[deliverable/linux.git] / arch / x86 / mm / dump_pagetables.c

/*
 * Debug helper to dump the current kernel pagetables of the system
 * so that we can see what the various memory ranges are set to.
 *
 * (C) Copyright 2008 Intel Corporation
 *
 * Author: Arjan van de Ven <arjan@linux.intel.com>
 *
 * 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/mm.h>
#include <linux/module.h>
#include <linux/seq_file.h>

#include <asm/pgtable.h>

/*
 * The dumper groups pagetable entries of the same type into one, and for
 * that it needs to keep some state when walking, and flush this state
 * when a "break" in the continuity is found.
 */
struct pg_state {
	int level;
	pgprot_t current_prot;
	unsigned long start_address;
	unsigned long current_address;
	const struct addr_marker *marker;
	unsigned long lines;
	bool to_dmesg;
};

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

/* indices for address_markers; keep sync'd w/ address_markers below */
enum address_markers_idx {
	USER_SPACE_NR = 0,
#ifdef CONFIG_X86_64
	KERNEL_SPACE_NR,
	LOW_KERNEL_NR,
	VMALLOC_START_NR,
	VMEMMAP_START_NR,
# ifdef CONFIG_X86_ESPFIX64
	ESPFIX_START_NR,
# endif
	HIGH_KERNEL_NR,
	MODULES_VADDR_NR,
	MODULES_END_NR,
#else
	KERNEL_SPACE_NR,
	VMALLOC_START_NR,
	VMALLOC_END_NR,
# ifdef CONFIG_HIGHMEM
	PKMAP_BASE_NR,
# endif
	FIXADDR_START_NR,
#endif
};

/* Address space markers hints */
static struct addr_marker address_markers[] = {
	{ 0, "User Space" },
#ifdef CONFIG_X86_64
	{ 0x8000000000000000UL, "Kernel Space" },
	{ PAGE_OFFSET,		"Low Kernel Mapping" },
	{ VMALLOC_START,        "vmalloc() Area" },
	{ VMEMMAP_START,        "Vmemmap" },
# ifdef CONFIG_X86_ESPFIX64
	{ ESPFIX_BASE_ADDR,	"ESPfix Area", 16 },
# endif
	{ __START_KERNEL_map,   "High Kernel Mapping" },
	{ MODULES_VADDR,        "Modules" },
	{ MODULES_END,          "End Modules" },
#else
	{ PAGE_OFFSET,          "Kernel Mapping" },
	{ 0/* VMALLOC_START */, "vmalloc() Area" },
	{ 0/*VMALLOC_END*/,     "vmalloc() End" },
# ifdef CONFIG_HIGHMEM
	{ 0/*PKMAP_BASE*/,      "Persisent kmap() Area" },
# endif
	{ 0/*FIXADDR_START*/,   "Fixmap Area" },
#endif
	{ -1, NULL }		/* End of list */
};

/* Multipliers for offsets within the PTEs */
#define PTE_LEVEL_MULT (PAGE_SIZE)
#define PMD_LEVEL_MULT (PTRS_PER_PTE * PTE_LEVEL_MULT)
#define PUD_LEVEL_MULT (PTRS_PER_PMD * PMD_LEVEL_MULT)
#define PGD_LEVEL_MULT (PTRS_PER_PUD * PUD_LEVEL_MULT)

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

#define pt_dump_cont_printf(m, to_dmesg, fmt, args...)		\
({								\
	if (to_dmesg)					\
		printk(KERN_CONT fmt, ##args);			\
	else							\
		if (m)						\
			seq_printf(m, fmt, ##args);		\
})

/*
 * Print a readable form of a pgprot_t to the seq_file
 */
static void printk_prot(struct seq_file *m, pgprot_t prot, int level, bool dmsg)
{
	pgprotval_t pr = pgprot_val(prot);
	static const char * const level_name[] =
		{ "cr3", "pgd", "pud", "pmd", "pte" };

	if (!pgprot_val(prot)) {
		/* Not present */
		pt_dump_cont_printf(m, dmsg, "                          ");
	} else {
		if (pr & _PAGE_USER)
			pt_dump_cont_printf(m, dmsg, "USR ");
		else
			pt_dump_cont_printf(m, dmsg, "    ");
		if (pr & _PAGE_RW)
			pt_dump_cont_printf(m, dmsg, "RW ");
		else
			pt_dump_cont_printf(m, dmsg, "ro ");
		if (pr & _PAGE_PWT)
			pt_dump_cont_printf(m, dmsg, "PWT ");
		else
			pt_dump_cont_printf(m, dmsg, "    ");
		if (pr & _PAGE_PCD)
			pt_dump_cont_printf(m, dmsg, "PCD ");
		else
			pt_dump_cont_printf(m, dmsg, "    ");

