_TAG_ENABLE,			0xffffffffffffffffull);

	/* Init_4 - PCIE System Configuration Register
	 *
	 * Adjust max speed based on system config
	 */
	val = in_be64(p->regs + PHB_PCIE_SCR);
	PHBDBG(p, "Default system config: 0x%016llx\n", val);
	val = SETFIELD(PHB_PCIE_SCR_MAXLINKSPEED, val, p->max_link_speed);
	out_be64(p->regs + PHB_PCIE_SCR, val);
	PHBDBG(p, "New system config    : 0x%016llx\n",
	       in_be64(p->regs + PHB_PCIE_SCR));

	/* Init_5 - deassert CFG reset */
	creset = in_be64(p->regs + PHB_PCIE_CRESET);
	PHBDBG(p, "Initial PHB CRESET is 0x%016llx\n", creset);
	creset &= ~PHB_PCIE_CRESET_CFG_CORE;
	out_be64(p->regs + PHB_PCIE_CRESET,			creset);

	/* Init_6..13 - PCIE DLP Lane EQ control */
	if (p->lane_eq) {
		out_be64(p->regs + PHB_PCIE_LANE_EQ_CNTL0, be64_to_cpu(p->lane_eq[0]));
		out_be64(p->regs + PHB_PCIE_LANE_EQ_CNTL1, be64_to_cpu(p->lane_eq[1]));
		out_be64(p->regs + PHB_PCIE_LANE_EQ_CNTL2, be64_to_cpu(p->lane_eq[2]));
		out_be64(p->regs + PHB_PCIE_LANE_EQ_CNTL3, be64_to_cpu(p->lane_eq[3]));
		out_be64(p->regs + PHB_PCIE_LANE_EQ_CNTL20, be64_to_cpu(p->lane_eq[4]));
		out_be64(p->regs + PHB_PCIE_LANE_EQ_CNTL21, be64_to_cpu(p->lane_eq[5]));
		if (p->rev == PHB4_REV_NIMBUS_DD10) {
			out_be64(p->regs + PHB_PCIE_LANE_EQ_CNTL22,
				 be64_to_cpu(p->lane_eq[6]));
			out_be64(p->regs + PHB_PCIE_LANE_EQ_CNTL23,
				 be64_to_cpu(p->lane_eq[7]));
		}
	}
	if (!p->lane_eq_en) {
		/* Read modify write and set to 2 bits */
		PHBDBG(p, "LINK: Disabling Lane EQ\n");
		val = in_be64(p->regs + PHB_PCIE_DLP_CTL);
		val |= PHB_PCIE_DLP_CTL_BYPASS_PH2 | PHB_PCIE_DLP_CTL_BYPASS_PH2;
		out_be64(p->regs + PHB_PCIE_DLP_CTL, val);
	}

	/* Init_14 - Clear link training */
	phb4_pcicfg_write32(&p->phb, 0, 0x78,
			    0x07FE0000 | p->max_link_speed);

	/* Init_15 - deassert cores reset */
	/*
	 * Lift the PHB resets but not PERST, this will be lifted
	 * later by the initial PERST state machine
	 */
	creset &= ~(PHB_PCIE_CRESET_TLDLP | PHB_PCIE_CRESET_PBL);
	creset |= PHB_PCIE_CRESET_PIPE_N;
	out_be64(p->regs + PHB_PCIE_CRESET,			   creset);

	/* Init_16 - Wait for DLP PGRESET to clear */
	if (!phb4_wait_dlp_reset(p))
		goto failed;

	/* Init_17 - PHB Control */
	val = PHB_CTRLR_IRQ_PGSZ_64K;
	if (p->rev == PHB4_REV_NIMBUS_DD10) {
		val |= SETFIELD(PHB_CTRLR_TVT_ADDR_SEL, 0ull, TVT_DD1_2_PER_PE);
	} else {
		val |= SETFIELD(PHB_CTRLR_TVT_ADDR_SEL, 0ull, TVT_2_PER_PE);
		if (PHB4_CAN_STORE_EOI(p))
			val |= PHB_CTRLR_IRQ_STORE_EOI;
	}

	if (!pci_eeh_mmio)
		val |= PHB_CTRLR_MMIO_EEH_DISABLE;

	out_be64(p->regs + PHB_CTRLR, val);

	/* Init_18..41 - Architected IODA3 inits */
	phb4_init_ioda3(p);

	/* Init_42..45 - Clear DLP error logs */
	out_be64(p->regs + 0x1aa0,			0xffffffffffffffffull);
	out_be64(p->regs + 0x1aa8,			0xffffffffffffffffull);
	out_be64(p->regs + 0x1ab0,			0xffffffffffffffffull);
	out_be64(p->regs + 0x1ab8,			0x0);


	/* Init_46..54 : Init root complex config space */
	if (!phb4_init_rc_cfg(p))
		goto failed;

	/* Init_55..121  : Setup error registers */
	phb4_init_errors(p);

