setup.c 26.1 KB
Newer Older
1 2 3 4 5 6 7 8 9 10
/*
 * Machine specific setup for xen
 *
 * Jeremy Fitzhardinge <jeremy@xensource.com>, XenSource Inc, 2007
 */

#include <linux/module.h>
#include <linux/sched.h>
#include <linux/mm.h>
#include <linux/pm.h>
11
#include <linux/memblock.h>
12
#include <linux/cpuidle.h>
13
#include <linux/cpufreq.h>
14 15

#include <asm/elf.h>
R
Roland McGrath 已提交
16
#include <asm/vdso.h>
17 18
#include <asm/e820.h>
#include <asm/setup.h>
19
#include <asm/acpi.h>
20
#include <asm/numa.h>
21 22 23
#include <asm/xen/hypervisor.h>
#include <asm/xen/hypercall.h>

24
#include <xen/xen.h>
25
#include <xen/page.h>
26
#include <xen/interface/callback.h>
27
#include <xen/interface/memory.h>
28 29
#include <xen/interface/physdev.h>
#include <xen/features.h>
30
#include <xen/hvc-console.h>
31
#include "xen-ops.h"
32
#include "vdso.h"
33
#include "p2m.h"
34
#include "mmu.h"
35

36
/* Amount of extra memory space we add to the e820 ranges */
37
struct xen_memory_region xen_extra_mem[XEN_EXTRA_MEM_MAX_REGIONS] __initdata;
38

39 40 41
/* Number of pages released from the initial allocation. */
unsigned long xen_released_pages;

42 43 44 45
/* E820 map used during setting up memory. */
static struct e820entry xen_e820_map[E820MAX] __initdata;
static u32 xen_e820_map_entries __initdata;

46 47 48 49 50 51 52 53 54 55 56 57 58
/*
 * Buffer used to remap identity mapped pages. We only need the virtual space.
 * The physical page behind this address is remapped as needed to different
 * buffer pages.
 */
#define REMAP_SIZE	(P2M_PER_PAGE - 3)
static struct {
	unsigned long	next_area_mfn;
	unsigned long	target_pfn;
	unsigned long	size;
	unsigned long	mfns[REMAP_SIZE];
} xen_remap_buf __initdata __aligned(PAGE_SIZE);
static unsigned long xen_remap_mfn __initdata = INVALID_P2M_ENTRY;
59

60 61 62 63 64 65 66 67 68 69 70 71
/* 
 * The maximum amount of extra memory compared to the base size.  The
 * main scaling factor is the size of struct page.  At extreme ratios
 * of base:extra, all the base memory can be filled with page
 * structures for the extra memory, leaving no space for anything
 * else.
 * 
 * 10x seems like a reasonable balance between scaling flexibility and
 * leaving a practically usable system.
 */
#define EXTRA_MEM_RATIO		(10)

72
static void __init xen_add_extra_mem(phys_addr_t start, phys_addr_t size)
73
{
74
	int i;
75

76 77 78 79 80 81 82 83 84 85 86 87 88 89 90
	for (i = 0; i < XEN_EXTRA_MEM_MAX_REGIONS; i++) {
		/* Add new region. */
		if (xen_extra_mem[i].size == 0) {
			xen_extra_mem[i].start = start;
			xen_extra_mem[i].size  = size;
			break;
		}
		/* Append to existing region. */
		if (xen_extra_mem[i].start + xen_extra_mem[i].size == start) {
			xen_extra_mem[i].size += size;
			break;
		}
	}
	if (i == XEN_EXTRA_MEM_MAX_REGIONS)
		printk(KERN_WARNING "Warning: not enough extra memory regions\n");
91

92
	memblock_reserve(start, size);
93
}
94

95
static void __init xen_del_extra_mem(phys_addr_t start, phys_addr_t size)
96 97
{
	int i;
98
	phys_addr_t start_r, size_r;
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
	for (i = 0; i < XEN_EXTRA_MEM_MAX_REGIONS; i++) {
		start_r = xen_extra_mem[i].start;
		size_r = xen_extra_mem[i].size;

		/* Start of region. */
		if (start_r == start) {
			BUG_ON(size > size_r);
			xen_extra_mem[i].start += size;
			xen_extra_mem[i].size -= size;
			break;
		}
		/* End of region. */
		if (start_r + size_r == start + size) {
			BUG_ON(size > size_r);
			xen_extra_mem[i].size -= size;
			break;
		}
		/* Mid of region. */
		if (start > start_r && start < start_r + size_r) {
			BUG_ON(start + size > start_r + size_r);
			xen_extra_mem[i].size = start - start_r;
			/* Calling memblock_reserve() again is okay. */
			xen_add_extra_mem(start + size, start_r + size_r -
					  (start + size));
			break;
		}
	}
	memblock_free(start, size);
}