		/* Bit 9 has a different meaning on level 3 vs 4 */
		if (level <= 3) {
			if (pr & _PAGE_PSE)
				pt_dump_cont_printf(m, dmsg, "PSE ");
			else
				pt_dump_cont_printf(m, dmsg, "    ");
		} else {
			if (pr & _PAGE_PAT)
				pt_dump_cont_printf(m, dmsg, "pat ");
			else
				pt_dump_cont_printf(m, dmsg, "    ");
		}
		if (pr & _PAGE_GLOBAL)
			pt_dump_cont_printf(m, dmsg, "GLB ");
		else
			pt_dump_cont_printf(m, dmsg, "    ");
		if (pr & _PAGE_NX)
			pt_dump_cont_printf(m, dmsg, "NX ");
		else
			pt_dump_cont_printf(m, dmsg, "x  ");
	}
	pt_dump_cont_printf(m, dmsg, "%s\n", level_name[level]);
}

/*
 * On 64 bits, sign-extend the 48 bit address to 64 bit
 */
static unsigned long normalize_addr(unsigned long u)
{
#ifdef CONFIG_X86_64
	return (signed long)(u << 16) >> 16;
#else
	return u;
#endif
}

/*
 * This function gets called on a break in a continuous series
 * of PTE entries; the next one is different so we need to
 * print what we collected so far.
 */
static void note_page(struct seq_file *m, struct pg_state *st,
		      pgprot_t new_prot, int level)
{
	pgprotval_t prot, cur;
	static const char units[] = "BKMGTPE";

	/*
	 * If we have a "break" in the series, we need to flush the state that
	 * we have now. "break" is either changing perms, levels or
	 * address space marker.
	 */
	prot = pgprot_val(new_prot) & PTE_FLAGS_MASK;
	cur = pgprot_val(st->current_prot) & PTE_FLAGS_MASK;

	if (!st->level) {
		/* First entry */
		st->current_prot = new_prot;
		st->level = level;
		st->marker = address_markers;
		st->lines = 0;
		pt_dump_seq_printf(m, st->to_dmesg, "---[ %s ]---\n",
				   st->marker->name);
	} else if (prot != cur || level != st->level ||
		   st->current_address >= st->marker[1].start_address) {
		const char *unit = units;
		unsigned long delta;
		int width = sizeof(unsigned long) * 2;

		/*
		 * Now print the actual finished series
		 */
		if (!st->marker->max_lines ||
		    st->lines < st->marker->max_lines) {
			pt_dump_seq_printf(m, st->to_dmesg,
					   "0x%0*lx-0x%0*lx   ",
					   width, st->start_address,
					   width, st->current_address);

			delta = st->current_address - st->start_address;
			while (!(delta & 1023) && unit[1]) {
				delta >>= 10;
				unit++;
			}
			pt_dump_cont_printf(m, st->to_dmesg, "%9lu%c ",
					    delta, *unit);
			printk_prot(m, st->current_prot, st->level,
				    st->to_dmesg);
		}
		st->lines++;

		/*
		 * We print markers for special areas of address space,
		 * such as the start of vmalloc space etc.
		 * This helps in the interpretation.
		 */
		if (st->current_address >= st->marker[1].start_address) {
			if (st->marker->max_lines &&
			    st->lines > st->marker->max_lines) {
				unsigned long nskip =
					st->lines - st->marker->max_lines;
				pt_dump_seq_printf(m, st->to_dmesg,
						   "... %lu entr%s skipped ... \n",
						   nskip,
						   nskip == 1 ? "y" : "ies");
			}
			st->marker++;
			st->lines = 0;
			pt_dump_seq_printf(m, st->to_dmesg, "---[ %s ]---\n",
					   st->marker->name);
		}

