Skip to contentSkip to conversation

page-types tool updates

PATCH New Wu Fengguang 1 participant 9 messages Sep 16, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover page-types tool updates

Hi Andrew,

Here are some updates for page-types. The first 2 patches are trivial additions of the KSM and HWPOISON bits, which should be OK for .32.

The latter patches are mainly code refactors (no behavior change) plus one final patch for adding the hwpoison/unpoison feature. They are tested OK, but could delay merging them if necessary.

Thanks,
Fengguang

#1 lore raw

PATCH 1/8 pagemap: export KPF_HWPOISON

This flag indicates a hardware detected memory corruption on the page. Any future access of the page data may bring down the machine.

---
 Documentation/vm/page-types.c |    2 ++
 Documentation/vm/pagemap.txt  |    4 ++++
 fs/proc/page.c                |    5 +++++
 3 files changed, 11 insertions(+)
fs/proc/page.c +5 -0
@@ -94,6 +94,7 @@ static const struct file_operations proc
 #define KPF_COMPOUND_TAIL	16
 #define KPF_HUGE		17
 #define KPF_UNEVICTABLE		18
+#define KPF_HWPOISON		19
 #define KPF_NOPAGE		20
 
 #define KPF_KSM			21
@@ -180,6 +181,10 @@ u64 page_uflags(struct page *page)
 	u |= kpf_copy_bit(k, KPF_UNEVICTABLE,	PG_unevictable);
 	u |= kpf_copy_bit(k, KPF_MLOCKED,	PG_mlocked);
 
+#ifdef CONFIG_MEMORY_FAILURE
+	u |= kpf_copy_bit(k, KPF_HWPOISON,	PG_hwpoison);
+#endif
+
 #ifdef CONFIG_IA64_UNCACHED_ALLOCATOR
 	u |= kpf_copy_bit(k, KPF_UNCACHED,	PG_uncached);
 #endif
Documentation/vm/page-types.c +2 -0
@@ -69,6 +69,7 @@
 #define KPF_COMPOUND_TAIL	16
 #define KPF_HUGE		17
 #define KPF_UNEVICTABLE		18
+#define KPF_HWPOISON		19
 #define KPF_NOPAGE		20
 
 /* [32-] kernel hacking assistances */
@@ -116,6 +117,7 @@ static char *page_flag_names[] = {
 	[KPF_COMPOUND_TAIL]	= "T:compound_tail",
 	[KPF_HUGE]		= "G:huge",
 	[KPF_UNEVICTABLE]	= "u:unevictable",
+	[KPF_HWPOISON]		= "X:hwpoison",
 	[KPF_NOPAGE]		= "n:nopage",
 
 	[KPF_RESERVED]		= "r:reserved",
Documentation/vm/pagemap.txt +4 -0
@@ -57,6 +57,7 @@ There are three components to pagemap:
     16. COMPOUND_TAIL
     16. HUGE
     18. UNEVICTABLE
+    19. HWPOISON
     20. NOPAGE
 
 Short descriptions to the page flags:
@@ -86,6 +87,9 @@ Short descriptions to the page flags:
 17. HUGE
     this is an integral part of a HugeTLB page
 
+19. HWPOISON
+    hardware detected memory corruption on this page: don't touch the data!
+
 20. NOPAGE
     no page frame exists at the requested address
 
#2 lore raw

PATCH 2/8 pagemap: document KPF_KSM and show it in page-types

In reply to Wu Fengguang

It indicates to the system admin that processes mapping such pages may be eating less physical memory than the reported numbers by legacy tools.

---
 Documentation/vm/page-types.c |    2 ++
 Documentation/vm/pagemap.txt  |    4 ++++
 2 files changed, 6 insertions(+)
Documentation/vm/pagemap.txt +4 -0
@@ -59,6 +59,7 @@ There are three components to pagemap:
     18. UNEVICTABLE
     19. HWPOISON
     20. NOPAGE
+    21. KSM
 
 Short descriptions to the page flags:
 
@@ -93,6 +94,9 @@ Short descriptions to the page flags:
 20. NOPAGE
     no page frame exists at the requested address
 
