Skip to contentSkip to conversation

maple_tree: remove maple_big_node.parent

PATCH Superseded Wei Yang 2 participants 4 messages Sep 7, 2024 – Sep 8, 2024

Subsystems (3)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Wei Yang

author
#0 lore raw

PATCH 1/2 maple_tree: remove maple_big_node.parent

The member parent of maple_big_node is never used.

Let's remove it.

CC Liam R. Howlett <[email protected]>
---
 lib/maple_tree.c | 1 -
 1 file changed, 1 deletion(-)
lib/maple_tree.c +0 -1
index 994dd4c8c051..d8f10773e451 100644
@@ -120,7 +120,6 @@ static const unsigned char mt_min_slots[] = {
 #define MAPLE_BIG_NODE_GAPS	(MAPLE_ARANGE64_SLOTS * 2 + 1)
 
 struct maple_big_node {
-	struct maple_pnode *parent;
 	unsigned long pivot[MAPLE_BIG_NODE_SLOTS - 1];
 	union {
 		struct maple_enode *slot[MAPLE_BIG_NODE_SLOTS];
Signature · 1 line
2.34.1
author
#1 lore raw

PATCH 2/2 maple_tree: memset maple_big_node as a whole

In maple_big_node, we define slot and padding/gap in a union. And based on current definition of MAPLE_BIG_NODE_SLOTS/GAPS, padding is always less then slot and part of the gap is overlapped by slot.

For example on 64bit system:

MAPLE_BIG_NODE_SLOT is 34
MAPLE_BIG_NODE_GAP is 21

With this knowledge, current code actually clear the whole maple_big_node and even clear some space twice.

Instead of clear maple_big_node each field separately, let's clear it in one memset.

CC Liam R. Howlett <[email protected]>

---
Liam:

This looks obvious, so I just run the ./maple test to see it doesn't break anything.

Do you think I need to add a benchmark and run a perf for this kind of change?

---
 lib/maple_tree.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
lib/maple_tree.c +1 -4
index d8f10773e451..911c5e04e634 100644
@@ -3134,10 +3134,7 @@ static inline void mast_fill_bnode(struct maple_subtree_state *mast,
 	bool cp = true;
 	unsigned char split;
 
-	memset(mast->bn->gap, 0, sizeof(unsigned long) * ARRAY_SIZE(mast->bn->gap));
-	memset(mast->bn->slot, 0, sizeof(unsigned long) * ARRAY_SIZE(mast->bn->slot));
-	memset(mast->bn->pivot, 0, sizeof(unsigned long) * ARRAY_SIZE(mast->bn->pivot));
-	mast->bn->b_end = 0;
+	memset(mast->bn, 0, sizeof(struct maple_big_node));
 
 	if (mte_is_root(mas->node)) {
 		cp = false;
Signature · 1 line
2.34.1
#2 lore raw

Re: [PATCH 2/2] maple_tree: memset maple_big_node as a whole

Quoted 3 lines · Wei Yang

In maple_big_node, we define slot and padding/gap in a union. And based on current definition of MAPLE_BIG_NODE_SLOTS/GAPS, padding is always less then slot and part of the gap is overlapped by slot.

       ^^^^- than
Quoted 12 lines

For example on 64bit system:

MAPLE_BIG_NODE_SLOT is 34
MAPLE_BIG_NODE_GAP is 21

With this knowledge, current code actually clear the whole maple_big_node and even clear some space twice.

Instead of clear maple_big_node each field separately, let's clear it in one memset.

Quoted 9 lines
CC Liam R. Howlett <[email protected]>

---
Liam:

This looks obvious, so I just run the ./maple test to see it doesn't break anything.

The big node also includes the type, which isn't cleared. However it is unconditionally set below, so your change log is not correct in the statement that it is fully cleared, but the code will work and it is worth fixing what you have found.

If you are sending more than one patch, it is better to make a cover letter to explain your series.

