Skip to contentSkip to conversation

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

PATCH Superseded Yafang Shao 5 participants 20 messages Feb 1, 2021 – Feb 2, 2021

Subsystems (5)

v2 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

v2 cover mm, vsprintf: 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

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

The documentation and test cases of pGp are also updated.

Below is the result of the test_printf,

[ 5091.307308] test_printf: loaded.
[ 5091.308285] test_printf: all 388 tests passed
[ 5091.309105] test_printf: unloaded.

v2:
- various coding improvement, per Joe, Miaohe, Vlastimil and Andy
- remove the prefix completely in patch #2, per Vlastimil
- Update the test cases, per Andy

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

 Documentation/core-api/printk-formats.rst |  2 +-
 lib/test_printf.c                         | 65 ++++++++++++++++++-----
 lib/vsprintf.c                            | 58 +++++++++++++++++++-
 mm/slub.c                                 | 13 ++---
 4 files changed, 116 insertions(+), 22 deletions(-)
Signature · 1 line
2.17.1
#1 lore raw

PATCH v2 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 34dcc09e2ec9..87ff086e68a4 100644
@@ -638,8 +638,9 @@ 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
#2 lore raw

PATCH v2 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 remove the prefix completely.
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

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

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

---
 mm/slub.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
mm/slub.c +5 -5
index 87ff086e68a4..2514c37ab4e4 100644
@@ -612,7 +612,7 @@ 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("%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
 	{
@@ -638,7 +638,7 @@ 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("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);
 
@@ -695,7 +695,7 @@ 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("Object 0x%p @offset=%tu fp=0x%p\n\n",
 	       p, p - addr, get_freepointer(s, p));
 
 	if (s->flags & SLAB_RED_ZONE)
@@ -788,7 +788,7 @@ 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("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);
@@ -3854,7 +3854,7 @@ 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("Object 0x%p @offset=%tu\n", p, p - addr);
 			print_tracking(s, p);
 		}
 	}
Signature · 1 line
2.17.1
#3 lore raw

PATCH v2 3/3 vsprintf: 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,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

The Documentation and test cases are also updated.

Cc David Hildenbrand <[email protected]>
Cc Joe Perches <[email protected]>
Cc Miaohe Lin <[email protected]>
Cc Vlastimil Babka <[email protected]>
Cc Andy Shevchenko <[email protected]>
---
 Documentation/core-api/printk-formats.rst |  2 +-
 lib/test_printf.c                         | 65 ++++++++++++++++++-----
 lib/vsprintf.c                            | 58 +++++++++++++++++++-
 3 files changed, 109 insertions(+), 16 deletions(-)
Documentation/core-api/printk-formats.rst +1 -1
index 6d26c5c6ac48..1374cdd04f0f 100644
@@ -538,7 +538,7 @@ Flags bitfields such as page flags, gfp_flags
 
 ::
 
-	%pGp	referenced|uptodate|lru|active|private
+	%pGp	Node 0,Zone 2,referenced|uptodate|lru|active|private
 	%pGg	GFP_USER|GFP_DMA32|GFP_NOWARN
 	%pGv	read|exec|mayread|maywrite|mayexec|denywrite
 
lib/test_printf.c +51 -14

84 lines not on this page · raw message

lib/vsprintf.c +57 -1

70 lines not on this page · raw message

Signature · 1 line
2.17.1
#4 lore raw

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

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,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

While debugfs is not an ABI, this format is exported in debugfs to userspace via mm/page_owner.c read_page_owner/print_page_owner.

Does changing the output format matter to anyone?

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 1944

Quoted 5 lines
lib/vsprintf.c +5 -0
+static
+char *format_page_flags(char *buf, char *end, unsigned long page_flags)
+{
+	unsigned long flags = page_flags & ((1UL << NR_PAGEFLAGS) - 1);
+	int size = ARRAY_SIZE(pfl);

There's no real value in used-once variables.

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1948

Quoted 4 lines
lib/vsprintf.c +4 -0
+	bool separator = false;
+	int i;
+
+	for (i = 0; i < size; i++) {

Use ARRAY_SIZE here instead

	for (i = 0; i < ARRAY_SIZE(pfl); i++) {
Quoted 6 lines
lib/vsprintf.c +6 -0
+		if (pfl[i].width == 0)
+			continue;
+
+		if (separator) {
+			if (buf < end)
+				*buf = ',';

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1964

Quoted 10 lines
lib/vsprintf.c +10 -0
+			buf++;
+		}
+
+
+		buf = string(buf, end, pfl[i].name, *pfl[i].spec);
+
+		buf = number(buf, end, (page_flags >> pfl[i].shift) & pfl[i].mask,
+			     *pfl[i].spec);
+		separator = true;
+	}

