gmap.c 64.0 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
/*
 *  KVM guest address space mapping code
 *
 *    Copyright IBM Corp. 2007, 2016
 *    Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com>
 */

#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/swap.h>
#include <linux/smp.h>
#include <linux/spinlock.h>
#include <linux/slab.h>
#include <linux/swapops.h>
#include <linux/ksm.h>
#include <linux/mman.h>

#include <asm/pgtable.h>
#include <asm/pgalloc.h>
#include <asm/gmap.h>
#include <asm/tlb.h>

24 25
#define GMAP_SHADOW_FAKE_TABLE 1ULL

26
/**
27
 * gmap_alloc - allocate and initialize a guest address space
28
 * @mm: pointer to the parent mm_struct
29
 * @limit: maximum address of the gmap address space
30 31 32
 *
 * Returns a guest address space structure.
 */
33
static struct gmap *gmap_alloc(unsigned long limit)
34 35 36 37 38 39
{
	struct gmap *gmap;
	struct page *page;
	unsigned long *table;
	unsigned long etype, atype;

40 41
	if (limit < _REGION3_SIZE) {
		limit = _REGION3_SIZE - 1;
42 43
		atype = _ASCE_TYPE_SEGMENT;
		etype = _SEGMENT_ENTRY_EMPTY;
44 45
	} else if (limit < _REGION2_SIZE) {
		limit = _REGION2_SIZE - 1;
46 47
		atype = _ASCE_TYPE_REGION3;
		etype = _REGION3_ENTRY_EMPTY;
48 49
	} else if (limit < _REGION1_SIZE) {
		limit = _REGION1_SIZE - 1;
50 51 52 53 54 55 56 57 58 59 60
		atype = _ASCE_TYPE_REGION2;
		etype = _REGION2_ENTRY_EMPTY;
	} else {
		limit = -1UL;
		atype = _ASCE_TYPE_REGION1;
		etype = _REGION1_ENTRY_EMPTY;
	}
	gmap = kzalloc(sizeof(struct gmap), GFP_KERNEL);
	if (!gmap)
		goto out;
	INIT_LIST_HEAD(&gmap->crst_list);
61 62
	INIT_LIST_HEAD(&gmap->children);
	INIT_LIST_HEAD(&gmap->pt_list);
63 64
	INIT_RADIX_TREE(&gmap->guest_to_host, GFP_KERNEL);
	INIT_RADIX_TREE(&gmap->host_to_guest, GFP_ATOMIC);
65
	INIT_RADIX_TREE(&gmap->host_to_rmap, GFP_ATOMIC);
66
	spin_lock_init(&gmap->guest_table_lock);
67
	spin_lock_init(&gmap->shadow_lock);
68
	atomic_set(&gmap->ref_count, 1);
69
	page = alloc_pages(GFP_KERNEL, CRST_ALLOC_ORDER);
70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86
	if (!page)
		goto out_free;
	page->index = 0;
	list_add(&page->lru, &gmap->crst_list);
	table = (unsigned long *) page_to_phys(page);
	crst_table_init(table, etype);
	gmap->table = table;
	gmap->asce = atype | _ASCE_TABLE_LENGTH |
		_ASCE_USER_BITS | __pa(table);
	gmap->asce_end = limit;
	return gmap;

out_free:
	kfree(gmap);
out:
	return NULL;
}
87 88 89 90 91 92 93 94 95 96 97

/**
 * gmap_create - create a guest address space
 * @mm: pointer to the parent mm_struct
 * @limit: maximum size of the gmap address space
 *
 * Returns a guest address space structure.
 */
struct gmap *gmap_create(struct mm_struct *mm, unsigned long limit)
{
	struct gmap *gmap;
98
	unsigned long gmap_asce;
99 100 101 102 103

	gmap = gmap_alloc(limit);
	if (!gmap)
		return NULL;
	gmap->mm = mm;
104
	spin_lock(&mm->context.lock);
105
	list_add_rcu(&gmap->list, &mm->context.gmap_list);
106 107 108 109 110
	if (list_is_singular(&mm->context.gmap_list))
		gmap_asce = gmap->asce;
	else
		gmap_asce = -1UL;
	WRITE_ONCE(mm->context.gmap_asce, gmap_asce);
111
	spin_unlock(&mm->context.lock);
112 113 114
	return gmap;
}
EXPORT_SYMBOL_GPL(gmap_create);
115 116 117 118

static void gmap_flush_tlb(struct gmap *gmap)
{
	if (MACHINE_HAS_IDTE)
119
		__tlb_flush_idte(gmap->asce);
120 121 122 123 124 125 126 127 128
	else
		__tlb_flush_global();
}

static void gmap_radix_tree_free(struct radix_tree_root *root)
{
	struct radix_tree_iter iter;
	unsigned long indices[16];
	unsigned long index;
H
Heiko Carstens 已提交
129
	void __rcu **slot;
130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147
	int i, nr;

	/* A radix tree is freed by deleting all of its entries */
	index = 0;
	do {
		nr = 0;
		radix_tree_for_each_slot(slot, root, &iter, index) {
			indices[nr] = iter.index;
			if (++nr == 16)
				break;
		}
		for (i = 0; i < nr; i++) {
			index = indices[i];
			radix_tree_delete(root, index);
		}
	} while (nr > 0);
}

148 149 150 151 152 153
static void gmap_rmap_radix_tree_free(struct radix_tree_root *root)
{
	struct gmap_rmap *rmap, *rnext, *head;
	struct radix_tree_iter iter;
	unsigned long indices[16];
	unsigned long index;
H
Heiko Carstens 已提交
154
	void __rcu **slot;
155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
	int i, nr;

	/* A radix tree is freed by deleting all of its entries */
	index = 0;
	do {
		nr = 0;
		radix_tree_for_each_slot(slot, root, &iter, index) {
			indices[nr] = iter.index;
			if (++nr == 16)
				break;
		}
		for (i = 0; i < nr; i++) {
			index = indices[i];
			head = radix_tree_delete(root, index);
			gmap_for_each_rmap_safe(rmap, rnext, head)
				kfree(rmap);
		}
	} while (nr > 0);
}

175 176 177
/**
 * gmap_free - free a guest address space
 * @gmap: pointer to the guest address space structure
178 179
 *
 * No locks required. There are no references to this gmap anymore.
180
 */
181
static void gmap_free(struct gmap *gmap)
182 183 184
{
	struct page *page, *next;

185 186 187
	/* Flush tlb of all gmaps (if not already done for shadows) */
	if (!(gmap_is_shadow(gmap) && gmap->removed))
		gmap_flush_tlb(gmap);
188 189
	/* Free all segment & region tables. */
	list_for_each_entry_safe(page, next, &gmap->crst_list, lru)
190
		__free_pages(page, CRST_ALLOC_ORDER);
191 192
	gmap_radix_tree_free(&gmap->guest_to_host);
	gmap_radix_tree_free(&gmap->host_to_guest);
193 194 195 196 197 198 199 200 201 202 203

	/* Free additional data for a shadow gmap */
	if (gmap_is_shadow(gmap)) {
		/* Free all page tables. */
		list_for_each_entry_safe(page, next, &gmap->pt_list, lru)
			page_table_free_pgste(page);
		gmap_rmap_radix_tree_free(&gmap->host_to_rmap);
		/* Release reference to the parent */
		gmap_put(gmap->parent);
	}

204 205
	kfree(gmap);
}
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

/**
 * gmap_get - increase reference counter for guest address space
 * @gmap: pointer to the guest address space structure
 *
 * Returns the gmap pointer
 */
struct gmap *gmap_get(struct gmap *gmap)
{
	atomic_inc(&gmap->ref_count);
	return gmap;
}
EXPORT_SYMBOL_GPL(gmap_get);

/**
 * gmap_put - decrease reference counter for guest address space
 * @gmap: pointer to the guest address space structure
 *
 * If the reference counter reaches zero the guest address space is freed.
 */
void gmap_put(struct gmap *gmap)
{
	if (atomic_dec_return(&gmap->ref_count) == 0)
		gmap_free(gmap);
}
EXPORT_SYMBOL_GPL(gmap_put);

/**
 * gmap_remove - remove a guest address space but do not free it yet
 * @gmap: pointer to the guest address space structure
 */
void gmap_remove(struct gmap *gmap)
{
239
	struct gmap *sg, *next;
240
	unsigned long gmap_asce;
241 242 243 244 245 246 247 248 249 250

	/* Remove all shadow gmaps linked to this gmap */
	if (!list_empty(&gmap->children)) {
		spin_lock(&gmap->shadow_lock);
		list_for_each_entry_safe(sg, next, &gmap->children, list) {
			list_del(&sg->list);
			gmap_put(sg);
		}
		spin_unlock(&gmap->shadow_lock);
	}
251
	/* Remove gmap from the pre-mm list */
252
	spin_lock(&gmap->mm->context.lock);
253
	list_del_rcu(&gmap->list);
254 255 256 257 258 259 260 261
	if (list_empty(&gmap->mm->context.gmap_list))
		gmap_asce = 0;
	else if (list_is_singular(&gmap->mm->context.gmap_list))
		gmap_asce = list_first_entry(&gmap->mm->context.gmap_list,
					     struct gmap, list)->asce;
	else
		gmap_asce = -1UL;
	WRITE_ONCE(gmap->mm->context.gmap_asce, gmap_asce);
262
	spin_unlock(&gmap->mm->context.lock);
263 264 265 266 267
	synchronize_rcu();
	/* Put reference */
	gmap_put(gmap);
}
EXPORT_SYMBOL_GPL(gmap_remove);
268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288

/**
 * gmap_enable - switch primary space to the guest address space
 * @gmap: pointer to the guest address space structure
 */
void gmap_enable(struct gmap *gmap)
{
	S390_lowcore.gmap = (unsigned long) gmap;
}
EXPORT_SYMBOL_GPL(gmap_enable);

/**
 * gmap_disable - switch back to the standard primary address space
 * @gmap: pointer to the guest address space structure
 */
void gmap_disable(struct gmap *gmap)
{
	S390_lowcore.gmap = 0UL;
}
EXPORT_SYMBOL_GPL(gmap_disable);

289 290 291 292 293 294 295 296 297 298 299
/**
 * gmap_get_enabled - get a pointer to the currently enabled gmap
 *
 * Returns a pointer to the currently enabled gmap. 0 if none is enabled.
 */
struct gmap *gmap_get_enabled(void)
{
	return (struct gmap *) S390_lowcore.gmap;
}
EXPORT_SYMBOL_GPL(gmap_get_enabled);

300 301 302 303 304 305 306 307 308 309
/*
 * gmap_alloc_table is assumed to be called with mmap_sem held
 */
static int gmap_alloc_table(struct gmap *gmap, unsigned long *table,
			    unsigned long init, unsigned long gaddr)
{
	struct page *page;
	unsigned long *new;

	/* since we dont free the gmap table until gmap_free we can unlock */
310
	page = alloc_pages(GFP_KERNEL, CRST_ALLOC_ORDER);
311 312 313 314
	if (!page)
		return -ENOMEM;
	new = (unsigned long *) page_to_phys(page);
	crst_table_init(new, init);
315
	spin_lock(&gmap->guest_table_lock);
316 317 318 319 320 321 322
	if (*table & _REGION_ENTRY_INVALID) {
		list_add(&page->lru, &gmap->crst_list);
		*table = (unsigned long) new | _REGION_ENTRY_LENGTH |
			(*table & _REGION_ENTRY_TYPE_MASK);
		page->index = gaddr;
		page = NULL;
	}
323
	spin_unlock(&gmap->guest_table_lock);
324
	if (page)
325
		__free_pages(page, CRST_ALLOC_ORDER);
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
	return 0;
}