	/* Init_122..123 : Wait for link
	 * NOTE: At this point the spec waits for the link to come up. We
	 * don't bother as we are doing a PERST soon.
	 */

	/* Init_124 :  NBW. XXX TODO */
	/* See enable_capi_mode() */

	/* Init_125 : Setup PCI command/status on root complex
	 * I don't know why the spec does this now and not earlier, so
	 * to be sure to get it right we might want to move it to the freset
	 * state machine, though the generic PCI layer will probably do
	 * this anyway (ie, enable MEM, etc... in the RC)

	 */
	phb4_pcicfg_write16(&p->phb, 0, PCI_CFG_CMD,
			    PCI_CFG_CMD_MEM_EN |
			    PCI_CFG_CMD_BUS_MASTER_EN);

	/* Clear errors */
	phb4_pcicfg_write16(&p->phb, 0, PCI_CFG_STAT,
			    PCI_CFG_STAT_SENT_TABORT |
			    PCI_CFG_STAT_RECV_TABORT |
			    PCI_CFG_STAT_RECV_MABORT |
			    PCI_CFG_STAT_SENT_SERR |
			    PCI_CFG_STAT_RECV_PERR);

	/* Init_126..130 - Re-enable error interrupts */
	out_be64(p->regs + PHB_ERR_IRQ_ENABLE,			0xca8880cc00000000ull);
	out_be64(p->regs + PHB_TXE_ERR_IRQ_ENABLE,		0x2008400e08200000ull);
	out_be64(p->regs + PHB_RXE_ARB_ERR_IRQ_ENABLE,		0xc40038fc01804070ull);
	out_be64(p->regs + PHB_RXE_MRG_ERR_IRQ_ENABLE,		0x00006100008000a8ull);
	if (p->rev == PHB4_REV_NIMBUS_DD10)
		out_be64(p->regs + PHB_RXE_TCE_ERR_IRQ_ENABLE,	0x6051005000000000ull);
	else
		out_be64(p->regs + PHB_RXE_TCE_ERR_IRQ_ENABLE,	0x60510050c0000000ull);

	/* Init_131 - Re-enable LEM error mask */
	out_be64(p->regs + PHB_LEM_ERROR_MASK,			0x0000000000000000ull);


	/* Init_132 - Enable DMA address speculation */
	out_be64(p->regs + PHB_TCE_SPEC_CTL,			0x0000000000000000ull);

	/* Init_133 - Timeout Control Register 1 */
	out_be64(p->regs + PHB_TIMEOUT_CTRL1,			0x0015150000150000ull);

	/* Init_134 - Timeout Control Register 2 */
	out_be64(p->regs + PHB_TIMEOUT_CTRL2,			0x0000181700000000ull);

	/* Init_135 - PBL Timeout Control Register */
	out_be64(p->regs + PHB_PBL_TIMEOUT_CTRL,		0x2013000000000000ull);

	/* Mark the PHB as functional which enables all the various sequences */
	p->broken = false;

	PHBDBG(p, "Initialization complete\n");

	return;

 failed:
	PHBERR(p, "Initialization failed\n");
	p->broken = true;
}

/* FIXME: Use scoms rather than MMIO incase we are fenced */
static bool phb4_read_capabilities(struct phb4 *p)
{
	uint64_t val;

	/* XXX Should make sure ETU is out of reset ! */

	/* Grab version and fit it in an int */
	val = phb4_read_reg_asb(p, PHB_VERSION);
	if (val == 0 || val == 0xffffffffffffffff) {
		PHBERR(p, "Failed to read version, PHB appears broken\n");
		return false;
	}

	p->rev = ((val >> 16) & 0x00ff0000) | (val & 0xffff);
	PHBDBG(p, "Core revision 0x%x\n", p->rev);

	/* Read EEH capabilities */
	val = in_be64(p->regs + PHB_PHB4_EEH_CAP);
	if (val == 0xffffffffffffffff) {
		PHBERR(p, "Failed to read EEH cap, PHB appears broken\n");
		return false;
	}
	p->max_num_pes = val >> 52;
	if (p->max_num_pes >= 512) {
		p->mrt_size = 16;
		p->mbt_size = 32;
		p->tvt_size = 512;
	} else {
		p->mrt_size = 8;
		p->mbt_size = 16;
		p->tvt_size = 256;
	}
	/* DD2.0 has twice has many TVEs */
	if (p->rev >= PHB4_REV_NIMBUS_DD20)
		p->tvt_size *= 2;

	val = in_be64(p->regs + PHB_PHB4_IRQ_CAP);
	if (val == 0xffffffffffffffff) {
		PHBERR(p, "Failed to read IRQ cap, PHB appears broken\n");
		return false;
	}
	p->num_irqs = val & 0xffff;

	/* This works for 512 PEs.  FIXME calculate for any hardware
	 * size returned above
	 */
	p->tbl_peltv_size = PELTV_TABLE_SIZE_MAX;

	p->tbl_pest_size = p->max_num_pes*16;

