[deliverable/linux.git] / arch / arm64 / mm / init.c

/*
 * Based on arch/arm/mm/init.c
 *
 * Copyright (C) 1995-2005 Russell King
 * Copyright (C) 2012 ARM Ltd.
 *
 * 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.
 *
 * 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, see <http://www.gnu.org/licenses/>.
 */

#include <linux/kernel.h>
#include <linux/export.h>
#include <linux/errno.h>
#include <linux/swap.h>
#include <linux/init.h>
#include <linux/bootmem.h>
#include <linux/cache.h>
#include <linux/mman.h>
#include <linux/nodemask.h>
#include <linux/initrd.h>
#include <linux/gfp.h>
#include <linux/memblock.h>
#include <linux/sort.h>
#include <linux/of_fdt.h>
#include <linux/dma-mapping.h>
#include <linux/dma-contiguous.h>
#include <linux/efi.h>
#include <linux/swiotlb.h>
#include <linux/vmalloc.h>

#include <asm/boot.h>
#include <asm/fixmap.h>
#include <asm/kasan.h>
#include <asm/kernel-pgtable.h>
#include <asm/memory.h>
#include <asm/numa.h>
#include <asm/sections.h>
#include <asm/setup.h>
#include <asm/sizes.h>
#include <asm/tlb.h>
#include <asm/alternative.h>

/*
 * We need to be able to catch inadvertent references to memstart_addr
 * that occur (potentially in generic code) before arm64_memblock_init()
 * executes, which assigns it its actual value. So use a default value
 * that cannot be mistaken for a real physical address.
 */
s64 memstart_addr __ro_after_init = -1;
phys_addr_t arm64_dma_phys_limit __ro_after_init;

#ifdef CONFIG_BLK_DEV_INITRD
static int __init early_initrd(char *p)
{
	unsigned long start, size;
	char *endp;

	start = memparse(p, &endp);
	if (*endp == ',') {
		size = memparse(endp + 1, NULL);

		initrd_start = start;
		initrd_end = start + size;
	}
	return 0;
}
early_param("initrd", early_initrd);
#endif

/*
 * Return the maximum physical address for ZONE_DMA (DMA_BIT_MASK(32)). It
 * currently assumes that for memory starting above 4G, 32-bit devices will
 * use a DMA offset.
 */
static phys_addr_t __init max_zone_dma_phys(void)
{
	phys_addr_t offset = memblock_start_of_DRAM() & GENMASK_ULL(63, 32);
	return min(offset + (1ULL << 32), memblock_end_of_DRAM());
}

#ifdef CONFIG_NUMA

static void __init zone_sizes_init(unsigned long min, unsigned long max)
{
	unsigned long max_zone_pfns[MAX_NR_ZONES]  = {0};

	if (IS_ENABLED(CONFIG_ZONE_DMA))
		max_zone_pfns[ZONE_DMA] = PFN_DOWN(max_zone_dma_phys());
	max_zone_pfns[ZONE_NORMAL] = max;

	free_area_init_nodes(max_zone_pfns);
}

#else

static void __init zone_sizes_init(unsigned long min, unsigned long max)
{
	struct memblock_region *reg;
	unsigned long zone_size[MAX_NR_ZONES], zhole_size[MAX_NR_ZONES];
	unsigned long max_dma = min;

	memset(zone_size, 0, sizeof(zone_size));

	/* 4GB maximum for 32-bit only capable devices */
#ifdef CONFIG_ZONE_DMA
	max_dma = PFN_DOWN(arm64_dma_phys_limit);
	zone_size[ZONE_DMA] = max_dma - min;
#endif
	zone_size[ZONE_NORMAL] = max - max_dma;

	memcpy(zhole_size, zone_size, sizeof(zhole_size));

	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)
			continue;

#ifdef CONFIG_ZONE_DMA
		if (start < max_dma) {
			unsigned long dma_end = min(end, max_dma);
			zhole_size[ZONE_DMA] -= dma_end - start;
		}
#endif
		if (end > max_dma) {
			unsigned long normal_end = min(end, max);
			unsigned long normal_start = max(start, max_dma);
			zhole_size[ZONE_NORMAL] -= normal_end - normal_start;
		}
	}

	free_area_init_node(0, zone_size, min, zhole_size);
}

#endif /* CONFIG_NUMA */

#ifdef CONFIG_HAVE_ARCH_PFN_VALID
int pfn_valid(unsigned long pfn)
{
	return memblock_is_map_memory(pfn << PAGE_SHIFT);
}
EXPORT_SYMBOL(pfn_valid);
#endif

#ifndef CONFIG_SPARSEMEM
static void __init arm64_memory_present(void)
{
}
#else
static void __init arm64_memory_present(void)
{
	struct memblock_region *reg;

	for_each_memblock(memory, reg) {
		int nid = memblock_get_region_node(reg);

		memory_present(nid, memblock_region_memory_base_pfn(reg),
				memblock_region_memory_end_pfn(reg));
	}
}
#endif

static phys_addr_t memory_limit = (phys_addr_t)ULLONG_MAX;

/*
 * Limit the memory size that was specified via FDT.
 */
static int __init early_mem(char *p)
{
	if (!p)
		return 1;

	memory_limit = memparse(p, &p) & PAGE_MASK;
	pr_notice("Memory limited to %lldMB\n", memory_limit >> 20);

	return 0;
}
early_param("mem", early_mem);

void __init arm64_memblock_init(void)
{
	const s64 linear_region_size = -(s64)PAGE_OFFSET;