Style question:
Might this array be more intelligible with pointers instead of indexes? Something like:

	struct page_flags_layout *p;

	for (p = pfl; p < pfl + ARRAY_SIZE(pfl); p++) {
		if (p->width == 0)
			continue;

		if (p > pfl) {
			if (buf < end)
				*buf = ',';
			buf++;
		}

		buf = string(buf, end, p->name, *p->spec);
		buf = number(buf, end, (page_flags >> p->shift) & p->mask, *p->spec);
	}

lib/vsprintf.c

Quoted 2 lines
lib/vsprintf.c +2 -0
+
+	if (flags) {

Maybe:

	if (page_flags & (BIT(NR_PAGEFLAGS) - 1)) {
Quoted 8 lines
lib/vsprintf.c +8 -0
+		if (buf < end)
+			*buf = ',';
+		buf++;
+	}
+
+	return buf;
+}
+
#5 lore raw

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

Quoted 7 lines · Joe Perches
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 9 lines
Quoted 5 lines

- Before the patch,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

While debugfs is not an ABI, this format is exported in debugfs to userspace via mm/page_owner.c read_page_owner/print_page_owner.

Right, the page_owner will be affected by this change.

Quoted 2 lines

Does changing the output format matter to anyone?

The user tools which parse the page_owner may be affected. If we don't want to affect the userspace tools, I think we can make a little change in page_owner as follows,

    unsigned long masked_flags = page->flags &  (BIT(NR_PAGEFLAGS) - 1);
    snprintf("..., %#lx(%pGp)\n", page->flags, &masked_flags);
Quoted 10 lines

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 1944

Quoted 5 lines
lib/vsprintf.c +5 -0
+static
+char *format_page_flags(char *buf, char *end, unsigned long page_flags)
+{
+     unsigned long flags = page_flags & ((1UL << NR_PAGEFLAGS) - 1);
+     int size = ARRAY_SIZE(pfl);

There's no real value in used-once variables.

Quoted 7 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1948

Quoted 4 lines
lib/vsprintf.c +4 -0
+     bool separator = false;
+     int i;
+
+     for (i = 0; i < size; i++) {

Use ARRAY_SIZE here instead

Sure

Quoted 13 lines
        for (i = 0; i < ARRAY_SIZE(pfl); i++) {
Quoted 11 lines
lib/vsprintf.c +11 -0
+             if (pfl[i].width == 0)
+                     continue;
+
+             if (separator) {
+                     if (buf < end)
+                             *buf = ',';
+                     buf++;
+             }
+
+
+             buf = string(buf, end, pfl[i].name, *pfl[i].spec);
Quoted 8 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1964

Quoted 5 lines
lib/vsprintf.c +5 -0
+
+             buf = number(buf, end, (page_flags >> pfl[i].shift) & pfl[i].mask,
+                          *pfl[i].spec);
+             separator = true;
+     }

Style question:
Might this array be more intelligible with pointers instead of indexes?

Good suggestion!
I will change it in the next version.

Quoted 14 lines

Something like:

        struct page_flags_layout *p;

        for (p = pfl; p < pfl + ARRAY_SIZE(pfl); p++) {
                if (p->width == 0)
                        continue;


                if (p > pfl) {
                        if (buf < end)
                                *buf = ',';
                        buf++;
                }

It doesn't work, because there's a 'continue' above, meaning that the p may be greater than pfl without doing anything.

Quoted 5 lines
                buf = string(buf, end, p->name, *p->spec);
                buf = number(buf, end, (page_flags >> p->shift) & p->mask, *p->spec);
        }
Quoted 7 lines

lib/vsprintf.c

Quoted 2 lines
lib/vsprintf.c +2 -0
+
+     if (flags) {

Maybe:

        if (page_flags & (BIT(NR_PAGEFLAGS) - 1)) {

Sure.

Quoted 10 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1974

Quoted 8 lines
lib/vsprintf.c +8 -0
+             if (buf < end)
+                     *buf = ',';
+             buf++;
+     }
+
+     return buf;
+}
+

--
Thanks
Yafang

#6 lore raw

Re: [PATCH v2 3/3] vsprintf: 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,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

For debugging purposes, it might be helpful to have the actual zone name (and to know if the value is sane). You could obtain it (without other modifications) via

const char zname = "Invalid";

if (zone < MAX_NR_ZONES)

	zname = first_online_pgdat()->node_zones[zone].name;

Similarly, it might also be helpful to indicate if a node is online/offline/invalid/.

const char nstate = "Invalid";

if (node_online(nid))

	nstate = "Online";
else if (node_possible(nid))
	nstate = "Offline";

Printing that in addition to the raw value could be helpful. Just some thoughts.

Signature · 3 lines
Thanks,

David / dhildenb
#7 lore raw

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

Quoted 25 lines · David Hildenbrand
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,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

For debugging purposes, it might be helpful to have the actual zone name (and to know if the value is sane). You could obtain it (without other modifications) via

const char zname = "Invalid";

if (zone < MAX_NR_ZONES)

	zname = first_online_pgdat()->node_zones[zone].name;

Similarly, it might also be helpful to indicate if a node is online/offline/invalid/.

