powernow-k8.c 31.1 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6
/*
 *   (c) 2003, 2004 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
 *
D
Dave Jones 已提交
7
 *  Support : mark.langsdorf@amd.com
L
Linus Torvalds 已提交
8 9 10 11 12 13 14 15 16 17
 *
 *  Based on the powernow-k7.c module written by Dave Jones.
 *  (C) 2003 Dave Jones <davej@codemonkey.org.uk> on behalf of SuSE Labs
 *  (C) 2004 Dominik Brodowski <linux@brodo.de>
 *  (C) 2004 Pavel Machek <pavel@suse.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, and others.
D
Dave Jones 已提交
18
 *  Originally developed by Paul Devriendt.
L
Linus Torvalds 已提交
19 20 21 22 23
 *  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 infrerred from
D
Dave Jones 已提交
24
 *     http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/30430.pdf
L
Linus Torvalds 已提交
25 26 27 28 29 30 31 32 33
 */

#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>
D
Dave Jones 已提交
34
#include <linux/cpumask.h>
L
Linus Torvalds 已提交
35 36 37 38 39 40 41 42 43 44 45 46

#include <asm/msr.h>
#include <asm/io.h>
#include <asm/delay.h>

#ifdef CONFIG_X86_POWERNOW_K8_ACPI
#include <linux/acpi.h>
#include <acpi/processor.h>
#endif

#define PFX "powernow-k8: "
#define BFX PFX "BIOS error: "
D
Dave Jones 已提交
47
#define VERSION "version 1.40.2"
L
Linus Torvalds 已提交
48 49 50 51 52 53 54
#include "powernow-k8.h"

/* serialize freq changes  */
static DECLARE_MUTEX(fidvid_sem);

static struct powernow_k8_data *powernow_data[NR_CPUS];

D
Dave Jones 已提交
55 56 57 58
#ifndef CONFIG_SMP
static cpumask_t cpu_core_map[1];
#endif

L
Linus Torvalds 已提交
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
/* 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);
}

/* Return a voltage in miliVolts, given an input vid */
static u32 find_millivolts_from_vid(struct powernow_k8_data *data, u32 vid)
{
	return 1550-vid*25;
}

/* 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;

	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;

	lo = MSR_S_LO_CHANGE_PENDING;
	while (lo & MSR_S_LO_CHANGE_PENDING) {
		if (i++ > 0x1000000) {
			printk(KERN_ERR PFX "detected change pending stuck\n");
			return 1;
		}
		rdmsr(MSR_FIDVID_STATUS, lo, hi);
	}

	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 stabalization 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;
	dprintk("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;

	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 | (data->currvid << MSR_C_LO_VID_SHIFT) | MSR_C_LO_INIT_FID_VID;

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

	wrmsr(MSR_FIDVID_CTL, lo, data->plllock * PLL_LOCK_CONVERSION);

	if (query_current_values_with_pending_wait(data))
		return 1;

	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;

	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 | (vid << MSR_C_LO_VID_SHIFT) | MSR_C_LO_INIT_FID_VID;

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

	wrmsr(MSR_FIDVID_CTL, lo, STOP_GRANT_5NS);

	if (query_current_values_with_pending_wait(data))
		return 1;

	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 0x1f 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 the 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))
		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;
	}

	dprintk("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 rvosteps = data->rvo;
	u32 savefid = data->currfid;
D
Dave Jones 已提交
283
	u32 maxvid, lo;
L
Linus Torvalds 已提交
284 285 286 287 288

	dprintk("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);

D
Dave Jones 已提交
289 290 291 292 293 294
	rdmsr(MSR_FIDVID_STATUS, lo, maxvid);
	maxvid = 0x1f & (maxvid >> 16);
	dprintk("ph1 maxvid=0x%x\n", maxvid);
	if (reqvid < maxvid) /* lower numbers are higher voltages */
		reqvid = maxvid;

