[linux-2.6-block.git] / arch / powerpc / mm / ptdump / ptdump.c

/*
 * Copyright 2016, Rashmica Gupta, IBM Corp.
 *
 * This traverses the kernel pagetables and dumps the
 * information about the used sections of memory to
 * /sys/kernel/debug/kernel_pagetables.
 *
 * Derived from the arm64 implementation:
 * Copyright (c) 2014, The Linux Foundation, Laura Abbott.
 * (C) Copyright 2008 Intel Corporation, Arjan van de Ven.
 *
 * 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; version 2
 * of the License.
 */
#include <linux/debugfs.h>
#include <linux/fs.h>
#include <linux/hugetlb.h>
#include <linux/io.h>
#include <linux/mm.h>
#include <linux/highmem.h>
#include <linux/sched.h>
#include <linux/seq_file.h>
#include <asm/fixmap.h>
#include <asm/pgtable.h>
#include <linux/const.h>
#include <asm/page.h>
#include <asm/pgalloc.h>

#include "ptdump.h"

#ifdef CONFIG_PPC32
#define KERN_VIRT_START	0
#endif

/*
 * To visualise what is happening,
 *
 *  - PTRS_PER_P** = how many entries there are in the corresponding P**
 *  - P**_SHIFT = how many bits of the address we use to index into the
 * corresponding P**
 *  - P**_SIZE is how much memory we can access through the table - not the
 * size of the table itself.
 * P**={PGD, PUD, PMD, PTE}
 *
 *
 * Each entry of the PGD points to a PUD. Each entry of a PUD points to a
 * PMD. Each entry of a PMD points to a PTE. And every PTE entry points to
 * a page.
 *
 * In the case where there are only 3 levels, the PUD is folded into the
 * PGD: every PUD has only one entry which points to the PMD.
 *
 * The page dumper groups page table entries of the same type into a single
 * description. It uses pg_state to track the range information while
 * iterating over the PTE entries. When the continuity is broken it then
 * dumps out a description of the range - ie PTEs that are virtually contiguous
 * with the same PTE flags are chunked together. This is to make it clear how
 * different areas of the kernel virtual memory are used.
 *
 */
struct pg_state {
	struct seq_file *seq;
	const struct addr_marker *marker;
	unsigned long start_address;
	unsigned long start_pa;
	unsigned long last_pa;
	unsigned int level;
	u64 current_flags;
};

struct addr_marker {
	unsigned long start_address;
	const char *name;
};

static struct addr_marker address_markers[] = {
	{ 0,	"Start of kernel VM" },
	{ 0,	"vmalloc() Area" },
	{ 0,	"vmalloc() End" },
#ifdef CONFIG_PPC64
	{ 0,	"isa I/O start" },
	{ 0,	"isa I/O end" },
	{ 0,	"phb I/O start" },
	{ 0,	"phb I/O end" },
	{ 0,	"I/O remap start" },
	{ 0,	"I/O remap end" },
	{ 0,	"vmemmap start" },
#else
	{ 0,	"Early I/O remap start" },
	{ 0,	"Early I/O remap end" },
#ifdef CONFIG_NOT_COHERENT_CACHE
	{ 0,	"Consistent mem start" },
	{ 0,	"Consistent mem end" },
#endif
#ifdef CONFIG_HIGHMEM
	{ 0,	"Highmem PTEs start" },
	{ 0,	"Highmem PTEs end" },
#endif
	{ 0,	"Fixmap start" },
	{ 0,	"Fixmap end" },
#endif
#ifdef CONFIG_KASAN
	{ 0,	"kasan shadow mem start" },
	{ 0,	"kasan shadow mem end" },
#endif
	{ -1,	NULL },
};

#define pt_dump_seq_printf(m, fmt, args...)	\
({						\
	if (m)					\
		seq_printf(m, fmt, ##args);	\
})

#define pt_dump_seq_putc(m, c)		\
({					\
	if (m)				\
		seq_putc(m, c);		\
})

static void dump_flag_info(struct pg_state *st, const struct flag_info
		*flag, u64 pte, int num)
{
	unsigned int i;

	for (i = 0; i < num; i++, flag++) {
		const char *s = NULL;
		u64 val;