/**
 * __gmap_segment_gaddr - find virtual address from segment pointer
 * @entry: pointer to a segment table entry in the guest address space
 *
 * Returns the virtual address in the guest address space for the segment
 */
static unsigned long __gmap_segment_gaddr(unsigned long *entry)
{
	struct page *page;
	unsigned long offset, mask;

	offset = (unsigned long) entry / sizeof(unsigned long);
	offset = (offset & (PTRS_PER_PMD - 1)) * PMD_SIZE;
	mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1);
	page = virt_to_page((void *)((unsigned long) entry & mask));
	return page->index + offset;
}

/**
 * __gmap_unlink_by_vmaddr - unlink a single segment via a host address
 * @gmap: pointer to the guest address space structure
 * @vmaddr: address in the host process address space
 *
 * Returns 1 if a TLB flush is required
 */
static int __gmap_unlink_by_vmaddr(struct gmap *gmap, unsigned long vmaddr)
{
	unsigned long *entry;
	int flush = 0;

359
	BUG_ON(gmap_is_shadow(gmap));
360 361 362
	spin_lock(&gmap->guest_table_lock);
	entry = radix_tree_delete(&gmap->host_to_guest, vmaddr >> PMD_SHIFT);
	if (entry) {
363 364
		flush = (*entry != _SEGMENT_ENTRY_EMPTY);
		*entry = _SEGMENT_ENTRY_EMPTY;
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
	}
	spin_unlock(&gmap->guest_table_lock);
	return flush;
}

/**
 * __gmap_unmap_by_gaddr - unmap a single segment via a guest address
 * @gmap: pointer to the guest address space structure
 * @gaddr: address in the guest address space
 *
 * Returns 1 if a TLB flush is required
 */
static int __gmap_unmap_by_gaddr(struct gmap *gmap, unsigned long gaddr)
{
	unsigned long vmaddr;

	vmaddr = (unsigned long) radix_tree_delete(&gmap->guest_to_host,
						   gaddr >> PMD_SHIFT);
	return vmaddr ? __gmap_unlink_by_vmaddr(gmap, vmaddr) : 0;
}

/**
 * gmap_unmap_segment - unmap segment from the guest address space
 * @gmap: pointer to the guest address space structure
 * @to: address in the guest address space
 * @len: length of the memory area to unmap
 *
 * Returns 0 if the unmap succeeded, -EINVAL if not.
 */
int gmap_unmap_segment(struct gmap *gmap, unsigned long to, unsigned long len)
{
	unsigned long off;
	int flush;

399
	BUG_ON(gmap_is_shadow(gmap));
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
	if ((to | len) & (PMD_SIZE - 1))
		return -EINVAL;
	if (len == 0 || to + len < to)
		return -EINVAL;

	flush = 0;
	down_write(&gmap->mm->mmap_sem);
	for (off = 0; off < len; off += PMD_SIZE)
		flush |= __gmap_unmap_by_gaddr(gmap, to + off);
	up_write(&gmap->mm->mmap_sem);
	if (flush)
		gmap_flush_tlb(gmap);
	return 0;
}
EXPORT_SYMBOL_GPL(gmap_unmap_segment);

/**
 * gmap_map_segment - map a segment to the guest address space
 * @gmap: pointer to the guest address space structure
 * @from: source address in the parent address space
 * @to: target address in the guest address space
 * @len: length of the memory area to map
 *
 * Returns 0 if the mmap succeeded, -EINVAL or -ENOMEM if not.
 */
int gmap_map_segment(struct gmap *gmap, unsigned long from,
		     unsigned long to, unsigned long len)
{
	unsigned long off;
	int flush;

431
	BUG_ON(gmap_is_shadow(gmap));
432 433 434
	if ((from | to | len) & (PMD_SIZE - 1))
		return -EINVAL;
	if (len == 0 || from + len < from || to + len < to ||
435
	    from + len - 1 > TASK_SIZE_MAX || to + len - 1 > gmap->asce_end)
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
		return -EINVAL;

	flush = 0;
	down_write(&gmap->mm->mmap_sem);
	for (off = 0; off < len; off += PMD_SIZE) {
		/* Remove old translation */
		flush |= __gmap_unmap_by_gaddr(gmap, to + off);
		/* Store new translation */
		if (radix_tree_insert(&gmap->guest_to_host,
				      (to + off) >> PMD_SHIFT,
				      (void *) from + off))
			break;
	}
	up_write(&gmap->mm->mmap_sem);
	if (flush)
		gmap_flush_tlb(gmap);
	if (off >= len)
		return 0;
	gmap_unmap_segment(gmap, to, len);
	return -ENOMEM;
}
EXPORT_SYMBOL_GPL(gmap_map_segment);

/**
 * __gmap_translate - translate a guest address to a user space address
 * @gmap: pointer to guest mapping meta data structure
 * @gaddr: guest address
 *
 * Returns user space address which corresponds to the guest address or
 * -EFAULT if no such mapping exists.
 * This function does not establish potentially missing page table entries.
 * The mmap_sem of the mm that belongs to the address space must be held
 * when this function gets called.
469 470
 *
 * Note: Can also be called for shadow gmaps.
471 472 473 474 475 476 477
 */
unsigned long __gmap_translate(struct gmap *gmap, unsigned long gaddr)
{
	unsigned long vmaddr;

	vmaddr = (unsigned long)
		radix_tree_lookup(&gmap->guest_to_host, gaddr >> PMD_SHIFT);
478
	/* Note: guest_to_host is empty for a shadow gmap */
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
	return vmaddr ? (vmaddr | (gaddr & ~PMD_MASK)) : -EFAULT;
}
EXPORT_SYMBOL_GPL(__gmap_translate);

/**
 * gmap_translate - translate a guest address to a user space address
 * @gmap: pointer to guest mapping meta data structure
 * @gaddr: guest address
 *
 * Returns user space address which corresponds to the guest address or
 * -EFAULT if no such mapping exists.
 * This function does not establish potentially missing page table entries.
 */
unsigned long gmap_translate(struct gmap *gmap, unsigned long gaddr)
{
	unsigned long rc;

	down_read(&gmap->mm->mmap_sem);
	rc = __gmap_translate(gmap, gaddr);
	up_read(&gmap->mm->mmap_sem);
	return rc;
}
EXPORT_SYMBOL_GPL(gmap_translate);

/**
 * gmap_unlink - disconnect a page table from the gmap shadow tables
 * @gmap: pointer to guest mapping meta data structure
 * @table: pointer to the host page table
 * @vmaddr: vm address associated with the host page table
 */
void gmap_unlink(struct mm_struct *mm, unsigned long *table,
		 unsigned long vmaddr)
{
	struct gmap *gmap;
	int flush;

515 516
	rcu_read_lock();
	list_for_each_entry_rcu(gmap, &mm->context.gmap_list, list) {
517 518 519 520
		flush = __gmap_unlink_by_vmaddr(gmap, vmaddr);
		if (flush)
			gmap_flush_tlb(gmap);
	}
521
	rcu_read_unlock();
522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540
}

/**
 * gmap_link - set up shadow page tables to connect a host to a guest address
 * @gmap: pointer to guest mapping meta data structure
 * @gaddr: guest address
 * @vmaddr: vm address
 *
 * Returns 0 on success, -ENOMEM for out of memory conditions, and -EFAULT
 * if the vm address is already mapped to a different guest segment.
 * The mmap_sem of the mm that belongs to the address space must be held
 * when this function gets called.
 */
int __gmap_link(struct gmap *gmap, unsigned long gaddr, unsigned long vmaddr)
{
	struct mm_struct *mm;
	unsigned long *table;
	spinlock_t *ptl;
	pgd_t *pgd;
541
	p4d_t *p4d;
542 543 544 545
	pud_t *pud;
	pmd_t *pmd;
	int rc;

546
	BUG_ON(gmap_is_shadow(gmap));
547 548 549
	/* Create higher level tables in the gmap page table */
	table = gmap->table;
	if ((gmap->asce & _ASCE_TYPE_MASK) >= _ASCE_TYPE_REGION1) {
550
		table += (gaddr & _REGION1_INDEX) >> _REGION1_SHIFT;
551 552
		if ((*table & _REGION_ENTRY_INVALID) &&
		    gmap_alloc_table(gmap, table, _REGION2_ENTRY_EMPTY,
553
				     gaddr & _REGION1_MASK))
554 555 556 557
			return -ENOMEM;
		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
	}
	if ((gmap->asce & _ASCE_TYPE_MASK) >= _ASCE_TYPE_REGION2) {
558
		table += (gaddr & _REGION2_INDEX) >> _REGION2_SHIFT;
559 560
		if ((*table & _REGION_ENTRY_INVALID) &&
		    gmap_alloc_table(gmap, table, _REGION3_ENTRY_EMPTY,
561
				     gaddr & _REGION2_MASK))
562 563 564 565
			return -ENOMEM;
		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
	}
	if ((gmap->asce & _ASCE_TYPE_MASK) >= _ASCE_TYPE_REGION3) {
566
		table += (gaddr & _REGION3_INDEX) >> _REGION3_SHIFT;
567 568
		if ((*table & _REGION_ENTRY_INVALID) &&
		    gmap_alloc_table(gmap, table, _SEGMENT_ENTRY_EMPTY,
569
				     gaddr & _REGION3_MASK))
570 571 572
			return -ENOMEM;
		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
	}
573
	table += (gaddr & _SEGMENT_INDEX) >> _SEGMENT_SHIFT;
574 575 576 577
	/* Walk the parent mm page table */
	mm = gmap->mm;
	pgd = pgd_offset(mm, vmaddr);
	VM_BUG_ON(pgd_none(*pgd));
578 579 580
	p4d = p4d_offset(pgd, vmaddr);
	VM_BUG_ON(p4d_none(*p4d));
	pud = pud_offset(p4d, vmaddr);
581
	VM_BUG_ON(pud_none(*pud));
582 583 584
	/* large puds cannot yet be handled */
	if (pud_large(*pud))
		return -EFAULT;
585 586 587 588 589 590 591 592 593 594 595
	pmd = pmd_offset(pud, vmaddr);
	VM_BUG_ON(pmd_none(*pmd));
	/* large pmds cannot yet be handled */
	if (pmd_large(*pmd))
		return -EFAULT;
	/* Link gmap segment table entry location to page table. */
	rc = radix_tree_preload(GFP_KERNEL);
	if (rc)
		return rc;
	ptl = pmd_lock(mm, pmd);
	spin_lock(&gmap->guest_table_lock);
596
	if (*table == _SEGMENT_ENTRY_EMPTY) {
597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693
		rc = radix_tree_insert(&gmap->host_to_guest,
				       vmaddr >> PMD_SHIFT, table);
		if (!rc)
			*table = pmd_val(*pmd);
	} else
		rc = 0;
	spin_unlock(&gmap->guest_table_lock);
	spin_unlock(ptl);
	radix_tree_preload_end();
	return rc;
}