Quoted 6 lines

const char nstate = "Invalid";

if (node_online(nid))

	nstate = "Online";
else if (node_possible(nid))
	nstate = "Offline";

Just remembering that we have to take care of nid limits:

if (nid >= 0 && nid < MAX_NUMNODES) {
	if (node_online(nid))
		nstate = "Online";
	else if (node_possible(nid))
		nstate = "Offline";
}
Signature · 3 lines
Thanks,

David / dhildenb
#8 lore raw

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

In reply to David Hildenbrand
Quoted 27 lines · David Hildenbrand
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,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

For debugging purposes, it might be helpful to have the actual zone name (and to know if the value is sane). You could obtain it (without other modifications) via

const char zname = "Invalid";

if (zone < MAX_NR_ZONES)

	zname = first_online_pgdat()->node_zones[zone].name;

Similarly, it might also be helpful to indicate if a node is online/offline/invalid/.

const char nstate = "Invalid";

Quoted 8 lines

if (node_online(nid))

	nstate = "Online";
else if (node_possible(nid))
	nstate = "Offline";

Printing that in addition to the raw value could be helpful. Just some thoughts.

printf() buffer is not a black hole, esp. when you get it messed with critical messages (Oops). I suggest to reduce a burden as much as possible. If you wish to get this, make it caller-configurable, i.e. adding another letter to the specifier.

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

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

Quoted 35 lines · Andy Shevchenko
Quoted 33 lines · David Hildenbrand
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,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

For debugging purposes, it might be helpful to have the actual zone name (and to know if the value is sane). You could obtain it (without other modifications) via

const char zname = "Invalid";

if (zone < MAX_NR_ZONES)

      zname = first_online_pgdat()->node_zones[zone].name;

Similarly, it might also be helpful to indicate if a node is online/offline/invalid/.

const char nstate = "Invalid";

if (node_online(nid))

      nstate = "Online";
else if (node_possible(nid))
      nstate = "Offline";
Quoted 8 lines
Quoted 2 lines

Printing that in addition to the raw value could be helpful. Just some thoughts.

printf() buffer is not a black hole, esp. when you get it messed with critical messages (Oops). I suggest to reduce a burden as much as possible. If you wish to get this, make it caller-configurable, i.e. adding another letter to the specifier.

I think David's suggestion will help us to identify issues.

You mean that we should use another one like "%pGpd" - means pGp debug
- to implement it ?

Signature · 2 lines
Thanks
Yafang
#10 lore raw

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

Quoted 4 lines · Yafang Shao
Quoted 2 lines · Andy Shevchenko
Quoted 1 line · David Hildenbrand

On 01.02.21 12:56, Yafang Shao wrote:

...

Quoted 9 lines
Quoted 8 lines
Quoted 2 lines

Printing that in addition to the raw value could be helpful. Just some thoughts.

printf() buffer is not a black hole, esp. when you get it messed with critical messages (Oops). I suggest to reduce a burden as much as possible. If you wish to get this, make it caller-configurable, i.e. adding another letter to the specifier.

Quoted 4 lines

I think David's suggestion will help us to identify issues.

You mean that we should use another one like "%pGpd" - means pGp debug
- to implement it ?

Yes.

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

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

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 7 lines

- Before the patch,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

The Documentation and test cases are also updated.

Thanks for an update, my comments below.

...

Documentation/core-api/printk-formats.rst Flags bitfields such as page flags, gfp_flags · line 541

Quoted 2 lines
Documentation/core-api/printk-formats.rst +1 -1
-	%pGp	referenced|uptodate|lru|active|private
+	%pGp	Node 0,Zone 2,referenced|uptodate|lru|active|private

Since of the nature of printf() buffer, I wonder if these should be at the end. I.o.w. the question is is the added material more important to user to see than the existed one?

...

lib/test_printf.c netdev_features(void) · line 581

Quoted 10 lines
lib/test_printf.c +10 -0
+static void __init
+page_flags_test(int section, int node, int zone, int last_cpupid,
+		int kasan_tag, int flags, const char *name, char *cmp_buf)
+{
+	unsigned long page_flags = 0;
+	unsigned long size = 0;
+
+#ifdef SECTION_IN_PAGE_FLAGS
+	page_flags |= (sec & SECTIONS_MASK) << SECTIONS_PGSHIFT;
+	snprintf(cmp_buf, BUF_SIZE, "Section %#x,", sec);

I would keep it in the same form as latter ones, i.e.

	snprintf(cmp_buf + size, BUF_SIZE - size, "Section %#x,", sec);

In this case it will be easier if at some point we might need to reshuffle.

