Skip to contentSkip to conversation

mm, printk: dump full information of page flags in pGp

PATCH Superseded Yafang Shao 6 participants 20 messages Jan 28, 2021 – Jan 29, 2021

Subsystems (4)

v1 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

cover mm, printk: dump full information of page flags in pGp

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

This patchset also includes some code cleanup in mm/slub.c.

Below is the example of the output in mm/slub.c.
- Before the patchset

[ 6155.716018] INFO: Slab 0x000000004027dd4f objects=33 used=3 fp=0x000000008cd1579c flags=0x17ffffc0010200

- After the patchset

[ 6315.235783] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(Node 0x0,Zone 0x2,Lastcpupid 0x1fffff,slab|head)

Yafang Shao (3):
mm, slub: use pGp to print page flags
mm, slub: don't combine pr_err with INFO
printk: dump full information of page flags in pGp

 lib/vsprintf.c | 42 +++++++++++++++++++++++++++++++++++++++++-
 mm/slub.c      | 13 +++++++------
 2 files changed, 48 insertions(+), 7 deletions(-)
Signature · 1 line
2.17.1
#1 lore raw

PATCH 1/3 mm, slub: use pGp to print page flags

As pGp has been already introduced in printk, we'd better use it to make the output human readable.

Before this change, the output is,

[ 6155.716018] INFO: Slab 0x000000004027dd4f objects=33 used=3 fp=0x000000008cd1579c flags=0x17ffffc0010200

While after this change, the output is,

[ 8846.517809] INFO: Slab 0x00000000f42a2c60 objects=33 used=3 fp=0x0000000060d32ca8 flags=0x17ffffc0010200(slab|head)
---
 mm/slub.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
mm/slub.c +3 -2
index 69742ab9a21d..4b9ab267afbc 100644
@@ -641,8 +641,9 @@ void print_tracking(struct kmem_cache *s, void *object)
 
 static void print_page_info(struct page *page)
 {
-	pr_err("INFO: Slab 0x%p objects=%u used=%u fp=0x%p flags=0x%04lx\n",
-	       page, page->objects, page->inuse, page->freelist, page->flags);
+	pr_err("INFO: Slab 0x%p objects=%u used=%u fp=0x%p flags=%#lx(%pGp)\n",
+	       page, page->objects, page->inuse, page->freelist,
+	       page->flags, &page->flags);
 
 }
 
Signature · 1 line
2.17.1
Reviewed-by on v1 1/3
#2 lore raw

Re: [PATCH 1/3] mm, slub: use pGp to print page flags

Quoted 11 lines · Yafang Shao

As pGp has been already introduced in printk, we'd better use it to make the output human readable.

Before this change, the output is,

[ 6155.716018] INFO: Slab 0x000000004027dd4f objects=33 used=3 fp=0x000000008cd1579c flags=0x17ffffc0010200

While after this change, the output is,

[ 8846.517809] INFO: Slab 0x00000000f42a2c60 objects=33 used=3 fp=0x0000000060d32ca8 flags=0x17ffffc0010200(slab|head)
Quoted 21 lines
---
 mm/slub.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
mm/slub.c +3 -2
index 69742ab9a21d..4b9ab267afbc 100644
@@ -641,8 +641,9 @@ void print_tracking(struct kmem_cache *s, void *object)
 
 static void print_page_info(struct page *page)
 {
-	pr_err("INFO: Slab 0x%p objects=%u used=%u fp=0x%p flags=0x%04lx\n",
-	       page, page->objects, page->inuse, page->freelist, page->flags);
+	pr_err("INFO: Slab 0x%p objects=%u used=%u fp=0x%p flags=%#lx(%pGp)\n",
+	       page, page->objects, page->inuse, page->freelist,
+	       page->flags, &page->flags);
 
 }
 
reviewer Acked-by on v1 1/3
#3 lore raw

Re: [PATCH 1/3] mm, slub: use pGp to print page flags

In reply to Yafang Shao
Quoted 11 lines · Yafang Shao

As pGp has been already introduced in printk, we'd better use it to make the output human readable.

Before this change, the output is,

[ 6155.716018] INFO: Slab 0x000000004027dd4f objects=33 used=3 fp=0x000000008cd1579c flags=0x17ffffc0010200

While after this change, the output is,

[ 8846.517809] INFO: Slab 0x00000000f42a2c60 objects=33 used=3 fp=0x0000000060d32ca8 flags=0x17ffffc0010200(slab|head)
#4 lore raw

PATCH 2/3 mm, slub: don't combine pr_err with INFO

In reply to Yafang Shao

It is strange to combine "pr_err" with "INFO", so let's clean them up. This patch is motivated by David's comment[1].

