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

// SPDX-License-Identifier: GPL-2.0-only
/*
 * 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.
 */
#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	PAGE_OFFSET
#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;
	bool check_wx;
	unsigned long wx_pages;
};

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_prot_wx(struct pg_state *st, unsigned long addr)
{
	if (!st->check_wx)
		return;

	if (!((st->current_flags & pgprot_val(PAGE_KERNEL_X)) == pgprot_val(PAGE_KERNEL_X)))
		return;

	WARN_ONCE(1, "powerpc/mm: Found insecure W+X mapping at address %p/%pS\n",
		  (void *)st->start_address, (void *)st->start_address);

	st->wx_pages += (addr - st->start_address) / PAGE_SIZE;
}

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) {
			note_prot_wx(st, addr);
			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;
}

#ifdef CONFIG_PPC_DEBUG_WX
void ptdump_check_wx(void)
{
	struct pg_state st = {
		.seq = NULL,
		.marker = address_markers,
		.check_wx = true,
	};

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

	walk_pagetables(&st);

	if (st.wx_pages)
		pr_warn("Checked W+X mappings: FAILED, %lu W+X pages found\n",
			st.wx_pages);
	else
		pr_info("Checked W+X mappings: passed, no W+X pages found\n");
}
#endif

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
b886d83c	1	// SPDX-License-Identifier: GPL-2.0-only
8eb07b18 RG	2	/*
	3	* Copyright 2016, Rashmica Gupta, IBM Corp.
	4	*
	5	* This traverses the kernel pagetables and dumps the
	6	* information about the used sections of memory to
	7	* /sys/kernel/debug/kernel_pagetables.
	8	*
	9	* Derived from the arm64 implementation:
	10	* Copyright (c) 2014, The Linux Foundation, Laura Abbott.
	11	* (C) Copyright 2008 Intel Corporation, Arjan van de Ven.
8eb07b18 RG	12	*/
	13	#include <linux/debugfs.h>
	14	#include <linux/fs.h>
bfb9956a	15	#include <linux/hugetlb.h>
8eb07b18 RG	16	#include <linux/io.h>
8eb07b18 RG	17	#include <linux/mm.h>
462951cd	18	#include <linux/highmem.h>
8eb07b18 RG	19	#include <linux/sched.h>
	20	#include <linux/seq_file.h>
	21	#include <asm/fixmap.h>
	22	#include <asm/pgtable.h>
	23	#include <linux/const.h>
	24	#include <asm/page.h>
	25	#include <asm/pgalloc.h>
	26
e66c3209	27	#include "ptdump.h"
97026b5a	28
6c01bbd2	29	#ifdef CONFIG_PPC32
453d87f6	30	#define KERN_VIRT_START PAGE_OFFSET
6c01bbd2 CL	31	#endif
6c01bbd2 CL	32
8eb07b18 RG	33	/*
	34	* To visualise what is happening,
	35	*
	36	* - PTRS_PER_P = how many entries there are in the corresponding P
	37	* - P**_SHIFT = how many bits of the address we use to index into the
	38	* corresponding P**
	39	* - P**_SIZE is how much memory we can access through the table - not the
	40	* size of the table itself.
	41	* P**={PGD, PUD, PMD, PTE}
	42	*
	43	*
	44	* Each entry of the PGD points to a PUD. Each entry of a PUD points to a
	45	* PMD. Each entry of a PMD points to a PTE. And every PTE entry points to
	46	* a page.
	47	*
	48	* In the case where there are only 3 levels, the PUD is folded into the
