Skip to contentSkip to conversation

Replace rcu_swap_protected() with rcu_replace_pointer()

PATCH In review Paul E. McKenney 3 participants 10 messages Dec 10, 2019 – Dec 12, 2019

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Replace rcu_swap_protected() with rcu_replace_pointer()

This commit replaces the use of rcu_swap_protected() with the more intuitively appealing rcu_replace_pointer() as a step towards removing rcu_swap_protected().

net/tipc/crypto.c +2 -2
index 990a872..64cf831 100644
@@ -258,7 +258,7 @@ static char *tipc_key_change_dump(struct tipc_key old, struct tipc_key new,
 	rcu_dereference_protected((rcu_ptr), lockdep_is_held(lock))
 
 #define tipc_aead_rcu_swap(rcu_ptr, ptr, lock)				\
-	rcu_swap_protected((rcu_ptr), (ptr), lockdep_is_held(lock))
+	rcu_replace_pointer((rcu_ptr), (ptr), lockdep_is_held(lock))
 
 #define tipc_aead_rcu_replace(rcu_ptr, ptr, lock)			\
 do {									\
@@ -1189,7 +1189,7 @@ static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)
 
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = rcu_replace_pointer(rx->aead[key.passive], tmp2, &rx->lock);
 		x = (key.passive - key.pending + new_pending) % KEY_MAX;
 		new_passive = (x <= 0) ? x + KEY_MAX : x;
 	}
#1 lore raw
Quoted 19 lines · Paul E. McKenney

This commit replaces the use of rcu_swap_protected() with the more intuitively appealing rcu_replace_pointer() as a step towards removing rcu_swap_protected().

net/tipc/crypto.c +0 -0
index 990a872..64cf831 100644

net/tipc/crypto.c static char *tipc_key_change_dump(struct tipc_key old, struct tipc_key new, · line 261

Quoted 5 lines
net/tipc/crypto.c +1 -1
@@ -258,7 +258,7 @@ static char *tipc_key_change_dump(struct tipc_key old, struct tipc_key new,
 	rcu_dereference_protected((rcu_ptr), lockdep_is_held(lock))
 
 #define tipc_aead_rcu_swap(rcu_ptr, ptr, lock)				\
-	rcu_swap_protected((rcu_ptr), (ptr), lockdep_is_held(lock))
+	rcu_replace_pointer((rcu_ptr), (ptr), lockdep_is_held(lock))

(ptr) = rcu_replace_pointer((rcu_ptr), (ptr), lockdep_is_held(lock))

net/tipc/crypto.c static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending) · line 1192

Quoted 9 lines
net/tipc/crypto.c +1 -1
 #define tipc_aead_rcu_replace(rcu_ptr, ptr, lock)			\
 do {									\
@@ -1189,7 +1189,7 @@ static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)
 
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = rcu_replace_pointer(rx->aead[key.passive], tmp2, &rx->lock);

tipc_aead_rcu_swap() is only called here in TIPC module. If we use rcu_replace_pointer() to switch pointers instead of calling tipc_aead_rcu_swap() macro, I think we should completely remove tipc_aead_rcu_swap().

Quoted 4 lines
net/tipc/crypto.c +0 -0
 		x = (key.passive - key.pending + new_pending) % KEY_MAX;
 		new_passive = (x <= 0) ? x + KEY_MAX : x;
 	}
#2 lore raw
Quoted 33 lines · Ying Xue
Quoted 19 lines · Paul E. McKenney

This commit replaces the use of rcu_swap_protected() with the more intuitively appealing rcu_replace_pointer() as a step towards removing rcu_swap_protected().

net/tipc/crypto.c +0 -0
index 990a872..64cf831 100644

net/tipc/crypto.c static char *tipc_key_change_dump(struct tipc_key old, struct tipc_key new, · line 261

Quoted 5 lines
net/tipc/crypto.c +1 -1
@@ -258,7 +258,7 @@ static char *tipc_key_change_dump(struct tipc_key old, struct tipc_key new,
 	rcu_dereference_protected((rcu_ptr), lockdep_is_held(lock))
 
