aboutsummaryrefslogtreecommitdiffstats
path: root/arch/m32r/mm/init.c
blob: 73e2205ebf5afba2c17ca848701c0ae026f52fab (plain) (blame)
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
/*
 *  linux/arch/m32r/mm/init.c
 *
 *  Copyright (c) 2001, 2002  Hitoshi Yamamoto
 *
 *  Some code taken from sh version.
 *    Copyright (C) 1999  Niibe Yutaka
 *    Based on linux/arch/i386/mm/init.c:
 *      Copyright (C) 1995  Linus Torvalds
 */

#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/pagemap.h>
#include <linux/bootmem.h>
#include <linux/swap.h>
#include <linux/highmem.h>
#include <linux/bitops.h>
#include <linux/nodemask.h>
#include <linux/pfn.h>
#include <linux/gfp.h>
#include <asm/types.h>
#include <asm/processor.h>
#include <asm/page.h>
#include <asm/pgtable.h>
#include <asm/pgalloc.h>
#include <asm/mmu_context.h>
#include <asm/setup.h>
#include <asm/tlb.h>

/* References to section boundaries */
extern char _text, _etext, _edata;
extern char __init_begin, __init_end;

pgd_t swapper_pg_dir[1024];

DEFINE_PER_CPU(struct mmu_gather, mmu_gathers);

/*
 * Cache of MMU context last used.
 */
#ifndef CONFIG_SMP
unsigned long mmu_context_cache_dat;
#else
unsigned long mmu_context_cache_dat[NR_CPUS];
#endif
static unsigned long hole_pages;

/*
 * function prototype
 */
void __init paging_init(void);
void __init mem_init(void);
void free_initmem(void);
#ifdef CONFIG_BLK_DEV_INITRD
void free_initrd_mem(unsigned long, unsigned long);
#endif

/* It'd be good if these lines were in the standard header file. */
#define START_PFN(nid)		(NODE_DATA(nid)->bdata->node_min_pfn)
#define MAX_LOW_PFN(nid)	(NODE_DATA(nid)->bdata->node_low_pfn)

#ifndef CONFIG_DISCONTIGMEM
unsigned long __init zone_sizes_init(void)
{
	unsigned long  zones_size[MAX_NR_ZONES] = {0, };
	unsigned long  max_dma;
	unsigned long  low;
	unsigned long  start_pfn;

#ifdef CONFIG_MMU
	start_pfn = START_PFN(0);
	max_dma = virt_to_phys((char *)MAX_DMA_ADDRESS) >> PAGE_SHIFT;
	low = MAX_LOW_PFN(0);

	if (low < max_dma){
		zones_size[ZONE_DMA] = low - start_pfn;
		zones_size[ZONE_NORMAL] = 0;
	} else {
		zones_size[ZONE_DMA] = low - start_pfn;
		zones_size[ZONE_NORMAL] = low - max_dma;
	}
#else
	zones_size[ZONE_DMA] = 0 >> PAGE_SHIFT;
	zones_size[ZONE_NORMAL] = __MEMORY_SIZE >> PAGE_SHIFT;
	start_pfn = __MEMORY_START >> PAGE_SHIFT;
#endif /* CONFIG_MMU */

	free_area_init_node(0, zones_size, start_pfn, 0);

	return 0;
}
#else	/* CONFIG_DISCONTIGMEM */
extern unsigned long zone_sizes_init(void);
#endif	/* CONFIG_DISCONTIGMEM */

/*======================================================================*
 * paging_init() : sets up the page tables
 *======================================================================*/
void __init paging_init(void)
{
#ifdef CONFIG_MMU
	int  i;
	pgd_t *pg_dir;

	/* We don't need kernel mapping as hardware support that. */
	pg_dir = swapper_pg_dir;

	for (i = 0 ; i < USER_PTRS_PER_PGD * 2 ; i++)
		pgd_val(pg_dir[i]) = 0;
#endif /* CONFIG_MMU */
	hole_pages = zone_sizes_init();
}

int __init reservedpages_count(void)
{
	int reservedpages, nid, i;

	reservedpages = 0;
	for_each_online_node(nid) {
		unsigned long flags;
		pgdat_resize_lock(NODE_DATA(nid), &flags);
		for (i = 0 ; i < MAX_LOW_PFN(nid) - START_PFN(nid) ; i++)
			if (PageReserved(nid_page_nr(nid, i)))
				reservedpages++;
		pgdat_resize_unlock(NODE_DATA(nid), &flags);
	}

	return reservedpages;
}

/*======================================================================*
 * mem_init() :
 * orig : arch/sh/mm/init.c
 *======================================================================*/
void __init mem_init(void)
{
	int codesize, reservedpages, datasize, initsize;
	int nid;
#ifndef CONFIG_MMU
	extern unsigned long memory_end;
#endif

	num_physpages = 0;
	for_each_online_node(nid)
		num_physpages += MAX_LOW_PFN(nid) - START_PFN(nid) + 1;

	num_physpages -= hole_pages;

#ifndef CONFIG_DISCONTIGMEM
	max_mapnr = num_physpages;
#endif	/* CONFIG_DISCONTIGMEM */

#ifdef CONFIG_MMU
	high_memory = (void *)__va(PFN_PHYS(MAX_LOW_PFN(0)));
#else
	high_memory = (void *)(memory_end & PAGE_MASK);
#endif /* CONFIG_MMU */

	/* clear the zero-page */
	memset(empty_zero_page, 0, PAGE_SIZE);

	/* this will put all low memory onto the freelists */
	for_each_online_node(nid)
		totalram_pages += free_all_bootmem_node(NODE_DATA(nid));

	reservedpages = reservedpages_count() - hole_pages;
	codesize = (unsigned long) &_etext - (unsigned long)&_text;
	datasize = (unsigned long) &_edata - (unsigned long)&_etext;
	initsize = (unsigned long) &__init_end - (unsigned long)&__init_begin;

	printk(KERN_INFO "Memory: %luk/%luk available (%dk kernel code, "
		"%dk reserved, %dk data, %dk init)\n",
		nr_free_pages() << (PAGE_SHIFT-10),
		num_physpages << (PAGE_SHIFT-10),
		codesize >> 10,
		reservedpages << (PAGE_SHIFT-10),
		datasize >> 10,
		initsize >> 10);
}

/*======================================================================*
 * free_initmem() :
 * orig : arch/sh/mm/init.c
 *======================================================================*/
void free_initmem(void)
{
	unsigned long addr;

	addr = (unsigned long)(&__init_begin);
	for (; addr < (unsigned long)(&__init_end); addr += PAGE_SIZE) {
		ClearPageReserved(virt_to_page(addr));
		init_page_count(virt_to_page(addr));
		free_page(addr);
		totalram_pages++;
	}
	printk (KERN_INFO "Freeing unused kernel memory: %dk freed\n", \
	  (int)(&__init_end - &__init_begin) >> 10);
}