+21. KSM
+    identical memory pages dynamically shared between one or more processes
+
     [IO related page flags]
  1. ERROR     IO error occurred
  3. UPTODATE  page has up-to-date data
Documentation/vm/page-types.c +2 -0
@@ -71,6 +71,7 @@
 #define KPF_UNEVICTABLE		18
 #define KPF_HWPOISON		19
 #define KPF_NOPAGE		20
+#define KPF_KSM			21
 
 /* [32-] kernel hacking assistances */
 #define KPF_RESERVED		32
@@ -119,6 +120,7 @@ static char *page_flag_names[] = {
 	[KPF_UNEVICTABLE]	= "u:unevictable",
 	[KPF_HWPOISON]		= "X:hwpoison",
 	[KPF_NOPAGE]		= "n:nopage",
+	[KPF_KSM]		= "x:ksm",
 
 	[KPF_RESERVED]		= "r:reserved",
 	[KPF_MLOCKED]		= "m:mlocked",
#3 lore raw

PATCH 3/8 page-types: add GPL note

In reply to Wu Fengguang
---
 Documentation/vm/page-types.c |    5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
Documentation/vm/page-types.c +4 -1
@@ -2,7 +2,10 @@
  * page-types: Tool for querying page flags
  *
  * Copyright (C) 2009 Intel corporation
- * Copyright (C) 2009 Wu Fengguang <[email protected]>
+ *
+ * Authors: Wu Fengguang <[email protected]>
+ *
+ * Released under the General Public License (GPL).
  */
 
 #define _LARGEFILE64_SOURCE
#4 lore raw

PATCH 4/8 page-types: introduce checked_open()

In reply to Wu Fengguang

This helps merge duplicate code (now and future)
and outstand the main logic.

---
 Documentation/vm/page-types.c |   24 ++++++++++++++----------
 1 file changed, 14 insertions(+), 10 deletions(-)
Documentation/vm/page-types.c +14 -10
@@ -211,6 +211,18 @@ static void fatal(const char *x, ...)
 	exit(EXIT_FAILURE);
 }
 
+int checked_open(const char *pathname, int flags)
+{
+	int fd = open(pathname, flags);
+
+	if (fd < 0) {
+		perror(pathname);
+		exit(EXIT_FAILURE);
+	}
+
+	return fd;
+}
+
 
 /*
  * page flag names
@@ -531,11 +543,7 @@ static void walk_addr_ranges(void)
 {
 	int i;
 
-	kpageflags_fd = open(PROC_KPAGEFLAGS, O_RDONLY);
-	if (kpageflags_fd < 0) {
-		perror(PROC_KPAGEFLAGS);
-		exit(EXIT_FAILURE);
-	}
+	kpageflags_fd = checked_open(PROC_KPAGEFLAGS, O_RDONLY);
 
 	if (!nr_addr_ranges)
 		add_addr_range(0, ULONG_MAX);
@@ -628,11 +636,7 @@ static void parse_pid(const char *str)
 	opt_pid = parse_number(str);
 
 	sprintf(buf, "/proc/%d/pagemap", opt_pid);
-	pagemap_fd = open(buf, O_RDONLY);
-	if (pagemap_fd < 0) {
-		perror(buf);
-		exit(EXIT_FAILURE);
-	}
+	pagemap_fd = checked_open(buf, O_RDONLY);
 
 	sprintf(buf, "/proc/%d/maps", opt_pid);
 	file = fopen(buf, "r");
#5 lore raw

PATCH 5/8 page-types: make standalone pagemap/kpageflags read routines

In reply to Wu Fengguang

Refactor the code to be more modular and easier to reuse.