Quoted 16 lines
lib/test_printf.c +16 -0
+	size = strlen(cmp_buf);
+#endif
+
+	page_flags |= (node & NODES_MASK) << NODES_PGSHIFT;
+	snprintf(cmp_buf + size, BUF_SIZE - size, "Node %d", node);
+	size = strlen(cmp_buf);
+
+	page_flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
+	snprintf(cmp_buf + size, BUF_SIZE - size, ",Zone %d", zone);
+	size = strlen(cmp_buf);
+
+#ifndef LAST_CPUPID_NOT_IN_PAGE_FLAGS
+	page_flags |= (last_cpupid & LAST_CPUPID_MASK) << LAST_CPUPID_PGSHIFT;
+	snprintf(cmp_buf + size, BUF_SIZE - size, ",Lastcpupid %#x", last_cpupid);
+	size = strlen(cmp_buf);
+#endif

lib/test_printf.c netdev_features(void) · line 607

Quoted 10 lines
lib/test_printf.c +10 -0
+
+#if defined(CONFIG_KASAN_SW_TAGS) || defined(CONFIG_KASAN_HW_TAGS)
+	page_flags |= (tag & KASAN_TAG_MASK) << KASAN_TAG_PGSHIFT;
+	snprintf(cmp_buf + size, BUF_SIZE - size, ",Kasantag %#x", tag);
+	size = strlen(cmp_buf);
+#endif
+
+	test(cmp_buf, "%pGp", &page_flags);
+
+	if (flags) {

If below will be always for flags set, I would rewrite this condition as

	if (!flags)
		return;

but it's up to you.

Quoted 5 lines
lib/test_printf.c +5 -0
+		page_flags |= flags;
+		snprintf(cmp_buf + size, BUF_SIZE - size, ",%s", name);
+		test(cmp_buf, "%pGp", &page_flags);
+	}
+}

...

lib/test_printf.c flags(void) · line 587

Quoted 4 lines
lib/test_printf.c +0 -3
-	flags = 0;
 
-	flags |= 1UL << PG_uptodate | 1UL << PG_dirty | 1UL << PG_lru
-		| 1UL << PG_active | 1UL << PG_swapbacked;

I would leave this untouched and reuse below...

lib/test_printf.c flags(void) · line 623

Quoted 3 lines
lib/test_printf.c +3 -0
+	cmp_buffer = kmalloc(BUF_SIZE, GFP_KERNEL);
+	if (!cmp_buffer)
+		return;

...as

lib/test_printf.c flags(void) · line 625

Quoted 1 line
lib/test_printf.c +1 -0
+	page_flags_test(0, 0, 0, 0, 0, 0, NULL, cmp_buffer);
	flags = 0;
	page_flags_test(0, 0, 0, 0, 0, flags, NULL, cmp_buffer);

lib/test_printf.c flags(void) · line 630

Quoted 5 lines
lib/test_printf.c +5 -0
+	page_flags_test(1, 1, 1, 0x1ffff, 1,
+			(1UL << PG_uptodate | 1UL << PG_dirty | 1UL << PG_lru
+			 | 1UL << PG_active | 1UL << PG_swapbacked),
+			"uptodate|dirty|lru|active|swapbacked",
+			cmp_buffer);
	flags |= 1UL << PG_uptodate | 1UL << PG_dirty | 1UL << PG_lru
		| 1UL << PG_active | 1UL << PG_swapbacked;
	page_flags_test(1, 1, 1, 0x1ffff, 1, flags,
			"uptodate|dirty|lru|active|swapbacked",
			cmp_buffer);

...

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1938

Quoted 12 lines
lib/vsprintf.c +12 -0
+static const struct page_flags_layout pfl[] = {
+	{SECTIONS_WIDTH, SECTIONS_PGSHIFT, SECTIONS_MASK,
+	 &default_dec_spec, "Section "},
+	{NODES_WIDTH, NODES_PGSHIFT, NODES_MASK,
+	 &default_dec_spec, "Node "},
+	{ZONES_WIDTH, ZONES_PGSHIFT, ZONES_MASK,
+	 &default_dec_spec, "Zone "},
+	{LAST_CPUPID_WIDTH, LAST_CPUPID_PGSHIFT, LAST_CPUPID_MASK,
+	 &default_flag_spec, "Lastcpupid "},
+	{KASAN_TAG_WIDTH, KASAN_TAG_PGSHIFT, KASAN_TAG_MASK,
+	 &default_flag_spec, "Kasantag "},
+};

Please add trailing space only once where it's needed (below in the code).

...

Quoted 10 lines
lib/vsprintf.c +10 -0
+static
+char *format_page_flags(char *buf, char *end, unsigned long page_flags)
+{
+	unsigned long flags = page_flags & ((1UL << NR_PAGEFLAGS) - 1);
+	int size = ARRAY_SIZE(pfl);
+	bool separator = false;
+	int i;
+
+	for (i = 0; i < size; i++) {
+		if (pfl[i].width == 0)

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1958

Quoted 10 lines
lib/vsprintf.c +9 -0
+			continue;
+
+		if (separator) {
+			if (buf < end)
+				*buf = ',';
+			buf++;
+		}
 
+
+

One blank line is enough.

