slice.c 21.5 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
/*
 * address space "slices" (meta-segments) support
 *
 * Copyright (C) 2007 Benjamin Herrenschmidt, IBM Corporation.
 *
 * Based on hugetlb implementation
 *
 * Copyright (C) 2003 David Gibson, IBM Corporation.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */

#undef DEBUG

#include <linux/kernel.h>
#include <linux/mm.h>
#include <linux/pagemap.h>
#include <linux/err.h>
#include <linux/spinlock.h>
32
#include <linux/export.h>
33
#include <linux/hugetlb.h>
34 35
#include <asm/mman.h>
#include <asm/mmu.h>
36
#include <asm/copro.h>
37
#include <asm/hugetlb.h>
38

R
Roel Kluin 已提交
39
static DEFINE_SPINLOCK(slice_convert_lock);
40 41 42 43 44 45 46 47 48
/*
 * One bit per slice. We have lower slices which cover 256MB segments
 * upto 4G range. That gets us 16 low slices. For the rest we track slices
 * in 1TB size.
 */
struct slice_mask {
	u64 low_slices;
	DECLARE_BITMAP(high_slices, SLICE_NUM_HIGH);
};
49 50 51 52 53 54 55 56

#ifdef DEBUG
int _slice_debug = 1;

static void slice_print_mask(const char *label, struct slice_mask mask)
{
	if (!_slice_debug)
		return;
57 58
	pr_devel("%s low_slice: %*pbl\n", label, (int)SLICE_NUM_LOW, &mask.low_slices);
	pr_devel("%s high_slice: %*pbl\n", label, (int)SLICE_NUM_HIGH, mask.high_slices);
59 60
}

61
#define slice_dbg(fmt...) do { if (_slice_debug) pr_devel(fmt); } while (0)
62 63 64 65 66 67 68 69

#else

static void slice_print_mask(const char *label, struct slice_mask mask) {}
#define slice_dbg(fmt...)

#endif

70 71
static void slice_range_to_mask(unsigned long start, unsigned long len,
				struct slice_mask *ret)
72 73
{
	unsigned long end = start + len - 1;
74

75 76
	ret->low_slices = 0;
	bitmap_zero(ret->high_slices, SLICE_NUM_HIGH);
77 78

	if (start < SLICE_LOW_TOP) {
79
		unsigned long mend = min(end, (SLICE_LOW_TOP - 1));
80

81
		ret->low_slices = (1u << (GET_LOW_SLICE_INDEX(mend) + 1))
82
			- (1u << GET_LOW_SLICE_INDEX(start));
83 84
	}

85 86 87 88
	if ((start + len) > SLICE_LOW_TOP) {
		unsigned long start_index = GET_HIGH_SLICE_INDEX(start);
		unsigned long align_end = ALIGN(end, (1UL << SLICE_HIGH_SHIFT));
		unsigned long count = GET_HIGH_SLICE_INDEX(align_end) - start_index;
89

90
		bitmap_set(ret->high_slices, start_index, count);
91
	}
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124
}

static int slice_area_is_free(struct mm_struct *mm, unsigned long addr,
			      unsigned long len)
{
	struct vm_area_struct *vma;

	if ((mm->task_size - len) < addr)
		return 0;
	vma = find_vma(mm, addr);
	return (!vma || (addr + len) <= vma->vm_start);
}

static int slice_low_has_vma(struct mm_struct *mm, unsigned long slice)
{
	return !slice_area_is_free(mm, slice << SLICE_LOW_SHIFT,
				   1ul << SLICE_LOW_SHIFT);
}

static int slice_high_has_vma(struct mm_struct *mm, unsigned long slice)
{
	unsigned long start = slice << SLICE_HIGH_SHIFT;
	unsigned long end = start + (1ul << SLICE_HIGH_SHIFT);

	/* Hack, so that each addresses is controlled by exactly one
	 * of the high or low area bitmaps, the first high area starts
	 * at 4GB, not 0 */
	if (start == 0)
		start = SLICE_LOW_TOP;

	return !slice_area_is_free(mm, start, end - start);
}

125
static void slice_mask_for_free(struct mm_struct *mm, struct slice_mask *ret)
126 127 128
{
	unsigned long i;

