init.c 19.2 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3
/*
 *  linux/arch/arm/mm/init.c
 *
R
Russell King 已提交
4
 *  Copyright (C) 1995-2005 Russell King
L
Linus Torvalds 已提交
5 6 7 8 9 10 11 12 13 14 15
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/swap.h>
#include <linux/init.h>
#include <linux/bootmem.h>
#include <linux/mman.h>
16
#include <linux/export.h>
L
Linus Torvalds 已提交
17 18
#include <linux/nodemask.h>
#include <linux/initrd.h>
G
Grant Likely 已提交
19
#include <linux/of_fdt.h>
20
#include <linux/highmem.h>
21
#include <linux/gfp.h>
R
Russell King 已提交
22
#include <linux/memblock.h>
23
#include <linux/dma-contiguous.h>
24
#include <linux/sizes.h>
25
#include <linux/stop_machine.h>
L
Linus Torvalds 已提交
26

27
#include <asm/cp15.h>
L
Linus Torvalds 已提交
28
#include <asm/mach-types.h>
29
#include <asm/memblock.h>
30
#include <asm/memory.h>
31
#include <asm/prom.h>
R
Russell King 已提交
32
#include <asm/sections.h>
L
Linus Torvalds 已提交
33
#include <asm/setup.h>
34
#include <asm/system_info.h>
L
Linus Torvalds 已提交
35
#include <asm/tlb.h>
36
#include <asm/fixmap.h>
L
Linus Torvalds 已提交
37 38 39 40

#include <asm/mach/arch.h>
#include <asm/mach/map.h>

41 42
#include "mm.h"

43 44 45 46 47 48 49 50
#ifdef CONFIG_CPU_CP15_MMU
unsigned long __init __clear_cr(unsigned long mask)
{
	cr_alignment = cr_alignment & ~mask;
	return cr_alignment;
}
#endif

51
static phys_addr_t phys_initrd_start __initdata = 0;
52 53
static unsigned long phys_initrd_size __initdata = 0;

54
static int __init early_initrd(char *p)
55
{
56 57
	phys_addr_t start;
	unsigned long size;
58
	char *endp;
59

60 61 62
	start = memparse(p, &endp);
	if (*endp == ',') {
		size = memparse(endp + 1, NULL);
63 64 65 66

		phys_initrd_start = start;
		phys_initrd_size = size;
	}
67
	return 0;
68
}
69
early_param("initrd", early_initrd);
70 71 72

static int __init parse_tag_initrd(const struct tag *tag)
{
R
Russell King 已提交
73
	pr_warn("ATAG_INITRD is deprecated; "
74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89
		"please update your bootloader.\n");
	phys_initrd_start = __virt_to_phys(tag->u.initrd.start);
	phys_initrd_size = tag->u.initrd.size;
	return 0;
}

__tagtable(ATAG_INITRD, parse_tag_initrd);

static int __init parse_tag_initrd2(const struct tag *tag)
{
	phys_initrd_start = tag->u.initrd.start;
	phys_initrd_size = tag->u.initrd.size;
	return 0;
}

__tagtable(ATAG_INITRD2, parse_tag_initrd2);
L
Linus Torvalds 已提交
90

91
static void __init find_limits(unsigned long *min, unsigned long *max_low,
92
			       unsigned long *max_high)
R
Russell King 已提交
93
{
L
Laura Abbott 已提交
94 95 96
	*max_low = PFN_DOWN(memblock_get_current_limit());
	*min = PFN_UP(memblock_start_of_DRAM());
	*max_high = PFN_DOWN(memblock_end_of_DRAM());
R
Russell King 已提交
97 98
}

99
#ifdef CONFIG_ZONE_DMA
100

101
phys_addr_t arm_dma_zone_size __read_mostly;
102 103
EXPORT_SYMBOL(arm_dma_zone_size);

