[patch] x86: add code to dump the (kernel) page tables for visual inspection

[patch] x86: add code to dump the (kernel) page tables for visual inspection

[Arjan van de Ven]

Subject: x86: add code to dump the (kernel) page tables for visual inspection by kernel developers
From: Arjan van de Ven <arjan@linux.intel.com>

This patch adds code to the kernel to have an (optional)
/proc/kernel_page_tables debug file that basically dumps the kernel
pagetables; this allows us kernel developers to verify that nothing fishy is
going on and that the various mappings are set up correctly. This was quite
useful in finding various change_page_attr() bugs, and is very likely to be
useful in the future as well.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
---
 arch/x86/Kconfig.debug        |   11 +
 arch/x86/mm/Makefile_64       |    1 
 arch/x86/mm/dump_pagetables.c |  301 ++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 313 insertions(+)

Index: linux.trees.git/arch/x86/Kconfig.debug
===================================================================
--- linux.trees.git.orig/arch/x86/Kconfig.debug
+++ linux.trees.git/arch/x86/Kconfig.debug
@@ -70,6 +70,17 @@ config DEBUG_PER_CPU_MAPS
 
 	  Say N if unsure.
 
+config X86_PTDUMP
+	bool "Export kernel pagetable layout to userspace in /proc"
+	depends on X86_64
+	help
+	  Say Y here if you want to show the kernel pagetable layout in
+	  a /proc file. This information is only useful for kernel developers
+	  who are working in architecture specific areas of the kernel.
+	  It is probably not a good idea to enable this feature in a production
+	  kernel.
+	  If in doubt, say "N"
+
 config DEBUG_RODATA
 	bool "Write protect kernel read-only data structures"
 	default y
Index: linux.trees.git/arch/x86/mm/Makefile_64
===================================================================
--- linux.trees.git.orig/arch/x86/mm/Makefile_64
+++ linux.trees.git/arch/x86/mm/Makefile_64
@@ -7,3 +7,4 @@ obj-$(CONFIG_HUGETLB_PAGE) += hugetlbpag
 obj-$(CONFIG_NUMA) += numa_64.o
 obj-$(CONFIG_K8_NUMA) += k8topology_64.o
 obj-$(CONFIG_ACPI_NUMA) += srat_64.o
+obj-$(CONFIG_X86_PTDUMP) += dump_pagetables.o
Index: linux.trees.git/arch/x86/mm/dump_pagetables.c
===================================================================
--- /dev/null
+++ linux.trees.git/arch/x86/mm/dump_pagetables.c
@@ -0,0 +1,301 @@
+/*
+ * 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/module.h>
+#include <linux/seq_file.h>
+#include <linux/proc_fs.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;
+	int printed_vmalloc;
+	int printed_modules;
+	int printed_vmemmap;
+	int printed_highmap;
+};
+
+/* Multipliers for offsets within the PTEs */
+#define LEVEL_4_MULT (PAGE_SIZE)
+#define LEVEL_3_MULT (512UL * LEVEL_4_MULT)
+#define LEVEL_2_MULT (512UL * LEVEL_3_MULT)
+#define LEVEL_1_MULT (512UL * LEVEL_2_MULT)
+
+
+/*
+ * 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)
+{
+	unsigned long pr = pgprot_val(prot);
+
+	if (pr & _PAGE_USER)
+		seq_printf(m, "USR ");
+	else
+		seq_printf(m, "    ");
+	if (pr & _PAGE_RW)
+		seq_printf(m, "RW ");
+	else
+		seq_printf(m, "ro ");
+	if (pr & _PAGE_PWT)
+		seq_printf(m, "PWT ");
+	else
+		seq_printf(m, "    ");
+	if (pr & _PAGE_PCD)
+		seq_printf(m, "PCD ");
+	else
+		seq_printf(m, "    ");
+
+	/* Bit 9 has a different meaning on level 3 vs 4 */
+	if (level <= 3) {
+		if (pr & _PAGE_PSE)
+			seq_printf(m, "PSE ");
+		else
+			seq_printf(m, "    ");
+	} else {
+		if (pr & _PAGE_PAT)
+			seq_printf(m, "pat ");
+		else
+			seq_printf(m, "    ");
+	}
+	if (pr & _PAGE_GLOBAL)
+		seq_printf(m, "GLB ");
+	else
+		seq_printf(m, "    ");
+	if (pr & _PAGE_NX)
+		seq_printf(m, "NX ");
+	else
+		seq_printf(m, "x  ");
+}
+
+/*
+ * Sign-extend the 48 bit address to 64 bit
+ */
+static unsigned long sign_extend(unsigned long u)
+{
+	if (u>>47)
+		u = u | (0xffffUL << 48);
+	return u;
+}
+
+/*
+ * 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)
+{