- before the patch

[ 8846.517809] INFO: Slab 0x00000000f42a2c60 objects=33 used=3 fp=0x0000000060d32ca8 flags=0x17ffffc0010200(slab|head)

- after the patch

[ 6312.639698] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(slab|head)

[1]. https://lore.kernel.org/linux-mm/[email protected]/#t

Cc David Hildenbrand <[email protected]>
---
 mm/slub.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
mm/slub.c +5 -5
index 4b9ab267afbc..18b4474c8fa2 100644
@@ -615,7 +615,7 @@ static void print_track(const char *s, struct track *t, unsigned long pr_time)
 	if (!t->addr)
 		return;
 
-	pr_err("INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
+	pr_err("ERR: %s in %pS age=%lu cpu=%u pid=%d\n",
 	       s, (void *)t->addr, pr_time - t->when, t->cpu, t->pid);
 #ifdef CONFIG_STACKTRACE
 	{
@@ -641,7 +641,7 @@ void print_tracking(struct kmem_cache *s, void *object)
 
 static void print_page_info(struct page *page)
 {
-	pr_err("INFO: Slab 0x%p objects=%u used=%u fp=0x%p flags=%#lx(%pGp)\n",
+	pr_err("ERR: Slab 0x%p objects=%u used=%u fp=0x%p flags=%#lx(%pGp)\n",
 	       page, page->objects, page->inuse, page->freelist,
 	       page->flags, &page->flags);
 
@@ -698,7 +698,7 @@ static void print_trailer(struct kmem_cache *s, struct page *page, u8 *p)
 
 	print_page_info(page);
 
-	pr_err("INFO: Object 0x%p @offset=%tu fp=0x%p\n\n",
+	pr_err("ERR: Object 0x%p @offset=%tu fp=0x%p\n\n",
 	       p, p - addr, get_freepointer(s, p));
 
 	if (s->flags & SLAB_RED_ZONE)
@@ -791,7 +791,7 @@ static int check_bytes_and_report(struct kmem_cache *s, struct page *page,
 		end--;
 
 	slab_bug(s, "%s overwritten", what);
-	pr_err("INFO: 0x%p-0x%p @offset=%tu. First byte 0x%x instead of 0x%x\n",
+	pr_err("ERR: 0x%p-0x%p @offset=%tu. First byte 0x%x instead of 0x%x\n",
 					fault, end - 1, fault - addr,
 					fault[0], value);
 	print_trailer(s, page, object);
@@ -3855,7 +3855,7 @@ static void list_slab_objects(struct kmem_cache *s, struct page *page,
 	for_each_object(p, s, addr, page->objects) {
 
 		if (!test_bit(__obj_to_index(s, addr, p), map)) {
-			pr_err("INFO: Object 0x%p @offset=%tu\n", p, p - addr);
+			pr_err("ERR: Object 0x%p @offset=%tu\n", p, p - addr);
 			print_tracking(s, p);
 		}
 	}
Signature · 1 line
2.17.1
#5 lore raw

Re: [PATCH 2/3] mm, slub: don't combine pr_err with INFO

Quoted 6 lines · Yafang Shao

It is strange to combine "pr_err" with "INFO", so let's clean them up. This patch is motivated by David's comment[1].

- before the patch

[ 8846.517809] INFO: Slab 0x00000000f42a2c60 objects=33 used=3 fp=0x0000000060d32ca8 flags=0x17ffffc0010200(slab|head)
Quoted 7 lines

- after the patch

[ 6312.639698] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(slab|head)

[1]. https://lore.kernel.org/linux-mm/[email protected]/#t

Cc David Hildenbrand <[email protected]>

These are usually printed as part of slab_bug() with its prominent banner. In that sense it's additional details, thus INFO. The details itself are not error, thus ERR makes little sense imho. How about removing the prefix completely, or just replacing with an ident to make it visually part of the BUG report.

Quoted 54 lines
---
 mm/slub.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
mm/slub.c +5 -5
index 4b9ab267afbc..18b4474c8fa2 100644
@@ -615,7 +615,7 @@ static void print_track(const char *s, struct track *t, unsigned long pr_time)
 	if (!t->addr)
 		return;
 
-	pr_err("INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
+	pr_err("ERR: %s in %pS age=%lu cpu=%u pid=%d\n",
 	       s, (void *)t->addr, pr_time - t->when, t->cpu, t->pid);
 #ifdef CONFIG_STACKTRACE
 	{
@@ -641,7 +641,7 @@ void print_tracking(struct kmem_cache *s, void *object)
 
 static void print_page_info(struct page *page)
 {
-	pr_err("INFO: Slab 0x%p objects=%u used=%u fp=0x%p flags=%#lx(%pGp)\n",
+	pr_err("ERR: Slab 0x%p objects=%u used=%u fp=0x%p flags=%#lx(%pGp)\n",
 	       page, page->objects, page->inuse, page->freelist,
 	       page->flags, &page->flags);
 
@@ -698,7 +698,7 @@ static void print_trailer(struct kmem_cache *s, struct page *page, u8 *p)
 
 	print_page_info(page);
 
-	pr_err("INFO: Object 0x%p @offset=%tu fp=0x%p\n\n",
+	pr_err("ERR: Object 0x%p @offset=%tu fp=0x%p\n\n",
 	       p, p - addr, get_freepointer(s, p));
 
 	if (s->flags & SLAB_RED_ZONE)
@@ -791,7 +791,7 @@ static int check_bytes_and_report(struct kmem_cache *s, struct page *page,
 		end--;
 
 	slab_bug(s, "%s overwritten", what);
-	pr_err("INFO: 0x%p-0x%p @offset=%tu. First byte 0x%x instead of 0x%x\n",
+	pr_err("ERR: 0x%p-0x%p @offset=%tu. First byte 0x%x instead of 0x%x\n",
 					fault, end - 1, fault - addr,
 					fault[0], value);
 	print_trailer(s, page, object);
@@ -3855,7 +3855,7 @@ static void list_slab_objects(struct kmem_cache *s, struct page *page,
 	for_each_object(p, s, addr, page->objects) {
 
 		if (!test_bit(__obj_to_index(s, addr, p), map)) {
-			pr_err("INFO: Object 0x%p @offset=%tu\n", p, p - addr);
+			pr_err("ERR: Object 0x%p @offset=%tu\n", p, p - addr);
 			print_tracking(s, p);
 		}
 	}
#6 lore raw