---
 Documentation/vm/page-types.c |  156 ++++++++++++++++++--------------
 1 file changed, 90 insertions(+), 66 deletions(-)
Documentation/vm/page-types.c +90 -66
@@ -154,8 +154,6 @@ static unsigned long	pg_start[MAX_VMAS];
 static unsigned long	pg_end[MAX_VMAS];
 static unsigned long	voffset;
 
-static int		pagemap_fd;
-
 #define MAX_BIT_FILTERS	64
 static int		nr_bit_filters;
 static uint64_t		opt_mask[MAX_BIT_FILTERS];
@@ -163,7 +161,7 @@ static uint64_t		opt_bits[MAX_BIT_FILTER
 
 static int		page_size;
 
-#define PAGES_BATCH	(64 << 10)	/* 64k pages */
+static int		pagemap_fd;
 static int		kpageflags_fd;
 
 #define HASH_SHIFT	13
@@ -219,6 +217,62 @@ int checked_open(const char *pathname, i
 	return fd;
 }
 
+/*
+ * pagemap/kpageflags routines
+ */
+
+static unsigned long do_u64_read(int fd, char *name,
+				 uint64_t *buf,
+				 unsigned long index,
+				 unsigned long count)
+{
+	long bytes;
+
+	if (index > ULONG_MAX / 8)
+		fatal("index overflow: %lu\n", index);
+
+	if (lseek(fd, index * 8, SEEK_SET) < 0) {
+		perror(name);
+		exit(EXIT_FAILURE);
+	}
+
+	bytes = read(fd, buf, count * 8);
+	if (bytes < 0) {
+		perror(name);
+		exit(EXIT_FAILURE);
+	}
+	if (bytes % 8)
+		fatal("partial read: %lu bytes\n", bytes);
+
+	return bytes / 8;
+}
+
+static unsigned long kpageflags_read(uint64_t *buf,
+				     unsigned long index,
+				     unsigned long pages)
+{
+	return do_u64_read(kpageflags_fd, PROC_KPAGEFLAGS, buf, index, pages);
+}
+
+static unsigned long pagemap_read(uint64_t *buf,
+				  unsigned long index,
+				  unsigned long pages)
+{
+	return do_u64_read(pagemap_fd, "/proc/pid/pagemap", buf, index, pages);
+}
+
+static unsigned long pagemap_pfn(uint64_t val)
+{
+	unsigned long pfn;
+
+	if (val & PM_PRESENT)
+		pfn = PM_PFRAME(val);
+	else
+		pfn = 0;
+
+	return pfn;
+}
+
 
 /*
  * page flag names
@@ -425,79 +479,53 @@ static void add_page(unsigned long offse
 	total_pages++;
 }
 
+#define KPAGEFLAGS_BATCH	(64 << 10)	/* 64k pages */
 static void walk_pfn(unsigned long index, unsigned long count)
 {
+	uint64_t buf[KPAGEFLAGS_BATCH];
 	unsigned long batch;
-	unsigned long n;
+	unsigned long pages;
 	unsigned long i;
 
-	if (index > ULONG_MAX / KPF_BYTES)
-		fatal("index overflow: %lu\n", index);
-
-	lseek(kpageflags_fd, index * KPF_BYTES, SEEK_SET);
-
 	while (count) {
-		uint64_t kpageflags_buf[KPF_BYTES * PAGES_BATCH];
-
-		batch = min_t(unsigned long, count, PAGES_BATCH);
-		n = read(kpageflags_fd, kpageflags_buf, batch * KPF_BYTES);
-		if (n == 0)
-			break;
-		if (n < 0) {
-			perror(PROC_KPAGEFLAGS);
-			exit(EXIT_FAILURE);
-		}
-
-		if (n % KPF_BYTES != 0)
-			fatal("partial read: %lu bytes\n", n);
-		n = n / KPF_BYTES;
+		batch = min_t(unsigned long, count, KPAGEFLAGS_BATCH);
+		pages = kpageflags_read(buf, index, batch);
+		if (pages == 0)
+			break;
 
-		for (i = 0; i < n; i++)
-			add_page(index + i, kpageflags_buf[i]);
+		for (i = 0; i < pages; i++)
+			add_page(index + i, buf[i]);
 
-		index += batch;
-		count -= batch;
+		index += pages;
+		count -= pages;
 	}
 }
 
-
-#define PAGEMAP_BATCH	4096
-static unsigned long task_pfn(unsigned long pgoff)
+#define PAGEMAP_BATCH	(64 << 10)
+static void walk_vma(unsigned long index, unsigned long count)
 {
-	static uint64_t buf[PAGEMAP_BATCH];
-	static unsigned long start;
-	static long count;
-	uint64_t pfn;
+	uint64_t buf[PAGEMAP_BATCH];
+	unsigned long batch;
+	unsigned long pages;
+	unsigned long pfn;
+	unsigned long i;
 