104 105 106 107 108 109
/*
 * The DMA mask corresponding to the maximum bus address allocatable
 * using GFP_DMA.  The default here places no restriction on DMA
 * allocations.  This must be the smallest DMA mask in the system,
 * so a successful GFP_DMA allocation will always satisfy this.
 */
110
phys_addr_t arm_dma_limit;
111
unsigned long arm_dma_pfn_limit;
112

113 114 115 116 117 118 119 120 121 122 123 124 125
static void __init arm_adjust_dma_zone(unsigned long *size, unsigned long *hole,
	unsigned long dma_size)
{
	if (size[0] <= dma_size)
		return;

	size[ZONE_NORMAL] = size[0] - dma_size;
	size[ZONE_DMA] = dma_size;
	hole[ZONE_NORMAL] = hole[0];
	hole[ZONE_DMA] = 0;
}
#endif

126
void __init setup_dma_zone(const struct machine_desc *mdesc)
127 128 129 130
{
#ifdef CONFIG_ZONE_DMA
	if (mdesc->dma_zone_size) {
		arm_dma_zone_size = mdesc->dma_zone_size;
131
		arm_dma_limit = PHYS_OFFSET + arm_dma_zone_size - 1;
132 133
	} else
		arm_dma_limit = 0xffffffff;
134
	arm_dma_pfn_limit = arm_dma_limit >> PAGE_SHIFT;
135 136 137
#endif
}

138
static void __init zone_sizes_init(unsigned long min, unsigned long max_low,
139
	unsigned long max_high)
140 141
{
	unsigned long zone_size[MAX_NR_ZONES], zhole_size[MAX_NR_ZONES];
142
	struct memblock_region *reg;
143

R
Russell King 已提交
144
	/*
R
Russell King 已提交
145
	 * initialise the zones.
R
Russell King 已提交
146 147 148 149
	 */
	memset(zone_size, 0, sizeof(zone_size));

	/*
R
Russell King 已提交
150 151 152
	 * The memory size has already been determined.  If we need
	 * to do anything fancy with the allocation of this memory
	 * to the zones, now is the time to do it.
R
Russell King 已提交
153
	 */
R
Russell King 已提交
154 155 156 157
	zone_size[0] = max_low - min;
#ifdef CONFIG_HIGHMEM
	zone_size[ZONE_HIGHMEM] = max_high - max_low;
#endif
R
Russell King 已提交
158 159

	/*
R
Russell King 已提交
160 161
	 * Calculate the size of the holes.
	 *  holes = node_size - sum(bank_sizes)
R
Russell King 已提交
162
	 */
R
Russell King 已提交
163
	memcpy(zhole_size, zone_size, sizeof(zhole_size));
164 165 166 167 168 169 170 171
	for_each_memblock(memory, reg) {
		unsigned long start = memblock_region_memory_base_pfn(reg);
		unsigned long end = memblock_region_memory_end_pfn(reg);

		if (start < max_low) {
			unsigned long low_end = min(end, max_low);
			zhole_size[0] -= low_end - start;
		}
R
Russell King 已提交
172
#ifdef CONFIG_HIGHMEM
173 174 175 176
		if (end > max_low) {
			unsigned long high_start = max(start, max_low);
			zhole_size[ZONE_HIGHMEM] -= end - high_start;
		}
R
Russell King 已提交
177 178
#endif
	}
R
Russell King 已提交
179

180
#ifdef CONFIG_ZONE_DMA
R
Russell King 已提交
181 182 183 184
	/*
	 * Adjust the sizes according to any special requirements for
	 * this machine type.
	 */
185
	if (arm_dma_zone_size)
186 187
		arm_adjust_dma_zone(zone_size, zhole_size,
			arm_dma_zone_size >> PAGE_SHIFT);
188
#endif
R
Russell King 已提交
189

R
Russell King 已提交
190
	free_area_init_node(0, zone_size, min, zhole_size);
L
Linus Torvalds 已提交
191 192
}