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1970

Quoted 12 lines
lib/vsprintf.c +12 -0
+		buf = string(buf, end, pfl[i].name, *pfl[i].spec);
+
+		buf = number(buf, end, (page_flags >> pfl[i].shift) & pfl[i].mask,
+			     *pfl[i].spec);
+		separator = true;
+	}
+
+	if (flags) {
+		if (buf < end)
+			*buf = ',';
+		buf++;
+	}

I think you may optimize above to avoid using the separator variable.

	DECLARE_BITMAP(mask, ARRAY_SIZE(pfl));
	unsigned long flags;
	unsigned long last;

	for (i = 0; i < ARRAY_SIZE(pfl); i++)
		__assign_bit(mask, pfl[i].width);

	last = find_last_bit(mask, ARRAY_SIZE(pfl));

	for_each_set_bit(i, mask, ARRAY_SIZE(pfl)) {
		flags = (page_flags >> pfl[i].shift) & pfl[i].mask;

		/* Format: Flag Name + ' ' (space) + Number + ',' (separator) */
		buf = string(buf, end, pfl[i].name, *pfl[i].spec);

		if (buf < end)
			*buf = ' ';
		buf++;

		buf = number(buf, end, flags, *pfl[i].spec);

		/* No separator for the last entry */
		if ((page_flags & (BIT(NR_PAGEFLAGS) - 1)) || (i != last)) {
			if (buf < end)
				*buf = ',';
			buf++;
		}
	}
Quoted 2 lines
lib/vsprintf.c +2 -0
+	return buf;
+}
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#12 lore raw

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

Quoted 17 lines · Andy Shevchenko
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

- Before the patch,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

The Documentation and test cases are also updated.

Thanks for an update, my comments below.

Quoted 9 lines

...

Documentation/core-api/printk-formats.rst Flags bitfields such as page flags, gfp_flags · line 541

Quoted 2 lines
Documentation/core-api/printk-formats.rst +1 -1
-     %pGp    referenced|uptodate|lru|active|private
+     %pGp    Node 0,Zone 2,referenced|uptodate|lru|active|private

Since of the nature of printf() buffer, I wonder if these should be at the end. I.o.w. the question is is the added material more important to user to see than the existed one?

The existing one should be more important than the added one. But the order of output will not match with the value for page->flags. E.g.

    flags=0x17ffffc0010200(slab|head,Node 0,Zone 2,Lastcpupid 0x1fffff)

It may be strange to compare the value with the string.

Quoted 15 lines

...

lib/test_printf.c netdev_features(void) · line 581

Quoted 10 lines
lib/test_printf.c +10 -0
+static void __init
+page_flags_test(int section, int node, int zone, int last_cpupid,
+             int kasan_tag, int flags, const char *name, char *cmp_buf)
+{
+     unsigned long page_flags = 0;
+     unsigned long size = 0;
+
+#ifdef SECTION_IN_PAGE_FLAGS
+     page_flags |= (sec & SECTIONS_MASK) << SECTIONS_PGSHIFT;
+     snprintf(cmp_buf, BUF_SIZE, "Section %#x,", sec);

I would keep it in the same form as latter ones, i.e.

Quoted 4 lines
        snprintf(cmp_buf + size, BUF_SIZE - size, "Section %#x,", sec);

In this case it will be easier if at some point we might need to reshuffle.

Good suggestion.

Quoted 66 lines
Quoted 16 lines
lib/test_printf.c +16 -0
+     size = strlen(cmp_buf);
+#endif
+
+     page_flags |= (node & NODES_MASK) << NODES_PGSHIFT;
+     snprintf(cmp_buf + size, BUF_SIZE - size, "Node %d", node);
+     size = strlen(cmp_buf);
+
+     page_flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
+     snprintf(cmp_buf + size, BUF_SIZE - size, ",Zone %d", zone);
+     size = strlen(cmp_buf);
+
+#ifndef LAST_CPUPID_NOT_IN_PAGE_FLAGS
+     page_flags |= (last_cpupid & LAST_CPUPID_MASK) << LAST_CPUPID_PGSHIFT;
+     snprintf(cmp_buf + size, BUF_SIZE - size, ",Lastcpupid %#x", last_cpupid);
+     size = strlen(cmp_buf);
+#endif

lib/test_printf.c netdev_features(void) · line 607

Quoted 10 lines
lib/test_printf.c +10 -0
+
+#if defined(CONFIG_KASAN_SW_TAGS) || defined(CONFIG_KASAN_HW_TAGS)
+     page_flags |= (tag & KASAN_TAG_MASK) << KASAN_TAG_PGSHIFT;
+     snprintf(cmp_buf + size, BUF_SIZE - size, ",Kasantag %#x", tag);
+     size = strlen(cmp_buf);
+#endif
+
+     test(cmp_buf, "%pGp", &page_flags);
+
+     if (flags) {

If below will be always for flags set, I would rewrite this condition as

        if (!flags)
                return;

but it's up to you.