Re: [PATCH 2/3] mm, slub: don't combine pr_err with INFO

Quoted 13 lines · Vlastimil Babka
Quoted 11 lines · Yafang Shao

It is strange to combine "pr_err" with "INFO", so let's clean them up. This patch is motivated by David's comment[1].

- before the patch

[ 8846.517809] INFO: Slab 0x00000000f42a2c60 objects=33 used=3 fp=0x0000000060d32ca8 flags=0x17ffffc0010200(slab|head)

- after the patch

[ 6312.639698] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(slab|head)

[1]. https://lore.kernel.org/linux-mm/[email protected]/#t

Quoted 8 lines
Quoted 2 lines
Cc David Hildenbrand <[email protected]>

These are usually printed as part of slab_bug() with its prominent banner. In that sense it's additional details, thus INFO. The details itself are not error, thus ERR makes little sense imho. How about removing the prefix completely, or just replacing with an ident to make it visually part of the BUG report.

Thanks for the explanation. I will remove the prefix completely in the next version.

Quoted 55 lines
Quoted 54 lines
---
 mm/slub.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
mm/slub.c +5 -5
index 4b9ab267afbc..18b4474c8fa2 100644
@@ -615,7 +615,7 @@ static void print_track(const char *s, struct track *t, unsigned long pr_time)
      if (!t->addr)
              return;
 
-     pr_err("INFO: %s in %pS age=%lu cpu=%u pid=%d\n",
+     pr_err("ERR: %s in %pS age=%lu cpu=%u pid=%d\n",
             s, (void *)t->addr, pr_time - t->when, t->cpu, t->pid);
 #ifdef CONFIG_STACKTRACE
      {
@@ -641,7 +641,7 @@ void print_tracking(struct kmem_cache *s, void *object)
 
 static void print_page_info(struct page *page)
 {
-     pr_err("INFO: Slab 0x%p objects=%u used=%u fp=0x%p flags=%#lx(%pGp)\n",
+     pr_err("ERR: Slab 0x%p objects=%u used=%u fp=0x%p flags=%#lx(%pGp)\n",
             page, page->objects, page->inuse, page->freelist,
             page->flags, &page->flags);
 
@@ -698,7 +698,7 @@ static void print_trailer(struct kmem_cache *s, struct page *page, u8 *p)
 
      print_page_info(page);
 
-     pr_err("INFO: Object 0x%p @offset=%tu fp=0x%p\n\n",
+     pr_err("ERR: Object 0x%p @offset=%tu fp=0x%p\n\n",
             p, p - addr, get_freepointer(s, p));
 
      if (s->flags & SLAB_RED_ZONE)
@@ -791,7 +791,7 @@ static int check_bytes_and_report(struct kmem_cache *s, struct page *page,
              end--;
 
      slab_bug(s, "%s overwritten", what);
-     pr_err("INFO: 0x%p-0x%p @offset=%tu. First byte 0x%x instead of 0x%x\n",
+     pr_err("ERR: 0x%p-0x%p @offset=%tu. First byte 0x%x instead of 0x%x\n",
                                      fault, end - 1, fault - addr,
                                      fault[0], value);
      print_trailer(s, page, object);
@@ -3855,7 +3855,7 @@ static void list_slab_objects(struct kmem_cache *s, struct page *page,
      for_each_object(p, s, addr, page->objects) {
 
              if (!test_bit(__obj_to_index(s, addr, p), map)) {
-                     pr_err("INFO: Object 0x%p @offset=%tu\n", p, p - addr);
+                     pr_err("ERR: Object 0x%p @offset=%tu\n", p, p - addr);
                      print_tracking(s, p);
              }
      }
Signature · 2 lines
Thanks
Yafang
#7 lore raw

PATCH 3/3 printk: dump full information of page flags in pGp

In reply to Yafang Shao

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

- Before the patch,

[ 6312.639698] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6315.235783] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(Node 0x0,Zone 0x2,Lastcpupid 0x1fffff,slab|head)
Cc David Hildenbrand <[email protected]>
---
 lib/vsprintf.c | 42 +++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 41 insertions(+), 1 deletion(-)
lib/vsprintf.c +41 -1
index 3b53c73580c5..bd809f4f1b82 100644
@@ -1916,6 +1916,46 @@ char *format_flags(char *buf, char *end, unsigned long flags,
 	return buf;
 }
 