/**
 * gmap_fault - resolve a fault on a guest address
 * @gmap: pointer to guest mapping meta data structure
 * @gaddr: guest address
 * @fault_flags: flags to pass down to handle_mm_fault()
 *
 * Returns 0 on success, -ENOMEM for out of memory conditions, and -EFAULT
 * if the vm address is already mapped to a different guest segment.
 */
int gmap_fault(struct gmap *gmap, unsigned long gaddr,
	       unsigned int fault_flags)
{
	unsigned long vmaddr;
	int rc;
	bool unlocked;

	down_read(&gmap->mm->mmap_sem);

retry:
	unlocked = false;
	vmaddr = __gmap_translate(gmap, gaddr);
	if (IS_ERR_VALUE(vmaddr)) {
		rc = vmaddr;
		goto out_up;
	}
	if (fixup_user_fault(current, gmap->mm, vmaddr, fault_flags,
			     &unlocked)) {
		rc = -EFAULT;
		goto out_up;
	}
	/*
	 * In the case that fixup_user_fault unlocked the mmap_sem during
	 * faultin redo __gmap_translate to not race with a map/unmap_segment.
	 */
	if (unlocked)
		goto retry;

	rc = __gmap_link(gmap, gaddr, vmaddr);
out_up:
	up_read(&gmap->mm->mmap_sem);
	return rc;
}
EXPORT_SYMBOL_GPL(gmap_fault);

/*
 * this function is assumed to be called with mmap_sem held
 */
void __gmap_zap(struct gmap *gmap, unsigned long gaddr)
{
	unsigned long vmaddr;
	spinlock_t *ptl;
	pte_t *ptep;

	/* Find the vm address for the guest address */
	vmaddr = (unsigned long) radix_tree_lookup(&gmap->guest_to_host,
						   gaddr >> PMD_SHIFT);
	if (vmaddr) {
		vmaddr |= gaddr & ~PMD_MASK;
		/* Get pointer to the page table entry */
		ptep = get_locked_pte(gmap->mm, vmaddr, &ptl);
		if (likely(ptep))
			ptep_zap_unused(gmap->mm, vmaddr, ptep, 0);
		pte_unmap_unlock(ptep, ptl);
	}
}
EXPORT_SYMBOL_GPL(__gmap_zap);

void gmap_discard(struct gmap *gmap, unsigned long from, unsigned long to)
{
	unsigned long gaddr, vmaddr, size;
	struct vm_area_struct *vma;

	down_read(&gmap->mm->mmap_sem);
	for (gaddr = from; gaddr < to;
	     gaddr = (gaddr + PMD_SIZE) & PMD_MASK) {
		/* Find the vm address for the guest address */
		vmaddr = (unsigned long)
			radix_tree_lookup(&gmap->guest_to_host,
					  gaddr >> PMD_SHIFT);
		if (!vmaddr)
			continue;
		vmaddr |= gaddr & ~PMD_MASK;
		/* Find vma in the parent mm */
		vma = find_vma(gmap->mm, vmaddr);
		size = min(to - gaddr, PMD_SIZE - (gaddr & ~PMD_MASK));
694
		zap_page_range(vma, vmaddr, size);
695 696 697 698 699 700 701 702 703
	}
	up_read(&gmap->mm->mmap_sem);
}
EXPORT_SYMBOL_GPL(gmap_discard);

static LIST_HEAD(gmap_notifier_list);
static DEFINE_SPINLOCK(gmap_notifier_lock);

/**
704
 * gmap_register_pte_notifier - register a pte invalidation callback
705 706
 * @nb: pointer to the gmap notifier block
 */
707
void gmap_register_pte_notifier(struct gmap_notifier *nb)
708 709
{
	spin_lock(&gmap_notifier_lock);
710
	list_add_rcu(&nb->list, &gmap_notifier_list);
711 712
	spin_unlock(&gmap_notifier_lock);
}
713
EXPORT_SYMBOL_GPL(gmap_register_pte_notifier);
714 715

/**
716
 * gmap_unregister_pte_notifier - remove a pte invalidation callback
717 718
 * @nb: pointer to the gmap notifier block
 */
719
void gmap_unregister_pte_notifier(struct gmap_notifier *nb)
720 721
{
	spin_lock(&gmap_notifier_lock);
722
	list_del_rcu(&nb->list);
723
	spin_unlock(&gmap_notifier_lock);
724
	synchronize_rcu();
725
}
726
EXPORT_SYMBOL_GPL(gmap_unregister_pte_notifier);
727

728 729 730 731 732 733 734 735 736 737 738 739 740 741 742
/**
 * gmap_call_notifier - call all registered invalidation callbacks
 * @gmap: pointer to guest mapping meta data structure
 * @start: start virtual address in the guest address space
 * @end: end virtual address in the guest address space
 */
static void gmap_call_notifier(struct gmap *gmap, unsigned long start,
			       unsigned long end)
{
	struct gmap_notifier *nb;

	list_for_each_entry(nb, &gmap_notifier_list, list)
		nb->notifier_call(gmap, start, end);
}

743
/**
744 745 746
 * gmap_table_walk - walk the gmap page tables
 * @gmap: pointer to guest mapping meta data structure
 * @gaddr: virtual address in the guest address space
747 748 749 750 751 752 753 754 755 756 757
 * @level: page table level to stop at
 *
 * Returns a table entry pointer for the given guest address and @level
 * @level=0 : returns a pointer to a page table table entry (or NULL)
 * @level=1 : returns a pointer to a segment table entry (or NULL)
 * @level=2 : returns a pointer to a region-3 table entry (or NULL)
 * @level=3 : returns a pointer to a region-2 table entry (or NULL)
 * @level=4 : returns a pointer to a region-1 table entry (or NULL)
 *
 * Returns NULL if the gmap page tables could not be walked to the
 * requested level.
758
 *
759
 * Note: Can also be called for shadow gmaps.
760 761
 */
static inline unsigned long *gmap_table_walk(struct gmap *gmap,
762
					     unsigned long gaddr, int level)
763 764 765
{
	unsigned long *table;

766 767 768 769 770 771
	if ((gmap->asce & _ASCE_TYPE_MASK) + 4 < (level * 4))
		return NULL;
	if (gmap_is_shadow(gmap) && gmap->removed)
		return NULL;
	if (gaddr & (-1UL << (31 + ((gmap->asce & _ASCE_TYPE_MASK) >> 2)*11)))
		return NULL;
772 773 774
	table = gmap->table;
	switch (gmap->asce & _ASCE_TYPE_MASK) {
	case _ASCE_TYPE_REGION1:
775
		table += (gaddr & _REGION1_INDEX) >> _REGION1_SHIFT;
776 777
		if (level == 4)
			break;
778 779 780 781 782
		if (*table & _REGION_ENTRY_INVALID)
			return NULL;
		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
		/* Fallthrough */
	case _ASCE_TYPE_REGION2:
783
		table += (gaddr & _REGION2_INDEX) >> _REGION2_SHIFT;
784 785
		if (level == 3)
			break;
786 787 788 789 790
		if (*table & _REGION_ENTRY_INVALID)
			return NULL;
		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
		/* Fallthrough */
	case _ASCE_TYPE_REGION3:
791
		table += (gaddr & _REGION3_INDEX) >> _REGION3_SHIFT;
792 793
		if (level == 2)
			break;
794 795 796 797 798
		if (*table & _REGION_ENTRY_INVALID)
			return NULL;
		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
		/* Fallthrough */
	case _ASCE_TYPE_SEGMENT:
799
		table += (gaddr & _SEGMENT_INDEX) >> _SEGMENT_SHIFT;
800 801 802 803 804
		if (level == 1)
			break;
		if (*table & _REGION_ENTRY_INVALID)
			return NULL;
		table = (unsigned long *)(*table & _SEGMENT_ENTRY_ORIGIN);
805
		table += (gaddr & _PAGE_INDEX) >> _PAGE_SHIFT;
806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823
	}
	return table;
}

/**
 * gmap_pte_op_walk - walk the gmap page table, get the page table lock
 *		      and return the pte pointer
 * @gmap: pointer to guest mapping meta data structure
 * @gaddr: virtual address in the guest address space
 * @ptl: pointer to the spinlock pointer
 *
 * Returns a pointer to the locked pte for a guest address, or NULL
 */
static pte_t *gmap_pte_op_walk(struct gmap *gmap, unsigned long gaddr,
			       spinlock_t **ptl)
{
	unsigned long *table;

824
	BUG_ON(gmap_is_shadow(gmap));
825
	/* Walk the gmap page table, lock and get pte pointer */
826
	table = gmap_table_walk(gmap, gaddr, 1); /* get segment pointer */
827
	if (!table || *table & _SEGMENT_ENTRY_INVALID)
828 829 830 831 832 833 834 835 836
		return NULL;
	return pte_alloc_map_lock(gmap->mm, (pmd_t *) table, gaddr, ptl);
}

/**
 * gmap_pte_op_fixup - force a page in and connect the gmap page table
 * @gmap: pointer to guest mapping meta data structure
 * @gaddr: virtual address in the guest address space
 * @vmaddr: address in the host process address space
837
 * @prot: indicates access rights: PROT_NONE, PROT_READ or PROT_WRITE
838 839 840 841 842 843
 *
 * Returns 0 if the caller can retry __gmap_translate (might fail again),
 * -ENOMEM if out of memory and -EFAULT if anything goes wrong while fixing
 * up or connecting the gmap page table.
 */
static int gmap_pte_op_fixup(struct gmap *gmap, unsigned long gaddr,
844
			     unsigned long vmaddr, int prot)
845 846
{
	struct mm_struct *mm = gmap->mm;
847
	unsigned int fault_flags;
848 849
	bool unlocked = false;

850
	BUG_ON(gmap_is_shadow(gmap));
851 852
	fault_flags = (prot == PROT_WRITE) ? FAULT_FLAG_WRITE : 0;
	if (fixup_user_fault(current, mm, vmaddr, fault_flags, &unlocked))
853 854 855 856 857 858
		return -EFAULT;
	if (unlocked)
		/* lost mmap_sem, caller has to retry __gmap_translate */
		return 0;
	/* Connect the page tables */
	return __gmap_link(gmap, gaddr, vmaddr);
859 860 861
}

/**
862 863 864 865 866 867 868 869
 * gmap_pte_op_end - release the page table lock
 * @ptl: pointer to the spinlock pointer
 */
static void gmap_pte_op_end(spinlock_t *ptl)
{
	spin_unlock(ptl);
}

870 871
/*
 * gmap_protect_range - remove access rights to memory and set pgste bits
872 873 874
 * @gmap: pointer to guest mapping meta data structure
 * @gaddr: virtual address in the guest address space
 * @len: size of area
875 876 877 878 879 880 881
 * @prot: indicates access rights: PROT_NONE, PROT_READ or PROT_WRITE
 * @bits: pgste notification bits to set
 *
 * Returns 0 if successfully protected, -ENOMEM if out of memory and
 * -EFAULT if gaddr is invalid (or mapping for shadows is missing).
 *
 * Called with sg->mm->mmap_sem in read.
882
 */
883 884
static int gmap_protect_range(struct gmap *gmap, unsigned long gaddr,
			      unsigned long len, int prot, unsigned long bits)
885
{
886
	unsigned long vmaddr;
887 888
	spinlock_t *ptl;
	pte_t *ptep;
889 890
	int rc;

891
	BUG_ON(gmap_is_shadow(gmap));
892 893 894 895 896 897 898 899 900 901 902
	while (len) {
		rc = -EAGAIN;
		ptep = gmap_pte_op_walk(gmap, gaddr, &ptl);
		if (ptep) {
			rc = ptep_force_prot(gmap->mm, gaddr, ptep, prot, bits);
			gmap_pte_op_end(ptl);
		}
		if (rc) {
			vmaddr = __gmap_translate(gmap, gaddr);
			if (IS_ERR_VALUE(vmaddr))
				return vmaddr;
903
			rc = gmap_pte_op_fixup(gmap, gaddr, vmaddr, prot);
904 905 906 907 908 909 910 911 912
			if (rc)
				return rc;
			continue;
		}
		gaddr += PAGE_SIZE;
		len -= PAGE_SIZE;
	}
	return 0;
}
913