#ifdef CONFIG_BLK_DEV_INITRD
/*======================================================================*
 * free_initrd_mem() :
 * orig : arch/sh/mm/init.c
 *======================================================================*/
void free_initrd_mem(unsigned long start, unsigned long end)
{
	unsigned long p;
	for (p = start; p < end; p += PAGE_SIZE) {
		ClearPageReserved(virt_to_page(p));
		init_page_count(virt_to_page(p));
		free_page(p);
		totalram_pages++;
	}
	printk (KERN_INFO "Freeing initrd memory: %ldk freed\n", (end - start) >> 10);
}
#endif
14'>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
/*
 *   (c) 2003-2010 Advanced Micro Devices, Inc.
 *  Your use of this code is subject to the terms and conditions of the
 *  GNU general public license version 2. See "COPYING" or
 *  http://www.gnu.org/licenses/gpl.html
 *
 *  Support : mark.langsdorf@amd.com
 *
 *  Based on the powernow-k7.c module written by Dave Jones.
 *  (C) 2003 Dave Jones on behalf of SuSE Labs
 *  (C) 2004 Dominik Brodowski <linux@brodo.de>
 *  (C) 2004 Pavel Machek <pavel@ucw.cz>
 *  Licensed under the terms of the GNU GPL License version 2.
 *  Based upon datasheets & sample CPUs kindly provided by AMD.
 *
 *  Valuable input gratefully received from Dave Jones, Pavel Machek,
 *  Dominik Brodowski, Jacob Shin, and others.
 *  Originally developed by Paul Devriendt.
 *  Processor information obtained from Chapter 9 (Power and Thermal Management)
 *  of the "BIOS and Kernel Developer's Guide for the AMD Athlon 64 and AMD
 *  Opteron Processors" available for download from www.amd.com
 *
 *  Tables for specific CPUs can be inferred from
 *     http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/30430.pdf
 */

#include <linux/kernel.h>
#include <linux/smp.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/cpufreq.h>
#include <linux/slab.h>
#include <linux/string.h>
#include <linux/cpumask.h>
#include <linux/sched.h>	/* for current / set_cpus_allowed() */
#include <linux/io.h>
#include <linux/delay.h>

#include <asm/msr.h>

#include <linux/acpi.h>
#include <linux/mutex.h>
#include <acpi/processor.h>

#define PFX "powernow-k8: "
#define VERSION "version 2.20.00"
#include "powernow-k8.h"
#include "mperf.h"

/* serialize freq changes  */
static DEFINE_MUTEX(fidvid_mutex);

static DEFINE_PER_CPU(struct powernow_k8_data *, powernow_data);

static int cpu_family = CPU_OPTERON;

/* array to map SW pstate number to acpi state */
static u32 ps_to_as[8];

/* core performance boost */
static bool cpb_capable, cpb_enabled;
static struct msr __percpu *msrs;

static struct cpufreq_driver cpufreq_amd64_driver;

#ifndef CONFIG_SMP
static inline const struct cpumask *cpu_core_mask(int cpu)
{
	return cpumask_of(0);
}
#endif

/* Return a frequency in MHz, given an input fid */
static u32 find_freq_from_fid(u32 fid)
{
	return 800 + (fid * 100);
}

/* Return a frequency in KHz, given an input fid */
static u32 find_khz_freq_from_fid(u32 fid)
{
	return 1000 * find_freq_from_fid(fid);
}

static u32 find_khz_freq_from_pstate(struct cpufreq_frequency_table *data,
				     u32 pstate)
{
	return data[ps_to_as[pstate]].frequency;
}

/* Return the vco fid for an input fid
 *
 * Each "low" fid has corresponding "high" fid, and you can get to "low" fids
 * only from corresponding high fids. This returns "high" fid corresponding to
 * "low" one.
 */
static u32 convert_fid_to_vco_fid(u32 fid)
{
	if (fid < HI_FID_TABLE_BOTTOM)
		return 8 + (2 * fid);
	else
		return fid;
}

/*
 * Return 1 if the pending bit is set. Unless we just instructed the processor
 * to transition to a new state, seeing this bit set is really bad news.
 */
static int pending_bit_stuck(void)
{
	u32 lo, hi;

	if (cpu_family == CPU_HW_PSTATE)
		return 0;

	rdmsr(MSR_FIDVID_STATUS, lo, hi);
	return lo & MSR_S_LO_CHANGE_PENDING ? 1 : 0;
}

/*
 * Update the global current fid / vid values from the status msr.
 * Returns 1 on error.
 */
static int query_current_values_with_pending_wait(struct powernow_k8_data *data)
{
	u32 lo, hi;
	u32 i = 0;

	if (cpu_family == CPU_HW_PSTATE) {
		rdmsr(MSR_PSTATE_STATUS, lo, hi);
		i = lo & HW_PSTATE_MASK;
		data->currpstate = i;

		/*
		 * a workaround for family 11h erratum 311 might cause
		 * an "out-of-range Pstate if the core is in Pstate-0
		 */
		if ((boot_cpu_data.x86 == 0x11) && (i >= data->numps))
			data->currpstate = HW_PSTATE_0;

		return 0;
	}
	do {
		if (i++ > 10000) {
			pr_debug("detected change pending stuck\n");
			return 1;
		}
		rdmsr(MSR_FIDVID_STATUS, lo, hi);
	} while (lo & MSR_S_LO_CHANGE_PENDING);

	data->currvid = hi & MSR_S_HI_CURRENT_VID;
	data->currfid = lo & MSR_S_LO_CURRENT_FID;

	return 0;
}

/* the isochronous relief time */
static void count_off_irt(struct powernow_k8_data *data)
{
	udelay((1 << data->irt) * 10);
	return;
}

/* the voltage stabilization time */
static void count_off_vst(struct powernow_k8_data *data)
{
	udelay(data->vstable * VST_UNITS_20US);
	return;
}

/* need to init the control msr to a safe value (for each cpu) */
static void fidvid_msr_init(void)
{
	u32 lo, hi;
	u8 fid, vid;

	rdmsr(MSR_FIDVID_STATUS, lo, hi);
	vid = hi & MSR_S_HI_CURRENT_VID;
	fid = lo & MSR_S_LO_CURRENT_FID;
	lo = fid | (vid << MSR_C_LO_VID_SHIFT);
	hi = MSR_C_HI_STP_GNT_BENIGN;
	pr_debug("cpu%d, init lo 0x%x, hi 0x%x\n", smp_processor_id(), lo, hi);
	wrmsr(MSR_FIDVID_CTL, lo, hi);
}

/* write the new fid value along with the other control fields to the msr */
static int write_new_fid(struct powernow_k8_data *data, u32 fid)
{
	u32 lo;
	u32 savevid = data->currvid;
	u32 i = 0;

	if ((fid & INVALID_FID_MASK) || (data->currvid & INVALID_VID_MASK)) {
		printk(KERN_ERR PFX "internal error - overflow on fid write\n");
		return 1;
	}

	lo = fid;
	lo |= (data->currvid << MSR_C_LO_VID_SHIFT);
	lo |= MSR_C_LO_INIT_FID_VID;

	pr_debug("writing fid 0x%x, lo 0x%x, hi 0x%x\n",
		fid, lo, data->plllock * PLL_LOCK_CONVERSION);

	do {
		wrmsr(MSR_FIDVID_CTL, lo, data->plllock * PLL_LOCK_CONVERSION);
		if (i++ > 100) {
			printk(KERN_ERR PFX
				"Hardware error - pending bit very stuck - "
				"no further pstate changes possible\n");
			return 1;
		}
	} while (query_current_values_with_pending_wait(data));

	count_off_irt(data);

	if (savevid != data->currvid) {
		printk(KERN_ERR PFX
			"vid change on fid trans, old 0x%x, new 0x%x\n",
			savevid, data->currvid);
		return 1;
	}

	if (fid != data->currfid) {
		printk(KERN_ERR PFX
			"fid trans failed, fid 0x%x, curr 0x%x\n", fid,
			data->currfid);
		return 1;
	}

	return 0;
}

/* Write a new vid to the hardware */
static int write_new_vid(struct powernow_k8_data *data, u32 vid)
{
	u32 lo;
	u32 savefid = data->currfid;
	int i = 0;

	if ((data->currfid & INVALID_FID_MASK) || (vid & INVALID_VID_MASK)) {
		printk(KERN_ERR PFX "internal error - overflow on vid write\n");
		return 1;
	}