	/*
	 * Ensure that the linear region takes up exactly half of the kernel
	 * virtual address space. This way, we can distinguish a linear address
	 * from a kernel/module/vmalloc address by testing a single bit.
	 */
	BUILD_BUG_ON(linear_region_size != BIT(VA_BITS - 1));

	/*
	 * Select a suitable value for the base of physical memory.
	 */
	memstart_addr = round_down(memblock_start_of_DRAM(),
				   ARM64_MEMSTART_ALIGN);

	/*
	 * Remove the memory that we will not be able to cover with the
	 * linear mapping. Take care not to clip the kernel which may be
	 * high in memory.
	 */
	memblock_remove(max_t(u64, memstart_addr + linear_region_size, __pa(_end)),
			ULLONG_MAX);
	if (memstart_addr + linear_region_size < memblock_end_of_DRAM()) {
		/* ensure that memstart_addr remains sufficiently aligned */
		memstart_addr = round_up(memblock_end_of_DRAM() - linear_region_size,
					 ARM64_MEMSTART_ALIGN);
		memblock_remove(0, memstart_addr);
	}

	/*
	 * Apply the memory limit if it was set. Since the kernel may be loaded
	 * high up in memory, add back the kernel region that must be accessible
	 * via the linear mapping.
	 */
	if (memory_limit != (phys_addr_t)ULLONG_MAX) {
		memblock_mem_limit_remove_map(memory_limit);
		memblock_add(__pa(_text), (u64)(_end - _text));
	}

	if (IS_ENABLED(CONFIG_BLK_DEV_INITRD) && initrd_start) {
		/*
		 * Add back the memory we just removed if it results in the
		 * initrd to become inaccessible via the linear mapping.
		 * Otherwise, this is a no-op
		 */
		u64 base = initrd_start & PAGE_MASK;
		u64 size = PAGE_ALIGN(initrd_end) - base;

		/*
		 * We can only add back the initrd memory if we don't end up
		 * with more memory than we can address via the linear mapping.
		 * It is up to the bootloader to position the kernel and the
		 * initrd reasonably close to each other (i.e., within 32 GB of
		 * each other) so that all granule/#levels combinations can
		 * always access both.
		 */
		if (WARN(base < memblock_start_of_DRAM() ||
			 base + size > memblock_start_of_DRAM() +
				       linear_region_size,
			"initrd not fully accessible via the linear mapping -- please check your bootloader ...\n")) {
			initrd_start = 0;
		} else {
			memblock_remove(base, size); /* clear MEMBLOCK_ flags */
			memblock_add(base, size);
			memblock_reserve(base, size);
		}
	}

	if (IS_ENABLED(CONFIG_RANDOMIZE_BASE)) {
		extern u16 memstart_offset_seed;
		u64 range = linear_region_size -
			    (memblock_end_of_DRAM() - memblock_start_of_DRAM());

		/*
		 * If the size of the linear region exceeds, by a sufficient
		 * margin, the size of the region that the available physical
		 * memory spans, randomize the linear region as well.
		 */
		if (memstart_offset_seed > 0 && range >= ARM64_MEMSTART_ALIGN) {
			range = range / ARM64_MEMSTART_ALIGN + 1;
			memstart_addr -= ARM64_MEMSTART_ALIGN *
					 ((range * memstart_offset_seed) >> 16);
		}
	}

	/*
	 * Register the kernel text, kernel data, initrd, and initial
	 * pagetables with memblock.
	 */
	memblock_reserve(__pa(_text), _end - _text);
#ifdef CONFIG_BLK_DEV_INITRD
	if (initrd_start) {
		memblock_reserve(initrd_start, initrd_end - initrd_start);

		/* the generic initrd code expects virtual addresses */
		initrd_start = __phys_to_virt(initrd_start);
		initrd_end = __phys_to_virt(initrd_end);
	}
#endif

	early_init_fdt_scan_reserved_mem();

	/* 4GB maximum for 32-bit only capable devices */
	if (IS_ENABLED(CONFIG_ZONE_DMA))
		arm64_dma_phys_limit = max_zone_dma_phys();
	else
		arm64_dma_phys_limit = PHYS_MASK + 1;
	dma_contiguous_reserve(arm64_dma_phys_limit);

	memblock_allow_resize();
}

void __init bootmem_init(void)
{
	unsigned long min, max;

	min = PFN_UP(memblock_start_of_DRAM());
	max = PFN_DOWN(memblock_end_of_DRAM());

	early_memtest(min << PAGE_SHIFT, max << PAGE_SHIFT);

	max_pfn = max_low_pfn = max;

	arm64_numa_init();
	/*
	 * Sparsemem tries to allocate bootmem in memory_present(), so must be
	 * done after the fixed reservations.
	 */
	arm64_memory_present();

	sparse_init();
	zone_sizes_init(min, max);

	high_memory = __va((max << PAGE_SHIFT) - 1) + 1;
	memblock_dump_all();
}

#ifndef CONFIG_SPARSEMEM_VMEMMAP
static inline void free_memmap(unsigned long start_pfn, unsigned long end_pfn)
{
	struct page *start_pg, *end_pg;
	unsigned long pg, pgend;

	/*
	 * Convert start_pfn/end_pfn to a struct page pointer.
	 */
	start_pg = pfn_to_page(start_pfn - 1) + 1;
	end_pg = pfn_to_page(end_pfn - 1) + 1;

