Skip to contentSkip to conversation

atomics: API cleanups

PATCH Superseded Mark Rutland 4 participants 33 messages May 29, 2018 – Jun 6, 2018

Subsystems (29)
Networking Arm (32-bit) x86 PowerPC +25 more subsystems

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Mark Rutland

author reviewer
#0 lore raw

v2 cover atomics: API cleanups

This series contains a few cleanups of the atomic API, fixing inconsistencies between atomic_* and atomic64_*, and minimizing repetition in arch code. This is nicer for arch code, and the improved regularity will help when generating the atomic headers in future.

Since v1 [1]:
* Add missing inc_not_zero #define for x86_32
* Remove newly redundant test op definitions for riscv
* Remove atomic_inc_not_zero_hint()
* Make inc/dec ops optional

The bulk of the patches reorganise things so architectures consistently provide <atomic>_fetch_add_unless(), with atomic_fetch_add_unless() provided as a wrapper by core code. A generic fallback is provided for <atomic>_fetch_add_unless(), based on <atomic>_read() and <atomic>_try_cmpxchg().

Other patches in the series add common fallbacks for:

* atomic64_inc_not_zero()
* <atomic>_inc_and_test()
* <atomic>_dec_and_test()
* <atomic>_sub_and_test()
* <atomic>add_negative()
* <atomic>_*inc*()
* <atomic>_*dec*()

... as almost all architectures provide (near-)identical implementation of these today. Where an architecture provides a non-trivial definition, it is updated to provide a matching preprocessor symbol, and the instrumented atomics are updated correspondingly.

The end result is a strongly negative diffstat, though <linux/atomic.h> grows by a reasonable amount. We could halve this in future by templating the various fallbacks for atomic{,64}_t with code generation.

I've pushed this out to my atomics/api-cleanup branch [2] on kernel.org.

Thanks,
Mark.

[1] https://lkml.kernel.org/r/[email protected]
[2] git://git.kernel.org/pub/scm/linux/kernel/git/mark/linux.git atomics/api-cleanup

Mark Rutland (16):
atomics/treewide: s/__atomic_add_unless/atomic_fetch_add_unless/ atomics/treewide: remove redundant atomic_inc_not_zero() definitions atomics/treewide: make atomic64_inc_not_zero() optional atomics/treewide: make atomic_fetch_add_unless() optional
atomics: prepare for atomic64_fetch_add_unless() atomics/generic: define atomic64_fetch_add_unless() atomics/alpha: define atomic64_fetch_add_unless() atomics/arc: define atomic64_fetch_add_unless() atomics/arm: define atomic64_fetch_add_unless() atomics/powerpc: define atomic64_fetch_add_unless() atomics/riscv: define atomic64_fetch_add_unless() atomics/treewide: make atomic64_fetch_add_unless() optional atomics/treewide: make test ops optional atomics/treewide: remove atomic_inc_not_zero_hint() atomics/treewide: make unconditional inc/dec ops optional atomics/treewide: make conditional inc/dec ops optional

 arch/alpha/include/asm/atomic.h           |  56 ++---
 arch/arc/include/asm/atomic.h             |  78 ++-----
 arch/arm/include/asm/atomic.h             |  55 ++---
 arch/arm64/include/asm/atomic.h           |  47 +---
 arch/h8300/include/asm/atomic.h           |  15 +-
 arch/hexagon/include/asm/atomic.h         |  18 +-
 arch/ia64/include/asm/atomic.h            |  81 -------
 arch/m68k/include/asm/atomic.h            |  24 +-
 arch/mips/include/asm/atomic.h            | 172 --------------
 arch/openrisc/include/asm/atomic.h        |   4 +-
 arch/parisc/include/asm/atomic.h          | 107 ---------
 arch/powerpc/include/asm/atomic.h         |  52 ++---
 arch/riscv/include/asm/atomic.h           | 149 +-----------
 arch/s390/include/asm/atomic.h            |  65 ------
 arch/sh/include/asm/atomic.h              |  35 ---
 arch/sparc/include/asm/atomic_32.h        |  24 +-
 arch/sparc/include/asm/atomic_64.h        |  65 +-----
 arch/sparc/lib/atomic32.c                 |   4 +-
 arch/x86/include/asm/atomic.h             |  30 +--
 arch/x86/include/asm/atomic64_32.h        |  61 +----
 arch/x86/include/asm/atomic64_64.h        |  48 +---
 arch/xtensa/include/asm/atomic.h          |  98 --------
 drivers/block/rbd.c                       |   2 +-
 drivers/infiniband/core/rdma_core.c       |   2 +-
 fs/afs/rxrpc.c                            |   2 +-
 include/asm-generic/atomic-instrumented.h |  69 +++++-
 include/asm-generic/atomic.h              |  33 ---
 include/asm-generic/atomic64.h            |  14 +-
 include/linux/atomic.h                    | 371 +++++++++++++++++++++++++-----
 kernel/bpf/syscall.c                      |   4 +-
 lib/atomic64.c                            |  13 +-
 net/atm/pppoatm.c                         |   2 +-
 net/rxrpc/call_object.c                   |   2 +-
 net/rxrpc/conn_object.c                   |   4 +-
 net/rxrpc/local_object.c                  |   2 +-
 net/rxrpc/peer_object.c                   |   2 +-
 36 files changed, 499 insertions(+), 1311 deletions(-)
Signature · 1 line
2.11.0
author reviewer
#1 lore raw

PATCH v2 1/16 atomics/treewide: s/__atomic_add_unless/atomic_fetch_add_unless/

While __atomic_add_unless was originally intended as a building-block for atomic_add_unless, it's now used in a number of places around the kernel. It's the only common atomic operation named __atomic*(), rather than atomic_*(), and for consistency it would be better named atomic_fetch_add_unless().

This lack of consistency is slightly confusing, and gets in the way of scripting atomics. Given that, let's clean things up and promote it to an offical part of the atomics API, in the form of atomic_fetch_add_unless().

This patch converts definitions and uses over to the new name, including the instrumented version, using the following script:

----
git grep -w __atomic_add_unless | while read line; do
sed -i '{s/\<__atomic_add_unless\>/atomic_fetch_add_unless/}' "${line%%:*}"; done
git grep -w __arch_atomic_add_unless | while read line; do
sed -i '{s/\<__arch_atomic_add_unless\>/arch_atomic_fetch_add_unless/}' "${line%%:*}"; done
----

Note that we do not have atomic{64,_long}_fetch_add_unless(), which will be introduced by later patches.

There should be no functional change as a result of this patch.

---
 arch/alpha/include/asm/atomic.h           | 4 ++--
 arch/arc/include/asm/atomic.h             | 4 ++--
 arch/arm/include/asm/atomic.h             | 4 ++--
 arch/arm64/include/asm/atomic.h           | 2 +-
 arch/h8300/include/asm/atomic.h           | 2 +-
 arch/hexagon/include/asm/atomic.h         | 4 ++--
 arch/ia64/include/asm/atomic.h            | 2 +-
 arch/m68k/include/asm/atomic.h            | 2 +-
 arch/mips/include/asm/atomic.h            | 4 ++--
 arch/openrisc/include/asm/atomic.h        | 4 ++--
 arch/parisc/include/asm/atomic.h          | 4 ++--
 arch/powerpc/include/asm/atomic.h         | 8 ++++----
 arch/riscv/include/asm/atomic.h           | 4 ++--
 arch/s390/include/asm/atomic.h            | 2 +-
 arch/sh/include/asm/atomic.h              | 4 ++--
 arch/sparc/include/asm/atomic_32.h        | 2 +-
 arch/sparc/include/asm/atomic_64.h        | 2 +-
 arch/sparc/lib/atomic32.c                 | 4 ++--
 arch/x86/include/asm/atomic.h             | 4 ++--
 arch/xtensa/include/asm/atomic.h          | 4 ++--
 drivers/block/rbd.c                       | 2 +-
 drivers/infiniband/core/rdma_core.c       | 2 +-
 fs/afs/rxrpc.c                            | 2 +-
 include/asm-generic/atomic-instrumented.h | 4 ++--
 include/asm-generic/atomic.h              | 4 ++--
 include/linux/atomic.h                    | 2 +-
 kernel/bpf/syscall.c                      | 4 ++--
 net/rxrpc/call_object.c                   | 2 +-
 net/rxrpc/conn_object.c                   | 4 ++--
 net/rxrpc/local_object.c                  | 2 +-
 net/rxrpc/peer_object.c                   | 2 +-
 31 files changed, 50 insertions(+), 50 deletions(-)
arch/alpha/include/asm/atomic.h +2 -2
index 767bfdd42992..392b15a4dd4f 100644
@@ -206,7 +206,7 @@ ATOMIC_OPS(xor, xor)
 #define atomic_xchg(v, new) (xchg(&((v)->counter), new))
 
 /**
- * __atomic_add_unless - add unless the number is a given value
+ * atomic_fetch_add_unless - add unless the number is a given value
  * @v: pointer of type atomic_t
  * @a: the amount to add to v...
  * @u: ...unless v is equal to u.
@@ -214,7 +214,7 @@ ATOMIC_OPS(xor, xor)
  * Atomically adds @a to @v, so long as it was not @u.
  * Returns the old value of @v.
  */
-static __inline__ int __atomic_add_unless(atomic_t *v, int a, int u)
+static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, new, old;
 	smp_mb();
arch/arc/include/asm/atomic.h +2 -2
index 11859287c52a..67121b5ff3a3 100644
@@ -309,7 +309,7 @@ ATOMIC_OPS(xor, ^=, CTOP_INST_AXOR_DI_R2_R2_R3)
 #undef ATOMIC_OP
 
 /**
- * __atomic_add_unless - add unless the number is a given value
+ * atomic_fetch_add_unless - add unless the number is a given value
  * @v: pointer of type atomic_t
  * @a: the amount to add to v...
  * @u: ...unless v is equal to u.
@@ -317,7 +317,7 @@ ATOMIC_OPS(xor, ^=, CTOP_INST_AXOR_DI_R2_R2_R3)
  * Atomically adds @a to @v, so long as it was not @u.
  * Returns the old value of @v
  */
-#define __atomic_add_unless(v, a, u)					\
+#define atomic_fetch_add_unless(v, a, u)					\
 ({									\
 	int c, old;							\
 									\
arch/arm/include/asm/atomic.h +2 -2
index 66d0e215a773..9d56d0727c9b 100644
@@ -130,7 +130,7 @@ static inline int atomic_cmpxchg_relaxed(atomic_t *ptr, int old, int new)
 }
 #define atomic_cmpxchg_relaxed		atomic_cmpxchg_relaxed
 
-static inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int oldval, newval;
 	unsigned long tmp;
@@ -215,7 +215,7 @@ static inline int atomic_cmpxchg(atomic_t *v, int old, int new)
 	return ret;
 }
 
-static inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, old;
 
arch/arm64/include/asm/atomic.h +1 -1
index c0235e0ff849..264d20339f74 100644
@@ -125,7 +125,7 @@
 #define atomic_dec_and_test(v)		(atomic_dec_return(v) == 0)
 #define atomic_sub_and_test(i, v)	(atomic_sub_return((i), (v)) == 0)
 #define atomic_add_negative(i, v)	(atomic_add_return((i), (v)) < 0)
-#define __atomic_add_unless(v, a, u)	___atomic_add_unless(v, a, u,)
+#define atomic_fetch_add_unless(v, a, u)	___atomic_add_unless(v, a, u,)
 #define atomic_andnot			atomic_andnot
 
 /*
arch/h8300/include/asm/atomic.h +1 -1
index 941e7554e886..4465cfc30a3a 100644
@@ -94,7 +94,7 @@ static inline int atomic_cmpxchg(atomic_t *v, int old, int new)
 	return ret;
 }
 
-static inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int ret;
 	h8300flags flags;
arch/hexagon/include/asm/atomic.h +2 -2
index fb3dfb2a667e..287aa9f394f3 100644
@@ -164,7 +164,7 @@ ATOMIC_OPS(xor)
 #undef ATOMIC_OP
 
 /**
- * __atomic_add_unless - add unless the number is a given value
+ * atomic_fetch_add_unless - add unless the number is a given value
  * @v: pointer to value
  * @a: amount to add
  * @u: unless value is equal to u
@@ -173,7 +173,7 @@ ATOMIC_OPS(xor)
  *
  */
 
-static inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int __oldval;
 	register int tmp;
arch/ia64/include/asm/atomic.h +1 -1
index 2524fb60fbc2..9d2ddde5f9d5 100644
@@ -215,7 +215,7 @@ ATOMIC64_FETCH_OP(xor, ^)
 	(cmpxchg(&((v)->counter), old, new))
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), new))
 
-static __inline__ int __atomic_add_unless(atomic_t *v, int a, int u)
+static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, old;
 	c = atomic_read(v);
arch/m68k/include/asm/atomic.h +1 -1
index e993e2860ee1..8022d9ea1213 100644
@@ -211,7 +211,7 @@ static inline int atomic_add_negative(int i, atomic_t *v)
 	return c != 0;
 }
 