L
Linus Torvalds 已提交
295 296 297 298 299 300 301
	while (data->currvid > reqvid) {
		dprintk("ph1: curr 0x%x, req vid 0x%x\n",
			data->currvid, reqvid);
		if (decrease_vid_code_by_step(data, reqvid, data->vidmvs))
			return 1;
	}

D
Dave Jones 已提交
302 303
	while ((rvosteps > 0) && ((data->rvo + data->currvid) > reqvid)) {
		if (data->currvid == maxvid) {
L
Linus Torvalds 已提交
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
			rvosteps = 0;
		} else {
			dprintk("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;
	}

	dprintk("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, savevid = data->currvid;

	if ((reqfid < HI_FID_TABLE_BOTTOM) && (data->currfid < HI_FID_TABLE_BOTTOM)) {
		printk(KERN_ERR PFX "ph2: illegal lo-lo transition 0x%x 0x%x\n",
			reqfid, data->currfid);
		return 1;
	}

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

	dprintk("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;

	while (vcofiddiff > 2) {
		if (reqfid > data->currfid) {
			if (data->currfid > LO_FID_TABLE_TOP) {
				if (write_new_fid(data, data->currfid + 2)) {
					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 - 2))
				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;
	}

	dprintk("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;

	dprintk("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) {
		dprintk("ph3 failed, currvid 0x%x\n", data->currvid);
		return 1;
	}

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

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

	return 0;
}

static int check_supported_cpu(unsigned int cpu)
{
	cpumask_t oldmask = CPU_MASK_ALL;
	u32 eax, ebx, ecx, edx;
	unsigned int rc = 0;

	oldmask = current->cpus_allowed;
	set_cpus_allowed(current, cpumask_of_cpu(cpu));
	schedule();

	if (smp_processor_id() != cpu) {
		printk(KERN_ERR "limiting to cpu %u failed\n", cpu);
		goto out;
	}

	if (current_cpu_data.x86_vendor != X86_VENDOR_AMD)
		goto out;

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

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

	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");
		goto out;
	}

	rc = 1;

out:
	set_cpus_allowed(current, oldmask);
	schedule();
	return rc;

}

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

	return 0;
}

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)
			printk(KERN_INFO PFX "   %d : fid 0x%x (%d MHz), vid 0x%x (%d mV)\n", j,
				data->powernow_table[j].index & 0xff,
				data->powernow_table[j].frequency/1000,
				data->powernow_table[j].index >> 8,
				find_millivolts_from_vid(data, data->powernow_table[j].index >> 8));
	}
	if (data->batps)
		printk(KERN_INFO PFX "Only %d pstates on battery\n", data->batps);
}

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++) {
		powernow_table[j].index = pst[j].fid; /* lower 8 bits */
		powernow_table[j].index |= (pst[j].vid << 8); /* upper 8 bits */
		powernow_table[j].frequency = find_khz_freq_from_fid(pst[j].fid);
	}
	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;
	}

	dprintk("cfid 0x%x, cvid 0x%x\n", data->currfid, data->currvid);
	data->powernow_table = powernow_table;
	print_basics(data);

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

	dprintk("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;

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

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

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

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

		dprintk("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;

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

		dprintk("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 BFX "numpst must be 1\n");
			return -ENODEV;
		}

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

		data->numps = psb->numps;
		dprintk("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
	 */
D
Dave Jones 已提交
687
	printk(KERN_INFO PFX "BIOS error - no PSB or ACPI _PSS objects\n");
L
Linus Torvalds 已提交
688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710
	return -ENODEV;
}

#ifdef CONFIG_X86_POWERNOW_K8_ACPI
static void powernow_k8_acpi_pst_values(struct powernow_k8_data *data, unsigned int index)
{
	if (!data->acpi_data.state_count)
		return;

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

static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data)
{
	int i;
	int cntlofreq = 0;
	struct cpufreq_frequency_table *powernow_table;

	if (acpi_processor_register_performance(&data->acpi_data, data->cpu)) {
D
Dave Jones 已提交
711
		dprintk("register performance failed: bad ACPI data\n");
L
Linus Torvalds 已提交
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
		return -EIO;
	}