	PHBDBG(p, "Found %d max PEs and %d IRQs \n",
	       p->max_num_pes, p->num_irqs);

	return true;
}

static void phb4_allocate_tables(struct phb4 *p)
{
	uint16_t *rte;
	uint32_t i;

	/* XXX Our current memalign implementation sucks,
	 *
	 * It will do the job, however it doesn't support freeing
	 * the memory and wastes space by always allocating twice
	 * as much as requested (size + alignment)
	 */
	p->tbl_rtt = (uint64_t)local_alloc(p->chip_id, RTT_TABLE_SIZE, RTT_TABLE_SIZE);
	assert(p->tbl_rtt);
	rte = (uint16_t *)(p->tbl_rtt);
	for (i = 0; i < RTT_TABLE_ENTRIES; i++, rte++)
		*rte = PHB4_RESERVED_PE_NUM(p);

	p->tbl_peltv = (uint64_t)local_alloc(p->chip_id, p->tbl_peltv_size, p->tbl_peltv_size);
	assert(p->tbl_peltv);
	memset((void *)p->tbl_peltv, 0, p->tbl_peltv_size);

	p->tbl_pest = (uint64_t)local_alloc(p->chip_id, p->tbl_pest_size, p->tbl_pest_size);
	assert(p->tbl_pest);
	memset((void *)p->tbl_pest, 0, p->tbl_pest_size);
}

static void phb4_add_properties(struct phb4 *p)
{
	struct dt_node *np = p->phb.dt_node;
	uint32_t lsibase, icsp = get_ics_phandle();
	uint64_t m32b, m64b, m64s;

	/* Add various properties that HB doesn't have to
	 * add, some of them simply because they result from
	 * policy decisions made in skiboot rather than in HB
	 * such as the MMIO windows going to PCI, interrupts,
	 * etc...
	 */
	dt_add_property_cells(np, "#address-cells", 3);
	dt_add_property_cells(np, "#size-cells", 2);
	dt_add_property_cells(np, "#interrupt-cells", 1);
	dt_add_property_cells(np, "bus-range", 0, 0xff);
	dt_add_property_cells(np, "clock-frequency", 0x200, 0); /* ??? */

	dt_add_property_cells(np, "interrupt-parent", icsp);

	/* XXX FIXME: add slot-name */
	//dt_property_cell("bus-width", 8); /* Figure it out from VPD ? */

	/* "ranges", we only expose M32 (PHB4 doesn't do IO)
	 *
	 * Note: The kernel expects us to have chopped of 64k from the
	 * M32 size (for the 32-bit MSIs). If we don't do that, it will
	 * get confused (OPAL does it)
	 */
	m32b = cleanup_addr(p->mm1_base);
	m64b = cleanup_addr(p->mm0_base);
	m64s = p->mm0_size;
	dt_add_property_cells(np, "ranges",
			      /* M32 space */
			      0x02000000, 0x00000000, M32_PCI_START,
			      hi32(m32b), lo32(m32b), 0, M32_PCI_SIZE - 0x10000);

	/* XXX FIXME: add opal-memwin32, dmawins, etc... */
	dt_add_property_u64s(np, "ibm,opal-m64-window", m64b, m64b, m64s);
	dt_add_property(np, "ibm,opal-single-pe", NULL, 0);
	dt_add_property_cells(np, "ibm,opal-num-pes", p->num_pes);
	dt_add_property_cells(np, "ibm,opal-reserved-pe",
			      PHB4_RESERVED_PE_NUM(p));
	dt_add_property_cells(np, "ibm,opal-msi-ranges",
			      p->base_msi, p->num_irqs - 8);
	/* M64 ranges start at 1 as MBT0 is used for M32 */
	dt_add_property_cells(np, "ibm,opal-available-m64-ranges",
			      1, p->mbt_size - 1);
	dt_add_property_cells(np, "ibm,supported-tce-sizes",
			      12, // 4K
			      16, // 64K
			      21, // 2M
			      30); // 1G

	/* Tell Linux about alignment limits for segment splits.
	 *
	 * XXX We currently only expose splits of 1 and "num PEs",
	 */
	dt_add_property_cells(np, "ibm,opal-m64-segment-splits",
			      /* Full split, number of segments: */
			      p->num_pes,
			      /* Encoding passed to the enable call */
			      OPAL_ENABLE_M64_SPLIT,
			      /* Alignement/size restriction in #bits*/
			      /* XXX VERIFY VALUE */
			      12,
			      /* Unused */
			      0,
			      /* single PE, number of segments: */
			      1,
			      /* Encoding passed to the enable call */
			      OPAL_ENABLE_M64_NON_SPLIT,
			      /* Alignement/size restriction in #bits*/
			      /* XXX VERIFY VALUE */
			      12,
			      /* Unused */
			      0);