-static __inline__ int __atomic_add_unless(atomic_t *v, int a, int u)
+static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, old;
 	c = atomic_read(v);
arch/mips/include/asm/atomic.h +2 -2
index 0ab176bdb8e8..02fc1553cf9b 100644
@@ -275,7 +275,7 @@ static __inline__ int atomic_sub_if_positive(int i, atomic_t * v)
 #define atomic_xchg(v, new) (xchg(&((v)->counter), (new)))
 
 /**
- * __atomic_add_unless - add unless the number is a given value
+ * atomic_fetch_add_unless - add unless the number is a given value
  * @v: pointer of type atomic_t
  * @a: the amount to add to v...
  * @u: ...unless v is equal to u.
@@ -283,7 +283,7 @@ static __inline__ int atomic_sub_if_positive(int i, atomic_t * v)
  * Atomically adds @a to @v, so long as it was not @u.
  * Returns the old value of @v.
  */
-static __inline__ int __atomic_add_unless(atomic_t *v, int a, int u)
+static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, old;
 	c = atomic_read(v);
arch/openrisc/include/asm/atomic.h +2 -2
index 146e1660f00e..b589fac39b92 100644
@@ -100,7 +100,7 @@ ATOMIC_OP(xor)
  *
  * This is often used through atomic_inc_not_zero()
  */
-static inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int old, tmp;
 
@@ -119,7 +119,7 @@ static inline int __atomic_add_unless(atomic_t *v, int a, int u)
 
 	return old;
 }
-#define __atomic_add_unless	__atomic_add_unless
+#define atomic_fetch_add_unless	atomic_fetch_add_unless
 
 #include <asm-generic/atomic.h>
 
arch/parisc/include/asm/atomic.h +2 -2
index 88bae6676c9b..7748abced766 100644
@@ -78,7 +78,7 @@ static __inline__ int atomic_read(const atomic_t *v)
 #define atomic_xchg(v, new) (xchg(&((v)->counter), new))
 
 /**
- * __atomic_add_unless - add unless the number is a given value
+ * atomic_fetch_add_unless - add unless the number is a given value
  * @v: pointer of type atomic_t
  * @a: the amount to add to v...
  * @u: ...unless v is equal to u.
@@ -86,7 +86,7 @@ static __inline__ int atomic_read(const atomic_t *v)
  * Atomically adds @a to @v, so long as it was not @u.
  * Returns the old value of @v.
  */
-static __inline__ int __atomic_add_unless(atomic_t *v, int a, int u)
+static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, old;
 	c = atomic_read(v);
arch/powerpc/include/asm/atomic.h +4 -4
index 682b3e6a1e21..1483261080a1 100644
@@ -218,7 +218,7 @@ static __inline__ int atomic_dec_return_relaxed(atomic_t *v)
 #define atomic_xchg_relaxed(v, new) xchg_relaxed(&((v)->counter), (new))
 
 /**
- * __atomic_add_unless - add unless the number is a given value
+ * atomic_fetch_add_unless - add unless the number is a given value
  * @v: pointer of type atomic_t
  * @a: the amount to add to v...
  * @u: ...unless v is equal to u.
@@ -226,13 +226,13 @@ static __inline__ int atomic_dec_return_relaxed(atomic_t *v)
  * Atomically adds @a to @v, so long as it was not @u.
  * Returns the old value of @v.
  */
-static __inline__ int __atomic_add_unless(atomic_t *v, int a, int u)
+static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int t;
 
 	__asm__ __volatile__ (
 	PPC_ATOMIC_ENTRY_BARRIER
-"1:	lwarx	%0,0,%1		# __atomic_add_unless\n\
+"1:	lwarx	%0,0,%1		# atomic_fetch_add_unless\n\
 	cmpw	0,%0,%3 \n\
 	beq	2f \n\
 	add	%0,%2,%0 \n"
@@ -538,7 +538,7 @@ static __inline__ int atomic64_add_unless(atomic64_t *v, long a, long u)
 
 	__asm__ __volatile__ (
 	PPC_ATOMIC_ENTRY_BARRIER
-"1:	ldarx	%0,0,%1		# __atomic_add_unless\n\
+"1:	ldarx	%0,0,%1		# atomic_fetch_add_unless\n\
 	cmpd	0,%0,%3 \n\
 	beq	2f \n\
 	add	%0,%2,%0 \n"
arch/riscv/include/asm/atomic.h +2 -2
index 855115ace98c..739e810c857e 100644
@@ -332,7 +332,7 @@ ATOMIC_OP(dec_and_test, dec, ==, 0, 64)
 #undef ATOMIC_OP
 
 /* This is required to provide a full barrier on success. */
-static __always_inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
        int prev, rc;
 
@@ -381,7 +381,7 @@ static __always_inline int atomic64_add_unless(atomic64_t *v, long a, long u)
  */
 static __always_inline int atomic_inc_not_zero(atomic_t *v)
 {
-        return __atomic_add_unless(v, 1, 0);
+        return atomic_fetch_add_unless(v, 1, 0);
 }
 
 #ifndef CONFIG_GENERIC_ATOMIC64
arch/s390/include/asm/atomic.h +1 -1
index 4b55532f15c4..c2858cdd8c29 100644
@@ -90,7 +90,7 @@ static inline int atomic_cmpxchg(atomic_t *v, int old, int new)
 	return __atomic_cmpxchg(&v->counter, old, new);
 }
 
-static inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, old;
 	c = atomic_read(v);
arch/sh/include/asm/atomic.h +2 -2
index 0fd0099f43cc..ef45931ebac5 100644
@@ -46,7 +46,7 @@
 #define atomic_cmpxchg(v, o, n)		(cmpxchg(&((v)->counter), (o), (n)))
 
 /**
- * __atomic_add_unless - add unless the number is a given value
+ * atomic_fetch_add_unless - add unless the number is a given value
  * @v: pointer of type atomic_t
  * @a: the amount to add to v...
  * @u: ...unless v is equal to u.
@@ -54,7 +54,7 @@
  * Atomically adds @a to @v, so long as it was not @u.
  * Returns the old value of @v.
  */
-static inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, old;
 	c = atomic_read(v);
arch/sparc/include/asm/atomic_32.h +1 -1
index d13ce517f4b9..a58f4b43bcc7 100644
@@ -27,7 +27,7 @@ int atomic_fetch_or(int, atomic_t *);
 int atomic_fetch_xor(int, atomic_t *);
 int atomic_cmpxchg(atomic_t *, int, int);
 int atomic_xchg(atomic_t *, int);
-int __atomic_add_unless(atomic_t *, int, int);
+int atomic_fetch_add_unless(atomic_t *, int, int);
 void atomic_set(atomic_t *, int);
 
 #define atomic_set_release(v, i)	atomic_set((v), (i))
arch/sparc/include/asm/atomic_64.h +1 -1
index 28db058d471b..f416fd3d2708 100644
@@ -89,7 +89,7 @@ static inline int atomic_xchg(atomic_t *v, int new)
 	return xchg(&v->counter, new);
 }
 
-static inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, old;
 	c = atomic_read(v);
arch/sparc/lib/atomic32.c +2 -2
index 465a901a0ada..281fa634bb1a 100644
@@ -95,7 +95,7 @@ int atomic_cmpxchg(atomic_t *v, int old, int new)
 }
 EXPORT_SYMBOL(atomic_cmpxchg);
 
-int __atomic_add_unless(atomic_t *v, int a, int u)
+int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int ret;
 	unsigned long flags;
@@ -107,7 +107,7 @@ int __atomic_add_unless(atomic_t *v, int a, int u)
 	spin_unlock_irqrestore(ATOMIC_HASH(v), flags);
 	return ret;
 }
-EXPORT_SYMBOL(__atomic_add_unless);
+EXPORT_SYMBOL(atomic_fetch_add_unless);
 
 /* Atomic operations are already serializing */
 void atomic_set(atomic_t *v, int i)
arch/x86/include/asm/atomic.h +2 -2
index 0db6bec95489..84ed0bd76aef 100644
@@ -254,7 +254,7 @@ static inline int arch_atomic_fetch_xor(int i, atomic_t *v)
 }
 
 /**
- * __arch_atomic_add_unless - add unless the number is already a given value
+ * arch_atomic_fetch_add_unless - add unless the number is already a given value
  * @v: pointer of type atomic_t
  * @a: the amount to add to v...
  * @u: ...unless v is equal to u.
@@ -262,7 +262,7 @@ static inline int arch_atomic_fetch_xor(int i, atomic_t *v)
  * Atomically adds @a to @v, so long as @v was not already @u.
  * Returns the old value of @v.
  */
-static __always_inline int __arch_atomic_add_unless(atomic_t *v, int a, int u)
+static __always_inline int arch_atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c = arch_atomic_read(v);
 
arch/xtensa/include/asm/atomic.h +2 -2
index e7a23f2a519a..4188e56c06c9 100644
@@ -275,7 +275,7 @@ ATOMIC_OPS(xor)
 #define atomic_xchg(v, new) (xchg(&((v)->counter), new))
 
 /**
- * __atomic_add_unless - add unless the number is a given value
+ * atomic_fetch_add_unless - add unless the number is a given value
  * @v: pointer of type atomic_t
  * @a: the amount to add to v...
  * @u: ...unless v is equal to u.
@@ -283,7 +283,7 @@ ATOMIC_OPS(xor)
  * Atomically adds @a to @v, so long as it was not @u.
  * Returns the old value of @v.
  */
-static __inline__ int __atomic_add_unless(atomic_t *v, int a, int u)
+static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, old;
 	c = atomic_read(v);
drivers/block/rbd.c +1 -1
index 33b36fea1d73..b21cfacdbf03 100644
@@ -61,7 +61,7 @@ static int atomic_inc_return_safe(atomic_t *v)
 {
 	unsigned int counter;
 
-	counter = (unsigned int)__atomic_add_unless(v, 1, 0);
+	counter = (unsigned int)atomic_fetch_add_unless(v, 1, 0);
 	if (counter <= (unsigned int)INT_MAX)
 		return (int)counter;
 
drivers/infiniband/core/rdma_core.c +1 -1
index a6e904973ba8..475910ffbcb6 100644
@@ -121,7 +121,7 @@ static int uverbs_try_lock_object(struct ib_uobject *uobj, bool exclusive)
 	 * this lock.
 	 */
 	if (!exclusive)
-		return __atomic_add_unless(&uobj->usecnt, 1, -1) == -1 ?
+		return atomic_fetch_add_unless(&uobj->usecnt, 1, -1) == -1 ?
 			-EBUSY : 0;
 
 	/* lock is either WRITE or DESTROY - should be exclusive */
fs/afs/rxrpc.c +1 -1
index 08735948f15d..b35115b71b34 100644
@@ -648,7 +648,7 @@ static void afs_wake_up_async_call(struct sock *sk, struct rxrpc_call *rxcall,
 	trace_afs_notify_call(rxcall, call);
 	call->need_attention = true;
 
-	u = __atomic_add_unless(&call->usage, 1, 0);
+	u = atomic_fetch_add_unless(&call->usage, 1, 0);
 	if (u != 0) {
 		trace_afs_call(call, afs_call_trace_wake, u,
 			       atomic_read(&call->net->nr_outstanding_calls),
include/asm-generic/atomic-instrumented.h +2 -2
index ec07f23678ea..b8b14cc2df6c 100644
@@ -84,10 +84,10 @@ static __always_inline bool atomic64_try_cmpxchg(atomic64_t *v, s64 *old, s64 ne
 }
 #endif
 
-static __always_inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	kasan_check_write(v, sizeof(*v));
-	return __arch_atomic_add_unless(v, a, u);
+	return arch_atomic_fetch_add_unless(v, a, u);
 }
 
 
include/asm-generic/atomic.h +2 -2
index abe6dd9ca2a8..10051ed6d088 100644
@@ -221,8 +221,8 @@ static inline void atomic_dec(atomic_t *v)
 #define atomic_xchg(ptr, v)		(xchg(&(ptr)->counter, (v)))
 #define atomic_cmpxchg(v, old, new)	(cmpxchg(&((v)->counter), (old), (new)))
 
-#ifndef __atomic_add_unless
-static inline int __atomic_add_unless(atomic_t *v, int a, int u)
+#ifndef atomic_fetch_add_unless
+static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, old;
 	c = atomic_read(v);
include/linux/atomic.h +1 -1
index 8b276fd9a127..ec36a1375c9d 100644
@@ -530,7 +530,7 @@
  */
 static inline int atomic_add_unless(atomic_t *v, int a, int u)
 {
-	return __atomic_add_unless(v, a, u) != u;
+	return atomic_fetch_add_unless(v, a, u) != u;
 }
 
 /**
kernel/bpf/syscall.c +2 -2
index 016ef9025827..dd04ecb6ccaf 100644
@@ -538,7 +538,7 @@ static struct bpf_map *bpf_map_inc_not_zero(struct bpf_map *map,
 {
 	int refold;
 