/*
 * Called during boot before the p2m list can take entries beyond the
 * hypervisor supplied p2m list. Entries in extra mem are to be regarded as
 * invalid.
 */
unsigned long __ref xen_chk_extra_mem(unsigned long pfn)
{
	int i;
138
	phys_addr_t addr = PFN_PHYS(pfn);
139

140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157
	for (i = 0; i < XEN_EXTRA_MEM_MAX_REGIONS; i++) {
		if (addr >= xen_extra_mem[i].start &&
		    addr < xen_extra_mem[i].start + xen_extra_mem[i].size)
			return INVALID_P2M_ENTRY;
	}

	return IDENTITY_FRAME(pfn);
}

/*
 * Mark all pfns of extra mem as invalid in p2m list.
 */
void __init xen_inv_extra_mem(void)
{
	unsigned long pfn, pfn_s, pfn_e;
	int i;

	for (i = 0; i < XEN_EXTRA_MEM_MAX_REGIONS; i++) {
158 159
		if (!xen_extra_mem[i].size)
			continue;
160 161 162 163
		pfn_s = PFN_DOWN(xen_extra_mem[i].start);
		pfn_e = PFN_UP(xen_extra_mem[i].start + xen_extra_mem[i].size);
		for (pfn = pfn_s; pfn < pfn_e; pfn++)
			set_phys_to_machine(pfn, INVALID_P2M_ENTRY);
164
	}
165 166
}

167 168 169 170 171
/*
 * Finds the next RAM pfn available in the E820 map after min_pfn.
 * This function updates min_pfn with the pfn found and returns
 * the size of that range or zero if not found.
 */
172
static unsigned long __init xen_find_pfn_range(unsigned long *min_pfn)
173
{
174
	const struct e820entry *entry = xen_e820_map;
175 176 177
	unsigned int i;
	unsigned long done = 0;

178
	for (i = 0; i < xen_e820_map_entries; i++, entry++) {
179 180 181 182 183 184
		unsigned long s_pfn;
		unsigned long e_pfn;

		if (entry->type != E820_RAM)
			continue;

185
		e_pfn = PFN_DOWN(entry->addr + entry->size);
186

187 188
		/* We only care about E820 after this */
		if (e_pfn < *min_pfn)
189 190
			continue;

191
		s_pfn = PFN_UP(entry->addr);
192 193 194

		/* If min_pfn falls within the E820 entry, we want to start
		 * at the min_pfn PFN.
195
		 */
196 197
		if (s_pfn <= *min_pfn) {
			done = e_pfn - *min_pfn;
198
		} else {
199 200
			done = e_pfn - s_pfn;
			*min_pfn = s_pfn;
201
		}
202 203
		break;
	}
204

205 206
	return done;
}
207

208 209 210 211 212 213 214 215 216 217 218 219 220 221
static int __init xen_free_mfn(unsigned long mfn)
{
	struct xen_memory_reservation reservation = {
		.address_bits = 0,
		.extent_order = 0,
		.domid        = DOMID_SELF
	};

	set_xen_guest_handle(reservation.extent_start, &mfn);
	reservation.nr_extents = 1;

	return HYPERVISOR_memory_op(XENMEM_decrease_reservation, &reservation);
}

222
/*
223
 * This releases a chunk of memory and then does the identity map. It's used
224 225 226
 * as a fallback if the remapping fails.
 */
static void __init xen_set_identity_and_release_chunk(unsigned long start_pfn,
227
			unsigned long end_pfn, unsigned long nr_pages)
228
{
229 230 231
	unsigned long pfn, end;
	int ret;

232 233
	WARN_ON(start_pfn > end_pfn);

234
	/* Release pages first. */
235 236 237 238 239 240 241 242 243 244 245 246
	end = min(end_pfn, nr_pages);
	for (pfn = start_pfn; pfn < end; pfn++) {
		unsigned long mfn = pfn_to_mfn(pfn);

		/* Make sure pfn exists to start with */
		if (mfn == INVALID_P2M_ENTRY || mfn_to_pfn(mfn) != pfn)
			continue;

		ret = xen_free_mfn(mfn);
		WARN(ret != 1, "Failed to release pfn %lx err=%d\n", pfn, ret);

		if (ret == 1) {
247
			xen_released_pages++;
248 249 250 251 252 253
			if (!__set_phys_to_machine(pfn, INVALID_P2M_ENTRY))
				break;
		} else
			break;
	}

