summaryrefslogtreecommitdiff
path: root/arch/x86/platform/uv/tlb_uv.c
blob: 4ea69690c3e49a6afbfa2035cb81b831362a3963 (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
// SPDX-License-Identifier: GPL-2.0-or-later
/*
 *	SGI UltraViolet TLB flush routines.
 *
 *	(c) 2008-2014 Cliff Wickman <cpw@sgi.com>, SGI.
 */
#include <linux/seq_file.h>
#include <linux/proc_fs.h>
#include <linux/debugfs.h>
#include <linux/kernel.h>
#include <linux/slab.h>
#include <linux/delay.h>

#include <asm/mmu_context.h>
#include <asm/uv/uv.h>
#include <asm/uv/uv_mmrs.h>
#include <asm/uv/uv_hub.h>
#include <asm/uv/uv_bau.h>
#include <asm/apic.h>
#include <asm/tsc.h>
#include <asm/irq_vectors.h>
#include <asm/timer.h>

static struct bau_operations ops __ro_after_init;

/* timeouts in nanoseconds (indexed by UVH_AGING_PRESCALE_SEL urgency7 30:28) */
static const int timeout_base_ns[] = {
		20,
		160,
		1280,
		10240,
		81920,
		655360,
		5242880,
		167772160
};

static int timeout_us;
static bool nobau = true;
static int nobau_perm;

/* tunables: */
static int max_concurr		= MAX_BAU_CONCURRENT;
static int max_concurr_const	= MAX_BAU_CONCURRENT;
static int plugged_delay	= PLUGGED_DELAY;
static int plugsb4reset		= PLUGSB4RESET;
static int giveup_limit		= GIVEUP_LIMIT;
static int timeoutsb4reset	= TIMEOUTSB4RESET;
static int ipi_reset_limit	= IPI_RESET_LIMIT;
static int complete_threshold	= COMPLETE_THRESHOLD;
static int congested_respns_us	= CONGESTED_RESPONSE_US;
static int congested_reps	= CONGESTED_REPS;
static int disabled_period	= DISABLED_PERIOD;

static struct tunables tunables[] = {
	{&max_concurr,           MAX_BAU_CONCURRENT}, /* must be [0] */
	{&plugged_delay,         PLUGGED_DELAY},
	{&plugsb4reset,          PLUGSB4RESET},
	{&timeoutsb4reset,       TIMEOUTSB4RESET},
	{&ipi_reset_limit,       IPI_RESET_LIMIT},
	{&complete_threshold,    COMPLETE_THRESHOLD},
	{&congested_respns_us,   CONGESTED_RESPONSE_US},
	{&congested_reps,        CONGESTED_REPS},
	{&disabled_period,       DISABLED_PERIOD},
	{&giveup_limit,          GIVEUP_LIMIT}
};

static struct dentry *tunables_dir;

/* these correspond to the statistics printed by ptc_seq_show() */
static char *stat_description[] = {
	"sent:     number of shootdown messages sent",
	"stime:    time spent sending messages",
	"numuvhubs: number of hubs targeted with shootdown",
	"numuvhubs16: number times 16 or more hubs targeted",
	"numuvhubs8: number times 8 or more hubs targeted",
	"numuvhubs4: number times 4 or more hubs targeted",
	"numuvhubs2: number times 2 or more hubs targeted",
	"numuvhubs1: number times 1 hub targeted",
	"numcpus:  number of cpus targeted with shootdown",
	"dto:      number of destination timeouts",
	"retries:  destination timeout retries sent",
	"rok:   :  destination timeouts successfully retried",
	"resetp:   ipi-style resource resets for plugs",
	"resett:   ipi-style resource resets for timeouts",
	"giveup:   fall-backs to ipi-style shootdowns",
	"sto:      number of source timeouts",
	"bz:       number of stay-busy's",
	"throt:    number times spun in throttle",
	"swack:   image of UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE",
	"recv:     shootdown messages received",
	"rtime:    time spent processing messages",
	"all:      shootdown all-tlb messages",
	"one:      shootdown one-tlb messages",
	"mult:     interrupts that found multiple messages",
	"none:     interrupts that found no messages",
	"retry:    number of retry messages processed",
	"canc:     number messages canceled by retries",
	"nocan:    number retries that found nothing to cancel",
	"reset:    number of ipi-style reset requests processed",
	"rcan:     number messages canceled by reset requests",
	"disable:  number times use of the BAU was disabled",
	"enable:   number times use of the BAU was re-enabled"
};

static int __init setup_bau(char *arg)
{
	int result;

	if (!arg)
		return -EINVAL;

	result = strtobool(arg, &nobau);
	if (result)
		return result;

	/* we need to flip the logic here, so that bau=y sets nobau to false */
	nobau = !nobau;

	if (!nobau)
		pr_info("UV BAU Enabled\n");
	else
		pr_info("UV BAU Disabled\n");

	return 0;
}
early_param("bau", setup_bau);

/* base pnode in this partition */
static int uv_base_pnode __read_mostly;

static DEFINE_PER_CPU(struct ptc_stats, ptcstats);
static DEFINE_PER_CPU(struct bau_control, bau_control);
static DEFINE_PER_CPU(cpumask_var_t, uv_flush_tlb_mask);

static void
set_bau_on(void)
{
	int cpu;
	struct bau_control *bcp;

	if (nobau_perm) {
		pr_info("BAU not initialized; cannot be turned on\n");
		return;
	}
	nobau = false;
	for_each_present_cpu(cpu) {
		bcp = &per_cpu(bau_control, cpu);
		bcp->nobau = false;
	}
	pr_info("BAU turned on\n");
	return;
}

static void
set_bau_off(void)
{
	int cpu;
	struct bau_control *bcp;

	nobau = true;
	for_each_present_cpu(cpu) {
		bcp = &per_cpu(bau_control, cpu);
		bcp->nobau = true;
	}
	pr_info("BAU turned off\n");
	return;
}

/*
 * Determine the first node on a uvhub. 'Nodes' are used for kernel
 * memory allocation.
 */
static int __init uvhub_to_first_node(int uvhub)
{
	int node, b;

	for_each_online_node(node) {
		b = uv_node_to_blade_id(node);
		if (uvhub == b)
			return node;
	}
	return -1;
}

/*
 * Determine the apicid of the first cpu on a uvhub.
 */
static int __init uvhub_to_first_apicid(int uvhub)
{
	int cpu;

	for_each_present_cpu(cpu)
		if (uvhub == uv_cpu_to_blade_id(cpu))
			return per_cpu(x86_cpu_to_apicid, cpu);
	return -1;
}

/*
 * Free a software acknowledge hardware resource by clearing its Pending
 * bit. This will return a reply to the sender.
 * If the message has timed out, a reply has already been sent by the
 * hardware but the resource has not been released. In that case our
 * clear of the Timeout bit (as well) will free the resource. No reply will
 * be sent (the hardware will only do one reply per message).
 */
static void reply_to_message(struct msg_desc *mdp, struct bau_control *bcp,
						int do_acknowledge)
{
	unsigned long dw;
	struct bau_pq_entry *msg;

	msg = mdp->msg;
	if (!msg->canceled && do_acknowledge) {
		dw = (msg->swack_vec << UV_SW_ACK_NPENDING) | msg->swack_vec;
		ops.write_l_sw_ack(dw);
	}
	msg->replied_to = 1;
	msg->swack_vec = 0;
}

/*
 * Process the receipt of a RETRY message
 */
static void bau_process_retry_msg(struct msg_desc *mdp,
					struct bau_control *bcp)
{
	int i;
	int cancel_count = 0;
	unsigned long msg_res;
	unsigned long mmr = 0;
	struct bau_pq_entry *msg = mdp->msg;
	struct bau_pq_entry *msg2;
	struct ptc_stats *stat = bcp->statp;

	stat->d_retries++;
	/*
	 * cancel any message from msg+1 to the retry itself
	 */
	for (msg2 = msg+1, i = 0; i < DEST_Q_SIZE; msg2++, i++) {
		if (msg2 > mdp->queue_last)
			msg2 = mdp->queue_first;
		if (msg2 == msg)
			break;

		/* same conditions for cancellation as do_reset */
		if ((msg2->replied_to == 0) && (msg2->canceled == 0) &&
		    (msg2->swack_vec) && ((msg2->swack_vec &
			msg->swack_vec) == 0) &&
		    (msg2->sending_cpu == msg->sending_cpu) &&
		    (msg2->msg_type != MSG_NOOP)) {
			mmr = ops.read_l_sw_ack();
			msg_res = msg2->swack_vec;
			/*
			 * This is a message retry; clear the resources held
			 * by the previous message only if they timed out.
			 * If it has not timed out we have an unexpected
			 * situation to report.
			 */
			if (mmr & (msg_res << UV_SW_ACK_NPENDING)) {
				unsigned long mr;
				/*
				 * Is the resource timed out?
				 * Make everyone ignore the cancelled message.
				 */
				msg2->canceled = 1;
				stat->d_canceled++;
				cancel_count++;
				mr = (msg_res << UV_SW_ACK_NPENDING) | msg_res;
				ops.write_l_sw_ack(mr);
			}
		}
	}
	if (!cancel_count)
		stat->d_nocanceled++;
}

/*
 * Do all the things a cpu should do for a TLB shootdown message.
 * Other cpu's may come here at the same time for this message.
 */
static void bau_process_message(struct msg_desc *mdp, struct bau_control *bcp,
						int do_acknowledge)
{
	short socket_ack_count = 0;
	short *sp;
	struct atomic_short *asp;
	struct ptc_stats *stat = bcp->statp;
	struct bau_pq_entry *msg = mdp->msg;
	struct bau_control *smaster = bcp->socket_master;