129 130
	ret->low_slices = 0;
	bitmap_zero(ret->high_slices, SLICE_NUM_HIGH);
131

132 133
	for (i = 0; i < SLICE_NUM_LOW; i++)
		if (!slice_low_has_vma(mm, i))
134
			ret->low_slices |= 1u << i;
135 136

	if (mm->task_size <= SLICE_LOW_TOP)
137
		return;
138

139
	for (i = 0; i < GET_HIGH_SLICE_INDEX(mm->context.addr_limit); i++)
140
		if (!slice_high_has_vma(mm, i))
141
			__set_bit(i, ret->high_slices);
142 143
}

144
static void slice_mask_for_size(struct mm_struct *mm, int psize, struct slice_mask *ret)
145
{
146 147
	unsigned char *hpsizes;
	int index, mask_index;
148
	unsigned long i;
149
	u64 lpsizes;
150

151 152
	ret->low_slices = 0;
	bitmap_zero(ret->high_slices, SLICE_NUM_HIGH);
153

154
	lpsizes = mm->context.low_slices_psize;
155
	for (i = 0; i < SLICE_NUM_LOW; i++)
156
		if (((lpsizes >> (i * 4)) & 0xf) == psize)
157
			ret->low_slices |= 1u << i;
158

159
	hpsizes = mm->context.high_slices_psize;
160
	for (i = 0; i < GET_HIGH_SLICE_INDEX(mm->context.addr_limit); i++) {
161 162 163
		mask_index = i & 0x1;
		index = i >> 1;
		if (((hpsizes[index] >> (mask_index * 4)) & 0xf) == psize)
164
			__set_bit(i, ret->high_slices);
165
	}
166 167
}

168 169
static int slice_check_fit(struct mm_struct *mm,
			   struct slice_mask mask, struct slice_mask available)
170
{
171
	DECLARE_BITMAP(result, SLICE_NUM_HIGH);
172
	unsigned long slice_count = GET_HIGH_SLICE_INDEX(mm->context.addr_limit);
173 174

	bitmap_and(result, mask.high_slices,
175
		   available.high_slices, slice_count);
176

177
	return (mask.low_slices & available.low_slices) == mask.low_slices &&
178
		bitmap_equal(result, mask.high_slices, slice_count);
179 180 181 182 183 184 185 186 187 188
}

static void slice_flush_segments(void *parm)
{
	struct mm_struct *mm = parm;
	unsigned long flags;

	if (mm != current->active_mm)
		return;

189
	copy_mm_to_paca(current->active_mm);
190 191 192 193 194 195 196 197

	local_irq_save(flags);
	slb_flush_and_rebolt();
	local_irq_restore(flags);
}

static void slice_convert(struct mm_struct *mm, struct slice_mask mask, int psize)
{
198
	int index, mask_index;
199
	/* Write the new slice psize bits */
200 201
	unsigned char *hpsizes;
	u64 lpsizes;
202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217
	unsigned long i, flags;

	slice_dbg("slice_convert(mm=%p, psize=%d)\n", mm, psize);
	slice_print_mask(" mask", mask);

	/* We need to use a spinlock here to protect against
	 * concurrent 64k -> 4k demotion ...
	 */
	spin_lock_irqsave(&slice_convert_lock, flags);

	lpsizes = mm->context.low_slices_psize;
	for (i = 0; i < SLICE_NUM_LOW; i++)
		if (mask.low_slices & (1u << i))
			lpsizes = (lpsizes & ~(0xful << (i * 4))) |
				(((unsigned long)psize) << (i * 4));

218
	/* Assign the value back */
219
	mm->context.low_slices_psize = lpsizes;
220 221

	hpsizes = mm->context.high_slices_psize;
222
	for (i = 0; i < GET_HIGH_SLICE_INDEX(mm->context.addr_limit); i++) {
223 224
		mask_index = i & 0x1;
		index = i >> 1;
225
		if (test_bit(i, mask.high_slices))
226 227 228 229
			hpsizes[index] = (hpsizes[index] &
					  ~(0xf << (mask_index * 4))) |
				(((unsigned long)psize) << (mask_index * 4));
	}
230 231

	slice_dbg(" lsps=%lx, hsps=%lx\n",
232 233
		  (unsigned long)mm->context.low_slices_psize,
		  (unsigned long)mm->context.high_slices_psize);
234 235 236

	spin_unlock_irqrestore(&slice_convert_lock, flags);