-	if (pgoff < start || pgoff >= start + count) {
-		if (lseek64(pagemap_fd,
-			    (uint64_t)pgoff * PM_ENTRY_BYTES,
-			    SEEK_SET) < 0) {
-			perror("pagemap seek");
-			exit(EXIT_FAILURE);
-		}
-		count = read(pagemap_fd, buf, sizeof(buf));
-		if (count == 0)
-			return 0;
-		if (count < 0) {
-			perror("pagemap read");
-			exit(EXIT_FAILURE);
-		}
-		if (count % PM_ENTRY_BYTES) {
-			fatal("pagemap read not aligned.\n");
-			exit(EXIT_FAILURE);
-		}
-		count /= PM_ENTRY_BYTES;
-		start = pgoff;
-	}
+	while (count) {
+		batch = min_t(unsigned long, count, PAGEMAP_BATCH);
+		pages = pagemap_read(buf, index, batch);
+		if (pages == 0)
+			break;
 
-	pfn = buf[pgoff - start];
-	if (pfn & PM_PRESENT)
-		pfn = PM_PFRAME(pfn);
-	else
-		pfn = 0;
+		for (i = 0; i < pages; i++) {
+			pfn = pagemap_pfn(buf[i]);
+			voffset = index + i;
+			if (pfn)
+				walk_pfn(pfn, 1);
+		}
 
-	return pfn;
+		index += pages;
+		count -= pages;
+	}
 }
 
 static void walk_task(unsigned long index, unsigned long count)
@@ -517,11 +545,7 @@ static void walk_task(unsigned long inde
 		index   = min_t(unsigned long, pg_end[i], end);
 
 		assert(voffset < index);
-		for (; voffset < index; voffset++) {
-			unsigned long pfn = task_pfn(voffset);
-			if (pfn)
-				walk_pfn(pfn, 1);
-		}
+		walk_vma(voffset, index - voffset);
 	}
 }
 
#6 lore raw

PATCH 6/8 page-types: make voffset local variables

In reply to Wu Fengguang
---
 Documentation/vm/page-types.c |   39 +++++++++++++++++---------------
 1 file changed, 21 insertions(+), 18 deletions(-)
Documentation/vm/page-types.c +21 -18
@@ -159,7 +159,6 @@ static unsigned long	opt_size[MAX_ADDR_R
 static int		nr_vmas;
 static unsigned long	pg_start[MAX_VMAS];
 static unsigned long	pg_end[MAX_VMAS];
-static unsigned long	voffset;
 
 #define MAX_BIT_FILTERS	64
 static int		nr_bit_filters;
@@ -328,7 +327,8 @@ static char *page_flag_longname(uint64_t
  * page list and summary
  */
 
-static void show_page_range(unsigned long offset, uint64_t flags)
+static void show_page_range(unsigned long voffset,
+			    unsigned long offset, uint64_t flags)
 {
 	static uint64_t      flags0;
 	static unsigned long voff;
@@ -354,7 +354,8 @@ static void show_page_range(unsigned lon
 	count  = 1;
 }
 
-static void show_page(unsigned long offset, uint64_t flags)
+static void show_page(unsigned long voffset,
+		      unsigned long offset, uint64_t flags)
 {
 	if (opt_pid)
 		printf("%lx\t", voffset);
@@ -435,7 +436,6 @@ static uint64_t well_known_flags(uint64_
 	return flags;
 }
 
-
 /*
  * page frame walker
  */
@@ -467,7 +467,8 @@ static int hash_slot(uint64_t flags)
 	exit(EXIT_FAILURE);
 }
 
-static void add_page(unsigned long offset, uint64_t flags)
+static void add_page(unsigned long voffset,
+		     unsigned long offset, uint64_t flags)
 {
 	flags = expand_overloaded_flags(flags);
 