	/*
	 * This must be a normal message, or retry of a normal message
	 */
	if (msg->address == TLB_FLUSH_ALL) {
		flush_tlb_local();
		stat->d_alltlb++;
	} else {
		flush_tlb_one_user(msg->address);
		stat->d_onetlb++;
	}
	stat->d_requestee++;

	/*
	 * One cpu on each uvhub has the additional job on a RETRY
	 * of releasing the resource held by the message that is
	 * being retried.  That message is identified by sending
	 * cpu number.
	 */
	if (msg->msg_type == MSG_RETRY && bcp == bcp->uvhub_master)
		bau_process_retry_msg(mdp, bcp);

	/*
	 * This is a swack message, so we have to reply to it.
	 * Count each responding cpu on the socket. This avoids
	 * pinging the count's cache line back and forth between
	 * the sockets.
	 */
	sp = &smaster->socket_acknowledge_count[mdp->msg_slot];
	asp = (struct atomic_short *)sp;
	socket_ack_count = atom_asr(1, asp);
	if (socket_ack_count == bcp->cpus_in_socket) {
		int msg_ack_count;
		/*
		 * Both sockets dump their completed count total into
		 * the message's count.
		 */
		*sp = 0;
		asp = (struct atomic_short *)&msg->acknowledge_count;
		msg_ack_count = atom_asr(socket_ack_count, asp);

		if (msg_ack_count == bcp->cpus_in_uvhub) {
			/*
			 * All cpus in uvhub saw it; reply
			 * (unless we are in the UV2 workaround)
			 */
			reply_to_message(mdp, bcp, do_acknowledge);
		}
	}

	return;
}

/*
 * Determine the first cpu on a pnode.
 */
static int pnode_to_first_cpu(int pnode, struct bau_control *smaster)
{
	int cpu;
	struct hub_and_pnode *hpp;

	for_each_present_cpu(cpu) {
		hpp = &smaster->thp[cpu];
		if (pnode == hpp->pnode)
			return cpu;
	}
	return -1;
}

/*
 * Last resort when we get a large number of destination timeouts is
 * to clear resources held by a given cpu.
 * Do this with IPI so that all messages in the BAU message queue
 * can be identified by their nonzero swack_vec field.
 *
 * This is entered for a single cpu on the uvhub.
 * The sender want's this uvhub to free a specific message's
 * swack resources.
 */
static void do_reset(void *ptr)
{
	int i;
	struct bau_control *bcp = &per_cpu(bau_control, smp_processor_id());
	struct reset_args *rap = (struct reset_args *)ptr;
	struct bau_pq_entry *msg;
	struct ptc_stats *stat = bcp->statp;

	stat->d_resets++;
	/*
	 * We're looking for the given sender, and
	 * will free its swack resource.
	 * If all cpu's finally responded after the timeout, its
	 * message 'replied_to' was set.
	 */
	for (msg = bcp->queue_first, i = 0; i < DEST_Q_SIZE; msg++, i++) {
		unsigned long msg_res;
		/* do_reset: same conditions for cancellation as
		   bau_process_retry_msg() */
		if ((msg->replied_to == 0) &&
		    (msg->canceled == 0) &&
		    (msg->sending_cpu == rap->sender) &&
		    (msg->swack_vec) &&
		    (msg->msg_type != MSG_NOOP)) {
			unsigned long mmr;
			unsigned long mr;
			/*
			 * make everyone else ignore this message
			 */
			msg->canceled = 1;
			/*
			 * only reset the resource if it is still pending
			 */
			mmr = ops.read_l_sw_ack();
			msg_res = msg->swack_vec;
			mr = (msg_res << UV_SW_ACK_NPENDING) | msg_res;
			if (mmr & msg_res) {
				stat->d_rcanceled++;
				ops.write_l_sw_ack(mr);
			}
		}
	}
	return;
}

/*
 * Use IPI to get all target uvhubs to release resources held by
 * a given sending cpu number.
 */
static void reset_with_ipi(struct pnmask *distribution, struct bau_control *bcp)
{
	int pnode;
	int apnode;
	int maskbits;
	int sender = bcp->cpu;
	cpumask_t *mask = bcp->uvhub_master->cpumask;
	struct bau_control *smaster = bcp->socket_master;
	struct reset_args reset_args;

	reset_args.sender = sender;
	cpumask_clear(mask);
	/* find a single cpu for each uvhub in this distribution mask */
	maskbits = sizeof(struct pnmask) * BITSPERBYTE;
	/* each bit is a pnode relative to the partition base pnode */
	for (pnode = 0; pnode < maskbits; pnode++) {
		int cpu;
		if (!bau_uvhub_isset(pnode, distribution))
			continue;
		apnode = pnode + bcp->partition_base_pnode;
		cpu = pnode_to_first_cpu(apnode, smaster);
		cpumask_set_cpu(cpu, mask);
	}

	/* IPI all cpus; preemption is already disabled */
	smp_call_function_many(mask, do_reset, (void *)&reset_args, 1);
	return;
}

/*
 * Not to be confused with cycles_2_ns() from tsc.c; this gives a relative
 * number, not an absolute. It converts a duration in cycles to a duration in
 * ns.
 */
static inline unsigned long long cycles_2_ns(unsigned long long cyc)
{
	struct cyc2ns_data data;
	unsigned long long ns;

	cyc2ns_read_begin(&data);
	ns = mul_u64_u32_shr(cyc, data.cyc2ns_mul, data.cyc2ns_shift);
	cyc2ns_read_end();

	return ns;
}

/*
 * The reverse of the above; converts a duration in ns to a duration in cycles.
 */
static inline unsigned long long ns_2_cycles(unsigned long long ns)
{
	struct cyc2ns_data data;
	unsigned long long cyc;

	cyc2ns_read_begin(&data);
	cyc = (ns << data.cyc2ns_shift) / data.cyc2ns_mul;
	cyc2ns_read_end();

	return cyc;
}

static inline unsigned long cycles_2_us(unsigned long long cyc)
{
	return cycles_2_ns(cyc) / NSEC_PER_USEC;
}

static inline cycles_t sec_2_cycles(unsigned long sec)
{
	return ns_2_cycles(sec * NSEC_PER_SEC);
}

static inline unsigned long long usec_2_cycles(unsigned long usec)
{
	return ns_2_cycles(usec * NSEC_PER_USEC);
}

/*
 * wait for all cpus on this hub to finish their sends and go quiet
 * leaves uvhub_quiesce set so that no new broadcasts are started by
 * bau_flush_send_and_wait()
 */
static inline void quiesce_local_uvhub(struct bau_control *hmaster)
{
	atom_asr(1, (struct atomic_short *)&hmaster->uvhub_quiesce);
}

/*
 * mark this quiet-requestor as done
 */
static inline void end_uvhub_quiesce(struct bau_control *hmaster)
{
	atom_asr(-1, (struct atomic_short *)&hmaster->uvhub_quiesce);
}

static unsigned long uv1_read_status(unsigned long mmr_offset, int right_shift)
{
	unsigned long descriptor_status;

	descriptor_status = uv_read_local_mmr(mmr_offset);
	descriptor_status >>= right_shift;
	descriptor_status &= UV_ACT_STATUS_MASK;
	return descriptor_status;
}

/*
 * Wait for completion of a broadcast software ack message
 * return COMPLETE, RETRY(PLUGGED or TIMEOUT) or GIVEUP
 */
static int uv1_wait_completion(struct bau_desc *bau_desc,
				struct bau_control *bcp, long try)
{
	unsigned long descriptor_status;
	cycles_t ttm;
	u64 mmr_offset = bcp->status_mmr;
	int right_shift = bcp->status_index;
	struct ptc_stats *stat = bcp->statp;

	descriptor_status = uv1_read_status(mmr_offset, right_shift);
	/* spin on the status MMR, waiting for it to go idle */
	while ((descriptor_status != DS_IDLE)) {
		/*
		 * Our software ack messages may be blocked because
		 * there are no swack resources available.  As long
		 * as none of them has timed out hardware will NACK
		 * our message and its state will stay IDLE.
		 */
		if (descriptor_status == DS_SOURCE_TIMEOUT) {
			stat->s_stimeout++;
			return FLUSH_GIVEUP;
		} else if (descriptor_status == DS_DESTINATION_TIMEOUT) {
			stat->s_dtimeout++;
			ttm = get_cycles();

			/*
			 * Our retries may be blocked by all destination
			 * swack resources being consumed, and a timeout
			 * pending.  In that case hardware returns the
			 * ERROR that looks like a destination timeout.
			 */
			if (cycles_2_us(ttm - bcp->send_message) < timeout_us) {
				bcp->conseccompletes = 0;
				return FLUSH_RETRY_PLUGGED;
			}

			bcp->conseccompletes = 0;
			return FLUSH_RETRY_TIMEOUT;
		} else {
			/*
			 * descriptor_status is still BUSY
			 */
			cpu_relax();
		}
		descriptor_status = uv1_read_status(mmr_offset, right_shift);
	}
	bcp->conseccompletes++;
	return FLUSH_COMPLETE;
}

/*
 * UV2 could have an extra bit of status in the ACTIVATION_STATUS_2 register.
 * But not currently used.
 */
static unsigned long uv2_3_read_status(unsigned long offset, int rshft, int desc)
{
	return ((read_lmmr(offset) >> rshft) & UV_ACT_STATUS_MASK) << 1;
}