Quoted 5 lines
lib/test_printf.c +5 -0
+             page_flags |= flags;
+             snprintf(cmp_buf + size, BUF_SIZE - size, ",%s", name);
+             test(cmp_buf, "%pGp", &page_flags);
+     }
+}

...

lib/test_printf.c flags(void) · line 587

Quoted 4 lines
lib/test_printf.c +0 -3
-     flags = 0;
 
-     flags |= 1UL << PG_uptodate | 1UL << PG_dirty | 1UL << PG_lru
-             | 1UL << PG_active | 1UL << PG_swapbacked;

I would leave this untouched and reuse below...

lib/test_printf.c flags(void) · line 623

Quoted 3 lines
lib/test_printf.c +3 -0
+     cmp_buffer = kmalloc(BUF_SIZE, GFP_KERNEL);
+     if (!cmp_buffer)
+             return;

...as

Quoted 1 line

+ page_flags_test(0, 0, 0, 0, 0, 0, NULL, cmp_buffer);

        flags = 0;
        page_flags_test(0, 0, 0, 0, 0, flags, NULL, cmp_buffer);

lib/test_printf.c flags(void) · line 630

Quoted 5 lines
lib/test_printf.c +5 -0
+     page_flags_test(1, 1, 1, 0x1ffff, 1,
+                     (1UL << PG_uptodate | 1UL << PG_dirty | 1UL << PG_lru
+                      | 1UL << PG_active | 1UL << PG_swapbacked),
+                     "uptodate|dirty|lru|active|swapbacked",
+                     cmp_buffer);
Quoted 6 lines
        flags |= 1UL << PG_uptodate | 1UL << PG_dirty | 1UL << PG_lru
                | 1UL << PG_active | 1UL << PG_swapbacked;
        page_flags_test(1, 1, 1, 0x1ffff, 1, flags,
                        "uptodate|dirty|lru|active|swapbacked",
                        cmp_buffer);

Sure, I will change them.

Quoted 37 lines

...

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1938

Quoted 12 lines
lib/vsprintf.c +12 -0
+static const struct page_flags_layout pfl[] = {
+     {SECTIONS_WIDTH, SECTIONS_PGSHIFT, SECTIONS_MASK,
+      &default_dec_spec, "Section "},
+     {NODES_WIDTH, NODES_PGSHIFT, NODES_MASK,
+      &default_dec_spec, "Node "},
+     {ZONES_WIDTH, ZONES_PGSHIFT, ZONES_MASK,
+      &default_dec_spec, "Zone "},
+     {LAST_CPUPID_WIDTH, LAST_CPUPID_PGSHIFT, LAST_CPUPID_MASK,
+      &default_flag_spec, "Lastcpupid "},
+     {KASAN_TAG_WIDTH, KASAN_TAG_PGSHIFT, KASAN_TAG_MASK,
+      &default_flag_spec, "Kasantag "},
+};

Please add trailing space only once where it's needed (below in the code).

...

Quoted 17 lines
lib/vsprintf.c +17 -0
+static
+char *format_page_flags(char *buf, char *end, unsigned long page_flags)
+{
+     unsigned long flags = page_flags & ((1UL << NR_PAGEFLAGS) - 1);
+     int size = ARRAY_SIZE(pfl);
+     bool separator = false;
+     int i;
+
+     for (i = 0; i < size; i++) {
+             if (pfl[i].width == 0)
+                     continue;
+
+             if (separator) {
+                     if (buf < end)
+                             *buf = ',';
+                     buf++;
+             }
Quoted 5 lines

Review

Quoted 2 lines
(quoted diff) +2 -0
+
+

One blank line is enough.

Thanks for pointing it out.

BTW, it seems we need to improve the check_patch to catch this kind of issue automatically.

Quoted 36 lines

lib/vsprintf.c char *format_flags(char *buf, char *end, unsigned long flags, · line 1970

Quoted 12 lines
lib/vsprintf.c +12 -0
+             buf = string(buf, end, pfl[i].name, *pfl[i].spec);
+
+             buf = number(buf, end, (page_flags >> pfl[i].shift) & pfl[i].mask,
+                          *pfl[i].spec);
+             separator = true;
+     }
+
+     if (flags) {
+             if (buf < end)
+                     *buf = ',';
+             buf++;
+     }

I think you may optimize above to avoid using the separator variable.