 #define tipc_aead_rcu_swap(rcu_ptr, ptr, lock)				\
-	rcu_swap_protected((rcu_ptr), (ptr), lockdep_is_held(lock))
+	rcu_replace_pointer((rcu_ptr), (ptr), lockdep_is_held(lock))

(ptr) = rcu_replace_pointer((rcu_ptr), (ptr), lockdep_is_held(lock))

Quoted 5 lines
net/tipc/crypto.c +0 -0
 #define tipc_aead_rcu_replace(rcu_ptr, ptr, lock)			\
 do {									\
@@ -1189,7 +1189,7 @@ static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)
 
Quoted 9 lines

net/tipc/crypto.c static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending) · line 1192

Quoted 4 lines
net/tipc/crypto.c +1 -1
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = rcu_replace_pointer(rx->aead[key.passive], tmp2, &rx->lock);

tipc_aead_rcu_swap() is only called here in TIPC module. If we use rcu_replace_pointer() to switch pointers instead of calling tipc_aead_rcu_swap() macro, I think we should completely remove tipc_aead_rcu_swap().

Good catch, thank you! How about the following instead?

Thanx, Paul

------------------------------------------------------------------------

commit 4ee8e2c68b076867b7a5af82a38010fffcab611c

Author: Paul E. McKenney <[email protected]>
Date:   Mon Dec 9 19:13:45 2019 -0800

net/tipc: Replace rcu_swap_protected() with rcu_replace_pointer()

This commit replaces the use of rcu_swap_protected() with the more intuitively appealing rcu_replace_pointer() as a step towards removing rcu_swap_protected().

net/tipc/crypto.c +1 -4
index 990a872..978d2db 100644
@@ -257,9 +257,6 @@ static char *tipc_key_change_dump(struct tipc_key old, struct tipc_key new,
 #define tipc_aead_rcu_ptr(rcu_ptr, lock)				\
 	rcu_dereference_protected((rcu_ptr), lockdep_is_held(lock))
 
-#define tipc_aead_rcu_swap(rcu_ptr, ptr, lock)				\
-	rcu_swap_protected((rcu_ptr), (ptr), lockdep_is_held(lock))
-
 #define tipc_aead_rcu_replace(rcu_ptr, ptr, lock)			\
 do {									\
 	typeof(rcu_ptr) __tmp = rcu_dereference_protected((rcu_ptr),	\
@@ -1189,7 +1186,7 @@ static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)
 
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = rcu_replace_pointer(rx->aead[key.passive], tmp2, &rx->lock);
 		x = (key.passive - key.pending + new_pending) % KEY_MAX;
 		new_passive = (x <= 0) ? x + KEY_MAX : x;
 	}
Acked-by
#3 lore raw
Quoted 10 lines · Paul E. McKenney

commit 4ee8e2c68b076867b7a5af82a38010fffcab611c

Author: Paul E. McKenney <[email protected]>
Date:   Mon Dec 9 19:13:45 2019 -0800

net/tipc: Replace rcu_swap_protected() with rcu_replace_pointer()

This commit replaces the use of rcu_swap_protected() with the more intuitively appealing rcu_replace_pointer() as a step towards removing rcu_swap_protected().

Quoted 10 lines
Quoted 24 lines
net/tipc/crypto.c +1 -4
index 990a872..978d2db 100644
@@ -257,9 +257,6 @@ static char *tipc_key_change_dump(struct tipc_key old, struct tipc_key new,
 #define tipc_aead_rcu_ptr(rcu_ptr, lock)				\
 	rcu_dereference_protected((rcu_ptr), lockdep_is_held(lock))
 
-#define tipc_aead_rcu_swap(rcu_ptr, ptr, lock)				\
-	rcu_swap_protected((rcu_ptr), (ptr), lockdep_is_held(lock))
-
 #define tipc_aead_rcu_replace(rcu_ptr, ptr, lock)			\
 do {									\
 	typeof(rcu_ptr) __tmp = rcu_dereference_protected((rcu_ptr),	\
@@ -1189,7 +1186,7 @@ static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)
 
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = rcu_replace_pointer(rx->aead[key.passive], tmp2, &rx->lock);
 		x = (key.passive - key.pending + new_pending) % KEY_MAX;
 		new_passive = (x <= 0) ? x + KEY_MAX : x;
 	}
Acked-by Ying Xue
#4 lore raw

Hi Ying, Paul,

Please see my comments inline. Thanks!