	lo = data->currfid;
	lo |= (vid << MSR_C_LO_VID_SHIFT);
	lo |= MSR_C_LO_INIT_FID_VID;

	pr_debug("writing vid 0x%x, lo 0x%x, hi 0x%x\n",
		vid, lo, STOP_GRANT_5NS);

	do {
		wrmsr(MSR_FIDVID_CTL, lo, STOP_GRANT_5NS);
		if (i++ > 100) {
			printk(KERN_ERR PFX "internal error - pending bit "
					"very stuck - no further pstate "
					"changes possible\n");
			return 1;
		}
	} while (query_current_values_with_pending_wait(data));

	if (savefid != data->currfid) {
		printk(KERN_ERR PFX "fid changed on vid trans, old "
			"0x%x new 0x%x\n",
		       savefid, data->currfid);
		return 1;
	}

	if (vid != data->currvid) {
		printk(KERN_ERR PFX "vid trans failed, vid 0x%x, "
				"curr 0x%x\n",
				vid, data->currvid);
		return 1;
	}

	return 0;
}

/*
 * Reduce the vid by the max of step or reqvid.
 * Decreasing vid codes represent increasing voltages:
 * vid of 0 is 1.550V, vid of 0x1e is 0.800V, vid of VID_OFF is off.
 */
static int decrease_vid_code_by_step(struct powernow_k8_data *data,
		u32 reqvid, u32 step)
{
	if ((data->currvid - reqvid) > step)
		reqvid = data->currvid - step;

	if (write_new_vid(data, reqvid))
		return 1;

	count_off_vst(data);

	return 0;
}

/* Change hardware pstate by single MSR write */
static int transition_pstate(struct powernow_k8_data *data, u32 pstate)
{
	wrmsr(MSR_PSTATE_CTRL, pstate, 0);
	data->currpstate = pstate;
	return 0;
}

/* Change Opteron/Athlon64 fid and vid, by the 3 phases. */
static int transition_fid_vid(struct powernow_k8_data *data,
		u32 reqfid, u32 reqvid)
{
	if (core_voltage_pre_transition(data, reqvid, reqfid))
		return 1;

	if (core_frequency_transition(data, reqfid))
		return 1;

	if (core_voltage_post_transition(data, reqvid))
		return 1;

	if (query_current_values_with_pending_wait(data))
		return 1;

	if ((reqfid != data->currfid) || (reqvid != data->currvid)) {
		printk(KERN_ERR PFX "failed (cpu%d): req 0x%x 0x%x, "
				"curr 0x%x 0x%x\n",
				smp_processor_id(),
				reqfid, reqvid, data->currfid, data->currvid);
		return 1;
	}

	pr_debug("transitioned (cpu%d): new fid 0x%x, vid 0x%x\n",
		smp_processor_id(), data->currfid, data->currvid);

	return 0;
}

/* Phase 1 - core voltage transition ... setup voltage */
static int core_voltage_pre_transition(struct powernow_k8_data *data,
		u32 reqvid, u32 reqfid)
{
	u32 rvosteps = data->rvo;
	u32 savefid = data->currfid;
	u32 maxvid, lo, rvomult = 1;

	pr_debug("ph1 (cpu%d): start, currfid 0x%x, currvid 0x%x, "
		"reqvid 0x%x, rvo 0x%x\n",
		smp_processor_id(),
		data->currfid, data->currvid, reqvid, data->rvo);

	if ((savefid < LO_FID_TABLE_TOP) && (reqfid < LO_FID_TABLE_TOP))
		rvomult = 2;
	rvosteps *= rvomult;
	rdmsr(MSR_FIDVID_STATUS, lo, maxvid);
	maxvid = 0x1f & (maxvid >> 16);
	pr_debug("ph1 maxvid=0x%x\n", maxvid);
	if (reqvid < maxvid) /* lower numbers are higher voltages */
		reqvid = maxvid;

	while (data->currvid > reqvid) {
		pr_debug("ph1: curr 0x%x, req vid 0x%x\n",
			data->currvid, reqvid);
		if (decrease_vid_code_by_step(data, reqvid, data->vidmvs))
			return 1;
	}

	while ((rvosteps > 0) &&
			((rvomult * data->rvo + data->currvid) > reqvid)) {
		if (data->currvid == maxvid) {
			rvosteps = 0;
		} else {
			pr_debug("ph1: changing vid for rvo, req 0x%x\n",
				data->currvid - 1);
			if (decrease_vid_code_by_step(data, data->currvid-1, 1))
				return 1;
			rvosteps--;
		}
	}

	if (query_current_values_with_pending_wait(data))
		return 1;

	if (savefid != data->currfid) {
		printk(KERN_ERR PFX "ph1 err, currfid changed 0x%x\n",
				data->currfid);
		return 1;
	}

	pr_debug("ph1 complete, currfid 0x%x, currvid 0x%x\n",
		data->currfid, data->currvid);

	return 0;
}

/* Phase 2 - core frequency transition */
static int core_frequency_transition(struct powernow_k8_data *data, u32 reqfid)
{
	u32 vcoreqfid, vcocurrfid, vcofiddiff;
	u32 fid_interval, savevid = data->currvid;

	if (data->currfid == reqfid) {
		printk(KERN_ERR PFX "ph2 null fid transition 0x%x\n",
				data->currfid);
		return 0;
	}

	pr_debug("ph2 (cpu%d): starting, currfid 0x%x, currvid 0x%x, "
		"reqfid 0x%x\n",
		smp_processor_id(),
		data->currfid, data->currvid, reqfid);

	vcoreqfid = convert_fid_to_vco_fid(reqfid);
	vcocurrfid = convert_fid_to_vco_fid(data->currfid);
	vcofiddiff = vcocurrfid > vcoreqfid ? vcocurrfid - vcoreqfid
	    : vcoreqfid - vcocurrfid;

	if ((reqfid <= LO_FID_TABLE_TOP) && (data->currfid <= LO_FID_TABLE_TOP))
		vcofiddiff = 0;

	while (vcofiddiff > 2) {
		(data->currfid & 1) ? (fid_interval = 1) : (fid_interval = 2);

		if (reqfid > data->currfid) {
			if (data->currfid > LO_FID_TABLE_TOP) {
				if (write_new_fid(data,
						data->currfid + fid_interval))
					return 1;
			} else {
				if (write_new_fid
				    (data,
				     2 + convert_fid_to_vco_fid(data->currfid)))
					return 1;
			}
		} else {
			if (write_new_fid(data, data->currfid - fid_interval))
				return 1;
		}

		vcocurrfid = convert_fid_to_vco_fid(data->currfid);
		vcofiddiff = vcocurrfid > vcoreqfid ? vcocurrfid - vcoreqfid
		    : vcoreqfid - vcocurrfid;
	}

	if (write_new_fid(data, reqfid))
		return 1;

	if (query_current_values_with_pending_wait(data))
		return 1;

	if (data->currfid != reqfid) {
		printk(KERN_ERR PFX
			"ph2: mismatch, failed fid transition, "
			"curr 0x%x, req 0x%x\n",
			data->currfid, reqfid);
		return 1;
	}

	if (savevid != data->currvid) {
		printk(KERN_ERR PFX "ph2: vid changed, save 0x%x, curr 0x%x\n",
			savevid, data->currvid);
		return 1;
	}

	pr_debug("ph2 complete, currfid 0x%x, currvid 0x%x\n",
		data->currfid, data->currvid);

	return 0;
}

/* Phase 3 - core voltage transition flow ... jump to the final vid. */
static int core_voltage_post_transition(struct powernow_k8_data *data,
		u32 reqvid)
{
	u32 savefid = data->currfid;
	u32 savereqvid = reqvid;