254
	set_phys_range_identity(start_pfn, end_pfn);
255 256 257
}

/*
258
 * Helper function to update the p2m and m2p tables and kernel mapping.
259
 */
260
static void __init xen_update_mem_tables(unsigned long pfn, unsigned long mfn)
261 262
{
	struct mmu_update update = {
263
		.ptr = ((uint64_t)mfn << PAGE_SHIFT) | MMU_MACHPHYS_UPDATE,
264 265 266 267
		.val = pfn
	};

	/* Update p2m */
268
	if (!set_phys_to_machine(pfn, mfn)) {
269 270
		WARN(1, "Failed to set p2m mapping for pfn=%ld mfn=%ld\n",
		     pfn, mfn);
271
		BUG();
272
	}
273 274 275 276 277

	/* Update m2p */
	if (HYPERVISOR_mmu_update(&update, 1, NULL, DOMID_SELF) < 0) {
		WARN(1, "Failed to set m2p mapping for mfn=%ld pfn=%ld\n",
		     mfn, pfn);
278
		BUG();
279 280
	}

281
	/* Update kernel mapping, but not for highmem. */
282
	if (pfn >= PFN_UP(__pa(high_memory - 1)))
283 284 285 286 287 288 289 290
		return;

	if (HYPERVISOR_update_va_mapping((unsigned long)__va(pfn << PAGE_SHIFT),
					 mfn_pte(mfn, PAGE_KERNEL), 0)) {
		WARN(1, "Failed to update kernel mapping for mfn=%ld pfn=%ld\n",
		      mfn, pfn);
		BUG();
	}
291
}
292

293 294
/*
 * This function updates the p2m and m2p tables with an identity map from
295 296 297 298 299 300 301 302 303
 * start_pfn to start_pfn+size and prepares remapping the underlying RAM of the
 * original allocation at remap_pfn. The information needed for remapping is
 * saved in the memory itself to avoid the need for allocating buffers. The
 * complete remap information is contained in a list of MFNs each containing
 * up to REMAP_SIZE MFNs and the start target PFN for doing the remap.
 * This enables us to preserve the original mfn sequence while doing the
 * remapping at a time when the memory management is capable of allocating
 * virtual and physical memory in arbitrary amounts, see 'xen_remap_memory' and
 * its callers.
304
 */
305
static void __init xen_do_set_identity_and_remap_chunk(
306
        unsigned long start_pfn, unsigned long size, unsigned long remap_pfn)
307
{
308 309
	unsigned long buf = (unsigned long)&xen_remap_buf;
	unsigned long mfn_save, mfn;
310
	unsigned long ident_pfn_iter, remap_pfn_iter;
311
	unsigned long ident_end_pfn = start_pfn + size;
312
	unsigned long left = size;
313
	unsigned int i, chunk;
314 315 316 317

	WARN_ON(size == 0);

	BUG_ON(xen_feature(XENFEAT_auto_translated_physmap));
318

319
	mfn_save = virt_to_mfn(buf);
320

321 322 323 324
	for (ident_pfn_iter = start_pfn, remap_pfn_iter = remap_pfn;
	     ident_pfn_iter < ident_end_pfn;
	     ident_pfn_iter += REMAP_SIZE, remap_pfn_iter += REMAP_SIZE) {
		chunk = (left < REMAP_SIZE) ? left : REMAP_SIZE;
325

326 327 328
		/* Map first pfn to xen_remap_buf */
		mfn = pfn_to_mfn(ident_pfn_iter);
		set_pte_mfn(buf, mfn, PAGE_KERNEL);
329

330 331 332 333 334 335
		/* Save mapping information in page */
		xen_remap_buf.next_area_mfn = xen_remap_mfn;
		xen_remap_buf.target_pfn = remap_pfn_iter;
		xen_remap_buf.size = chunk;
		for (i = 0; i < chunk; i++)
			xen_remap_buf.mfns[i] = pfn_to_mfn(ident_pfn_iter + i);
336

337 338
		/* Put remap buf into list. */
		xen_remap_mfn = mfn;
339

340
		/* Set identity map */
341
		set_phys_range_identity(ident_pfn_iter, ident_pfn_iter + chunk);
342

343
		left -= chunk;
344
	}
345

346 347
	/* Restore old xen_remap_buf mapping */
	set_pte_mfn(buf, mfn_save, PAGE_KERNEL);
348 349
}