237
	copro_flush_all_slbs(mm);
238 239
}

240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260
/*
 * Compute which slice addr is part of;
 * set *boundary_addr to the start or end boundary of that slice
 * (depending on 'end' parameter);
 * return boolean indicating if the slice is marked as available in the
 * 'available' slice_mark.
 */
static bool slice_scan_available(unsigned long addr,
				 struct slice_mask available,
				 int end,
				 unsigned long *boundary_addr)
{
	unsigned long slice;
	if (addr < SLICE_LOW_TOP) {
		slice = GET_LOW_SLICE_INDEX(addr);
		*boundary_addr = (slice + end) << SLICE_LOW_SHIFT;
		return !!(available.low_slices & (1u << slice));
	} else {
		slice = GET_HIGH_SLICE_INDEX(addr);
		*boundary_addr = (slice + end) ?
			((slice + end) << SLICE_HIGH_SHIFT) : SLICE_LOW_TOP;
261
		return !!test_bit(slice, available.high_slices);
262 263 264
	}
}

265 266 267
static unsigned long slice_find_area_bottomup(struct mm_struct *mm,
					      unsigned long len,
					      struct slice_mask available,
268
					      int psize, unsigned long high_limit)
269 270
{
	int pshift = max_t(int, mmu_psize_defs[psize].shift, PAGE_SHIFT);
271 272
	unsigned long addr, found, next_end;
	struct vm_unmapped_area_info info;
273

274 275 276 277
	info.flags = 0;
	info.length = len;
	info.align_mask = PAGE_MASK & ((1ul << pshift) - 1);
	info.align_offset = 0;
278

279
	addr = TASK_UNMAPPED_BASE;
280 281 282 283
	/*
	 * Check till the allow max value for this mmap request
	 */
	while (addr < high_limit) {
284 285
		info.low_limit = addr;
		if (!slice_scan_available(addr, available, 1, &addr))
286
			continue;
287 288 289 290 291 292 293 294

 next_slice:
		/*
		 * At this point [info.low_limit; addr) covers
		 * available slices only and ends at a slice boundary.
		 * Check if we need to reduce the range, or if we can
		 * extend it to cover the next available slice.
		 */
295 296
		if (addr >= high_limit)
			addr = high_limit;
297 298 299
		else if (slice_scan_available(addr, available, 1, &next_end)) {
			addr = next_end;
			goto next_slice;
300
		}
301 302 303 304 305
		info.high_limit = addr;

		found = vm_unmapped_area(&info);
		if (!(found & ~PAGE_MASK))
			return found;
306 307 308 309 310 311 312 313
	}

	return -ENOMEM;
}

static unsigned long slice_find_area_topdown(struct mm_struct *mm,
					     unsigned long len,
					     struct slice_mask available,
314
					     int psize, unsigned long high_limit)
315 316
{
	int pshift = max_t(int, mmu_psize_defs[psize].shift, PAGE_SHIFT);
317 318 319 320 321 322 323
	unsigned long addr, found, prev;
	struct vm_unmapped_area_info info;

	info.flags = VM_UNMAPPED_AREA_TOPDOWN;
	info.length = len;
	info.align_mask = PAGE_MASK & ((1ul << pshift) - 1);
	info.align_offset = 0;
324 325

	addr = mm->mmap_base;
326 327 328 329 330 331 332 333 334
	/*
	 * If we are trying to allocate above DEFAULT_MAP_WINDOW
	 * Add the different to the mmap_base.
	 * Only for that request for which high_limit is above
	 * DEFAULT_MAP_WINDOW we should apply this.
	 */
	if (high_limit  > DEFAULT_MAP_WINDOW)
		addr += mm->context.addr_limit - DEFAULT_MAP_WINDOW;

335 336 337
	while (addr > PAGE_SIZE) {
		info.high_limit = addr;
		if (!slice_scan_available(addr - 1, available, 0, &addr))
338 339
			continue;