+struct page_flags_layout {
+	int width;
+	int shift;
+	int mask;
+	char *name;
+};
+
+struct page_flags_layout pfl[] = {
+	{SECTIONS_WIDTH, SECTIONS_PGSHIFT, SECTIONS_MASK, "Section "},
+	{NODES_WIDTH, NODES_PGSHIFT, NODES_MASK, "Node "},
+	{ZONES_WIDTH, ZONES_PGSHIFT, ZONES_MASK, "Zone "},
+	{LAST_CPUPID_WIDTH, LAST_CPUPID_PGSHIFT, LAST_CPUPID_MASK, "Lastcpupid "},
+	{KASAN_TAG_WIDTH, KASAN_TAG_PGSHIFT, KASAN_TAG_MASK, "Kasantag "},
+};
+
+static
+char *format_layout(char *buf, char *end, unsigned long flags)
+{
+	int i;
+
+	for (i = 0; i < sizeof(pfl) / sizeof(struct page_flags_layout) && buf < end; i++) {
+		if (pfl[i].width == 0)
+			continue;
+
+		buf = string(buf, end, pfl[i].name, default_str_spec);
+
+		if (buf >= end)
+			break;
+		buf = number(buf, end, (flags >> pfl[i].shift) & pfl[i].mask,
+			     default_flag_spec);
+
+		if (buf >= end)
+			break;
+		*buf = ',';
+		buf++;
+	}
+
+	return buf;
+}
+
 static noinline_for_stack
 char *flags_string(char *buf, char *end, void *flags_ptr,
 		   struct printf_spec spec, const char *fmt)
@@ -1929,7 +1969,7 @@ char *flags_string(char *buf, char *end, void *flags_ptr,
 	switch (fmt[1]) {
 	case 'p':
 		flags = *(unsigned long *)flags_ptr;
-		/* Remove zone id */
+		buf = format_layout(buf, end, flags & ~((1UL << NR_PAGEFLAGS) - 1));
 		flags &= (1UL << NR_PAGEFLAGS) - 1;
 		names = pageflag_names;
 		break;
Signature · 1 line
2.17.1
#8 lore raw

Re: [PATCH 3/3] printk: dump full information of page flags in pGp

Quoted 4 lines · Yafang Shao

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

[]

lib/vsprintf.c

Quoted 1 line
lib/vsprintf.c +0 -0

[]

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1926

Quoted 12 lines
lib/vsprintf.c +8 -0
@@ -1916,6 +1916,46 @@ char *format_flags(char *buf, char *end, unsigned long flags,
  	return buf;
  }
 
+struct page_flags_layout {
+	int width;
+	int shift;
+	int mask;
+	char *name;
+};
+
+struct page_flags_layout pfl[] = {

static const struct page_flags_layout pfl[] = {

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1935

Quoted 9 lines
lib/vsprintf.c +9 -0
+	{SECTIONS_WIDTH, SECTIONS_PGSHIFT, SECTIONS_MASK, "Section "},
+	{NODES_WIDTH, NODES_PGSHIFT, NODES_MASK, "Node "},
+	{ZONES_WIDTH, ZONES_PGSHIFT, ZONES_MASK, "Zone "},
+	{LAST_CPUPID_WIDTH, LAST_CPUPID_PGSHIFT, LAST_CPUPID_MASK, "Lastcpupid "},
+	{KASAN_TAG_WIDTH, KASAN_TAG_PGSHIFT, KASAN_TAG_MASK, "Kasantag "},
+};
+
+static
+char *format_layout(char *buf, char *end, unsigned long flags)

poor name. perhaps format_page_flags

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1939

Quoted 4 lines
lib/vsprintf.c +4 -0
+{
+	int i;
+
+	for (i = 0; i < sizeof(pfl) / sizeof(struct page_flags_layout) && buf < end; i++) {
	for (i = 0; i < ARRAY_SIZE(pfl) && buf < end; i++) {
Quoted 7 lines
lib/vsprintf.c +0 -0
@@ -1929,7 +1969,7 @@ char *flags_string(char *buf, char *end, void *flags_ptr,
  	switch (fmt[1]) {

  case 'p':
  flags = *(unsigned long *)flags_ptr;

lib/vsprintf.c +1 -1
-		/* Remove zone id */
+		buf = format_layout(buf, end, flags & ~((1UL << NR_PAGEFLAGS) - 1));

  flags &= (1UL << NR_PAGEFLAGS) - 1;

Perhaps store the bitshift into a temp and use the temp twice

		foo = BIT(NR_PAGEFLAGS) - 1;

		buf = format_layout(buf, end, flags & ~foo);
		flags &= foo;
#9 lore raw

Re: [PATCH 3/3] printk: dump full information of page flags in pGp

Quoted 13 lines · Joe Perches
Quoted 4 lines · Yafang Shao

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

[]

lib/vsprintf.c

Quoted 1 line
lib/vsprintf.c +0 -0

[]

Quoted 4 lines
lib/vsprintf.c +0 -0
@@ -1916,6 +1916,46 @@ char *format_flags(char *buf, char *end, unsigned long flags,
      return buf;
 }
 
Quoted 10 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1926

Quoted 8 lines
lib/vsprintf.c +8 -0
+struct page_flags_layout {
+     int width;
+     int shift;
+     int mask;
+     char *name;
+};
+
+struct page_flags_layout pfl[] = {

static const struct page_flags_layout pfl[] = {

Sure.