	pr_debug("ph3 (cpu%d): starting, currfid 0x%x, currvid 0x%x\n",
		smp_processor_id(),
		data->currfid, data->currvid);

	if (reqvid != data->currvid) {
		if (write_new_vid(data, reqvid))
			return 1;

		if (savefid != data->currfid) {
			printk(KERN_ERR PFX
			       "ph3: bad fid change, save 0x%x, curr 0x%x\n",
			       savefid, data->currfid);
			return 1;
		}

		if (data->currvid != reqvid) {
			printk(KERN_ERR PFX
			       "ph3: failed vid transition\n, "
			       "req 0x%x, curr 0x%x",
			       reqvid, data->currvid);
			return 1;
		}
	}

	if (query_current_values_with_pending_wait(data))
		return 1;

	if (savereqvid != data->currvid) {
		pr_debug("ph3 failed, currvid 0x%x\n", data->currvid);
		return 1;
	}

	if (savefid != data->currfid) {
		pr_debug("ph3 failed, currfid changed 0x%x\n",
			data->currfid);
		return 1;
	}

	pr_debug("ph3 complete, currfid 0x%x, currvid 0x%x\n",
		data->currfid, data->currvid);

	return 0;
}

static void check_supported_cpu(void *_rc)
{
	u32 eax, ebx, ecx, edx;
	int *rc = _rc;

	*rc = -ENODEV;

	if (__this_cpu_read(cpu_info.x86_vendor) != X86_VENDOR_AMD)
		return;

	eax = cpuid_eax(CPUID_PROCESSOR_SIGNATURE);
	if (((eax & CPUID_XFAM) != CPUID_XFAM_K8) &&
	    ((eax & CPUID_XFAM) < CPUID_XFAM_10H))
		return;

	if ((eax & CPUID_XFAM) == CPUID_XFAM_K8) {
		if (((eax & CPUID_USE_XFAM_XMOD) != CPUID_USE_XFAM_XMOD) ||
		    ((eax & CPUID_XMOD) > CPUID_XMOD_REV_MASK)) {
			printk(KERN_INFO PFX
				"Processor cpuid %x not supported\n", eax);
			return;
		}

		eax = cpuid_eax(CPUID_GET_MAX_CAPABILITIES);
		if (eax < CPUID_FREQ_VOLT_CAPABILITIES) {
			printk(KERN_INFO PFX
			       "No frequency change capabilities detected\n");
			return;
		}

		cpuid(CPUID_FREQ_VOLT_CAPABILITIES, &eax, &ebx, &ecx, &edx);
		if ((edx & P_STATE_TRANSITION_CAPABLE)
			!= P_STATE_TRANSITION_CAPABLE) {
			printk(KERN_INFO PFX
				"Power state transitions not supported\n");
			return;
		}
	} else { /* must be a HW Pstate capable processor */
		cpuid(CPUID_FREQ_VOLT_CAPABILITIES, &eax, &ebx, &ecx, &edx);
		if ((edx & USE_HW_PSTATE) == USE_HW_PSTATE)
			cpu_family = CPU_HW_PSTATE;
		else
			return;
	}

	*rc = 0;
}

static int check_pst_table(struct powernow_k8_data *data, struct pst_s *pst,
		u8 maxvid)
{
	unsigned int j;
	u8 lastfid = 0xff;

	for (j = 0; j < data->numps; j++) {
		if (pst[j].vid > LEAST_VID) {
			printk(KERN_ERR FW_BUG PFX "vid %d invalid : 0x%x\n",
			       j, pst[j].vid);
			return -EINVAL;
		}
		if (pst[j].vid < data->rvo) {
			/* vid + rvo >= 0 */
			printk(KERN_ERR FW_BUG PFX "0 vid exceeded with pstate"
			       " %d\n", j);
			return -ENODEV;
		}
		if (pst[j].vid < maxvid + data->rvo) {
			/* vid + rvo >= maxvid */
			printk(KERN_ERR FW_BUG PFX "maxvid exceeded with pstate"
			       " %d\n", j);
			return -ENODEV;
		}
		if (pst[j].fid > MAX_FID) {
			printk(KERN_ERR FW_BUG PFX "maxfid exceeded with pstate"
			       " %d\n", j);
			return -ENODEV;
		}
		if (j && (pst[j].fid < HI_FID_TABLE_BOTTOM)) {
			/* Only first fid is allowed to be in "low" range */
			printk(KERN_ERR FW_BUG PFX "two low fids - %d : "
			       "0x%x\n", j, pst[j].fid);
			return -EINVAL;
		}
		if (pst[j].fid < lastfid)
			lastfid = pst[j].fid;
	}
	if (lastfid & 1) {
		printk(KERN_ERR FW_BUG PFX "lastfid invalid\n");
		return -EINVAL;
	}
	if (lastfid > LO_FID_TABLE_TOP)
		printk(KERN_INFO FW_BUG PFX
			"first fid not from lo freq table\n");

	return 0;
}

static void invalidate_entry(struct cpufreq_frequency_table *powernow_table,
		unsigned int entry)
{
	powernow_table[entry].frequency = CPUFREQ_ENTRY_INVALID;
}

static void print_basics(struct powernow_k8_data *data)
{
	int j;
	for (j = 0; j < data->numps; j++) {
		if (data->powernow_table[j].frequency !=
				CPUFREQ_ENTRY_INVALID) {
			if (cpu_family == CPU_HW_PSTATE) {
				printk(KERN_INFO PFX
					"   %d : pstate %d (%d MHz)\n", j,
					data->powernow_table[j].index,
					data->powernow_table[j].frequency/1000);
			} else {
				printk(KERN_INFO PFX
					"fid 0x%x (%d MHz), vid 0x%x\n",
					data->powernow_table[j].index & 0xff,
					data->powernow_table[j].frequency/1000,
					data->powernow_table[j].index >> 8);
			}
		}
	}
	if (data->batps)
		printk(KERN_INFO PFX "Only %d pstates on battery\n",
				data->batps);
}

static u32 freq_from_fid_did(u32 fid, u32 did)
{
	u32 mhz = 0;

	if (boot_cpu_data.x86 == 0x10)
		mhz = (100 * (fid + 0x10)) >> did;
	else if (boot_cpu_data.x86 == 0x11)
		mhz = (100 * (fid + 8)) >> did;
	else
		BUG();

	return mhz * 1000;
}

static int fill_powernow_table(struct powernow_k8_data *data,
		struct pst_s *pst, u8 maxvid)
{
	struct cpufreq_frequency_table *powernow_table;
	unsigned int j;

	if (data->batps) {
		/* use ACPI support to get full speed on mains power */
		printk(KERN_WARNING PFX
			"Only %d pstates usable (use ACPI driver for full "
			"range\n", data->batps);
		data->numps = data->batps;
	}

	for (j = 1; j < data->numps; j++) {
		if (pst[j-1].fid >= pst[j].fid) {
			printk(KERN_ERR PFX "PST out of sequence\n");
			return -EINVAL;
		}
	}

	if (data->numps < 2) {
		printk(KERN_ERR PFX "no p states to transition\n");
		return -ENODEV;
	}

	if (check_pst_table(data, pst, maxvid))
		return -EINVAL;

	powernow_table = kmalloc((sizeof(struct cpufreq_frequency_table)
		* (data->numps + 1)), GFP_KERNEL);
	if (!powernow_table) {
		printk(KERN_ERR PFX "powernow_table memory alloc failure\n");
		return -ENOMEM;
	}