193
#ifdef CONFIG_HAVE_ARCH_PFN_VALID
194 195
int pfn_valid(unsigned long pfn)
{
196
	return memblock_is_map_memory(__pfn_to_phys(pfn));
197 198
}
EXPORT_SYMBOL(pfn_valid);
199
#endif
200

201
#ifndef CONFIG_SPARSEMEM
202
static void __init arm_memory_present(void)
203 204 205
{
}
#else
206
static void __init arm_memory_present(void)
207
{
208 209
	struct memblock_region *reg;

210
	for_each_memblock(memory, reg)
211 212
		memory_present(0, memblock_region_memory_base_pfn(reg),
			       memblock_region_memory_end_pfn(reg));
213
}
214 215
#endif

216 217
static bool arm_memblock_steal_permitted = true;

218
phys_addr_t __init arm_memblock_steal(phys_addr_t size, phys_addr_t align)
219 220 221 222 223
{
	phys_addr_t phys;

	BUG_ON(!arm_memblock_steal_permitted);

224
	phys = memblock_alloc_base(size, align, MEMBLOCK_ALLOC_ANYWHERE);
225 226 227 228 229 230
	memblock_free(phys, size);
	memblock_remove(phys, size);

	return phys;
}

231
static void __init arm_initrd_init(void)
R
Russell King 已提交
232 233
{
#ifdef CONFIG_BLK_DEV_INITRD
234
	/* FDT scan will populate initrd_start */
235
	if (initrd_start && !phys_initrd_size) {
236 237 238
		phys_initrd_start = __virt_to_phys(initrd_start);
		phys_initrd_size = initrd_end - initrd_start;
	}
239
	initrd_start = initrd_end = 0;
240 241
	if (phys_initrd_size &&
	    !memblock_is_region_memory(phys_initrd_start, phys_initrd_size)) {
242 243
		pr_err("INITRD: 0x%08llx+0x%08lx is not a memory region - disabling initrd\n",
		       (u64)phys_initrd_start, phys_initrd_size);
244 245
		phys_initrd_start = phys_initrd_size = 0;
	}
246 247
	if (phys_initrd_size &&
	    memblock_is_region_reserved(phys_initrd_start, phys_initrd_size)) {
248 249
		pr_err("INITRD: 0x%08llx+0x%08lx overlaps in-use memory region - disabling initrd\n",
		       (u64)phys_initrd_start, phys_initrd_size);
250 251
		phys_initrd_start = phys_initrd_size = 0;
	}
R
Russell King 已提交
252 253 254 255 256 257 258 259
	if (phys_initrd_size) {
		memblock_reserve(phys_initrd_start, phys_initrd_size);

		/* Now convert initrd to virtual addresses */
		initrd_start = __phys_to_virt(phys_initrd_start);
		initrd_end = initrd_start + phys_initrd_size;
	}
#endif
260 261 262 263 264 265 266 267
}

void __init arm_memblock_init(const struct machine_desc *mdesc)
{
	/* Register the kernel text, kernel data and initrd with memblock. */
	memblock_reserve(__pa(KERNEL_START), KERNEL_END - KERNEL_START);

	arm_initrd_init();
R
Russell King 已提交
268 269 270

	arm_mm_memblock_reserve();

271 272 273 274
	/* reserve any platform specific memblock areas */
	if (mdesc->reserve)
		mdesc->reserve();

275
	early_init_fdt_reserve_self();
276 277
	early_init_fdt_scan_reserved_mem();

278
	/* reserve memory for DMA contiguous allocations */
279
	dma_contiguous_reserve(arm_dma_limit);
280

281
	arm_memblock_steal_permitted = false;
R
Russell King 已提交
282 283 284
	memblock_dump_all();
}

285
void __init bootmem_init(void)
L
Linus Torvalds 已提交
286
{
R
Russell King 已提交
287
	unsigned long min, max_low, max_high;
L
Linus Torvalds 已提交
288

289
	memblock_allow_resize();
R
Russell King 已提交
290 291
	max_low = max_high = 0;