		st->start_address = st->current_address;
		st->current_prot = new_prot;
		st->level = level;
	}
}

static void walk_pte_level(struct seq_file *m, struct pg_state *st, pmd_t addr,
							unsigned long P)
{
	int i;
	pte_t *start;

	start = (pte_t *) pmd_page_vaddr(addr);
	for (i = 0; i < PTRS_PER_PTE; i++) {
		pgprot_t prot = pte_pgprot(*start);

		st->current_address = normalize_addr(P + i * PTE_LEVEL_MULT);
		note_page(m, st, prot, 4);
		start++;
	}
}

#if PTRS_PER_PMD > 1

static void walk_pmd_level(struct seq_file *m, struct pg_state *st, pud_t addr,
							unsigned long P)
{
	int i;
	pmd_t *start;

	start = (pmd_t *) pud_page_vaddr(addr);
	for (i = 0; i < PTRS_PER_PMD; i++) {
		st->current_address = normalize_addr(P + i * PMD_LEVEL_MULT);
		if (!pmd_none(*start)) {
			pgprotval_t prot = pmd_val(*start) & PTE_FLAGS_MASK;

			if (pmd_large(*start) || !pmd_present(*start))
				note_page(m, st, __pgprot(prot), 3);
			else
				walk_pte_level(m, st, *start,
					       P + i * PMD_LEVEL_MULT);
		} else
			note_page(m, st, __pgprot(0), 3);
		start++;
	}
}

#else
#define walk_pmd_level(m,s,a,p) walk_pte_level(m,s,__pmd(pud_val(a)),p)
#define pud_large(a) pmd_large(__pmd(pud_val(a)))
#define pud_none(a)  pmd_none(__pmd(pud_val(a)))
#endif

#if PTRS_PER_PUD > 1

static void walk_pud_level(struct seq_file *m, struct pg_state *st, pgd_t addr,
							unsigned long P)
{
	int i;
	pud_t *start;

	start = (pud_t *) pgd_page_vaddr(addr);

	for (i = 0; i < PTRS_PER_PUD; i++) {
		st->current_address = normalize_addr(P + i * PUD_LEVEL_MULT);
		if (!pud_none(*start)) {
			pgprotval_t prot = pud_val(*start) & PTE_FLAGS_MASK;

			if (pud_large(*start) || !pud_present(*start))
				note_page(m, st, __pgprot(prot), 2);
			else
				walk_pmd_level(m, st, *start,
					       P + i * PUD_LEVEL_MULT);
		} else
			note_page(m, st, __pgprot(0), 2);

		start++;
	}
}

#else
#define walk_pud_level(m,s,a,p) walk_pmd_level(m,s,__pud(pgd_val(a)),p)
#define pgd_large(a) pud_large(__pud(pgd_val(a)))
#define pgd_none(a)  pud_none(__pud(pgd_val(a)))
#endif

void ptdump_walk_pgd_level(struct seq_file *m, pgd_t *pgd)
{
#ifdef CONFIG_X86_64
	pgd_t *start = (pgd_t *) &init_level4_pgt;
#else
	pgd_t *start = swapper_pg_dir;
#endif
	int i;
	struct pg_state st = {};

	if (pgd) {
		start = pgd;
		st.to_dmesg = true;
	}

	for (i = 0; i < PTRS_PER_PGD; i++) {
		st.current_address = normalize_addr(i * PGD_LEVEL_MULT);
		if (!pgd_none(*start)) {
			pgprotval_t prot = pgd_val(*start) & PTE_FLAGS_MASK;

			if (pgd_large(*start) || !pgd_present(*start))
				note_page(m, &st, __pgprot(prot), 1);
			else
				walk_pud_level(m, &st, *start,
					       i * PGD_LEVEL_MULT);
		} else
			note_page(m, &st, __pgprot(0), 1);

		start++;
	}

	/* Flush out the last page */
	st.current_address = normalize_addr(PTRS_PER_PGD*PGD_LEVEL_MULT);
	note_page(m, &st, __pgprot(0), 0);
}

static int ptdump_show(struct seq_file *m, void *v)
{
	ptdump_walk_pgd_level(m, NULL);
	return 0;
}