350 351 352 353 354 355 356 357 358 359 360
/*
 * This function takes a contiguous pfn range that needs to be identity mapped
 * and:
 *
 *  1) Finds a new range of pfns to use to remap based on E820 and remap_pfn.
 *  2) Calls the do_ function to actually do the mapping/remapping work.
 *
 * The goal is to not allocate additional memory but to remap the existing
 * pages. In the case of an error the underlying memory is simply released back
 * to Xen and not remapped.
 */
361
static unsigned long __init xen_set_identity_and_remap_chunk(
362
	unsigned long start_pfn, unsigned long end_pfn, unsigned long nr_pages,
363
	unsigned long remap_pfn)
364 365 366 367 368 369 370 371 372 373 374 375 376 377
{
	unsigned long pfn;
	unsigned long i = 0;
	unsigned long n = end_pfn - start_pfn;

	while (i < n) {
		unsigned long cur_pfn = start_pfn + i;
		unsigned long left = n - i;
		unsigned long size = left;
		unsigned long remap_range_size;

		/* Do not remap pages beyond the current allocation */
		if (cur_pfn >= nr_pages) {
			/* Identity map remaining pages */
378
			set_phys_range_identity(cur_pfn, cur_pfn + size);
379 380 381 382 383
			break;
		}
		if (cur_pfn + size > nr_pages)
			size = nr_pages - cur_pfn;

384
		remap_range_size = xen_find_pfn_range(&remap_pfn);
385 386 387
		if (!remap_range_size) {
			pr_warning("Unable to find available pfn range, not remapping identity pages\n");
			xen_set_identity_and_release_chunk(cur_pfn,
388
						cur_pfn + left, nr_pages);
389 390 391 392 393 394
			break;
		}
		/* Adjust size to fit in current e820 RAM region */
		if (size > remap_range_size)
			size = remap_range_size;

395
		xen_do_set_identity_and_remap_chunk(cur_pfn, size, remap_pfn);
396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413

		/* Update variables to reflect new mappings. */
		i += size;
		remap_pfn += size;
	}

	/*
	 * If the PFNs are currently mapped, the VA mapping also needs
	 * to be updated to be 1:1.
	 */
	for (pfn = start_pfn; pfn <= max_pfn_mapped && pfn < end_pfn; pfn++)
		(void)HYPERVISOR_update_va_mapping(
			(unsigned long)__va(pfn << PAGE_SHIFT),
			mfn_pte(pfn, PAGE_KERNEL_IO), 0);

	return remap_pfn;
}

414
static void __init xen_set_identity_and_remap(unsigned long nr_pages)
415
{
416
	phys_addr_t start = 0;
417
	unsigned long last_pfn = nr_pages;
418
	const struct e820entry *entry = xen_e820_map;
419 420
	int i;

421 422 423
	/*
	 * Combine non-RAM regions and gaps until a RAM region (or the
	 * end of the map) is reached, then set the 1:1 map and
424
	 * remap the memory in those non-RAM regions.
425 426 427 428 429 430 431
	 *
	 * The combined non-RAM regions are rounded to a whole number
	 * of pages so any partial pages are accessible via the 1:1
	 * mapping.  This is needed for some BIOSes that put (for
	 * example) the DMI tables in a reserved region that begins on
	 * a non-page boundary.
	 */
432
	for (i = 0; i < xen_e820_map_entries; i++, entry++) {
433
		phys_addr_t end = entry->addr + entry->size;
434
		if (entry->type == E820_RAM || i == xen_e820_map_entries - 1) {
435 436
			unsigned long start_pfn = PFN_DOWN(start);
			unsigned long end_pfn = PFN_UP(end);
437

438 439
			if (entry->type == E820_RAM)
				end_pfn = PFN_UP(entry->addr);
440

441
			if (start_pfn < end_pfn)
442
				last_pfn = xen_set_identity_and_remap_chunk(
443
						start_pfn, end_pfn, nr_pages,
444
						last_pfn);
445
			start = end;
446 447
		}
	}
448

449
	pr_info("Released %ld page(s)\n", xen_released_pages);
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

/*
 * Remap the memory prepared in xen_do_set_identity_and_remap_chunk().
 * The remap information (which mfn remap to which pfn) is contained in the
 * to be remapped memory itself in a linked list anchored at xen_remap_mfn.
 * This scheme allows to remap the different chunks in arbitrary order while
 * the resulting mapping will be independant from the order.
 */
void __init xen_remap_memory(void)
{
	unsigned long buf = (unsigned long)&xen_remap_buf;
	unsigned long mfn_save, mfn, pfn;
	unsigned long remapped = 0;
	unsigned int i;
	unsigned long pfn_s = ~0UL;
	unsigned long len = 0;

	mfn_save = virt_to_mfn(buf);

	while (xen_remap_mfn != INVALID_P2M_ENTRY) {
		/* Map the remap information */
		set_pte_mfn(buf, xen_remap_mfn, PAGE_KERNEL);