	for (j = 0; j < data->numps; j++) {
		int freq;
		powernow_table[j].index = pst[j].fid; /* lower 8 bits */
		powernow_table[j].index |= (pst[j].vid << 8); /* upper 8 bits */
		freq = find_khz_freq_from_fid(pst[j].fid);
		powernow_table[j].frequency = freq;
	}
	powernow_table[data->numps].frequency = CPUFREQ_TABLE_END;
	powernow_table[data->numps].index = 0;

	if (query_current_values_with_pending_wait(data)) {
		kfree(powernow_table);
		return -EIO;
	}

	pr_debug("cfid 0x%x, cvid 0x%x\n", data->currfid, data->currvid);
	data->powernow_table = powernow_table;
	if (cpumask_first(cpu_core_mask(data->cpu)) == data->cpu)
		print_basics(data);

	for (j = 0; j < data->numps; j++)
		if ((pst[j].fid == data->currfid) &&
		    (pst[j].vid == data->currvid))
			return 0;

	pr_debug("currfid/vid do not match PST, ignoring\n");
	return 0;
}

/* Find and validate the PSB/PST table in BIOS. */
static int find_psb_table(struct powernow_k8_data *data)
{
	struct psb_s *psb;
	unsigned int i;
	u32 mvs;
	u8 maxvid;
	u32 cpst = 0;
	u32 thiscpuid;

	for (i = 0xc0000; i < 0xffff0; i += 0x10) {
		/* Scan BIOS looking for the signature. */
		/* It can not be at ffff0 - it is too big. */

		psb = phys_to_virt(i);
		if (memcmp(psb, PSB_ID_STRING, PSB_ID_STRING_LEN) != 0)
			continue;

		pr_debug("found PSB header at 0x%p\n", psb);

		pr_debug("table vers: 0x%x\n", psb->tableversion);
		if (psb->tableversion != PSB_VERSION_1_4) {
			printk(KERN_ERR FW_BUG PFX "PSB table is not v1.4\n");
			return -ENODEV;
		}

		pr_debug("flags: 0x%x\n", psb->flags1);
		if (psb->flags1) {
			printk(KERN_ERR FW_BUG PFX "unknown flags\n");
			return -ENODEV;
		}

		data->vstable = psb->vstable;
		pr_debug("voltage stabilization time: %d(*20us)\n",
				data->vstable);

		pr_debug("flags2: 0x%x\n", psb->flags2);
		data->rvo = psb->flags2 & 3;
		data->irt = ((psb->flags2) >> 2) & 3;
		mvs = ((psb->flags2) >> 4) & 3;
		data->vidmvs = 1 << mvs;
		data->batps = ((psb->flags2) >> 6) & 3;

		pr_debug("ramp voltage offset: %d\n", data->rvo);
		pr_debug("isochronous relief time: %d\n", data->irt);
		pr_debug("maximum voltage step: %d - 0x%x\n", mvs, data->vidmvs);

		pr_debug("numpst: 0x%x\n", psb->num_tables);
		cpst = psb->num_tables;
		if ((psb->cpuid == 0x00000fc0) ||
		    (psb->cpuid == 0x00000fe0)) {
			thiscpuid = cpuid_eax(CPUID_PROCESSOR_SIGNATURE);
			if ((thiscpuid == 0x00000fc0) ||
			    (thiscpuid == 0x00000fe0))
				cpst = 1;
		}
		if (cpst != 1) {
			printk(KERN_ERR FW_BUG PFX "numpst must be 1\n");
			return -ENODEV;
		}

		data->plllock = psb->plllocktime;
		pr_debug("plllocktime: 0x%x (units 1us)\n", psb->plllocktime);
		pr_debug("maxfid: 0x%x\n", psb->maxfid);
		pr_debug("maxvid: 0x%x\n", psb->maxvid);
		maxvid = psb->maxvid;

		data->numps = psb->numps;
		pr_debug("numpstates: 0x%x\n", data->numps);
		return fill_powernow_table(data,
				(struct pst_s *)(psb+1), maxvid);
	}
	/*
	 * If you see this message, complain to BIOS manufacturer. If
	 * he tells you "we do not support Linux" or some similar
	 * nonsense, remember that Windows 2000 uses the same legacy
	 * mechanism that the old Linux PSB driver uses. Tell them it
	 * is broken with Windows 2000.
	 *
	 * The reference to the AMD documentation is chapter 9 in the
	 * BIOS and Kernel Developer's Guide, which is available on
	 * www.amd.com
	 */
	printk(KERN_ERR FW_BUG PFX "No PSB or ACPI _PSS objects\n");
	printk(KERN_ERR PFX "Make sure that your BIOS is up to date"
		" and Cool'N'Quiet support is enabled in BIOS setup\n");
	return -ENODEV;
}

static void powernow_k8_acpi_pst_values(struct powernow_k8_data *data,
		unsigned int index)
{
	u64 control;

	if (!data->acpi_data.state_count || (cpu_family == CPU_HW_PSTATE))
		return;

	control = data->acpi_data.states[index].control;
	data->irt = (control >> IRT_SHIFT) & IRT_MASK;
	data->rvo = (control >> RVO_SHIFT) & RVO_MASK;
	data->exttype = (control >> EXT_TYPE_SHIFT) & EXT_TYPE_MASK;
	data->plllock = (control >> PLL_L_SHIFT) & PLL_L_MASK;
	data->vidmvs = 1 << ((control >> MVS_SHIFT) & MVS_MASK);
	data->vstable = (control >> VST_SHIFT) & VST_MASK;
}

static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data)
{
	struct cpufreq_frequency_table *powernow_table;
	int ret_val = -ENODEV;
	u64 control, status;

	if (acpi_processor_register_performance(&data->acpi_data, data->cpu)) {
		pr_debug("register performance failed: bad ACPI data\n");
		return -EIO;
	}

	/* verify the data contained in the ACPI structures */
	if (data->acpi_data.state_count <= 1) {
		pr_debug("No ACPI P-States\n");
		goto err_out;
	}

	control = data->acpi_data.control_register.space_id;
	status = data->acpi_data.status_register.space_id;

	if ((control != ACPI_ADR_SPACE_FIXED_HARDWARE) ||
	    (status != ACPI_ADR_SPACE_FIXED_HARDWARE)) {
		pr_debug("Invalid control/status registers (%llx - %llx)\n",
			control, status);
		goto err_out;
	}

	/* fill in data->powernow_table */
	powernow_table = kmalloc((sizeof(struct cpufreq_frequency_table)
		* (data->acpi_data.state_count + 1)), GFP_KERNEL);
	if (!powernow_table) {
		pr_debug("powernow_table memory alloc failure\n");
		goto err_out;
	}

	/* fill in data */
	data->numps = data->acpi_data.state_count;
	powernow_k8_acpi_pst_values(data, 0);

	if (cpu_family == CPU_HW_PSTATE)
		ret_val = fill_powernow_table_pstate(data, powernow_table);
	else
		ret_val = fill_powernow_table_fidvid(data, powernow_table);
	if (ret_val)
		goto err_out_mem;

	powernow_table[data->acpi_data.state_count].frequency =
		CPUFREQ_TABLE_END;
	powernow_table[data->acpi_data.state_count].index = 0;
	data->powernow_table = powernow_table;

	if (cpumask_first(cpu_core_mask(data->cpu)) == data->cpu)
		print_basics(data);

	/* notify BIOS that we exist */
	acpi_processor_notify_smm(THIS_MODULE);

	if (!zalloc_cpumask_var(&data->acpi_data.shared_cpu_map, GFP_KERNEL)) {
		printk(KERN_ERR PFX
				"unable to alloc powernow_k8_data cpumask\n");
		ret_val = -ENOMEM;
		goto err_out_mem;
	}

	return 0;

err_out_mem:
	kfree(powernow_table);

err_out:
	acpi_processor_unregister_performance(&data->acpi_data, data->cpu);