BR/Tuong

-----Original Message-----
From: Ying Xue <[email protected]>
Sent: Wednesday, December 11, 2019 8:32 AM
To: [email protected]
Cc: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [tipc-discussion] [PATCH net/tipc] Replace rcu_swap_protected() with rcu_replace_pointer()

Quoted 11 lines · Paul E. McKenney

commit 4ee8e2c68b076867b7a5af82a38010fffcab611c

Author: Paul E. McKenney <[email protected]>
Date:   Mon Dec 9 19:13:45 2019 -0800

net/tipc: Replace rcu_swap_protected() with rcu_replace_pointer()

This commit replaces the use of rcu_swap_protected() with the more intuitively appealing rcu_replace_pointer() as a step towards removing rcu_swap_protected().

Link:

https://lore.kernel.org/lkml/CAHk-=wiAsJLw1egFEE=Z7-GGtM6wcvtyytXZA1+BHqta4g
[email protected]/

Quoted 9 lines

net/tipc/crypto.c static char *tipc_key_change_dump(struct tipc_key old,

Quoted 5 lines
net/tipc/crypto.c +0 -0
index 990a872..978d2db 100644
@@ -257,9 +257,6 @@ static char *tipc_key_change_dump(struct tipc_key old,

struct tipc_key new,

net/tipc/crypto.c static char *tipc_key_change_dump(struct tipc_key old, struct tipc_key new,

Quoted 4 lines
net/tipc/crypto.c +0 -1
 #define tipc_aead_rcu_ptr(rcu_ptr, lock)				\
 	rcu_dereference_protected((rcu_ptr), lockdep_is_held(lock))
 
-#define tipc_aead_rcu_swap(rcu_ptr, ptr, lock)

\

net/tipc/crypto.c static bool tipc_crypto_key_try_align(struct

Quoted 6 lines
net/tipc/crypto.c +0 -2
-	rcu_swap_protected((rcu_ptr), (ptr), lockdep_is_held(lock))
-
 #define tipc_aead_rcu_replace(rcu_ptr, ptr, lock)			\
 do {									\
 	typeof(rcu_ptr) __tmp = rcu_dereference_protected((rcu_ptr),	\
@@ -1189,7 +1186,7 @@ static bool tipc_crypto_key_try_align(struct

tipc_crypto *rx, u8 new_pending)

net/tipc/crypto.c static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)

Quoted 5 lines
net/tipc/crypto.c +1 -1
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = rcu_replace_pointer(rx->aead[key.passive], tmp2,

&rx->lock);
The 3rd parameter should be the lockdep condition checking instead of the spinlock's pointer i.e. "lockdep_is_held(&rx->lock)"?
That's why I'd prefer to use the 'tipc_aead_rcu_swap ()' macro, which is clear & concise at least for the context here. It might be re-used later as well...

net/tipc/crypto.c static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)

Quoted 4 lines
net/tipc/crypto.c +0 -0
 		x = (key.passive - key.pending + new_pending) % KEY_MAX;
 		new_passive = (x <= 0) ? x + KEY_MAX : x;
 	}
 

_______________________________________________
tipc-discussion mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/tipc-discussion

#5 lore raw
Quoted 3 lines · Tuong Lien Tong

Hi Ying, Paul,

Please see my comments inline. Thanks!

Good catch!