292
	find_limits(&min, &max_low, &max_high);
R
Russell King 已提交
293

V
Vladimir Murzin 已提交
294 295 296
	early_memtest((phys_addr_t)min << PAGE_SHIFT,
		      (phys_addr_t)max_low << PAGE_SHIFT);

R
Russell King 已提交
297 298 299 300
	/*
	 * Sparsemem tries to allocate bootmem in memory_present(),
	 * so must be done after the fixed reservations
	 */
301
	arm_memory_present();
L
Linus Torvalds 已提交
302

303 304 305 306 307 308
	/*
	 * sparse_init() needs the bootmem allocator up and running.
	 */
	sparse_init();

	/*
R
Russell King 已提交
309
	 * Now free the memory - free_area_init_node needs
310 311 312
	 * the sparse mem_map arrays initialized by sparse_init()
	 * for memmap_init_zone(), otherwise all PFNs are invalid.
	 */
313
	zone_sizes_init(min, max_low, max_high);
314

R
Russell King 已提交
315 316 317 318 319
	/*
	 * This doesn't seem to be used by the Linux memory manager any
	 * more, but is used by ll_rw_block.  If we can get rid of it, we
	 * also get rid of some of the stuff above as well.
	 */
320 321 322
	min_low_pfn = min;
	max_low_pfn = max_low;
	max_pfn = max_high;
R
Russell King 已提交
323
}
L
Linus Torvalds 已提交
324

325 326 327 328 329 330 331
/*
 * Poison init memory with an undefined instruction (ARM) or a branch to an
 * undefined instruction (Thumb).
 */
static inline void poison_init_mem(void *s, size_t count)
{
	u32 *p = (u32 *)s;
332
	for (; count != 0; count -= 4)
333 334 335
		*p++ = 0xe7fddef0;
}

336
static inline void
R
Russell King 已提交
337
free_memmap(unsigned long start_pfn, unsigned long end_pfn)
338 339
{
	struct page *start_pg, *end_pg;
340
	phys_addr_t pg, pgend;
341 342 343 344

	/*
	 * Convert start_pfn/end_pfn to a struct page pointer.
	 */
345
	start_pg = pfn_to_page(start_pfn - 1) + 1;
346
	end_pg = pfn_to_page(end_pfn - 1) + 1;
347 348 349 350 351

	/*
	 * Convert to physical addresses, and
	 * round start upwards and end downwards.
	 */
352 353
	pg = PAGE_ALIGN(__pa(start_pg));
	pgend = __pa(end_pg) & PAGE_MASK;
354 355 356 357 358 359

	/*
	 * If there are free pages between these,
	 * free the section of the memmap array.
	 */
	if (pg < pgend)
360
		memblock_free_early(pg, pgend - pg);
361 362 363 364 365
}

/*
 * The mem_map array can get very big.  Free the unused area of the memory map.
 */
L
Laura Abbott 已提交
366
static void __init free_unused_memmap(void)
367
{
L
Laura Abbott 已提交
368 369
	unsigned long start, prev_end = 0;
	struct memblock_region *reg;
370 371

	/*
372 373
	 * This relies on each bank being in address order.
	 * The banks are sorted previously in bootmem_init().
374
	 */
L
Laura Abbott 已提交
375 376
	for_each_memblock(memory, reg) {
		start = memblock_region_memory_base_pfn(reg);
377

378 379 380 381 382
#ifdef CONFIG_SPARSEMEM
		/*
		 * Take care not to free memmap entries that don't exist
		 * due to SPARSEMEM sections which aren't present.
		 */
L
Laura Abbott 已提交
383 384
		start = min(start,
				 ALIGN(prev_end, PAGES_PER_SECTION));
385 386 387 388 389 390
#else
		/*
		 * Align down here since the VM subsystem insists that the
		 * memmap entries are valid from the bank start aligned to
		 * MAX_ORDER_NR_PAGES.
		 */
L
Laura Abbott 已提交
391
		start = round_down(start, MAX_ORDER_NR_PAGES);
392
#endif
393 394 395 396
		/*
		 * If we had a previous bank, and there is a space
		 * between the current bank and the previous, free it.
		 */
L
Laura Abbott 已提交
397 398
		if (prev_end && prev_end < start)
			free_memmap(prev_end, start);
399