	49	* PGD: every PUD has only one entry which points to the PMD.
	50	*
	51	* The page dumper groups page table entries of the same type into a single
	52	* description. It uses pg_state to track the range information while
	53	* iterating over the PTE entries. When the continuity is broken it then
	54	* dumps out a description of the range - ie PTEs that are virtually contiguous
	55	* with the same PTE flags are chunked together. This is to make it clear how
	56	* different areas of the kernel virtual memory are used.
	57	*
	58	*/
	59	struct pg_state {
	60	struct seq_file *seq;
	61	const struct addr_marker *marker;
	62	unsigned long start_address;
aaa22952 OH	63	unsigned long start_pa;
aaa22952 OH	64	unsigned long last_pa;
8eb07b18 RG	65	unsigned int level;
8eb07b18 RG	66	u64 current_flags;
453d87f6 RC	67	bool check_wx;
453d87f6 RC	68	unsigned long wx_pages;
8eb07b18 RG	69	};
	70
	71	struct addr_marker {
	72	unsigned long start_address;
	73	const char *name;
	74	};
	75
	76	static struct addr_marker address_markers[] = {
	77	{ 0, "Start of kernel VM" },
	78	{ 0, "vmalloc() Area" },
	79	{ 0, "vmalloc() End" },
6c01bbd2	80	#ifdef CONFIG_PPC64
8eb07b18 RG	81	{ 0, "isa I/O start" },
	82	{ 0, "isa I/O end" },
	83	{ 0, "phb I/O start" },
	84	{ 0, "phb I/O end" },
	85	{ 0, "I/O remap start" },
	86	{ 0, "I/O remap end" },
	87	{ 0, "vmemmap start" },
6c01bbd2 CL	88	#else
	89	{ 0, "Early I/O remap start" },
	90	{ 0, "Early I/O remap end" },
	91	#ifdef CONFIG_NOT_COHERENT_CACHE
	92	{ 0, "Consistent mem start" },
	93	{ 0, "Consistent mem end" },
	94	#endif
	95	#ifdef CONFIG_HIGHMEM
	96	{ 0, "Highmem PTEs start" },
	97	{ 0, "Highmem PTEs end" },
	98	#endif
	99	{ 0, "Fixmap start" },
	100	{ 0, "Fixmap end" },
b4abe38f CL	101	#endif
	102	#ifdef CONFIG_KASAN
	103	{ 0, "kasan shadow mem start" },
	104	{ 0, "kasan shadow mem end" },
6c01bbd2	105	#endif
8eb07b18 RG	106	{ -1, NULL },
	107	};
	108
5f18cbdb RC	109	#define pt_dump_seq_printf(m, fmt, args...) \
	110	({ \
	111	if (m) \
	112	seq_printf(m, fmt, ##args); \
	113	})
	114
	115	#define pt_dump_seq_putc(m, c) \
	116	({ \
	117	if (m) \
	118	seq_putc(m, c); \
	119	})
	120
8eb07b18 RG	121	static void dump_flag_info(struct pg_state *st, const struct flag_info
	122	*flag, u64 pte, int num)
	123	{
	124	unsigned int i;
	125
	126	for (i = 0; i < num; i++, flag++) {
	127	const char *s = NULL;
	128	u64 val;
	129
	130	/* flag not defined so don't check it */
	131	if (flag->mask == 0)
	132	continue;
	133	/* Some 'flags' are actually values */
	134	if (flag->is_val) {
	135	val = pte & flag->val;
	136	if (flag->shift)
	137	val = val >> flag->shift;
5f18cbdb	138	pt_dump_seq_printf(st->seq, " %s:%llx", flag->set, val);
8eb07b18 RG	139	} else {
	140	if ((pte & flag->mask) == flag->val)
	141	s = flag->set;
	142	else
	143	s = flag->clear;
	144	if (s)
5f18cbdb	145	pt_dump_seq_printf(st->seq, " %s", s);
8eb07b18 RG	146	}
	147	st->current_flags &= ~flag->mask;
	148	}
	149	if (st->current_flags != 0)
5f18cbdb	150	pt_dump_seq_printf(st->seq, " unknown flags:%llx", st->current_flags);