	/* data->acpi_data.state_count informs us at ->exit()
	 * whether ACPI was used */
	data->acpi_data.state_count = 0;

	return ret_val;
}

static int fill_powernow_table_pstate(struct powernow_k8_data *data,
		struct cpufreq_frequency_table *powernow_table)
{
	int i;
	u32 hi = 0, lo = 0;
	rdmsr(MSR_PSTATE_CUR_LIMIT, lo, hi);
	data->max_hw_pstate = (lo & HW_PSTATE_MAX_MASK) >> HW_PSTATE_MAX_SHIFT;

	for (i = 0; i < data->acpi_data.state_count; i++) {
		u32 index;

		index = data->acpi_data.states[i].control & HW_PSTATE_MASK;
		if (index > data->max_hw_pstate) {
			printk(KERN_ERR PFX "invalid pstate %d - "
					"bad value %d.\n", i, index);
			printk(KERN_ERR PFX "Please report to BIOS "
					"manufacturer\n");
			invalidate_entry(powernow_table, i);
			continue;
		}

		ps_to_as[index] = i;

		/* Frequency may be rounded for these */
		if ((boot_cpu_data.x86 == 0x10 && boot_cpu_data.x86_model < 10)
				 || boot_cpu_data.x86 == 0x11) {

			rdmsr(MSR_PSTATE_DEF_BASE + index, lo, hi);
			if (!(hi & HW_PSTATE_VALID_MASK)) {
				pr_debug("invalid pstate %d, ignoring\n", index);
				invalidate_entry(powernow_table, i);
				continue;
			}

			powernow_table[i].frequency =
				freq_from_fid_did(lo & 0x3f, (lo >> 6) & 7);
		} else
			powernow_table[i].frequency =
				data->acpi_data.states[i].core_frequency * 1000;

		powernow_table[i].index = index;
	}
	return 0;
}

static int fill_powernow_table_fidvid(struct powernow_k8_data *data,
		struct cpufreq_frequency_table *powernow_table)
{
	int i;

	for (i = 0; i < data->acpi_data.state_count; i++) {
		u32 fid;
		u32 vid;
		u32 freq, index;
		u64 status, control;

		if (data->exttype) {
			status =  data->acpi_data.states[i].status;
			fid = status & EXT_FID_MASK;
			vid = (status >> VID_SHIFT) & EXT_VID_MASK;
		} else {
			control =  data->acpi_data.states[i].control;
			fid = control & FID_MASK;
			vid = (control >> VID_SHIFT) & VID_MASK;
		}

		pr_debug("   %d : fid 0x%x, vid 0x%x\n", i, fid, vid);

		index = fid | (vid<<8);
		powernow_table[i].index = index;

		freq = find_khz_freq_from_fid(fid);
		powernow_table[i].frequency = freq;

		/* verify frequency is OK */
		if ((freq > (MAX_FREQ * 1000)) || (freq < (MIN_FREQ * 1000))) {
			pr_debug("invalid freq %u kHz, ignoring\n", freq);
			invalidate_entry(powernow_table, i);
			continue;
		}

		/* verify voltage is OK -
		 * BIOSs are using "off" to indicate invalid */
		if (vid == VID_OFF) {
			pr_debug("invalid vid %u, ignoring\n", vid);
			invalidate_entry(powernow_table, i);
			continue;
		}

		if (freq != (data->acpi_data.states[i].core_frequency * 1000)) {
			printk(KERN_INFO PFX "invalid freq entries "
				"%u kHz vs. %u kHz\n", freq,
				(unsigned int)
				(data->acpi_data.states[i].core_frequency
				 * 1000));
			invalidate_entry(powernow_table, i);
			continue;
		}
	}
	return 0;
}

static void powernow_k8_cpu_exit_acpi(struct powernow_k8_data *data)
{
	if (data->acpi_data.state_count)
		acpi_processor_unregister_performance(&data->acpi_data,
				data->cpu);
	free_cpumask_var(data->acpi_data.shared_cpu_map);
}

static int get_transition_latency(struct powernow_k8_data *data)
{
	int max_latency = 0;
	int i;
	for (i = 0; i < data->acpi_data.state_count; i++) {
		int cur_latency = data->acpi_data.states[i].transition_latency
			+ data->acpi_data.states[i].bus_master_latency;
		if (cur_latency > max_latency)
			max_latency = cur_latency;
	}
	if (max_latency == 0) {
		/*
		 * Fam 11h and later may return 0 as transition latency. This
		 * is intended and means "very fast". While cpufreq core and
		 * governors currently can handle that gracefully, better set it
		 * to 1 to avoid problems in the future.
		 */
		if (boot_cpu_data.x86 < 0x11)
			printk(KERN_ERR FW_WARN PFX "Invalid zero transition "
				"latency\n");
		max_latency = 1;
	}
	/* value in usecs, needs to be in nanoseconds */
	return 1000 * max_latency;
}

/* Take a frequency, and issue the fid/vid transition command */
static int transition_frequency_fidvid(struct powernow_k8_data *data,
		unsigned int index)
{
	u32 fid = 0;
	u32 vid = 0;
	int res, i;
	struct cpufreq_freqs freqs;

	pr_debug("cpu %d transition to index %u\n", smp_processor_id(), index);

	/* fid/vid correctness check for k8 */
	/* fid are the lower 8 bits of the index we stored into
	 * the cpufreq frequency table in find_psb_table, vid
	 * are the upper 8 bits.
	 */
	fid = data->powernow_table[index].index & 0xFF;
	vid = (data->powernow_table[index].index & 0xFF00) >> 8;

	pr_debug("table matched fid 0x%x, giving vid 0x%x\n", fid, vid);

	if (query_current_values_with_pending_wait(data))
		return 1;

	if ((data->currvid == vid) && (data->currfid == fid)) {
		pr_debug("target matches current values (fid 0x%x, vid 0x%x)\n",
			fid, vid);
		return 0;
	}

	pr_debug("cpu %d, changing to fid 0x%x, vid 0x%x\n",
		smp_processor_id(), fid, vid);
	freqs.old = find_khz_freq_from_fid(data->currfid);
	freqs.new = find_khz_freq_from_fid(fid);

	for_each_cpu(i, data->available_cores) {
		freqs.cpu = i;
		cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE);
	}

	res = transition_fid_vid(data, fid, vid);
	if (res)
		return res;

	freqs.new = find_khz_freq_from_fid(data->currfid);

	for_each_cpu(i, data->available_cores) {
		freqs.cpu = i;
		cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE);
	}
	return res;
}

/* Take a frequency, and issue the hardware pstate transition command */
static int transition_frequency_pstate(struct powernow_k8_data *data,
		unsigned int index)
{
	u32 pstate = 0;
	int res, i;
	struct cpufreq_freqs freqs;

	pr_debug("cpu %d transition to index %u\n", smp_processor_id(), index);

	/* get MSR index for hardware pstate transition */
	pstate = index & HW_PSTATE_MASK;
	if (pstate > data->max_hw_pstate)
		return -EINVAL;

	freqs.old = find_khz_freq_from_pstate(data->powernow_table,
			data->currpstate);
	freqs.new = find_khz_freq_from_pstate(data->powernow_table, pstate);

	for_each_cpu(i, data->available_cores) {
		freqs.cpu = i;
		cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE);
	}

	res = transition_pstate(data, pstate);
	freqs.new = find_khz_freq_from_pstate(data->powernow_table, pstate);

	for_each_cpu(i, data->available_cores) {
		freqs.cpu = i;
		cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE);
	}
	return res;
}