Quoted 35 lines

BR/Tuong

-----Original Message-----
From: Ying Xue <[email protected]>
Sent: Wednesday, December 11, 2019 8:32 AM
To: [email protected]
Cc: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
Subject: Re: [tipc-discussion] [PATCH net/tipc] Replace rcu_swap_protected() with rcu_replace_pointer()

Quoted 11 lines · Paul E. McKenney

commit 4ee8e2c68b076867b7a5af82a38010fffcab611c

Author: Paul E. McKenney <[email protected]>
Date:   Mon Dec 9 19:13:45 2019 -0800

net/tipc: Replace rcu_swap_protected() with rcu_replace_pointer()

This commit replaces the use of rcu_swap_protected() with the more intuitively appealing rcu_replace_pointer() as a step towards removing rcu_swap_protected().

Link:

https://lore.kernel.org/lkml/CAHk-=wiAsJLw1egFEE=Z7-GGtM6wcvtyytXZA1+BHqta4g
[email protected]/

Quoted 8 lines
Quoted 1 line

As in the following? If so, I will be very happy to apply your Acked-by.

Thanx, Paul

------------------------------------------------------------------------

commit 4c0855120704e7a578dc6862ae57babf6dc9bc77

Author: Paul E. McKenney <[email protected]>
Date:   Mon Dec 9 19:13:45 2019 -0800

net/tipc: Replace rcu_swap_protected() with rcu_replace_pointer()

This commit replaces the use of rcu_swap_protected() with the more intuitively appealing rcu_replace_pointer() as a step towards removing rcu_swap_protected().

net/tipc/crypto.c +2 -2
index 990a872..39a13b4 100644
@@ -258,7 +258,7 @@ static char *tipc_key_change_dump(struct tipc_key old, struct tipc_key new,
 	rcu_dereference_protected((rcu_ptr), lockdep_is_held(lock))
 
 #define tipc_aead_rcu_swap(rcu_ptr, ptr, lock)				\
-	rcu_swap_protected((rcu_ptr), (ptr), lockdep_is_held(lock))
+	rcu_replace_pointer((rcu_ptr), (ptr), lockdep_is_held(lock))
 
 #define tipc_aead_rcu_replace(rcu_ptr, ptr, lock)			\
 do {									\
@@ -1189,7 +1189,7 @@ static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)
 
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
 		x = (key.passive - key.pending + new_pending) % KEY_MAX;
 		new_passive = (x <= 0) ? x + KEY_MAX : x;
 	}
#7 lore raw
In reply to Tuong Lien Tong
Quoted 12 lines · Tuong Lien Tong

net/tipc/crypto.c static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)

Quoted 5 lines
net/tipc/crypto.c +1 -1
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = rcu_replace_pointer(rx->aead[key.passive], tmp2,

&rx->lock);
The 3rd parameter should be the lockdep condition checking instead of the spinlock's pointer i.e. "lockdep_is_held(&rx->lock)"?
That's why I'd prefer to use the 'tipc_aead_rcu_swap ()' macro, which is clear & concise at least for the context here. It might be re-used later as well...

Right. The 3rd parameter of rcu_replace_pointer() should be "lockdep_is_held(&rx->lock)" instead of "&rx->lock".

#8 lore raw
Quoted 14 lines · Ying Xue
Quoted 12 lines · Tuong Lien Tong

Review

Quoted 5 lines
(quoted diff) +1 -1
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = rcu_replace_pointer(rx->aead[key.passive], tmp2,

&rx->lock);
The 3rd parameter should be the lockdep condition checking instead of the spinlock's pointer i.e. "lockdep_is_held(&rx->lock)"?
That's why I'd prefer to use the 'tipc_aead_rcu_swap ()' macro, which is clear & concise at least for the context here. It might be re-used later as well...

Quoted 2 lines

Right. The 3rd parameter of rcu_replace_pointer() should be "lockdep_is_held(&rx->lock)" instead of "&rx->lock".

Like this?