400 401 402 403 404
		/*
		 * Align up here since the VM subsystem insists that the
		 * memmap entries are valid from the bank end aligned to
		 * MAX_ORDER_NR_PAGES.
		 */
L
Laura Abbott 已提交
405 406
		prev_end = ALIGN(memblock_region_memory_end_pfn(reg),
				 MAX_ORDER_NR_PAGES);
407
	}
408 409

#ifdef CONFIG_SPARSEMEM
L
Laura Abbott 已提交
410 411 412
	if (!IS_ALIGNED(prev_end, PAGES_PER_SECTION))
		free_memmap(prev_end,
			    ALIGN(prev_end, PAGES_PER_SECTION));
413
#endif
414 415
}

416 417 418
#ifdef CONFIG_HIGHMEM
static inline void free_area_high(unsigned long pfn, unsigned long end)
{
419 420
	for (; pfn < end; pfn++)
		free_highmem_page(pfn_to_page(pfn));
421 422 423
}
#endif

424 425 426
static void __init free_highpages(void)
{
#ifdef CONFIG_HIGHMEM
427
	unsigned long max_low = max_low_pfn;
428
	struct memblock_region *mem, *res;
429 430

	/* set highmem page free */
431 432 433 434 435 436 437 438
	for_each_memblock(memory, mem) {
		unsigned long start = memblock_region_memory_base_pfn(mem);
		unsigned long end = memblock_region_memory_end_pfn(mem);

		/* Ignore complete lowmem entries */
		if (end <= max_low)
			continue;

439 440 441
		if (memblock_is_nomap(mem))
			continue;

442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461
		/* Truncate partial highmem entries */
		if (start < max_low)
			start = max_low;

		/* Find and exclude any reserved regions */
		for_each_memblock(reserved, res) {
			unsigned long res_start, res_end;

			res_start = memblock_region_reserved_base_pfn(res);
			res_end = memblock_region_reserved_end_pfn(res);

			if (res_end < start)
				continue;
			if (res_start < start)
				res_start = start;
			if (res_start > end)
				res_start = end;
			if (res_end > end)
				res_end = end;
			if (res_start != start)
462
				free_area_high(start, res_start);
463 464 465 466 467 468 469
			start = res_end;
			if (start == end)
				break;
		}

		/* And now free anything which remains */
		if (start < end)
470
			free_area_high(start, end);
471 472 473 474
	}
#endif
}

L
Linus Torvalds 已提交
475 476 477 478 479 480 481
/*
 * mem_init() marks the free areas in the mem_map and tells us how much
 * memory is free.  This is done after various parts of the system have
 * claimed their memory after the kernel image.
 */
void __init mem_init(void)
{
482 483 484 485 486
#ifdef CONFIG_HAVE_TCM
	/* These pointers are filled in on TCM detection */
	extern u32 dtcm_end;
	extern u32 itcm_end;
#endif
L
Linus Torvalds 已提交
487

488
	set_max_mapnr(pfn_to_page(max_pfn) - mem_map);
L
Linus Torvalds 已提交
489 490

	/* this will put all unused low memory onto the freelists */
L
Laura Abbott 已提交
491
	free_unused_memmap();
492
	free_all_bootmem();
L
Linus Torvalds 已提交
493 494 495

#ifdef CONFIG_SA1111
	/* now that our DMA memory is actually so designated, we can free it */
496
	free_reserved_area(__va(PHYS_OFFSET), swapper_pg_dir, -1, NULL);
L
Linus Torvalds 已提交
497 498
#endif