		BUG_ON(xen_remap_mfn != xen_remap_buf.mfns[0]);

		pfn = xen_remap_buf.target_pfn;
		for (i = 0; i < xen_remap_buf.size; i++) {
			mfn = xen_remap_buf.mfns[i];
			xen_update_mem_tables(pfn, mfn);
			remapped++;
			pfn++;
		}
		if (pfn_s == ~0UL || pfn == pfn_s) {
			pfn_s = xen_remap_buf.target_pfn;
			len += xen_remap_buf.size;
		} else if (pfn_s + len == xen_remap_buf.target_pfn) {
			len += xen_remap_buf.size;
		} else {
489
			xen_del_extra_mem(PFN_PHYS(pfn_s), PFN_PHYS(len));
490 491 492 493 494 495 496 497 498
			pfn_s = xen_remap_buf.target_pfn;
			len = xen_remap_buf.size;
		}

		mfn = xen_remap_mfn;
		xen_remap_mfn = xen_remap_buf.next_area_mfn;
	}

	if (pfn_s != ~0UL && len)
499
		xen_del_extra_mem(PFN_PHYS(pfn_s), PFN_PHYS(len));
500 501 502 503 504 505

	set_pte_mfn(buf, mfn_save, PAGE_KERNEL);

	pr_info("Remapped %ld page(s)\n", remapped);
}

506 507 508 509 510 511
static unsigned long __init xen_get_max_pages(void)
{
	unsigned long max_pages = MAX_DOMAIN_PAGES;
	domid_t domid = DOMID_SELF;
	int ret;

512 513 514 515 516 517 518 519 520 521 522 523 524 525 526
	/*
	 * For the initial domain we use the maximum reservation as
	 * the maximum page.
	 *
	 * For guest domains the current maximum reservation reflects
	 * the current maximum rather than the static maximum. In this
	 * case the e820 map provided to us will cover the static
	 * maximum region.
	 */
	if (xen_initial_domain()) {
		ret = HYPERVISOR_memory_op(XENMEM_maximum_reservation, &domid);
		if (ret > 0)
			max_pages = ret;
	}

527 528 529
	return min(max_pages, MAX_DOMAIN_PAGES);
}

530 531
static void __init xen_align_and_add_e820_region(phys_addr_t start,
						 phys_addr_t size, int type)
532
{
533
	phys_addr_t end = start + size;
534 535 536 537

	/* Align RAM regions to page boundaries. */
	if (type == E820_RAM) {
		start = PAGE_ALIGN(start);
538
		end &= ~((phys_addr_t)PAGE_SIZE - 1);
539 540 541 542 543
	}

	e820_add_region(start, end - start, type);
}

544
static void __init xen_ignore_unusable(void)
545
{
546
	struct e820entry *entry = xen_e820_map;
547 548
	unsigned int i;

549
	for (i = 0; i < xen_e820_map_entries; i++, entry++) {
550 551 552 553 554
		if (entry->type == E820_UNUSABLE)
			entry->type = E820_RAM;
	}
}

555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576
static unsigned long __init xen_count_remap_pages(unsigned long max_pfn)
{
	unsigned long extra = 0;
	const struct e820entry *entry = xen_e820_map;
	int i;

	for (i = 0; i < xen_e820_map_entries; i++, entry++) {
		unsigned long start_pfn = PFN_DOWN(entry->addr);
		unsigned long end_pfn = PFN_UP(entry->addr + entry->size);

		if (start_pfn >= max_pfn)
			break;
		if (entry->type == E820_RAM)
			continue;
		if (end_pfn >= max_pfn)
			end_pfn = max_pfn;
		extra += end_pfn - start_pfn;
	}

	return extra;
}

577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599
bool __init xen_is_e820_reserved(phys_addr_t start, phys_addr_t size)
{
	struct e820entry *entry;
	unsigned mapcnt;
	phys_addr_t end;

	if (!size)
		return false;

	end = start + size;
	entry = xen_e820_map;

	for (mapcnt = 0; mapcnt < xen_e820_map_entries; mapcnt++) {
		if (entry->type == E820_RAM && entry->addr <= start &&
		    (entry->addr + entry->size) >= end)
			return false;

		entry++;
	}

	return true;
}

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
/*
 * Find a free area in physical memory not yet reserved and compliant with
 * E820 map.
 * Used to relocate pre-allocated areas like initrd or p2m list which are in
 * conflict with the to be used E820 map.
 * In case no area is found, return 0. Otherwise return the physical address
 * of the area which is already reserved for convenience.
 */
phys_addr_t __init xen_find_free_area(phys_addr_t size)
{
	unsigned mapcnt;
	phys_addr_t addr, start;
	struct e820entry *entry = xen_e820_map;