/*
 * Entered when a bau descriptor has gone into a permanent busy wait because
 * of a hardware bug.
 * Workaround the bug.
 */
static int handle_uv2_busy(struct bau_control *bcp)
{
	struct ptc_stats *stat = bcp->statp;

	stat->s_uv2_wars++;
	bcp->busy = 1;
	return FLUSH_GIVEUP;
}

static int uv2_3_wait_completion(struct bau_desc *bau_desc,
				struct bau_control *bcp, long try)
{
	unsigned long descriptor_stat;
	cycles_t ttm;
	u64 mmr_offset = bcp->status_mmr;
	int right_shift = bcp->status_index;
	int desc = bcp->uvhub_cpu;
	long busy_reps = 0;
	struct ptc_stats *stat = bcp->statp;

	descriptor_stat = uv2_3_read_status(mmr_offset, right_shift, desc);

	/* spin on the status MMR, waiting for it to go idle */
	while (descriptor_stat != UV2H_DESC_IDLE) {
		if (descriptor_stat == UV2H_DESC_SOURCE_TIMEOUT) {
			/*
			 * A h/w bug on the destination side may
			 * have prevented the message being marked
			 * pending, thus it doesn't get replied to
			 * and gets continually nacked until it times
			 * out with a SOURCE_TIMEOUT.
			 */
			stat->s_stimeout++;
			return FLUSH_GIVEUP;
		} else if (descriptor_stat == UV2H_DESC_DEST_TIMEOUT) {
			ttm = get_cycles();

			/*
			 * Our retries may be blocked by all destination
			 * swack resources being consumed, and a timeout
			 * pending.  In that case hardware returns the
			 * ERROR that looks like a destination timeout.
			 * Without using the extended status we have to
			 * deduce from the short time that this was a
			 * strong nack.
			 */
			if (cycles_2_us(ttm - bcp->send_message) < timeout_us) {
				bcp->conseccompletes = 0;
				stat->s_plugged++;
				/* FLUSH_RETRY_PLUGGED causes hang on boot */
				return FLUSH_GIVEUP;
			}
			stat->s_dtimeout++;
			bcp->conseccompletes = 0;
			/* FLUSH_RETRY_TIMEOUT causes hang on boot */
			return FLUSH_GIVEUP;
		} else {
			busy_reps++;
			if (busy_reps > 1000000) {
				/* not to hammer on the clock */
				busy_reps = 0;
				ttm = get_cycles();
				if ((ttm - bcp->send_message) > bcp->timeout_interval)
					return handle_uv2_busy(bcp);
			}
			/*
			 * descriptor_stat is still BUSY
			 */
			cpu_relax();
		}
		descriptor_stat = uv2_3_read_status(mmr_offset, right_shift, desc);
	}
	bcp->conseccompletes++;
	return FLUSH_COMPLETE;
}

/*
 * Returns the status of current BAU message for cpu desc as a bit field
 * [Error][Busy][Aux]
 */
static u64 read_status(u64 status_mmr, int index, int desc)
{
	u64 stat;

	stat = ((read_lmmr(status_mmr) >> index) & UV_ACT_STATUS_MASK) << 1;
	stat |= (read_lmmr(UVH_LB_BAU_SB_ACTIVATION_STATUS_2) >> desc) & 0x1;

	return stat;
}

static int uv4_wait_completion(struct bau_desc *bau_desc,
				struct bau_control *bcp, long try)
{
	struct ptc_stats *stat = bcp->statp;
	u64 descriptor_stat;
	u64 mmr = bcp->status_mmr;
	int index = bcp->status_index;
	int desc = bcp->uvhub_cpu;

	descriptor_stat = read_status(mmr, index, desc);

	/* spin on the status MMR, waiting for it to go idle */
	while (descriptor_stat != UV2H_DESC_IDLE) {
		switch (descriptor_stat) {
		case UV2H_DESC_SOURCE_TIMEOUT:
			stat->s_stimeout++;
			return FLUSH_GIVEUP;

		case UV2H_DESC_DEST_TIMEOUT:
			stat->s_dtimeout++;
			bcp->conseccompletes = 0;
			return FLUSH_RETRY_TIMEOUT;

		case UV2H_DESC_DEST_STRONG_NACK:
			stat->s_plugged++;
			bcp->conseccompletes = 0;
			return FLUSH_RETRY_PLUGGED;

		case UV2H_DESC_DEST_PUT_ERR:
			bcp->conseccompletes = 0;
			return FLUSH_GIVEUP;

		default:
			/* descriptor_stat is still BUSY */
			cpu_relax();
		}
		descriptor_stat = read_status(mmr, index, desc);
	}
	bcp->conseccompletes++;
	return FLUSH_COMPLETE;
}

/*
 * Our retries are blocked by all destination sw ack resources being
 * in use, and a timeout is pending. In that case hardware immediately
 * returns the ERROR that looks like a destination timeout.
 */
static void destination_plugged(struct bau_desc *bau_desc,
			struct bau_control *bcp,
			struct bau_control *hmaster, struct ptc_stats *stat)
{
	udelay(bcp->plugged_delay);
	bcp->plugged_tries++;

	if (bcp->plugged_tries >= bcp->plugsb4reset) {
		bcp->plugged_tries = 0;

		quiesce_local_uvhub(hmaster);

		spin_lock(&hmaster->queue_lock);
		reset_with_ipi(&bau_desc->distribution, bcp);
		spin_unlock(&hmaster->queue_lock);

		end_uvhub_quiesce(hmaster);

		bcp->ipi_attempts++;
		stat->s_resets_plug++;
	}
}

static void destination_timeout(struct bau_desc *bau_desc,
			struct bau_control *bcp, struct bau_control *hmaster,
			struct ptc_stats *stat)
{
	hmaster->max_concurr = 1;
	bcp->timeout_tries++;
	if (bcp->timeout_tries >= bcp->timeoutsb4reset) {
		bcp->timeout_tries = 0;

		quiesce_local_uvhub(hmaster);

		spin_lock(&hmaster->queue_lock);
		reset_with_ipi(&bau_desc->distribution, bcp);
		spin_unlock(&hmaster->queue_lock);

		end_uvhub_quiesce(hmaster);

		bcp->ipi_attempts++;
		stat->s_resets_timeout++;
	}
}

/*
 * Stop all cpus on a uvhub from using the BAU for a period of time.
 * This is reversed by check_enable.
 */
static void disable_for_period(struct bau_control *bcp, struct ptc_stats *stat)
{
	int tcpu;
	struct bau_control *tbcp;
	struct bau_control *hmaster;
	cycles_t tm1;

	hmaster = bcp->uvhub_master;
	spin_lock(&hmaster->disable_lock);
	if (!bcp->baudisabled) {
		stat->s_bau_disabled++;
		tm1 = get_cycles();
		for_each_present_cpu(tcpu) {
			tbcp = &per_cpu(bau_control, tcpu);
			if (tbcp->uvhub_master == hmaster) {
				tbcp->baudisabled = 1;
				tbcp->set_bau_on_time =
					tm1 + bcp->disabled_period;
			}
		}
	}
	spin_unlock(&hmaster->disable_lock);
}

static void count_max_concurr(int stat, struct bau_control *bcp,
				struct bau_control *hmaster)
{
	bcp->plugged_tries = 0;
	bcp->timeout_tries = 0;
	if (stat != FLUSH_COMPLETE)
		return;
	if (bcp->conseccompletes <= bcp->complete_threshold)
		return;
	if (hmaster->max_concurr >= hmaster->max_concurr_const)
		return;
	hmaster->max_concurr++;
}

static void record_send_stats(cycles_t time1, cycles_t time2,
		struct bau_control *bcp, struct ptc_stats *stat,
		int completion_status, int try)
{
	cycles_t elapsed;

	if (time2 > time1) {
		elapsed = time2 - time1;
		stat->s_time += elapsed;

		if ((completion_status == FLUSH_COMPLETE) && (try == 1)) {
			bcp->period_requests++;
			bcp->period_time += elapsed;
			if ((elapsed > usec_2_cycles(bcp->cong_response_us)) &&
			    (bcp->period_requests > bcp->cong_reps) &&
			    ((bcp->period_time / bcp->period_requests) >
					usec_2_cycles(bcp->cong_response_us))) {
				stat->s_congested++;
				disable_for_period(bcp, stat);
			}
		}
	} else
		stat->s_requestor--;

	if (completion_status == FLUSH_COMPLETE && try > 1)
		stat->s_retriesok++;
	else if (completion_status == FLUSH_GIVEUP) {
		stat->s_giveup++;
		if (get_cycles() > bcp->period_end)
			bcp->period_giveups = 0;
		bcp->period_giveups++;
		if (bcp->period_giveups == 1)
			bcp->period_end = get_cycles() + bcp->disabled_period;
		if (bcp->period_giveups > bcp->giveup_limit) {
			disable_for_period(bcp, stat);
			stat->s_giveuplimit++;
		}
	}
}

/*
 * Because of a uv1 hardware bug only a limited number of concurrent
 * requests can be made.
 */
static void uv1_throttle(struct bau_control *hmaster, struct ptc_stats *stat)
{
	spinlock_t *lock = &hmaster->uvhub_lock;
	atomic_t *v;

	v = &hmaster->active_descriptor_count;
	if (!atomic_inc_unless_ge(lock, v, hmaster->max_concurr)) {
		stat->s_throttles++;
		do {
			cpu_relax();
		} while (!atomic_inc_unless_ge(lock, v, hmaster->max_concurr));
	}
}

/*
 * Handle the completion status of a message send.
 */
static void handle_cmplt(int completion_status, struct bau_desc *bau_desc,
			struct bau_control *bcp, struct bau_control *hmaster,
			struct ptc_stats *stat)
{
	if (completion_status == FLUSH_RETRY_PLUGGED)
		destination_plugged(bau_desc, bcp, hmaster, stat);
	else if (completion_status == FLUSH_RETRY_TIMEOUT)
		destination_timeout(bau_desc, bcp, hmaster, stat);
}