340
 prev_slice:
341
		/*
342 343 344 345
		 * At this point [addr; info.high_limit) covers
		 * available slices only and starts at a slice boundary.
		 * Check if we need to reduce the range, or if we can
		 * extend it to cover the previous available slice.
346
		 */
347 348 349 350 351 352 353
		if (addr < PAGE_SIZE)
			addr = PAGE_SIZE;
		else if (slice_scan_available(addr - 1, available, 0, &prev)) {
			addr = prev;
			goto prev_slice;
		}
		info.low_limit = addr;
354

355 356 357
		found = vm_unmapped_area(&info);
		if (!(found & ~PAGE_MASK))
			return found;
358 359 360 361 362 363 364 365
	}

	/*
	 * A failed mmap() very likely causes application failure,
	 * so fall back to the bottom-up function here. This scenario
	 * can happen with large stack limits and large mmap()
	 * allocations.
	 */
366
	return slice_find_area_bottomup(mm, len, available, psize, high_limit);
367 368 369 370 371
}


static unsigned long slice_find_area(struct mm_struct *mm, unsigned long len,
				     struct slice_mask mask, int psize,
372
				     int topdown, unsigned long high_limit)
373 374
{
	if (topdown)
375
		return slice_find_area_topdown(mm, len, mask, psize, high_limit);
376
	else
377
		return slice_find_area_bottomup(mm, len, mask, psize, high_limit);
378 379
}

380 381 382
static inline void slice_or_mask(struct slice_mask *dst, struct slice_mask *src)
{
	DECLARE_BITMAP(result, SLICE_NUM_HIGH);
383

384 385 386 387 388 389 390 391 392 393 394 395 396 397
	dst->low_slices |= src->low_slices;
	bitmap_or(result, dst->high_slices, src->high_slices, SLICE_NUM_HIGH);
	bitmap_copy(dst->high_slices, result, SLICE_NUM_HIGH);
}

static inline void slice_andnot_mask(struct slice_mask *dst, struct slice_mask *src)
{
	DECLARE_BITMAP(result, SLICE_NUM_HIGH);

	dst->low_slices &= ~src->low_slices;

	bitmap_andnot(result, dst->high_slices, src->high_slices, SLICE_NUM_HIGH);
	bitmap_copy(dst->high_slices, result, SLICE_NUM_HIGH);
}
398 399 400 401 402 403 404

#ifdef CONFIG_PPC_64K_PAGES
#define MMU_PAGE_BASE	MMU_PAGE_64K
#else
#define MMU_PAGE_BASE	MMU_PAGE_4K
#endif

405 406
unsigned long slice_get_unmapped_area(unsigned long addr, unsigned long len,
				      unsigned long flags, unsigned int psize,
407
				      int topdown)
408
{
409
	struct slice_mask mask;
410
	struct slice_mask good_mask;
411 412
	struct slice_mask potential_mask;
	struct slice_mask compat_mask;
413 414 415
	int fixed = (flags & MAP_FIXED);
	int pshift = max_t(int, mmu_psize_defs[psize].shift, PAGE_SHIFT);
	struct mm_struct *mm = current->mm;
416
	unsigned long newaddr;
417
	unsigned long high_limit;
418

419 420 421 422 423 424 425 426 427 428 429 430 431 432
	/*
	 * Check if we need to expland slice area.
	 */
	if (unlikely(addr > mm->context.addr_limit && addr < TASK_SIZE)) {
		mm->context.addr_limit = TASK_SIZE;
		on_each_cpu(slice_flush_segments, mm, 1);
	}
	/*
	 * This mmap request can allocate upt to 512TB
	 */
	if (addr > DEFAULT_MAP_WINDOW)
		high_limit = mm->context.addr_limit;
	else
		high_limit = DEFAULT_MAP_WINDOW;
433 434 435 436 437 438 439 440 441 442 443 444 445
	/*
	 * init different masks
	 */
	mask.low_slices = 0;
	bitmap_zero(mask.high_slices, SLICE_NUM_HIGH);

	/* silence stupid warning */;
	potential_mask.low_slices = 0;
	bitmap_zero(potential_mask.high_slices, SLICE_NUM_HIGH);

	compat_mask.low_slices = 0;
	bitmap_zero(compat_mask.high_slices, SLICE_NUM_HIGH);