        DECLARE_BITMAP(mask, ARRAY_SIZE(pfl));
        unsigned long flags;
        unsigned long last;

        for (i = 0; i < ARRAY_SIZE(pfl); i++)
                __assign_bit(mask, pfl[i].width);

        last = find_last_bit(mask, ARRAY_SIZE(pfl));

        for_each_set_bit(i, mask, ARRAY_SIZE(pfl)) {
                flags = (page_flags >> pfl[i].shift) & pfl[i].mask;

                /* Format: Flag Name + ' ' (space) + Number + ',' (separator) */
                buf = string(buf, end, pfl[i].name, *pfl[i].spec);

                if (buf < end)
                        *buf = ' ';
                buf++;

                buf = number(buf, end, flags, *pfl[i].spec);
Quoted 8 lines
                /* No separator for the last entry */
                if ((page_flags & (BIT(NR_PAGEFLAGS) - 1)) || (i != last)) {
                        if (buf < end)
                                *buf = ',';
                        buf++;
                }
        }

Good suggestion. I will think about it.

Signature · 2 lines
Thanks
Yafang
#13 lore raw

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

Quoted 3 lines · Yafang Shao
Quoted 1 line · Andy Shevchenko

On Mon, Feb 01, 2021 at 07:56:10PM +0800, Yafang Shao wrote:

...

Quoted 20 lines
Quoted 10 lines
Quoted 7 lines

- Before the patch,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

The Documentation and test cases are also updated.

Thanks for an update, my comments below.

Quoted 9 lines

...

Documentation/core-api/printk-formats.rst Flags bitfields such as page flags, gfp_flags · line 541

Quoted 2 lines
Documentation/core-api/printk-formats.rst +1 -1
-     %pGp    referenced|uptodate|lru|active|private
+     %pGp    Node 0,Zone 2,referenced|uptodate|lru|active|private

Since of the nature of printf() buffer, I wonder if these should be at the end. I.o.w. the question is is the added material more important to user to see than the existed one?

Quoted 5 lines

The existing one should be more important than the added one. But the order of output will not match with the value for page->flags. E.g.

    flags=0x17ffffc0010200(slab|head,Node 0,Zone 2,Lastcpupid 0x1fffff)

It may be strange to compare the value with the string.

More I'm looking at it, more I'm thinking it should have different specifiers for each group of desired flags to be printed.

So, you leave %pGp as is and then add another letter to add more details, so user will choose what and in which order they want.

For example, let's assume %pGp == %pGpf and P is a new specifier for what you are initially adding here:

%pGpfP => referenced|uptodate|lru|active|private,Node 0,Zone 2 %pGpPf => Node 0,Zone 2,referenced|uptodate|lru|active|private

and so on.

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

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

Quoted 4 lines · Andy Shevchenko
Quoted 3 lines · Yafang Shao
Quoted 1 line · Andy Shevchenko

On Mon, Feb 01, 2021 at 07:56:10PM +0800, Yafang Shao wrote:

...

Quoted 12 lines
Quoted 5 lines

The existing one should be more important than the added one. But the order of output will not match with the value for page->flags. E.g.

    flags=0x17ffffc0010200(slab|head,Node 0,Zone 2,Lastcpupid 0x1fffff)

It may be strange to compare the value with the string.

More I'm looking at it, more I'm thinking it should have different specifiers for each group of desired flags to be printed.

So, you leave %pGp as is and then add another letter to add more details, so user will choose what and in which order they want.

Quoted 7 lines

For example, let's assume %pGp == %pGpf and P is a new specifier for what you are initially adding here:

%pGpfP => referenced|uptodate|lru|active|private,Node 0,Zone 2 %pGpPf => Node 0,Zone 2,referenced|uptodate|lru|active|private

and so on.

And I agree with Matthew about format, but it doesn't oppose my suggestion above.

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

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

In reply to Andy Shevchenko
Quoted 40 lines · Andy Shevchenko
Quoted 3 lines · Yafang Shao
Quoted 1 line · Andy Shevchenko

On Mon, Feb 01, 2021 at 07:56:10PM +0800, Yafang Shao wrote:

...

Quoted 20 lines
Quoted 10 lines
Quoted 7 lines

- Before the patch,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

The Documentation and test cases are also updated.

Thanks for an update, my comments below.

Quoted 9 lines

...

Documentation/core-api/printk-formats.rst Flags bitfields such as page flags, gfp_flags · line 541

Quoted 2 lines
Documentation/core-api/printk-formats.rst +1 -1
-     %pGp    referenced|uptodate|lru|active|private
+     %pGp    Node 0,Zone 2,referenced|uptodate|lru|active|private

Since of the nature of printf() buffer, I wonder if these should be at the end. I.o.w. the question is is the added material more important to user to see than the existed one?

Quoted 5 lines

The existing one should be more important than the added one. But the order of output will not match with the value for page->flags. E.g.

    flags=0x17ffffc0010200(slab|head,Node 0,Zone 2,Lastcpupid 0x1fffff)

It may be strange to compare the value with the string.

More I'm looking at it, more I'm thinking it should have different specifiers for each group of desired flags to be printed.