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

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

static int pt_dump_init(void)
{
	struct dentry *pe;

#ifdef CONFIG_X86_32
	/* Not a compile-time constant on x86-32 */
	address_markers[VMALLOC_START_NR].start_address = VMALLOC_START;
	address_markers[VMALLOC_END_NR].start_address = VMALLOC_END;
# ifdef CONFIG_HIGHMEM
	address_markers[PKMAP_BASE_NR].start_address = PKMAP_BASE;
# endif
	address_markers[FIXADDR_START_NR].start_address = FIXADDR_START;
#endif

	pe = debugfs_create_file("kernel_page_tables", 0600, NULL, NULL,
				 &ptdump_fops);
	if (!pe)
		return -ENOMEM;

	return 0;
}

__initcall(pt_dump_init);
MODULE_LICENSE("GPL");
MODULE_AUTHOR("Arjan van de Ven <arjan@linux.intel.com>");
MODULE_DESCRIPTION("Kernel debugging helper that dumps pagetables");
Commit	Line	Data
926e5392 AV	1	/*
	2	* Debug helper to dump the current kernel pagetables of the system
	3	* so that we can see what the various memory ranges are set to.
	4	*
	5	* (C) Copyright 2008 Intel Corporation
	6	*
	7	* Author: Arjan van de Ven <arjan@linux.intel.com>
	8	*
	9	* This program is free software; you can redistribute it and/or
	10	* modify it under the terms of the GNU General Public License
	11	* as published by the Free Software Foundation; version 2
	12	* of the License.
	13	*/
	14
fe770bf0 PA	15	#include <linux/debugfs.h>
fe770bf0 PA	16	#include <linux/mm.h>
926e5392 AV	17	#include <linux/module.h>
926e5392 AV	18	#include <linux/seq_file.h>
926e5392 AV	19
	20	#include <asm/pgtable.h>
	21
	22	/*
	23	* The dumper groups pagetable entries of the same type into one, and for
	24	* that it needs to keep some state when walking, and flush this state
	25	* when a "break" in the continuity is found.
	26	*/
	27	struct pg_state {
	28	int level;
	29	pgprot_t current_prot;
	30	unsigned long start_address;
	31	unsigned long current_address;
fe770bf0	32	const struct addr_marker *marker;
3891a04a	33	unsigned long lines;
ef6bea6d	34	bool to_dmesg;
926e5392 AV	35	};
926e5392 AV	36
fe770bf0 PA	37	struct addr_marker {
	38	unsigned long start_address;
	39	const char *name;
3891a04a	40	unsigned long max_lines;
fe770bf0 PA	41	};
fe770bf0 PA	42
92851e2f AS	43	/* indices for address_markers; keep sync'd w/ address_markers below */
	44	enum address_markers_idx {
	45	USER_SPACE_NR = 0,
	46	#ifdef CONFIG_X86_64
	47	KERNEL_SPACE_NR,
	48	LOW_KERNEL_NR,
	49	VMALLOC_START_NR,
	50	VMEMMAP_START_NR,
8a5a5d15	51	# ifdef CONFIG_X86_ESPFIX64
3891a04a	52	ESPFIX_START_NR,
8a5a5d15	53	# endif
92851e2f AS	54	HIGH_KERNEL_NR,
	55	MODULES_VADDR_NR,
	56	MODULES_END_NR,
	57	#else
	58	KERNEL_SPACE_NR,
	59	VMALLOC_START_NR,
	60	VMALLOC_END_NR,
	61	# ifdef CONFIG_HIGHMEM
	62	PKMAP_BASE_NR,
	63	# endif
	64	FIXADDR_START_NR,
	65	#endif
	66	};
	67
fe770bf0 PA	68	/* Address space markers hints */
	69	static struct addr_marker address_markers[] = {
	70	{ 0, "User Space" },