914 915 916
/**
 * gmap_mprotect_notify - change access rights for a range of ptes and
 *                        call the notifier if any pte changes again
917 918 919
 * @gmap: pointer to guest mapping meta data structure
 * @gaddr: virtual address in the guest address space
 * @len: size of area
920
 * @prot: indicates access rights: PROT_NONE, PROT_READ or PROT_WRITE
921
 *
922 923 924 925 926
 * Returns 0 if for each page in the given range a gmap mapping exists,
 * the new access rights could be set and the notifier could be armed.
 * If the gmap mapping is missing for one or more pages -EFAULT is
 * returned. If no memory could be allocated -ENOMEM is returned.
 * This function establishes missing page table entries.
927
 */
928 929
int gmap_mprotect_notify(struct gmap *gmap, unsigned long gaddr,
			 unsigned long len, int prot)
930
{
931
	int rc;
932

933
	if ((gaddr & ~PAGE_MASK) || (len & ~PAGE_MASK) || gmap_is_shadow(gmap))
934
		return -EINVAL;
935
	if (!MACHINE_HAS_ESOP && prot == PROT_READ)
936 937
		return -EINVAL;
	down_read(&gmap->mm->mmap_sem);
938 939 940 941 942 943 944 945 946 947 948 949 950 951
	rc = gmap_protect_range(gmap, gaddr, len, prot, PGSTE_IN_BIT);
	up_read(&gmap->mm->mmap_sem);
	return rc;
}
EXPORT_SYMBOL_GPL(gmap_mprotect_notify);

/**
 * gmap_read_table - get an unsigned long value from a guest page table using
 *                   absolute addressing, without marking the page referenced.
 * @gmap: pointer to guest mapping meta data structure
 * @gaddr: virtual address in the guest address space
 * @val: pointer to the unsigned long value to return
 *
 * Returns 0 if the value was read, -ENOMEM if out of memory and -EFAULT
952 953
 * if reading using the virtual address failed. -EINVAL if called on a gmap
 * shadow.
954 955 956 957 958 959 960 961 962 963
 *
 * Called with gmap->mm->mmap_sem in read.
 */
int gmap_read_table(struct gmap *gmap, unsigned long gaddr, unsigned long *val)
{
	unsigned long address, vmaddr;
	spinlock_t *ptl;
	pte_t *ptep, pte;
	int rc;

964 965 966
	if (gmap_is_shadow(gmap))
		return -EINVAL;

967
	while (1) {
968 969 970
		rc = -EAGAIN;
		ptep = gmap_pte_op_walk(gmap, gaddr, &ptl);
		if (ptep) {
971 972 973 974 975 976 977 978 979
			pte = *ptep;
			if (pte_present(pte) && (pte_val(pte) & _PAGE_READ)) {
				address = pte_val(pte) & PAGE_MASK;
				address += gaddr & ~PAGE_MASK;
				*val = *(unsigned long *) address;
				pte_val(*ptep) |= _PAGE_YOUNG;
				/* Do *NOT* clear the _PAGE_INVALID bit! */
				rc = 0;
			}
980
			gmap_pte_op_end(ptl);
981
		}
982 983 984 985 986
		if (!rc)
			break;
		vmaddr = __gmap_translate(gmap, gaddr);
		if (IS_ERR_VALUE(vmaddr)) {
			rc = vmaddr;
987 988
			break;
		}
989
		rc = gmap_pte_op_fixup(gmap, gaddr, vmaddr, PROT_READ);
990 991 992 993 994
		if (rc)
			break;
	}
	return rc;
}
995
EXPORT_SYMBOL_GPL(gmap_read_table);
996 997

/**
998 999 1000 1001
 * gmap_insert_rmap - add a rmap to the host_to_rmap radix tree
 * @sg: pointer to the shadow guest address space structure
 * @vmaddr: vm address associated with the rmap
 * @rmap: pointer to the rmap structure
1002
 *
1003
 * Called with the sg->guest_table_lock
1004
 */
1005 1006
static inline void gmap_insert_rmap(struct gmap *sg, unsigned long vmaddr,
				    struct gmap_rmap *rmap)
1007
{
H
Heiko Carstens 已提交
1008
	void __rcu **slot;
1009

1010 1011 1012 1013 1014
	BUG_ON(!gmap_is_shadow(sg));
	slot = radix_tree_lookup_slot(&sg->host_to_rmap, vmaddr >> PAGE_SHIFT);
	if (slot) {
		rmap->next = radix_tree_deref_slot_protected(slot,
							&sg->guest_table_lock);
1015
		radix_tree_replace_slot(&sg->host_to_rmap, slot, rmap);
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
	} else {
		rmap->next = NULL;
		radix_tree_insert(&sg->host_to_rmap, vmaddr >> PAGE_SHIFT,
				  rmap);
	}
}

/**
 * gmap_protect_rmap - modify access rights to memory and create an rmap
 * @sg: pointer to the shadow guest address space structure
 * @raddr: rmap address in the shadow gmap
 * @paddr: address in the parent guest address space
 * @len: length of the memory area to protect
 * @prot: indicates access rights: none, read-only or read-write
 *
 * Returns 0 if successfully protected and the rmap was created, -ENOMEM
 * if out of memory and -EFAULT if paddr is invalid.
 */
static int gmap_protect_rmap(struct gmap *sg, unsigned long raddr,
			     unsigned long paddr, unsigned long len, int prot)
{
	struct gmap *parent;
	struct gmap_rmap *rmap;
	unsigned long vmaddr;
	spinlock_t *ptl;
	pte_t *ptep;
	int rc;

	BUG_ON(!gmap_is_shadow(sg));
	parent = sg->parent;
	while (len) {
		vmaddr = __gmap_translate(parent, paddr);
		if (IS_ERR_VALUE(vmaddr))
			return vmaddr;
		rmap = kzalloc(sizeof(*rmap), GFP_KERNEL);
		if (!rmap)
			return -ENOMEM;
		rmap->raddr = raddr;
		rc = radix_tree_preload(GFP_KERNEL);
1055
		if (rc) {
1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072
			kfree(rmap);
			return rc;
		}
		rc = -EAGAIN;
		ptep = gmap_pte_op_walk(parent, paddr, &ptl);
		if (ptep) {
			spin_lock(&sg->guest_table_lock);
			rc = ptep_force_prot(parent->mm, paddr, ptep, prot,
					     PGSTE_VSIE_BIT);
			if (!rc)
				gmap_insert_rmap(sg, vmaddr, rmap);
			spin_unlock(&sg->guest_table_lock);
			gmap_pte_op_end(ptl);
		}
		radix_tree_preload_end();
		if (rc) {
			kfree(rmap);
1073
			rc = gmap_pte_op_fixup(parent, paddr, vmaddr, prot);
1074
			if (rc)
1075
				return rc;
1076 1077
			continue;
		}
1078
		paddr += PAGE_SIZE;
1079
		len -= PAGE_SIZE;
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
	return 0;
}

#define _SHADOW_RMAP_MASK	0x7
#define _SHADOW_RMAP_REGION1	0x5
#define _SHADOW_RMAP_REGION2	0x4
#define _SHADOW_RMAP_REGION3	0x3
#define _SHADOW_RMAP_SEGMENT	0x2
#define _SHADOW_RMAP_PGTABLE	0x1

/**
 * gmap_idte_one - invalidate a single region or segment table entry
 * @asce: region or segment table *origin* + table-type bits
 * @vaddr: virtual address to identify the table entry to flush
 *
 * The invalid bit of a single region or segment table entry is set
 * and the associated TLB entries depending on the entry are flushed.
 * The table-type of the @asce identifies the portion of the @vaddr
 * that is used as the invalidation index.
 */
static inline void gmap_idte_one(unsigned long asce, unsigned long vaddr)
{
	asm volatile(
		"	.insn	rrf,0xb98e0000,%0,%1,0,0"
		: : "a" (asce), "a" (vaddr) : "cc", "memory");
}

/**
 * gmap_unshadow_page - remove a page from a shadow page table
 * @sg: pointer to the shadow guest address space structure
 * @raddr: rmap address in the shadow guest address space
 *
 * Called with the sg->guest_table_lock
 */
static void gmap_unshadow_page(struct gmap *sg, unsigned long raddr)
{
	unsigned long *table;

	BUG_ON(!gmap_is_shadow(sg));
	table = gmap_table_walk(sg, raddr, 0); /* get page table pointer */
	if (!table || *table & _PAGE_INVALID)
		return;
1123
	gmap_call_notifier(sg, raddr, raddr + _PAGE_SIZE - 1);
1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140
	ptep_unshadow_pte(sg->mm, raddr, (pte_t *) table);
}

/**
 * __gmap_unshadow_pgt - remove all entries from a shadow page table
 * @sg: pointer to the shadow guest address space structure
 * @raddr: rmap address in the shadow guest address space
 * @pgt: pointer to the start of a shadow page table
 *
 * Called with the sg->guest_table_lock
 */
static void __gmap_unshadow_pgt(struct gmap *sg, unsigned long raddr,
				unsigned long *pgt)
{
	int i;

	BUG_ON(!gmap_is_shadow(sg));
1141
	for (i = 0; i < _PAGE_ENTRIES; i++, raddr += _PAGE_SIZE)
1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158
		pgt[i] = _PAGE_INVALID;
}

/**
 * gmap_unshadow_pgt - remove a shadow page table from a segment entry
 * @sg: pointer to the shadow guest address space structure
 * @raddr: address in the shadow guest address space
 *
 * Called with the sg->guest_table_lock
 */
static void gmap_unshadow_pgt(struct gmap *sg, unsigned long raddr)
{
	unsigned long sto, *ste, *pgt;
	struct page *page;

	BUG_ON(!gmap_is_shadow(sg));
	ste = gmap_table_walk(sg, raddr, 1); /* get segment pointer */
1159
	if (!ste || !(*ste & _SEGMENT_ENTRY_ORIGIN))
1160
		return;
1161 1162
	gmap_call_notifier(sg, raddr, raddr + _SEGMENT_SIZE - 1);
	sto = (unsigned long) (ste - ((raddr & _SEGMENT_INDEX) >> _SEGMENT_SHIFT));
1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183
	gmap_idte_one(sto | _ASCE_TYPE_SEGMENT, raddr);
	pgt = (unsigned long *)(*ste & _SEGMENT_ENTRY_ORIGIN);
	*ste = _SEGMENT_ENTRY_EMPTY;
	__gmap_unshadow_pgt(sg, raddr, pgt);
	/* Free page table */
	page = pfn_to_page(__pa(pgt) >> PAGE_SHIFT);
	list_del(&page->lru);
	page_table_free_pgste(page);
}

/**
 * __gmap_unshadow_sgt - remove all entries from a shadow segment table
 * @sg: pointer to the shadow guest address space structure
 * @raddr: rmap address in the shadow guest address space
 * @sgt: pointer to the start of a shadow segment table
 *
 * Called with the sg->guest_table_lock
 */
static void __gmap_unshadow_sgt(struct gmap *sg, unsigned long raddr,
				unsigned long *sgt)
{
H
Heiko Carstens 已提交
1184
	unsigned long *pgt;
1185 1186 1187 1188
	struct page *page;
	int i;