		/* flag not defined so don't check it */
		if (flag->mask == 0)
			continue;
		/* Some 'flags' are actually values */
		if (flag->is_val) {
			val = pte & flag->val;
			if (flag->shift)
				val = val >> flag->shift;
			pt_dump_seq_printf(st->seq, "  %s:%llx", flag->set, val);
		} else {
			if ((pte & flag->mask) == flag->val)
				s = flag->set;
			else
				s = flag->clear;
			if (s)
				pt_dump_seq_printf(st->seq, "  %s", s);
		}
		st->current_flags &= ~flag->mask;
	}
	if (st->current_flags != 0)
		pt_dump_seq_printf(st->seq, "  unknown flags:%llx", st->current_flags);
}

static void dump_addr(struct pg_state *st, unsigned long addr)
{
	static const char units[] = "KMGTPE";
	const char *unit = units;
	unsigned long delta;

#ifdef CONFIG_PPC64
#define REG		"0x%016lx"
#else
#define REG		"0x%08lx"
#endif

	pt_dump_seq_printf(st->seq, REG "-" REG " ", st->start_address, addr - 1);
	if (st->start_pa == st->last_pa && st->start_address + PAGE_SIZE != addr) {
		pt_dump_seq_printf(st->seq, "[" REG "]", st->start_pa);
		delta = PAGE_SIZE >> 10;
	} else {
		pt_dump_seq_printf(st->seq, " " REG " ", st->start_pa);
		delta = (addr - st->start_address) >> 10;
	}
	/* Work out what appropriate unit to use */
	while (!(delta & 1023) && unit[1]) {
		delta >>= 10;
		unit++;
	}
	pt_dump_seq_printf(st->seq, "%9lu%c", delta, *unit);

}

static void note_page(struct pg_state *st, unsigned long addr,
	       unsigned int level, u64 val)
{
	u64 flag = val & pg_level[level].mask;
	u64 pa = val & PTE_RPN_MASK;

	/* At first no level is set */
	if (!st->level) {
		st->level = level;
		st->current_flags = flag;
		st->start_address = addr;
		st->start_pa = pa;
		st->last_pa = pa;
		pt_dump_seq_printf(st->seq, "---[ %s ]---\n", st->marker->name);
	/*
	 * Dump the section of virtual memory when:
	 *   - the PTE flags from one entry to the next differs.
	 *   - we change levels in the tree.
	 *   - the address is in a different section of memory and is thus
	 *   used for a different purpose, regardless of the flags.
	 *   - the pa of this page is not adjacent to the last inspected page
	 */
	} else if (flag != st->current_flags || level != st->level ||
		   addr >= st->marker[1].start_address ||
		   (pa != st->last_pa + PAGE_SIZE &&
		    (pa != st->start_pa || st->start_pa != st->last_pa))) {

		/* Check the PTE flags */
		if (st->current_flags) {
			dump_addr(st, addr);

			/* Dump all the flags */
			if (pg_level[st->level].flag)
				dump_flag_info(st, pg_level[st->level].flag,
					  st->current_flags,
					  pg_level[st->level].num);

			pt_dump_seq_putc(st->seq, '\n');
		}

		/*
		 * Address indicates we have passed the end of the
		 * current section of virtual memory
		 */
		while (addr >= st->marker[1].start_address) {
			st->marker++;
			pt_dump_seq_printf(st->seq, "---[ %s ]---\n", st->marker->name);
		}
		st->start_address = addr;
		st->start_pa = pa;
		st->last_pa = pa;
		st->current_flags = flag;
		st->level = level;
	} else {
		st->last_pa = pa;
	}
}

static void walk_pte(struct pg_state *st, pmd_t *pmd, unsigned long start)
{
	pte_t *pte = pte_offset_kernel(pmd, 0);
	unsigned long addr;
	unsigned int i;

	for (i = 0; i < PTRS_PER_PTE; i++, pte++) {
		addr = start + i * PAGE_SIZE;
		note_page(st, addr, 4, pte_val(*pte));