499
	free_highpages();
500

501
	mem_init_print_info(NULL);
L
Linus Torvalds 已提交
502

503 504 505 506
#define MLK(b, t) b, t, ((t) - (b)) >> 10
#define MLM(b, t) b, t, ((t) - (b)) >> 20
#define MLK_ROUNDUP(b, t) b, t, DIV_ROUND_UP(((t) - (b)), SZ_1K)

R
Russell King 已提交
507
	pr_notice("Virtual kernel memory layout:\n"
508
			"    vector  : 0x%08lx - 0x%08lx   (%4ld kB)\n"
509 510 511 512
#ifdef CONFIG_HAVE_TCM
			"    DTCM    : 0x%08lx - 0x%08lx   (%4ld kB)\n"
			"    ITCM    : 0x%08lx - 0x%08lx   (%4ld kB)\n"
#endif
513 514 515 516 517 518
			"    fixmap  : 0x%08lx - 0x%08lx   (%4ld kB)\n"
			"    vmalloc : 0x%08lx - 0x%08lx   (%4ld MB)\n"
			"    lowmem  : 0x%08lx - 0x%08lx   (%4ld MB)\n"
#ifdef CONFIG_HIGHMEM
			"    pkmap   : 0x%08lx - 0x%08lx   (%4ld MB)\n"
#endif
519
#ifdef CONFIG_MODULES
520
			"    modules : 0x%08lx - 0x%08lx   (%4ld MB)\n"
521
#endif
R
Russell King 已提交
522 523 524 525
			"      .text : 0x%p" " - 0x%p" "   (%4td kB)\n"
			"      .init : 0x%p" " - 0x%p" "   (%4td kB)\n"
			"      .data : 0x%p" " - 0x%p" "   (%4td kB)\n"
			"       .bss : 0x%p" " - 0x%p" "   (%4td kB)\n",
526

527
			MLK(VECTORS_BASE, VECTORS_BASE + PAGE_SIZE),
528
#ifdef CONFIG_HAVE_TCM
529 530
			MLK(DTCM_OFFSET, (unsigned long) dtcm_end),
			MLK(ITCM_OFFSET, (unsigned long) itcm_end),
531
#endif
M
Mark Salter 已提交
532
			MLK(FIXADDR_START, FIXADDR_END),
533
			MLM(VMALLOC_START, VMALLOC_END),
534 535 536 537 538
			MLM(PAGE_OFFSET, (unsigned long)high_memory),
#ifdef CONFIG_HIGHMEM
			MLM(PKMAP_BASE, (PKMAP_BASE) + (LAST_PKMAP) *
				(PAGE_SIZE)),
#endif
539
#ifdef CONFIG_MODULES
540
			MLM(MODULES_VADDR, MODULES_END),
541
#endif
542 543

			MLK_ROUNDUP(_text, _etext),
544
			MLK_ROUNDUP(__init_begin, __init_end),
545 546
			MLK_ROUNDUP(_sdata, _edata),
			MLK_ROUNDUP(__bss_start, __bss_stop));
547 548 549 550 551

#undef MLK
#undef MLM
#undef MLK_ROUNDUP

552 553 554 555 556 557 558 559 560 561 562 563 564 565
	/*
	 * Check boundaries twice: Some fundamental inconsistencies can
	 * be detected at build time already.
	 */
#ifdef CONFIG_MMU
	BUILD_BUG_ON(TASK_SIZE				> MODULES_VADDR);
	BUG_ON(TASK_SIZE 				> MODULES_VADDR);
#endif

#ifdef CONFIG_HIGHMEM
	BUILD_BUG_ON(PKMAP_BASE + LAST_PKMAP * PAGE_SIZE > PAGE_OFFSET);
	BUG_ON(PKMAP_BASE + LAST_PKMAP * PAGE_SIZE	> PAGE_OFFSET);
#endif