/* Driver entry point to switch to the target frequency */
static int powernowk8_target(struct cpufreq_policy *pol,
		unsigned targfreq, unsigned relation)
{
	cpumask_var_t oldmask;
	struct powernow_k8_data *data = per_cpu(powernow_data, pol->cpu);
	u32 checkfid;
	u32 checkvid;
	unsigned int newstate;
	int ret = -EIO;

	if (!data)
		return -EINVAL;

	checkfid = data->currfid;
	checkvid = data->currvid;

	/* only run on specific CPU from here on. */
	/* This is poor form: use a workqueue or smp_call_function_single */
	if (!alloc_cpumask_var(&oldmask, GFP_KERNEL))
		return -ENOMEM;

	cpumask_copy(oldmask, tsk_cpus_allowed(current));
	set_cpus_allowed_ptr(current, cpumask_of(pol->cpu));

	if (smp_processor_id() != pol->cpu) {
		printk(KERN_ERR PFX "limiting to cpu %u failed\n", pol->cpu);
		goto err_out;
	}

	if (pending_bit_stuck()) {
		printk(KERN_ERR PFX "failing targ, change pending bit set\n");
		goto err_out;
	}

	pr_debug("targ: cpu %d, %d kHz, min %d, max %d, relation %d\n",
		pol->cpu, targfreq, pol->min, pol->max, relation);

	if (query_current_values_with_pending_wait(data))
		goto err_out;

	if (cpu_family != CPU_HW_PSTATE) {
		pr_debug("targ: curr fid 0x%x, vid 0x%x\n",
		data->currfid, data->currvid);

		if ((checkvid != data->currvid) ||
		    (checkfid != data->currfid)) {
			printk(KERN_INFO PFX
				"error - out of sync, fix 0x%x 0x%x, "
				"vid 0x%x 0x%x\n",
				checkfid, data->currfid,
				checkvid, data->currvid);
		}
	}

	if (cpufreq_frequency_table_target(pol, data->powernow_table,
				targfreq, relation, &newstate))
		goto err_out;

	mutex_lock(&fidvid_mutex);

	powernow_k8_acpi_pst_values(data, newstate);

	if (cpu_family == CPU_HW_PSTATE)
		ret = transition_frequency_pstate(data,
			data->powernow_table[newstate].index);
	else
		ret = transition_frequency_fidvid(data, newstate);
	if (ret) {
		printk(KERN_ERR PFX "transition frequency failed\n");
		ret = 1;
		mutex_unlock(&fidvid_mutex);
		goto err_out;
	}
	mutex_unlock(&fidvid_mutex);

	if (cpu_family == CPU_HW_PSTATE)
		pol->cur = find_khz_freq_from_pstate(data->powernow_table,
				data->powernow_table[newstate].index);
	else
		pol->cur = find_khz_freq_from_fid(data->currfid);
	ret = 0;

err_out:
	set_cpus_allowed_ptr(current, oldmask);
	free_cpumask_var(oldmask);
	return ret;
}

/* Driver entry point to verify the policy and range of frequencies */
static int powernowk8_verify(struct cpufreq_policy *pol)
{
	struct powernow_k8_data *data = per_cpu(powernow_data, pol->cpu);

	if (!data)
		return -EINVAL;

	return cpufreq_frequency_table_verify(pol, data->powernow_table);
}

struct init_on_cpu {
	struct powernow_k8_data *data;
	int rc;
};

static void __cpuinit powernowk8_cpu_init_on_cpu(void *_init_on_cpu)
{
	struct init_on_cpu *init_on_cpu = _init_on_cpu;

	if (pending_bit_stuck()) {
		printk(KERN_ERR PFX "failing init, change pending bit set\n");
		init_on_cpu->rc = -ENODEV;
		return;
	}

	if (query_current_values_with_pending_wait(init_on_cpu->data)) {
		init_on_cpu->rc = -ENODEV;
		return;
	}

	if (cpu_family == CPU_OPTERON)
		fidvid_msr_init();

	init_on_cpu->rc = 0;
}

/* per CPU init entry point to the driver */
static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol)
{
	static const char ACPI_PSS_BIOS_BUG_MSG[] =
		KERN_ERR FW_BUG PFX "No compatible ACPI _PSS objects found.\n"
		FW_BUG PFX "Try again with latest BIOS.\n";
	struct powernow_k8_data *data;
	struct init_on_cpu init_on_cpu;
	int rc;
	struct cpuinfo_x86 *c = &cpu_data(pol->cpu);

	if (!cpu_online(pol->cpu))
		return -ENODEV;

	smp_call_function_single(pol->cpu, check_supported_cpu, &rc, 1);
	if (rc)
		return -ENODEV;

	data = kzalloc(sizeof(struct powernow_k8_data), GFP_KERNEL);
	if (!data) {
		printk(KERN_ERR PFX "unable to alloc powernow_k8_data");
		return -ENOMEM;
	}

	data->cpu = pol->cpu;
	data->currpstate = HW_PSTATE_INVALID;

	if (powernow_k8_cpu_init_acpi(data)) {
		/*
		 * Use the PSB BIOS structure. This is only available on
		 * an UP version, and is deprecated by AMD.
		 */
		if (num_online_cpus() != 1) {
			printk_once(ACPI_PSS_BIOS_BUG_MSG);
			goto err_out;
		}
		if (pol->cpu != 0) {
			printk(KERN_ERR FW_BUG PFX "No ACPI _PSS objects for "
			       "CPU other than CPU0. Complain to your BIOS "
			       "vendor.\n");
			goto err_out;
		}
		rc = find_psb_table(data);
		if (rc)
			goto err_out;

		/* Take a crude guess here.
		 * That guess was in microseconds, so multiply with 1000 */
		pol->cpuinfo.transition_latency = (
			 ((data->rvo + 8) * data->vstable * VST_UNITS_20US) +
			 ((1 << data->irt) * 30)) * 1000;
	} else /* ACPI _PSS objects available */
		pol->cpuinfo.transition_latency = get_transition_latency(data);

	/* only run on specific CPU from here on */
	init_on_cpu.data = data;
	smp_call_function_single(data->cpu, powernowk8_cpu_init_on_cpu,
				 &init_on_cpu, 1);
	rc = init_on_cpu.rc;
	if (rc != 0)
		goto err_out_exit_acpi;

	if (cpu_family == CPU_HW_PSTATE)
		cpumask_copy(pol->cpus, cpumask_of(pol->cpu));
	else
		cpumask_copy(pol->cpus, cpu_core_mask(pol->cpu));
	data->available_cores = pol->cpus;

	if (cpu_family == CPU_HW_PSTATE)
		pol->cur = find_khz_freq_from_pstate(data->powernow_table,
				data->currpstate);
	else
		pol->cur = find_khz_freq_from_fid(data->currfid);
	pr_debug("policy current frequency %d kHz\n", pol->cur);

	/* min/max the cpu is capable of */
	if (cpufreq_frequency_table_cpuinfo(pol, data->powernow_table)) {
		printk(KERN_ERR FW_BUG PFX "invalid powernow_table\n");
		powernow_k8_cpu_exit_acpi(data);
		kfree(data->powernow_table);
		kfree(data);
		return -EINVAL;
	}

