aboutsummaryrefslogtreecommitdiff
path: root/arch/powerpc/mm/slice.c
blob: f833dba2a0287ea06218f180620b8634300a58e7 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
/*
 * 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>
#include <linux/module.h>
#include <asm/mman.h>
#include <asm/mmu.h>
#include <asm/spu.h>

static spinlock_t slice_convert_lock = SPIN_LOCK_UNLOCKED;


#ifdef DEBUG
int _slice_debug = 1;

static void slice_print_mask(const char *label, struct slice_mask mask)
{
	char	*p, buf[16 + 3 + 16 + 1];
	int	i;

	if (!_slice_debug)
		return;
	p = buf;
	for (i = 0; i < SLICE_NUM_LOW; i++)
		*(p++) = (mask.low_slices & (1 << i)) ? '1' : '0';
	*(p++) = ' ';
	*(p++) = '-';
	*(p++) = ' ';
	for (i = 0; i < SLICE_NUM_HIGH; i++)
		*(p++) = (mask.high_slices & (1 << i)) ? '1' : '0';
	*(p++) = 0;

	printk(KERN_DEBUG "%s:%s\n", label, buf);
}

#define slice_dbg(fmt...) do { if (_slice_debug) pr_debug(fmt); } while(0)

#else

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

#endif

static struct slice_mask slice_range_to_mask(unsigned long start,
					     unsigned long len)
{
	unsigned long end = start + len - 1;
	struct slice_mask ret = { 0, 0 };

	if (start < SLICE_LOW_TOP) {
		unsigned long mend = min(end, SLICE_LOW_TOP);
		unsigned long mstart = min(start, SLICE_LOW_TOP);

		ret.low_slices = (1u << (GET_LOW_SLICE_INDEX(mend) + 1))
			- (1u << GET_LOW_SLICE_INDEX(mstart));
	}

	if ((start + len) > SLICE_LOW_TOP)
		ret.high_slices = (1u << (GET_HIGH_SLICE_INDEX(end) + 1))
			- (1u << GET_HIGH_SLICE_INDEX(start));

	return ret;
}

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);
}

static struct slice_mask slice_mask_for_free(struct mm_struct *mm)
{
	struct slice_mask ret = { 0, 0 };
	unsigned long i;

	for (i = 0; i < SLICE_NUM_LOW; i++)
		if (!slice_low_has_vma(mm, i))
			ret.low_slices |= 1u << i;

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

	for (i = 0; i < SLICE_NUM_HIGH; i++)
		if (!slice_high_has_vma(mm, i))
			ret.high_slices |= 1u << i;

	return ret;
}

static struct slice_mask slice_mask_for_size(struct mm_struct *mm, int psize)
{
	struct slice_mask ret = { 0, 0 };
	unsigned long i;
	u64 psizes;

	psizes = mm->context.low_slices_psize;
	for (i = 0; i < SLICE_NUM_LOW; i++)
		if (((psizes >> (i * 4)) & 0xf) == psize)
			ret.low_slices |= 1u << i;

	psizes = mm->context.high_slices_psize;
	for (i = 0; i < SLICE_NUM_HIGH; i++)
		if (((psizes >> (i * 4)) & 0xf) == psize)
			ret.high_slices |= 1u << i;

	return ret;
}

static int slice_check_fit(struct slice_mask mask, struct slice_mask available)
{
	return (mask.low_slices & available.low_slices) == mask.low_slices &&
		(mask.high_slices & available.high_slices) == mask.high_slices;
}

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

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

	/* update the paca copy of the context struct */
	get_paca()->context = current->active_mm->context;

	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)
{
	/* Write the new slice psize bits */
	u64 lpsizes, hpsizes;
	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));

	hpsizes = mm->context.high_slices_psize;
	for (i = 0; i < SLICE_NUM_HIGH; i++)
		if (mask.high_slices & (1u << i))
			hpsizes = (hpsizes & ~(0xful << (i * 4))) |
				(((unsigned long)psize) << (i * 4));

	mm->context.low_slices_psize = lpsizes;
	mm->context.high_slices_psize = hpsizes;

	slice_dbg(" lsps=%lx, hsps=%lx\n",
		  mm->context.low_slices_psize,
		  mm->context.high_slices_psize);

	spin_unlock_irqrestore(&slice_convert_lock, flags);
	mb();