	71	#ifdef CONFIG_X86_64
	72	{ 0x8000000000000000UL, "Kernel Space" },
684eb016	73	{ PAGE_OFFSET, "Low Kernel Mapping" },
fe770bf0	74	{ VMALLOC_START, "vmalloc() Area" },
fe770bf0	75	{ VMEMMAP_START, "Vmemmap" },
8a5a5d15	76	# ifdef CONFIG_X86_ESPFIX64
3891a04a	77	{ ESPFIX_BASE_ADDR, "ESPfix Area", 16 },
8a5a5d15	78	# endif
fe770bf0	79	{ __START_KERNEL_map, "High Kernel Mapping" },
9a79cf9c YL	80	{ MODULES_VADDR, "Modules" },
9a79cf9c YL	81	{ MODULES_END, "End Modules" },
fe770bf0 PA	82	#else
	83	{ PAGE_OFFSET, "Kernel Mapping" },
	84	{ 0/* VMALLOC_START */, "vmalloc() Area" },
	85	{ 0/VMALLOC_END/, "vmalloc() End" },
	86	# ifdef CONFIG_HIGHMEM
	87	{ 0/PKMAP_BASE/, "Persisent kmap() Area" },
	88	# endif
	89	{ 0/FIXADDR_START/, "Fixmap Area" },
	90	#endif
	91	{ -1, NULL } /* End of list */
	92	};
926e5392	93
fe770bf0 PA	94	/* Multipliers for offsets within the PTEs */
	95	#define PTE_LEVEL_MULT (PAGE_SIZE)
	96	#define PMD_LEVEL_MULT (PTRS_PER_PTE * PTE_LEVEL_MULT)
	97	#define PUD_LEVEL_MULT (PTRS_PER_PMD * PMD_LEVEL_MULT)
	98	#define PGD_LEVEL_MULT (PTRS_PER_PUD * PUD_LEVEL_MULT)
926e5392	99
ef6bea6d BP	100	#define pt_dump_seq_printf(m, to_dmesg, fmt, args...) \
	101	({ \
	102	if (to_dmesg) \
	103	printk(KERN_INFO fmt, ##args); \
	104	else \
	105	if (m) \
	106	seq_printf(m, fmt, ##args); \
	107	})
	108
	109	#define pt_dump_cont_printf(m, to_dmesg, fmt, args...) \
	110	({ \
	111	if (to_dmesg) \
	112	printk(KERN_CONT fmt, ##args); \
	113	else \
	114	if (m) \
	115	seq_printf(m, fmt, ##args); \
	116	})
	117
926e5392 AV	118	/*
	119	* Print a readable form of a pgprot_t to the seq_file
	120	*/
ef6bea6d	121	static void printk_prot(struct seq_file *m, pgprot_t prot, int level, bool dmsg)
926e5392	122	{
fe770bf0 PA	123	pgprotval_t pr = pgprot_val(prot);
	124	static const char * const level_name[] =
	125	{ "cr3", "pgd", "pud", "pmd", "pte" };
	126
	127	if (!pgprot_val(prot)) {
	128	/* Not present */
ef6bea6d	129	pt_dump_cont_printf(m, dmsg, " ");
fe770bf0 PA	130	} else {
fe770bf0 PA	131	if (pr & _PAGE_USER)
ef6bea6d	132	pt_dump_cont_printf(m, dmsg, "USR ");
926e5392	133	else
ef6bea6d	134	pt_dump_cont_printf(m, dmsg, " ");
fe770bf0	135	if (pr & _PAGE_RW)
ef6bea6d	136	pt_dump_cont_printf(m, dmsg, "RW ");
fe770bf0	137	else
ef6bea6d	138	pt_dump_cont_printf(m, dmsg, "ro ");
fe770bf0	139	if (pr & _PAGE_PWT)
ef6bea6d	140	pt_dump_cont_printf(m, dmsg, "PWT ");
fe770bf0	141	else
ef6bea6d	142	pt_dump_cont_printf(m, dmsg, " ");
fe770bf0	143	if (pr & _PAGE_PCD)
ef6bea6d	144	pt_dump_cont_printf(m, dmsg, "PCD ");
926e5392	145	else
ef6bea6d	146	pt_dump_cont_printf(m, dmsg, " ");
fe770bf0 PA	147
	148	/* Bit 9 has a different meaning on level 3 vs 4 */
	149	if (level <= 3) {
	150	if (pr & _PAGE_PSE)
ef6bea6d	151	pt_dump_cont_printf(m, dmsg, "PSE ");
fe770bf0	152	else
ef6bea6d	153	pt_dump_cont_printf(m, dmsg, " ");
fe770bf0 PA	154	} else {
fe770bf0 PA	155	if (pr & _PAGE_PAT)