	BUG_ON(!gmap_is_shadow(sg));
1189
	for (i = 0; i < _CRST_ENTRIES; i++, raddr += _SEGMENT_SIZE) {
1190
		if (!(sgt[i] & _SEGMENT_ENTRY_ORIGIN))
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
			continue;
		pgt = (unsigned long *)(sgt[i] & _REGION_ENTRY_ORIGIN);
		sgt[i] = _SEGMENT_ENTRY_EMPTY;
		__gmap_unshadow_pgt(sg, raddr, pgt);
		/* Free page table */
		page = pfn_to_page(__pa(pgt) >> PAGE_SHIFT);
		list_del(&page->lru);
		page_table_free_pgste(page);
	}
}

/**
 * gmap_unshadow_sgt - remove a shadow segment table from a region-3 entry
 * @sg: pointer to the shadow guest address space structure
 * @raddr: rmap address in the shadow guest address space
 *
 * Called with the shadow->guest_table_lock
 */
static void gmap_unshadow_sgt(struct gmap *sg, unsigned long raddr)
{
	unsigned long r3o, *r3e, *sgt;
	struct page *page;

	BUG_ON(!gmap_is_shadow(sg));
	r3e = gmap_table_walk(sg, raddr, 2); /* get region-3 pointer */
1216
	if (!r3e || !(*r3e & _REGION_ENTRY_ORIGIN))
1217
		return;
1218 1219
	gmap_call_notifier(sg, raddr, raddr + _REGION3_SIZE - 1);
	r3o = (unsigned long) (r3e - ((raddr & _REGION3_INDEX) >> _REGION3_SHIFT));
1220 1221 1222 1223 1224 1225 1226
	gmap_idte_one(r3o | _ASCE_TYPE_REGION3, raddr);
	sgt = (unsigned long *)(*r3e & _REGION_ENTRY_ORIGIN);
	*r3e = _REGION3_ENTRY_EMPTY;
	__gmap_unshadow_sgt(sg, raddr, sgt);
	/* Free segment table */
	page = pfn_to_page(__pa(sgt) >> PAGE_SHIFT);
	list_del(&page->lru);
1227
	__free_pages(page, CRST_ALLOC_ORDER);
1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240
}

/**
 * __gmap_unshadow_r3t - remove all entries from a shadow region-3 table
 * @sg: pointer to the shadow guest address space structure
 * @raddr: address in the shadow guest address space
 * @r3t: pointer to the start of a shadow region-3 table
 *
 * Called with the sg->guest_table_lock
 */
static void __gmap_unshadow_r3t(struct gmap *sg, unsigned long raddr,
				unsigned long *r3t)
{
H
Heiko Carstens 已提交
1241
	unsigned long *sgt;
1242 1243 1244 1245
	struct page *page;
	int i;

	BUG_ON(!gmap_is_shadow(sg));
1246
	for (i = 0; i < _CRST_ENTRIES; i++, raddr += _REGION3_SIZE) {
1247
		if (!(r3t[i] & _REGION_ENTRY_ORIGIN))
1248 1249 1250 1251 1252 1253 1254
			continue;
		sgt = (unsigned long *)(r3t[i] & _REGION_ENTRY_ORIGIN);
		r3t[i] = _REGION3_ENTRY_EMPTY;
		__gmap_unshadow_sgt(sg, raddr, sgt);
		/* Free segment table */
		page = pfn_to_page(__pa(sgt) >> PAGE_SHIFT);
		list_del(&page->lru);
1255
		__free_pages(page, CRST_ALLOC_ORDER);
1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272
	}
}

/**
 * gmap_unshadow_r3t - remove a shadow region-3 table from a region-2 entry
 * @sg: pointer to the shadow guest address space structure
 * @raddr: rmap address in the shadow guest address space
 *
 * Called with the sg->guest_table_lock
 */
static void gmap_unshadow_r3t(struct gmap *sg, unsigned long raddr)
{
	unsigned long r2o, *r2e, *r3t;
	struct page *page;

	BUG_ON(!gmap_is_shadow(sg));
	r2e = gmap_table_walk(sg, raddr, 3); /* get region-2 pointer */
1273
	if (!r2e || !(*r2e & _REGION_ENTRY_ORIGIN))
1274
		return;
1275 1276
	gmap_call_notifier(sg, raddr, raddr + _REGION2_SIZE - 1);
	r2o = (unsigned long) (r2e - ((raddr & _REGION2_INDEX) >> _REGION2_SHIFT));
1277 1278 1279 1280 1281 1282 1283
	gmap_idte_one(r2o | _ASCE_TYPE_REGION2, raddr);
	r3t = (unsigned long *)(*r2e & _REGION_ENTRY_ORIGIN);
	*r2e = _REGION2_ENTRY_EMPTY;
	__gmap_unshadow_r3t(sg, raddr, r3t);
	/* Free region 3 table */
	page = pfn_to_page(__pa(r3t) >> PAGE_SHIFT);
	list_del(&page->lru);
1284
	__free_pages(page, CRST_ALLOC_ORDER);
1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297
}

/**
 * __gmap_unshadow_r2t - remove all entries from a shadow region-2 table
 * @sg: pointer to the shadow guest address space structure
 * @raddr: rmap address in the shadow guest address space
 * @r2t: pointer to the start of a shadow region-2 table
 *
 * Called with the sg->guest_table_lock
 */
static void __gmap_unshadow_r2t(struct gmap *sg, unsigned long raddr,
				unsigned long *r2t)
{
H
Heiko Carstens 已提交
1298
	unsigned long *r3t;
1299 1300 1301 1302
	struct page *page;
	int i;

	BUG_ON(!gmap_is_shadow(sg));
1303
	for (i = 0; i < _CRST_ENTRIES; i++, raddr += _REGION2_SIZE) {
1304
		if (!(r2t[i] & _REGION_ENTRY_ORIGIN))
1305 1306 1307 1308 1309 1310 1311
			continue;
		r3t = (unsigned long *)(r2t[i] & _REGION_ENTRY_ORIGIN);
		r2t[i] = _REGION2_ENTRY_EMPTY;
		__gmap_unshadow_r3t(sg, raddr, r3t);
		/* Free region 3 table */
		page = pfn_to_page(__pa(r3t) >> PAGE_SHIFT);
		list_del(&page->lru);
1312
		__free_pages(page, CRST_ALLOC_ORDER);
1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329
	}
}

/**
 * gmap_unshadow_r2t - remove a shadow region-2 table from a region-1 entry
 * @sg: pointer to the shadow guest address space structure
 * @raddr: rmap address in the shadow guest address space
 *
 * Called with the sg->guest_table_lock
 */
static void gmap_unshadow_r2t(struct gmap *sg, unsigned long raddr)
{
	unsigned long r1o, *r1e, *r2t;
	struct page *page;

	BUG_ON(!gmap_is_shadow(sg));
	r1e = gmap_table_walk(sg, raddr, 4); /* get region-1 pointer */
1330
	if (!r1e || !(*r1e & _REGION_ENTRY_ORIGIN))
1331
		return;
1332 1333
	gmap_call_notifier(sg, raddr, raddr + _REGION1_SIZE - 1);
	r1o = (unsigned long) (r1e - ((raddr & _REGION1_INDEX) >> _REGION1_SHIFT));
1334 1335 1336 1337 1338 1339 1340
	gmap_idte_one(r1o | _ASCE_TYPE_REGION1, raddr);
	r2t = (unsigned long *)(*r1e & _REGION_ENTRY_ORIGIN);
	*r1e = _REGION1_ENTRY_EMPTY;
	__gmap_unshadow_r2t(sg, raddr, r2t);
	/* Free region 2 table */
	page = pfn_to_page(__pa(r2t) >> PAGE_SHIFT);
	list_del(&page->lru);
1341
	__free_pages(page, CRST_ALLOC_ORDER);
1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360
}

/**
 * __gmap_unshadow_r1t - remove all entries from a shadow region-1 table
 * @sg: pointer to the shadow guest address space structure
 * @raddr: rmap address in the shadow guest address space
 * @r1t: pointer to the start of a shadow region-1 table
 *
 * Called with the shadow->guest_table_lock
 */
static void __gmap_unshadow_r1t(struct gmap *sg, unsigned long raddr,
				unsigned long *r1t)
{
	unsigned long asce, *r2t;
	struct page *page;
	int i;

	BUG_ON(!gmap_is_shadow(sg));
	asce = (unsigned long) r1t | _ASCE_TYPE_REGION1;
1361
	for (i = 0; i < _CRST_ENTRIES; i++, raddr += _REGION1_SIZE) {
1362
		if (!(r1t[i] & _REGION_ENTRY_ORIGIN))
1363 1364 1365 1366 1367 1368 1369 1370 1371
			continue;
		r2t = (unsigned long *)(r1t[i] & _REGION_ENTRY_ORIGIN);
		__gmap_unshadow_r2t(sg, raddr, r2t);
		/* Clear entry and flush translation r1t -> r2t */
		gmap_idte_one(asce, raddr);
		r1t[i] = _REGION1_ENTRY_EMPTY;
		/* Free region 2 table */
		page = pfn_to_page(__pa(r2t) >> PAGE_SHIFT);
		list_del(&page->lru);
1372
		__free_pages(page, CRST_ALLOC_ORDER);
1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390
	}
}

/**
 * gmap_unshadow - remove a shadow page table completely
 * @sg: pointer to the shadow guest address space structure
 *
 * Called with sg->guest_table_lock
 */
static void gmap_unshadow(struct gmap *sg)
{
	unsigned long *table;

	BUG_ON(!gmap_is_shadow(sg));
	if (sg->removed)
		return;
	sg->removed = 1;
	gmap_call_notifier(sg, 0, -1UL);
1391
	gmap_flush_tlb(sg);
1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412
	table = (unsigned long *)(sg->asce & _ASCE_ORIGIN);
	switch (sg->asce & _ASCE_TYPE_MASK) {
	case _ASCE_TYPE_REGION1:
		__gmap_unshadow_r1t(sg, 0, table);
		break;
	case _ASCE_TYPE_REGION2:
		__gmap_unshadow_r2t(sg, 0, table);
		break;
	case _ASCE_TYPE_REGION3:
		__gmap_unshadow_r3t(sg, 0, table);
		break;
	case _ASCE_TYPE_SEGMENT:
		__gmap_unshadow_sgt(sg, 0, table);
		break;
	}
}

/**
 * gmap_find_shadow - find a specific asce in the list of shadow tables
 * @parent: pointer to the parent gmap
 * @asce: ASCE for which the shadow table is created
1413
 * @edat_level: edat level to be used for the shadow translation
1414 1415
 *
 * Returns the pointer to a gmap if a shadow table with the given asce is
1416 1417
 * already available, ERR_PTR(-EAGAIN) if another one is just being created,
 * otherwise NULL
1418
 */
1419 1420
static struct gmap *gmap_find_shadow(struct gmap *parent, unsigned long asce,
				     int edat_level)
1421 1422 1423 1424
{
	struct gmap *sg;

	list_for_each_entry(sg, &parent->children, list) {
1425 1426
		if (sg->orig_asce != asce || sg->edat_level != edat_level ||
		    sg->removed)
1427
			continue;
1428 1429
		if (!sg->initialized)
			return ERR_PTR(-EAGAIN);
1430 1431 1432 1433 1434 1435
		atomic_inc(&sg->ref_count);
		return sg;
	}
	return NULL;
}

1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455
/**
 * gmap_shadow_valid - check if a shadow guest address space matches the
 *                     given properties and is still valid
 * @sg: pointer to the shadow guest address space structure
 * @asce: ASCE for which the shadow table is requested
 * @edat_level: edat level to be used for the shadow translation
 *
 * Returns 1 if the gmap shadow is still valid and matches the given
 * properties, the caller can continue using it. Returns 0 otherwise, the
 * caller has to request a new shadow gmap in this case.
 *
 */
int gmap_shadow_valid(struct gmap *sg, unsigned long asce, int edat_level)
{
	if (sg->removed)
		return 0;
	return sg->orig_asce == asce && sg->edat_level == edat_level;
}
EXPORT_SYMBOL_GPL(gmap_shadow_valid);