	}
}

static void walk_pmd(struct pg_state *st, pud_t *pud, unsigned long start)
{
	pmd_t *pmd = pmd_offset(pud, 0);
	unsigned long addr;
	unsigned int i;

	for (i = 0; i < PTRS_PER_PMD; i++, pmd++) {
		addr = start + i * PMD_SIZE;
		if (!pmd_none(*pmd) && !pmd_huge(*pmd))
			/* pmd exists */
			walk_pte(st, pmd, addr);
		else
			note_page(st, addr, 3, pmd_val(*pmd));
	}
}

static void walk_pud(struct pg_state *st, pgd_t *pgd, unsigned long start)
{
	pud_t *pud = pud_offset(pgd, 0);
	unsigned long addr;
	unsigned int i;

	for (i = 0; i < PTRS_PER_PUD; i++, pud++) {
		addr = start + i * PUD_SIZE;
		if (!pud_none(*pud) && !pud_huge(*pud))
			/* pud exists */
			walk_pmd(st, pud, addr);
		else
			note_page(st, addr, 2, pud_val(*pud));
	}
}

static void walk_pagetables(struct pg_state *st)
{
	pgd_t *pgd = pgd_offset_k(0UL);
	unsigned int i;
	unsigned long addr;

	addr = st->start_address;

	/*
	 * Traverse the linux pagetable structure and dump pages that are in
	 * the hash pagetable.
	 */
	for (i = 0; i < PTRS_PER_PGD; i++, pgd++, addr += PGDIR_SIZE) {
		if (!pgd_none(*pgd) && !pgd_huge(*pgd))
			/* pgd exists */
			walk_pud(st, pgd, addr);
		else
			note_page(st, addr, 1, pgd_val(*pgd));
	}
}

static void populate_markers(void)
{
	int i = 0;

	address_markers[i++].start_address = PAGE_OFFSET;
	address_markers[i++].start_address = VMALLOC_START;
	address_markers[i++].start_address = VMALLOC_END;
#ifdef CONFIG_PPC64
	address_markers[i++].start_address = ISA_IO_BASE;
	address_markers[i++].start_address = ISA_IO_END;
	address_markers[i++].start_address = PHB_IO_BASE;
	address_markers[i++].start_address = PHB_IO_END;
	address_markers[i++].start_address = IOREMAP_BASE;
	address_markers[i++].start_address = IOREMAP_END;
	/* What is the ifdef about? */
#ifdef CONFIG_PPC_BOOK3S_64
	address_markers[i++].start_address =  H_VMEMMAP_START;
#else
	address_markers[i++].start_address =  VMEMMAP_BASE;
#endif
#else /* !CONFIG_PPC64 */
	address_markers[i++].start_address = ioremap_bot;
	address_markers[i++].start_address = IOREMAP_TOP;
#ifdef CONFIG_NOT_COHERENT_CACHE
	address_markers[i++].start_address = IOREMAP_TOP;
	address_markers[i++].start_address = IOREMAP_TOP +
					     CONFIG_CONSISTENT_SIZE;
#endif
#ifdef CONFIG_HIGHMEM
	address_markers[i++].start_address = PKMAP_BASE;
	address_markers[i++].start_address = PKMAP_ADDR(LAST_PKMAP);
#endif
	address_markers[i++].start_address = FIXADDR_START;
	address_markers[i++].start_address = FIXADDR_TOP;
#ifdef CONFIG_KASAN
	address_markers[i++].start_address = KASAN_SHADOW_START;
	address_markers[i++].start_address = KASAN_SHADOW_END;
#endif
#endif /* CONFIG_PPC64 */
}

static int ptdump_show(struct seq_file *m, void *v)
{
	struct pg_state st = {
		.seq = m,
		.marker = address_markers,
	};

	if (radix_enabled())
		st.start_address = PAGE_OFFSET;
	else
		st.start_address = KERN_VIRT_START;

	/* Traverse kernel page tables */
	walk_pagetables(&st);
	note_page(&st, 0, 0, 0);
	return 0;
}


static int ptdump_open(struct inode *inode, struct file *file)
{
	return single_open(file, ptdump_show, NULL);
}

static const struct file_operations ptdump_fops = {
	.open		= ptdump_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
	.release	= single_release,
};

static void build_pgtable_complete_mask(void)
{
	unsigned int i, j;

	for (i = 0; i < ARRAY_SIZE(pg_level); i++)
		if (pg_level[i].flag)
			for (j = 0; j < pg_level[i].num; j++)
				pg_level[i].mask |= pg_level[i].flag[j].mask;
}

static int ptdump_init(void)
{
	struct dentry *debugfs_file;