-	refold = __atomic_add_unless(&map->refcnt, 1, 0);
+	refold = atomic_fetch_add_unless(&map->refcnt, 1, 0);
 
 	if (refold >= BPF_MAX_REFCNT) {
 		__bpf_map_put(map, false);
@@ -1115,7 +1115,7 @@ struct bpf_prog *bpf_prog_inc_not_zero(struct bpf_prog *prog)
 {
 	int refold;
 
-	refold = __atomic_add_unless(&prog->aux->refcnt, 1, 0);
+	refold = atomic_fetch_add_unless(&prog->aux->refcnt, 1, 0);
 
 	if (refold >= BPF_MAX_REFCNT) {
 		__bpf_prog_put(prog, false);
net/rxrpc/call_object.c +1 -1
index f6734d8cb01a..9486293fef5c 100644
@@ -415,7 +415,7 @@ void rxrpc_incoming_call(struct rxrpc_sock *rx,
 bool rxrpc_queue_call(struct rxrpc_call *call)
 {
 	const void *here = __builtin_return_address(0);
-	int n = __atomic_add_unless(&call->usage, 1, 0);
+	int n = atomic_fetch_add_unless(&call->usage, 1, 0);
 	if (n == 0)
 		return false;
 	if (rxrpc_queue_work(&call->processor))
net/rxrpc/conn_object.c +2 -2
index 4c77a78a252a..77440a356b14 100644
@@ -266,7 +266,7 @@ void rxrpc_kill_connection(struct rxrpc_connection *conn)
 bool rxrpc_queue_conn(struct rxrpc_connection *conn)
 {
 	const void *here = __builtin_return_address(0);
-	int n = __atomic_add_unless(&conn->usage, 1, 0);
+	int n = atomic_fetch_add_unless(&conn->usage, 1, 0);
 	if (n == 0)
 		return false;
 	if (rxrpc_queue_work(&conn->processor))
@@ -309,7 +309,7 @@ rxrpc_get_connection_maybe(struct rxrpc_connection *conn)
 	const void *here = __builtin_return_address(0);
 
 	if (conn) {
-		int n = __atomic_add_unless(&conn->usage, 1, 0);
+		int n = atomic_fetch_add_unless(&conn->usage, 1, 0);
 		if (n > 0)
 			trace_rxrpc_conn(conn, rxrpc_conn_got, n + 1, here);
 		else
net/rxrpc/local_object.c +1 -1
index b493e6b62740..777c3ed4cfc0 100644
@@ -305,7 +305,7 @@ struct rxrpc_local *rxrpc_get_local_maybe(struct rxrpc_local *local)
 	const void *here = __builtin_return_address(0);
 
 	if (local) {
-		int n = __atomic_add_unless(&local->usage, 1, 0);
+		int n = atomic_fetch_add_unless(&local->usage, 1, 0);
 		if (n > 0)
 			trace_rxrpc_local(local, rxrpc_local_got, n + 1, here);
 		else
net/rxrpc/peer_object.c +1 -1
index 1b7e8107b3ae..1cf3b408017a 100644
@@ -406,7 +406,7 @@ struct rxrpc_peer *rxrpc_get_peer_maybe(struct rxrpc_peer *peer)
 	const void *here = __builtin_return_address(0);
 
 	if (peer) {
-		int n = __atomic_add_unless(&peer->usage, 1, 0);
+		int n = atomic_fetch_add_unless(&peer->usage, 1, 0);
 		if (n > 0)
 			trace_rxrpc_peer(peer, rxrpc_peer_got, n + 1, here);
 		else
Signature · 1 line
2.11.0
Acked-by on v2 1/16
#2 lore raw

Re: [PATCHv2 01/16] atomics/treewide: s/__atomic_add_unless/atomic_fetch_add_unless/

Quoted 94 lines · On Tue, 29 May 2018 08:43:31 PDT (-0700), [email protected] wrote:

While __atomic_add_unless was originally intended as a building-block for atomic_add_unless, it's now used in a number of places around the kernel. It's the only common atomic operation named __atomic*(), rather than atomic_*(), and for consistency it would be better named atomic_fetch_add_unless().

This lack of consistency is slightly confusing, and gets in the way of scripting atomics. Given that, let's clean things up and promote it to an offical part of the atomics API, in the form of atomic_fetch_add_unless().

This patch converts definitions and uses over to the new name, including the instrumented version, using the following script:

----
git grep -w __atomic_add_unless | while read line; do
sed -i '{s/\<__atomic_add_unless\>/atomic_fetch_add_unless/}' "${line%%:*}"; done
git grep -w __arch_atomic_add_unless | while read line; do
sed -i '{s/\<__arch_atomic_add_unless\>/arch_atomic_fetch_add_unless/}' "${line%%:*}"; done
----

Note that we do not have atomic{64,_long}_fetch_add_unless(), which will be introduced by later patches.

There should be no functional change as a result of this patch.

---
 arch/alpha/include/asm/atomic.h           | 4 ++--
 arch/arc/include/asm/atomic.h             | 4 ++--
 arch/arm/include/asm/atomic.h             | 4 ++--
 arch/arm64/include/asm/atomic.h           | 2 +-
 arch/h8300/include/asm/atomic.h           | 2 +-
 arch/hexagon/include/asm/atomic.h         | 4 ++--
 arch/ia64/include/asm/atomic.h            | 2 +-
 arch/m68k/include/asm/atomic.h            | 2 +-
 arch/mips/include/asm/atomic.h            | 4 ++--
 arch/openrisc/include/asm/atomic.h        | 4 ++--
 arch/parisc/include/asm/atomic.h          | 4 ++--
 arch/powerpc/include/asm/atomic.h         | 8 ++++----
 arch/riscv/include/asm/atomic.h           | 4 ++--
 arch/s390/include/asm/atomic.h            | 2 +-
 arch/sh/include/asm/atomic.h              | 4 ++--
 arch/sparc/include/asm/atomic_32.h        | 2 +-
 arch/sparc/include/asm/atomic_64.h        | 2 +-
 arch/sparc/lib/atomic32.c                 | 4 ++--
 arch/x86/include/asm/atomic.h             | 4 ++--
 arch/xtensa/include/asm/atomic.h          | 4 ++--
 drivers/block/rbd.c                       | 2 +-
 drivers/infiniband/core/rdma_core.c       | 2 +-
 fs/afs/rxrpc.c                            | 2 +-
 include/asm-generic/atomic-instrumented.h | 4 ++--
 include/asm-generic/atomic.h              | 4 ++--
 include/linux/atomic.h                    | 2 +-
 kernel/bpf/syscall.c                      | 4 ++--
 net/rxrpc/call_object.c                   | 2 +-
 net/rxrpc/conn_object.c                   | 4 ++--
 net/rxrpc/local_object.c                  | 2 +-
 net/rxrpc/peer_object.c                   | 2 +-
 31 files changed, 50 insertions(+), 50 deletions(-)

[...]

arch/riscv/include/asm/atomic.h +2 -2
index 855115ace98c..739e810c857e 100644
@@ -332,7 +332,7 @@ ATOMIC_OP(dec_and_test, dec, ==, 0, 64)
 #undef ATOMIC_OP
 
 /* This is required to provide a full barrier on success. */
-static __always_inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
        int prev, rc;
 
@@ -381,7 +381,7 @@ static __always_inline int atomic64_add_unless(atomic64_t *v, long a, long u)
  */
 static __always_inline int atomic_inc_not_zero(atomic_t *v)
 {
-        return __atomic_add_unless(v, 1, 0);
+        return atomic_fetch_add_unless(v, 1, 0);
 }
 
 #ifndef CONFIG_GENERIC_ATOMIC64
arch/s390/include/asm/atomic.h +0 -0
index 4b55532f15c4..c2858cdd8c29 100644

arch/s390/include/asm/atomic.h static inline int atomic_cmpxchg(atomic_t *v, int old, int new) · line 96

Quoted 8 lines
arch/s390/include/asm/atomic.h +1 -1
@@ -90,7 +90,7 @@ static inline int atomic_cmpxchg(atomic_t *v, int old, int new)
 	return __atomic_cmpxchg(&v->counter, old, new);
 }
 
-static inline int __atomic_add_unless(atomic_t *v, int a, int u)
+static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	int c, old;
 	c = atomic_read(v);

Thanks!

author reviewer
#3 lore raw

PATCH v2 2/16 atomics/treewide: remove redundant atomic_inc_not_zero() definitions

In reply to Mark Rutland

When atomic_inc_not_zero(v) isn't defined, <linux/atomic.h> will define it as falling back to atomic_add_unless((v), 1, 0), so there's no need for arch code to do so.

There should be no functional change as a result of this patch.

---
 arch/arc/include/asm/atomic.h     | 2 --
 arch/hexagon/include/asm/atomic.h | 2 --
 arch/riscv/include/asm/atomic.h   | 9 ---------
 3 files changed, 13 deletions(-)
arch/arc/include/asm/atomic.h +0 -2
index 67121b5ff3a3..cecdf3403caf 100644
@@ -336,8 +336,6 @@ ATOMIC_OPS(xor, ^=, CTOP_INST_AXOR_DI_R2_R2_R3)
 	c;								\
 })
 
-#define atomic_inc_not_zero(v)		atomic_add_unless((v), 1, 0)
-
 #define atomic_inc(v)			atomic_add(1, v)
 #define atomic_dec(v)			atomic_sub(1, v)
 
arch/hexagon/include/asm/atomic.h +0 -2
index 287aa9f394f3..d2feeba93c44 100644
@@ -197,8 +197,6 @@ static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 	return __oldval;
 }
 
-#define atomic_inc_not_zero(v) atomic_add_unless((v), 1, 0)
-
 #define atomic_inc(v) atomic_add(1, (v))
 #define atomic_dec(v) atomic_sub(1, (v))
 
arch/riscv/include/asm/atomic.h +0 -9
index 739e810c857e..0e27e050ba14 100644
@@ -375,15 +375,6 @@ static __always_inline int atomic64_add_unless(atomic64_t *v, long a, long u)
 }
 #endif
 
-/*
- * The extra atomic operations that are constructed from one of the core
- * LR/SC-based operations above.
- */
-static __always_inline int atomic_inc_not_zero(atomic_t *v)
-{
-        return atomic_fetch_add_unless(v, 1, 0);
-}
-
 #ifndef CONFIG_GENERIC_ATOMIC64
 static __always_inline long atomic64_inc_not_zero(atomic64_t *v)
 {
Signature · 1 line
2.11.0
Acked-by on v2 2/16
#4 lore raw

Re: [PATCHv2 02/16] atomics/treewide: remove redundant atomic_inc_not_zero() definitions

Quoted 26 lines · On Tue, 29 May 2018 08:43:32 PDT (-0700), [email protected] wrote:

When atomic_inc_not_zero(v) isn't defined, <linux/atomic.h> will define it as falling back to atomic_add_unless((v), 1, 0), so there's no need for arch code to do so.

There should be no functional change as a result of this patch.

---
 arch/arc/include/asm/atomic.h     | 2 --
 arch/hexagon/include/asm/atomic.h | 2 --
 arch/riscv/include/asm/atomic.h   | 9 ---------
 3 files changed, 13 deletions(-)

[...]

arch/riscv/include/asm/atomic.h +0 -2
index 739e810c857e..0e27e050ba14 100644
@@ -375,15 +375,6 @@ static __always_inline int atomic64_add_unless(atomic64_t *v, long a, long u)
 }
 #endif
 
-/*
- * The extra atomic operations that are constructed from one of the core

arch/riscv/include/asm/atomic.h static __always_inline int atomic64_add_unless(atomic64_t *v, long a, long u) · line 380

Quoted 10 lines
arch/riscv/include/asm/atomic.h +0 -7
- * LR/SC-based operations above.
- */
-static __always_inline int atomic_inc_not_zero(atomic_t *v)
-{
-        return atomic_fetch_add_unless(v, 1, 0);
-}
-
 #ifndef CONFIG_GENERIC_ATOMIC64
 static __always_inline long atomic64_inc_not_zero(atomic64_t *v)
 {

Thanks!

author reviewer
#5 lore raw

PATCH v2 3/16 atomics/treewide: make atomic64_inc_not_zero() optional

In reply to Mark Rutland

We define a trivial fallback for atomic_inc_not_zero(), but don't do the same for atmic64_inc_not_zero(), leading most architectures to define the same boilerplate.

Let's add a fallback in <linux/atomic.h>, and remove the redundant implementations. Note that atomic64_add_unless() is always defined in <linux/atomic.h>, and promotes its arguments to the requisite types, so we need not do this explicitly.

There should be no functional change as a result of this patch.