ef6bea6d	156	pt_dump_cont_printf(m, dmsg, "pat ");
fe770bf0	157	else
ef6bea6d	158	pt_dump_cont_printf(m, dmsg, " ");
fe770bf0 PA	159	}
fe770bf0 PA	160	if (pr & _PAGE_GLOBAL)
ef6bea6d	161	pt_dump_cont_printf(m, dmsg, "GLB ");
fe770bf0	162	else
ef6bea6d	163	pt_dump_cont_printf(m, dmsg, " ");
fe770bf0	164	if (pr & _PAGE_NX)
ef6bea6d	165	pt_dump_cont_printf(m, dmsg, "NX ");
fe770bf0	166	else
ef6bea6d	167	pt_dump_cont_printf(m, dmsg, "x ");
926e5392	168	}
ef6bea6d	169	pt_dump_cont_printf(m, dmsg, "%s\n", level_name[level]);
926e5392 AV	170	}
	171
	172	/*
fe770bf0	173	* On 64 bits, sign-extend the 48 bit address to 64 bit
926e5392	174	*/
fe770bf0	175	static unsigned long normalize_addr(unsigned long u)
926e5392	176	{
fe770bf0 PA	177	#ifdef CONFIG_X86_64
	178	return (signed long)(u << 16) >> 16;
	179	#else
926e5392	180	return u;
fe770bf0	181	#endif
926e5392 AV	182	}
	183
	184	/*
	185	* This function gets called on a break in a continuous series
	186	* of PTE entries; the next one is different so we need to
	187	* print what we collected so far.
	188	*/
	189	static void note_page(struct seq_file m, struct pg_state st,
fe770bf0	190	pgprot_t new_prot, int level)
926e5392	191	{
fe770bf0	192	pgprotval_t prot, cur;
3891a04a	193	static const char units[] = "BKMGTPE";
926e5392 AV	194
	195	/*
	196	* If we have a "break" in the series, we need to flush the state that
fe770bf0 PA	197	* we have now. "break" is either changing perms, levels or
fe770bf0 PA	198	* address space marker.
926e5392	199	*/
27990eac JF	200	prot = pgprot_val(new_prot) & PTE_FLAGS_MASK;
27990eac JF	201	cur = pgprot_val(st->current_prot) & PTE_FLAGS_MASK;
926e5392	202
fe770bf0 PA	203	if (!st->level) {
	204	/* First entry */
	205	st->current_prot = new_prot;
	206	st->level = level;
	207	st->marker = address_markers;
3891a04a	208	st->lines = 0;
ef6bea6d BP	209	pt_dump_seq_printf(m, st->to_dmesg, "---[ %s ]---\n",
ef6bea6d BP	210	st->marker->name);
fe770bf0 PA	211	} else if (prot != cur \|\| level != st->level \|\|
	212	st->current_address >= st->marker[1].start_address) {
	213	const char *unit = units;
926e5392	214	unsigned long delta;
6424fb38	215	int width = sizeof(unsigned long) * 2;
926e5392	216
926e5392 AV	217	/*
	218	* Now print the actual finished series
	219	*/
3891a04a PA	220	if (!st->marker->max_lines \|\|
	221	st->lines < st->marker->max_lines) {
	222	pt_dump_seq_printf(m, st->to_dmesg,
	223	"0x%0lx-0x%0lx ",
	224	width, st->start_address,
	225	width, st->current_address);
926e5392	226
3891a04a PA	227	delta = st->current_address - st->start_address;
	228	while (!(delta & 1023) && unit[1]) {
	229	delta >>= 10;
	230	unit++;
	231	}
	232	pt_dump_cont_printf(m, st->to_dmesg, "%9lu%c ",
	233	delta, *unit);
	234	printk_prot(m, st->current_prot, st->level,
	235	st->to_dmesg);
926e5392	236	}
3891a04a	237	st->lines++;
fe770bf0 PA	238
	239	/*
	240	* We print markers for special areas of address space,
	241	* such as the start of vmalloc space etc.