	/*
	 * Convert to physical addresses, and round start upwards and end
	 * downwards.
	 */
	pg = (unsigned long)PAGE_ALIGN(__pa(start_pg));
	pgend = (unsigned long)__pa(end_pg) & PAGE_MASK;

	/*
	 * If there are free pages between these, free the section of the
	 * memmap array.
	 */
	if (pg < pgend)
		free_bootmem(pg, pgend - pg);
}

/*
 * The mem_map array can get very big. Free the unused area of the memory map.
 */
static void __init free_unused_memmap(void)
{
	unsigned long start, prev_end = 0;
	struct memblock_region *reg;

	for_each_memblock(memory, reg) {
		start = __phys_to_pfn(reg->base);

#ifdef CONFIG_SPARSEMEM
		/*
		 * Take care not to free memmap entries that don't exist due
		 * to SPARSEMEM sections which aren't present.
		 */
		start = min(start, ALIGN(prev_end, PAGES_PER_SECTION));
#endif
		/*
		 * If we had a previous bank, and there is a space between the
		 * current bank and the previous, free it.
		 */
		if (prev_end && prev_end < start)
			free_memmap(prev_end, start);

		/*
		 * Align up here since the VM subsystem insists that the
		 * memmap entries are valid from the bank end aligned to
		 * MAX_ORDER_NR_PAGES.
		 */
		prev_end = ALIGN(__phys_to_pfn(reg->base + reg->size),
				 MAX_ORDER_NR_PAGES);
	}

#ifdef CONFIG_SPARSEMEM
	if (!IS_ALIGNED(prev_end, PAGES_PER_SECTION))
		free_memmap(prev_end, ALIGN(prev_end, PAGES_PER_SECTION));
#endif
}
#endif	/* !CONFIG_SPARSEMEM_VMEMMAP */

/*
 * 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)
{
	if (swiotlb_force || max_pfn > (arm64_dma_phys_limit >> PAGE_SHIFT))
		swiotlb_init(1);

	set_max_mapnr(pfn_to_page(max_pfn) - mem_map);

#ifndef CONFIG_SPARSEMEM_VMEMMAP
	free_unused_memmap();
#endif
	/* this will put all unused low memory onto the freelists */
	free_all_bootmem();

	mem_init_print_info(NULL);

#define MLK(b, t) b, t, ((t) - (b)) >> 10
#define MLM(b, t) b, t, ((t) - (b)) >> 20
#define MLG(b, t) b, t, ((t) - (b)) >> 30
#define MLK_ROUNDUP(b, t) b, t, DIV_ROUND_UP(((t) - (b)), SZ_1K)

	pr_notice("Virtual kernel memory layout:\n");
#ifdef CONFIG_KASAN
	pr_cont("    kasan   : 0x%16lx - 0x%16lx   (%6ld GB)\n",
		MLG(KASAN_SHADOW_START, KASAN_SHADOW_END));
#endif
	pr_cont("    modules : 0x%16lx - 0x%16lx   (%6ld MB)\n",
		MLM(MODULES_VADDR, MODULES_END));
	pr_cont("    vmalloc : 0x%16lx - 0x%16lx   (%6ld GB)\n",
		MLG(VMALLOC_START, VMALLOC_END));
	pr_cont("      .text : 0x%p" " - 0x%p" "   (%6ld KB)\n",
		MLK_ROUNDUP(_text, _etext));
	pr_cont("    .rodata : 0x%p" " - 0x%p" "   (%6ld KB)\n",
		MLK_ROUNDUP(__start_rodata, __init_begin));
	pr_cont("      .init : 0x%p" " - 0x%p" "   (%6ld KB)\n",
		MLK_ROUNDUP(__init_begin, __init_end));
	pr_cont("      .data : 0x%p" " - 0x%p" "   (%6ld KB)\n",
		MLK_ROUNDUP(_sdata, _edata));
	pr_cont("       .bss : 0x%p" " - 0x%p" "   (%6ld KB)\n",
		MLK_ROUNDUP(__bss_start, __bss_stop));
	pr_cont("    fixed   : 0x%16lx - 0x%16lx   (%6ld KB)\n",
		MLK(FIXADDR_START, FIXADDR_TOP));
	pr_cont("    PCI I/O : 0x%16lx - 0x%16lx   (%6ld MB)\n",
		MLM(PCI_IO_START, PCI_IO_END));
#ifdef CONFIG_SPARSEMEM_VMEMMAP
	pr_cont("    vmemmap : 0x%16lx - 0x%16lx   (%6ld GB maximum)\n",
		MLG(VMEMMAP_START, VMEMMAP_START + VMEMMAP_SIZE));
	pr_cont("              0x%16lx - 0x%16lx   (%6ld MB actual)\n",
		MLM((unsigned long)phys_to_page(memblock_start_of_DRAM()),
		    (unsigned long)virt_to_page(high_memory)));
#endif
	pr_cont("    memory  : 0x%16lx - 0x%16lx   (%6ld MB)\n",
		MLM(__phys_to_virt(memblock_start_of_DRAM()),
		    (unsigned long)high_memory));

#undef MLK
#undef MLM
#undef MLK_ROUNDUP

	/*
	 * Check boundaries twice: Some fundamental inconsistencies can be
	 * detected at build time already.
	 */
#ifdef CONFIG_COMPAT
	BUILD_BUG_ON(TASK_SIZE_32			> TASK_SIZE_64);
#endif