Quoted 13 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1935

Quoted 10 lines
lib/vsprintf.c +9 -0
+     {SECTIONS_WIDTH, SECTIONS_PGSHIFT, SECTIONS_MASK, "Section "},
+     {NODES_WIDTH, NODES_PGSHIFT, NODES_MASK, "Node "},
+     {ZONES_WIDTH, ZONES_PGSHIFT, ZONES_MASK, "Zone "},
+     {LAST_CPUPID_WIDTH, LAST_CPUPID_PGSHIFT, LAST_CPUPID_MASK, "Lastcpupid "},
+     {KASAN_TAG_WIDTH, KASAN_TAG_PGSHIFT, KASAN_TAG_MASK, "Kasantag "},
+};
+
+static
+char *format_layout(char *buf, char *end, unsigned long flags)

poor name. perhaps format_page_flags

Thanks for the suggestion.

Quoted 7 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1939

Quoted 4 lines
lib/vsprintf.c +4 -0
+{
+     int i;
+
+     for (i = 0; i < sizeof(pfl) / sizeof(struct page_flags_layout) && buf < end; i++) {
        for (i = 0; i < ARRAY_SIZE(pfl) && buf < end; i++) {

Sure.

Quoted 11 lines

lib/vsprintf.c char *flags_string(char *buf, char *end, void *flags_ptr, · line 1973

Quoted 8 lines
lib/vsprintf.c +1 -1
@@ -1929,7 +1969,7 @@ char *flags_string(char *buf, char *end, void *flags_ptr,
      switch (fmt[1]) {
      case 'p':
              flags = *(unsigned long *)flags_ptr;
-             /* Remove zone id */
+             buf = format_layout(buf, end, flags & ~((1UL << NR_PAGEFLAGS) - 1));
              flags &= (1UL << NR_PAGEFLAGS) - 1;

Perhaps store the bitshift into a temp and use the temp twice

Quoted 6 lines
                foo = BIT(NR_PAGEFLAGS) - 1;

                buf = format_layout(buf, end, flags & ~foo);
                flags &= foo;

Thanks for the suggestion. I will change them all.

Signature · 2 lines
Thanks
Yafang
#10 lore raw

Re: [PATCH 3/3] printk: dump full information of page flags in pGp

In reply to Yafang Shao

Hi:

Quoted 11 lines · Yafang Shao

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

- Before the patch,

[ 6312.639698] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6315.235783] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(Node 0x0,Zone 0x2,Lastcpupid 0x1fffff,slab|head)

Thanks. This really helps!

Quoted 11 lines
Cc David Hildenbrand <[email protected]>
---
 lib/vsprintf.c | 42 +++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 41 insertions(+), 1 deletion(-)
lib/vsprintf.c +0 -0
index 3b53c73580c5..bd809f4f1b82 100644

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1923

Quoted 8 lines
lib/vsprintf.c +5 -0
@@ -1916,6 +1916,46 @@ char *format_flags(char *buf, char *end, unsigned long flags,
 	return buf;
 }
 
+struct page_flags_layout {
+	int width;
+	int shift;
+	int mask;
+	char *name;

Should we add const for name ?

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1926

Quoted 3 lines
lib/vsprintf.c +3 -0
+};
+
+struct page_flags_layout pfl[] = {

Should we add static const for pfl[] as we won't change its value and use it outside this file ?

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1939

Quoted 13 lines
lib/vsprintf.c +13 -0
+	{SECTIONS_WIDTH, SECTIONS_PGSHIFT, SECTIONS_MASK, "Section "},
+	{NODES_WIDTH, NODES_PGSHIFT, NODES_MASK, "Node "},
+	{ZONES_WIDTH, ZONES_PGSHIFT, ZONES_MASK, "Zone "},
+	{LAST_CPUPID_WIDTH, LAST_CPUPID_PGSHIFT, LAST_CPUPID_MASK, "Lastcpupid "},
+	{KASAN_TAG_WIDTH, KASAN_TAG_PGSHIFT, KASAN_TAG_MASK, "Kasantag "},
+};
+
+static
+char *format_layout(char *buf, char *end, unsigned long flags)
+{
+	int i;
+
+	for (i = 0; i < sizeof(pfl) / sizeof(struct page_flags_layout) && buf < end; i++) {

I think we can use ARRAY_SIZE here.

Quoted 24 lines
lib/vsprintf.c +19 -0
+		if (pfl[i].width == 0)
+			continue;
+
+		buf = string(buf, end, pfl[i].name, default_str_spec);
+
+		if (buf >= end)
+			break;
+		buf = number(buf, end, (flags >> pfl[i].shift) & pfl[i].mask,
+			     default_flag_spec);
+
+		if (buf >= end)
+			break;
+		*buf = ',';
+		buf++;
+	}
+
+	return buf;
+}
+
 static noinline_for_stack
 char *flags_string(char *buf, char *end, void *flags_ptr,
 		   struct printf_spec spec, const char *fmt)

lib/vsprintf.c char *flags_string(char *buf, char *end, void *flags_ptr, · line 1975

Quoted 8 lines
lib/vsprintf.c +1 -1
@@ -1929,7 +1969,7 @@ char *flags_string(char *buf, char *end, void *flags_ptr,
 	switch (fmt[1]) {
 	case 'p':
 		flags = *(unsigned long *)flags_ptr;
-		/* Remove zone id */
+		buf = format_layout(buf, end, flags & ~((1UL << NR_PAGEFLAGS) - 1));
 		flags &= (1UL << NR_PAGEFLAGS) - 1;
 		names = pageflag_names;
 		break;

Many thanks.