	/* XXX this is sub-optimal but will do for now */
	on_each_cpu(slice_flush_segments, mm, 0, 1);
#ifdef CONFIG_SPU_BASE
	spu_flush_all_slbs(mm);
#endif
}

static unsigned long slice_find_area_bottomup(struct mm_struct *mm,
					      unsigned long len,
					      struct slice_mask available,
					      int psize, int use_cache)
{
	struct vm_area_struct *vma;
	unsigned long start_addr, addr;
	struct slice_mask mask;
	int pshift = max_t(int, mmu_psize_defs[psize].shift, PAGE_SHIFT);

	if (use_cache) {
		if (len <= mm->cached_hole_size) {
			start_addr = addr = TASK_UNMAPPED_BASE;
			mm->cached_hole_size = 0;
		} else
			start_addr = addr = mm->free_area_cache;
	} else
		start_addr = addr = TASK_UNMAPPED_BASE;

full_search:
	for (;;) {
		addr = _ALIGN_UP(addr, 1ul << pshift);
		if ((TASK_SIZE - len) < addr)
			break;
		vma = find_vma(mm, addr);
		BUG_ON(vma && (addr >= vma->vm_end));

		mask = slice_range_to_mask(addr, len);
		if (!slice_check_fit(mask, available)) {
			if (addr < SLICE_LOW_TOP)
				addr = _ALIGN_UP(addr + 1,  1ul << SLICE_LOW_SHIFT);
			else
				addr = _ALIGN_UP(addr + 1,  1ul << SLICE_HIGH_SHIFT);
			continue;
		}
		if (!vma || addr + len <= vma->vm_start) {
			/*
			 * Remember the place where we stopped the search:
			 */
			if (use_cache)
				mm->free_area_cache = addr + len;
			return addr;
		}
		if (use_cache && (addr + mm->cached_hole_size) < vma->vm_start)
		        mm->cached_hole_size = vma->vm_start - addr;
		addr = vma->vm_end;
	}

	/* Make sure we didn't miss any holes */
	if (use_cache && start_addr != TASK_UNMAPPED_BASE) {
		start_addr = addr = TASK_UNMAPPED_BASE;
		mm->cached_hole_size = 0;
		goto full_search;
	}
	return -ENOMEM;
}

static unsigned long slice_find_area_topdown(struct mm_struct *mm,
					     unsigned long len,
					     struct slice_mask available,
					     int psize, int use_cache)
{
	struct vm_area_struct *vma;
	unsigned long addr;
	struct slice_mask mask;
	int pshift = max_t(int, mmu_psize_defs[psize].shift, PAGE_SHIFT);

	/* check if free_area_cache is useful for us */
	if (use_cache) {
		if (len <= mm->cached_hole_size) {
			mm->cached_hole_size = 0;
			mm->free_area_cache = mm->mmap_base;
		}

		/* either no address requested or can't fit in requested
		 * address hole
		 */
		addr = mm->free_area_cache;

		/* make sure it can fit in the remaining address space */
		if (addr > len) {
			addr = _ALIGN_DOWN(addr - len, 1ul << pshift);
			mask = slice_range_to_mask(addr, len);
			if (slice_check_fit(mask, available) &&
			    slice_area_is_free(mm, addr, len))
					/* remember the address as a hint for
					 * next time
					 */
					return (mm->free_area_cache = addr);
		}
	}

	addr = mm->mmap_base;
	while (addr > len) {
		/* Go down by chunk size */
		addr = _ALIGN_DOWN(addr - len, 1ul << pshift);

		/* Check for hit with different page size */
		mask = slice_range_to_mask(addr, len);
		if (!slice_check_fit(mask, available)) {
			if (addr < SLICE_LOW_TOP)
				addr = _ALIGN_DOWN(addr, 1ul << SLICE_LOW_SHIFT);
			else if (addr < (1ul << SLICE_HIGH_SHIFT))
				addr = SLICE_LOW_TOP;
			else
				addr = _ALIGN_DOWN(addr, 1ul << SLICE_HIGH_SHIFT);
			continue;
		}

		/*
		 * Lookup failure means no vma is above this address,
		 * else if new region fits below vma->vm_start,
		 * return with success:
		 */
		vma = find_vma(mm, addr);
		if (!vma || (addr + len) <= vma->vm_start) {
			/* remember the address as a hint for next time */
			if (use_cache)
				mm->free_area_cache = addr;
			return addr;
		}

		/* remember the largest hole we saw so far */
		if (use_cache && (addr + mm->cached_hole_size) < vma->vm_start)
		        mm->cached_hole_size = vma->vm_start - addr;

		/* try just below the current vma->vm_start */
		addr = vma->vm_start;
	}

	/*
	 * 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.
	 */
	addr = slice_find_area_bottomup(mm, len, available, psize, 0);

	/*
	 * Restore the topdown base:
	 */
	if (use_cache) {
		mm->free_area_cache = mm->mmap_base;
		mm->cached_hole_size = ~0UL;
	}

	return addr;
}


static unsigned long slice_find_area(struct mm_struct *mm, unsigned long len,
				     struct slice_mask mask, int psize,
				     int topdown, int use_cache)
{
	if (topdown)
		return slice_find_area_topdown(mm, len, mask, psize, use_cache);
	else
		return slice_find_area_bottomup(mm, len, mask, psize, use_cache);
}

unsigned long slice_get_unmapped_area(unsigned long addr, unsigned long len,
				      unsigned long flags, unsigned int psize,
				      int topdown, int use_cache)
{
	struct slice_mask mask;
	struct slice_mask good_mask;
	struct slice_mask potential_mask = {0,0} /* silence stupid warning */;
	int pmask_set = 0;
	int fixed = (flags & MAP_FIXED);
	int pshift = max_t(int, mmu_psize_defs[psize].shift, PAGE_SHIFT);
	struct mm_struct *mm = current->mm;