446 447
	/* Sanity checks */
	BUG_ON(mm->task_size == 0);
448
	VM_BUG_ON(radix_enabled());
449 450

	slice_dbg("slice_get_unmapped_area(mm=%p, psize=%d...\n", mm, psize);
451 452
	slice_dbg(" addr=%lx, len=%lx, flags=%lx, topdown=%d\n",
		  addr, len, flags, topdown);
453 454 455

	if (len > mm->task_size)
		return -ENOMEM;
456 457
	if (len & ((1ul << pshift) - 1))
		return -EINVAL;
458 459 460
	if (fixed && (addr & ((1ul << pshift) - 1)))
		return -EINVAL;
	if (fixed && addr > (mm->task_size - len))
461
		return -ENOMEM;
462 463 464 465 466

	/* If hint, make sure it matches our alignment restrictions */
	if (!fixed && addr) {
		addr = _ALIGN_UP(addr, 1ul << pshift);
		slice_dbg(" aligned addr=%lx\n", addr);
467 468 469 470
		/* Ignore hint if it's too large or overlaps a VMA */
		if (addr > mm->task_size - len ||
		    !slice_area_is_free(mm, addr, len))
			addr = 0;
471 472
	}

473
	/* First make up a "good" mask of slices that have the right size
474 475
	 * already
	 */
476
	slice_mask_for_size(mm, psize, &good_mask);
477 478
	slice_print_mask(" good_mask", good_mask);

479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496
	/*
	 * Here "good" means slices that are already the right page size,
	 * "compat" means slices that have a compatible page size (i.e.
	 * 4k in a 64k pagesize kernel), and "free" means slices without
	 * any VMAs.
	 *
	 * If MAP_FIXED:
	 *	check if fits in good | compat => OK
	 *	check if fits in good | compat | free => convert free
	 *	else bad
	 * If have hint:
	 *	check if hint fits in good => OK
	 *	check if hint fits in good | free => convert free
	 * Otherwise:
	 *	search in good, found => OK
	 *	search in good | free, found => convert free
	 *	search in good | compat | free, found => convert free.
	 */
497

498 499 500
#ifdef CONFIG_PPC_64K_PAGES
	/* If we support combo pages, we can allow 64k pages in 4k slices */
	if (psize == MMU_PAGE_64K) {
501
		slice_mask_for_size(mm, MMU_PAGE_4K, &compat_mask);
502
		if (fixed)
503
			slice_or_mask(&good_mask, &compat_mask);
504 505
	}
#endif
506

507 508
	/* First check hint if it's valid or if we have MAP_FIXED */
	if (addr != 0 || fixed) {
509
		/* Build a mask for the requested range */
510
		slice_range_to_mask(addr, len, &mask);
511 512 513 514 515
		slice_print_mask(" mask", mask);

		/* Check if we fit in the good mask. If we do, we just return,
		 * nothing else to do
		 */
516
		if (slice_check_fit(mm, mask, good_mask)) {
517 518 519
			slice_dbg(" fits good !\n");
			return addr;
		}
520 521 522
	} else {
		/* Now let's see if we can find something in the existing
		 * slices for that size
523
		 */
524 525
		newaddr = slice_find_area(mm, len, good_mask,
					  psize, topdown, high_limit);
526 527 528 529 530 531
		if (newaddr != -ENOMEM) {
			/* Found within the good mask, we don't have to setup,
			 * we thus return directly
			 */
			slice_dbg(" found area at 0x%lx\n", newaddr);
			return newaddr;
532 533 534
		}
	}

535 536 537
	/* We don't fit in the good mask, check what other slices are
	 * empty and thus can be converted
	 */
538
	slice_mask_for_free(mm, &potential_mask);
539
	slice_or_mask(&potential_mask, &good_mask);
540 541
	slice_print_mask(" potential", potential_mask);

542
	if ((addr != 0 || fixed) && slice_check_fit(mm, mask, potential_mask)) {
543 544 545 546 547
		slice_dbg(" fits potential !\n");
		goto convert;
	}

	/* If we have MAP_FIXED and failed the above steps, then error out */
548 549 550 551 552
	if (fixed)
		return -EBUSY;

	slice_dbg(" search...\n");

553 554
	/* If we had a hint that didn't work out, see if we can fit
	 * anywhere in the good area.
555
	 */
556
	if (addr) {
557 558
		addr = slice_find_area(mm, len, good_mask,
				       psize, topdown, high_limit);
559 560 561 562
		if (addr != -ENOMEM) {
			slice_dbg(" found area at 0x%lx\n", addr);
			return addr;
		}
563 564 565
	}