1456 1457 1458 1459
/**
 * gmap_shadow - create/find a shadow guest address space
 * @parent: pointer to the parent gmap
 * @asce: ASCE for which the shadow table is created
1460
 * @edat_level: edat level to be used for the shadow translation
1461 1462 1463 1464 1465 1466
 *
 * The pages of the top level page table referred by the asce parameter
 * will be set to read-only and marked in the PGSTEs of the kvm process.
 * The shadow table will be removed automatically on any change to the
 * PTE mapping for the source table.
 *
1467 1468 1469
 * Returns a guest address space structure, ERR_PTR(-ENOMEM) if out of memory,
 * ERR_PTR(-EAGAIN) if the caller has to retry and ERR_PTR(-EFAULT) if the
 * parent gmap table could not be protected.
1470
 */
1471 1472
struct gmap *gmap_shadow(struct gmap *parent, unsigned long asce,
			 int edat_level)
1473 1474 1475 1476 1477 1478 1479
{
	struct gmap *sg, *new;
	unsigned long limit;
	int rc;

	BUG_ON(gmap_is_shadow(parent));
	spin_lock(&parent->shadow_lock);
1480
	sg = gmap_find_shadow(parent, asce, edat_level);
1481 1482 1483 1484 1485
	spin_unlock(&parent->shadow_lock);
	if (sg)
		return sg;
	/* Create a new shadow gmap */
	limit = -1UL >> (33 - (((asce & _ASCE_TYPE_MASK) >> 2) * 11));
1486 1487
	if (asce & _ASCE_REAL_SPACE)
		limit = -1UL;
1488 1489
	new = gmap_alloc(limit);
	if (!new)
1490
		return ERR_PTR(-ENOMEM);
1491 1492 1493
	new->mm = parent->mm;
	new->parent = gmap_get(parent);
	new->orig_asce = asce;
1494
	new->edat_level = edat_level;
1495 1496 1497
	new->initialized = false;
	spin_lock(&parent->shadow_lock);
	/* Recheck if another CPU created the same shadow */
1498
	sg = gmap_find_shadow(parent, asce, edat_level);
1499 1500 1501 1502 1503
	if (sg) {
		spin_unlock(&parent->shadow_lock);
		gmap_free(new);
		return sg;
	}
1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516
	if (asce & _ASCE_REAL_SPACE) {
		/* only allow one real-space gmap shadow */
		list_for_each_entry(sg, &parent->children, list) {
			if (sg->orig_asce & _ASCE_REAL_SPACE) {
				spin_lock(&sg->guest_table_lock);
				gmap_unshadow(sg);
				spin_unlock(&sg->guest_table_lock);
				list_del(&sg->list);
				gmap_put(sg);
				break;
			}
		}
	}
1517 1518
	atomic_set(&new->ref_count, 2);
	list_add(&new->list, &parent->children);
1519 1520 1521 1522 1523 1524
	if (asce & _ASCE_REAL_SPACE) {
		/* nothing to protect, return right away */
		new->initialized = true;
		spin_unlock(&parent->shadow_lock);
		return new;
	}
1525 1526
	spin_unlock(&parent->shadow_lock);
	/* protect after insertion, so it will get properly invalidated */
1527 1528
	down_read(&parent->mm->mmap_sem);
	rc = gmap_protect_range(parent, asce & _ASCE_ORIGIN,
1529
				((asce & _ASCE_TABLE_LENGTH) + 1) * PAGE_SIZE,
1530 1531
				PROT_READ, PGSTE_VSIE_BIT);
	up_read(&parent->mm->mmap_sem);
1532 1533
	spin_lock(&parent->shadow_lock);
	new->initialized = true;
1534
	if (rc) {
1535
		list_del(&new->list);
1536
		gmap_free(new);
1537 1538 1539 1540
		new = ERR_PTR(rc);
	}
	spin_unlock(&parent->shadow_lock);
	return new;
1541 1542 1543 1544 1545 1546 1547 1548
}
EXPORT_SYMBOL_GPL(gmap_shadow);

/**
 * gmap_shadow_r2t - create an empty shadow region 2 table
 * @sg: pointer to the shadow guest address space structure
 * @saddr: faulting address in the shadow gmap
 * @r2t: parent gmap address of the region 2 table to get shadowed
1549
 * @fake: r2t references contiguous guest memory block, not a r2t
1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561
 *
 * The r2t parameter specifies the address of the source table. The
 * four pages of the source table are made read-only in the parent gmap
 * address space. A write to the source table area @r2t will automatically
 * remove the shadow r2 table and all of its decendents.
 *
 * Returns 0 if successfully shadowed or already shadowed, -EAGAIN if the
 * shadow table structure is incomplete, -ENOMEM if out of memory and
 * -EFAULT if an address in the parent gmap could not be resolved.
 *
 * Called with sg->mm->mmap_sem in read.
 */
1562 1563
int gmap_shadow_r2t(struct gmap *sg, unsigned long saddr, unsigned long r2t,
		    int fake)
1564 1565 1566 1567 1568 1569 1570 1571
{
	unsigned long raddr, origin, offset, len;
	unsigned long *s_r2t, *table;
	struct page *page;
	int rc;

	BUG_ON(!gmap_is_shadow(sg));
	/* Allocate a shadow region second table */
1572
	page = alloc_pages(GFP_KERNEL, CRST_ALLOC_ORDER);
1573 1574 1575
	if (!page)
		return -ENOMEM;
	page->index = r2t & _REGION_ENTRY_ORIGIN;
1576 1577
	if (fake)
		page->index |= GMAP_SHADOW_FAKE_TABLE;
1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588
	s_r2t = (unsigned long *) page_to_phys(page);
	/* Install shadow region second table */
	spin_lock(&sg->guest_table_lock);
	table = gmap_table_walk(sg, saddr, 4); /* get region-1 pointer */
	if (!table) {
		rc = -EAGAIN;		/* Race with unshadow */
		goto out_free;
	}
	if (!(*table & _REGION_ENTRY_INVALID)) {
		rc = 0;			/* Already established */
		goto out_free;
1589 1590 1591
	} else if (*table & _REGION_ENTRY_ORIGIN) {
		rc = -EAGAIN;		/* Race with shadow */
		goto out_free;
1592 1593
	}
	crst_table_init(s_r2t, _REGION2_ENTRY_EMPTY);
1594 1595 1596
	/* mark as invalid as long as the parent table is not protected */
	*table = (unsigned long) s_r2t | _REGION_ENTRY_LENGTH |
		 _REGION_ENTRY_TYPE_R1 | _REGION_ENTRY_INVALID;
1597 1598
	if (sg->edat_level >= 1)
		*table |= (r2t & _REGION_ENTRY_PROTECT);
1599
	list_add(&page->lru, &sg->crst_list);
1600 1601 1602 1603 1604 1605
	if (fake) {
		/* nothing to protect for fake tables */
		*table &= ~_REGION_ENTRY_INVALID;
		spin_unlock(&sg->guest_table_lock);
		return 0;
	}
1606 1607
	spin_unlock(&sg->guest_table_lock);
	/* Make r2t read-only in parent gmap page table */
1608
	raddr = (saddr & _REGION1_MASK) | _SHADOW_RMAP_REGION1;
1609
	origin = r2t & _REGION_ENTRY_ORIGIN;
1610 1611
	offset = ((r2t & _REGION_ENTRY_OFFSET) >> 6) * PAGE_SIZE;
	len = ((r2t & _REGION_ENTRY_LENGTH) + 1) * PAGE_SIZE - offset;
1612
	rc = gmap_protect_rmap(sg, raddr, origin + offset, len, PROT_READ);
1613 1614 1615 1616 1617 1618 1619 1620 1621
	spin_lock(&sg->guest_table_lock);
	if (!rc) {
		table = gmap_table_walk(sg, saddr, 4);
		if (!table || (*table & _REGION_ENTRY_ORIGIN) !=
			      (unsigned long) s_r2t)
			rc = -EAGAIN;		/* Race with unshadow */
		else
			*table &= ~_REGION_ENTRY_INVALID;
	} else {
1622 1623
		gmap_unshadow_r2t(sg, raddr);
	}
1624
	spin_unlock(&sg->guest_table_lock);
1625 1626 1627
	return rc;
out_free:
	spin_unlock(&sg->guest_table_lock);
1628
	__free_pages(page, CRST_ALLOC_ORDER);
1629 1630
	return rc;
}
1631 1632 1633 1634 1635 1636 1637
EXPORT_SYMBOL_GPL(gmap_shadow_r2t);

/**
 * gmap_shadow_r3t - create a shadow region 3 table
 * @sg: pointer to the shadow guest address space structure
 * @saddr: faulting address in the shadow gmap
 * @r3t: parent gmap address of the region 3 table to get shadowed
1638
 * @fake: r3t references contiguous guest memory block, not a r3t
1639 1640 1641 1642 1643 1644 1645
 *
 * Returns 0 if successfully shadowed or already shadowed, -EAGAIN if the
 * shadow table structure is incomplete, -ENOMEM if out of memory and
 * -EFAULT if an address in the parent gmap could not be resolved.
 *
 * Called with sg->mm->mmap_sem in read.
 */
1646 1647
int gmap_shadow_r3t(struct gmap *sg, unsigned long saddr, unsigned long r3t,
		    int fake)
1648 1649 1650 1651 1652 1653 1654 1655
{
	unsigned long raddr, origin, offset, len;
	unsigned long *s_r3t, *table;
	struct page *page;
	int rc;

	BUG_ON(!gmap_is_shadow(sg));
	/* Allocate a shadow region second table */
1656
	page = alloc_pages(GFP_KERNEL, CRST_ALLOC_ORDER);
1657 1658 1659
	if (!page)
		return -ENOMEM;
	page->index = r3t & _REGION_ENTRY_ORIGIN;
1660 1661
	if (fake)
		page->index |= GMAP_SHADOW_FAKE_TABLE;
1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672
	s_r3t = (unsigned long *) page_to_phys(page);
	/* Install shadow region second table */
	spin_lock(&sg->guest_table_lock);
	table = gmap_table_walk(sg, saddr, 3); /* get region-2 pointer */
	if (!table) {
		rc = -EAGAIN;		/* Race with unshadow */
		goto out_free;
	}
	if (!(*table & _REGION_ENTRY_INVALID)) {
		rc = 0;			/* Already established */
		goto out_free;
1673 1674
	} else if (*table & _REGION_ENTRY_ORIGIN) {
		rc = -EAGAIN;		/* Race with shadow */
1675 1676
	}
	crst_table_init(s_r3t, _REGION3_ENTRY_EMPTY);
1677 1678 1679
	/* mark as invalid as long as the parent table is not protected */
	*table = (unsigned long) s_r3t | _REGION_ENTRY_LENGTH |
		 _REGION_ENTRY_TYPE_R2 | _REGION_ENTRY_INVALID;
1680 1681
	if (sg->edat_level >= 1)
		*table |= (r3t & _REGION_ENTRY_PROTECT);
1682
	list_add(&page->lru, &sg->crst_list);
1683 1684 1685 1686 1687 1688
	if (fake) {
		/* nothing to protect for fake tables */
		*table &= ~_REGION_ENTRY_INVALID;
		spin_unlock(&sg->guest_table_lock);
		return 0;
	}
1689 1690
	spin_unlock(&sg->guest_table_lock);
	/* Make r3t read-only in parent gmap page table */
1691
	raddr = (saddr & _REGION2_MASK) | _SHADOW_RMAP_REGION2;
1692
	origin = r3t & _REGION_ENTRY_ORIGIN;
1693 1694
	offset = ((r3t & _REGION_ENTRY_OFFSET) >> 6) * PAGE_SIZE;
	len = ((r3t & _REGION_ENTRY_LENGTH) + 1) * PAGE_SIZE - offset;
1695
	rc = gmap_protect_rmap(sg, raddr, origin + offset, len, PROT_READ);
1696 1697 1698 1699 1700 1701 1702 1703 1704
	spin_lock(&sg->guest_table_lock);
	if (!rc) {
		table = gmap_table_walk(sg, saddr, 3);
		if (!table || (*table & _REGION_ENTRY_ORIGIN) !=
			      (unsigned long) s_r3t)
			rc = -EAGAIN;		/* Race with unshadow */
		else
			*table &= ~_REGION_ENTRY_INVALID;
	} else {
1705 1706
		gmap_unshadow_r3t(sg, raddr);
	}
1707
	spin_unlock(&sg->guest_table_lock);
1708 1709 1710
	return rc;
out_free:
	spin_unlock(&sg->guest_table_lock);
1711
	__free_pages(page, CRST_ALLOC_ORDER);
1712 1713 1714 1715 1716 1717 1718 1719 1720
	return rc;
}
EXPORT_SYMBOL_GPL(gmap_shadow_r3t);