	/* The interrupt maps will be generated in the RC node by the
	 * PCI code based on the content of this structure:
	 */
	lsibase = p->base_lsi;
	p->phb.lstate.int_size = 2;
	p->phb.lstate.int_val[0][0] = lsibase + PHB4_LSI_PCIE_INTA;
	p->phb.lstate.int_val[0][1] = 1;
	p->phb.lstate.int_val[1][0] = lsibase + PHB4_LSI_PCIE_INTB;
	p->phb.lstate.int_val[1][1] = 1;
	p->phb.lstate.int_val[2][0] = lsibase + PHB4_LSI_PCIE_INTC;
	p->phb.lstate.int_val[2][1] = 1;
	p->phb.lstate.int_val[3][0] = lsibase + PHB4_LSI_PCIE_INTD;
	p->phb.lstate.int_val[3][1] = 1;
	p->phb.lstate.int_parent[0] = icsp;
	p->phb.lstate.int_parent[1] = icsp;
	p->phb.lstate.int_parent[2] = icsp;
	p->phb.lstate.int_parent[3] = icsp;

	/* Indicators for variable tables */
	dt_add_property_cells(np, "ibm,opal-rtt-table",
		hi32(p->tbl_rtt), lo32(p->tbl_rtt), RTT_TABLE_SIZE);
	dt_add_property_cells(np, "ibm,opal-peltv-table",
		hi32(p->tbl_peltv), lo32(p->tbl_peltv), p->tbl_peltv_size);
	dt_add_property_cells(np, "ibm,opal-pest-table",
		hi32(p->tbl_pest), lo32(p->tbl_pest), p->tbl_pest_size);

	dt_add_property_cells(np, "ibm,phb-diag-data-size",
			      sizeof(struct OpalIoPhb4ErrorData));

	/* Indicate to Linux that CAPP timebase sync is supported */
	dt_add_property_string(np, "ibm,capp-timebase-sync", NULL);

	/* Tell Linux Compare/Mask indication values */
	dt_add_property_cells(np, "ibm,phb-indications", CAPIIND, ASNIND,
			      NBWIND);
}

static bool phb4_calculate_windows(struct phb4 *p)
{
	const struct dt_property *prop;

	/* Get PBCQ MMIO windows from device-tree */
	prop = dt_require_property(p->phb.dt_node,
				   "ibm,mmio-windows", -1);
	assert(prop->len >= (2 * sizeof(uint64_t)));

	p->mm0_base = ((const uint64_t *)prop->prop)[0];
	p->mm0_size = ((const uint64_t *)prop->prop)[1];
	if (prop->len > 16) {
		p->mm1_base = ((const uint64_t *)prop->prop)[2];
		p->mm1_size = ((const uint64_t *)prop->prop)[3];
	}

	/* Sort them so that 0 is big and 1 is small */
	if (p->mm1_size && p->mm1_size > p->mm0_size) {
		uint64_t b = p->mm0_base;
		uint64_t s = p->mm0_size;
		p->mm0_base = p->mm1_base;
		p->mm0_size = p->mm1_size;
		p->mm1_base = b;
		p->mm1_size = s;
	}

	/* If 1 is too small, ditch it */
	if (p->mm1_size < M32_PCI_SIZE)
		p->mm1_size = 0;

	/* If 1 doesn't exist, carve it out of 0 */
	if (p->mm1_size == 0) {
		p->mm0_size /= 2;
		p->mm1_base = p->mm0_base + p->mm0_size;
		p->mm1_size = p->mm0_size;
	}

	/* Crop mm1 to our desired size */
	if (p->mm1_size > M32_PCI_SIZE)
		p->mm1_size = M32_PCI_SIZE;

	return true;
}

static void phb4_err_interrupt(struct irq_source *is, uint32_t isn)
{
	struct phb4 *p = is->data;

	PHBDBG(p, "Got interrupt 0x%08x\n", isn);

#if 0
	/* Update pending event */
	opal_update_pending_evt(OPAL_EVENT_PCI_ERROR,
				OPAL_EVENT_PCI_ERROR);

	/* If the PHB is broken, go away */
	if (p->broken)
		return;

	/*
	 * Mark the PHB has pending error so that the OS
	 * can handle it at late point.
	 */
	phb3_set_err_pending(p, true);
#endif
}

static uint64_t phb4_lsi_attributes(struct irq_source *is __unused,
				uint32_t isn __unused)
{
#ifndef DISABLE_ERR_INTS
	struct phb3 *p = is->data;
	uint32_t idx = isn - p->base_lsi;

	if (idx == PHB3_LSI_PCIE_INF || idx == PHB3_LSI_PCIE_ER)
		return IRQ_ATTR_TARGET_OPAL | IRQ_ATTR_TARGET_RARE;
#endif
	return IRQ_ATTR_TARGET_LINUX;
}

static int64_t phb4_ndd1_lsi_set_xive(struct irq_source *is, uint32_t isn,
				     uint16_t server, uint8_t priority)
{
	struct phb4 *p = is->data;
	uint32_t idx = isn - p->base_lsi;

	if (idx > 8)
		return OPAL_PARAMETER;

	phb_lock(&p->phb);

	phb4_ioda_sel(p, IODA3_TBL_LIST, idx, false);