#11 lore raw

Re: [PATCH 3/3] printk: dump full information of page flags in pGp

Quoted 31 lines · Miaohe Lin

Hi:

Quoted 11 lines · Yafang Shao

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

- Before the patch,

[ 6312.639698] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6315.235783] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(Node 0x0,Zone 0x2,Lastcpupid 0x1fffff,slab|head)

Thanks. This really helps!

Quoted 14 lines
Cc David Hildenbrand <[email protected]>
---
 lib/vsprintf.c | 42 +++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 41 insertions(+), 1 deletion(-)
lib/vsprintf.c +0 -0
index 3b53c73580c5..bd809f4f1b82 100644
@@ -1916,6 +1916,46 @@ char *format_flags(char *buf, char *end, unsigned long flags,
      return buf;
 }
 
Quoted 8 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1923

Quoted 5 lines
lib/vsprintf.c +5 -0
+struct page_flags_layout {
+     int width;
+     int shift;
+     int mask;
+     char *name;

Should we add const for name ?

Good suggestion.

Quoted 6 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1926

Quoted 3 lines
lib/vsprintf.c +3 -0
+};
+
+struct page_flags_layout pfl[] = {

Should we add static const for pfl[] as we won't change its value and use it outside this file ?

Sure.

Quoted 7 lines
Quoted 7 lines
lib/vsprintf.c +7 -0
+     {SECTIONS_WIDTH, SECTIONS_PGSHIFT, SECTIONS_MASK, "Section "},
+     {NODES_WIDTH, NODES_PGSHIFT, NODES_MASK, "Node "},
+     {ZONES_WIDTH, ZONES_PGSHIFT, ZONES_MASK, "Zone "},
+     {LAST_CPUPID_WIDTH, LAST_CPUPID_PGSHIFT, LAST_CPUPID_MASK, "Lastcpupid "},
+     {KASAN_TAG_WIDTH, KASAN_TAG_PGSHIFT, KASAN_TAG_MASK, "Kasantag "},
+};
+
Quoted 9 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1939

Quoted 6 lines
lib/vsprintf.c +6 -0
+static
+char *format_layout(char *buf, char *end, unsigned long flags)
+{
+     int i;
+
+     for (i = 0; i < sizeof(pfl) / sizeof(struct page_flags_layout) && buf < end; i++) {

I think we can use ARRAY_SIZE here.

Sure.

Quoted 33 lines
Quoted 24 lines
lib/vsprintf.c +19 -0
+             if (pfl[i].width == 0)
+                     continue;
+
+             buf = string(buf, end, pfl[i].name, default_str_spec);
+
+             if (buf >= end)
+                     break;
+             buf = number(buf, end, (flags >> pfl[i].shift) & pfl[i].mask,
+                          default_flag_spec);
+
+             if (buf >= end)
+                     break;
+             *buf = ',';
+             buf++;
+     }
+
+     return buf;
+}
+
 static noinline_for_stack
 char *flags_string(char *buf, char *end, void *flags_ptr,
                 struct printf_spec spec, const char *fmt)

lib/vsprintf.c char *flags_string(char *buf, char *end, void *flags_ptr, · line 1975

Quoted 8 lines
lib/vsprintf.c +1 -1
@@ -1929,7 +1969,7 @@ char *flags_string(char *buf, char *end, void *flags_ptr,
      switch (fmt[1]) {
      case 'p':
              flags = *(unsigned long *)flags_ptr;
-             /* Remove zone id */
+             buf = format_layout(buf, end, flags & ~((1UL << NR_PAGEFLAGS) - 1));
              flags &= (1UL << NR_PAGEFLAGS) - 1;
              names = pageflag_names;
              break;

Many thanks.

Signature · 2 lines
Thanks
Yafang
#12 lore raw

Re: [PATCH 3/3] printk: dump full information of page flags in pGp

In reply to Yafang Shao
Quoted 10 lines · Yafang Shao

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

- Before the patch,

[ 6312.639698] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6315.235783] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(Node 0x0,Zone 0x2,Lastcpupid 0x1fffff,slab|head)

node, zone could be decimal?
Thanks

Quoted 68 lines
Cc David Hildenbrand <[email protected]>
---
 lib/vsprintf.c | 42 +++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 41 insertions(+), 1 deletion(-)
lib/vsprintf.c +41 -1
index 3b53c73580c5..bd809f4f1b82 100644
@@ -1916,6 +1916,46 @@ char *format_flags(char *buf, char *end, unsigned long flags,
 	return buf;
 }
 