566
	if (PAGE_SIZE >= 16384 && get_num_physpages() <= 128) {
L
Linus Torvalds 已提交
567 568 569 570 571 572 573 574 575 576
		extern int sysctl_overcommit_memory;
		/*
		 * On a machine this small we won't get
		 * anywhere without overcommit, so turn
		 * it on by default.
		 */
		sysctl_overcommit_memory = OVERCOMMIT_ALWAYS;
	}
}

577
#ifdef CONFIG_DEBUG_RODATA
578
struct section_perm {
579
	const char *name;
580 581 582 583
	unsigned long start;
	unsigned long end;
	pmdval_t mask;
	pmdval_t prot;
584
	pmdval_t clear;
585 586
};

587 588 589
/* First section-aligned location at or after __start_rodata. */
extern char __start_rodata_section_aligned[];

590
static struct section_perm nx_perms[] = {
591 592
	/* Make pages tables, etc before _stext RW (set NX). */
	{
593
		.name	= "pre-text NX",
594 595 596 597 598 599 600
		.start	= PAGE_OFFSET,
		.end	= (unsigned long)_stext,
		.mask	= ~PMD_SECT_XN,
		.prot	= PMD_SECT_XN,
	},
	/* Make init RW (set NX). */
	{
601
		.name	= "init NX",
602 603 604 605 606
		.start	= (unsigned long)__init_begin,
		.end	= (unsigned long)_sdata,
		.mask	= ~PMD_SECT_XN,
		.prot	= PMD_SECT_XN,
	},
607 608
	/* Make rodata NX (set RO in ro_perms below). */
	{
609
		.name	= "rodata NX",
610
		.start  = (unsigned long)__start_rodata_section_aligned,
611 612 613 614
		.end    = (unsigned long)__init_begin,
		.mask   = ~PMD_SECT_XN,
		.prot   = PMD_SECT_XN,
	},
615 616
};

617 618 619
static struct section_perm ro_perms[] = {
	/* Make kernel code and rodata RX (set RO). */
	{
620
		.name	= "text/rodata RO",
621 622 623
		.start  = (unsigned long)_stext,
		.end    = (unsigned long)__init_begin,
#ifdef CONFIG_ARM_LPAE
624 625
		.mask   = ~L_PMD_SECT_RDONLY,
		.prot   = L_PMD_SECT_RDONLY,
626 627 628 629 630 631 632 633
#else
		.mask   = ~(PMD_SECT_APX | PMD_SECT_AP_WRITE),
		.prot   = PMD_SECT_APX | PMD_SECT_AP_WRITE,
		.clear  = PMD_SECT_AP_WRITE,
#endif
	},
};

634 635 636 637 638 639
/*
 * Updates section permissions only for the current mm (sections are
 * copied into each mm). During startup, this is the init_mm. Is only
 * safe to be called with preemption disabled, as under stop_machine().
 */
static inline void section_update(unsigned long addr, pmdval_t mask,
640
				  pmdval_t prot, struct mm_struct *mm)
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
{
	pmd_t *pmd;

	pmd = pmd_offset(pud_offset(pgd_offset(mm, addr), addr), addr);

#ifdef CONFIG_ARM_LPAE
	pmd[0] = __pmd((pmd_val(pmd[0]) & mask) | prot);
#else
	if (addr & SECTION_SIZE)
		pmd[1] = __pmd((pmd_val(pmd[1]) & mask) | prot);
	else
		pmd[0] = __pmd((pmd_val(pmd[0]) & mask) | prot);
#endif
	flush_pmd_entry(pmd);
	local_flush_tlb_kernel_range(addr, addr + SECTION_SIZE);
}