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

	if ((data->acpi_data.control_register.space_id != ACPI_ADR_SPACE_FIXED_HARDWARE) ||
		(data->acpi_data.status_register.space_id != ACPI_ADR_SPACE_FIXED_HARDWARE)) {
		dprintk("Invalid control/status registers (%x - %x)\n",
			data->acpi_data.control_register.space_id,
			data->acpi_data.status_register.space_id);
		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) {
		dprintk("powernow_table memory alloc failure\n");
		goto err_out;
	}

	for (i = 0; i < data->acpi_data.state_count; i++) {
		u32 fid = data->acpi_data.states[i].control & FID_MASK;
		u32 vid = (data->acpi_data.states[i].control >> VID_SHIFT) & VID_MASK;

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

		powernow_table[i].index = fid; /* lower 8 bits */
		powernow_table[i].index |= (vid << 8); /* upper 8 bits */
		powernow_table[i].frequency = find_khz_freq_from_fid(fid);

		/* verify frequency is OK */
		if ((powernow_table[i].frequency > (MAX_FREQ * 1000)) ||
			(powernow_table[i].frequency < (MIN_FREQ * 1000))) {
			dprintk("invalid freq %u kHz, ignoring\n", powernow_table[i].frequency);
			powernow_table[i].frequency = CPUFREQ_ENTRY_INVALID;
			continue;
		}

		/* verify voltage is OK - BIOSs are using "off" to indicate invalid */
		if (vid == 0x1f) {
			dprintk("invalid vid %u, ignoring\n", vid);
			powernow_table[i].frequency = CPUFREQ_ENTRY_INVALID;
			continue;
		}

D
Dave Jones 已提交
762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778
		/* verify only 1 entry from the lo frequency table */
		if (fid < HI_FID_TABLE_BOTTOM) {
			if (cntlofreq) {
				/* if both entries are the same, ignore this
				 * one... 
				 */
				if ((powernow_table[i].frequency != powernow_table[cntlofreq].frequency) ||
				    (powernow_table[i].index != powernow_table[cntlofreq].index)) {
					printk(KERN_ERR PFX "Too many lo freq table entries\n");
					goto err_out_mem;
				}

				dprintk("double low frequency table entry, ignoring it.\n");
				powernow_table[i].frequency = CPUFREQ_ENTRY_INVALID;
				continue;
			} else
				cntlofreq = i;
L
Linus Torvalds 已提交
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
		}

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

	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;

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

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

	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 -ENODEV;
}

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);
}

#else
static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data) { return -ENODEV; }
static void powernow_k8_cpu_exit_acpi(struct powernow_k8_data *data) { return; }
static void powernow_k8_acpi_pst_values(struct powernow_k8_data *data, unsigned int index) { return; }
#endif /* CONFIG_X86_POWERNOW_K8_ACPI */

/* Take a frequency, and issue the fid/vid transition command */
static int transition_frequency(struct powernow_k8_data *data, unsigned int index)
{
	u32 fid;
	u32 vid;
D
Dave Jones 已提交
833
	int res, i;
L
Linus Torvalds 已提交
834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857
	struct cpufreq_freqs freqs;

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

	/* 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;

	dprintk("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)) {
		dprintk("target matches current values (fid 0x%x, vid 0x%x)\n",
			fid, vid);
		return 0;
	}

	if ((fid < HI_FID_TABLE_BOTTOM) && (data->currfid < HI_FID_TABLE_BOTTOM)) {
D
Dave Jones 已提交
858 859
		printk(KERN_ERR PFX
		       "ignoring illegal change in lo freq table-%x to 0x%x\n",
L
Linus Torvalds 已提交
860 861 862 863 864 865 866 867 868 869
		       data->currfid, fid);
		return 1;
	}

	dprintk("cpu %d, changing to fid 0x%x, vid 0x%x\n",
		smp_processor_id(), fid, vid);