---
 arch/alpha/include/asm/atomic.h           |  2 --
 arch/arc/include/asm/atomic.h             |  1 -
 arch/arm/include/asm/atomic.h             |  1 -
 arch/arm64/include/asm/atomic.h           |  2 --
 arch/ia64/include/asm/atomic.h            |  2 --
 arch/mips/include/asm/atomic.h            |  2 --
 arch/parisc/include/asm/atomic.h          |  2 --
 arch/powerpc/include/asm/atomic.h         |  1 +
 arch/riscv/include/asm/atomic.h           |  7 -------
 arch/s390/include/asm/atomic.h            |  1 -
 arch/sparc/include/asm/atomic_64.h        |  2 --
 arch/x86/include/asm/atomic64_32.h        |  2 +-
 arch/x86/include/asm/atomic64_64.h        |  2 --
 include/asm-generic/atomic-instrumented.h |  3 +++
 include/asm-generic/atomic64.h            |  1 -
 include/linux/atomic.h                    | 11 +++++++++++
 16 files changed, 16 insertions(+), 26 deletions(-)
arch/alpha/include/asm/atomic.h +0 -2
index 392b15a4dd4f..eb0f25e4c5dd 100644
@@ -296,8 +296,6 @@ static inline long atomic64_dec_if_positive(atomic64_t *v)
 	return old - 1;
 }
 
-#define atomic64_inc_not_zero(v) atomic64_add_unless((v), 1, 0)
-
 #define atomic_add_negative(a, v) (atomic_add_return((a), (v)) < 0)
 #define atomic64_add_negative(a, v) (atomic64_add_return((a), (v)) < 0)
 
arch/arc/include/asm/atomic.h +0 -1
index cecdf3403caf..1406825b5e7d 100644
@@ -603,7 +603,6 @@ static inline int atomic64_add_unless(atomic64_t *v, long long a, long long u)
 #define atomic64_dec(v)			atomic64_sub(1LL, (v))
 #define atomic64_dec_return(v)		atomic64_sub_return(1LL, (v))
 #define atomic64_dec_and_test(v)	(atomic64_dec_return((v)) == 0)
-#define atomic64_inc_not_zero(v)	atomic64_add_unless((v), 1LL, 0LL)
 
 #endif	/* !CONFIG_GENERIC_ATOMIC64 */
 
arch/arm/include/asm/atomic.h +0 -1
index 9d56d0727c9b..02f3894faa48 100644
@@ -534,7 +534,6 @@ static inline int atomic64_add_unless(atomic64_t *v, long long a, long long u)
 #define atomic64_dec(v)			atomic64_sub(1LL, (v))
 #define atomic64_dec_return_relaxed(v)	atomic64_sub_return_relaxed(1LL, (v))
 #define atomic64_dec_and_test(v)	(atomic64_dec_return((v)) == 0)
-#define atomic64_inc_not_zero(v)	atomic64_add_unless((v), 1LL, 0LL)
 
 #endif /* !CONFIG_GENERIC_ATOMIC64 */
 #endif
arch/arm64/include/asm/atomic.h +0 -2
index 264d20339f74..ad50412889c5 100644
@@ -204,7 +204,5 @@
 #define atomic64_add_unless(v, a, u)	(___atomic_add_unless(v, a, u, 64) != u)
 #define atomic64_andnot			atomic64_andnot
 
-#define atomic64_inc_not_zero(v)	atomic64_add_unless((v), 1, 0)
-
 #endif
 #endif
arch/ia64/include/asm/atomic.h +0 -2
index 9d2ddde5f9d5..93d48b823220 100644
@@ -246,8 +246,6 @@ static __inline__ long atomic64_add_unless(atomic64_t *v, long a, long u)
 	return c != (u);
 }
 
-#define atomic64_inc_not_zero(v) atomic64_add_unless((v), 1, 0)
-
 static __inline__ long atomic64_dec_if_positive(atomic64_t *v)
 {
 	long c, old, dec;
arch/mips/include/asm/atomic.h +0 -2
index 02fc1553cf9b..502e691c6393 100644
@@ -644,8 +644,6 @@ static __inline__ int atomic64_add_unless(atomic64_t *v, long a, long u)
 	return c != (u);
 }
 
-#define atomic64_inc_not_zero(v) atomic64_add_unless((v), 1, 0)
-
 #define atomic64_dec_return(v) atomic64_sub_return(1, (v))
 #define atomic64_inc_return(v) atomic64_add_return(1, (v))
 
arch/parisc/include/asm/atomic.h +0 -2
index 7748abced766..3fd0243bf405 100644
@@ -305,8 +305,6 @@ static __inline__ int atomic64_add_unless(atomic64_t *v, long a, long u)
 	return c != (u);
 }
 
-#define atomic64_inc_not_zero(v) atomic64_add_unless((v), 1, 0)
-
 /*
  * atomic64_dec_if_positive - decrement by 1 if old value positive
  * @v: pointer of type atomic_t
arch/powerpc/include/asm/atomic.h +1 -0
index 1483261080a1..e59620ee4f6b 100644
@@ -582,6 +582,7 @@ static __inline__ int atomic64_inc_not_zero(atomic64_t *v)
 
 	return t1 != 0;
 }
+#define atomic64_inc_not_zero(v) atomic64_inc_not_zero((v))
 
 #endif /* __powerpc64__ */
 
arch/riscv/include/asm/atomic.h +0 -7
index 0e27e050ba14..18259e90f57e 100644
@@ -375,13 +375,6 @@ static __always_inline int atomic64_add_unless(atomic64_t *v, long a, long u)
 }
 #endif
 
-#ifndef CONFIG_GENERIC_ATOMIC64
-static __always_inline long atomic64_inc_not_zero(atomic64_t *v)
-{
-        return atomic64_add_unless(v, 1, 0);
-}
-#endif
-
 /*
  * atomic_{cmp,}xchg is required to have exactly the same ordering semantics as
  * {cmp,}xchg and the operations that return, so they need a full barrier.
arch/s390/include/asm/atomic.h +0 -1
index c2858cdd8c29..66dac30a4fe1 100644
@@ -212,6 +212,5 @@ static inline long atomic64_dec_if_positive(atomic64_t *v)
 #define atomic64_dec(_v)		atomic64_sub(1, _v)
 #define atomic64_dec_return(_v)		atomic64_sub_return(1, _v)
 #define atomic64_dec_and_test(_v)	(atomic64_sub_return(1, _v) == 0)
-#define atomic64_inc_not_zero(v)	atomic64_add_unless((v), 1, 0)
 
 #endif /* __ARCH_S390_ATOMIC__  */
arch/sparc/include/asm/atomic_64.h +0 -2
index f416fd3d2708..07830a316464 100644
@@ -123,8 +123,6 @@ static inline long atomic64_add_unless(atomic64_t *v, long a, long u)
 	return c != (u);
 }
 
-#define atomic64_inc_not_zero(v) atomic64_add_unless((v), 1, 0)
-
 long atomic64_dec_if_positive(atomic64_t *v);
 
 #endif /* !(__ARCH_SPARC64_ATOMIC__) */
arch/x86/include/asm/atomic64_32.h +1 -1
index 92212bf0484f..2a33cc17801b 100644
@@ -295,7 +295,7 @@ static inline int arch_atomic64_add_unless(atomic64_t *v, long long a,
 	return (int)a;
 }
 
-
+#define arch_atomic64_inc_not_zero arch_atomic64_inc_not_zero
 static inline int arch_atomic64_inc_not_zero(atomic64_t *v)
 {
 	int r;
arch/x86/include/asm/atomic64_64.h +0 -2
index 6106b59d3260..6f95023894b7 100644
@@ -207,8 +207,6 @@ static inline bool arch_atomic64_add_unless(atomic64_t *v, long a, long u)
 	return true;
 }
 
-#define arch_atomic64_inc_not_zero(v) arch_atomic64_add_unless((v), 1, 0)
-
 /*
  * arch_atomic64_dec_if_positive - decrement by 1 if old value positive
  * @v: pointer of type atomic_t
include/asm-generic/atomic-instrumented.h +3 -0
index b8b14cc2df6c..52b7dcb41769 100644
@@ -205,11 +205,14 @@ static __always_inline s64 atomic64_dec_return(atomic64_t *v)
 	return arch_atomic64_dec_return(v);
 }
 
+#ifdef arch_atomic64_inc_not_zero
+#define atomic64_inc_not_zero atomic64_inc_not_zero
 static __always_inline s64 atomic64_inc_not_zero(atomic64_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic64_inc_not_zero(v);
 }
+#endif
 
 static __always_inline s64 atomic64_dec_if_positive(atomic64_t *v)
 {
include/asm-generic/atomic64.h +0 -1
index 8d28eb010d0d..e9a94b7a4d0a 100644
@@ -62,6 +62,5 @@ extern int	 atomic64_add_unless(atomic64_t *v, long long a, long long u);
 #define atomic64_dec(v)			atomic64_sub(1LL, (v))
 #define atomic64_dec_return(v)		atomic64_sub_return(1LL, (v))
 #define atomic64_dec_and_test(v)	(atomic64_dec_return((v)) == 0)
-#define atomic64_inc_not_zero(v) 	atomic64_add_unless((v), 1LL, 0LL)
 
 #endif  /*  _ASM_GENERIC_ATOMIC64_H  */
include/linux/atomic.h +11 -0
index ec36a1375c9d..e83f12ec7f2d 100644
@@ -1048,6 +1048,17 @@ static inline int atomic_dec_if_positive(atomic_t *v)
 #define atomic64_try_cmpxchg_release	atomic64_try_cmpxchg
 #endif /* atomic64_try_cmpxchg */
 
+/**
+ * atomic64_inc_not_zero - increment unless the number is zero
+ * @v: pointer of type atomic64_t
+ *
+ * Atomically increments @v by 1, so long as @v is non-zero.
+ * Returns non-zero if @v was non-zero, and zero otherwise.
+ */
+#ifndef atomic64_inc_not_zero
+#define atomic64_inc_not_zero(v)	atomic64_add_unless((v), 1, 0)
+#endif
+
 #ifndef atomic64_andnot
 static inline void atomic64_andnot(long long i, atomic64_t *v)
 {
Signature · 1 line
2.11.0
Acked-by on v2 3/16
#6 lore raw

Re: [PATCHv2 03/16] atomics/treewide: make atomic64_inc_not_zero() optional

Quoted 3 lines · On Tue, 29 May 2018 08:43:33 PDT (-0700), [email protected] wrote:

We define a trivial fallback for atomic_inc_not_zero(), but don't do the same for atmic64_inc_not_zero(), leading most architectures to define the same boilerplate.

atmic64

Quoted 38 lines

Let's add a fallback in <linux/atomic.h>, and remove the redundant implementations. Note that atomic64_add_unless() is always defined in <linux/atomic.h>, and promotes its arguments to the requisite types, so we need not do this explicitly.

There should be no functional change as a result of this patch.

---
 arch/alpha/include/asm/atomic.h           |  2 --
 arch/arc/include/asm/atomic.h             |  1 -
 arch/arm/include/asm/atomic.h             |  1 -
 arch/arm64/include/asm/atomic.h           |  2 --
 arch/ia64/include/asm/atomic.h            |  2 --
 arch/mips/include/asm/atomic.h            |  2 --
 arch/parisc/include/asm/atomic.h          |  2 --
 arch/powerpc/include/asm/atomic.h         |  1 +
 arch/riscv/include/asm/atomic.h           |  7 -------
 arch/s390/include/asm/atomic.h            |  1 -
 arch/sparc/include/asm/atomic_64.h        |  2 --
 arch/x86/include/asm/atomic64_32.h        |  2 +-
 arch/x86/include/asm/atomic64_64.h        |  2 --
 include/asm-generic/atomic-instrumented.h |  3 +++
 include/asm-generic/atomic64.h            |  1 -
 include/linux/atomic.h                    | 11 +++++++++++
 16 files changed, 16 insertions(+), 26 deletions(-)

[...]

arch/riscv/include/asm/atomic.h +0 -0
index 0e27e050ba14..18259e90f57e 100644
@@ -375,13 +375,6 @@ static __always_inline int atomic64_add_unless(atomic64_t *v, long a, long u)
 }
 #endif
 

arch/riscv/include/asm/atomic.h static __always_inline int atomic64_add_unless(atomic64_t *v, long a, long u) · line 380

Quoted 10 lines
arch/riscv/include/asm/atomic.h +0 -7
-#ifndef CONFIG_GENERIC_ATOMIC64
-static __always_inline long atomic64_inc_not_zero(atomic64_t *v)
-{
-        return atomic64_add_unless(v, 1, 0);
-}
-#endif
-
 /*
  * atomic_{cmp,}xchg is required to have exactly the same ordering semantics as
  * {cmp,}xchg and the operations that return, so they need a full barrier.

Thanks!

author reviewer
#7 lore raw

Re: [PATCHv2 03/16] atomics/treewide: make atomic64_inc_not_zero() optional

Quoted 6 lines · Palmer Dabbelt
Quoted 3 lines · On Tue, 29 May 2018 08:43:33 PDT (-0700), [email protected] wrote:

We define a trivial fallback for atomic_inc_not_zero(), but don't do the same for atmic64_inc_not_zero(), leading most architectures to define the same boilerplate.

atmic64

Cheers for the spot; fixed now!

Mark.

author reviewer
#8 lore raw

PATCH v2 4/16 atomics/treewide: make atomic_fetch_add_unless() optional

In reply to Mark Rutland

Several architectures these have a near-identical implementation based on atomic_read() and atomic_cmpxchg() that we can instead define in <linux/atomic.h>, so let's do so, using something close to the existing x86 implementation with try_cmpxchg().