/*
 * Send a broadcast and wait for it to complete.
 *
 * The flush_mask contains the cpus the broadcast is to be sent to including
 * cpus that are on the local uvhub.
 *
 * Returns 0 if all flushing represented in the mask was done.
 * Returns 1 if it gives up entirely and the original cpu mask is to be
 * returned to the kernel.
 */
static int uv_flush_send_and_wait(struct cpumask *flush_mask,
				  struct bau_control *bcp,
				  struct bau_desc *bau_desc)
{
	int seq_number = 0;
	int completion_stat = 0;
	int uv1 = 0;
	long try = 0;
	unsigned long index;
	cycles_t time1;
	cycles_t time2;
	struct ptc_stats *stat = bcp->statp;
	struct bau_control *hmaster = bcp->uvhub_master;
	struct uv1_bau_msg_header *uv1_hdr = NULL;
	struct uv2_3_bau_msg_header *uv2_3_hdr = NULL;

	if (bcp->uvhub_version == UV_BAU_V1) {
		uv1 = 1;
		uv1_throttle(hmaster, stat);
	}

	while (hmaster->uvhub_quiesce)
		cpu_relax();

	time1 = get_cycles();
	if (uv1)
		uv1_hdr = &bau_desc->header.uv1_hdr;
	else
		/* uv2 and uv3 */
		uv2_3_hdr = &bau_desc->header.uv2_3_hdr;

	do {
		if (try == 0) {
			if (uv1)
				uv1_hdr->msg_type = MSG_REGULAR;
			else
				uv2_3_hdr->msg_type = MSG_REGULAR;
			seq_number = bcp->message_number++;
		} else {
			if (uv1)
				uv1_hdr->msg_type = MSG_RETRY;
			else
				uv2_3_hdr->msg_type = MSG_RETRY;
			stat->s_retry_messages++;
		}

		if (uv1)
			uv1_hdr->sequence = seq_number;
		else
			uv2_3_hdr->sequence = seq_number;
		index = (1UL << AS_PUSH_SHIFT) | bcp->uvhub_cpu;
		bcp->send_message = get_cycles();

		write_mmr_activation(index);

		try++;
		completion_stat = ops.wait_completion(bau_desc, bcp, try);

		handle_cmplt(completion_stat, bau_desc, bcp, hmaster, stat);

		if (bcp->ipi_attempts >= bcp->ipi_reset_limit) {
			bcp->ipi_attempts = 0;
			stat->s_overipilimit++;
			completion_stat = FLUSH_GIVEUP;
			break;
		}
		cpu_relax();
	} while ((completion_stat == FLUSH_RETRY_PLUGGED) ||
		 (completion_stat == FLUSH_RETRY_TIMEOUT));

	time2 = get_cycles();

	count_max_concurr(completion_stat, bcp, hmaster);

	while (hmaster->uvhub_quiesce)
		cpu_relax();

	atomic_dec(&hmaster->active_descriptor_count);

	record_send_stats(time1, time2, bcp, stat, completion_stat, try);

	if (completion_stat == FLUSH_GIVEUP)
		/* FLUSH_GIVEUP will fall back to using IPI's for tlb flush */
		return 1;
	return 0;
}

/*
 * The BAU is disabled for this uvhub. When the disabled time period has
 * expired re-enable it.
 * Return 0 if it is re-enabled for all cpus on this uvhub.
 */
static int check_enable(struct bau_control *bcp, struct ptc_stats *stat)
{
	int tcpu;
	struct bau_control *tbcp;
	struct bau_control *hmaster;

	hmaster = bcp->uvhub_master;
	spin_lock(&hmaster->disable_lock);
	if (bcp->baudisabled && (get_cycles() >= bcp->set_bau_on_time)) {
		stat->s_bau_reenabled++;
		for_each_present_cpu(tcpu) {
			tbcp = &per_cpu(bau_control, tcpu);
			if (tbcp->uvhub_master == hmaster) {
				tbcp->baudisabled = 0;
				tbcp->period_requests = 0;
				tbcp->period_time = 0;
				tbcp->period_giveups = 0;
			}
		}
		spin_unlock(&hmaster->disable_lock);
		return 0;
	}
	spin_unlock(&hmaster->disable_lock);
	return -1;
}

static void record_send_statistics(struct ptc_stats *stat, int locals, int hubs,
				int remotes, struct bau_desc *bau_desc)
{
	stat->s_requestor++;
	stat->s_ntargcpu += remotes + locals;
	stat->s_ntargremotes += remotes;
	stat->s_ntarglocals += locals;

	/* uvhub statistics */
	hubs = bau_uvhub_weight(&bau_desc->distribution);
	if (locals) {
		stat->s_ntarglocaluvhub++;
		stat->s_ntargremoteuvhub += (hubs - 1);
	} else
		stat->s_ntargremoteuvhub += hubs;

	stat->s_ntarguvhub += hubs;

	if (hubs >= 16)
		stat->s_ntarguvhub16++;
	else if (hubs >= 8)
		stat->s_ntarguvhub8++;
	else if (hubs >= 4)
		stat->s_ntarguvhub4++;
	else if (hubs >= 2)
		stat->s_ntarguvhub2++;
	else
		stat->s_ntarguvhub1++;
}

/*
 * Translate a cpu mask to the uvhub distribution mask in the BAU
 * activation descriptor.
 */
static int set_distrib_bits(struct cpumask *flush_mask, struct bau_control *bcp,
			struct bau_desc *bau_desc, int *localsp, int *remotesp)
{
	int cpu;
	int pnode;
	int cnt = 0;
	struct hub_and_pnode *hpp;

	for_each_cpu(cpu, flush_mask) {
		/*
		 * The distribution vector is a bit map of pnodes, relative
		 * to the partition base pnode (and the partition base nasid
		 * in the header).
		 * Translate cpu to pnode and hub using a local memory array.
		 */
		hpp = &bcp->socket_master->thp[cpu];
		pnode = hpp->pnode - bcp->partition_base_pnode;
		bau_uvhub_set(pnode, &bau_desc->distribution);
		cnt++;
		if (hpp->uvhub == bcp->uvhub)
			(*localsp)++;
		else
			(*remotesp)++;
	}
	if (!cnt)
		return 1;
	return 0;
}

/*
 * globally purge translation cache of a virtual address or all TLB's
 * @cpumask: mask of all cpu's in which the address is to be removed
 * @mm: mm_struct containing virtual address range
 * @start: start virtual address to be removed from TLB
 * @end: end virtual address to be remove from TLB
 * @cpu: the current cpu
 *
 * This is the entry point for initiating any UV global TLB shootdown.
 *
 * Purges the translation caches of all specified processors of the given
 * virtual address, or purges all TLB's on specified processors.
 *
 * The caller has derived the cpumask from the mm_struct.  This function
 * is called only if there are bits set in the mask. (e.g. flush_tlb_page())
 *
 * The cpumask is converted into a uvhubmask of the uvhubs containing
 * those cpus.
 *
 * Note that this function should be called with preemption disabled.
 *
 * Returns NULL if all remote flushing was done.
 * Returns pointer to cpumask if some remote flushing remains to be
 * done.  The returned pointer is valid till preemption is re-enabled.
 */
const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
					  const struct flush_tlb_info *info)
{
	unsigned int cpu = smp_processor_id();
	int locals = 0, remotes = 0, hubs = 0;
	struct bau_desc *bau_desc;
	struct cpumask *flush_mask;
	struct ptc_stats *stat;
	struct bau_control *bcp;
	unsigned long descriptor_status, status, address;

	bcp = &per_cpu(bau_control, cpu);

	if (bcp->nobau)
		return cpumask;

	stat = bcp->statp;
	stat->s_enters++;

	if (bcp->busy) {
		descriptor_status =
			read_lmmr(UVH_LB_BAU_SB_ACTIVATION_STATUS_0);
		status = ((descriptor_status >> (bcp->uvhub_cpu *
			UV_ACT_STATUS_SIZE)) & UV_ACT_STATUS_MASK) << 1;
		if (status == UV2H_DESC_BUSY)
			return cpumask;
		bcp->busy = 0;
	}

	/* bau was disabled due to slow response */
	if (bcp->baudisabled) {
		if (check_enable(bcp, stat)) {
			stat->s_ipifordisabled++;
			return cpumask;
		}
	}

	/*
	 * Each sending cpu has a per-cpu mask which it fills from the caller's
	 * cpu mask.  All cpus are converted to uvhubs and copied to the
	 * activation descriptor.
	 */
	flush_mask = (struct cpumask *)per_cpu(uv_flush_tlb_mask, cpu);
	/* don't actually do a shootdown of the local cpu */
	cpumask_andnot(flush_mask, cpumask, cpumask_of(cpu));

	if (cpumask_test_cpu(cpu, cpumask))
		stat->s_ntargself++;

	bau_desc = bcp->descriptor_base;
	bau_desc += (ITEMS_PER_DESC * bcp->uvhub_cpu);
	bau_uvhubs_clear(&bau_desc->distribution, UV_DISTRIBUTION_SIZE);
	if (set_distrib_bits(flush_mask, bcp, bau_desc, &locals, &remotes))
		return NULL;

	record_send_statistics(stat, locals, hubs, remotes, bau_desc);

	if (!info->end || (info->end - info->start) <= PAGE_SIZE)
		address = info->start;
	else
		address = TLB_FLUSH_ALL;

	switch (bcp->uvhub_version) {
	case UV_BAU_V1:
	case UV_BAU_V2:
	case UV_BAU_V3:
		bau_desc->payload.uv1_2_3.address = address;
		bau_desc->payload.uv1_2_3.sending_cpu = cpu;
		break;
	case UV_BAU_V4:
		bau_desc->payload.uv4.address = address;
		bau_desc->payload.uv4.sending_cpu = cpu;
		bau_desc->payload.uv4.qualifier = BAU_DESC_QUALIFIER;
		break;
	}