	populate_markers();
	build_pgtable_complete_mask();
	debugfs_file = debugfs_create_file("kernel_page_tables", 0400, NULL,
			NULL, &ptdump_fops);
	return debugfs_file ? 0 : -ENOMEM;
}
device_initcall(ptdump_init);
Commit	Line	Data
8eb07b18 RG	1	/*
	2	* Copyright 2016, Rashmica Gupta, IBM Corp.
	3	*
	4	* This traverses the kernel pagetables and dumps the
	5	* information about the used sections of memory to
	6	* /sys/kernel/debug/kernel_pagetables.
	7	*
	8	* Derived from the arm64 implementation:
	9	* Copyright (c) 2014, The Linux Foundation, Laura Abbott.
	10	* (C) Copyright 2008 Intel Corporation, Arjan van de Ven.
	11	*
	12	* This program is free software; you can redistribute it and/or
	13	* modify it under the terms of the GNU General Public License
	14	* as published by the Free Software Foundation; version 2
	15	* of the License.
	16	*/
	17	#include <linux/debugfs.h>
	18	#include <linux/fs.h>
bfb9956a	19	#include <linux/hugetlb.h>
8eb07b18 RG	20	#include <linux/io.h>
8eb07b18 RG	21	#include <linux/mm.h>
462951cd	22	#include <linux/highmem.h>
8eb07b18 RG	23	#include <linux/sched.h>
	24	#include <linux/seq_file.h>
	25	#include <asm/fixmap.h>
	26	#include <asm/pgtable.h>
	27	#include <linux/const.h>
	28	#include <asm/page.h>
	29	#include <asm/pgalloc.h>
	30
e66c3209	31	#include "ptdump.h"
97026b5a	32
6c01bbd2 CL	33	#ifdef CONFIG_PPC32
	34	#define KERN_VIRT_START 0
	35	#endif
	36
8eb07b18 RG	37	/*
	38	* To visualise what is happening,
	39	*
	40	* - PTRS_PER_P = how many entries there are in the corresponding P
	41	* - P**_SHIFT = how many bits of the address we use to index into the
	42	* corresponding P**
	43	* - P**_SIZE is how much memory we can access through the table - not the
	44	* size of the table itself.
	45	* P**={PGD, PUD, PMD, PTE}
	46	*
	47	*
	48	* Each entry of the PGD points to a PUD. Each entry of a PUD points to a
	49	* PMD. Each entry of a PMD points to a PTE. And every PTE entry points to
	50	* a page.
	51	*
	52	* In the case where there are only 3 levels, the PUD is folded into the
	53	* PGD: every PUD has only one entry which points to the PMD.
	54	*
	55	* The page dumper groups page table entries of the same type into a single
	56	* description. It uses pg_state to track the range information while
	57	* iterating over the PTE entries. When the continuity is broken it then
	58	* dumps out a description of the range - ie PTEs that are virtually contiguous
	59	* with the same PTE flags are chunked together. This is to make it clear how
	60	* different areas of the kernel virtual memory are used.
	61	*
	62	*/
	63	struct pg_state {
	64	struct seq_file *seq;
	65	const struct addr_marker *marker;
	66	unsigned long start_address;
aaa22952 OH	67	unsigned long start_pa;
aaa22952 OH	68	unsigned long last_pa;
8eb07b18 RG	69	unsigned int level;
	70	u64 current_flags;
	71	};
	72
	73	struct addr_marker {
	74	unsigned long start_address;
	75	const char *name;
	76	};
	77
	78	static struct addr_marker address_markers[] = {
	79	{ 0, "Start of kernel VM" },
	80	{ 0, "vmalloc() Area" },
	81	{ 0, "vmalloc() End" },
6c01bbd2	82	#ifdef CONFIG_PPC64
8eb07b18 RG	83	{ 0, "isa I/O start" },
	84	{ 0, "isa I/O end" },
	85	{ 0, "phb I/O start" },
	86	{ 0, "phb I/O end" },
	87	{ 0, "I/O remap start" },