	freqs.cpu = data->cpu;
	freqs.old = find_khz_freq_from_fid(data->currfid);
	freqs.new = find_khz_freq_from_fid(fid);
D
Dave Jones 已提交
870 871 872 873
	for_each_cpu_mask(i, cpu_core_map[data->cpu]) {
		freqs.cpu = i;
		cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE);
	}
L
Linus Torvalds 已提交
874 875 876 877

	res = transition_fid_vid(data, fid, vid);

	freqs.new = find_khz_freq_from_fid(data->currfid);
D
Dave Jones 已提交
878 879 880 881
	for_each_cpu_mask(i, cpu_core_map[data->cpu]) {
		freqs.cpu = i;
		cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE);
        }
L
Linus Torvalds 已提交
882 883 884 885 886 887 888 889 890 891 892 893
	return res;
}

/* Driver entry point to switch to the target frequency */
static int powernowk8_target(struct cpufreq_policy *pol, unsigned targfreq, unsigned relation)
{
	cpumask_t oldmask = CPU_MASK_ALL;
	struct powernow_k8_data *data = powernow_data[pol->cpu];
	u32 checkfid = data->currfid;
	u32 checkvid = data->currvid;
	unsigned int newstate;
	int ret = -EIO;
D
Dave Jones 已提交
894
	int i;
L
Linus Torvalds 已提交
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

	/* only run on specific CPU from here on */
	oldmask = current->cpus_allowed;
	set_cpus_allowed(current, cpumask_of_cpu(pol->cpu));
	schedule();

	if (smp_processor_id() != pol->cpu) {
		printk(KERN_ERR "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;
	}

	dprintk("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)) {
		ret = -EIO;
		goto err_out;
	}

	dprintk("targ: curr fid 0x%x, vid 0x%x\n",
		data->currfid, data->currvid);

	if ((checkvid != data->currvid) || (checkfid != data->currfid)) {
D
Dave Jones 已提交
923 924 925
		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);
L
Linus Torvalds 已提交
926 927 928 929 930
	}

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

D
Dave Jones 已提交
931 932 933 934 935 936 937 938 939 940 941
	down(&fidvid_sem);

	for_each_cpu_mask(i, cpu_core_map[pol->cpu]) {
		/* make sure the sibling is initialized */
		if (!powernow_data[i]) {
                        ret = 0;
                        up(&fidvid_sem);
                        goto err_out;
                }
	}

L
Linus Torvalds 已提交
942 943 944 945 946
	powernow_k8_acpi_pst_values(data, newstate);

	if (transition_frequency(data, newstate)) {
		printk(KERN_ERR PFX "transition frequency failed\n");
		ret = 1;
D
Dave Jones 已提交
947
		up(&fidvid_sem);
L
Linus Torvalds 已提交
948 949 950
		goto err_out;
	}

D
Dave Jones 已提交
951 952 953 954 955 956 957
	/* Update all the fid/vids of our siblings */
	for_each_cpu_mask(i, cpu_core_map[pol->cpu]) {
		powernow_data[i]->currvid = data->currvid;
		powernow_data[i]->currfid = data->currfid;
	}	
	up(&fidvid_sem);

L
Linus Torvalds 已提交
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
	pol->cur = find_khz_freq_from_fid(data->currfid);
	ret = 0;

err_out:
	set_cpus_allowed(current, oldmask);
	schedule();

	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 = powernow_data[pol->cpu];

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

/* per CPU init entry point to the driver */
static int __init powernowk8_cpu_init(struct cpufreq_policy *pol)
{
	struct powernow_k8_data *data;
	cpumask_t oldmask = CPU_MASK_ALL;
	int rc;