	/*
	 * Make sure we chose the upper bound of sizeof(struct page)
	 * correctly.
	 */
	BUILD_BUG_ON(sizeof(struct page) > (1 << STRUCT_PAGE_MAX_SHIFT));

	if (PAGE_SIZE >= 16384 && get_num_physpages() <= 128) {
		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;
	}
}

void free_initmem(void)
{
	free_reserved_area(__va(__pa(__init_begin)), __va(__pa(__init_end)),
			   0, "unused kernel");
	/*
	 * Unmap the __init region but leave the VM area in place. This
	 * prevents the region from being reused for kernel modules, which
	 * is not supported by kallsyms.
	 */
	unmap_kernel_range((u64)__init_begin, (u64)(__init_end - __init_begin));
}

#ifdef CONFIG_BLK_DEV_INITRD

static int keep_initrd __initdata;

void __init free_initrd_mem(unsigned long start, unsigned long end)
{
	if (!keep_initrd)
		free_reserved_area((void *)start, (void *)end, 0, "initrd");
}

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

__setup("keepinitrd", keepinitrd_setup);
#endif

/*
 * Dump out memory limit information on panic.
 */
static int dump_mem_limit(struct notifier_block *self, unsigned long v, void *p)
{
	if (memory_limit != (phys_addr_t)ULLONG_MAX) {
		pr_emerg("Memory Limit: %llu MB\n", memory_limit >> 20);
	} else {
		pr_emerg("Memory Limit: none\n");
	}
	return 0;
}

static struct notifier_block mem_limit_notifier = {
	.notifier_call = dump_mem_limit,
};