	242	* This helps in the interpretation.
	243	*/
	244	if (st->current_address >= st->marker[1].start_address) {
3891a04a PA	245	if (st->marker->max_lines &&
	246	st->lines > st->marker->max_lines) {
	247	unsigned long nskip =
	248	st->lines - st->marker->max_lines;
	249	pt_dump_seq_printf(m, st->to_dmesg,
	250	"... %lu entr%s skipped ... \n",
	251	nskip,
	252	nskip == 1 ? "y" : "ies");
	253	}
fe770bf0	254	st->marker++;
3891a04a	255	st->lines = 0;
ef6bea6d BP	256	pt_dump_seq_printf(m, st->to_dmesg, "---[ %s ]---\n",
ef6bea6d BP	257	st->marker->name);
926e5392	258	}
fe770bf0	259
926e5392 AV	260	st->start_address = st->current_address;
	261	st->current_prot = new_prot;
	262	st->level = level;
fe770bf0	263	}
926e5392 AV	264	}
926e5392 AV	265
fe770bf0	266	static void walk_pte_level(struct seq_file m, struct pg_state st, pmd_t addr,
926e5392 AV	267	unsigned long P)
	268	{
	269	int i;
	270	pte_t *start;
	271
	272	start = (pte_t *) pmd_page_vaddr(addr);
	273	for (i = 0; i < PTRS_PER_PTE; i++) {
	274	pgprot_t prot = pte_pgprot(*start);
	275
fe770bf0	276	st->current_address = normalize_addr(P + i * PTE_LEVEL_MULT);
926e5392 AV	277	note_page(m, st, prot, 4);
	278	start++;
	279	}
	280	}
	281
fe770bf0	282	#if PTRS_PER_PMD > 1
926e5392	283
fe770bf0	284	static void walk_pmd_level(struct seq_file m, struct pg_state st, pud_t addr,
926e5392 AV	285	unsigned long P)
	286	{
	287	int i;
	288	pmd_t *start;
	289
	290	start = (pmd_t *) pud_page_vaddr(addr);
	291	for (i = 0; i < PTRS_PER_PMD; i++) {
fe770bf0	292	st->current_address = normalize_addr(P + i * PMD_LEVEL_MULT);
926e5392	293	if (!pmd_none(*start)) {
77be1fab	294	pgprotval_t prot = pmd_val(*start) & PTE_FLAGS_MASK;
926e5392	295
fe770bf0	296	if (pmd_large(start) \|\| !pmd_present(start))
926e5392 AV	297	note_page(m, st, __pgprot(prot), 3);
926e5392 AV	298	else
fe770bf0 PA	299	walk_pte_level(m, st, *start,
fe770bf0 PA	300	P + i * PMD_LEVEL_MULT);
926e5392 AV	301	} else
	302	note_page(m, st, __pgprot(0), 3);
	303	start++;
	304	}
	305	}
	306
fe770bf0 PA	307	#else
	308	#define walk_pmd_level(m,s,a,p) walk_pte_level(m,s,__pmd(pud_val(a)),p)
	309	#define pud_large(a) pmd_large(__pmd(pud_val(a)))
	310	#define pud_none(a) pmd_none(__pmd(pud_val(a)))
	311	#endif
926e5392	312
fe770bf0 PA	313	#if PTRS_PER_PUD > 1
	314
	315	static void walk_pud_level(struct seq_file m, struct pg_state st, pgd_t addr,
926e5392 AV	316	unsigned long P)
	317	{
	318	int i;
	319	pud_t *start;
	320
	321	start = (pud_t *) pgd_page_vaddr(addr);
	322
	323	for (i = 0; i < PTRS_PER_PUD; i++) {
fe770bf0	324	st->current_address = normalize_addr(P + i * PUD_LEVEL_MULT);
926e5392	325	if (!pud_none(*start)) {
77be1fab	326	pgprotval_t prot = pud_val(*start) & PTE_FLAGS_MASK;
926e5392	327
fe770bf0	328	if (pud_large(start) \|\| !pud_present(start))
926e5392 AV	329	note_page(m, st, __pgprot(prot), 2);
926e5392 AV	330	else
fe770bf0 PA	331	walk_pmd_level(m, st, *start,