@@ -478,16 +479,18 @@ static void add_page(unsigned long offse
 		return;
 
 	if (opt_list == 1)
-		show_page_range(offset, flags);
+		show_page_range(voffset, offset, flags);
 	else if (opt_list == 2)
-		show_page(offset, flags);
+		show_page(voffset, offset, flags);
 
 	nr_pages[hash_slot(flags)]++;
 	total_pages++;
 }
 
 #define KPAGEFLAGS_BATCH	(64 << 10)	/* 64k pages */
-static void walk_pfn(unsigned long index, unsigned long count)
+static void walk_pfn(unsigned long voffset,
+		     unsigned long index,
+		     unsigned long count)
 {
 	uint64_t buf[KPAGEFLAGS_BATCH];
 	unsigned long batch;
@@ -501,7 +504,7 @@ static void walk_pfn(unsigned long index
 			break;
 
 		for (i = 0; i < pages; i++)
-			add_page(index + i, buf[i]);
+			add_page(voffset + i, index + i, buf[i]);
 
 		index += pages;
 		count -= pages;
@@ -525,9 +528,8 @@ static void walk_vma(unsigned long index
 
 		for (i = 0; i < pages; i++) {
 			pfn = pagemap_pfn(buf[i]);
-			voffset = index + i;
 			if (pfn)
-				walk_pfn(pfn, 1);
+				walk_pfn(index + i, pfn, 1);
 		}
 
 		index += pages;
@@ -537,8 +539,9 @@ static void walk_vma(unsigned long index
 
 static void walk_task(unsigned long index, unsigned long count)
 {
-	int i = 0;
 	const unsigned long end = index + count;
+	unsigned long start;
+	int i = 0;
 
 	while (index < end) {
 
@@ -548,11 +551,11 @@ static void walk_task(unsigned long inde
 		if (pg_start[i] >= end)
 			return;
 
-		voffset = max_t(unsigned long, pg_start[i], index);
-		index   = min_t(unsigned long, pg_end[i], end);
+		start = max_t(unsigned long, pg_start[i], index);
+		index = min_t(unsigned long, pg_end[i], end);
 
-		assert(voffset < index);
-		walk_vma(voffset, index - voffset);
+		assert(start < index);
+		walk_vma(start, index - start);
 	}
 }
 
@@ -577,7 +580,7 @@ static void walk_addr_ranges(void)
 
 	for (i = 0; i < nr_addr_ranges; i++)
 		if (!opt_pid)
-			walk_pfn(opt_offset[i], opt_size[i]);
+			walk_pfn(0, opt_offset[i], opt_size[i]);
 		else
 			walk_task(opt_offset[i], opt_size[i]);
 
@@ -879,7 +882,7 @@ int main(int argc, char *argv[])
 	walk_addr_ranges();
 
 	if (opt_list == 1)
-		show_page_range(0, 0);  /* drain the buffer */
+		show_page_range(0, 0, 0);  /* drain the buffer */
 
 	if (opt_no_summary)
 		return 0;
#7 lore raw

PATCH 7/8 page-types: introduce kpageflags_flags()

In reply to Wu Fengguang
---
 Documentation/vm/page-types.c |   15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)
Documentation/vm/page-types.c +11 -4
@@ -436,6 +436,16 @@ static uint64_t well_known_flags(uint64_
 	return flags;
 }
 
+static uint64_t kpageflags_flags(uint64_t flags)
+{
+	flags = expand_overloaded_flags(flags);
+
+	if (!opt_raw)
+		flags = well_known_flags(flags);
+
+	return flags;
+}
+
 /*
  * page frame walker
  */
@@ -470,10 +480,7 @@ static int hash_slot(uint64_t flags)
 static void add_page(unsigned long voffset,
 		     unsigned long offset, uint64_t flags)
 {
-	flags = expand_overloaded_flags(flags);
-
-	if (!opt_raw)
-		flags = well_known_flags(flags);
+	flags = kpageflags_flags(flags);
 
 	if (!bit_mask_ok(flags))
 		return;
#8 lore raw

PATCH 8/8 page-types: add hwpoison/unpoison feature

In reply to Wu Fengguang

For hwpoison stress testing.
The debugfs mount point is assumed to be /debug/.