static int __init register_mem_limit_dumper(void)
{
	atomic_notifier_chain_register(&panic_notifier_list,
				       &mem_limit_notifier);
	return 0;
}
__initcall(register_mem_limit_dumper);
Commit	Line	Data
c1cc1552 CM	1	/*
	2	* Based on arch/arm/mm/init.c
	3	*
	4	* Copyright (C) 1995-2005 Russell King
	5	* Copyright (C) 2012 ARM Ltd.
	6	*
	7	* This program is free software; you can redistribute it and/or modify
	8	* it under the terms of the GNU General Public License version 2 as
	9	* published by the Free Software Foundation.
	10	*
	11	* This program is distributed in the hope that it will be useful,
	12	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	14	* GNU General Public License for more details.
	15	*
	16	* You should have received a copy of the GNU General Public License
	17	* along with this program. If not, see <http://www.gnu.org/licenses/>.
	18	*/
	19
	20	#include <linux/kernel.h>
	21	#include <linux/export.h>
	22	#include <linux/errno.h>
	23	#include <linux/swap.h>
	24	#include <linux/init.h>
	25	#include <linux/bootmem.h>
5a9e3e15	26	#include <linux/cache.h>
c1cc1552 CM	27	#include <linux/mman.h>
	28	#include <linux/nodemask.h>
	29	#include <linux/initrd.h>
	30	#include <linux/gfp.h>
	31	#include <linux/memblock.h>
	32	#include <linux/sort.h>
	33	#include <linux/of_fdt.h>
19e7640d	34	#include <linux/dma-mapping.h>
6ac2104d	35	#include <linux/dma-contiguous.h>
86c8b27a	36	#include <linux/efi.h>
a1e50a82	37	#include <linux/swiotlb.h>
dae8c235	38	#include <linux/vmalloc.h>
c1cc1552	39
a7f8de16	40	#include <asm/boot.h>
08375198	41	#include <asm/fixmap.h>
f9040773	42	#include <asm/kasan.h>
a7f8de16	43	#include <asm/kernel-pgtable.h>
aa03c428	44	#include <asm/memory.h>
1a2db300	45	#include <asm/numa.h>
c1cc1552 CM	46	#include <asm/sections.h>
	47	#include <asm/setup.h>
	48	#include <asm/sizes.h>
	49	#include <asm/tlb.h>
e039ee4e	50	#include <asm/alternative.h>
c1cc1552	51
a7f8de16 AB	52	/*
	53	* We need to be able to catch inadvertent references to memstart_addr
	54	* that occur (potentially in generic code) before arm64_memblock_init()
	55	* executes, which assigns it its actual value. So use a default value
	56	* that cannot be mistaken for a real physical address.
	57	*/
5a9e3e15 JZ	58	s64 memstart_addr __ro_after_init = -1;
5a9e3e15 JZ	59	phys_addr_t arm64_dma_phys_limit __ro_after_init;
c1cc1552	60
ec2eaa73	61	#ifdef CONFIG_BLK_DEV_INITRD
c1cc1552 CM	62	static int __init early_initrd(char *p)
	63	{
	64	unsigned long start, size;
	65	char *endp;
	66
	67	start = memparse(p, &endp);
	68	if (*endp == ',') {
	69	size = memparse(endp + 1, NULL);
	70
a89dea58 AB	71	initrd_start = start;
a89dea58 AB	72	initrd_end = start + size;
c1cc1552 CM	73	}
	74	return 0;
	75	}
	76	early_param("initrd", early_initrd);
ec2eaa73	77	#endif
c1cc1552	78
d50314a6 CM	79	/*
	80	* Return the maximum physical address for ZONE_DMA (DMA_BIT_MASK(32)). It
	81	* currently assumes that for memory starting above 4G, 32-bit devices will
	82	* use a DMA offset.
	83	*/
a7c61a34	84	static phys_addr_t __init max_zone_dma_phys(void)
d50314a6 CM	85	{
	86	phys_addr_t offset = memblock_start_of_DRAM() & GENMASK_ULL(63, 32);
	87	return min(offset + (1ULL << 32), memblock_end_of_DRAM());
	88	}
	89
1a2db300 GK	90	#ifdef CONFIG_NUMA
	91
	92	static void __init zone_sizes_init(unsigned long min, unsigned long max)
	93	{
	94	unsigned long max_zone_pfns[MAX_NR_ZONES] = {0};
	95
	96	if (IS_ENABLED(CONFIG_ZONE_DMA))
	97	max_zone_pfns[ZONE_DMA] = PFN_DOWN(max_zone_dma_phys());
	98	max_zone_pfns[ZONE_NORMAL] = max;
	99
	100	free_area_init_nodes(max_zone_pfns);
	101	}
	102
	103	#else
	104
c1cc1552 CM	105	static void __init zone_sizes_init(unsigned long min, unsigned long max)
	106	{
	107	struct memblock_region *reg;
	108	unsigned long zone_size[MAX_NR_ZONES], zhole_size[MAX_NR_ZONES];
19e7640d	109	unsigned long max_dma = min;
c1cc1552 CM	110
	111	memset(zone_size, 0, sizeof(zone_size));
	112
c1cc1552	113	/* 4GB maximum for 32-bit only capable devices */
86a5906e RM	114	#ifdef CONFIG_ZONE_DMA
	115	max_dma = PFN_DOWN(arm64_dma_phys_limit);
	116	zone_size[ZONE_DMA] = max_dma - min;
	117	#endif
19e7640d	118	zone_size[ZONE_NORMAL] = max - max_dma;
c1cc1552 CM	119
	120	memcpy(zhole_size, zone_size, sizeof(zhole_size));
	121
	122	for_each_memblock(memory, reg) {
	123	unsigned long start = memblock_region_memory_base_pfn(reg);
	124	unsigned long end = memblock_region_memory_end_pfn(reg);
	125
	126	if (start >= max)
	127	continue;
19e7640d	128
86a5906e RM	129	#ifdef CONFIG_ZONE_DMA
86a5906e RM	130	if (start < max_dma) {
19e7640d CM	131	unsigned long dma_end = min(end, max_dma);