	for (mapcnt = 0; mapcnt < xen_e820_map_entries; mapcnt++, entry++) {
		if (entry->type != E820_RAM || entry->size < size)
			continue;
		start = entry->addr;
		for (addr = start; addr < start + size; addr += PAGE_SIZE) {
			if (!memblock_is_reserved(addr))
				continue;
			start = addr + PAGE_SIZE;
			if (start + size > entry->addr + entry->size)
				break;
		}
		if (addr >= start + size) {
			memblock_reserve(start, size);
			return start;
		}
	}

	return 0;
}

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
/*
 * Reserve Xen mfn_list.
 * See comment above "struct start_info" in <xen/interface/xen.h>
 * We tried to make the the memblock_reserve more selective so
 * that it would be clear what region is reserved. Sadly we ran
 * in the problem wherein on a 64-bit hypervisor with a 32-bit
 * initial domain, the pt_base has the cr3 value which is not
 * neccessarily where the pagetable starts! As Jan put it: "
 * Actually, the adjustment turns out to be correct: The page
 * tables for a 32-on-64 dom0 get allocated in the order "first L1",
 * "first L2", "first L3", so the offset to the page table base is
 * indeed 2. When reading xen/include/public/xen.h's comment
 * very strictly, this is not a violation (since there nothing is said
 * that the first thing in the page table space is pointed to by
 * pt_base; I admit that this seems to be implied though, namely
 * do I think that it is implied that the page table space is the
 * range [pt_base, pt_base + nt_pt_frames), whereas that
 * range here indeed is [pt_base - 2, pt_base - 2 + nt_pt_frames),
 * which - without a priori knowledge - the kernel would have
 * difficulty to figure out)." - so lets just fall back to the
 * easy way and reserve the whole region.
 */
static void __init xen_reserve_xen_mfnlist(void)
{
	if (xen_start_info->mfn_list >= __START_KERNEL_map) {
		memblock_reserve(__pa(xen_start_info->mfn_list),
				 xen_start_info->pt_base -
				 xen_start_info->mfn_list);
		return;
	}

	memblock_reserve(PFN_PHYS(xen_start_info->first_p2m_pfn),
			 PFN_PHYS(xen_start_info->nr_p2m_frames));
}

669 670 671 672 673 674
/**
 * machine_specific_memory_setup - Hook for machine specific memory setup.
 **/
char * __init xen_memory_setup(void)
{
	unsigned long max_pfn = xen_start_info->nr_pages;
675 676
	phys_addr_t mem_end, addr, size, chunk_size;
	u32 type;
677 678
	int rc;
	struct xen_memory_map memmap;
679
	unsigned long max_pages;
680
	unsigned long extra_pages = 0;
681
	int i;
682
	int op;
683

684
	max_pfn = min(MAX_DOMAIN_PAGES, max_pfn);
685 686 687
	mem_end = PFN_PHYS(max_pfn);

	memmap.nr_entries = E820MAX;
688
	set_xen_guest_handle(memmap.buffer, xen_e820_map);
689

690 691 692 693
	op = xen_initial_domain() ?
		XENMEM_machine_memory_map :
		XENMEM_memory_map;
	rc = HYPERVISOR_memory_op(op, &memmap);
694
	if (rc == -ENOSYS) {
695
		BUG_ON(xen_initial_domain());
696
		memmap.nr_entries = 1;
697 698
		xen_e820_map[0].addr = 0ULL;
		xen_e820_map[0].size = mem_end;
699
		/* 8MB slack (to balance backend allocations). */
700 701
		xen_e820_map[0].size += 8ULL << 20;
		xen_e820_map[0].type = E820_RAM;
702 703 704
		rc = 0;
	}
	BUG_ON(rc);
705
	BUG_ON(memmap.nr_entries == 0);
706
	xen_e820_map_entries = memmap.nr_entries;
707

708 709 710 711 712 713 714 715 716
	/*
	 * Xen won't allow a 1:1 mapping to be created to UNUSABLE
	 * regions, so if we're using the machine memory map leave the
	 * region as RAM as it is in the pseudo-physical map.
	 *
	 * UNUSABLE regions in domUs are not handled and will need
	 * a patch in the future.
	 */
	if (xen_initial_domain())
717
		xen_ignore_unusable();
718

719
	/* Make sure the Xen-supplied memory map is well-ordered. */
720 721
	sanitize_e820_map(xen_e820_map, xen_e820_map_entries,
			  &xen_e820_map_entries);
722 723 724 725 726

	max_pages = xen_get_max_pages();
	if (max_pages > max_pfn)
		extra_pages += max_pages - max_pfn;

727 728
	/* How many extra pages do we need due to remapping? */
	extra_pages += xen_count_remap_pages(max_pfn);
729