Where an architecture provides its own atomic_fetch_add_unless(), it must define a preprocessor symbol for it. The instrumented atomics are updated accordingly.

Note that arch/arc's existing atomic_fetch_add_unless() had redundant barriers, as these are already present in its atomic_cmpxchg() implementation.

There should be no functional change as a result of this patch.

---
 arch/alpha/include/asm/atomic.h           |  2 +-
 arch/arc/include/asm/atomic.h             | 28 ----------------------------
 arch/arm/include/asm/atomic.h             | 11 +----------
 arch/arm64/include/asm/atomic.h           |  1 -
 arch/h8300/include/asm/atomic.h           |  1 +
 arch/hexagon/include/asm/atomic.h         |  1 +
 arch/ia64/include/asm/atomic.h            | 16 ----------------
 arch/m68k/include/asm/atomic.h            | 15 ---------------
 arch/mips/include/asm/atomic.h            | 24 ------------------------
 arch/parisc/include/asm/atomic.h          | 24 ------------------------
 arch/powerpc/include/asm/atomic.h         |  1 +
 arch/riscv/include/asm/atomic.h           |  1 +
 arch/s390/include/asm/atomic.h            | 15 ---------------
 arch/sh/include/asm/atomic.h              | 25 -------------------------
 arch/sparc/include/asm/atomic_32.h        |  2 ++
 arch/sparc/include/asm/atomic_64.h        | 15 ---------------
 arch/x86/include/asm/atomic.h             | 21 ---------------------
 arch/xtensa/include/asm/atomic.h          | 24 ------------------------
 include/asm-generic/atomic-instrumented.h |  4 +++-
 include/asm-generic/atomic.h              | 11 -----------
 include/linux/atomic.h                    | 23 +++++++++++++++++++++++
 21 files changed, 34 insertions(+), 231 deletions(-)
arch/alpha/include/asm/atomic.h +1 -1
index eb0f25e4c5dd..4a800a3424a3 100644
@@ -235,7 +235,7 @@ static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 	smp_mb();
 	return old;
 }
-
+#define atomic_fetch_add_unless atomic_fetch_add_unless
 
 /**
  * atomic64_add_unless - add unless the number is a given value
arch/arc/include/asm/atomic.h +0 -28
index 1406825b5e7d..60da80481c5d 100644
@@ -308,34 +308,6 @@ ATOMIC_OPS(xor, ^=, CTOP_INST_AXOR_DI_R2_R2_R3)
 #undef ATOMIC_OP_RETURN
 #undef ATOMIC_OP
 
-/**
- * atomic_fetch_add_unless - add unless the number is a given value
- * @v: pointer of type atomic_t
- * @a: the amount to add to v...
- * @u: ...unless v is equal to u.
- *
- * Atomically adds @a to @v, so long as it was not @u.
- * Returns the old value of @v
- */
-#define atomic_fetch_add_unless(v, a, u)					\
-({									\
-	int c, old;							\
-									\
-	/*								\
-	 * Explicit full memory barrier needed before/after as		\
-	 * LLOCK/SCOND thmeselves don't provide any such semantics	\
-	 */								\
-	smp_mb();							\
-									\
-	c = atomic_read(v);						\
-	while (c != (u) && (old = atomic_cmpxchg((v), c, c + (a))) != c)\
-		c = old;						\
-									\
-	smp_mb();							\
-									\
-	c;								\
-})
-
 #define atomic_inc(v)			atomic_add(1, v)
 #define atomic_dec(v)			atomic_sub(1, v)
 
arch/arm/include/asm/atomic.h +1 -10
index 02f3894faa48..74460aa00fa0 100644
@@ -156,6 +156,7 @@ static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 
 	return oldval;
 }
+#define atomic_fetch_add_unless		atomic_fetch_add_unless
 
 #else /* ARM_ARCH_6 */
 
@@ -215,16 +216,6 @@ static inline int atomic_cmpxchg(atomic_t *v, int old, int new)
 	return ret;
 }
 
-static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
-{
-	int c, old;
-
-	c = atomic_read(v);
-	while (c != u && (old = atomic_cmpxchg((v), c, c + a)) != c)
-		c = old;
-	return c;
-}
-
 #endif /* __LINUX_ARM_ARCH__ */
 
 #define ATOMIC_OPS(op, c_op, asm_op)					\
arch/arm64/include/asm/atomic.h +0 -1
index ad50412889c5..22c8c43d6689 100644
@@ -125,7 +125,6 @@
 #define atomic_dec_and_test(v)		(atomic_dec_return(v) == 0)
 #define atomic_sub_and_test(i, v)	(atomic_sub_return((i), (v)) == 0)
 #define atomic_add_negative(i, v)	(atomic_add_return((i), (v)) < 0)
-#define atomic_fetch_add_unless(v, a, u)	___atomic_add_unless(v, a, u,)
 #define atomic_andnot			atomic_andnot
 
 /*
arch/h8300/include/asm/atomic.h +1 -0
index 4465cfc30a3a..250bb3baac23 100644
@@ -106,5 +106,6 @@ static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 	arch_local_irq_restore(flags);
 	return ret;
 }
+#define atomic_fetch_add_unless		atomic_fetch_add_unless
 
 #endif /* __ARCH_H8300_ATOMIC __ */
arch/hexagon/include/asm/atomic.h +1 -0
index d2feeba93c44..86c67e9adbfa 100644
@@ -196,6 +196,7 @@ static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 	);
 	return __oldval;
 }
+#define atomic_fetch_add_unless atomic_fetch_add_unless
 
 #define atomic_inc(v) atomic_add(1, (v))
 #define atomic_dec(v) atomic_sub(1, (v))
arch/ia64/include/asm/atomic.h +0 -16
index 93d48b823220..cfe44086338e 100644
@@ -215,22 +215,6 @@ ATOMIC64_FETCH_OP(xor, ^)
 	(cmpxchg(&((v)->counter), old, new))
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), new))
 
-static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
-{
-	int c, old;
-	c = atomic_read(v);
-	for (;;) {
-		if (unlikely(c == (u)))
-			break;
-		old = atomic_cmpxchg((v), c, c + (a));
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c;
-}
-
-
 static __inline__ long atomic64_add_unless(atomic64_t *v, long a, long u)
 {
 	long c, old;
arch/m68k/include/asm/atomic.h +0 -15
index 8022d9ea1213..596882cda224 100644
@@ -211,19 +211,4 @@ static inline int atomic_add_negative(int i, atomic_t *v)
 	return c != 0;
 }
 
-static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
-{
-	int c, old;
-	c = atomic_read(v);
-	for (;;) {
-		if (unlikely(c == (u)))
-			break;
-		old = atomic_cmpxchg((v), c, c + (a));
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c;
-}
-
 #endif /* __ARCH_M68K_ATOMIC __ */
arch/mips/include/asm/atomic.h +0 -24
index 502e691c6393..794734e730d9 100644
@@ -274,30 +274,6 @@ static __inline__ int atomic_sub_if_positive(int i, atomic_t * v)
 #define atomic_cmpxchg(v, o, n) (cmpxchg(&((v)->counter), (o), (n)))
 #define atomic_xchg(v, new) (xchg(&((v)->counter), (new)))
 
-/**
- * atomic_fetch_add_unless - add unless the number is a given value
- * @v: pointer of type atomic_t
- * @a: the amount to add to v...
- * @u: ...unless v is equal to u.
- *
- * Atomically adds @a to @v, so long as it was not @u.
- * Returns the old value of @v.
- */
-static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
-{
-	int c, old;
-	c = atomic_read(v);
-	for (;;) {
-		if (unlikely(c == (u)))
-			break;
-		old = atomic_cmpxchg((v), c, c + (a));
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c;
-}
-
 #define atomic_dec_return(v) atomic_sub_return(1, (v))
 #define atomic_inc_return(v) atomic_add_return(1, (v))
 
arch/parisc/include/asm/atomic.h +0 -24
index 3fd0243bf405..b2b6261d05e7 100644
@@ -77,30 +77,6 @@ static __inline__ int atomic_read(const atomic_t *v)
 #define atomic_cmpxchg(v, o, n) (cmpxchg(&((v)->counter), (o), (n)))
 #define atomic_xchg(v, new) (xchg(&((v)->counter), new))
 
-/**
- * atomic_fetch_add_unless - add unless the number is a given value
- * @v: pointer of type atomic_t
- * @a: the amount to add to v...
- * @u: ...unless v is equal to u.
- *
- * Atomically adds @a to @v, so long as it was not @u.
- * Returns the old value of @v.
- */
-static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
-{
-	int c, old;
-	c = atomic_read(v);
-	for (;;) {
-		if (unlikely(c == (u)))
-			break;
-		old = atomic_cmpxchg((v), c, c + (a));
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c;
-}
-
 #define ATOMIC_OP(op, c_op)						\
 static __inline__ void atomic_##op(int i, atomic_t *v)			\
 {									\
arch/powerpc/include/asm/atomic.h +1 -0
index e59620ee4f6b..b5646c079c16 100644
@@ -248,6 +248,7 @@ static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 
 	return t;
 }
+#define atomic_fetch_add_unless atomic_fetch_add_unless
 
 /**
  * atomic_inc_not_zero - increment unless the number is zero
arch/riscv/include/asm/atomic.h +1 -0
index 18259e90f57e..5f161daefcd2 100644
@@ -349,6 +349,7 @@ static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 		: "memory");
 	return prev;
 }
+#define atomic_fetch_add_unless atomic_fetch_add_unless
 
 #ifndef CONFIG_GENERIC_ATOMIC64
 static __always_inline long __atomic64_add_unless(atomic64_t *v, long a, long u)
arch/s390/include/asm/atomic.h +0 -15
index 66dac30a4fe1..26c6b713a7a3 100644
@@ -90,21 +90,6 @@ static inline int atomic_cmpxchg(atomic_t *v, int old, int new)
 	return __atomic_cmpxchg(&v->counter, old, new);
 }
 
-static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
-{
-	int c, old;
-	c = atomic_read(v);
-	for (;;) {
-		if (unlikely(c == u))
-			break;
-		old = atomic_cmpxchg(v, c, c + a);
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c;
-}
-
 #define ATOMIC64_INIT(i)  { (i) }
 
 static inline long atomic64_read(const atomic64_t *v)
arch/sh/include/asm/atomic.h +0 -25
index ef45931ebac5..422fac764ca1 100644
@@ -45,31 +45,6 @@
 #define atomic_xchg(v, new)		(xchg(&((v)->counter), new))
 #define atomic_cmpxchg(v, o, n)		(cmpxchg(&((v)->counter), (o), (n)))
 
-/**
- * atomic_fetch_add_unless - add unless the number is a given value
- * @v: pointer of type atomic_t
- * @a: the amount to add to v...
- * @u: ...unless v is equal to u.
- *
- * Atomically adds @a to @v, so long as it was not @u.
- * Returns the old value of @v.
- */
-static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
-{
-	int c, old;
-	c = atomic_read(v);
-	for (;;) {
-		if (unlikely(c == (u)))
-			break;
-		old = atomic_cmpxchg((v), c, c + (a));
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-
-	return c;
-}
-
 #endif /* CONFIG_CPU_J2 */
 
 #endif /* __ASM_SH_ATOMIC_H */
arch/sparc/include/asm/atomic_32.h +2 -0
index a58f4b43bcc7..9d7a15acc0c5 100644
@@ -30,6 +30,8 @@ int atomic_xchg(atomic_t *, int);
 int atomic_fetch_add_unless(atomic_t *, int, int);
 void atomic_set(atomic_t *, int);
 
+#define atomic_fetch_add_unless	atomic_fetch_add_unless
+
 #define atomic_set_release(v, i)	atomic_set((v), (i))
 
 #define atomic_read(v)          READ_ONCE((v)->counter)
arch/sparc/include/asm/atomic_64.h +0 -15
index 07830a316464..e4f1c93db31f 100644
@@ -89,21 +89,6 @@ static inline int atomic_xchg(atomic_t *v, int new)
 	return xchg(&v->counter, new);
 }
 
-static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
-{
-	int c, old;
-	c = atomic_read(v);
-	for (;;) {
-		if (unlikely(c == (u)))
-			break;
-		old = atomic_cmpxchg((v), c, c + (a));
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c;
-}
-
 #define atomic64_cmpxchg(v, o, n) \
 	((__typeof__((v)->counter))cmpxchg(&((v)->counter), (o), (n)))
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), new))
arch/x86/include/asm/atomic.h +0 -21
index 84ed0bd76aef..616327ac9d39 100644
@@ -253,27 +253,6 @@ static inline int arch_atomic_fetch_xor(int i, atomic_t *v)
 	return val;
 }
 