19e7640d CM	132	zhole_size[ZONE_DMA] -= dma_end - start;
c1cc1552	133	}
86a5906e	134	#endif
19e7640d	135	if (end > max_dma) {
c1cc1552	136	unsigned long normal_end = min(end, max);
19e7640d	137	unsigned long normal_start = max(start, max_dma);
c1cc1552 CM	138	zhole_size[ZONE_NORMAL] -= normal_end - normal_start;
	139	}
	140	}
	141
	142	free_area_init_node(0, zone_size, min, zhole_size);
	143	}
	144
1a2db300 GK	145	#endif /* CONFIG_NUMA */
1a2db300 GK	146
c1cc1552 CM	147	#ifdef CONFIG_HAVE_ARCH_PFN_VALID
	148	int pfn_valid(unsigned long pfn)
	149	{
68709f45	150	return memblock_is_map_memory(pfn << PAGE_SHIFT);
c1cc1552 CM	151	}
	152	EXPORT_SYMBOL(pfn_valid);
	153	#endif
	154
	155	#ifndef CONFIG_SPARSEMEM
a7c61a34	156	static void __init arm64_memory_present(void)
c1cc1552 CM	157	{
	158	}
	159	#else
a7c61a34	160	static void __init arm64_memory_present(void)
c1cc1552 CM	161	{
	162	struct memblock_region *reg;
	163
1a2db300	164	for_each_memblock(memory, reg) {
ea2cbee3 MR	165	int nid = memblock_get_region_node(reg);
ea2cbee3 MR	166
1a2db300 GK	167	memory_present(nid, memblock_region_memory_base_pfn(reg),
	168	memblock_region_memory_end_pfn(reg));
	169	}
c1cc1552 CM	170	}
	171	#endif
	172
6083fe74 MR	173	static phys_addr_t memory_limit = (phys_addr_t)ULLONG_MAX;
	174
	175	/*
	176	* Limit the memory size that was specified via FDT.
	177	*/
	178	static int __init early_mem(char *p)
	179	{
	180	if (!p)
	181	return 1;
	182
	183	memory_limit = memparse(p, &p) & PAGE_MASK;
	184	pr_notice("Memory limited to %lldMB\n", memory_limit >> 20);
	185
	186	return 0;
	187	}
	188	early_param("mem", early_mem);
	189
c1cc1552 CM	190	void __init arm64_memblock_init(void)
c1cc1552 CM	191	{
a7f8de16 AB	192	const s64 linear_region_size = -(s64)PAGE_OFFSET;
a7f8de16 AB	193
6d2aa549 AB	194	/*
	195	* Ensure that the linear region takes up exactly half of the kernel
	196	* virtual address space. This way, we can distinguish a linear address
	197	* from a kernel/module/vmalloc address by testing a single bit.
	198	*/
	199	BUILD_BUG_ON(linear_region_size != BIT(VA_BITS - 1));
	200
a7f8de16 AB	201	/*
	202	* Select a suitable value for the base of physical memory.
	203	*/
	204	memstart_addr = round_down(memblock_start_of_DRAM(),
	205	ARM64_MEMSTART_ALIGN);
	206
	207	/*
	208	* Remove the memory that we will not be able to cover with the
	209	* linear mapping. Take care not to clip the kernel which may be
	210	* high in memory.
	211	*/
020d044f	212	memblock_remove(max_t(u64, memstart_addr + linear_region_size, __pa(_end)),
a7f8de16	213	ULLONG_MAX);
2958987f AB	214	if (memstart_addr + linear_region_size < memblock_end_of_DRAM()) {
	215	/* ensure that memstart_addr remains sufficiently aligned */
	216	memstart_addr = round_up(memblock_end_of_DRAM() - linear_region_size,
	217	ARM64_MEMSTART_ALIGN);
	218	memblock_remove(0, memstart_addr);
	219	}
a7f8de16 AB	220
	221	/*
	222	* Apply the memory limit if it was set. Since the kernel may be loaded
	223	* high up in memory, add back the kernel region that must be accessible
	224	* via the linear mapping.
	225	*/
	226	if (memory_limit != (phys_addr_t)ULLONG_MAX) {
cb0a6502	227	memblock_mem_limit_remove_map(memory_limit);
a7f8de16 AB	228	memblock_add(__pa(_text), (u64)(_end - _text));
a7f8de16 AB	229	}
6083fe74	230
177e15f0 AB	231	if (IS_ENABLED(CONFIG_BLK_DEV_INITRD) && initrd_start) {
	232	/*
	233	* Add back the memory we just removed if it results in the
	234	* initrd to become inaccessible via the linear mapping.
	235	* Otherwise, this is a no-op
	236	*/
	237	u64 base = initrd_start & PAGE_MASK;
	238	u64 size = PAGE_ALIGN(initrd_end) - base;
	239
	240	/*
	241	* We can only add back the initrd memory if we don't end up
	242	* with more memory than we can address via the linear mapping.
	243	* It is up to the bootloader to position the kernel and the
	244	* initrd reasonably close to each other (i.e., within 32 GB of
	245	* each other) so that all granule/#levels combinations can
	246	* always access both.
	247	*/
	248	if (WARN(base < memblock_start_of_DRAM() \|\|
	249	base + size > memblock_start_of_DRAM() +
	250	linear_region_size,
	251	"initrd not fully accessible via the linear mapping -- please check your bootloader ...\n")) {
	252	initrd_start = 0;
	253	} else {
	254	memblock_remove(base, size); /* clear MEMBLOCK_ flags */
	255	memblock_add(base, size);
	256	memblock_reserve(base, size);
	257	}
	258	}
	259
c031a421 AB	260	if (IS_ENABLED(CONFIG_RANDOMIZE_BASE)) {
	261	extern u16 memstart_offset_seed;
	262	u64 range = linear_region_size -
	263	(memblock_end_of_DRAM() - memblock_start_of_DRAM());
	264
	265	/*