	/* Check for APERF/MPERF support in hardware */
	if (cpu_has(c, X86_FEATURE_APERFMPERF))
		cpufreq_amd64_driver.getavg = cpufreq_get_measured_perf;

	cpufreq_frequency_table_get_attr(data->powernow_table, pol->cpu);

	if (cpu_family == CPU_HW_PSTATE)
		pr_debug("cpu_init done, current pstate 0x%x\n",
				data->currpstate);
	else
		pr_debug("cpu_init done, current fid 0x%x, vid 0x%x\n",
			data->currfid, data->currvid);

	per_cpu(powernow_data, pol->cpu) = data;

	return 0;

err_out_exit_acpi:
	powernow_k8_cpu_exit_acpi(data);

err_out:
	kfree(data);
	return -ENODEV;
}

static int __devexit powernowk8_cpu_exit(struct cpufreq_policy *pol)
{
	struct powernow_k8_data *data = per_cpu(powernow_data, pol->cpu);

	if (!data)
		return -EINVAL;

	powernow_k8_cpu_exit_acpi(data);

	cpufreq_frequency_table_put_attr(pol->cpu);

	kfree(data->powernow_table);
	kfree(data);
	per_cpu(powernow_data, pol->cpu) = NULL;

	return 0;
}

static void query_values_on_cpu(void *_err)
{
	int *err = _err;
	struct powernow_k8_data *data = __this_cpu_read(powernow_data);

	*err = query_current_values_with_pending_wait(data);
}

static unsigned int powernowk8_get(unsigned int cpu)
{
	struct powernow_k8_data *data = per_cpu(powernow_data, cpu);
	unsigned int khz = 0;
	int err;

	if (!data)
		return 0;

	smp_call_function_single(cpu, query_values_on_cpu, &err, true);
	if (err)
		goto out;

	if (cpu_family == CPU_HW_PSTATE)
		khz = find_khz_freq_from_pstate(data->powernow_table,
						data->currpstate);
	else
		khz = find_khz_freq_from_fid(data->currfid);


out:
	return khz;
}

static void _cpb_toggle_msrs(bool t)
{
	int cpu;

	get_online_cpus();

	rdmsr_on_cpus(cpu_online_mask, MSR_K7_HWCR, msrs);

	for_each_cpu(cpu, cpu_online_mask) {
		struct msr *reg = per_cpu_ptr(msrs, cpu);
		if (t)
			reg->l &= ~BIT(25);
		else
			reg->l |= BIT(25);
	}
	wrmsr_on_cpus(cpu_online_mask, MSR_K7_HWCR, msrs);

	put_online_cpus();
}

/*
 * Switch on/off core performance boosting.
 *
 * 0=disable
 * 1=enable.
 */
static void cpb_toggle(bool t)
{
	if (!cpb_capable)
		return;

	if (t && !cpb_enabled) {
		cpb_enabled = true;
		_cpb_toggle_msrs(t);
		printk(KERN_INFO PFX "Core Boosting enabled.\n");
	} else if (!t && cpb_enabled) {
		cpb_enabled = false;
		_cpb_toggle_msrs(t);
		printk(KERN_INFO PFX "Core Boosting disabled.\n");
	}
}

static ssize_t store_cpb(struct cpufreq_policy *policy, const char *buf,
				 size_t count)
{
	int ret = -EINVAL;
	unsigned long val = 0;

	ret = strict_strtoul(buf, 10, &val);
	if (!ret && (val == 0 || val == 1) && cpb_capable)
		cpb_toggle(val);
	else
		return -EINVAL;

	return count;
}

static ssize_t show_cpb(struct cpufreq_policy *policy, char *buf)
{
	return sprintf(buf, "%u\n", cpb_enabled);
}

#define define_one_rw(_name) \
static struct freq_attr _name = \
__ATTR(_name, 0644, show_##_name, store_##_name)

define_one_rw(cpb);

static struct freq_attr *powernow_k8_attr[] = {
	&cpufreq_freq_attr_scaling_available_freqs,
	&cpb,
	NULL,
};

static struct cpufreq_driver cpufreq_amd64_driver = {
	.verify		= powernowk8_verify,
	.target		= powernowk8_target,
	.bios_limit	= acpi_processor_get_bios_limit,
	.init		= powernowk8_cpu_init,
	.exit		= __devexit_p(powernowk8_cpu_exit),
	.get		= powernowk8_get,
	.name		= "powernow-k8",
	.owner		= THIS_MODULE,
	.attr		= powernow_k8_attr,
};

/*
 * Clear the boost-disable flag on the CPU_DOWN path so that this cpu
 * cannot block the remaining ones from boosting. On the CPU_UP path we
 * simply keep the boost-disable flag in sync with the current global
 * state.
 */
static int cpb_notify(struct notifier_block *nb, unsigned long action,
		      void *hcpu)
{
	unsigned cpu = (long)hcpu;
	u32 lo, hi;

	switch (action) {
	case CPU_UP_PREPARE:
	case CPU_UP_PREPARE_FROZEN:

		if (!cpb_enabled) {
			rdmsr_on_cpu(cpu, MSR_K7_HWCR, &lo, &hi);
			lo |= BIT(25);
			wrmsr_on_cpu(cpu, MSR_K7_HWCR, lo, hi);
		}
		break;

	case CPU_DOWN_PREPARE:
	case CPU_DOWN_PREPARE_FROZEN:
		rdmsr_on_cpu(cpu, MSR_K7_HWCR, &lo, &hi);
		lo &= ~BIT(25);
		wrmsr_on_cpu(cpu, MSR_K7_HWCR, lo, hi);
		break;

	default:
		break;
	}

	return NOTIFY_OK;
}

static struct notifier_block cpb_nb = {
	.notifier_call		= cpb_notify,
};

/* driver entry point for init */
static int __cpuinit powernowk8_init(void)
{
	unsigned int i, supported_cpus = 0, cpu;
	int rv;

	for_each_online_cpu(i) {
		int rc;
		smp_call_function_single(i, check_supported_cpu, &rc, 1);
		if (rc == 0)
			supported_cpus++;
	}

	if (supported_cpus != num_online_cpus())
		return -ENODEV;

	printk(KERN_INFO PFX "Found %d %s (%d cpu cores) (" VERSION ")\n",
		num_online_nodes(), boot_cpu_data.x86_model_id, supported_cpus);

	if (boot_cpu_has(X86_FEATURE_CPB)) {

		cpb_capable = true;

		msrs = msrs_alloc();
		if (!msrs) {
			printk(KERN_ERR "%s: Error allocating msrs!\n", __func__);
			return -ENOMEM;
		}

		register_cpu_notifier(&cpb_nb);

		rdmsr_on_cpus(cpu_online_mask, MSR_K7_HWCR, msrs);

		for_each_cpu(cpu, cpu_online_mask) {
			struct msr *reg = per_cpu_ptr(msrs, cpu);
			cpb_enabled |= !(!!(reg->l & BIT(25)));
		}

		printk(KERN_INFO PFX "Core Performance Boosting: %s.\n",
			(cpb_enabled ? "on" : "off"));
	}

	rv = cpufreq_register_driver(&cpufreq_amd64_driver);
	if (rv < 0 && boot_cpu_has(X86_FEATURE_CPB)) {
		unregister_cpu_notifier(&cpb_nb);
		msrs_free(msrs);
		msrs = NULL;
	}
	return rv;
}

/* driver entry point for term */
static void __exit powernowk8_exit(void)
{
	pr_debug("exit\n");

	if (boot_cpu_has(X86_FEATURE_CPB)) {
		msrs_free(msrs);
		msrs = NULL;

		unregister_cpu_notifier(&cpb_nb);
	}

	cpufreq_unregister_driver(&cpufreq_amd64_driver);
}

MODULE_AUTHOR("Paul Devriendt <paul.devriendt@amd.com> and "
		"Mark Langsdorf <mark.langsdorf@amd.com>");
MODULE_DESCRIPTION("AMD Athlon 64 and Opteron processor frequency driver.");
MODULE_LICENSE("GPL");

late_initcall(powernowk8_init);
module_exit(powernowk8_exit);