-/**
- * arch_atomic_fetch_add_unless - add unless the number is already a given value
- * @v: pointer of type atomic_t
- * @a: the amount to add to v...
- * @u: ...unless v is equal to u.
- *
- * Atomically adds @a to @v, so long as @v was not already @u.
- * Returns the old value of @v.
- */
-static __always_inline int arch_atomic_fetch_add_unless(atomic_t *v, int a, int u)
-{
-	int c = arch_atomic_read(v);
-
-	do {
-		if (unlikely(c == u))
-			break;
-	} while (!arch_atomic_try_cmpxchg(v, &c, c + a));
-
-	return c;
-}
-
 #ifdef CONFIG_X86_32
 # include <asm/atomic64_32.h>
 #else
arch/xtensa/include/asm/atomic.h +0 -24
index 4188e56c06c9..f4c9f82c40c6 100644
@@ -274,30 +274,6 @@ ATOMIC_OPS(xor)
 #define atomic_cmpxchg(v, o, n) ((int)cmpxchg(&((v)->counter), (o), (n)))
 #define atomic_xchg(v, new) (xchg(&((v)->counter), new))
 
-/**
- * atomic_fetch_add_unless - add unless the number is a given value
- * @v: pointer of type atomic_t
- * @a: the amount to add to v...
- * @u: ...unless v is equal to u.
- *
- * Atomically adds @a to @v, so long as it was not @u.
- * Returns the old value of @v.
- */
-static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
-{
-	int c, old;
-	c = atomic_read(v);
-	for (;;) {
-		if (unlikely(c == (u)))
-			break;
-		old = atomic_cmpxchg((v), c, c + (a));
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c;
-}
-
 #endif /* __KERNEL__ */
 
 #endif /* _XTENSA_ATOMIC_H */
include/asm-generic/atomic-instrumented.h +3 -1
index 52b7dcb41769..6e0818c182e2 100644
@@ -84,12 +84,14 @@ static __always_inline bool atomic64_try_cmpxchg(atomic64_t *v, s64 *old, s64 ne
 }
 #endif
 
+#ifdef arch_atomic_fetch_add_unless
+#define atomic_fetch_add_unless atomic_fetch_add_unless
 static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic_fetch_add_unless(v, a, u);
 }
-
+#endif
 
 static __always_inline bool atomic64_add_unless(atomic64_t *v, s64 a, s64 u)
 {
include/asm-generic/atomic.h +0 -11
index 10051ed6d088..757e45821220 100644
@@ -221,15 +221,4 @@ static inline void atomic_dec(atomic_t *v)
 #define atomic_xchg(ptr, v)		(xchg(&(ptr)->counter, (v)))
 #define atomic_cmpxchg(v, old, new)	(cmpxchg(&((v)->counter), (old), (new)))
 
-#ifndef atomic_fetch_add_unless
-static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
-{
-	int c, old;
-	c = atomic_read(v);
-	while (c != u && (old = atomic_cmpxchg(v, c, c + a)) != c)
-		c = old;
-	return c;
-}
-#endif
-
 #endif /* __ASM_GENERIC_ATOMIC_H */
include/linux/atomic.h +23 -0
index e83f12ec7f2d..1105c0b37f27 100644
@@ -520,6 +520,29 @@
 #endif /* xchg_relaxed */
 
 /**
+ * atomic_fetch_add_unless - add unless the number is already a given value
+ * @v: pointer of type atomic_t
+ * @a: the amount to add to v...
+ * @u: ...unless v is equal to u.
+ *
+ * Atomically adds @a to @v, so long as @v was not already @u.
+ * Returns original value of @v
+ */
+#ifndef atomic_fetch_add_unless
+static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
+{
+	int c = atomic_read(v);
+
+	do {
+		if (unlikely(c == u))
+			break;
+	} while (!atomic_try_cmpxchg(v, &c, c + a));
+
+	return c;
+}
+#endif
+
+/**
  * atomic_add_unless - add unless the number is already a given value
  * @v: pointer of type atomic_t
  * @a: the amount to add to v...
Signature · 1 line
2.11.0
Acked-by on v2 4/16
#9 lore raw

Re: [PATCHv2 04/16] atomics/treewide: make atomic_fetch_add_unless() optional

Quoted 51 lines · On Tue, 29 May 2018 08:43:34 PDT (-0700), [email protected] wrote:

Several architectures these have a near-identical implementation based on atomic_read() and atomic_cmpxchg() that we can instead define in <linux/atomic.h>, so let's do so, using something close to the existing x86 implementation with try_cmpxchg().

Where an architecture provides its own atomic_fetch_add_unless(), it must define a preprocessor symbol for it. The instrumented atomics are updated accordingly.

Note that arch/arc's existing atomic_fetch_add_unless() had redundant barriers, as these are already present in its atomic_cmpxchg() implementation.

There should be no functional change as a result of this patch.

---
 arch/alpha/include/asm/atomic.h           |  2 +-
 arch/arc/include/asm/atomic.h             | 28 ----------------------------
 arch/arm/include/asm/atomic.h             | 11 +----------
 arch/arm64/include/asm/atomic.h           |  1 -
 arch/h8300/include/asm/atomic.h           |  1 +
 arch/hexagon/include/asm/atomic.h         |  1 +
 arch/ia64/include/asm/atomic.h            | 16 ----------------
 arch/m68k/include/asm/atomic.h            | 15 ---------------
 arch/mips/include/asm/atomic.h            | 24 ------------------------
 arch/parisc/include/asm/atomic.h          | 24 ------------------------
 arch/powerpc/include/asm/atomic.h         |  1 +
 arch/riscv/include/asm/atomic.h           |  1 +
 arch/s390/include/asm/atomic.h            | 15 ---------------
 arch/sh/include/asm/atomic.h              | 25 -------------------------
 arch/sparc/include/asm/atomic_32.h        |  2 ++
 arch/sparc/include/asm/atomic_64.h        | 15 ---------------
 arch/x86/include/asm/atomic.h             | 21 ---------------------
 arch/xtensa/include/asm/atomic.h          | 24 ------------------------
 include/asm-generic/atomic-instrumented.h |  4 +++-
 include/asm-generic/atomic.h              | 11 -----------
 include/linux/atomic.h                    | 23 +++++++++++++++++++++++
 21 files changed, 34 insertions(+), 231 deletions(-)

[...]

arch/riscv/include/asm/atomic.h +0 -0
index 18259e90f57e..5f161daefcd2 100644

arch/riscv/include/asm/atomic.h static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u) · line 355

Quoted 7 lines
arch/riscv/include/asm/atomic.h +1 -0
@@ -349,6 +349,7 @@ static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 		: "memory");
 	return prev;
 }
+#define atomic_fetch_add_unless atomic_fetch_add_unless
 
 #ifndef CONFIG_GENERIC_ATOMIC64
 static __always_inline long __atomic64_add_unless(atomic64_t *v, long a, long u)

Thanks!

author reviewer
#10 lore raw

PATCH v2 5/16 atomics: prepare for atomic64_fetch_add_unless()

In reply to Mark Rutland

Currently architecture must implement atomic_fetch_add_unless(), with common code providing atomic_add_unless(). Architectures must also implement atmic64_add_unless() directly, with no corresponding atomic64_fetch_add_unless().

This divergenece is unfortunate, and means that the APIs for atomic_t, atomic64_t, and atomic_long_t differ.

In preparation for unifying things, with architectures providing atomic64_fetch_add_unless, this patch adds a generic atomic64_add_unless() which will use atomic64_fetch_add_unless(). The instrumented atomics are updated to take this case into account.

There should be no functional change as a result of this patch.

Cc Boqun Feng <[email protected]>
Cc Will Deacon <[email protected]>
Cc Arnd Bergmann <[email protected]>
Cc Richard Henderson <[email protected]>
Cc Ivan Kokshaysky <[email protected]>
Cc Matt Turner <[email protected]>
Cc Vineet Gupta <[email protected]>
Cc Russell King <[email protected]>
Cc Benjamin Herrenschmidt <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Michael Ellerman <[email protected]>
Cc Palmer Dabbelt <[email protected]>
Cc Albert Ou <[email protected]>
---
 include/asm-generic/atomic-instrumented.h |  9 +++++++++
 include/linux/atomic.h                    | 16 ++++++++++++++++
 2 files changed, 25 insertions(+)
include/asm-generic/atomic-instrumented.h +9 -0
index 6e0818c182e2..e22d7e5f4ce7 100644
@@ -93,11 +93,20 @@ static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 }
 #endif
 
+#ifdef arch_atomic64_fetch_add_unless
+#define atomic64_fetch_add_unless atomic64_fetch_add_unless
+static __always_inline int atomic64_fetch_add_unless(atomic64_t *v, s64 a, s64 u)
+{
+	kasan_check_write(v, sizeof(*v));
+	return arch_atomic64_fetch_add_unless(v, a, u);
+}
+#else
 static __always_inline bool atomic64_add_unless(atomic64_t *v, s64 a, s64 u)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic64_add_unless(v, a, u);
 }
+#endif
 
 static __always_inline void atomic_inc(atomic_t *v)
 {
include/linux/atomic.h +16 -0
index 1105c0b37f27..8d93209052e1 100644
@@ -1072,6 +1072,22 @@ static inline int atomic_dec_if_positive(atomic_t *v)
 #endif /* atomic64_try_cmpxchg */
 
 /**
+ * atomic64_add_unless - add unless the number is already a given value
+ * @v: pointer of type atomic_t
+ * @a: the amount to add to v...
+ * @u: ...unless v is equal to u.
+ *
+ * Atomically adds @a to @v, so long as @v was not already @u.
+ * Returns non-zero if @v was not @u, and zero otherwise.
+ */
+#ifdef atomic64_fetch_add_unless
+static inline int atomic64_add_unless(atomic64_t *v, long long a, long long u)
+{
+	return atomic64_fetch_add_unless(v, a, u) != u;
+}
+#endif
+
+/**
  * atomic64_inc_not_zero - increment unless the number is zero
  * @v: pointer of type atomic64_t
  *
Signature · 1 line
2.11.0
maintainer
#11 lore raw

Re: [PATCHv2 05/16] atomics: prepare for atomic64_fetch_add_unless()

Quoted 8 lines · Mark Rutland

/**
+ * atomic64_add_unless - add unless the number is already a given value
+ * @v: pointer of type atomic_t
+ * @a: the amount to add to v...
+ * @u: ...unless v is equal to u.
+ *
+ * Atomically adds @a to @v, so long as @v was not already @u.
+ * Returns non-zero if @v was not @u, and zero otherwise.

I always get confused by that wording; would something like: "Returns true if the addition was done" not be more clear?

include/linux/atomic.h static inline int atomic_dec_if_positive(atomic_t *v) · line 1084

Quoted 3 lines

+ */

include/linux/atomic.h +2 -0
+#ifdef atomic64_fetch_add_unless
+static inline int atomic64_add_unless(atomic64_t *v, long long a, long long u)

Do we want to make that a "bool' return?

Quoted 4 lines
include/linux/atomic.h +4 -0
+{
+	return atomic64_fetch_add_unless(v, a, u) != u;
+}
+#endif
author reviewer
#12 lore raw

Re: [PATCHv2 05/16] atomics: prepare for atomic64_fetch_add_unless()

Quoted 10 lines · Peter Zijlstra
Quoted 8 lines · Mark Rutland

/**
+ * atomic64_add_unless - add unless the number is already a given value
+ * @v: pointer of type atomic_t
+ * @a: the amount to add to v...
+ * @u: ...unless v is equal to u.
+ *
+ * Atomically adds @a to @v, so long as @v was not already @u.
+ * Returns non-zero if @v was not @u, and zero otherwise.

Quoted 2 lines

I always get confused by that wording; would something like: "Returns true if the addition was done" not be more clear?

Sounds clearer to me; I just stole the wording from the existing atomic_add_unless().

I guess you'll want similar for the conditional inc/dec ops, e.g.

/**
* atomic_inc_not_zero - increment unless the number is zero
* @v: pointer of type atomic_t
*
* Atomically increments @v by 1, so long as @v is non-zero.
* Returns non-zero if @v was non-zero, and zero otherwise. */

Quoted 5 lines

include/linux/atomic.h static inline int atomic_dec_if_positive(atomic_t *v) · line 1084

Quoted 3 lines

+ */

include/linux/atomic.h +2 -0
+#ifdef atomic64_fetch_add_unless
+static inline int atomic64_add_unless(atomic64_t *v, long long a, long long u)

Do we want to make that a "bool' return?

I think so -- that's what the instrumented wrappers (and x86) do today anyhow, and what I ended up using for the generated headers.

I'll spin a prep patch cleaning up the existing fallbacks in <linux/atomic.h>, along with the comment fixup above, then rework the additions likewise.

Thanks,
Mark.

#13 lore raw

Re: [PATCHv2 05/16] atomics: prepare for atomic64_fetch_add_unless()

Quoted 19 lines · Mark Rutland
Quoted 10 lines · Peter Zijlstra
Quoted 8 lines · Mark Rutland

/**
+ * atomic64_add_unless - add unless the number is already a given value
+ * @v: pointer of type atomic_t
+ * @a: the amount to add to v...
+ * @u: ...unless v is equal to u.
+ *
+ * Atomically adds @a to @v, so long as @v was not already @u.
+ * Returns non-zero if @v was not @u, and zero otherwise.