---
 Documentation/vm/page-types.c |   73 +++++++++++++++++++++++++++++++-
 1 file changed, 72 insertions(+), 1 deletion(-)
Documentation/vm/page-types.c +72 -1
@@ -170,6 +170,13 @@ static int		page_size;
 static int		pagemap_fd;
 static int		kpageflags_fd;
 
+static int		opt_hwpoison;
+static int		opt_unpoison;
+
+static char		*hwpoison_debug_fs = "/debug/hwpoison";
+static int		hwpoison_inject_fd;
+static int		hwpoison_forget_fd;
+
 #define HASH_SHIFT	13
 #define HASH_SIZE	(1 << HASH_SHIFT)
 #define HASH_MASK	(HASH_SIZE - 1)
@@ -447,6 +454,53 @@ static uint64_t kpageflags_flags(uint64_
 }
 
 /*
+ * page actions
+ */
+
+static void prepare_hwpoison_fd(void)
+{
+	char buf[100];
+
+	if (opt_hwpoison && !hwpoison_inject_fd) {
+		sprintf(buf, "%s/corrupt-pfn", hwpoison_debug_fs);
+		hwpoison_inject_fd = checked_open(buf, O_WRONLY);
+	}
+
+	if (opt_unpoison && !hwpoison_forget_fd) {
+		sprintf(buf, "%s/renew-pfn", hwpoison_debug_fs);
+		hwpoison_forget_fd = checked_open(buf, O_WRONLY);
+	}
+}
+
+static int hwpoison_page(unsigned long offset)
+{
+	char buf[100];
+	int len;
+
+	len = sprintf(buf, "0x%lx\n", offset);
+	len = write(hwpoison_inject_fd, buf, len);
+	if (len < 0) {
+		perror("hwpoison inject");
+		return len;
+	}
+	return 0;
+}
+
+static int unpoison_page(unsigned long offset)
+{
+	char buf[100];
+	int len;
+
+	len = sprintf(buf, "0x%lx\n", offset);
+	len = write(hwpoison_forget_fd, buf, len);
+	if (len < 0) {
+		perror("hwpoison forget");
+		return len;
+	}
+	return 0;
+}
+
+/*
  * page frame walker
  */
 
@@ -485,6 +539,11 @@ static void add_page(unsigned long voffs
 	if (!bit_mask_ok(flags))
 		return;
 
+	if (opt_hwpoison)
+		hwpoison_page(offset);
+	if (opt_unpoison)
+		unpoison_page(offset);
+
 	if (opt_list == 1)
 		show_page_range(voffset, offset, flags);
 	else if (opt_list == 2)
@@ -624,6 +683,8 @@ static void usage(void)
 "            -l|--list                 Show page details in ranges\n"
 "            -L|--list-each            Show page details one by one\n"
 "            -N|--no-summary           Don't show summay info\n"
+"            -X|--hwpoison             hwpoison pages\n"
+"            -x|--unpoison             unpoison pages\n"
 "            -h|--help                 Show this usage message\n"
 "addr-spec:\n"
 "            N                         one page at offset N (unit: pages)\n"
@@ -833,6 +894,8 @@ static struct option opts[] = {
 	{ "list"      , 0, NULL, 'l' },
 	{ "list-each" , 0, NULL, 'L' },
 	{ "no-summary", 0, NULL, 'N' },
+	{ "hwpoison"  , 0, NULL, 'X' },
+	{ "unpoison"  , 0, NULL, 'x' },
 	{ "help"      , 0, NULL, 'h' },
 	{ NULL        , 0, NULL, 0 }
 };
@@ -844,7 +907,7 @@ int main(int argc, char *argv[])
 	page_size = getpagesize();
 
 	while ((c = getopt_long(argc, argv,
-				"rp:f:a:b:lLNh", opts, NULL)) != -1) {
+				"rp:f:a:b:lLNXxh", opts, NULL)) != -1) {
 		switch (c) {
 		case 'r':
 			opt_raw = 1;
@@ -870,6 +933,14 @@ int main(int argc, char *argv[])
 		case 'N':
 			opt_no_summary = 1;
 			break;
+		case 'X':
+			opt_hwpoison = 1;
+			prepare_hwpoison_fd();
+			break;
+		case 'x':
+			opt_unpoison = 1;
+			prepare_hwpoison_fd();
+			break;
 		case 'h':
 			usage();
 			exit(0);