It is probably worth re-spinning the series to fix the comment, but these changes look good. I'll have a closer look later.

Quoted 2 lines

Do you think I need to add a benchmark and run a perf for this kind of change?

No, you are doing less work so this should be better as long as it is correct. Zeroing can be expensive on some archs so this is good.

Quoted 23 lines
---
 lib/maple_tree.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
lib/maple_tree.c +1 -4
index d8f10773e451..911c5e04e634 100644
@@ -3134,10 +3134,7 @@ static inline void mast_fill_bnode(struct maple_subtree_state *mast,
 	bool cp = true;
 	unsigned char split;
 
-	memset(mast->bn->gap, 0, sizeof(unsigned long) * ARRAY_SIZE(mast->bn->gap));
-	memset(mast->bn->slot, 0, sizeof(unsigned long) * ARRAY_SIZE(mast->bn->slot));
-	memset(mast->bn->pivot, 0, sizeof(unsigned long) * ARRAY_SIZE(mast->bn->pivot));
-	mast->bn->b_end = 0;
+	memset(mast->bn, 0, sizeof(struct maple_big_node));
 
 	if (mte_is_root(mas->node)) {
 		cp = false;
Signature · 1 line
2.34.1
author
#3 lore raw

Re: [PATCH 2/2] maple_tree: memset maple_big_node as a whole

Quoted 24 lines · Liam R. Howlett
Quoted 3 lines · Wei Yang

In maple_big_node, we define slot and padding/gap in a union. And based on current definition of MAPLE_BIG_NODE_SLOTS/GAPS, padding is always less then slot and part of the gap is overlapped by slot.

      ^^^^- than
Quoted 18 lines

For example on 64bit system:

MAPLE_BIG_NODE_SLOT is 34
MAPLE_BIG_NODE_GAP is 21

With this knowledge, current code actually clear the whole maple_big_node and even clear some space twice.

Instead of clear maple_big_node each field separately, let's clear it in one memset.

CC Liam R. Howlett <[email protected]>

---
Liam:

Quoted 9 lines
Quoted 3 lines

This looks obvious, so I just run the ./maple test to see it doesn't break anything.

The big node also includes the type, which isn't cleared. However it is unconditionally set below, so your change log is not correct in the statement that it is fully cleared, but the code will work and it is worth fixing what you have found.

I will mention this next time.

Quoted 3 lines

If you are sending more than one patch, it is better to make a cover letter to explain your series.

Ok, I will add a cover letter next time.

Quoted 3 lines

It is probably worth re-spinning the series to fix the comment, but these changes look good. I'll have a closer look later.

Thank,

I will re-arrange the change log and add a cover letter for v2.

Quoted 29 lines
Quoted 2 lines

Do you think I need to add a benchmark and run a perf for this kind of change?

No, you are doing less work so this should be better as long as it is correct. Zeroing can be expensive on some archs so this is good.

Quoted 23 lines
---
 lib/maple_tree.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
lib/maple_tree.c +1 -4
index d8f10773e451..911c5e04e634 100644
@@ -3134,10 +3134,7 @@ static inline void mast_fill_bnode(struct maple_subtree_state *mast,
 	bool cp = true;
 	unsigned char split;
 
-	memset(mast->bn->gap, 0, sizeof(unsigned long) * ARRAY_SIZE(mast->bn->gap));
-	memset(mast->bn->slot, 0, sizeof(unsigned long) * ARRAY_SIZE(mast->bn->slot));
-	memset(mast->bn->pivot, 0, sizeof(unsigned long) * ARRAY_SIZE(mast->bn->pivot));
-	mast->bn->b_end = 0;
+	memset(mast->bn, 0, sizeof(struct maple_big_node));
 
 	if (mte_is_root(mas->node)) {
 		cp = false;
Signature · 1 line
2.34.1
Signature · 2 lines
Wei Yang
Help you, Help me