8eb07b18 RG	151	}
	152
	153	static void dump_addr(struct pg_state *st, unsigned long addr)
	154	{
	155	static const char units[] = "KMGTPE";
	156	const char *unit = units;
	157	unsigned long delta;
	158
78a18dbf	159	#ifdef CONFIG_PPC64
cabe8138	160	#define REG "0x%016lx"
78a18dbf	161	#else
cabe8138	162	#define REG "0x%08lx"
78a18dbf	163	#endif
aaa22952	164
5f18cbdb	165	pt_dump_seq_printf(st->seq, REG "-" REG " ", st->start_address, addr - 1);
cabe8138	166	if (st->start_pa == st->last_pa && st->start_address + PAGE_SIZE != addr) {
5f18cbdb	167	pt_dump_seq_printf(st->seq, "[" REG "]", st->start_pa);
cabe8138 CL	168	delta = PAGE_SIZE >> 10;
cabe8138 CL	169	} else {
5f18cbdb	170	pt_dump_seq_printf(st->seq, " " REG " ", st->start_pa);
cabe8138 CL	171	delta = (addr - st->start_address) >> 10;
cabe8138 CL	172	}
8eb07b18 RG	173	/* Work out what appropriate unit to use */
	174	while (!(delta & 1023) && unit[1]) {
	175	delta >>= 10;
	176	unit++;
	177	}
5f18cbdb	178	pt_dump_seq_printf(st->seq, "%9lu%c", delta, *unit);
8eb07b18 RG	179
	180	}
	181
453d87f6 RC	182	static void note_prot_wx(struct pg_state *st, unsigned long addr)
	183	{
	184	if (!st->check_wx)
	185	return;
	186
	187	if (!((st->current_flags & pgprot_val(PAGE_KERNEL_X)) == pgprot_val(PAGE_KERNEL_X)))
	188	return;
	189
	190	WARN_ONCE(1, "powerpc/mm: Found insecure W+X mapping at address %p/%pS\n",
	191	(void )st->start_address, (void )st->start_address);
	192
	193	st->wx_pages += (addr - st->start_address) / PAGE_SIZE;
	194	}
	195
8eb07b18 RG	196	static void note_page(struct pg_state *st, unsigned long addr,
	197	unsigned int level, u64 val)
	198	{
	199	u64 flag = val & pg_level[level].mask;
aaa22952 OH	200	u64 pa = val & PTE_RPN_MASK;
aaa22952 OH	201
8eb07b18 RG	202	/* At first no level is set */
	203	if (!st->level) {
	204	st->level = level;
	205	st->current_flags = flag;
	206	st->start_address = addr;
aaa22952 OH	207	st->start_pa = pa;
aaa22952 OH	208	st->last_pa = pa;
5f18cbdb	209	pt_dump_seq_printf(st->seq, "---[ %s ]---\n", st->marker->name);
8eb07b18 RG	210	/*
	211	* Dump the section of virtual memory when:
	212	* - the PTE flags from one entry to the next differs.
	213	* - we change levels in the tree.
	214	* - the address is in a different section of memory and is thus
	215	* used for a different purpose, regardless of the flags.
aaa22952	216	* - the pa of this page is not adjacent to the last inspected page
8eb07b18 RG	217	*/
8eb07b18 RG	218	} else if (flag != st->current_flags \|\| level != st->level \|\|
aaa22952	219	addr >= st->marker[1].start_address \|\|
cabe8138 CL	220	(pa != st->last_pa + PAGE_SIZE &&
cabe8138 CL	221	(pa != st->start_pa \|\| st->start_pa != st->last_pa))) {
8eb07b18 RG	222
	223	/* Check the PTE flags */
	224	if (st->current_flags) {
453d87f6	225	note_prot_wx(st, addr);
8eb07b18 RG	226	dump_addr(st, addr);
	227
	228	/* Dump all the flags */
	229	if (pg_level[st->level].flag)
	230	dump_flag_info(st, pg_level[st->level].flag,
	231	st->current_flags,
	232	pg_level[st->level].num);
	233