	/* Mask using P=0,Q=1, unmask using P=1,Q=0 followed by EOI */
	/* XXX FIXME: A quick mask/umask can make us shoot an interrupt
	 * more than once to a queue. We need to keep track better.
	 *
	 * Thankfully, this is only on Nimubs DD1 and for LSIs, so
	 * will go away soon enough.
	 */
	if (priority == 0xff)
		out_be64(p->regs + PHB_IODA_DATA0, IODA3_LIST_Q);
	else {
		out_be64(p->regs + PHB_IODA_DATA0, IODA3_LIST_P);
		__irq_source_eoi(is, isn);
	}

	phb_unlock(&p->phb);

	return 0;
}

static const struct irq_source_ops phb4_ndd1_lsi_ops = {
	.set_xive = phb4_ndd1_lsi_set_xive,
	.interrupt = phb4_err_interrupt,
	.attributes = phb4_lsi_attributes,
};

static const struct irq_source_ops phb4_lsi_ops = {
	.interrupt = phb4_err_interrupt,
	.attributes = phb4_lsi_attributes,
};

#ifdef HAVE_BIG_ENDIAN
static u64 lane_eq_default[8] = {
	0x5454545454545454, 0x5454545454545454,
	0x5454545454545454, 0x5454545454545454,
	0x7777777777777777, 0x7777777777777777,
	0x7777777777777777, 0x7777777777777777
};
#else
#error lane_eq_default needs to be big endian (device tree property)
#endif

static void phb4_create(struct dt_node *np)
{
	const struct dt_property *prop;
	struct phb4 *p = zalloc(sizeof(struct phb4));
	struct pci_slot *slot;
	size_t lane_eq_len, lane_eq_len_req;
	struct dt_node *iplp;
	char *path;
	uint32_t irq_base, irq_flags;
	int i;
	struct proc_chip *chip;

	assert(p);

	/* Populate base stuff */
	p->index = dt_prop_get_u32(np, "ibm,phb-index");
	p->chip_id = dt_prop_get_u32(np, "ibm,chip-id");
	chip = get_chip(p->chip_id);
	p->regs = (void *)dt_get_address(np, 0, NULL);
	p->int_mmio = (void *)dt_get_address(np, 1, NULL);
	p->phb.dt_node = np;
	p->phb.ops = &phb4_ops;
	p->phb.phb_type = phb_type_pcie_v4;
	p->phb.scan_map = 0x1; /* Only device 0 to scan */

	if (!phb4_calculate_windows(p))
		return;

	/* Get the various XSCOM register bases from the device-tree */
	prop = dt_require_property(np, "ibm,xscom-bases", 5 * sizeof(uint32_t));
	p->pe_xscom = ((const uint32_t *)prop->prop)[0];
	p->pe_stk_xscom = ((const uint32_t *)prop->prop)[1];
	p->pci_xscom = ((const uint32_t *)prop->prop)[2];
	p->pci_stk_xscom = ((const uint32_t *)prop->prop)[3];
	p->etu_xscom = ((const uint32_t *)prop->prop)[4];

	/*
	 * We skip the initial PERST assertion requested by the generic code
	 * when doing a cold boot because we are coming out of cold boot already
	 * so we save boot time that way. The PERST state machine will still
	 * handle waiting for the link to come up, it will just avoid actually
	 * asserting & deasserting the PERST output
	 *
	 * For a hot IPL, we still do a PERST
	 *
	 * Note: In absence of property (ie, FSP-less), we stick to the old
	 * behaviour and set skip_perst to true
	 */
	p->skip_perst = true; /* Default */

	iplp = dt_find_by_path(dt_root, "ipl-params/ipl-params");
	if (iplp) {
		const char *ipl_type = dt_prop_get_def(iplp, "cec-major-type", NULL);
		if (ipl_type && (!strcmp(ipl_type, "hot")))
			p->skip_perst = false;
	}

	/* By default link is assumed down */
	p->has_link = false;

	/* We register the PHB before we initialize it so we
	 * get a useful OPAL ID for it
	 */
	pci_register_phb(&p->phb, phb4_get_opal_id(p->chip_id, p->index));

	/* Create slot structure */
	slot = phb4_slot_create(&p->phb);
	if (!slot)
		PHBERR(p, "Cannot create PHB slot\n");

	/* Hello ! */
	path = dt_get_path(np);
	PHBINF(p, "Found %s @%p\n", path, p->regs);
	PHBINF(p, "  M32 [0x%016llx..0x%016llx]\n",
	       p->mm1_base, p->mm1_base + p->mm1_size - 1);
	PHBINF(p, "  M64 [0x%016llx..0x%016llx]\n",
	       p->mm0_base, p->mm0_base + p->mm0_size - 1);
	free(path);

	/* Find base location code from root node */
	p->phb.base_loc_code = dt_prop_get_def(dt_root,
					       "ibm,io-base-loc-code", NULL);
	if (!p->phb.base_loc_code)
		PHBDBG(p, "Base location code not found !\n");