/**
 * gmap_shadow_sgt - create a shadow segment table
 * @sg: pointer to the shadow guest address space structure
 * @saddr: faulting address in the shadow gmap
 * @sgt: parent gmap address of the segment table to get shadowed
1721
 * @fake: sgt references contiguous guest memory block, not a sgt
1722 1723 1724 1725 1726 1727 1728
 *
 * Returns: 0 if successfully shadowed or already shadowed, -EAGAIN if the
 * shadow table structure is incomplete, -ENOMEM if out of memory and
 * -EFAULT if an address in the parent gmap could not be resolved.
 *
 * Called with sg->mm->mmap_sem in read.
 */
1729 1730
int gmap_shadow_sgt(struct gmap *sg, unsigned long saddr, unsigned long sgt,
		    int fake)
1731 1732 1733 1734 1735 1736
{
	unsigned long raddr, origin, offset, len;
	unsigned long *s_sgt, *table;
	struct page *page;
	int rc;

1737
	BUG_ON(!gmap_is_shadow(sg) || (sgt & _REGION3_ENTRY_LARGE));
1738
	/* Allocate a shadow segment table */
1739
	page = alloc_pages(GFP_KERNEL, CRST_ALLOC_ORDER);
1740 1741 1742
	if (!page)
		return -ENOMEM;
	page->index = sgt & _REGION_ENTRY_ORIGIN;
1743 1744
	if (fake)
		page->index |= GMAP_SHADOW_FAKE_TABLE;
1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755
	s_sgt = (unsigned long *) page_to_phys(page);
	/* Install shadow region second table */
	spin_lock(&sg->guest_table_lock);
	table = gmap_table_walk(sg, saddr, 2); /* get region-3 pointer */
	if (!table) {
		rc = -EAGAIN;		/* Race with unshadow */
		goto out_free;
	}
	if (!(*table & _REGION_ENTRY_INVALID)) {
		rc = 0;			/* Already established */
		goto out_free;
1756 1757 1758
	} else if (*table & _REGION_ENTRY_ORIGIN) {
		rc = -EAGAIN;		/* Race with shadow */
		goto out_free;
1759 1760
	}
	crst_table_init(s_sgt, _SEGMENT_ENTRY_EMPTY);
1761 1762 1763
	/* mark as invalid as long as the parent table is not protected */
	*table = (unsigned long) s_sgt | _REGION_ENTRY_LENGTH |
		 _REGION_ENTRY_TYPE_R3 | _REGION_ENTRY_INVALID;
1764 1765
	if (sg->edat_level >= 1)
		*table |= sgt & _REGION_ENTRY_PROTECT;
1766
	list_add(&page->lru, &sg->crst_list);
1767 1768 1769 1770 1771 1772
	if (fake) {
		/* nothing to protect for fake tables */
		*table &= ~_REGION_ENTRY_INVALID;
		spin_unlock(&sg->guest_table_lock);
		return 0;
	}
1773 1774
	spin_unlock(&sg->guest_table_lock);
	/* Make sgt read-only in parent gmap page table */
1775
	raddr = (saddr & _REGION3_MASK) | _SHADOW_RMAP_REGION3;
1776
	origin = sgt & _REGION_ENTRY_ORIGIN;
1777 1778
	offset = ((sgt & _REGION_ENTRY_OFFSET) >> 6) * PAGE_SIZE;
	len = ((sgt & _REGION_ENTRY_LENGTH) + 1) * PAGE_SIZE - offset;
1779
	rc = gmap_protect_rmap(sg, raddr, origin + offset, len, PROT_READ);
1780 1781 1782 1783 1784 1785 1786 1787 1788
	spin_lock(&sg->guest_table_lock);
	if (!rc) {
		table = gmap_table_walk(sg, saddr, 2);
		if (!table || (*table & _REGION_ENTRY_ORIGIN) !=
			      (unsigned long) s_sgt)
			rc = -EAGAIN;		/* Race with unshadow */
		else
			*table &= ~_REGION_ENTRY_INVALID;
	} else {
1789 1790
		gmap_unshadow_sgt(sg, raddr);
	}
1791
	spin_unlock(&sg->guest_table_lock);
1792 1793 1794
	return rc;
out_free:
	spin_unlock(&sg->guest_table_lock);
1795
	__free_pages(page, CRST_ALLOC_ORDER);
1796 1797 1798 1799 1800 1801 1802 1803 1804 1805
	return rc;
}
EXPORT_SYMBOL_GPL(gmap_shadow_sgt);

/**
 * gmap_shadow_lookup_pgtable - find a shadow page table
 * @sg: pointer to the shadow guest address space structure
 * @saddr: the address in the shadow aguest address space
 * @pgt: parent gmap address of the page table to get shadowed
 * @dat_protection: if the pgtable is marked as protected by dat
1806
 * @fake: pgt references contiguous guest memory block, not a pgtable
1807 1808 1809 1810 1811 1812 1813
 *
 * Returns 0 if the shadow page table was found and -EAGAIN if the page
 * table was not found.
 *
 * Called with sg->mm->mmap_sem in read.
 */
int gmap_shadow_pgt_lookup(struct gmap *sg, unsigned long saddr,
1814 1815
			   unsigned long *pgt, int *dat_protection,
			   int *fake)
1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826
{
	unsigned long *table;
	struct page *page;
	int rc;

	BUG_ON(!gmap_is_shadow(sg));
	spin_lock(&sg->guest_table_lock);
	table = gmap_table_walk(sg, saddr, 1); /* get segment pointer */
	if (table && !(*table & _SEGMENT_ENTRY_INVALID)) {
		/* Shadow page tables are full pages (pte+pgste) */
		page = pfn_to_page(*table >> PAGE_SHIFT);
1827
		*pgt = page->index & ~GMAP_SHADOW_FAKE_TABLE;
1828
		*dat_protection = !!(*table & _SEGMENT_ENTRY_PROTECT);
1829
		*fake = !!(page->index & GMAP_SHADOW_FAKE_TABLE);
1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844
		rc = 0;
	} else  {
		rc = -EAGAIN;
	}
	spin_unlock(&sg->guest_table_lock);
	return rc;

}
EXPORT_SYMBOL_GPL(gmap_shadow_pgt_lookup);

/**
 * gmap_shadow_pgt - instantiate a shadow page table
 * @sg: pointer to the shadow guest address space structure
 * @saddr: faulting address in the shadow gmap
 * @pgt: parent gmap address of the page table to get shadowed
1845
 * @fake: pgt references contiguous guest memory block, not a pgtable
1846 1847 1848 1849 1850 1851 1852
 *
 * Returns 0 if successfully shadowed or already shadowed, -EAGAIN if the
 * shadow table structure is incomplete, -ENOMEM if out of memory,
 * -EFAULT if an address in the parent gmap could not be resolved and
 *
 * Called with gmap->mm->mmap_sem in read
 */
1853 1854
int gmap_shadow_pgt(struct gmap *sg, unsigned long saddr, unsigned long pgt,
		    int fake)
1855 1856 1857 1858 1859 1860
{
	unsigned long raddr, origin;
	unsigned long *s_pgt, *table;
	struct page *page;
	int rc;

1861
	BUG_ON(!gmap_is_shadow(sg) || (pgt & _SEGMENT_ENTRY_LARGE));
1862 1863 1864 1865 1866
	/* Allocate a shadow page table */
	page = page_table_alloc_pgste(sg->mm);
	if (!page)
		return -ENOMEM;
	page->index = pgt & _SEGMENT_ENTRY_ORIGIN;
1867 1868
	if (fake)
		page->index |= GMAP_SHADOW_FAKE_TABLE;
1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879
	s_pgt = (unsigned long *) page_to_phys(page);
	/* Install shadow page table */
	spin_lock(&sg->guest_table_lock);
	table = gmap_table_walk(sg, saddr, 1); /* get segment pointer */
	if (!table) {
		rc = -EAGAIN;		/* Race with unshadow */
		goto out_free;
	}
	if (!(*table & _SEGMENT_ENTRY_INVALID)) {
		rc = 0;			/* Already established */
		goto out_free;
1880 1881 1882
	} else if (*table & _SEGMENT_ENTRY_ORIGIN) {
		rc = -EAGAIN;		/* Race with shadow */
		goto out_free;
1883
	}
1884
	/* mark as invalid as long as the parent table is not protected */
1885
	*table = (unsigned long) s_pgt | _SEGMENT_ENTRY |
1886
		 (pgt & _SEGMENT_ENTRY_PROTECT) | _SEGMENT_ENTRY_INVALID;
1887
	list_add(&page->lru, &sg->pt_list);
1888 1889 1890 1891 1892 1893
	if (fake) {
		/* nothing to protect for fake tables */
		*table &= ~_SEGMENT_ENTRY_INVALID;
		spin_unlock(&sg->guest_table_lock);
		return 0;
	}
1894 1895
	spin_unlock(&sg->guest_table_lock);
	/* Make pgt read-only in parent gmap page table (not the pgste) */
1896
	raddr = (saddr & _SEGMENT_MASK) | _SHADOW_RMAP_SEGMENT;
1897 1898
	origin = pgt & _SEGMENT_ENTRY_ORIGIN & PAGE_MASK;
	rc = gmap_protect_rmap(sg, raddr, origin, PAGE_SIZE, PROT_READ);
1899 1900 1901 1902 1903 1904 1905 1906 1907
	spin_lock(&sg->guest_table_lock);
	if (!rc) {
		table = gmap_table_walk(sg, saddr, 1);
		if (!table || (*table & _SEGMENT_ENTRY_ORIGIN) !=
			      (unsigned long) s_pgt)
			rc = -EAGAIN;		/* Race with unshadow */
		else
			*table &= ~_SEGMENT_ENTRY_INVALID;
	} else {
1908 1909
		gmap_unshadow_pgt(sg, raddr);
	}
1910
	spin_unlock(&sg->guest_table_lock);
1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923
	return rc;
out_free:
	spin_unlock(&sg->guest_table_lock);
	page_table_free_pgste(page);
	return rc;