	88	{ 0, "I/O remap end" },
	89	{ 0, "vmemmap start" },
6c01bbd2 CL	90	#else
	91	{ 0, "Early I/O remap start" },
	92	{ 0, "Early I/O remap end" },
	93	#ifdef CONFIG_NOT_COHERENT_CACHE
	94	{ 0, "Consistent mem start" },
	95	{ 0, "Consistent mem end" },
	96	#endif
	97	#ifdef CONFIG_HIGHMEM
	98	{ 0, "Highmem PTEs start" },
	99	{ 0, "Highmem PTEs end" },
	100	#endif
	101	{ 0, "Fixmap start" },
	102	{ 0, "Fixmap end" },
b4abe38f CL	103	#endif
	104	#ifdef CONFIG_KASAN
	105	{ 0, "kasan shadow mem start" },
	106	{ 0, "kasan shadow mem end" },
6c01bbd2	107	#endif
8eb07b18 RG	108	{ -1, NULL },
	109	};
	110
5f18cbdb RC	111	#define pt_dump_seq_printf(m, fmt, args...) \
	112	({ \
	113	if (m) \
	114	seq_printf(m, fmt, ##args); \
	115	})
	116
	117	#define pt_dump_seq_putc(m, c) \
	118	({ \
	119	if (m) \
	120	seq_putc(m, c); \
	121	})
	122
8eb07b18 RG	123	static void dump_flag_info(struct pg_state *st, const struct flag_info
	124	*flag, u64 pte, int num)
	125	{
	126	unsigned int i;
	127
	128	for (i = 0; i < num; i++, flag++) {
	129	const char *s = NULL;
	130	u64 val;
	131
	132	/* flag not defined so don't check it */
	133	if (flag->mask == 0)
	134	continue;
	135	/* Some 'flags' are actually values */
	136	if (flag->is_val) {
	137	val = pte & flag->val;
	138	if (flag->shift)
	139	val = val >> flag->shift;
5f18cbdb	140	pt_dump_seq_printf(st->seq, " %s:%llx", flag->set, val);
8eb07b18 RG	141	} else {
	142	if ((pte & flag->mask) == flag->val)
	143	s = flag->set;
	144	else
	145	s = flag->clear;
	146	if (s)
5f18cbdb	147	pt_dump_seq_printf(st->seq, " %s", s);
8eb07b18 RG	148	}
	149	st->current_flags &= ~flag->mask;
	150	}
	151	if (st->current_flags != 0)
5f18cbdb	152	pt_dump_seq_printf(st->seq, " unknown flags:%llx", st->current_flags);
8eb07b18 RG	153	}
	154
	155	static void dump_addr(struct pg_state *st, unsigned long addr)
	156	{
	157	static const char units[] = "KMGTPE";
	158	const char *unit = units;
	159	unsigned long delta;
	160
78a18dbf	161	#ifdef CONFIG_PPC64
cabe8138	162	#define REG "0x%016lx"
78a18dbf	163	#else
cabe8138	164	#define REG "0x%08lx"
78a18dbf	165	#endif
aaa22952	166
5f18cbdb	167	pt_dump_seq_printf(st->seq, REG "-" REG " ", st->start_address, addr - 1);
cabe8138	168	if (st->start_pa == st->last_pa && st->start_address + PAGE_SIZE != addr) {
5f18cbdb	169	pt_dump_seq_printf(st->seq, "[" REG "]", st->start_pa);
cabe8138 CL	170	delta = PAGE_SIZE >> 10;
cabe8138 CL	171	} else {
5f18cbdb	172	pt_dump_seq_printf(st->seq, " " REG " ", st->start_pa);
cabe8138 CL	173	delta = (addr - st->start_address) >> 10;
cabe8138 CL	174	}
8eb07b18 RG	175	/* Work out what appropriate unit to use */
	176	while (!(delta & 1023) && unit[1]) {
	177	delta >>= 10;
	178	unit++;
	179	}
5f18cbdb	180	pt_dump_seq_printf(st->seq, "%9lu%c", delta, *unit);
8eb07b18 RG	181
	182	}
	183
	184	static void note_page(struct pg_state *st, unsigned long addr,
	185	unsigned int level, u64 val)
	186	{
	187	u64 flag = val & pg_level[level].mask;
aaa22952 OH	188	u64 pa = val & PTE_RPN_MASK;
aaa22952 OH	189
8eb07b18 RG	190	/* At first no level is set */
	191	if (!st->level) {
	192	st->level = level;
	193	st->current_flags = flag;
	194	st->start_address = addr;
aaa22952 OH	195	st->start_pa = pa;
aaa22952 OH	196	st->last_pa = pa;