	/* Now let's see if we can find something in the existing slices
566
	 * for that size plus free slices
567
	 */
568 569
	addr = slice_find_area(mm, len, potential_mask,
			       psize, topdown, high_limit);
570 571 572 573

#ifdef CONFIG_PPC_64K_PAGES
	if (addr == -ENOMEM && psize == MMU_PAGE_64K) {
		/* retry the search with 4k-page slices included */
574
		slice_or_mask(&potential_mask, &compat_mask);
575 576
		addr = slice_find_area(mm, len, potential_mask,
				       psize, topdown, high_limit);
577 578 579
	}
#endif

580 581 582
	if (addr == -ENOMEM)
		return -ENOMEM;

583
	slice_range_to_mask(addr, len, &mask);
584 585 586 587
	slice_dbg(" found potential area at 0x%lx\n", addr);
	slice_print_mask(" mask", mask);

 convert:
588 589 590
	slice_andnot_mask(&mask, &good_mask);
	slice_andnot_mask(&mask, &compat_mask);
	if (mask.low_slices || !bitmap_empty(mask.high_slices, SLICE_NUM_HIGH)) {
591 592
		slice_convert(mm, mask, psize);
		if (psize > MMU_PAGE_BASE)
593
			on_each_cpu(slice_flush_segments, mm, 1);
594
	}
595 596 597 598 599 600 601 602 603 604 605 606
	return addr;

}
EXPORT_SYMBOL_GPL(slice_get_unmapped_area);

unsigned long arch_get_unmapped_area(struct file *filp,
				     unsigned long addr,
				     unsigned long len,
				     unsigned long pgoff,
				     unsigned long flags)
{
	return slice_get_unmapped_area(addr, len, flags,
607
				       current->mm->context.user_psize, 0);
608 609 610 611 612 613 614 615 616
}

unsigned long arch_get_unmapped_area_topdown(struct file *filp,
					     const unsigned long addr0,
					     const unsigned long len,
					     const unsigned long pgoff,
					     const unsigned long flags)
{
	return slice_get_unmapped_area(addr0, len, flags,
617
				       current->mm->context.user_psize, 1);
618 619 620 621
}

unsigned int get_slice_psize(struct mm_struct *mm, unsigned long addr)
{
622 623
	unsigned char *hpsizes;
	int index, mask_index;
624

625 626 627 628 629 630 631 632 633 634
	/*
	 * Radix doesn't use slice, but can get enabled along with MMU_SLICE
	 */
	if (radix_enabled()) {
#ifdef CONFIG_PPC_64K_PAGES
		return MMU_PAGE_64K;
#else
		return MMU_PAGE_4K;
#endif
	}
635
	if (addr < SLICE_LOW_TOP) {
636 637
		u64 lpsizes;
		lpsizes = mm->context.low_slices_psize;
638
		index = GET_LOW_SLICE_INDEX(addr);
639
		return (lpsizes >> (index * 4)) & 0xf;
640
	}
641 642 643 644
	hpsizes = mm->context.high_slices_psize;
	index = GET_HIGH_SLICE_INDEX(addr);
	mask_index = index & 0x1;
	return (hpsizes[index >> 1] >> (mask_index * 4)) & 0xf;
645 646 647 648 649 650 651 652 653 654 655
}
EXPORT_SYMBOL_GPL(get_slice_psize);

/*
 * This is called by hash_page when it needs to do a lazy conversion of
 * an address space from real 64K pages to combo 4K pages (typically
 * when hitting a non cacheable mapping on a processor or hypervisor
 * that won't allow them for 64K pages).
 *
 * This is also called in init_new_context() to change back the user
 * psize from whatever the parent context had it set to
656
 * N.B. This may be called before mm->context.id has been set.
657 658 659 660 661 662 663
 *
 * This function will only change the content of the {low,high)_slice_psize
 * masks, it will not flush SLBs as this shall be handled lazily by the
 * caller.
 */
void slice_set_user_psize(struct mm_struct *mm, unsigned int psize)
{
664 665 666
	int index, mask_index;
	unsigned char *hpsizes;
	unsigned long flags, lpsizes;
667 668 669 670 671
	unsigned int old_psize;
	int i;

	slice_dbg("slice_set_user_psize(mm=%p, psize=%d)\n", mm, psize);