+struct page_flags_layout {
+	int width;
+	int shift;
+	int mask;
+	char *name;
+};
+
+struct page_flags_layout pfl[] = {
+	{SECTIONS_WIDTH, SECTIONS_PGSHIFT, SECTIONS_MASK, "Section "},
+	{NODES_WIDTH, NODES_PGSHIFT, NODES_MASK, "Node "},
+	{ZONES_WIDTH, ZONES_PGSHIFT, ZONES_MASK, "Zone "},
+	{LAST_CPUPID_WIDTH, LAST_CPUPID_PGSHIFT, LAST_CPUPID_MASK, "Lastcpupid "},
+	{KASAN_TAG_WIDTH, KASAN_TAG_PGSHIFT, KASAN_TAG_MASK, "Kasantag "},
+};
+
+static
+char *format_layout(char *buf, char *end, unsigned long flags)
+{
+	int i;
+
+	for (i = 0; i < sizeof(pfl) / sizeof(struct page_flags_layout) && buf < end; i++) {
+		if (pfl[i].width == 0)
+			continue;
+
+		buf = string(buf, end, pfl[i].name, default_str_spec);
+
+		if (buf >= end)
+			break;
+		buf = number(buf, end, (flags >> pfl[i].shift) & pfl[i].mask,
+			     default_flag_spec);
+
+		if (buf >= end)
+			break;
+		*buf = ',';
+		buf++;
+	}
+
+	return buf;
+}
+
 static noinline_for_stack
 char *flags_string(char *buf, char *end, void *flags_ptr,
 		   struct printf_spec spec, const char *fmt)
@@ -1929,7 +1969,7 @@ char *flags_string(char *buf, char *end, void *flags_ptr,
 	switch (fmt[1]) {
 	case 'p':
 		flags = *(unsigned long *)flags_ptr;
-		/* Remove zone id */
+		buf = format_layout(buf, end, flags & ~((1UL << NR_PAGEFLAGS) - 1));
 		flags &= (1UL << NR_PAGEFLAGS) - 1;
 		names = pageflag_names;
 		break;
#13 lore raw

Re: [PATCH 3/3] printk: dump full information of page flags in pGp

Quoted 7 lines · Vlastimil Babka
Quoted 5 lines · Yafang Shao

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

Quoted 7 lines
Quoted 5 lines

- Before the patch,

[ 6312.639698] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6315.235783] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(Node 0x0,Zone 0x2,Lastcpupid 0x1fffff,slab|head)

node, zone could be decimal?

Make sense to me. Will change it.

Quoted 70 lines
Quoted 69 lines
Cc David Hildenbrand <[email protected]>
---
 lib/vsprintf.c | 42 +++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 41 insertions(+), 1 deletion(-)
lib/vsprintf.c +41 -1
index 3b53c73580c5..bd809f4f1b82 100644
@@ -1916,6 +1916,46 @@ char *format_flags(char *buf, char *end, unsigned long flags,
      return buf;
 }
 
+struct page_flags_layout {
+     int width;
+     int shift;
+     int mask;
+     char *name;
+};
+
+struct page_flags_layout pfl[] = {
+     {SECTIONS_WIDTH, SECTIONS_PGSHIFT, SECTIONS_MASK, "Section "},
+     {NODES_WIDTH, NODES_PGSHIFT, NODES_MASK, "Node "},
+     {ZONES_WIDTH, ZONES_PGSHIFT, ZONES_MASK, "Zone "},
+     {LAST_CPUPID_WIDTH, LAST_CPUPID_PGSHIFT, LAST_CPUPID_MASK, "Lastcpupid "},
+     {KASAN_TAG_WIDTH, KASAN_TAG_PGSHIFT, KASAN_TAG_MASK, "Kasantag "},
+};
+
+static
+char *format_layout(char *buf, char *end, unsigned long flags)
+{
+     int i;
+
+     for (i = 0; i < sizeof(pfl) / sizeof(struct page_flags_layout) && buf < end; i++) {
+             if (pfl[i].width == 0)
+                     continue;
+
+             buf = string(buf, end, pfl[i].name, default_str_spec);
+
+             if (buf >= end)
+                     break;
+             buf = number(buf, end, (flags >> pfl[i].shift) & pfl[i].mask,
+                          default_flag_spec);
+
+             if (buf >= end)
+                     break;
+             *buf = ',';
+             buf++;
+     }
+
+     return buf;
+}
+
 static noinline_for_stack
 char *flags_string(char *buf, char *end, void *flags_ptr,
                 struct printf_spec spec, const char *fmt)
@@ -1929,7 +1969,7 @@ char *flags_string(char *buf, char *end, void *flags_ptr,
      switch (fmt[1]) {
      case 'p':
              flags = *(unsigned long *)flags_ptr;
-             /* Remove zone id */
+             buf = format_layout(buf, end, flags & ~((1UL << NR_PAGEFLAGS) - 1));
              flags &= (1UL << NR_PAGEFLAGS) - 1;
              names = pageflag_names;
              break;
Signature · 2 lines
Thanks
Yafang
#14 lore raw

Re: [PATCH 3/3] printk: dump full information of page flags in pGp

In reply to Yafang Shao
Quoted 11 lines · Yafang Shao

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

- Before the patch,

[ 6312.639698] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6315.235783] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(Node 0x0,Zone 0x2,Lastcpupid 0x1fffff,slab|head)

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1939

Quoted 3 lines
lib/vsprintf.c +3 -0
+	int i;
+
+	for (i = 0; i < sizeof(pfl) / sizeof(struct page_flags_layout) && buf < end; i++) {

'buf < end' is redundant.

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1946

Quoted 7 lines
lib/vsprintf.c +6 -0
+		if (pfl[i].width == 0)
+			continue;
+
+		buf = string(buf, end, pfl[i].name, default_str_spec);
 
+		if (buf >= end)
+			break;

Can you rather use usual patter, i.e.

	if (buf < end) {
		...do something...
	}
	buf++; // or whatever increase should be done

Moreover, number() and string() IIRC have the proper checks embedded into them.