So, you leave %pGp as is and then add another letter to add more details, so user will choose what and in which order they want.

Quoted 7 lines

For example, let's assume %pGp == %pGpf and P is a new specifier for what you are initially adding here:

%pGpfP => referenced|uptodate|lru|active|private,Node 0,Zone 2 %pGpPf => Node 0,Zone 2,referenced|uptodate|lru|active|private

and so on.

Thanks for your suggestion. I will think about it.

Signature · 2 lines
Thanks
Yafang
#16 lore raw

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

In reply to Yafang Shao
Quoted 5 lines · Yafang Shao

- Before the patch,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)

I would suggest it will be easier to parse as:

flags=0x17ffffc0010200(slab|head|node=0|zone=2|lastcpupid=0x1fffff)

That should alleviate the concerns about debugfs format change -- we've never guaranteed that flag names won't change, and they now look enough like flags that parsers shouldn't fall over them.

#17 lore raw

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

Quoted 8 lines · Matthew Wilcox
Quoted 5 lines · Yafang Shao

- Before the patch,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)
Quoted 7 lines

I would suggest it will be easier to parse as:

flags=0x17ffffc0010200(slab|head|node=0|zone=2|lastcpupid=0x1fffff)

That should alleviate the concerns about debugfs format change -- we've never guaranteed that flag names won't change, and they now look enough like flags that parsers shouldn't fall over them.

Good suggestion!
I will do it as you suggested in the next version.

Signature · 2 lines
Thanks
Yafang
#18 lore raw

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

In reply to Matthew Wilcox
Quoted 7 lines · Matthew Wilcox
Quoted 5 lines · Yafang Shao

- Before the patch,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)
Quoted 7 lines

I would suggest it will be easier to parse as:

flags=0x17ffffc0010200(slab|head|node=0|zone=2|lastcpupid=0x1fffff)

That should alleviate the concerns about debugfs format change -- we've never guaranteed that flag names won't change, and they now look enough like flags that parsers shouldn't fall over them.

Seems sensible and would make the generating code simpler too.

But is it worth the vsprintf code expansion for the 5 current uses?

mm/debug.c:     pr_warn("%sflags: %#lx(%pGp)%s\n", type, head->flags, &head->flags,
mm/memory-failure.c:                    pr_info("soft offline: %#lx: %s migration failed %d, type %lx (%pGp)\n",
mm/memory-failure.c:            pr_info("soft offline: %#lx: %s isolation failed, page count %d, type %lx (%pGp)\n",
mm/memory-failure.c:            pr_info("%s: %#lx: unknown page type: %lx (%pGp)\n",
mm/page_owner.c:                        "PFN %lu type %s Block %lu type %s Flags %#lx(%pGp)\n",

Wouldn't it be more sensible just to put this code in a new function call in mm?

#19 lore raw

Re: [PATCH v2 3/3] vsprintf: dump full information of page flags in pGp

Quoted 20 lines · Joe Perches
Quoted 7 lines · Matthew Wilcox
Quoted 5 lines · Yafang Shao

- Before the patch,

[ 6343.396602] Slab 0x000000004382e02b objects=33 used=3 fp=0x000000009ae06ffc flags=0x17ffffc0010200(slab|head)

- After the patch,

[ 6871.296131] Slab 0x00000000c0e19a37 objects=33 used=3 fp=0x00000000c4902159 flags=0x17ffffc0010200(Node 0,Zone 2,Lastcpupid 0x1fffff,slab|head)
Quoted 7 lines

I would suggest it will be easier to parse as:

flags=0x17ffffc0010200(slab|head|node=0|zone=2|lastcpupid=0x1fffff)

That should alleviate the concerns about debugfs format change -- we've never guaranteed that flag names won't change, and they now look enough like flags that parsers shouldn't fall over them.

Seems sensible and would make the generating code simpler too.

But is it worth the vsprintf code expansion for the 5 current uses?

Quoted 8 lines
mm/debug.c:     pr_warn("%sflags: %#lx(%pGp)%s\n", type, head->flags, &head->flags,
mm/memory-failure.c:                    pr_info("soft offline: %#lx: %s migration failed %d, type %lx (%pGp)\n",
mm/memory-failure.c:            pr_info("soft offline: %#lx: %s isolation failed, page count %d, type %lx (%pGp)\n",
mm/memory-failure.c:            pr_info("%s: %#lx: unknown page type: %lx (%pGp)\n",
mm/page_owner.c:                        "PFN %lu type %s Block %lu type %s Flags %#lx(%pGp)\n",

Wouldn't it be more sensible just to put this code in a new function call in mm?

Does it matter whether the code lives in vsprintf.c or mm/debug.c? It's built into the kernel core either way. I'm not a huge fan of the current way %pFoo is handled, but unless/until it's drastically revised, I don't think this proposed patch makes anything worse.