	/*
	 * uv_flush_send_and_wait returns 0 if all cpu's were messaged,
	 * or 1 if it gave up and the original cpumask should be returned.
	 */
	if (!uv_flush_send_and_wait(flush_mask, bcp, bau_desc))
		return NULL;
	else
		return cpumask;
}

/*
 * Search the message queue for any 'other' unprocessed message with the
 * same software acknowledge resource bit vector as the 'msg' message.
 */
static struct bau_pq_entry *find_another_by_swack(struct bau_pq_entry *msg,
						  struct bau_control *bcp)
{
	struct bau_pq_entry *msg_next = msg + 1;
	unsigned char swack_vec = msg->swack_vec;

	if (msg_next > bcp->queue_last)
		msg_next = bcp->queue_first;
	while (msg_next != msg) {
		if ((msg_next->canceled == 0) && (msg_next->replied_to == 0) &&
				(msg_next->swack_vec == swack_vec))
			return msg_next;
		msg_next++;
		if (msg_next > bcp->queue_last)
			msg_next = bcp->queue_first;
	}
	return NULL;
}

/*
 * UV2 needs to work around a bug in which an arriving message has not
 * set a bit in the UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE register.
 * Such a message must be ignored.
 */
static void process_uv2_message(struct msg_desc *mdp, struct bau_control *bcp)
{
	unsigned long mmr_image;
	unsigned char swack_vec;
	struct bau_pq_entry *msg = mdp->msg;
	struct bau_pq_entry *other_msg;

	mmr_image = ops.read_l_sw_ack();
	swack_vec = msg->swack_vec;

	if ((swack_vec & mmr_image) == 0) {
		/*
		 * This message was assigned a swack resource, but no
		 * reserved acknowlegment is pending.
		 * The bug has prevented this message from setting the MMR.
		 */
		/*
		 * Some message has set the MMR 'pending' bit; it might have
		 * been another message.  Look for that message.
		 */
		other_msg = find_another_by_swack(msg, bcp);
		if (other_msg) {
			/*
			 * There is another. Process this one but do not
			 * ack it.
			 */
			bau_process_message(mdp, bcp, 0);
			/*
			 * Let the natural processing of that other message
			 * acknowledge it. Don't get the processing of sw_ack's
			 * out of order.
			 */
			return;
		}
	}

	/*
	 * Either the MMR shows this one pending a reply or there is no
	 * other message using this sw_ack, so it is safe to acknowledge it.
	 */
	bau_process_message(mdp, bcp, 1);

	return;
}

/*
 * The BAU message interrupt comes here. (registered by set_intr_gate)
 * See entry_64.S
 *
 * We received a broadcast assist message.
 *
 * Interrupts are disabled; this interrupt could represent
 * the receipt of several messages.
 *
 * All cores/threads on this hub get this interrupt.
 * The last one to see it does the software ack.
 * (the resource will not be freed until noninterruptable cpus see this
 *  interrupt; hardware may timeout the s/w ack and reply ERROR)
 */
void uv_bau_message_interrupt(struct pt_regs *regs)
{
	int count = 0;
	cycles_t time_start;
	struct bau_pq_entry *msg;
	struct bau_control *bcp;
	struct ptc_stats *stat;
	struct msg_desc msgdesc;

	ack_APIC_irq();
	kvm_set_cpu_l1tf_flush_l1d();
	time_start = get_cycles();

	bcp = &per_cpu(bau_control, smp_processor_id());
	stat = bcp->statp;

	msgdesc.queue_first = bcp->queue_first;
	msgdesc.queue_last = bcp->queue_last;

	msg = bcp->bau_msg_head;
	while (msg->swack_vec) {
		count++;

		msgdesc.msg_slot = msg - msgdesc.queue_first;
		msgdesc.msg = msg;
		if (bcp->uvhub_version == UV_BAU_V2)
			process_uv2_message(&msgdesc, bcp);
		else
			/* no error workaround for uv1 or uv3 */
			bau_process_message(&msgdesc, bcp, 1);

		msg++;
		if (msg > msgdesc.queue_last)
			msg = msgdesc.queue_first;
		bcp->bau_msg_head = msg;
	}
	stat->d_time += (get_cycles() - time_start);
	if (!count)
		stat->d_nomsg++;
	else if (count > 1)
		stat->d_multmsg++;
}

/*
 * Each target uvhub (i.e. a uvhub that has cpu's) needs to have
 * shootdown message timeouts enabled.  The timeout does not cause
 * an interrupt, but causes an error message to be returned to
 * the sender.
 */
static void __init enable_timeouts(void)
{
	int uvhub;
	int nuvhubs;
	int pnode;
	unsigned long mmr_image;

	nuvhubs = uv_num_possible_blades();

	for (uvhub = 0; uvhub < nuvhubs; uvhub++) {
		if (!uv_blade_nr_possible_cpus(uvhub))
			continue;

		pnode = uv_blade_to_pnode(uvhub);
		mmr_image = read_mmr_misc_control(pnode);
		/*
		 * Set the timeout period and then lock it in, in three
		 * steps; captures and locks in the period.
		 *
		 * To program the period, the SOFT_ACK_MODE must be off.
		 */
		mmr_image &= ~(1L << SOFTACK_MSHIFT);
		write_mmr_misc_control(pnode, mmr_image);
		/*
		 * Set the 4-bit period.
		 */
		mmr_image &= ~((unsigned long)0xf << SOFTACK_PSHIFT);
		mmr_image |= (SOFTACK_TIMEOUT_PERIOD << SOFTACK_PSHIFT);
		write_mmr_misc_control(pnode, mmr_image);
		/*
		 * UV1:
		 * Subsequent reversals of the timebase bit (3) cause an
		 * immediate timeout of one or all INTD resources as
		 * indicated in bits 2:0 (7 causes all of them to timeout).
		 */
		mmr_image |= (1L << SOFTACK_MSHIFT);
		if (is_uv2_hub()) {
			/* do not touch the legacy mode bit */
			/* hw bug workaround; do not use extended status */
			mmr_image &= ~(1L << UV2_EXT_SHFT);
		} else if (is_uv3_hub()) {
			mmr_image &= ~(1L << PREFETCH_HINT_SHFT);
			mmr_image |= (1L << SB_STATUS_SHFT);
		}
		write_mmr_misc_control(pnode, mmr_image);
	}
}

static void *ptc_seq_start(struct seq_file *file, loff_t *offset)
{
	if (*offset < num_possible_cpus())
		return offset;
	return NULL;
}

static void *ptc_seq_next(struct seq_file *file, void *data, loff_t *offset)
{
	(*offset)++;
	if (*offset < num_possible_cpus())
		return offset;
	return NULL;
}

static void ptc_seq_stop(struct seq_file *file, void *data)
{
}

/*
 * Display the statistics thru /proc/sgi_uv/ptc_statistics
 * 'data' points to the cpu number
 * Note: see the descriptions in stat_description[].
 */
static int ptc_seq_show(struct seq_file *file, void *data)
{
	struct ptc_stats *stat;
	struct bau_control *bcp;
	int cpu;

	cpu = *(loff_t *)data;
	if (!cpu) {
		seq_puts(file,
			 "# cpu bauoff sent stime self locals remotes ncpus localhub ");
		seq_puts(file, "remotehub numuvhubs numuvhubs16 numuvhubs8 ");
		seq_puts(file,
			 "numuvhubs4 numuvhubs2 numuvhubs1 dto snacks retries ");
		seq_puts(file,
			 "rok resetp resett giveup sto bz throt disable ");
		seq_puts(file,
			 "enable wars warshw warwaits enters ipidis plugged ");
		seq_puts(file,
			 "ipiover glim cong swack recv rtime all one mult ");
		seq_puts(file, "none retry canc nocan reset rcan\n");
	}
	if (cpu < num_possible_cpus() && cpu_online(cpu)) {
		bcp = &per_cpu(bau_control, cpu);
		if (bcp->nobau) {
			seq_printf(file, "cpu %d bau disabled\n", cpu);
			return 0;
		}
		stat = bcp->statp;
		/* source side statistics */
		seq_printf(file,
			"cpu %d %d %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld ",
			   cpu, bcp->nobau, stat->s_requestor,
			   cycles_2_us(stat->s_time),
			   stat->s_ntargself, stat->s_ntarglocals,
			   stat->s_ntargremotes, stat->s_ntargcpu,
			   stat->s_ntarglocaluvhub, stat->s_ntargremoteuvhub,
			   stat->s_ntarguvhub, stat->s_ntarguvhub16);
		seq_printf(file, "%ld %ld %ld %ld %ld %ld ",
			   stat->s_ntarguvhub8, stat->s_ntarguvhub4,
			   stat->s_ntarguvhub2, stat->s_ntarguvhub1,
			   stat->s_dtimeout, stat->s_strongnacks);
		seq_printf(file, "%ld %ld %ld %ld %ld %ld %ld %ld ",
			   stat->s_retry_messages, stat->s_retriesok,
			   stat->s_resets_plug, stat->s_resets_timeout,
			   stat->s_giveup, stat->s_stimeout,
			   stat->s_busy, stat->s_throttles);
		seq_printf(file, "%ld %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld ",
			   stat->s_bau_disabled, stat->s_bau_reenabled,
			   stat->s_uv2_wars, stat->s_uv2_wars_hw,
			   stat->s_uv2_war_waits, stat->s_enters,
			   stat->s_ipifordisabled, stat->s_plugged,
			   stat->s_overipilimit, stat->s_giveuplimit,
			   stat->s_congested);