5f18cbdb	197	pt_dump_seq_printf(st->seq, "---[ %s ]---\n", st->marker->name);
8eb07b18 RG	198	/*
	199	* Dump the section of virtual memory when:
	200	* - the PTE flags from one entry to the next differs.
	201	* - we change levels in the tree.
	202	* - the address is in a different section of memory and is thus
	203	* used for a different purpose, regardless of the flags.
aaa22952	204	* - the pa of this page is not adjacent to the last inspected page
8eb07b18 RG	205	*/
8eb07b18 RG	206	} else if (flag != st->current_flags \|\| level != st->level \|\|
aaa22952	207	addr >= st->marker[1].start_address \|\|
cabe8138 CL	208	(pa != st->last_pa + PAGE_SIZE &&
cabe8138 CL	209	(pa != st->start_pa \|\| st->start_pa != st->last_pa))) {
8eb07b18 RG	210
	211	/* Check the PTE flags */
	212	if (st->current_flags) {
	213	dump_addr(st, addr);
	214
	215	/* Dump all the flags */
	216	if (pg_level[st->level].flag)
	217	dump_flag_info(st, pg_level[st->level].flag,
	218	st->current_flags,
	219	pg_level[st->level].num);
	220
5f18cbdb	221	pt_dump_seq_putc(st->seq, '\n');
8eb07b18 RG	222	}
	223
	224	/*
	225	* Address indicates we have passed the end of the
	226	* current section of virtual memory
	227	*/
	228	while (addr >= st->marker[1].start_address) {
	229	st->marker++;
5f18cbdb	230	pt_dump_seq_printf(st->seq, "---[ %s ]---\n", st->marker->name);
8eb07b18 RG	231	}
8eb07b18 RG	232	st->start_address = addr;
aaa22952 OH	233	st->start_pa = pa;
aaa22952 OH	234	st->last_pa = pa;
8eb07b18 RG	235	st->current_flags = flag;
8eb07b18 RG	236	st->level = level;
aaa22952 OH	237	} else {
aaa22952 OH	238	st->last_pa = pa;
8eb07b18 RG	239	}
	240	}
	241
	242	static void walk_pte(struct pg_state st, pmd_t pmd, unsigned long start)
	243	{
	244	pte_t *pte = pte_offset_kernel(pmd, 0);
	245	unsigned long addr;
	246	unsigned int i;
	247
	248	for (i = 0; i < PTRS_PER_PTE; i++, pte++) {
	249	addr = start + i * PAGE_SIZE;
	250	note_page(st, addr, 4, pte_val(*pte));
	251
	252	}
	253	}
	254
	255	static void walk_pmd(struct pg_state st, pud_t pud, unsigned long start)
	256	{
	257	pmd_t *pmd = pmd_offset(pud, 0);
	258	unsigned long addr;
	259	unsigned int i;
	260
	261	for (i = 0; i < PTRS_PER_PMD; i++, pmd++) {
	262	addr = start + i * PMD_SIZE;
bfb9956a	263	if (!pmd_none(pmd) && !pmd_huge(pmd))
8eb07b18 RG	264	/* pmd exists */
	265	walk_pte(st, pmd, addr);
	266	else
	267	note_page(st, addr, 3, pmd_val(*pmd));
	268	}
	269	}
	270
	271	static void walk_pud(struct pg_state st, pgd_t pgd, unsigned long start)
	272	{
	273	pud_t *pud = pud_offset(pgd, 0);
	274	unsigned long addr;
	275	unsigned int i;
	276
	277	for (i = 0; i < PTRS_PER_PUD; i++, pud++) {
	278	addr = start + i * PUD_SIZE;
bfb9956a	279	if (!pud_none(pud) && !pud_huge(pud))
8eb07b18 RG	280	/* pud exists */
	281	walk_pmd(st, pud, addr);
	282	else
	283	note_page(st, addr, 2, pud_val(*pud));
	284	}
	285	}
	286
	287	static void walk_pagetables(struct pg_state *st)
	288	{
	289	pgd_t *pgd = pgd_offset_k(0UL);
	290	unsigned int i;
	291	unsigned long addr;
	292
0d923962 ME	293	addr = st->start_address;
0d923962 ME	294
8eb07b18 RG	295	/*
	296	* Traverse the linux pagetable structure and dump pages that are in
	297	* the hash pagetable.
	298	*/
0d923962	299	for (i = 0; i < PTRS_PER_PGD; i++, pgd++, addr += PGDIR_SIZE) {