Quoted 2 lines

I always get confused by that wording; would something like: "Returns true if the addition was done" not be more clear?

Sounds clearer to me; I just stole the wording from the existing atomic_add_unless().

I guess you'll want similar for the conditional inc/dec ops, e.g.

Quoted 7 lines

/**
* atomic_inc_not_zero - increment unless the number is zero
* @v: pointer of type atomic_t
*
* Atomically increments @v by 1, so long as @v is non-zero.
* Returns non-zero if @v was non-zero, and zero otherwise. */

If we're bike-shedding .. :)

I think "so long as" is overly wordy and not helpful. It'd be clearer just as:

* Atomically increments @v by 1, if @v is non-zero.

cheers

author reviewer
#14 lore raw

Re: [PATCHv2 05/16] atomics: prepare for atomic64_fetch_add_unless()

Quoted 28 lines · Michael Ellerman
Quoted 19 lines · Mark Rutland
Quoted 10 lines · Peter Zijlstra
Quoted 8 lines · Mark Rutland

/**
+ * atomic64_add_unless - add unless the number is already a given value
+ * @v: pointer of type atomic_t
+ * @a: the amount to add to v...
+ * @u: ...unless v is equal to u.
+ *
+ * Atomically adds @a to @v, so long as @v was not already @u.
+ * Returns non-zero if @v was not @u, and zero otherwise.

Quoted 2 lines

I always get confused by that wording; would something like: "Returns true if the addition was done" not be more clear?

Sounds clearer to me; I just stole the wording from the existing atomic_add_unless().

I guess you'll want similar for the conditional inc/dec ops, e.g.

Quoted 7 lines

/**
* atomic_inc_not_zero - increment unless the number is zero
* @v: pointer of type atomic_t
*
* Atomically increments @v by 1, so long as @v is non-zero.
* Returns non-zero if @v was non-zero, and zero otherwise. */

Quoted 6 lines

If we're bike-shedding .. :)

I think "so long as" is overly wordy and not helpful. It'd be clearer just as:

* Atomically increments @v by 1, if @v is non-zero.

I agree; done.

Mark.

#15 lore raw

Re: [PATCHv2 05/16] atomics: prepare for atomic64_fetch_add_unless()

Quoted 29 lines · Mark Rutland
Quoted 28 lines · Michael Ellerman
Quoted 26 lines · Mark Rutland
Quoted 10 lines · Peter Zijlstra
Quoted 8 lines · Mark Rutland

/**
+ * atomic64_add_unless - add unless the number is already a given value
+ * @v: pointer of type atomic_t
+ * @a: the amount to add to v...
+ * @u: ...unless v is equal to u.
+ *
+ * Atomically adds @a to @v, so long as @v was not already @u.
+ * Returns non-zero if @v was not @u, and zero otherwise.

Quoted 2 lines

I always get confused by that wording; would something like: "Returns true if the addition was done" not be more clear?

Sounds clearer to me; I just stole the wording from the existing atomic_add_unless().

I guess you'll want similar for the conditional inc/dec ops, e.g.

/**
* atomic_inc_not_zero - increment unless the number is zero
* @v: pointer of type atomic_t
*
* Atomically increments @v by 1, so long as @v is non-zero.
* Returns non-zero if @v was non-zero, and zero otherwise. */

Quoted 8 lines
Quoted 6 lines

If we're bike-shedding .. :)

I think "so long as" is overly wordy and not helpful. It'd be clearer just as:

* Atomically increments @v by 1, if @v is non-zero.

I agree; done.

Thanks.

cheers

author reviewer
#16 lore raw

PATCH v2 6/16 atomics/generic: define atomic64_fetch_add_unless()

In reply to Mark Rutland

As a step towards unifying the atomic/atomic64/atomic_long APIs, this patch converts the generic implementation of atomic64_add_unless() into a generic implementation of atomic64_fetch_add_unless().

A wrapper in <linux/atomic.h> will build atomic_add_unless() atop of this, provided it is given a preprocessor definition.

No functional change is intended as a result of this patch.

---
 include/asm-generic/atomic64.h |  3 ++-
 lib/atomic64.c                 | 13 ++++++-------
 2 files changed, 8 insertions(+), 8 deletions(-)
include/asm-generic/atomic64.h +2 -1
index e9a94b7a4d0a..b0aa425361ad 100644
@@ -52,7 +52,8 @@ ATOMIC64_OPS(xor)
 extern long long atomic64_dec_if_positive(atomic64_t *v);
 extern long long atomic64_cmpxchg(atomic64_t *v, long long o, long long n);
 extern long long atomic64_xchg(atomic64_t *v, long long new);
-extern int	 atomic64_add_unless(atomic64_t *v, long long a, long long u);
+extern long long atomic64_fetch_add_unless(atomic64_t *v, long long a, long long u);
+#define atomic64_fetch_add_unless atomic64_fetch_add_unless
 
 #define atomic64_add_negative(a, v)	(atomic64_add_return((a), (v)) < 0)
 #define atomic64_inc(v)			atomic64_add(1LL, (v))
lib/atomic64.c +6 -7
index 53c2d5edc826..fb2a38af5ae8 100644
@@ -178,18 +178,17 @@ long long atomic64_xchg(atomic64_t *v, long long new)
 }
 EXPORT_SYMBOL(atomic64_xchg);
 
-int atomic64_add_unless(atomic64_t *v, long long a, long long u)
+long long atomic64_fetch_add_unless(atomic64_t *v, long long a, long long u)
 {
 	unsigned long flags;
 	raw_spinlock_t *lock = lock_addr(v);
-	int ret = 0;
+	long long val;
 
 	raw_spin_lock_irqsave(lock, flags);
-	if (v->counter != u) {
+	val = v->counter;
+	if (val != u)
 		v->counter += a;
-		ret = 1;
-	}
 	raw_spin_unlock_irqrestore(lock, flags);
-	return ret;
+	return val;
 }
-EXPORT_SYMBOL(atomic64_add_unless);
+EXPORT_SYMBOL(atomic64_fetch_add_unless);
Signature · 1 line
2.11.0
author reviewer
#17 lore raw

PATCH v2 7/16 atomics/alpha: define atomic64_fetch_add_unless()

In reply to Mark Rutland

As a step towards unifying the atomic/atomic64/atomic_long APIs, this patch converts the arch/alpha implementation of atomic64_add_unless() into an implementation of atomic64_fetch_add_unless().

A wrapper in <linux/atomic.h> will build atomic_add_unless() atop of this, provided it is given a preprocessor definition.

No functional change is intended as a result of this patch.

Cc Boqun Feng <[email protected]>
Cc Will Deacon <[email protected]>
Cc Richard Henderson <[email protected]>
Cc Ivan Kokshaysky <[email protected]>
Cc Matt Turner <[email protected]>
---
 arch/alpha/include/asm/atomic.h | 23 ++++++++++++-----------
 1 file changed, 12 insertions(+), 11 deletions(-)
arch/alpha/include/asm/atomic.h +12 -11
index 4a800a3424a3..dcb7bbeeae02 100644
@@ -238,35 +238,36 @@ static __inline__ int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 #define atomic_fetch_add_unless atomic_fetch_add_unless
 
 /**
- * atomic64_add_unless - add unless the number is a given value
+ * atomic64_fetch_add_unless - add unless the number is a given value
  * @v: pointer of type atomic64_t
  * @a: the amount to add to v...
  * @u: ...unless v is equal to u.
  *
  * Atomically adds @a to @v, so long as it was not @u.
- * Returns true iff @v was not @u.
+ * Returns the old value of @v.
  */
-static __inline__ int atomic64_add_unless(atomic64_t *v, long a, long u)
+static __inline__ int atomic64_fetch_add_unless(atomic64_t *v, long a, long u)
 {
-	long c, tmp;
+	long c, new, old;
 	smp_mb();
 	__asm__ __volatile__(
-	"1:	ldq_l	%[tmp],%[mem]\n"
-	"	cmpeq	%[tmp],%[u],%[c]\n"
-	"	addq	%[tmp],%[a],%[tmp]\n"
+	"1:	ldq_l	%[old],%[mem]\n"
+	"	cmpeq	%[old],%[u],%[c]\n"
+	"	addq	%[old],%[a],%[new]\n"
 	"	bne	%[c],2f\n"
-	"	stq_c	%[tmp],%[mem]\n"
-	"	beq	%[tmp],3f\n"
+	"	stq_c	%[new],%[mem]\n"
+	"	beq	%[new],3f\n"
 	"2:\n"
 	".subsection 2\n"
 	"3:	br	1b\n"
 	".previous"
-	: [tmp] "=&r"(tmp), [c] "=&r"(c)
+	: [old] "=&r"(old), [new] "=&r"(new), [c] "=&r"(c)
 	: [mem] "m"(*v), [a] "rI"(a), [u] "rI"(u)
 	: "memory");
 	smp_mb();
-	return !c;
+	return old;
 }
+#define atomic64_fetch_add_unless atomic64_fetch_add_unless
 
 /*
  * atomic64_dec_if_positive - decrement by 1 if old value positive
Signature · 1 line
2.11.0
author reviewer
#18 lore raw

PATCH v2 8/16 atomics/arc: define atomic64_fetch_add_unless()

In reply to Mark Rutland

As a step towards unifying the atomic/atomic64/atomic_long APIs, this patch converts the arch/arc implementation of atomic64_add_unless() into an implementation of atomic64_fetch_add_unless().

A wrapper in <linux/atomic.h> will build atomic_add_unless() atop of this, provided it is given a preprocessor definition.

No functional change is intended as a result of this patch.

---
 arch/arc/include/asm/atomic.h | 25 ++++++++++++-------------
 1 file changed, 12 insertions(+), 13 deletions(-)
arch/arc/include/asm/atomic.h +12 -13
index 60da80481c5d..335ddc21ed03 100644
@@ -531,41 +531,40 @@ static inline long long atomic64_dec_if_positive(atomic64_t *v)
 }
 
 /**
- * atomic64_add_unless - add unless the number is a given value
+ * atomic64_fetch_add_unless - add unless the number is a given value
  * @v: pointer of type atomic64_t
  * @a: the amount to add to v...
  * @u: ...unless v is equal to u.
  *
- * if (v != u) { v += a; ret = 1} else {ret = 0}
- * Returns 1 iff @v was not @u (i.e. if add actually happened)
+ * Atomically adds @a to @v, so long as it was not @u.
+ * Returns the old value of @v
  */
-static inline int atomic64_add_unless(atomic64_t *v, long long a, long long u)
+static inline long long atomic64_fetch_add_unless(atomic64_t *v, long long a,
+						  long long u)
 {
-	long long val;
-	int op_done;
+	long long old, temp;
 
 	smp_mb();
 
 	__asm__ __volatile__(
 	"1:	llockd  %0, [%2]	\n"
-	"	mov	%1, 1		\n"
 	"	brne	%L0, %L4, 2f	# continue to add since v != u \n"
 	"	breq.d	%H0, %H4, 3f	# return since v == u \n"
-	"	mov	%1, 0		\n"
 	"2:				\n"
-	"	add.f   %L0, %L0, %L3	\n"
-	"	adc     %H0, %H0, %H3	\n"
-	"	scondd  %0, [%2]	\n"
+	"	add.f   %L1, %L0, %L3	\n"
+	"	adc     %H1, %H0, %H3	\n"
+	"	scondd  %1, [%2]	\n"
 	"	bnz     1b		\n"
 	"3:				\n"
-	: "=&r"(val), "=&r" (op_done)
+	: "=&r"(old), "=&r" (temp)
 	: "r"(&v->counter), "r"(a), "r"(u)
 	: "cc");	/* memory clobber comes from smp_mb() */
 
 	smp_mb();
 
-	return op_done;
+	return old;
 }
+#define atomic64_fetch_add_unless atomic64_fetch_add_unless
 
 #define atomic64_add_negative(a, v)	(atomic64_add_return((a), (v)) < 0)
 #define atomic64_inc(v)			atomic64_add(1LL, (v))
Signature · 1 line
2.11.0
author reviewer
#19 lore raw

PATCH v2 9/16 atomics/arm: define atomic64_fetch_add_unless()

In reply to Mark Rutland

As a step towards unifying the atomic/atomic64/atomic_long APIs, this patch converts the arch/arm implementation of atomic64_add_unless() into an implementation of atomic64_fetch_add_unless().

A wrapper in <linux/atomic.h> will build atomic_add_unless() atop of this, provided it is given a preprocessor definition.

No functional change is intended as a result of this patch.