		/* destination side statistics */
		seq_printf(file,
			"%lx %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld\n",
			   ops.read_g_sw_ack(uv_cpu_to_pnode(cpu)),
			   stat->d_requestee, cycles_2_us(stat->d_time),
			   stat->d_alltlb, stat->d_onetlb, stat->d_multmsg,
			   stat->d_nomsg, stat->d_retries, stat->d_canceled,
			   stat->d_nocanceled, stat->d_resets,
			   stat->d_rcanceled);
	}
	return 0;
}

/*
 * Display the tunables thru debugfs
 */
static ssize_t tunables_read(struct file *file, char __user *userbuf,
				size_t count, loff_t *ppos)
{
	char *buf;
	int ret;

	buf = kasprintf(GFP_KERNEL, "%s %s %s\n%d %d %d %d %d %d %d %d %d %d\n",
		"max_concur plugged_delay plugsb4reset timeoutsb4reset",
		"ipi_reset_limit complete_threshold congested_response_us",
		"congested_reps disabled_period giveup_limit",
		max_concurr, plugged_delay, plugsb4reset,
		timeoutsb4reset, ipi_reset_limit, complete_threshold,
		congested_respns_us, congested_reps, disabled_period,
		giveup_limit);

	if (!buf)
		return -ENOMEM;

	ret = simple_read_from_buffer(userbuf, count, ppos, buf, strlen(buf));
	kfree(buf);
	return ret;
}

/*
 * handle a write to /proc/sgi_uv/ptc_statistics
 * -1: reset the statistics
 *  0: display meaning of the statistics
 */
static ssize_t ptc_proc_write(struct file *file, const char __user *user,
				size_t count, loff_t *data)
{
	int cpu;
	int i;
	int elements;
	long input_arg;
	char optstr[64];
	struct ptc_stats *stat;

	if (count == 0 || count > sizeof(optstr))
		return -EINVAL;
	if (copy_from_user(optstr, user, count))
		return -EFAULT;
	optstr[count - 1] = '\0';

	if (!strcmp(optstr, "on")) {
		set_bau_on();
		return count;
	} else if (!strcmp(optstr, "off")) {
		set_bau_off();
		return count;
	}

	if (kstrtol(optstr, 10, &input_arg) < 0) {
		pr_debug("%s is invalid\n", optstr);
		return -EINVAL;
	}

	if (input_arg == 0) {
		elements = ARRAY_SIZE(stat_description);
		pr_debug("# cpu:      cpu number\n");
		pr_debug("Sender statistics:\n");
		for (i = 0; i < elements; i++)
			pr_debug("%s\n", stat_description[i]);
	} else if (input_arg == -1) {
		for_each_present_cpu(cpu) {
			stat = &per_cpu(ptcstats, cpu);
			memset(stat, 0, sizeof(struct ptc_stats));
		}
	}

	return count;
}

static int local_atoi(const char *name)
{
	int val = 0;

	for (;; name++) {
		switch (*name) {
		case '0' ... '9':
			val = 10*val+(*name-'0');
			break;
		default:
			return val;
		}
	}
}

/*
 * Parse the values written to /sys/kernel/debug/sgi_uv/bau_tunables.
 * Zero values reset them to defaults.
 */
static int parse_tunables_write(struct bau_control *bcp, char *instr,
				int count)
{
	char *p;
	char *q;
	int cnt = 0;
	int val;
	int e = ARRAY_SIZE(tunables);

	p = instr + strspn(instr, WHITESPACE);
	q = p;
	for (; *p; p = q + strspn(q, WHITESPACE)) {
		q = p + strcspn(p, WHITESPACE);
		cnt++;
		if (q == p)
			break;
	}
	if (cnt != e) {
		pr_info("bau tunable error: should be %d values\n", e);
		return -EINVAL;
	}

	p = instr + strspn(instr, WHITESPACE);
	q = p;
	for (cnt = 0; *p; p = q + strspn(q, WHITESPACE), cnt++) {
		q = p + strcspn(p, WHITESPACE);
		val = local_atoi(p);
		switch (cnt) {
		case 0:
			if (val == 0) {
				max_concurr = MAX_BAU_CONCURRENT;
				max_concurr_const = MAX_BAU_CONCURRENT;
				continue;
			}
			if (val < 1 || val > bcp->cpus_in_uvhub) {
				pr_debug(
				"Error: BAU max concurrent %d is invalid\n",
				val);
				return -EINVAL;
			}
			max_concurr = val;
			max_concurr_const = val;
			continue;
		default:
			if (val == 0)
				*tunables[cnt].tunp = tunables[cnt].deflt;
			else
				*tunables[cnt].tunp = val;
			continue;
		}
	}
	return 0;
}

/*
 * Handle a write to debugfs. (/sys/kernel/debug/sgi_uv/bau_tunables)
 */
static ssize_t tunables_write(struct file *file, const char __user *user,
				size_t count, loff_t *data)
{
	int cpu;
	int ret;
	char instr[100];
	struct bau_control *bcp;

	if (count == 0 || count > sizeof(instr)-1)
		return -EINVAL;
	if (copy_from_user(instr, user, count))
		return -EFAULT;

	instr[count] = '\0';

	cpu = get_cpu();
	bcp = &per_cpu(bau_control, cpu);
	ret = parse_tunables_write(bcp, instr, count);
	put_cpu();
	if (ret)
		return ret;

	for_each_present_cpu(cpu) {
		bcp = &per_cpu(bau_control, cpu);
		bcp->max_concurr         = max_concurr;
		bcp->max_concurr_const   = max_concurr;
		bcp->plugged_delay       = plugged_delay;
		bcp->plugsb4reset        = plugsb4reset;
		bcp->timeoutsb4reset     = timeoutsb4reset;
		bcp->ipi_reset_limit     = ipi_reset_limit;
		bcp->complete_threshold  = complete_threshold;
		bcp->cong_response_us    = congested_respns_us;
		bcp->cong_reps           = congested_reps;
		bcp->disabled_period     = sec_2_cycles(disabled_period);
		bcp->giveup_limit        = giveup_limit;
	}
	return count;
}

static const struct seq_operations uv_ptc_seq_ops = {
	.start		= ptc_seq_start,
	.next		= ptc_seq_next,
	.stop		= ptc_seq_stop,
	.show		= ptc_seq_show
};

static int ptc_proc_open(struct inode *inode, struct file *file)
{
	return seq_open(file, &uv_ptc_seq_ops);
}

static int tunables_open(struct inode *inode, struct file *file)
{
	return 0;
}

static const struct proc_ops uv_ptc_proc_ops = {
	.proc_open	= ptc_proc_open,
	.proc_read	= seq_read,
	.proc_write	= ptc_proc_write,
	.proc_lseek	= seq_lseek,
	.proc_release	= seq_release,
};

static const struct file_operations tunables_fops = {
	.open		= tunables_open,
	.read		= tunables_read,
	.write		= tunables_write,
	.llseek		= default_llseek,
};

static int __init uv_ptc_init(void)
{
	struct proc_dir_entry *proc_uv_ptc;

	if (!is_uv_system())
		return 0;

	proc_uv_ptc = proc_create(UV_PTC_BASENAME, 0444, NULL,
				  &uv_ptc_proc_ops);
	if (!proc_uv_ptc) {
		pr_err("unable to create %s proc entry\n",
		       UV_PTC_BASENAME);
		return -EINVAL;
	}

	tunables_dir = debugfs_create_dir(UV_BAU_TUNABLES_DIR, NULL);
	debugfs_create_file(UV_BAU_TUNABLES_FILE, 0600, tunables_dir, NULL,
			    &tunables_fops);
	return 0;
}

/*
 * Initialize the sending side's sending buffers.
 */
static void activation_descriptor_init(int node, int pnode, int base_pnode)
{
	int i;
	int cpu;
	int uv1 = 0;
	unsigned long gpa;
	unsigned long m;
	unsigned long n;
	size_t dsize;
	struct bau_desc *bau_desc;
	struct bau_desc *bd2;
	struct uv1_bau_msg_header *uv1_hdr;
	struct uv2_3_bau_msg_header *uv2_3_hdr;
	struct bau_control *bcp;

	/*
	 * each bau_desc is 64 bytes; there are 8 (ITEMS_PER_DESC)
	 * per cpu; and one per cpu on the uvhub (ADP_SZ)
	 */
	dsize = sizeof(struct bau_desc) * ADP_SZ * ITEMS_PER_DESC;
	bau_desc = kmalloc_node(dsize, GFP_KERNEL, node);
	BUG_ON(!bau_desc);

	gpa = uv_gpa(bau_desc);
	n = uv_gpa_to_gnode(gpa);
	m = ops.bau_gpa_to_offset(gpa);
	if (is_uv1_hub())
		uv1 = 1;