Thanx, Paul

------------------------------------------------------------------------

commit 575bb4ba1b22383656760feb3d122e11656ccdfd

Author: Paul E. McKenney <[email protected]>
Date:   Mon Dec 9 19:13:45 2019 -0800

net/tipc: Replace rcu_swap_protected() with rcu_replace_pointer()

This commit replaces the use of rcu_swap_protected() with the more intuitively appealing rcu_replace_pointer() as a step towards removing rcu_swap_protected().

net/tipc/crypto.c +1 -4
index 990a872..c8c47fc 100644
@@ -257,9 +257,6 @@ static char *tipc_key_change_dump(struct tipc_key old, struct tipc_key new,
 #define tipc_aead_rcu_ptr(rcu_ptr, lock)				\
 	rcu_dereference_protected((rcu_ptr), lockdep_is_held(lock))
 
-#define tipc_aead_rcu_swap(rcu_ptr, ptr, lock)				\
-	rcu_swap_protected((rcu_ptr), (ptr), lockdep_is_held(lock))
-
 #define tipc_aead_rcu_replace(rcu_ptr, ptr, lock)			\
 do {									\
 	typeof(rcu_ptr) __tmp = rcu_dereference_protected((rcu_ptr),	\
@@ -1189,7 +1186,7 @@ static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)
 
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = rcu_replace_pointer(rx->aead[key.passive], tmp2, lockdep_is_held(&rx->lock));
 		x = (key.passive - key.pending + new_pending) % KEY_MAX;
 		new_passive = (x <= 0) ? x + KEY_MAX : x;
 	}
#9 lore raw
Quoted 15 lines · Paul E. McKenney
Quoted 14 lines · Ying Xue
Quoted 12 lines · Tuong Lien Tong

net/tipc/crypto.c static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)

Quoted 5 lines
net/tipc/crypto.c +1 -1
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = rcu_replace_pointer(rx->aead[key.passive], tmp2,

&rx->lock);
The 3rd parameter should be the lockdep condition checking instead of the spinlock's pointer i.e. "lockdep_is_held(&rx->lock)"?
That's why I'd prefer to use the 'tipc_aead_rcu_swap ()' macro, which is clear & concise at least for the context here. It might be re-used later as well...

Quoted 4 lines
Quoted 2 lines

Right. The 3rd parameter of rcu_replace_pointer() should be "lockdep_is_held(&rx->lock)" instead of "&rx->lock".

Like this?

Yes, I think it's better to set the 3rd parameter of rcu_replace_pointer() with "lockdep_is_held(&rx->lock)".

Quoted 50 lines

Thanx, Paul

------------------------------------------------------------------------

commit 575bb4ba1b22383656760feb3d122e11656ccdfd

Author: Paul E. McKenney <[email protected]>
Date:   Mon Dec 9 19:13:45 2019 -0800

net/tipc: Replace rcu_swap_protected() with rcu_replace_pointer()

This commit replaces the use of rcu_swap_protected() with the more intuitively appealing rcu_replace_pointer() as a step towards removing rcu_swap_protected().

net/tipc/crypto.c +1 -4
index 990a872..c8c47fc 100644
@@ -257,9 +257,6 @@ static char *tipc_key_change_dump(struct tipc_key old, struct tipc_key new,
 #define tipc_aead_rcu_ptr(rcu_ptr, lock)				\
 	rcu_dereference_protected((rcu_ptr), lockdep_is_held(lock))
 
-#define tipc_aead_rcu_swap(rcu_ptr, ptr, lock)				\
-	rcu_swap_protected((rcu_ptr), (ptr), lockdep_is_held(lock))
-
 #define tipc_aead_rcu_replace(rcu_ptr, ptr, lock)			\
 do {									\
 	typeof(rcu_ptr) __tmp = rcu_dereference_protected((rcu_ptr),	\
@@ -1189,7 +1186,7 @@ static bool tipc_crypto_key_try_align(struct tipc_crypto *rx, u8 new_pending)
 
 	/* Move passive key if any */
 	if (key.passive) {
-		tipc_aead_rcu_swap(rx->aead[key.passive], tmp2, &rx->lock);
+		tmp2 = rcu_replace_pointer(rx->aead[key.passive], tmp2, lockdep_is_held(&rx->lock));
 		x = (key.passive - key.pending + new_pending) % KEY_MAX;
 		new_passive = (x <= 0) ? x + KEY_MAX : x;
 	}