fe770bf0 PA	332	P + i * PUD_LEVEL_MULT);
926e5392 AV	333	} else
	334	note_page(m, st, __pgprot(0), 2);
	335
	336	start++;
	337	}
	338	}
	339
fe770bf0 PA	340	#else
	341	#define walk_pud_level(m,s,a,p) walk_pmd_level(m,s,__pud(pgd_val(a)),p)
	342	#define pgd_large(a) pud_large(__pud(pgd_val(a)))
	343	#define pgd_none(a) pud_none(__pud(pgd_val(a)))
	344	#endif
	345
ef6bea6d	346	void ptdump_walk_pgd_level(struct seq_file m, pgd_t pgd)
926e5392	347	{
fe770bf0	348	#ifdef CONFIG_X86_64
926e5392	349	pgd_t start = (pgd_t ) &init_level4_pgt;
fe770bf0 PA	350	#else
	351	pgd_t *start = swapper_pg_dir;
	352	#endif
926e5392	353	int i;
ef6bea6d	354	struct pg_state st = {};
926e5392	355
ef6bea6d BP	356	if (pgd) {
	357	start = pgd;
	358	st.to_dmesg = true;
	359	}
926e5392 AV	360
926e5392 AV	361	for (i = 0; i < PTRS_PER_PGD; i++) {
fe770bf0 PA	362	st.current_address = normalize_addr(i * PGD_LEVEL_MULT);
fe770bf0 PA	363	if (!pgd_none(*start)) {
77be1fab	364	pgprotval_t prot = pgd_val(*start) & PTE_FLAGS_MASK;
fe770bf0 PA	365
	366	if (pgd_large(start) \|\| !pgd_present(start))
	367	note_page(m, &st, __pgprot(prot), 1);
	368	else
	369	walk_pud_level(m, &st, *start,
	370	i * PGD_LEVEL_MULT);
	371	} else
926e5392	372	note_page(m, &st, __pgprot(0), 1);
fe770bf0	373
926e5392 AV	374	start++;
926e5392 AV	375	}
fe770bf0 PA	376
	377	/* Flush out the last page */
	378	st.current_address = normalize_addr(PTRS_PER_PGD*PGD_LEVEL_MULT);
	379	note_page(m, &st, __pgprot(0), 0);
926e5392 AV	380	}
	381
	382	static int ptdump_show(struct seq_file m, void v)
	383	{
ef6bea6d	384	ptdump_walk_pgd_level(m, NULL);
926e5392 AV	385	return 0;
	386	}
	387
	388	static int ptdump_open(struct inode inode, struct file filp)
	389	{
	390	return single_open(filp, ptdump_show, NULL);
	391	}
	392
	393	static const struct file_operations ptdump_fops = {
	394	.open = ptdump_open,
	395	.read = seq_read,
	396	.llseek = seq_lseek,
	397	.release = single_release,
	398	};
	399
a4928cff	400	static int pt_dump_init(void)
926e5392 AV	401	{
	402	struct dentry *pe;
	403
fe770bf0 PA	404	#ifdef CONFIG_X86_32
fe770bf0 PA	405	/* Not a compile-time constant on x86-32 */
92851e2f AS	406	address_markers[VMALLOC_START_NR].start_address = VMALLOC_START;
92851e2f AS	407	address_markers[VMALLOC_END_NR].start_address = VMALLOC_END;
fe770bf0	408	# ifdef CONFIG_HIGHMEM
92851e2f	409	address_markers[PKMAP_BASE_NR].start_address = PKMAP_BASE;
fe770bf0	410	# endif
92851e2f	411	address_markers[FIXADDR_START_NR].start_address = FIXADDR_START;
fe770bf0 PA	412	#endif
fe770bf0 PA	413
926e5392 AV	414	pe = debugfs_create_file("kernel_page_tables", 0600, NULL, NULL,
	415	&ptdump_fops);
	416	if (!pe)
	417	return -ENOMEM;
	418
	419	return 0;
	420	}
	421
	422	__initcall(pt_dump_init);
	423	MODULE_LICENSE("GPL");
	424	MODULE_AUTHOR("Arjan van de Ven <arjan@linux.intel.com>");
	425	MODULE_DESCRIPTION("Kernel debugging helper that dumps pagetables");