	/* the 14-bit pnode */
	write_mmr_descriptor_base(pnode,
		(n << UVH_LB_BAU_SB_DESCRIPTOR_BASE_NODE_ID_SHFT | m));
	/*
	 * Initializing all 8 (ITEMS_PER_DESC) descriptors for each
	 * cpu even though we only use the first one; one descriptor can
	 * describe a broadcast to 256 uv hubs.
	 */
	for (i = 0, bd2 = bau_desc; i < (ADP_SZ * ITEMS_PER_DESC); i++, bd2++) {
		memset(bd2, 0, sizeof(struct bau_desc));
		if (uv1) {
			uv1_hdr = &bd2->header.uv1_hdr;
			uv1_hdr->swack_flag = 1;
			/*
			 * The base_dest_nasid set in the message header
			 * is the nasid of the first uvhub in the partition.
			 * The bit map will indicate destination pnode numbers
			 * relative to that base. They may not be consecutive
			 * if nasid striding is being used.
			 */
			uv1_hdr->base_dest_nasid =
			                          UV_PNODE_TO_NASID(base_pnode);
			uv1_hdr->dest_subnodeid  = UV_LB_SUBNODEID;
			uv1_hdr->command         = UV_NET_ENDPOINT_INTD;
			uv1_hdr->int_both        = 1;
			/*
			 * all others need to be set to zero:
			 *   fairness chaining multilevel count replied_to
			 */
		} else {
			/*
			 * BIOS uses legacy mode, but uv2 and uv3 hardware always
			 * uses native mode for selective broadcasts.
			 */
			uv2_3_hdr = &bd2->header.uv2_3_hdr;
			uv2_3_hdr->swack_flag      = 1;
			uv2_3_hdr->base_dest_nasid =
			                          UV_PNODE_TO_NASID(base_pnode);
			uv2_3_hdr->dest_subnodeid  = UV_LB_SUBNODEID;
			uv2_3_hdr->command         = UV_NET_ENDPOINT_INTD;
		}
	}
	for_each_present_cpu(cpu) {
		if (pnode != uv_blade_to_pnode(uv_cpu_to_blade_id(cpu)))
			continue;
		bcp = &per_cpu(bau_control, cpu);
		bcp->descriptor_base = bau_desc;
	}
}

/*
 * initialize the destination side's receiving buffers
 * entered for each uvhub in the partition
 * - node is first node (kernel memory notion) on the uvhub
 * - pnode is the uvhub's physical identifier
 */
static void pq_init(int node, int pnode)
{
	int cpu;
	size_t plsize;
	char *cp;
	void *vp;
	unsigned long gnode, first, last, tail;
	struct bau_pq_entry *pqp;
	struct bau_control *bcp;

	plsize = (DEST_Q_SIZE + 1) * sizeof(struct bau_pq_entry);
	vp = kmalloc_node(plsize, GFP_KERNEL, node);
	BUG_ON(!vp);

	pqp = (struct bau_pq_entry *)vp;
	cp = (char *)pqp + 31;
	pqp = (struct bau_pq_entry *)(((unsigned long)cp >> 5) << 5);

	for_each_present_cpu(cpu) {
		if (pnode != uv_cpu_to_pnode(cpu))
			continue;
		/* for every cpu on this pnode: */
		bcp = &per_cpu(bau_control, cpu);
		bcp->queue_first	= pqp;
		bcp->bau_msg_head	= pqp;
		bcp->queue_last		= pqp + (DEST_Q_SIZE - 1);
	}

	first = ops.bau_gpa_to_offset(uv_gpa(pqp));
	last = ops.bau_gpa_to_offset(uv_gpa(pqp + (DEST_Q_SIZE - 1)));

	/*
	 * Pre UV4, the gnode is required to locate the payload queue
	 * and the payload queue tail must be maintained by the kernel.
	 */
	bcp = &per_cpu(bau_control, smp_processor_id());
	if (bcp->uvhub_version <= UV_BAU_V3) {
		tail = first;
		gnode = uv_gpa_to_gnode(uv_gpa(pqp));
		first = (gnode << UV_PAYLOADQ_GNODE_SHIFT) | tail;
		write_mmr_payload_tail(pnode, tail);
	}

	ops.write_payload_first(pnode, first);
	ops.write_payload_last(pnode, last);

	/* in effect, all msg_type's are set to MSG_NOOP */
	memset(pqp, 0, sizeof(struct bau_pq_entry) * DEST_Q_SIZE);
}

/*
 * Initialization of each UV hub's structures
 */
static void __init init_uvhub(int uvhub, int vector, int base_pnode)
{
	int node;
	int pnode;
	unsigned long apicid;

	node = uvhub_to_first_node(uvhub);
	pnode = uv_blade_to_pnode(uvhub);

	activation_descriptor_init(node, pnode, base_pnode);

	pq_init(node, pnode);
	/*
	 * The below initialization can't be in firmware because the
	 * messaging IRQ will be determined by the OS.
	 */
	apicid = uvhub_to_first_apicid(uvhub) | uv_apicid_hibits;
	write_mmr_data_config(pnode, ((apicid << 32) | vector));
}

/*
 * We will set BAU_MISC_CONTROL with a timeout period.
 * But the BIOS has set UVH_AGING_PRESCALE_SEL and UVH_TRANSACTION_TIMEOUT.
 * So the destination timeout period has to be calculated from them.
 */
static int calculate_destination_timeout(void)
{
	unsigned long mmr_image;
	int mult1;
	int mult2;
	int index;
	int base;
	int ret;
	unsigned long ts_ns;

	if (is_uv1_hub()) {
		mult1 = SOFTACK_TIMEOUT_PERIOD & BAU_MISC_CONTROL_MULT_MASK;
		mmr_image = uv_read_local_mmr(UVH_AGING_PRESCALE_SEL);
		index = (mmr_image >> BAU_URGENCY_7_SHIFT) & BAU_URGENCY_7_MASK;
		mmr_image = uv_read_local_mmr(UVH_TRANSACTION_TIMEOUT);
		mult2 = (mmr_image >> BAU_TRANS_SHIFT) & BAU_TRANS_MASK;
		ts_ns = timeout_base_ns[index];
		ts_ns *= (mult1 * mult2);
		ret = ts_ns / 1000;
	} else {
		/* same destination timeout for uv2 and uv3 */
		/* 4 bits  0/1 for 10/80us base, 3 bits of multiplier */
		mmr_image = uv_read_local_mmr(UVH_LB_BAU_MISC_CONTROL);
		mmr_image = (mmr_image & UV_SA_MASK) >> UV_SA_SHFT;
		if (mmr_image & (1L << UV2_ACK_UNITS_SHFT))
			base = 80;
		else
			base = 10;
		mult1 = mmr_image & UV2_ACK_MASK;
		ret = mult1 * base;
	}
	return ret;
}

static void __init init_per_cpu_tunables(void)
{
	int cpu;
	struct bau_control *bcp;

	for_each_present_cpu(cpu) {
		bcp = &per_cpu(bau_control, cpu);
		bcp->baudisabled		= 0;
		if (nobau)
			bcp->nobau		= true;
		bcp->statp			= &per_cpu(ptcstats, cpu);
		/* time interval to catch a hardware stay-busy bug */
		bcp->timeout_interval		= usec_2_cycles(2*timeout_us);
		bcp->max_concurr		= max_concurr;
		bcp->max_concurr_const		= max_concurr;
		bcp->plugged_delay		= plugged_delay;
		bcp->plugsb4reset		= plugsb4reset;
		bcp->timeoutsb4reset		= timeoutsb4reset;
		bcp->ipi_reset_limit		= ipi_reset_limit;
		bcp->complete_threshold		= complete_threshold;
		bcp->cong_response_us		= congested_respns_us;
		bcp->cong_reps			= congested_reps;
		bcp->disabled_period		= sec_2_cycles(disabled_period);
		bcp->giveup_limit		= giveup_limit;
		spin_lock_init(&bcp->queue_lock);
		spin_lock_init(&bcp->uvhub_lock);
		spin_lock_init(&bcp->disable_lock);
	}
}

/*
 * Scan all cpus to collect blade and socket summaries.
 */
static int __init get_cpu_topology(int base_pnode,
					struct uvhub_desc *uvhub_descs,
					unsigned char *uvhub_mask)
{
	int cpu;
	int pnode;
	int uvhub;
	int socket;
	struct bau_control *bcp;
	struct uvhub_desc *bdp;
	struct socket_desc *sdp;

	for_each_present_cpu(cpu) {
		bcp = &per_cpu(bau_control, cpu);

		memset(bcp, 0, sizeof(struct bau_control));

		pnode = uv_cpu_hub_info(cpu)->pnode;
		if ((pnode - base_pnode) >= UV_DISTRIBUTION_SIZE) {
			pr_emerg(
				"cpu %d pnode %d-%d beyond %d; BAU disabled\n",
				cpu, pnode, base_pnode, UV_DISTRIBUTION_SIZE);
			return 1;
		}

		bcp->osnode = cpu_to_node(cpu);
		bcp->partition_base_pnode = base_pnode;

		uvhub = uv_cpu_hub_info(cpu)->numa_blade_id;
		*(uvhub_mask + (uvhub/8)) |= (1 << (uvhub%8));
		bdp = &uvhub_descs[uvhub];

		bdp->num_cpus++;
		bdp->uvhub = uvhub;
		bdp->pnode = pnode;

		/* kludge: 'assuming' one node per socket, and assuming that
		   disabling a socket just leaves a gap in node numbers */
		socket = bcp->osnode & 1;
		bdp->socket_mask |= (1 << socket);
		sdp = &bdp->socket[socket];
		sdp->cpu_number[sdp->num_cpus] = cpu;
		sdp->num_cpus++;
		if (sdp->num_cpus > MAX_CPUS_PER_SOCKET) {
			pr_emerg("%d cpus per socket invalid\n",
				sdp->num_cpus);
			return 1;
		}
	}
	return 0;
}

/*
 * Each socket is to get a local array of pnodes/hubs.
 */
static void make_per_cpu_thp(struct bau_control *smaster)
{
	int cpu;
	size_t hpsz = sizeof(struct hub_and_pnode) * num_possible_cpus();

	smaster->thp = kzalloc_node(hpsz, GFP_KERNEL, smaster->osnode);
	for_each_present_cpu(cpu) {
		smaster->thp[cpu].pnode = uv_cpu_hub_info(cpu)->pnode;
		smaster->thp[cpu].uvhub = uv_cpu_hub_info(cpu)->numa_blade_id;
	}
}