5f18cbdb	234	pt_dump_seq_putc(st->seq, '\n');
8eb07b18 RG	235	}
	236
	237	/*
	238	* Address indicates we have passed the end of the
	239	* current section of virtual memory
	240	*/
	241	while (addr >= st->marker[1].start_address) {
	242	st->marker++;
5f18cbdb	243	pt_dump_seq_printf(st->seq, "---[ %s ]---\n", st->marker->name);
8eb07b18 RG	244	}
8eb07b18 RG	245	st->start_address = addr;
aaa22952 OH	246	st->start_pa = pa;
aaa22952 OH	247	st->last_pa = pa;
8eb07b18 RG	248	st->current_flags = flag;
8eb07b18 RG	249	st->level = level;
aaa22952 OH	250	} else {
aaa22952 OH	251	st->last_pa = pa;
8eb07b18 RG	252	}
	253	}
	254
	255	static void walk_pte(struct pg_state st, pmd_t pmd, unsigned long start)
	256	{
	257	pte_t *pte = pte_offset_kernel(pmd, 0);
	258	unsigned long addr;
	259	unsigned int i;
	260
	261	for (i = 0; i < PTRS_PER_PTE; i++, pte++) {
	262	addr = start + i * PAGE_SIZE;
	263	note_page(st, addr, 4, pte_val(*pte));
	264
	265	}
	266	}
	267
	268	static void walk_pmd(struct pg_state st, pud_t pud, unsigned long start)
	269	{
	270	pmd_t *pmd = pmd_offset(pud, 0);
	271	unsigned long addr;
	272	unsigned int i;
	273
	274	for (i = 0; i < PTRS_PER_PMD; i++, pmd++) {
	275	addr = start + i * PMD_SIZE;
bfb9956a	276	if (!pmd_none(pmd) && !pmd_huge(pmd))
8eb07b18 RG	277	/* pmd exists */
	278	walk_pte(st, pmd, addr);
	279	else
	280	note_page(st, addr, 3, pmd_val(*pmd));
	281	}
	282	}
	283
	284	static void walk_pud(struct pg_state st, pgd_t pgd, unsigned long start)
	285	{
	286	pud_t *pud = pud_offset(pgd, 0);
	287	unsigned long addr;
	288	unsigned int i;
	289
	290	for (i = 0; i < PTRS_PER_PUD; i++, pud++) {
	291	addr = start + i * PUD_SIZE;
bfb9956a	292	if (!pud_none(pud) && !pud_huge(pud))
8eb07b18 RG	293	/* pud exists */
	294	walk_pmd(st, pud, addr);
	295	else
	296	note_page(st, addr, 2, pud_val(*pud));
	297	}
	298	}
	299
	300	static void walk_pagetables(struct pg_state *st)
	301	{
	302	pgd_t *pgd = pgd_offset_k(0UL);
	303	unsigned int i;
	304	unsigned long addr;
	305
0d923962 ME	306	addr = st->start_address;
0d923962 ME	307
8eb07b18 RG	308	/*
	309	* Traverse the linux pagetable structure and dump pages that are in
	310	* the hash pagetable.
	311	*/
0d923962	312	for (i = 0; i < PTRS_PER_PGD; i++, pgd++, addr += PGDIR_SIZE) {
bfb9956a	313	if (!pgd_none(pgd) && !pgd_huge(pgd))
8eb07b18 RG	314	/* pgd exists */
	315	walk_pud(st, pgd, addr);
	316	else
	317	note_page(st, addr, 1, pgd_val(*pgd));
	318	}
	319	}
	320
	321	static void populate_markers(void)
	322	{
6c01bbd2 CL	323	int i = 0;
	324
	325	address_markers[i++].start_address = PAGE_OFFSET;
	326	address_markers[i++].start_address = VMALLOC_START;
	327	address_markers[i++].start_address = VMALLOC_END;
	328	#ifdef CONFIG_PPC64
	329	address_markers[i++].start_address = ISA_IO_BASE;
	330	address_markers[i++].start_address = ISA_IO_END;
	331	address_markers[i++].start_address = PHB_IO_BASE;
	332	address_markers[i++].start_address = PHB_IO_END;
	333	address_markers[i++].start_address = IOREMAP_BASE;
	334	address_markers[i++].start_address = IOREMAP_END;