730 731 732 733 734 735 736 737 738 739 740 741 742 743
	/*
	 * Clamp the amount of extra memory to a EXTRA_MEM_RATIO
	 * factor the base size.  On non-highmem systems, the base
	 * size is the full initial memory allocation; on highmem it
	 * is limited to the max size of lowmem, so that it doesn't
	 * get completely filled.
	 *
	 * In principle there could be a problem in lowmem systems if
	 * the initial memory is also very large with respect to
	 * lowmem, but we won't try to deal with that here.
	 */
	extra_pages = min(EXTRA_MEM_RATIO * min(max_pfn, PFN_DOWN(MAXMEM)),
			  extra_pages);
	i = 0;
744 745
	addr = xen_e820_map[0].addr;
	size = xen_e820_map[0].size;
746
	while (i < xen_e820_map_entries) {
747 748
		chunk_size = size;
		type = xen_e820_map[i].type;
749 750 751

		if (type == E820_RAM) {
			if (addr < mem_end) {
752
				chunk_size = min(size, mem_end - addr);
753
			} else if (extra_pages) {
754 755 756 757
				chunk_size = min(size, PFN_PHYS(extra_pages));
				extra_pages -= PFN_DOWN(chunk_size);
				xen_add_extra_mem(addr, chunk_size);
				xen_max_p2m_pfn = PFN_DOWN(addr + chunk_size);
758 759
			} else
				type = E820_UNUSABLE;
760 761
		}

762
		xen_align_and_add_e820_region(addr, chunk_size, type);
763

764 765 766
		addr += chunk_size;
		size -= chunk_size;
		if (size == 0) {
767
			i++;
768 769 770 771 772
			if (i < xen_e820_map_entries) {
				addr = xen_e820_map[i].addr;
				size = xen_e820_map[i].size;
			}
		}
773
	}
774

775 776 777 778 779 780 781
	/*
	 * Set the rest as identity mapped, in case PCI BARs are
	 * located here.
	 *
	 * PFNs above MAX_P2M_PFN are considered identity mapped as
	 * well.
	 */
782
	set_phys_range_identity(addr / PAGE_SIZE, ~0ul);
783

784
	/*
785 786
	 * In domU, the ISA region is normal, usable memory, but we
	 * reserve ISA memory anyway because too many things poke
787 788 789 790
	 * about in there.
	 */
	e820_add_region(ISA_START_ADDRESS, ISA_END_ADDRESS - ISA_START_ADDRESS,
			E820_RESERVED);
791

792 793
	sanitize_e820_map(e820.map, ARRAY_SIZE(e820.map), &e820.nr_map);

794 795 796 797 798 799 800 801 802 803 804
	/*
	 * Check whether the kernel itself conflicts with the target E820 map.
	 * Failing now is better than running into weird problems later due
	 * to relocating (and even reusing) pages with kernel text or data.
	 */
	if (xen_is_e820_reserved(__pa_symbol(_text),
			__pa_symbol(__bss_stop) - __pa_symbol(_text))) {
		xen_raw_console_write("Xen hypervisor allocated kernel memory conflicts with E820 map\n");
		BUG();
	}

805 806 807 808 809 810
	/*
	 * Check for a conflict of the hypervisor supplied page tables with
	 * the target E820 map.
	 */
	xen_pt_check_e820();

811 812
	xen_reserve_xen_mfnlist();

813 814 815 816 817 818
	/*
	 * Set identity map on non-RAM pages and prepare remapping the
	 * underlying RAM.
	 */
	xen_set_identity_and_remap(max_pfn);

819 820 821
	return "Xen";
}

822 823 824 825 826 827 828 829 830 831
/*
 * Machine specific memory setup for auto-translated guests.
 */
char * __init xen_auto_xlated_memory_setup(void)
{
	struct xen_memory_map memmap;
	int i;
	int rc;

	memmap.nr_entries = E820MAX;
832
	set_xen_guest_handle(memmap.buffer, xen_e820_map);
833 834 835 836 837

	rc = HYPERVISOR_memory_op(XENMEM_memory_map, &memmap);
	if (rc < 0)
		panic("No memory map (%d)\n", rc);

838 839 840 841
	xen_e820_map_entries = memmap.nr_entries;

	sanitize_e820_map(xen_e820_map, ARRAY_SIZE(xen_e820_map),
			  &xen_e820_map_entries);
842

843 844 845
	for (i = 0; i < xen_e820_map_entries; i++)
		e820_add_region(xen_e820_map[i].addr, xen_e820_map[i].size,
				xen_e820_map[i].type);
846