	/*
	 * Grab CEC IO VPD load info from the root of the device-tree,
	 * on P8 there's a single such VPD for the whole machine
	 */
	prop = dt_find_property(dt_root, "ibm,io-vpd");
	if (!prop) {
		/* LX VPD Lid not already loaded */
		vpd_iohub_load(dt_root);
	}

	/* Obtain informatin about the PHB from the hardware directly */
	if (!phb4_read_capabilities(p))
		goto failed;

	/* Priority order: NVRAM -> dt -> GEN2 dd1 -> GEN3 dd2.00 -> GEN4 */
	p->max_link_speed = 4;
	if (p->rev == PHB4_REV_NIMBUS_DD10)
		p->max_link_speed = 2;
	if (p->rev == PHB4_REV_NIMBUS_DD20 &&
	    ((0xf & chip->ec_level) == 0) && chip->ec_rev == 0)
		p->max_link_speed = 3;
	if (dt_has_node_property(np, "ibm,max-link-speed", NULL))
		p->max_link_speed = dt_prop_get_u32(np, "ibm,max-link-speed");
	if (pcie_max_link_speed)
		p->max_link_speed = pcie_max_link_speed;
	if (p->max_link_speed > 4) /* clamp to 4 */
		p->max_link_speed = 4;
	PHBINF(p, "Max link speed: GEN%i\n", p->max_link_speed);

	/* Check for lane equalization values from HB or HDAT */
	p->lane_eq_en = true;
	p->lane_eq = dt_prop_get_def_size(np, "ibm,lane-eq", NULL, &lane_eq_len);
	if (p->rev == PHB4_REV_NIMBUS_DD10)
		lane_eq_len_req = 8 * 8;
	else
		lane_eq_len_req = 6 * 8;
	if (p->lane_eq) {
		if (lane_eq_len < lane_eq_len_req) {
			PHBERR(p, "Device-tree has ibm,lane-eq too short: %ld"
			       " (want %ld)\n", lane_eq_len, lane_eq_len_req);
			p->lane_eq = NULL;
		}
	} else {
		PHBDBG(p, "Using default lane equalization settings\n");
		p->lane_eq = lane_eq_default;
	}
	if (p->lane_eq) {
		PHBDBG(p, "Override lane equalization settings:\n");
		for (i = 0 ; i < lane_eq_len_req/(8 * 2) ; i++)
			PHBDBG(p, "  0x%016llx 0x%016llx\n",
			       be64_to_cpu(p->lane_eq[2 * i]),
			       be64_to_cpu(p->lane_eq[2 * i + 1]));
	}

	/* Allocate a block of interrupts. We need to know if it needs
	 * 2K or 4K interrupts ... for now we just use 4K but that
	 * needs to be fixed
	 */
	irq_base = xive_alloc_hw_irqs(p->chip_id, p->num_irqs, p->num_irqs);
	if (irq_base == XIVE_IRQ_ERROR) {
		PHBERR(p, "Failed to allocate %d interrupt sources\n",
		       p->num_irqs);
		goto failed;
	}
	p->base_msi = irq_base;
	p->base_lsi = irq_base + p->num_irqs - 8;
	p->irq_port = xive_get_notify_port(p->chip_id,
					   XIVE_HW_SRC_PHBn(p->index));

	if (p->rev == PHB4_REV_NIMBUS_DD10)
		p->num_pes = p->max_num_pes/2;
	else
		p->num_pes = p->max_num_pes;

	/* Allocate the SkiBoot internal in-memory tables for the PHB */
	phb4_allocate_tables(p);

	phb4_add_properties(p);

	/* Clear IODA3 cache */
	phb4_init_ioda_cache(p);

	/* Get the HW up and running */
	phb4_init_hw(p, true);

	/* Load capp microcode into capp unit */
	load_capp_ucode(p);

	/* Compute XIVE source flags depending on PHB revision */
	irq_flags = 0;
	if (p->rev == PHB4_REV_NIMBUS_DD10)
		irq_flags |= XIVE_SRC_SHIFT_BUG;
	if (PHB4_CAN_STORE_EOI(p))
		irq_flags |= XIVE_SRC_STORE_EOI;
	else
		irq_flags |= XIVE_SRC_TRIGGER_PAGE;

	/* Register all interrupt sources with XIVE */
	xive_register_hw_source(p->base_msi, p->num_irqs - 8, 16,
				p->int_mmio, irq_flags, NULL, NULL);

	xive_register_hw_source(p->base_lsi, 8, 16,
				p->int_mmio + ((p->num_irqs - 8) << 16),
				XIVE_SRC_LSI | XIVE_SRC_SHIFT_BUG,
				p,
				(p->rev == PHB4_REV_NIMBUS_DD10) ?
				&phb4_ndd1_lsi_ops : &phb4_lsi_ops);