/*
 * Each uvhub is to get a local cpumask.
 */
static void make_per_hub_cpumask(struct bau_control *hmaster)
{
	int sz = sizeof(cpumask_t);

	hmaster->cpumask = kzalloc_node(sz, GFP_KERNEL, hmaster->osnode);
}

/*
 * Initialize all the per_cpu information for the cpu's on a given socket,
 * given what has been gathered into the socket_desc struct.
 * And reports the chosen hub and socket masters back to the caller.
 */
static int scan_sock(struct socket_desc *sdp, struct uvhub_desc *bdp,
			struct bau_control **smasterp,
			struct bau_control **hmasterp)
{
	int i, cpu, uvhub_cpu;
	struct bau_control *bcp;

	for (i = 0; i < sdp->num_cpus; i++) {
		cpu = sdp->cpu_number[i];
		bcp = &per_cpu(bau_control, cpu);
		bcp->cpu = cpu;
		if (i == 0) {
			*smasterp = bcp;
			if (!(*hmasterp))
				*hmasterp = bcp;
		}
		bcp->cpus_in_uvhub = bdp->num_cpus;
		bcp->cpus_in_socket = sdp->num_cpus;
		bcp->socket_master = *smasterp;
		bcp->uvhub = bdp->uvhub;
		if (is_uv1_hub())
			bcp->uvhub_version = UV_BAU_V1;
		else if (is_uv2_hub())
			bcp->uvhub_version = UV_BAU_V2;
		else if (is_uv3_hub())
			bcp->uvhub_version = UV_BAU_V3;
		else if (is_uv4_hub())
			bcp->uvhub_version = UV_BAU_V4;
		else {
			pr_emerg("uvhub version not 1, 2, 3, or 4\n");
			return 1;
		}
		bcp->uvhub_master = *hmasterp;
		uvhub_cpu = uv_cpu_blade_processor_id(cpu);
		bcp->uvhub_cpu = uvhub_cpu;

		/*
		 * The ERROR and BUSY status registers are located pairwise over
		 * the STATUS_0 and STATUS_1 mmrs; each an array[32] of 2 bits.
		 */
		if (uvhub_cpu < UV_CPUS_PER_AS) {
			bcp->status_mmr = UVH_LB_BAU_SB_ACTIVATION_STATUS_0;
			bcp->status_index = uvhub_cpu * UV_ACT_STATUS_SIZE;
		} else {
			bcp->status_mmr = UVH_LB_BAU_SB_ACTIVATION_STATUS_1;
			bcp->status_index = (uvhub_cpu - UV_CPUS_PER_AS)
						* UV_ACT_STATUS_SIZE;
		}

		if (bcp->uvhub_cpu >= MAX_CPUS_PER_UVHUB) {
			pr_emerg("%d cpus per uvhub invalid\n",
				bcp->uvhub_cpu);
			return 1;
		}
	}
	return 0;
}

/*
 * Summarize the blade and socket topology into the per_cpu structures.
 */
static int __init summarize_uvhub_sockets(int nuvhubs,
			struct uvhub_desc *uvhub_descs,
			unsigned char *uvhub_mask)
{
	int socket;
	int uvhub;
	unsigned short socket_mask;

	for (uvhub = 0; uvhub < nuvhubs; uvhub++) {
		struct uvhub_desc *bdp;
		struct bau_control *smaster = NULL;
		struct bau_control *hmaster = NULL;

		if (!(*(uvhub_mask + (uvhub/8)) & (1 << (uvhub%8))))
			continue;

		bdp = &uvhub_descs[uvhub];
		socket_mask = bdp->socket_mask;
		socket = 0;
		while (socket_mask) {
			struct socket_desc *sdp;
			if ((socket_mask & 1)) {
				sdp = &bdp->socket[socket];
				if (scan_sock(sdp, bdp, &smaster, &hmaster))
					return 1;
				make_per_cpu_thp(smaster);
			}
			socket++;
			socket_mask = (socket_mask >> 1);
		}
		make_per_hub_cpumask(hmaster);
	}
	return 0;
}

/*
 * initialize the bau_control structure for each cpu
 */
static int __init init_per_cpu(int nuvhubs, int base_part_pnode)
{
	struct uvhub_desc *uvhub_descs;
	unsigned char *uvhub_mask = NULL;

	if (is_uv3_hub() || is_uv2_hub() || is_uv1_hub())
		timeout_us = calculate_destination_timeout();

	uvhub_descs = kcalloc(nuvhubs, sizeof(struct uvhub_desc), GFP_KERNEL);
	if (!uvhub_descs)
		goto fail;

	uvhub_mask = kzalloc((nuvhubs+7)/8, GFP_KERNEL);
	if (!uvhub_mask)
		goto fail;

	if (get_cpu_topology(base_part_pnode, uvhub_descs, uvhub_mask))
		goto fail;

	if (summarize_uvhub_sockets(nuvhubs, uvhub_descs, uvhub_mask))
		goto fail;

	kfree(uvhub_descs);
	kfree(uvhub_mask);
	init_per_cpu_tunables();
	return 0;

fail:
	kfree(uvhub_descs);
	kfree(uvhub_mask);
	return 1;
}

static const struct bau_operations uv1_bau_ops __initconst = {
	.bau_gpa_to_offset       = uv_gpa_to_offset,
	.read_l_sw_ack           = read_mmr_sw_ack,
	.read_g_sw_ack           = read_gmmr_sw_ack,
	.write_l_sw_ack          = write_mmr_sw_ack,
	.write_g_sw_ack          = write_gmmr_sw_ack,
	.write_payload_first     = write_mmr_payload_first,
	.write_payload_last      = write_mmr_payload_last,
	.wait_completion	 = uv1_wait_completion,
};

static const struct bau_operations uv2_3_bau_ops __initconst = {
	.bau_gpa_to_offset       = uv_gpa_to_offset,
	.read_l_sw_ack           = read_mmr_sw_ack,
	.read_g_sw_ack           = read_gmmr_sw_ack,
	.write_l_sw_ack          = write_mmr_sw_ack,
	.write_g_sw_ack          = write_gmmr_sw_ack,
	.write_payload_first     = write_mmr_payload_first,
	.write_payload_last      = write_mmr_payload_last,
	.wait_completion	 = uv2_3_wait_completion,
};

static const struct bau_operations uv4_bau_ops __initconst = {
	.bau_gpa_to_offset       = uv_gpa_to_soc_phys_ram,
	.read_l_sw_ack           = read_mmr_proc_sw_ack,
	.read_g_sw_ack           = read_gmmr_proc_sw_ack,
	.write_l_sw_ack          = write_mmr_proc_sw_ack,
	.write_g_sw_ack          = write_gmmr_proc_sw_ack,
	.write_payload_first     = write_mmr_proc_payload_first,
	.write_payload_last      = write_mmr_proc_payload_last,
	.wait_completion         = uv4_wait_completion,
};

/*
 * Initialization of BAU-related structures
 */
static int __init uv_bau_init(void)
{
	int uvhub;
	int pnode;
	int nuvhubs;
	int cur_cpu;
	int cpus;
	int vector;
	cpumask_var_t *mask;

	if (!is_uv_system())
		return 0;

	if (is_uv4_hub())
		ops = uv4_bau_ops;
	else if (is_uv3_hub())
		ops = uv2_3_bau_ops;
	else if (is_uv2_hub())
		ops = uv2_3_bau_ops;
	else if (is_uv1_hub())
		ops = uv1_bau_ops;

	nuvhubs = uv_num_possible_blades();
	if (nuvhubs < 2) {
		pr_crit("UV: BAU disabled - insufficient hub count\n");
		goto err_bau_disable;
	}

	for_each_possible_cpu(cur_cpu) {
		mask = &per_cpu(uv_flush_tlb_mask, cur_cpu);
		zalloc_cpumask_var_node(mask, GFP_KERNEL, cpu_to_node(cur_cpu));
	}

	uv_base_pnode = 0x7fffffff;
	for (uvhub = 0; uvhub < nuvhubs; uvhub++) {
		cpus = uv_blade_nr_possible_cpus(uvhub);
		if (cpus && (uv_blade_to_pnode(uvhub) < uv_base_pnode))
			uv_base_pnode = uv_blade_to_pnode(uvhub);
	}

	/* software timeouts are not supported on UV4 */
	if (is_uv3_hub() || is_uv2_hub() || is_uv1_hub())
		enable_timeouts();

	if (init_per_cpu(nuvhubs, uv_base_pnode)) {
		pr_crit("UV: BAU disabled - per CPU init failed\n");
		goto err_bau_disable;
	}

	vector = UV_BAU_MESSAGE;
	for_each_possible_blade(uvhub) {
		if (uv_blade_nr_possible_cpus(uvhub))
			init_uvhub(uvhub, vector, uv_base_pnode);
	}

	for_each_possible_blade(uvhub) {
		if (uv_blade_nr_possible_cpus(uvhub)) {
			unsigned long val;
			unsigned long mmr;
			pnode = uv_blade_to_pnode(uvhub);
			/* INIT the bau */
			val = 1L << 63;
			write_gmmr_activation(pnode, val);
			mmr = 1; /* should be 1 to broadcast to both sockets */
			if (!is_uv1_hub())
				write_mmr_data_broadcast(pnode, mmr);
		}
	}

	return 0;

err_bau_disable:

	for_each_possible_cpu(cur_cpu)
		free_cpumask_var(per_cpu(uv_flush_tlb_mask, cur_cpu));

	set_bau_off();
	nobau_perm = 1;

	return -EINVAL;
}
core_initcall(uv_bau_init);
fs_initcall(uv_ptc_init);