+	unsigned long prot, cur;
+
+	/*
+	 * If we have a "break" in the series, we need to flush the state that
+	 * we have now. "break" is either changing perms or a different level.
+	 */
+	prot = pgprot_val(new_prot) & ~(PTE_MASK);
+	cur = pgprot_val(st->current_prot) & ~(PTE_MASK);
+
+	if ((prot != cur || level != st->level) &&
+				st->current_address != st->start_address) {
+		char unit = 'K';
+		unsigned long delta;
+
+		/*
+		 * We print markers for special areas of address space,
+		 * such as the start of vmalloc space etc.
+		 * This helps in the interpretation.
+		 */
+		if (!st->printed_vmalloc &&
+				st->start_address >= VMALLOC_START) {
+			seq_printf(m, "---[ VMALLOC SPACE ]---\n");
+			st->printed_vmalloc = 1;
+		}
+		if (!st->printed_modules &&
+				st->start_address >= MODULES_VADDR) {
+			seq_printf(m, "---[ MODULES SPACE ]---\n");
+			st->printed_modules = 1;
+		}
+		if (st->printed_modules < 2 &&
+				st->start_address >= MODULES_END) {
+			seq_printf(m, "---[ END MODULES SPACE ]---\n");
+			st->printed_modules = 2;
+		}
+		if (!st->printed_vmemmap &&
+				st->start_address >= VMEMMAP_START) {
+			seq_printf(m, "---[ VMMEMMAP SPACE ]---\n");
+			st->printed_vmemmap = 1;
+		}
+		if (!st->printed_highmap &&
+				st->start_address >= __START_KERNEL_map) {
+			seq_printf(m, "---[ HIGH KERNEL MAPPING ]---\n");
+			st->printed_highmap = 1;
+		}
+
+		/*
+		 * Now print the actual finished series
+		 */
+		seq_printf(m, "[ %016lx -  %016lx   ",
+				st->start_address, st->current_address);
+
+		delta = (st->current_address - st->start_address) >> 10;
+		if ((delta & 1023) == 0) {
+			delta = delta >> 10;
+			unit = 'M';
+		}
+		if (pgprot_val(st->current_prot)) {
+			seq_printf(m, "Size %9lu%cb ", delta, unit);
+			printk_prot(m, st->current_prot, st->level);
+			seq_printf(m, "L%i]\n", st->level);
+		} else {
+			/* don't print protections on non-present memory */
+			seq_printf(m, "%14lu%cb", delta, unit);
+			seq_printf(m, "                           L%i]\n",
+					st->level);
+		}
+		st->start_address = st->current_address;
+		st->current_prot = new_prot;
+		st->level = level;
+	};
+}
+
+static void walk_level_4(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 = sign_extend(P + i * LEVEL_4_MULT);
+		note_page(m, st, prot, 4);
+		start++;
+	}
+}
+
+
+static void walk_level_3(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 = sign_extend(P + i * LEVEL_3_MULT);
+		if (!pmd_none(*start)) {
+			unsigned long prot;
+
+			prot = pmd_val(*start) & ~(PTE_MASK);
+			/* Deal with 2Mb pages */
+			if (pmd_large(*start))
+				note_page(m, st, __pgprot(prot), 3);
+			else
+				walk_level_4(m, st, *start,
+							P + i * LEVEL_3_MULT);
+		} else
+			note_page(m, st, __pgprot(0), 3);
+		start++;
+	}
+}
+
+
+static void walk_level_2(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++) {
+		if (!pud_none(*start)) {
+			unsigned long prot;
+
+			prot = pud_val(*start) & ~(PTE_MASK);
+			/* Deal with 1Gb pages */
+			if (pud_large(*start))
+				note_page(m, st, __pgprot(prot), 2);
+			else
+				walk_level_3(m, st, *start,
+					P + i * LEVEL_2_MULT);
+		} else
+			note_page(m, st, __pgprot(0), 2);
+
+		start++;
+	}
+}
+
+static void walk_level_1(struct seq_file *m)
+{
+	pgd_t *start = (pgd_t *) &init_level4_pgt;
+	int i;
+	struct pg_state st;
+
+	memset(&st, 0, sizeof(st));
+	st.level = 1;
+
+	for (i = 0; i < PTRS_PER_PGD; i++) {
+		if (!pgd_none(*start))
+			walk_level_2(m, &st, *start, i * LEVEL_1_MULT);
+		else
+			note_page(m, &st, __pgprot(0), 1);
+		start++;
+	}
+}
+
+static int ptdump_show(struct seq_file *m, void *v)
+{
+	seq_puts(m, "Kernel pagetable dump\n");
+	walk_level_1(m);
+	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,
+};
+
+int pt_dump_init(void)
+{
+	struct proc_dir_entry *pe;
+
+	pe = create_proc_entry("kernel_page_tables", 0600, NULL);
+	if (!pe)
+		return -ENOMEM;
+
+	pe->proc_fops = &ptdump_fops;
+	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");