	/* Platform additional setup */
	if (platform.pci_setup_phb)
		platform.pci_setup_phb(&p->phb, p->index);

	dt_add_property_string(np, "status", "okay");

	return;

 failed:
	p->broken = true;

	/* Tell Linux it's broken */
	dt_add_property_string(np, "status", "error");
}

static void phb4_probe_stack(struct dt_node *stk_node, uint32_t pec_index,
			     uint32_t nest_base, uint32_t pci_base)
{
	uint32_t pci_stack, nest_stack, etu_base, gcid, phb_num, stk_index;
	uint64_t val, phb_bar = 0, irq_bar = 0, bar_en;
	uint64_t mmio0_bar = 0, mmio0_bmask, mmio0_sz;
	uint64_t mmio1_bar, mmio1_bmask, mmio1_sz;
	uint64_t reg[4];
	void *foo;
	uint64_t mmio_win[4];
	unsigned int mmio_win_sz;
	struct dt_node *np;
	char *path;
	uint64_t capp_ucode_base;
	unsigned int max_link_speed;

	gcid = dt_get_chip_id(stk_node);
	stk_index = dt_prop_get_u32(stk_node, "reg");
	phb_num = dt_prop_get_u32(stk_node, "ibm,phb-index");
	path = dt_get_path(stk_node);
	prlog(PR_INFO, "PHB: Chip %d Found PHB4 PBCQ%d Stack %d at %s\n",
	      gcid, pec_index, stk_index, path);
	free(path);

	pci_stack = pci_base + 0x40 * (stk_index + 1);
	nest_stack = nest_base + 0x40 * (stk_index + 1);
	etu_base = pci_base + 0x100 + 0x40 * stk_index;

	prlog(PR_DEBUG, "PHB[%d:%d] X[PE]=0x%08x/0x%08x X[PCI]=0x%08x/0x%08x X[ETU]=0x%08x\n",
	      gcid, phb_num, nest_base, nest_stack, pci_base, pci_stack, etu_base);

	/* Default BAR enables */
	bar_en = 0;

	/* Initialize PHB register BAR */
	phys_map_get(gcid, PHB4_REG_SPC, phb_num, &phb_bar, NULL);
	xscom_write(gcid, nest_stack + XPEC_NEST_STK_PHB_REG_BAR, phb_bar << 8);
	bar_en |= XPEC_NEST_STK_BAR_EN_PHB;


	/* Same with INT BAR (ESB) */
	phys_map_get(gcid, PHB4_XIVE_ESB, phb_num, &irq_bar, NULL);
	xscom_write(gcid, nest_stack + XPEC_NEST_STK_IRQ_BAR, irq_bar << 8);
	bar_en |= XPEC_NEST_STK_BAR_EN_INT;


	/* Same with MMIO windows */
	phys_map_get(gcid, PHB4_64BIT_MMIO, phb_num, &mmio0_bar, &mmio0_sz);
	mmio0_bmask =  (~(mmio0_sz - 1)) & 0x00FFFFFFFFFFFFFFULL;
	xscom_write(gcid, nest_stack + XPEC_NEST_STK_MMIO_BAR0, mmio0_bar << 8);
	xscom_write(gcid, nest_stack + XPEC_NEST_STK_MMIO_BAR0_MASK, mmio0_bmask << 8);

	phys_map_get(gcid, PHB4_32BIT_MMIO, phb_num, &mmio1_bar, &mmio1_sz);
	mmio1_bmask =  (~(mmio1_sz - 1)) & 0x00FFFFFFFFFFFFFFULL;
	xscom_write(gcid, nest_stack + XPEC_NEST_STK_MMIO_BAR1, mmio1_bar << 8);
	xscom_write(gcid, nest_stack + XPEC_NEST_STK_MMIO_BAR1_MASK, mmio1_bmask << 8);
	bar_en |= XPEC_NEST_STK_BAR_EN_MMIO0 | XPEC_NEST_STK_BAR_EN_MMIO1;

	/* Build MMIO windows list */
	mmio_win_sz = 0;
	if (mmio0_bar) {
		mmio_win[mmio_win_sz++] = mmio0_bar;
		mmio_win[mmio_win_sz++] = mmio0_sz;
		bar_en |= XPEC_NEST_STK_BAR_EN_MMIO0;
	}
	if (mmio1_bar) {
		mmio_win[mmio_win_sz++] = mmio1_bar;
		mmio_win[mmio_win_sz++] = mmio1_sz;
		bar_en |= XPEC_NEST_STK_BAR_EN_MMIO1;
	}

	/* Set the appropriate enables */
	xscom_read(gcid, nest_stack + XPEC_NEST_STK_BAR_EN, &val);
	val |= bar_en;
	xscom_write(gcid, nest_stack + XPEC_NEST_STK_BAR_EN, val);

	/* No MMIO windows ? Barf ! */
	if (mmio_win_sz == 0) {
		prerror("PHB[%d:%d] No MMIO windows enabled !\n", gcid, phb_num);
		return;
	}