672
	VM_BUG_ON(radix_enabled());
673 674 675 676 677 678 679 680 681 682 683 684 685 686 687
	spin_lock_irqsave(&slice_convert_lock, flags);

	old_psize = mm->context.user_psize;
	slice_dbg(" old_psize=%d\n", old_psize);
	if (old_psize == psize)
		goto bail;

	mm->context.user_psize = psize;
	wmb();

	lpsizes = mm->context.low_slices_psize;
	for (i = 0; i < SLICE_NUM_LOW; i++)
		if (((lpsizes >> (i * 4)) & 0xf) == old_psize)
			lpsizes = (lpsizes & ~(0xful << (i * 4))) |
				(((unsigned long)psize) << (i * 4));
688 689
	/* Assign the value back */
	mm->context.low_slices_psize = lpsizes;
690 691

	hpsizes = mm->context.high_slices_psize;
692 693 694 695 696 697 698 699 700 701
	for (i = 0; i < SLICE_NUM_HIGH; i++) {
		mask_index = i & 0x1;
		index = i >> 1;
		if (((hpsizes[index] >> (mask_index * 4)) & 0xf) == old_psize)
			hpsizes[index] = (hpsizes[index] &
					  ~(0xf << (mask_index * 4))) |
				(((unsigned long)psize) << (mask_index * 4));
	}


702 703 704


	slice_dbg(" lsps=%lx, hsps=%lx\n",
705 706
		  (unsigned long)mm->context.low_slices_psize,
		  (unsigned long)mm->context.high_slices_psize);
707 708 709 710 711

 bail:
	spin_unlock_irqrestore(&slice_convert_lock, flags);
}

712 713 714
void slice_set_range_psize(struct mm_struct *mm, unsigned long start,
			   unsigned long len, unsigned int psize)
{
715
	struct slice_mask mask;
716

717
	VM_BUG_ON(radix_enabled());
718 719

	slice_range_to_mask(start, len, &mask);
720 721 722
	slice_convert(mm, mask, psize);
}

723
#ifdef CONFIG_HUGETLB_PAGE
724
/*
725
 * is_hugepage_only_range() is used by generic code to verify whether
726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746
 * a normal mmap mapping (non hugetlbfs) is valid on a given area.
 *
 * until the generic code provides a more generic hook and/or starts
 * calling arch get_unmapped_area for MAP_FIXED (which our implementation
 * here knows how to deal with), we hijack it to keep standard mappings
 * away from us.
 *
 * because of that generic code limitation, MAP_FIXED mapping cannot
 * "convert" back a slice with no VMAs to the standard page size, only
 * get_unmapped_area() can. It would be possible to fix it here but I
 * prefer working on fixing the generic code instead.
 *
 * WARNING: This will not work if hugetlbfs isn't enabled since the
 * generic code will redefine that function as 0 in that. This is ok
 * for now as we only use slices with hugetlbfs enabled. This should
 * be fixed as the generic code gets fixed.
 */
int is_hugepage_only_range(struct mm_struct *mm, unsigned long addr,
			   unsigned long len)
{
	struct slice_mask mask, available;
747
	unsigned int psize = mm->context.user_psize;
748

749 750 751
	if (radix_enabled())
		return 0;

752 753
	slice_range_to_mask(addr, len, &mask);
	slice_mask_for_size(mm, psize, &available);
754 755 756 757
#ifdef CONFIG_PPC_64K_PAGES
	/* We need to account for 4k slices too */
	if (psize == MMU_PAGE_64K) {
		struct slice_mask compat_mask;
758
		slice_mask_for_size(mm, MMU_PAGE_4K, &compat_mask);
759
		slice_or_mask(&available, &compat_mask);
760 761
	}
#endif
762 763 764 765 766 767 768

#if 0 /* too verbose */
	slice_dbg("is_hugepage_only_range(mm=%p, addr=%lx, len=%lx)\n",
		 mm, addr, len);
	slice_print_mask(" mask", mask);
	slice_print_mask(" available", available);
#endif
769
	return !slice_check_fit(mm, mask, available);
770
}
771
#endif