}
EXPORT_SYMBOL_GPL(gmap_shadow_pgt);

/**
 * gmap_shadow_page - create a shadow page mapping
 * @sg: pointer to the shadow guest address space structure
 * @saddr: faulting address in the shadow gmap
1924
 * @pte: pte in parent gmap address space to get shadowed
1925 1926 1927 1928 1929 1930 1931
 *
 * Returns 0 if successfully shadowed or already shadowed, -EAGAIN if the
 * shadow table structure is incomplete, -ENOMEM if out of memory and
 * -EFAULT if an address in the parent gmap could not be resolved.
 *
 * Called with sg->mm->mmap_sem in read.
 */
1932
int gmap_shadow_page(struct gmap *sg, unsigned long saddr, pte_t pte)
1933 1934 1935
{
	struct gmap *parent;
	struct gmap_rmap *rmap;
1936
	unsigned long vmaddr, paddr;
1937 1938
	spinlock_t *ptl;
	pte_t *sptep, *tptep;
1939
	int prot;
1940 1941 1942 1943
	int rc;

	BUG_ON(!gmap_is_shadow(sg));
	parent = sg->parent;
1944
	prot = (pte_val(pte) & _PAGE_PROTECT) ? PROT_READ : PROT_WRITE;
1945 1946 1947 1948 1949 1950 1951

	rmap = kzalloc(sizeof(*rmap), GFP_KERNEL);
	if (!rmap)
		return -ENOMEM;
	rmap->raddr = (saddr & PAGE_MASK) | _SHADOW_RMAP_PGTABLE;

	while (1) {
1952
		paddr = pte_val(pte) & PAGE_MASK;
1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972
		vmaddr = __gmap_translate(parent, paddr);
		if (IS_ERR_VALUE(vmaddr)) {
			rc = vmaddr;
			break;
		}
		rc = radix_tree_preload(GFP_KERNEL);
		if (rc)
			break;
		rc = -EAGAIN;
		sptep = gmap_pte_op_walk(parent, paddr, &ptl);
		if (sptep) {
			spin_lock(&sg->guest_table_lock);
			/* Get page table pointer */
			tptep = (pte_t *) gmap_table_walk(sg, saddr, 0);
			if (!tptep) {
				spin_unlock(&sg->guest_table_lock);
				gmap_pte_op_end(ptl);
				radix_tree_preload_end();
				break;
			}
1973
			rc = ptep_shadow_pte(sg->mm, saddr, sptep, tptep, pte);
1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985
			if (rc > 0) {
				/* Success and a new mapping */
				gmap_insert_rmap(sg, vmaddr, rmap);
				rmap = NULL;
				rc = 0;
			}
			gmap_pte_op_end(ptl);
			spin_unlock(&sg->guest_table_lock);
		}
		radix_tree_preload_end();
		if (!rc)
			break;
1986
		rc = gmap_pte_op_fixup(parent, paddr, vmaddr, prot);
1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
		if (rc)
			break;
	}
	kfree(rmap);
	return rc;
}
EXPORT_SYMBOL_GPL(gmap_shadow_page);

/**
 * gmap_shadow_notify - handle notifications for shadow gmap
 *
 * Called with sg->parent->shadow_lock.
 */
static void gmap_shadow_notify(struct gmap *sg, unsigned long vmaddr,
2001
			       unsigned long gaddr)
2002 2003
{
	struct gmap_rmap *rmap, *rnext, *head;
2004
	unsigned long start, end, bits, raddr;
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

	BUG_ON(!gmap_is_shadow(sg));

	spin_lock(&sg->guest_table_lock);
	if (sg->removed) {
		spin_unlock(&sg->guest_table_lock);
		return;
	}
	/* Check for top level table */
	start = sg->orig_asce & _ASCE_ORIGIN;
2015
	end = start + ((sg->orig_asce & _ASCE_TABLE_LENGTH) + 1) * PAGE_SIZE;
2016 2017
	if (!(sg->orig_asce & _ASCE_REAL_SPACE) && gaddr >= start &&
	    gaddr < end) {
2018 2019 2020 2021 2022 2023 2024 2025
		/* The complete shadow table has to go */
		gmap_unshadow(sg);
		spin_unlock(&sg->guest_table_lock);
		list_del(&sg->list);
		gmap_put(sg);
		return;
	}
	/* Remove the page table tree from on specific entry */
2026
	head = radix_tree_delete(&sg->host_to_rmap, vmaddr >> PAGE_SHIFT);
2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050
	gmap_for_each_rmap_safe(rmap, rnext, head) {
		bits = rmap->raddr & _SHADOW_RMAP_MASK;
		raddr = rmap->raddr ^ bits;
		switch (bits) {
		case _SHADOW_RMAP_REGION1:
			gmap_unshadow_r2t(sg, raddr);
			break;
		case _SHADOW_RMAP_REGION2:
			gmap_unshadow_r3t(sg, raddr);
			break;
		case _SHADOW_RMAP_REGION3:
			gmap_unshadow_sgt(sg, raddr);
			break;
		case _SHADOW_RMAP_SEGMENT:
			gmap_unshadow_pgt(sg, raddr);
			break;
		case _SHADOW_RMAP_PGTABLE:
			gmap_unshadow_page(sg, raddr);
			break;
		}
		kfree(rmap);
	}
	spin_unlock(&sg->guest_table_lock);
}
2051 2052 2053 2054 2055 2056

/**
 * ptep_notify - call all invalidation callbacks for a specific pte.
 * @mm: pointer to the process mm_struct
 * @addr: virtual address in the process address space
 * @pte: pointer to the page table entry
2057
 * @bits: bits from the pgste that caused the notify call
2058 2059 2060 2061
 *
 * This function is assumed to be called with the page table lock held
 * for the pte to notify.
 */
2062 2063
void ptep_notify(struct mm_struct *mm, unsigned long vmaddr,
		 pte_t *pte, unsigned long bits)
2064
{
2065
	unsigned long offset, gaddr = 0;
2066
	unsigned long *table;
2067
	struct gmap *gmap, *sg, *next;
2068 2069

	offset = ((unsigned long) pte) & (255 * sizeof(pte_t));
2070
	offset = offset * (PAGE_SIZE / sizeof(pte_t));
2071 2072 2073
	rcu_read_lock();
	list_for_each_entry_rcu(gmap, &mm->context.gmap_list, list) {
		spin_lock(&gmap->guest_table_lock);
2074 2075
		table = radix_tree_lookup(&gmap->host_to_guest,
					  vmaddr >> PMD_SHIFT);
2076 2077 2078
		if (table)
			gaddr = __gmap_segment_gaddr(table) + offset;
		spin_unlock(&gmap->guest_table_lock);
2079 2080 2081 2082 2083 2084 2085
		if (!table)
			continue;

		if (!list_empty(&gmap->children) && (bits & PGSTE_VSIE_BIT)) {
			spin_lock(&gmap->shadow_lock);
			list_for_each_entry_safe(sg, next,
						 &gmap->children, list)
2086
				gmap_shadow_notify(sg, vmaddr, gaddr);
2087 2088 2089
			spin_unlock(&gmap->shadow_lock);
		}
		if (bits & PGSTE_IN_BIT)
2090
			gmap_call_notifier(gmap, gaddr, gaddr + PAGE_SIZE - 1);
2091
	}
2092
	rcu_read_unlock();
2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113
}
EXPORT_SYMBOL_GPL(ptep_notify);

static inline void thp_split_mm(struct mm_struct *mm)
{
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	struct vm_area_struct *vma;
	unsigned long addr;

	for (vma = mm->mmap; vma != NULL; vma = vma->vm_next) {
		for (addr = vma->vm_start;
		     addr < vma->vm_end;
		     addr += PAGE_SIZE)
			follow_page(vma, addr, FOLL_SPLIT);
		vma->vm_flags &= ~VM_HUGEPAGE;
		vma->vm_flags |= VM_NOHUGEPAGE;
	}
	mm->def_flags |= VM_NOHUGEPAGE;
#endif
}

2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144
/*
 * Remove all empty zero pages from the mapping for lazy refaulting
 * - This must be called after mm->context.has_pgste is set, to avoid
 *   future creation of zero pages
 * - This must be called after THP was enabled
 */
static int __zap_zero_pages(pmd_t *pmd, unsigned long start,
			   unsigned long end, struct mm_walk *walk)
{
	unsigned long addr;

	for (addr = start; addr != end; addr += PAGE_SIZE) {
		pte_t *ptep;
		spinlock_t *ptl;

		ptep = pte_offset_map_lock(walk->mm, pmd, addr, &ptl);
		if (is_zero_pfn(pte_pfn(*ptep)))
			ptep_xchg_direct(walk->mm, addr, ptep, __pte(_PAGE_INVALID));
		pte_unmap_unlock(ptep, ptl);
	}
	return 0;
}

static inline void zap_zero_pages(struct mm_struct *mm)
{
	struct mm_walk walk = { .pmd_entry = __zap_zero_pages };

	walk.mm = mm;
	walk_page_range(0, TASK_SIZE, &walk);
}

2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161
/*
 * switch on pgstes for its userspace process (for kvm)
 */
int s390_enable_sie(void)
{
	struct mm_struct *mm = current->mm;

	/* Do we have pgstes? if yes, we are done */
	if (mm_has_pgste(mm))
		return 0;
	/* Fail if the page tables are 2K */
	if (!mm_alloc_pgste(mm))
		return -EINVAL;
	down_write(&mm->mmap_sem);
	mm->context.has_pgste = 1;
	/* split thp mappings and disable thp for future mappings */
	thp_split_mm(mm);
2162
	zap_zero_pages(mm);
2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230
	up_write(&mm->mmap_sem);
	return 0;
}
EXPORT_SYMBOL_GPL(s390_enable_sie);

/*
 * Enable storage key handling from now on and initialize the storage
 * keys with the default key.
 */
static int __s390_enable_skey(pte_t *pte, unsigned long addr,
			      unsigned long next, struct mm_walk *walk)
{
	/* Clear storage key */
	ptep_zap_key(walk->mm, addr, pte);
	return 0;
}

int s390_enable_skey(void)
{
	struct mm_walk walk = { .pte_entry = __s390_enable_skey };
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma;
	int rc = 0;

	down_write(&mm->mmap_sem);
	if (mm_use_skey(mm))
		goto out_up;

	mm->context.use_skey = 1;
	for (vma = mm->mmap; vma; vma = vma->vm_next) {
		if (ksm_madvise(vma, vma->vm_start, vma->vm_end,
				MADV_UNMERGEABLE, &vma->vm_flags)) {
			mm->context.use_skey = 0;
			rc = -ENOMEM;
			goto out_up;
		}
	}
	mm->def_flags &= ~VM_MERGEABLE;

	walk.mm = mm;
	walk_page_range(0, TASK_SIZE, &walk);

out_up:
	up_write(&mm->mmap_sem);
	return rc;
}
EXPORT_SYMBOL_GPL(s390_enable_skey);

/*
 * Reset CMMA state, make all pages stable again.
 */
static int __s390_reset_cmma(pte_t *pte, unsigned long addr,
			     unsigned long next, struct mm_walk *walk)
{
	ptep_zap_unused(walk->mm, addr, pte, 1);
	return 0;
}

void s390_reset_cmma(struct mm_struct *mm)
{
	struct mm_walk walk = { .pte_entry = __s390_reset_cmma };

	down_write(&mm->mmap_sem);
	walk.mm = mm;
	walk_page_range(0, TASK_SIZE, &walk);
	up_write(&mm->mmap_sem);
}
EXPORT_SYMBOL_GPL(s390_reset_cmma);