	/* Clear errors in PFIR and NFIR */
	xscom_write(gcid, pci_stack + XPEC_PCI_STK_PCI_FIR, 0);
	xscom_write(gcid, nest_stack + XPEC_NEST_STK_PCI_NFIR, 0);

	/* Check ETU reset */
	xscom_read(gcid, pci_stack + XPEC_PCI_STK_ETU_RESET, &val);
	prlog_once(PR_DEBUG, "ETU reset: %llx\n", val);
	xscom_write(gcid, pci_stack + XPEC_PCI_STK_ETU_RESET, 0);
	time_wait_ms(1);

	// show we can read phb mmio space
	foo = (void *)(phb_bar + 0x800); // phb version register
	prlog_once(PR_DEBUG, "Version reg: 0x%016llx\n", in_be64(foo));

	/* Create PHB node */
	reg[0] = phb_bar;
	reg[1] = 0x1000;
	reg[2] = irq_bar;
	reg[3] = 0x10000000;

	np = dt_new_addr(dt_root, "pciex", reg[0]);
	if (!np)
		return;

	dt_add_property_strings(np, "compatible", "ibm,power9-pciex", "ibm,ioda3-phb");
	dt_add_property_strings(np, "device_type", "pciex");
	dt_add_property(np, "reg", reg, sizeof(reg));

	/* Everything else is handled later by skiboot, we just
	 * stick a few hints here
	 */
	dt_add_property_cells(np, "ibm,xscom-bases",
			      nest_base, nest_stack, pci_base, pci_stack, etu_base);
	dt_add_property(np, "ibm,mmio-windows", mmio_win, 8 * mmio_win_sz);
	dt_add_property_cells(np, "ibm,phb-index", phb_num);
	dt_add_property_cells(np, "ibm,phb-stack", stk_node->phandle);
	dt_add_property_cells(np, "ibm,phb-stack-index", stk_index);
	dt_add_property_cells(np, "ibm,chip-id", gcid);
	if (dt_has_node_property(stk_node, "ibm,hub-id", NULL))
		dt_add_property_cells(np, "ibm,hub-id",
				      dt_prop_get_u32(stk_node, "ibm,hub-id"));
	if (dt_has_node_property(stk_node, "ibm,loc-code", NULL)) {
		const char *lc = dt_prop_get(stk_node, "ibm,loc-code");
		dt_add_property_string(np, "ibm,loc-code", lc);
	}
	if (dt_has_node_property(stk_node, "ibm,lane-eq", NULL)) {
		size_t leq_size;
		const void *leq = dt_prop_get_def_size(stk_node, "ibm,lane-eq",
						       NULL, &leq_size);
		if (leq != NULL && leq_size >= 6 * 8)
			dt_add_property(np, "ibm,lane-eq", leq, leq_size);
	}
	if (dt_has_node_property(stk_node, "ibm,capp-ucode", NULL)) {
		capp_ucode_base = dt_prop_get_u32(stk_node, "ibm,capp-ucode");
		dt_add_property_cells(np, "ibm,capp-ucode", capp_ucode_base);
	}
	if (dt_has_node_property(stk_node, "ibm,max-link-speed", NULL)) {
		max_link_speed = dt_prop_get_u32(stk_node, "ibm,max-link-speed");
		dt_add_property_cells(np, "ibm,max-link-speed", max_link_speed);
	}
	dt_add_property_cells(np, "ibm,capi-flags",
			      OPAL_PHB_CAPI_FLAG_SNOOP_CONTROL);

	add_chip_dev_associativity(np);
}

static void phb4_probe_pbcq(struct dt_node *pbcq)
{
	uint32_t nest_base, pci_base, pec_index;
	struct dt_node *stk;

	nest_base = dt_get_address(pbcq, 0, NULL);
	pci_base = dt_get_address(pbcq, 1, NULL);
	pec_index = dt_prop_get_u32(pbcq, "ibm,pec-index");

	dt_for_each_child(pbcq, stk) {
		if (dt_node_is_enabled(stk))
			phb4_probe_stack(stk, pec_index, nest_base, pci_base);
	}
}

void probe_phb4(void)
{
	struct dt_node *np;

	verbose_eeh = nvram_query_eq("pci-eeh-verbose", "true");
	/* REMOVEME: force this for now until we stabalise PCIe */
	verbose_eeh = 1;
	if (verbose_eeh)
		prlog(PR_INFO, "PHB4: Verbose EEH enabled\n");

	pci_tracing = nvram_query_eq("pci-tracing", "true");
	pci_eeh_mmio = !nvram_query_eq("pci-eeh-mmio", "disabled");
	pci_retry_all = nvram_query_eq("pci-retry-all", "true");

	/* Look for PBCQ XSCOM nodes */
	dt_for_each_compatible(dt_root, np, "ibm,power9-pbcq")
		phb4_probe_pbcq(np);

	/* Look for newly created PHB nodes */
	dt_for_each_compatible(dt_root, np, "ibm,power9-pciex")
		phb4_create(np);
}
OpenPOWER on IntegriCloud