847
	xen_reserve_xen_mfnlist();
848 849 850 851

	return "Xen";
}

852 853
/*
 * Set the bit indicating "nosegneg" library variants should be used.
854 855
 * We only need to bother in pure 32-bit mode; compat 32-bit processes
 * can have un-truncated segments, so wrapping around is allowed.
856
 */
857
static void __init fiddle_vdso(void)
858
{
859
#ifdef CONFIG_X86_32
860 861 862 863 864
	/*
	 * This could be called before selected_vdso32 is initialized, so
	 * just fiddle with both possible images.  vdso_image_32_syscall
	 * can't be selected, since it only exists on 64-bit systems.
	 */
865
	u32 *mask;
866 867
	mask = vdso_image_32_int80.data +
		vdso_image_32_int80.sym_VDSO32_NOTE_MASK;
868
	*mask |= 1 << VDSO_NOTE_NONEGSEG_BIT;
869 870
	mask = vdso_image_32_sysenter.data +
		vdso_image_32_sysenter.sym_VDSO32_NOTE_MASK;
871
	*mask |= 1 << VDSO_NOTE_NONEGSEG_BIT;
872
#endif
873 874
}

875
static int register_callback(unsigned type, const void *func)
876
{
877 878 879
	struct callback_register callback = {
		.type = type,
		.address = XEN_CALLBACK(__KERNEL_CS, func),
880 881 882
		.flags = CALLBACKF_mask_events,
	};

883 884 885
	return HYPERVISOR_callback_op(CALLBACKOP_register, &callback);
}

886
void xen_enable_sysenter(void)
887
{
888
	int ret;
889
	unsigned sysenter_feature;
890 891

#ifdef CONFIG_X86_32
892
	sysenter_feature = X86_FEATURE_SEP;
893
#else
894
	sysenter_feature = X86_FEATURE_SYSENTER32;
895
#endif
896

897 898 899
	if (!boot_cpu_has(sysenter_feature))
		return;

900
	ret = register_callback(CALLBACKTYPE_sysenter, xen_sysenter_target);
901 902
	if(ret != 0)
		setup_clear_cpu_cap(sysenter_feature);
903 904
}

905
void xen_enable_syscall(void)
906 907 908 909 910 911
{
#ifdef CONFIG_X86_64
	int ret;

	ret = register_callback(CALLBACKTYPE_syscall, xen_syscall_target);
	if (ret != 0) {
912
		printk(KERN_ERR "Failed to set syscall callback: %d\n", ret);
913 914 915 916 917
		/* Pretty fatal; 64-bit userspace has no other
		   mechanism for syscalls. */
	}

	if (boot_cpu_has(X86_FEATURE_SYSCALL32)) {
918 919
		ret = register_callback(CALLBACKTYPE_syscall32,
					xen_syscall32_target);
920
		if (ret != 0)
921
			setup_clear_cpu_cap(X86_FEATURE_SYSCALL32);
922 923 924
	}
#endif /* CONFIG_X86_64 */
}
925

926
void __init xen_pvmmu_arch_setup(void)
927 928 929 930
{
	HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_4gb_segments);
	HYPERVISOR_vm_assist(VMASST_CMD_enable, VMASST_TYPE_writable_pagetables);

931 932
	HYPERVISOR_vm_assist(VMASST_CMD_enable,
			     VMASST_TYPE_pae_extended_cr3);
933

934 935 936
	if (register_callback(CALLBACKTYPE_event, xen_hypervisor_callback) ||
	    register_callback(CALLBACKTYPE_failsafe, xen_failsafe_callback))
		BUG();
937

938
	xen_enable_sysenter();
939
	xen_enable_syscall();
940 941 942 943 944 945 946 947 948
}

/* This function is not called for HVM domains */
void __init xen_arch_setup(void)
{
	xen_panic_handler_init();
	if (!xen_feature(XENFEAT_auto_translated_physmap))
		xen_pvmmu_arch_setup();

949 950 951 952 953 954 955 956 957 958 959
#ifdef CONFIG_ACPI
	if (!(xen_start_info->flags & SIF_INITDOMAIN)) {
		printk(KERN_INFO "ACPI in unprivileged domain disabled\n");
		disable_acpi();
	}
#endif

	memcpy(boot_command_line, xen_start_info->cmd_line,
	       MAX_GUEST_CMDLINE > COMMAND_LINE_SIZE ?
	       COMMAND_LINE_SIZE : MAX_GUEST_CMDLINE);

960
	/* Set up idle, making sure it calls safe_halt() pvop */
961
	disable_cpuidle();
962
	disable_cpufreq();
963
	WARN_ON(xen_set_default_idle());
964
	fiddle_vdso();
965 966 967
#ifdef CONFIG_NUMA
	numa_off = 1;
#endif
968
}
新手
引导
客服 返回
顶部