0034d395	335	/* What is the ifdef about? */
4e003747	336	#ifdef CONFIG_PPC_BOOK3S_64
0034d395	337	address_markers[i++].start_address = H_VMEMMAP_START;
8eb07b18	338	#else
6c01bbd2 CL	339	address_markers[i++].start_address = VMEMMAP_BASE;
	340	#endif
	341	#else /* !CONFIG_PPC64 */
	342	address_markers[i++].start_address = ioremap_bot;
	343	address_markers[i++].start_address = IOREMAP_TOP;
	344	#ifdef CONFIG_NOT_COHERENT_CACHE
	345	address_markers[i++].start_address = IOREMAP_TOP;
	346	address_markers[i++].start_address = IOREMAP_TOP +
	347	CONFIG_CONSISTENT_SIZE;
	348	#endif
	349	#ifdef CONFIG_HIGHMEM
	350	address_markers[i++].start_address = PKMAP_BASE;
	351	address_markers[i++].start_address = PKMAP_ADDR(LAST_PKMAP);
8eb07b18	352	#endif
6c01bbd2 CL	353	address_markers[i++].start_address = FIXADDR_START;
6c01bbd2 CL	354	address_markers[i++].start_address = FIXADDR_TOP;
b4abe38f CL	355	#ifdef CONFIG_KASAN
	356	address_markers[i++].start_address = KASAN_SHADOW_START;
	357	address_markers[i++].start_address = KASAN_SHADOW_END;
	358	#endif
6c01bbd2	359	#endif /* CONFIG_PPC64 */
8eb07b18 RG	360	}
	361
	362	static int ptdump_show(struct seq_file m, void v)
	363	{
	364	struct pg_state st = {
	365	.seq = m,
8eb07b18 RG	366	.marker = address_markers,
8eb07b18 RG	367	};
0d923962 ME	368
	369	if (radix_enabled())
	370	st.start_address = PAGE_OFFSET;
	371	else
	372	st.start_address = KERN_VIRT_START;
	373
8eb07b18 RG	374	/* Traverse kernel page tables */
	375	walk_pagetables(&st);
	376	note_page(&st, 0, 0, 0);
	377	return 0;
	378	}
	379
	380
	381	static int ptdump_open(struct inode inode, struct file file)
	382	{
	383	return single_open(file, ptdump_show, NULL);
	384	}
	385
	386	static const struct file_operations ptdump_fops = {
	387	.open = ptdump_open,
	388	.read = seq_read,
	389	.llseek = seq_lseek,
	390	.release = single_release,
	391	};
	392
	393	static void build_pgtable_complete_mask(void)
	394	{
	395	unsigned int i, j;
	396
	397	for (i = 0; i < ARRAY_SIZE(pg_level); i++)
	398	if (pg_level[i].flag)
	399	for (j = 0; j < pg_level[i].num; j++)
	400	pg_level[i].mask \|= pg_level[i].flag[j].mask;
	401	}
	402
453d87f6 RC	403	#ifdef CONFIG_PPC_DEBUG_WX
	404	void ptdump_check_wx(void)
	405	{
	406	struct pg_state st = {
	407	.seq = NULL,
	408	.marker = address_markers,
	409	.check_wx = true,
	410	};
	411
	412	if (radix_enabled())
	413	st.start_address = PAGE_OFFSET;
	414	else
	415	st.start_address = KERN_VIRT_START;
	416
	417	walk_pagetables(&st);
	418
	419	if (st.wx_pages)
	420	pr_warn("Checked W+X mappings: FAILED, %lu W+X pages found\n",
	421	st.wx_pages);
	422	else
	423	pr_info("Checked W+X mappings: passed, no W+X pages found\n");
	424	}
	425	#endif
	426
8eb07b18 RG	427	static int ptdump_init(void)
	428	{
	429	struct dentry *debugfs_file;
	430
	431	populate_markers();
	432	build_pgtable_complete_mask();
2505820f	433	debugfs_file = debugfs_create_file("kernel_page_tables", 0400, NULL,
8eb07b18 RG	434	NULL, &ptdump_fops);
	435	return debugfs_file ? 0 : -ENOMEM;
	436	}
	437	device_initcall(ptdump_init);