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

	data = kmalloc(sizeof(struct powernow_k8_data), GFP_KERNEL);
	if (!data) {
		printk(KERN_ERR PFX "unable to alloc powernow_k8_data");
		return -ENOMEM;
	}
	memset(data,0,sizeof(struct powernow_k8_data));

	data->cpu = pol->cpu;

	if (powernow_k8_cpu_init_acpi(data)) {
		/*
		 * Use the PSB BIOS structure. This is only availabe on
		 * an UP version, and is deprecated by AMD.
		 */

		if ((num_online_cpus() != 1) || (num_possible_cpus() != 1)) {
D
Dave Jones 已提交
1002
			printk(KERN_ERR PFX "MP systems not supported by PSB BIOS structure\n");
L
Linus Torvalds 已提交
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
			kfree(data);
			return -ENODEV;
		}
		if (pol->cpu != 0) {
			printk(KERN_ERR PFX "init not cpu 0\n");
			kfree(data);
			return -ENODEV;
		}
		rc = find_psb_table(data);
		if (rc) {
			kfree(data);
			return -ENODEV;
		}
	}

	/* only run on specific CPU from here on */
	oldmask = current->cpus_allowed;
	set_cpus_allowed(current, cpumask_of_cpu(pol->cpu));
	schedule();

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

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

	if (query_current_values_with_pending_wait(data))
		goto err_out;

	fidvid_msr_init();

	/* run on any CPU again */
	set_cpus_allowed(current, oldmask);
	schedule();

	pol->governor = CPUFREQ_DEFAULT_GOVERNOR;
D
Dave Jones 已提交
1043
	pol->cpus = cpu_core_map[pol->cpu];
L
Linus Torvalds 已提交
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

	/* 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)
	    + (3 * (1 << data->irt) * 10)) * 1000;

	pol->cur = find_khz_freq_from_fid(data->currfid);
	dprintk("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 PFX "invalid powernow_table\n");
		powernow_k8_cpu_exit_acpi(data);
		kfree(data->powernow_table);
		kfree(data);
		return -EINVAL;
	}

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

	printk("cpu_init done, current fid 0x%x, vid 0x%x\n",
	       data->currfid, data->currvid);

	powernow_data[pol->cpu] = data;

	return 0;

err_out:
	set_cpus_allowed(current, oldmask);
	schedule();
	powernow_k8_cpu_exit_acpi(data);

	kfree(data);
	return -ENODEV;
}

static int __devexit powernowk8_cpu_exit (struct cpufreq_policy *pol)
{
	struct powernow_k8_data *data = 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);

	return 0;
}

static unsigned int powernowk8_get (unsigned int cpu)
{
	struct powernow_k8_data *data = powernow_data[cpu];
	cpumask_t oldmask = current->cpus_allowed;
	unsigned int khz = 0;

	set_cpus_allowed(current, cpumask_of_cpu(cpu));
	if (smp_processor_id() != cpu) {
		printk(KERN_ERR PFX "limiting to CPU %d failed in powernowk8_get\n", cpu);
		set_cpus_allowed(current, oldmask);
		return 0;
	}
	preempt_disable();
D
Dave Jones 已提交
1110
	
L
Linus Torvalds 已提交
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
	if (query_current_values_with_pending_wait(data))
		goto out;

	khz = find_khz_freq_from_fid(data->currfid);

 out:
	preempt_enable_no_resched();
	set_cpus_allowed(current, oldmask);

	return khz;
}

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

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

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

	for (i=0; i<NR_CPUS; i++) {
		if (!cpu_online(i))
			continue;
		if (check_supported_cpu(i))
			supported_cpus++;
	}

	if (supported_cpus == num_online_cpus()) {
		printk(KERN_INFO PFX "Found %d AMD Athlon 64 / Opteron processors (" VERSION ")\n",
			supported_cpus);
		return cpufreq_register_driver(&cpufreq_amd64_driver);
	}

	return -ENODEV;
}

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

	cpufreq_unregister_driver(&cpufreq_amd64_driver);
}

D
Dave Jones 已提交
1168
MODULE_AUTHOR("Paul Devriendt <paul.devriendt@amd.com> and Mark Langsdorf <mark.langsdorf@amd.com.");
L
Linus Torvalds 已提交
1169 1170 1171 1172 1173
MODULE_DESCRIPTION("AMD Athlon 64 and Opteron processor frequency driver.");
MODULE_LICENSE("GPL");

late_initcall(powernowk8_init);
module_exit(powernowk8_exit);
D
Dave Jones 已提交
1174