-- 
If you want to reach me at my work email, use arjan@linux.intel.com
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

Re: [patch] x86: add code to dump the (kernel) page tables for visual inspection

[Eric Dumazet]

Arjan van de Ven a écrit :
> Subject: x86: add code to dump the (kernel) page tables for visual inspection by kernel developers
> From: Arjan van de Ven <arjan@linux.intel.com>
> 
> This patch adds code to the kernel to have an (optional)
> /proc/kernel_page_tables debug file that basically dumps the kernel
> pagetables; this allows us kernel developers to verify that nothing fishy is
> going on and that the various mappings are set up correctly. This was quite
> useful in finding various change_page_attr() bugs, and is very likely to be
> useful in the future as well.
> 

Seems nice Arjan, but could we also add NUMA information ?

Re: [patch] x86: add code to dump the (kernel) page tables for visual inspection

[Arjan van de Ven]

On Tue, 05 Feb 2008 21:46:52 +0100
Eric Dumazet <dada1@cosmosbay.com> wrote:

> Arjan van de Ven a écrit :
> > Subject: x86: add code to dump the (kernel) page tables for visual
> > inspection by kernel developers From: Arjan van de Ven
> > <arjan@linux.intel.com>
> > 
> > This patch adds code to the kernel to have an (optional)
> > /proc/kernel_page_tables debug file that basically dumps the kernel
> > pagetables; this allows us kernel developers to verify that nothing
> > fishy is going on and that the various mappings are set up
> > correctly. This was quite useful in finding various
> > change_page_attr() bugs, and is very likely to be useful in the
> > future as well.
> > 
> 
> Seems nice Arjan, but could we also add NUMA information ?
> 


ok maybe I'm missing something, but the kernel VA space currently
is totally not numa dependent in any shape or form... 
so can you explain what kind of information you're looking for?
(I guess I just don't really understand your request; can you expand a bit on it?)


-- 
If you want to reach me at my work email, use arjan@linux.intel.com
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org

Re: [patch] x86: add code to dump the (kernel) page tables for visual inspection

[Pavel Machek]

On Mon 2008-02-04 22:26:41, Arjan van de Ven wrote:
> Subject: x86: add code to dump the (kernel) page tables for visual inspection by kernel developers
> From: Arjan van de Ven <arjan@linux.intel.com>
> 
> This patch adds code to the kernel to have an (optional)
> /proc/kernel_page_tables debug file that basically dumps the kernel

'No new crap goes to /proc?'

It does not seem to be per process -> it should probably go to
debugfs.

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html