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1953

Quoted 7 lines
lib/vsprintf.c +6 -0
+		buf = number(buf, end, (flags >> pfl[i].shift) & pfl[i].mask,
+			     default_flag_spec);
 
+		if (buf >= end)
+			break;
+		*buf = ',';
+		buf++;

Here is a very standard pattern can be used, see code around

		if (buf < end)
			*buf = ',';
		buf++;
Quoted 1 line
(quoted diff) +1 -0
+	}
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#15 lore raw

Re: [PATCH 3/3] printk: dump full information of page flags in pGp

Quoted 10 lines · Andy Shevchenko
Quoted 8 lines · Yafang Shao

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

- Before the patch,

[ 6312.639698] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(slab|head)
Quoted 9 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1939

Quoted 6 lines

- After the patch,

[ 6315.235783] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(Node 0x0,Zone 0x2,Lastcpupid 0x1fffff,slab|head)
lib/vsprintf.c +3 -0
+     int i;
+
+     for (i = 0; i < sizeof(pfl) / sizeof(struct page_flags_layout) && buf < end; i++) {

'buf < end' is redundant.

Thanks for pointing this out.

Quoted 8 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1946

Quoted 7 lines
lib/vsprintf.c +6 -0
+             if (pfl[i].width == 0)
+                     continue;
+
+             buf = string(buf, end, pfl[i].name, default_str_spec);
 
+             if (buf >= end)
+                     break;
Quoted 9 lines

Can you rather use usual patter, i.e.

        if (buf < end) {
                ...do something...
        }
        buf++; // or whatever increase should be done

Moreover, number() and string() IIRC have the proper checks embedded into them.

I will take a look at the detail in these two functions.

Quoted 8 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1953

Quoted 7 lines
lib/vsprintf.c +6 -0
+             buf = number(buf, end, (flags >> pfl[i].shift) & pfl[i].mask,
+                          default_flag_spec);
 
+             if (buf >= end)
+                     break;
+             *buf = ',';
+             buf++;
Quoted 8 lines

Here is a very standard pattern can be used, see code around

                if (buf < end)
                        *buf = ',';
                buf++;
Quoted 1 line

+ }

Thanks for the explanation.
I will change it as you suggested.

Signature · 2 lines
Thanks
Yafang
#16 lore raw

Re: [PATCH 3/3] printk: dump full information of page flags in pGp

Quoted 2 lines · Yafang Shao

On Thu, Jan 28, 2021 at 8:11 PM Andy Shevchenko
<[email protected]> wrote:

...

Quoted 2 lines

Thanks for the explanation.
I will change it as you suggested.

You are welcome!

Just note, that kasprintf() should work for this as well as for the rest printf() cases. That's why it's very important to return proper buffer pointer.

Also read man snprintf(3) RETURN VALUE section to understand it.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#17 lore raw

Re: [PATCH 3/3] printk: dump full information of page flags in pGp

Quoted 10 lines · Andy Shevchenko
Quoted 2 lines · Yafang Shao

On Thu, Jan 28, 2021 at 8:11 PM Andy Shevchenko
<[email protected]> wrote:

...

Quoted 2 lines

Thanks for the explanation.
I will change it as you suggested.

Quoted 7 lines

You are welcome!

Just note, that kasprintf() should work for this as well as for the rest printf() cases. That's why it's very important to return proper buffer pointer.

Also read man snprintf(3) RETURN VALUE section to understand it.

Thanks for the information. I will take a look at it.

Signature · 2 lines
Thanks
Yafang
#18 lore raw
In reply to Yafang Shao
Quoted 5 lines · Yafang Shao

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

Quoted 8 lines

This patchset also includes some code cleanup in mm/slub.c.

Below is the example of the output in mm/slub.c.
- Before the patchset

[ 6155.716018] INFO: Slab 0x000000004027dd4f objects=33 used=3 fp=0x000000008cd1579c flags=0x17ffffc0010200

- After the patchset

[ 6315.235783] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(Node 0x0,Zone 0x2,Lastcpupid 0x1fffff,slab|head)

Please, add a corresponding test cases to test_printf.c. W/o test cases NAK.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#19 lore raw
Quoted 13 lines · Andy Shevchenko
Quoted 11 lines · Yafang Shao

Currently the pGp only shows the names of page flags, rather than the full information including section, node, zone, last cpupid and kasan tag. While it is not easy to parse these information manually because there're so many flavors. Let's interpret them in pGp as well.

This patchset also includes some code cleanup in mm/slub.c.

Below is the example of the output in mm/slub.c.
- Before the patchset

[ 6155.716018] INFO: Slab 0x000000004027dd4f objects=33 used=3 fp=0x000000008cd1579c flags=0x17ffffc0010200
Quoted 6 lines
Quoted 2 lines

- After the patchset

[ 6315.235783] ERR: Slab 0x000000006d1133b9 objects=33 used=3 fp=0x000000006d0779d1 flags=0x17ffffc0010200(Node 0x0,Zone 0x2,Lastcpupid 0x1fffff,slab|head)

Please, add a corresponding test cases to test_printf.c. W/o test cases NAK.

Sure. Will add the test cases in the next version.

Signature · 2 lines
Thanks
Yafang