/* Make sure extended page tables are in use. */
static inline bool arch_has_strict_perms(void)
{
	if (cpu_architecture() < CPU_ARCH_ARMv6)
		return false;

	return !!(get_cr() & CR_XP);
}

667 668 669 670 671 672 673 674 675 676 677 678
void set_section_perms(struct section_perm *perms, int n, bool set,
			struct mm_struct *mm)
{
	size_t i;
	unsigned long addr;

	if (!arch_has_strict_perms())
		return;

	for (i = 0; i < n; i++) {
		if (!IS_ALIGNED(perms[i].start, SECTION_SIZE) ||
		    !IS_ALIGNED(perms[i].end, SECTION_SIZE)) {
679 680
			pr_err("BUG: %s section %lx-%lx not aligned to %lx\n",
				perms[i].name, perms[i].start, perms[i].end,
681 682 683 684 685 686 687 688 689 690 691
				SECTION_SIZE);
			continue;
		}

		for (addr = perms[i].start;
		     addr < perms[i].end;
		     addr += SECTION_SIZE)
			section_update(addr, perms[i].mask,
				set ? perms[i].prot : perms[i].clear, mm);
	}

692 693
}

694
static void update_sections_early(struct section_perm perms[], int n)
695
{
696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718
	struct task_struct *t, *s;

	read_lock(&tasklist_lock);
	for_each_process(t) {
		if (t->flags & PF_KTHREAD)
			continue;
		for_each_thread(t, s)
			set_section_perms(perms, n, true, s->mm);
	}
	read_unlock(&tasklist_lock);
	set_section_perms(perms, n, true, current->active_mm);
	set_section_perms(perms, n, true, &init_mm);
}

int __fix_kernmem_perms(void *unused)
{
	update_sections_early(nx_perms, ARRAY_SIZE(nx_perms));
	return 0;
}

void fix_kernmem_perms(void)
{
	stop_machine(__fix_kernmem_perms, NULL, NULL);
719
}
720

721 722 723 724 725 726
int __mark_rodata_ro(void *unused)
{
	update_sections_early(ro_perms, ARRAY_SIZE(ro_perms));
	return 0;
}

727 728
void mark_rodata_ro(void)
{
729
	stop_machine(__mark_rodata_ro, NULL, NULL);
730 731 732 733
}

void set_kernel_text_rw(void)
{
734 735
	set_section_perms(ro_perms, ARRAY_SIZE(ro_perms), false,
				current->active_mm);
736 737 738 739
}

void set_kernel_text_ro(void)
{
740 741
	set_section_perms(ro_perms, ARRAY_SIZE(ro_perms), true,
				current->active_mm);
742 743
}

744 745
#else
static inline void fix_kernmem_perms(void) { }
746
#endif /* CONFIG_DEBUG_RODATA */
747 748

void free_tcmmem(void)
L
Linus Torvalds 已提交
749
{
750
#ifdef CONFIG_HAVE_TCM
751
	extern char __tcm_start, __tcm_end;
752

753
	poison_init_mem(&__tcm_start, &__tcm_end - &__tcm_start);
754
	free_reserved_area(&__tcm_start, &__tcm_end, -1, "TCM link");
755
#endif
756 757 758 759 760 761
}

void free_initmem(void)
{
	fix_kernmem_perms();
	free_tcmmem();
762

763
	poison_init_mem(__init_begin, __init_end - __init_begin);
N
Nicolas Pitre 已提交
764
	if (!machine_is_integrator() && !machine_is_cintegrator())
765
		free_initmem_default(-1);
L
Linus Torvalds 已提交
766 767 768 769 770 771 772 773
}

#ifdef CONFIG_BLK_DEV_INITRD

static int keep_initrd;

void free_initrd_mem(unsigned long start, unsigned long end)
{
774
	if (!keep_initrd) {
775 776 777 778 779
		if (start == initrd_start)
			start = round_down(start, PAGE_SIZE);
		if (end == initrd_end)
			end = round_up(end, PAGE_SIZE);

780
		poison_init_mem((void *)start, PAGE_ALIGN(end) - start);
781
		free_reserved_area((void *)start, (void *)end, -1, "initrd");
782
	}
L
Linus Torvalds 已提交
783 784 785 786 787 788 789 790 791 792
}

static int __init keepinitrd_setup(char *__unused)
{
	keep_initrd = 1;
	return 1;
}

__setup("keepinitrd", keepinitrd_setup);
#endif