	266	* If the size of the linear region exceeds, by a sufficient
	267	* margin, the size of the region that the available physical
	268	* memory spans, randomize the linear region as well.
	269	*/
	270	if (memstart_offset_seed > 0 && range >= ARM64_MEMSTART_ALIGN) {
	271	range = range / ARM64_MEMSTART_ALIGN + 1;
	272	memstart_addr -= ARM64_MEMSTART_ALIGN *
	273	((range * memstart_offset_seed) >> 16);
	274	}
	275	}
6083fe74	276
bd00cd5f MR	277	/*
	278	* Register the kernel text, kernel data, initrd, and initial
	279	* pagetables with memblock.
	280	*/
c1cc1552 CM	281	memblock_reserve(__pa(_text), _end - _text);
c1cc1552 CM	282	#ifdef CONFIG_BLK_DEV_INITRD
a89dea58 AB	283	if (initrd_start) {
	284	memblock_reserve(initrd_start, initrd_end - initrd_start);
	285
	286	/* the generic initrd code expects virtual addresses */
	287	initrd_start = __phys_to_virt(initrd_start);
	288	initrd_end = __phys_to_virt(initrd_end);
	289	}
c1cc1552 CM	290	#endif
c1cc1552 CM	291
0ceac9e0	292	early_init_fdt_scan_reserved_mem();
2d5a5612 CM	293
	294	/* 4GB maximum for 32-bit only capable devices */
	295	if (IS_ENABLED(CONFIG_ZONE_DMA))
a1e50a82 CM	296	arm64_dma_phys_limit = max_zone_dma_phys();
	297	else
	298	arm64_dma_phys_limit = PHYS_MASK + 1;
	299	dma_contiguous_reserve(arm64_dma_phys_limit);
6ac2104d	300
c1cc1552	301	memblock_allow_resize();
c1cc1552 CM	302	}
	303
	304	void __init bootmem_init(void)
	305	{
	306	unsigned long min, max;
	307
	308	min = PFN_UP(memblock_start_of_DRAM());
	309	max = PFN_DOWN(memblock_end_of_DRAM());
	310
36dd9086 VM	311	early_memtest(min << PAGE_SHIFT, max << PAGE_SHIFT);
36dd9086 VM	312
1a2db300 GK	313	max_pfn = max_low_pfn = max;
	314
	315	arm64_numa_init();
c1cc1552 CM	316	/*
	317	* Sparsemem tries to allocate bootmem in memory_present(), so must be
	318	* done after the fixed reservations.
	319	*/
	320	arm64_memory_present();
	321
	322	sparse_init();
	323	zone_sizes_init(min, max);
	324
	325	high_memory = __va((max << PAGE_SHIFT) - 1) + 1;
1a2db300	326	memblock_dump_all();
c1cc1552 CM	327	}
c1cc1552 CM	328
c1cc1552 CM	329	#ifndef CONFIG_SPARSEMEM_VMEMMAP
	330	static inline void free_memmap(unsigned long start_pfn, unsigned long end_pfn)
	331	{
	332	struct page start_pg, end_pg;
	333	unsigned long pg, pgend;
	334
	335	/*
	336	* Convert start_pfn/end_pfn to a struct page pointer.
	337	*/
	338	start_pg = pfn_to_page(start_pfn - 1) + 1;
	339	end_pg = pfn_to_page(end_pfn - 1) + 1;
	340
	341	/*
	342	* Convert to physical addresses, and round start upwards and end
	343	* downwards.
	344	*/
	345	pg = (unsigned long)PAGE_ALIGN(__pa(start_pg));
	346	pgend = (unsigned long)__pa(end_pg) & PAGE_MASK;
	347
	348	/*
	349	* If there are free pages between these, free the section of the
	350	* memmap array.
	351	*/
	352	if (pg < pgend)
	353	free_bootmem(pg, pgend - pg);
	354	}
	355
	356	/*
	357	* The mem_map array can get very big. Free the unused area of the memory map.
	358	*/
	359	static void __init free_unused_memmap(void)
	360	{
	361	unsigned long start, prev_end = 0;
	362	struct memblock_region *reg;
	363
	364	for_each_memblock(memory, reg) {
	365	start = __phys_to_pfn(reg->base);
	366
	367	#ifdef CONFIG_SPARSEMEM
	368	/*
	369	* Take care not to free memmap entries that don't exist due
	370	* to SPARSEMEM sections which aren't present.
	371	*/
	372	start = min(start, ALIGN(prev_end, PAGES_PER_SECTION));
	373	#endif
	374	/*
	375	* If we had a previous bank, and there is a space between the
	376	* current bank and the previous, free it.
	377	*/
	378	if (prev_end && prev_end < start)
	379	free_memmap(prev_end, start);
	380
	381	/*
	382	* Align up here since the VM subsystem insists that the
	383	* memmap entries are valid from the bank end aligned to
	384	* MAX_ORDER_NR_PAGES.
	385	*/
b9bcc919	386	prev_end = ALIGN(__phys_to_pfn(reg->base + reg->size),
c1cc1552 CM	387	MAX_ORDER_NR_PAGES);
	388	}
	389
	390	#ifdef CONFIG_SPARSEMEM
	391	if (!IS_ALIGNED(prev_end, PAGES_PER_SECTION))
	392	free_memmap(prev_end, ALIGN(prev_end, PAGES_PER_SECTION));
	393	#endif
	394	}
	395	#endif /* !CONFIG_SPARSEMEM_VMEMMAP */
	396
	397	/*
	398	* mem_init() marks the free areas in the mem_map and tells us how much memory
	399	* is free. This is done after various parts of the system have claimed their
	400	* memory after the kernel image.
	401	*/
	402	void __init mem_init(void)
	403	{
b67a8b29 JZ	404	if (swiotlb_force \|\| max_pfn > (arm64_dma_phys_limit >> PAGE_SHIFT))
b67a8b29 JZ	405	swiotlb_init(1);
a1e50a82	406
a6583c7c	407	set_max_mapnr(pfn_to_page(max_pfn) - mem_map);
c1cc1552 CM	408
c1cc1552 CM	409	#ifndef CONFIG_SPARSEMEM_VMEMMAP
c1cc1552 CM	410	free_unused_memmap();