bfb9956a	300	if (!pgd_none(pgd) && !pgd_huge(pgd))
8eb07b18 RG	301	/* pgd exists */
	302	walk_pud(st, pgd, addr);
	303	else
	304	note_page(st, addr, 1, pgd_val(*pgd));
	305	}
	306	}
	307
	308	static void populate_markers(void)
	309	{
6c01bbd2 CL	310	int i = 0;
	311
	312	address_markers[i++].start_address = PAGE_OFFSET;
	313	address_markers[i++].start_address = VMALLOC_START;
	314	address_markers[i++].start_address = VMALLOC_END;
	315	#ifdef CONFIG_PPC64
	316	address_markers[i++].start_address = ISA_IO_BASE;
	317	address_markers[i++].start_address = ISA_IO_END;
	318	address_markers[i++].start_address = PHB_IO_BASE;
	319	address_markers[i++].start_address = PHB_IO_END;
	320	address_markers[i++].start_address = IOREMAP_BASE;
	321	address_markers[i++].start_address = IOREMAP_END;
0034d395	322	/* What is the ifdef about? */
4e003747	323	#ifdef CONFIG_PPC_BOOK3S_64
0034d395	324	address_markers[i++].start_address = H_VMEMMAP_START;
8eb07b18	325	#else
6c01bbd2 CL	326	address_markers[i++].start_address = VMEMMAP_BASE;
	327	#endif
	328	#else /* !CONFIG_PPC64 */
	329	address_markers[i++].start_address = ioremap_bot;
	330	address_markers[i++].start_address = IOREMAP_TOP;
	331	#ifdef CONFIG_NOT_COHERENT_CACHE
	332	address_markers[i++].start_address = IOREMAP_TOP;
	333	address_markers[i++].start_address = IOREMAP_TOP +
	334	CONFIG_CONSISTENT_SIZE;
	335	#endif
	336	#ifdef CONFIG_HIGHMEM
	337	address_markers[i++].start_address = PKMAP_BASE;
	338	address_markers[i++].start_address = PKMAP_ADDR(LAST_PKMAP);
8eb07b18	339	#endif
6c01bbd2 CL	340	address_markers[i++].start_address = FIXADDR_START;
6c01bbd2 CL	341	address_markers[i++].start_address = FIXADDR_TOP;
b4abe38f CL	342	#ifdef CONFIG_KASAN
	343	address_markers[i++].start_address = KASAN_SHADOW_START;
	344	address_markers[i++].start_address = KASAN_SHADOW_END;
	345	#endif
6c01bbd2	346	#endif /* CONFIG_PPC64 */
8eb07b18 RG	347	}
	348
	349	static int ptdump_show(struct seq_file m, void v)
	350	{
	351	struct pg_state st = {
	352	.seq = m,
8eb07b18 RG	353	.marker = address_markers,
8eb07b18 RG	354	};
0d923962 ME	355
	356	if (radix_enabled())
	357	st.start_address = PAGE_OFFSET;
	358	else
	359	st.start_address = KERN_VIRT_START;
	360
8eb07b18 RG	361	/* Traverse kernel page tables */
	362	walk_pagetables(&st);
	363	note_page(&st, 0, 0, 0);
	364	return 0;
	365	}
	366
	367
	368	static int ptdump_open(struct inode inode, struct file file)
	369	{
	370	return single_open(file, ptdump_show, NULL);
	371	}
	372
	373	static const struct file_operations ptdump_fops = {
	374	.open = ptdump_open,
	375	.read = seq_read,
	376	.llseek = seq_lseek,
	377	.release = single_release,
	378	};
	379
	380	static void build_pgtable_complete_mask(void)
	381	{
	382	unsigned int i, j;
	383
	384	for (i = 0; i < ARRAY_SIZE(pg_level); i++)
	385	if (pg_level[i].flag)
	386	for (j = 0; j < pg_level[i].num; j++)
	387	pg_level[i].mask \|= pg_level[i].flag[j].mask;
	388	}
	389
	390	static int ptdump_init(void)
	391	{
	392	struct dentry *debugfs_file;
	393
	394	populate_markers();
	395	build_pgtable_complete_mask();
2505820f	396	debugfs_file = debugfs_create_file("kernel_page_tables", 0400, NULL,
8eb07b18 RG	397	NULL, &ptdump_fops);
	398	return debugfs_file ? 0 : -ENOMEM;
	399	}
	400	device_initcall(ptdump_init);