	/* Sanity checks */
	BUG_ON(mm->task_size == 0);

	slice_dbg("slice_get_unmapped_area(mm=%p, psize=%d...\n", mm, psize);
	slice_dbg(" addr=%lx, len=%lx, flags=%lx, topdown=%d, use_cache=%d\n",
		  addr, len, flags, topdown, use_cache);

	if (len > mm->task_size)
		return -ENOMEM;
	if (fixed && (addr & ((1ul << pshift) - 1)))
		return -EINVAL;
	if (fixed && addr > (mm->task_size - len))
		return -EINVAL;

	/* 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);
	}

	/* First makeup a "good" mask of slices that have the right size
	 * already
	 */
	good_mask = slice_mask_for_size(mm, psize);
	slice_print_mask(" good_mask", good_mask);

	/* First check hint if it's valid or if we have MAP_FIXED */
	if ((addr != 0 || fixed) && (mm->task_size - len) >= addr) {

		/* Don't bother with hint if it overlaps a VMA */
		if (!fixed && !slice_area_is_free(mm, addr, len))
			goto search;

		/* Build a mask for the requested range */
		mask = slice_range_to_mask(addr, len);
		slice_print_mask(" mask", mask);

		/* Check if we fit in the good mask. If we do, we just return,
		 * nothing else to do
		 */
		if (slice_check_fit(mask, good_mask)) {
			slice_dbg(" fits good !\n");
			return addr;
		}

		/* We don't fit in the good mask, check what other slices are
		 * empty and thus can be converted
		 */
		potential_mask = slice_mask_for_free(mm);
		potential_mask.low_slices |= good_mask.low_slices;
		potential_mask.high_slices |= good_mask.high_slices;
		pmask_set = 1;
		slice_print_mask(" potential", potential_mask);
		if (slice_check_fit(mask, potential_mask)) {
			slice_dbg(" fits potential !\n");
			goto convert;
		}
	}

	/* If we have MAP_FIXED and failed the above step, then error out */
	if (fixed)
		return -EBUSY;

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

	/* Now let's see if we can find something in the existing slices
	 * for that size
	 */
	addr = slice_find_area(mm, len, good_mask, psize, topdown, use_cache);
	if (addr != -ENOMEM) {
		/* Found within the good mask, we don't have to setup,
		 * we thus return directly
		 */
		slice_dbg(" found area at 0x%lx\n", addr);
		return addr;
	}

	/* Won't fit, check what can be converted */
	if (!pmask_set) {
		potential_mask = slice_mask_for_free(mm);
		potential_mask.low_slices |= good_mask.low_slices;
		potential_mask.high_slices |= good_mask.high_slices;
		pmask_set = 1;
		slice_print_mask(" potential", potential_mask);
	}

	/* Now let's see if we can find something in the existing slices
	 * for that size
	 */
	addr = slice_find_area(mm, len, potential_mask, psize, topdown,
			       use_cache);
	if (addr == -ENOMEM)
		return -ENOMEM;

	mask = slice_range_to_mask(addr, len);
	slice_dbg(" found potential area at 0x%lx\n", addr);
	slice_print_mask(" mask", mask);

 convert:
	slice_convert(mm, mask, psize);
	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,
				       current->mm->context.user_psize,
				       0, 1);
}

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,
				       current->mm->context.user_psize,
				       1, 1);
}

unsigned int get_slice_psize(struct mm_struct *mm, unsigned long addr)
{
	u64 psizes;
	int index;

	if (addr < SLICE_LOW_TOP) {
		psizes = mm->context.low_slices_psize;
		index = GET_LOW_SLICE_INDEX(addr);
	} else {
		psizes = mm->context.high_slices_psize;
		index = GET_HIGH_SLICE_INDEX(addr);
	}

	return (psizes >> (index * 4)) & 0xf;
}
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
 *
 * 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)
{
	unsigned long flags, lpsizes, hpsizes;
	unsigned int old_psize;
	int i;

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

	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));

	hpsizes = mm->context.high_slices_psize;
	for (i = 0; i < SLICE_NUM_HIGH; i++)
		if (((hpsizes >> (i * 4)) & 0xf) == old_psize)
			hpsizes = (hpsizes & ~(0xful << (i * 4))) |
				(((unsigned long)psize) << (i * 4));

	mm->context.low_slices_psize = lpsizes;
	mm->context.high_slices_psize = hpsizes;

	slice_dbg(" lsps=%lx, hsps=%lx\n",
		  mm->context.low_slices_psize,
		  mm->context.high_slices_psize);

 bail:
	spin_unlock_irqrestore(&slice_convert_lock, flags);
}

/*
 * is_hugepage_only_range() is used by generic code to verify wether
 * 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;

	mask = slice_range_to_mask(addr, len);
	available = slice_mask_for_size(mm, mm->context.user_psize);

#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
	return !slice_check_fit(mask, available);
}