c1cc1552 CM	411	#endif
bee4ebd1	412	/* this will put all unused low memory onto the freelists */
0c988534	413	free_all_bootmem();
c1cc1552	414
6879ea83	415	mem_init_print_info(NULL);
c1cc1552 CM	416
	417	#define MLK(b, t) b, t, ((t) - (b)) >> 10
	418	#define MLM(b, t) b, t, ((t) - (b)) >> 20
08375198	419	#define MLG(b, t) b, t, ((t) - (b)) >> 30
c1cc1552 CM	420	#define MLK_ROUNDUP(b, t) b, t, DIV_ROUND_UP(((t) - (b)), SZ_1K)
c1cc1552 CM	421
f09f1bac	422	pr_notice("Virtual kernel memory layout:\n");
ee7f881b	423	#ifdef CONFIG_KASAN
f09f1bac CM	424	pr_cont(" kasan : 0x%16lx - 0x%16lx (%6ld GB)\n",
f09f1bac CM	425	MLG(KASAN_SHADOW_START, KASAN_SHADOW_END));
ee7f881b	426	#endif
f09f1bac CM	427	pr_cont(" modules : 0x%16lx - 0x%16lx (%6ld MB)\n",
	428	MLM(MODULES_VADDR, MODULES_END));
	429	pr_cont(" vmalloc : 0x%16lx - 0x%16lx (%6ld GB)\n",
	430	MLG(VMALLOC_START, VMALLOC_END));
d32351c8	431	pr_cont(" .text : 0x%p" " - 0x%p" " (%6ld KB)\n",
9fdc14c5	432	MLK_ROUNDUP(_text, _etext));
d32351c8	433	pr_cont(" .rodata : 0x%p" " - 0x%p" " (%6ld KB)\n",
9fdc14c5	434	MLK_ROUNDUP(__start_rodata, __init_begin));
d32351c8 KW	435	pr_cont(" .init : 0x%p" " - 0x%p" " (%6ld KB)\n",
	436	MLK_ROUNDUP(__init_begin, __init_end));
	437	pr_cont(" .data : 0x%p" " - 0x%p" " (%6ld KB)\n",
f09f1bac	438	MLK_ROUNDUP(_sdata, _edata));
9974723e KW	439	pr_cont(" .bss : 0x%p" " - 0x%p" " (%6ld KB)\n",
9974723e KW	440	MLK_ROUNDUP(__bss_start, __bss_stop));
3e1907d5 AB	441	pr_cont(" fixed : 0x%16lx - 0x%16lx (%6ld KB)\n",
	442	MLK(FIXADDR_START, FIXADDR_TOP));
	443	pr_cont(" PCI I/O : 0x%16lx - 0x%16lx (%6ld MB)\n",
	444	MLM(PCI_IO_START, PCI_IO_END));
c1cc1552	445	#ifdef CONFIG_SPARSEMEM_VMEMMAP
d32351c8 KW	446	pr_cont(" vmemmap : 0x%16lx - 0x%16lx (%6ld GB maximum)\n",
	447	MLG(VMEMMAP_START, VMEMMAP_START + VMEMMAP_SIZE));
	448	pr_cont(" 0x%16lx - 0x%16lx (%6ld MB actual)\n",
f09f1bac CM	449	MLM((unsigned long)phys_to_page(memblock_start_of_DRAM()),
f09f1bac CM	450	(unsigned long)virt_to_page(high_memory)));
c1cc1552	451	#endif
f09f1bac CM	452	pr_cont(" memory : 0x%16lx - 0x%16lx (%6ld MB)\n",
	453	MLM(__phys_to_virt(memblock_start_of_DRAM()),
	454	(unsigned long)high_memory));
c1cc1552 CM	455
	456	#undef MLK
	457	#undef MLM
	458	#undef MLK_ROUNDUP
	459
	460	/*
	461	* Check boundaries twice: Some fundamental inconsistencies can be
	462	* detected at build time already.
	463	*/
	464	#ifdef CONFIG_COMPAT
	465	BUILD_BUG_ON(TASK_SIZE_32 > TASK_SIZE_64);
	466	#endif
c1cc1552	467
3e1907d5 AB	468	/*
	469	* Make sure we chose the upper bound of sizeof(struct page)
	470	* correctly.
	471	*/
	472	BUILD_BUG_ON(sizeof(struct page) > (1 << STRUCT_PAGE_MAX_SHIFT));
	473
bee4ebd1	474	if (PAGE_SIZE >= 16384 && get_num_physpages() <= 128) {
c1cc1552 CM	475	extern int sysctl_overcommit_memory;
	476	/*
	477	* On a machine this small we won't get anywhere without
	478	* overcommit, so turn it on by default.
	479	*/
	480	sysctl_overcommit_memory = OVERCOMMIT_ALWAYS;
	481	}
	482	}
	483
	484	void free_initmem(void)
	485	{
d386825c AB	486	free_reserved_area(__va(__pa(__init_begin)), __va(__pa(__init_end)),
d386825c AB	487	0, "unused kernel");
dae8c235 KW	488	/*
	489	* Unmap the __init region but leave the VM area in place. This
	490	* prevents the region from being reused for kernel modules, which
	491	* is not supported by kallsyms.
	492	*/
	493	unmap_kernel_range((u64)__init_begin, (u64)(__init_end - __init_begin));
c1cc1552 CM	494	}
	495
	496	#ifdef CONFIG_BLK_DEV_INITRD
	497
662ba3db	498	static int keep_initrd __initdata;
c1cc1552	499
662ba3db	500	void __init free_initrd_mem(unsigned long start, unsigned long end)
c1cc1552	501	{
0145058c	502	if (!keep_initrd)
9af5b807	503	free_reserved_area((void )start, (void )end, 0, "initrd");
c1cc1552 CM	504	}
	505
	506	static int __init keepinitrd_setup(char *__unused)
	507	{
	508	keep_initrd = 1;
	509	return 1;
	510	}
	511
	512	__setup("keepinitrd", keepinitrd_setup);
	513	#endif
a7f8de16 AB	514
	515	/*
	516	* Dump out memory limit information on panic.
	517	*/
	518	static int dump_mem_limit(struct notifier_block self, unsigned long v, void p)
	519	{
	520	if (memory_limit != (phys_addr_t)ULLONG_MAX) {
	521	pr_emerg("Memory Limit: %llu MB\n", memory_limit >> 20);
	522	} else {
	523	pr_emerg("Memory Limit: none\n");
	524	}
	525	return 0;
	526	}
	527
	528	static struct notifier_block mem_limit_notifier = {
	529	.notifier_call = dump_mem_limit,
	530	};
	531
	532	static int __init register_mem_limit_dumper(void)
	533	{
	534	atomic_notifier_chain_register(&panic_notifier_list,
	535	&mem_limit_notifier);
	536	return 0;
	537	}
	538	__initcall(register_mem_limit_dumper);