---
 arch/arm/include/asm/atomic.h | 20 ++++++++++----------
 1 file changed, 10 insertions(+), 10 deletions(-)
arch/arm/include/asm/atomic.h +10 -10
index 74460aa00fa0..852e1fee72b0 100644
@@ -486,11 +486,11 @@ static inline long long atomic64_dec_if_positive(atomic64_t *v)
 	return result;
 }
 
-static inline int atomic64_add_unless(atomic64_t *v, long long a, long long u)
+static inline long long atomic64_fetch_add_unless(atomic64_t *v, long long a,
+						  long long u)
 {
-	long long val;
+	long long oldval, newval;
 	unsigned long tmp;
-	int ret = 1;
 
 	smp_mb();
 	prefetchw(&v->counter);
@@ -499,23 +499,23 @@ static inline int atomic64_add_unless(atomic64_t *v, long long a, long long u)
 "1:	ldrexd	%0, %H0, [%4]\n"
 "	teq	%0, %5\n"
 "	teqeq	%H0, %H5\n"
-"	moveq	%1, #0\n"
 "	beq	2f\n"
-"	adds	%Q0, %Q0, %Q6\n"
-"	adc	%R0, %R0, %R6\n"
-"	strexd	%2, %0, %H0, [%4]\n"
+"	adds	%Q1, %Q0, %Q6\n"
+"	adc	%R1, %R0, %R6\n"
+"	strexd	%2, %1, %H1, [%4]\n"
 "	teq	%2, #0\n"
 "	bne	1b\n"
 "2:"
-	: "=&r" (val), "+r" (ret), "=&r" (tmp), "+Qo" (v->counter)
+	: "=&r" (oldval), "=&r" (newval), "=&r" (tmp), "+Qo" (v->counter)
 	: "r" (&v->counter), "r" (u), "r" (a)
 	: "cc");
 
-	if (ret)
+	if (oldval != u)
 		smp_mb();
 
-	return ret;
+	return oldval;
 }
+#define atomic64_fetch_add_unless atomic64_fetch_add_unless
 
 #define atomic64_add_negative(a, v)	(atomic64_add_return((a), (v)) < 0)
 #define atomic64_inc(v)			atomic64_add(1LL, (v))
Signature · 1 line
2.11.0
author reviewer
#20 lore raw

PATCH v2 10/16 atomics/powerpc: define atomic64_fetch_add_unless()

In reply to Mark Rutland

As a step towards unifying the atomic/atomic64/atomic_long APIs, this patch converts the arch/powerpc implementation of atomic64_add_unless() into an implementation of atomic64_fetch_add_unless().

A wrapper in <linux/atomic.h> will build atomic_add_unless() atop of this, provided it is given a preprocessor definition.

No functional change is intended as a result of this patch.

---
 arch/powerpc/include/asm/atomic.h | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)
arch/powerpc/include/asm/atomic.h +5 -4
index b5646c079c16..233dbf31911c 100644
@@ -525,7 +525,7 @@ static __inline__ long atomic64_dec_if_positive(atomic64_t *v)
 #define atomic64_xchg_relaxed(v, new) xchg_relaxed(&((v)->counter), (new))
 
 /**
- * atomic64_add_unless - add unless the number is a given value
+ * atomic64_fetch_add_unless - add unless the number is a given value
  * @v: pointer of type atomic64_t
  * @a: the amount to add to v...
  * @u: ...unless v is equal to u.
@@ -533,13 +533,13 @@ static __inline__ long atomic64_dec_if_positive(atomic64_t *v)
  * Atomically adds @a to @v, so long as it was not @u.
  * Returns the old value of @v.
  */
-static __inline__ int atomic64_add_unless(atomic64_t *v, long a, long u)
+static __inline__ long atomic64_fetch_add_unless(atomic64_t *v, long a, long u)
 {
 	long t;
 
 	__asm__ __volatile__ (
 	PPC_ATOMIC_ENTRY_BARRIER
-"1:	ldarx	%0,0,%1		# atomic_fetch_add_unless\n\
+"1:	ldarx	%0,0,%1		# atomic64_fetch_add_unless\n\
 	cmpd	0,%0,%3 \n\
 	beq	2f \n\
 	add	%0,%2,%0 \n"
@@ -552,8 +552,9 @@ static __inline__ int atomic64_add_unless(atomic64_t *v, long a, long u)
 	: "r" (&v->counter), "r" (a), "r" (u)
 	: "cc", "memory");
 
-	return t != u;
+	return t;
 }
+#define atomic64_fetch_add_unless atomic64_fetch_add_unless
 
 /**
  * atomic_inc64_not_zero - increment unless the number is zero
Signature · 1 line
2.11.0
author reviewer
#21 lore raw

PATCH v2 11/16 atomics/riscv: define atomic64_fetch_add_unless()

In reply to Mark Rutland

As a step towards unifying the atomic/atomic64/atomic_long APIs, this patch converts the arch/riscv implementation of atomic64_add_unless() into an implementation of atomic64_fetch_add_unless().

A wrapper in <linux/atomic.h> will build atomic_add_unless() atop of this, provided it is given a preprocessor definition.

No functional change is intended as a result of this patch.

---
 arch/riscv/include/asm/atomic.h | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)
arch/riscv/include/asm/atomic.h +2 -6
index 5f161daefcd2..d959bbaaad41 100644
@@ -352,7 +352,7 @@ static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 #define atomic_fetch_add_unless atomic_fetch_add_unless
 
 #ifndef CONFIG_GENERIC_ATOMIC64
-static __always_inline long __atomic64_add_unless(atomic64_t *v, long a, long u)
+static __always_inline long atomic64_fetch_add_unless(atomic64_t *v, long a, long u)
 {
        long prev, rc;
 
@@ -369,11 +369,7 @@ static __always_inline long __atomic64_add_unless(atomic64_t *v, long a, long u)
 		: "memory");
 	return prev;
 }
-
-static __always_inline int atomic64_add_unless(atomic64_t *v, long a, long u)
-{
-	return __atomic64_add_unless(v, a, u) != u;
-}
+#define atomic64_fetch_add_unless atomic64_fetch_add_unless
 #endif
 
 /*
Signature · 1 line
2.11.0
#22 lore raw

Re: [PATCHv2 11/16] atomics/riscv: define atomic64_fetch_add_unless()

Quoted 35 lines · On Tue, 29 May 2018 08:43:41 PDT (-0700), [email protected] wrote:

As a step towards unifying the atomic/atomic64/atomic_long APIs, this patch converts the arch/riscv implementation of atomic64_add_unless() into an implementation of atomic64_fetch_add_unless().

A wrapper in <linux/atomic.h> will build atomic_add_unless() atop of this, provided it is given a preprocessor definition.

No functional change is intended as a result of this patch.

---
 arch/riscv/include/asm/atomic.h | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)
arch/riscv/include/asm/atomic.h +1 -1
index 5f161daefcd2..d959bbaaad41 100644
@@ -352,7 +352,7 @@ static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 #define atomic_fetch_add_unless atomic_fetch_add_unless
 
 #ifndef CONFIG_GENERIC_ATOMIC64
-static __always_inline long __atomic64_add_unless(atomic64_t *v, long a, long u)
+static __always_inline long atomic64_fetch_add_unless(atomic64_t *v, long a, long u)
 {
        long prev, rc;
 
@@ -369,11 +369,7 @@ static __always_inline long __atomic64_add_unless(atomic64_t *v, long a, long u)
 		: "memory");
 	return prev;

arch/riscv/include/asm/atomic.h static __always_inline long __atomic64_add_unless(atomic64_t *v, long a, long u) · line 375

Quoted 10 lines
arch/riscv/include/asm/atomic.h +1 -5
 }
-
-static __always_inline int atomic64_add_unless(atomic64_t *v, long a, long u)
-{
-	return __atomic64_add_unless(v, a, u) != u;
-}
+#define atomic64_fetch_add_unless atomic64_fetch_add_unless
 #endif
 
 /*

For some reason I remember there being a reason we were doing this in such an odd fashion but I can't remember what it was any more. Assuming this still builds, feel free to add an

Acked-by Palmer Dabbelt <[email protected]>

Thanks!

author reviewer
#23 lore raw

PATCH v2 12/16 atomics/treewide: make atomic64_fetch_add_unless() optional

In reply to Mark Rutland

Architectures with atomic64_fetch_add_unless provide a preprocessor symbol if they do so, and all other architectures have trivial C implementations of atomic64_add_unless() which are near-identical.

Let's unify the trivial definitions of atomic64_fetch_add_unless() in <linux/atomic.h>, so that we always have both atomic64_fetch_add_unless() and atomic64_add_unless() with less boilerplate code.

This means that atomic64_add_unless() is always implemented in core code, and the instrumented atomics are updated accordingly.

There should be no functional change as a result of this patch.

---
 arch/arm64/include/asm/atomic.h           | 12 ------------
 arch/ia64/include/asm/atomic.h            | 15 ---------------
 arch/mips/include/asm/atomic.h            | 24 ------------------------
 arch/parisc/include/asm/atomic.h          | 24 ------------------------
 arch/s390/include/asm/atomic.h            | 16 ----------------
 arch/sparc/include/asm/atomic_64.h        | 15 ---------------
 arch/x86/include/asm/atomic64_64.h        | 19 -------------------
 include/asm-generic/atomic-instrumented.h |  6 ------
 include/linux/atomic.h                    | 26 ++++++++++++++++++++++++--
 9 files changed, 24 insertions(+), 133 deletions(-)
arch/arm64/include/asm/atomic.h +0 -12
index 22c8c43d6689..82db0e4febd4 100644
@@ -40,17 +40,6 @@
 
 #include <asm/cmpxchg.h>
 
-#define ___atomic_add_unless(v, a, u, sfx)				\
-({									\
-	typeof((v)->counter) c, old;					\
-									\
-	c = atomic##sfx##_read(v);					\
-	while (c != (u) &&						\
-	      (old = atomic##sfx##_cmpxchg((v), c, c + (a))) != c)	\
-		c = old;						\
-	c;								\
- })
-
 #define ATOMIC_INIT(i)	{ (i) }
 
 #define atomic_read(v)			READ_ONCE((v)->counter)
@@ -200,7 +189,6 @@
 #define atomic64_dec_and_test(v)	(atomic64_dec_return(v) == 0)
 #define atomic64_sub_and_test(i, v)	(atomic64_sub_return((i), (v)) == 0)
 #define atomic64_add_negative(i, v)	(atomic64_add_return((i), (v)) < 0)
-#define atomic64_add_unless(v, a, u)	(___atomic_add_unless(v, a, u, 64) != u)
 #define atomic64_andnot			atomic64_andnot
 
 #endif
arch/ia64/include/asm/atomic.h +0 -15
index cfe44086338e..0f80a3eafaba 100644
@@ -215,21 +215,6 @@ ATOMIC64_FETCH_OP(xor, ^)
 	(cmpxchg(&((v)->counter), old, new))
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), new))
 
-static __inline__ long atomic64_add_unless(atomic64_t *v, long a, long u)
-{
-	long c, old;
-	c = atomic64_read(v);
-	for (;;) {
-		if (unlikely(c == (u)))
-			break;
-		old = atomic64_cmpxchg((v), c, c + (a));
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c != (u);
-}
-
 static __inline__ long atomic64_dec_if_positive(atomic64_t *v)
 {
 	long c, old, dec;
arch/mips/include/asm/atomic.h +0 -24
index 794734e730d9..d42b27df1548 100644
@@ -596,30 +596,6 @@ static __inline__ long atomic64_sub_if_positive(long i, atomic64_t * v)
 	((__typeof__((v)->counter))cmpxchg(&((v)->counter), (o), (n)))
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), (new)))
 
-/**
- * atomic64_add_unless - add unless the number is a given value
- * @v: pointer of type atomic64_t
- * @a: the amount to add to v...
- * @u: ...unless v is equal to u.
- *
- * Atomically adds @a to @v, so long as it was not @u.
- * Returns true iff @v was not @u.
- */
-static __inline__ int atomic64_add_unless(atomic64_t *v, long a, long u)
-{
-	long c, old;
-	c = atomic64_read(v);
-	for (;;) {
-		if (unlikely(c == (u)))
-			break;
-		old = atomic64_cmpxchg((v), c, c + (a));
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c != (u);
-}
-
 #define atomic64_dec_return(v) atomic64_sub_return(1, (v))
 #define atomic64_inc_return(v) atomic64_add_return(1, (v))
 
arch/parisc/include/asm/atomic.h +0 -24
index b2b6261d05e7..f53ba2d6ff67 100644
@@ -257,30 +257,6 @@ atomic64_read(const atomic64_t *v)
 	((__typeof__((v)->counter))cmpxchg(&((v)->counter), (o), (n)))
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), new))
 
-/**
- * atomic64_add_unless - add unless the number is a given value
- * @v: pointer of type atomic64_t
- * @a: the amount to add to v...
- * @u: ...unless v is equal to u.
- *
- * Atomically adds @a to @v, so long as it was not @u.
- * Returns the old value of @v.
- */
-static __inline__ int atomic64_add_unless(atomic64_t *v, long a, long u)
-{
-	long c, old;
-	c = atomic64_read(v);
-	for (;;) {
-		if (unlikely(c == (u)))
-			break;
-		old = atomic64_cmpxchg((v), c, c + (a));
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c != (u);
-}
-
 /*
  * atomic64_dec_if_positive - decrement by 1 if old value positive
  * @v: pointer of type atomic_t
arch/s390/include/asm/atomic.h +0 -16
index 26c6b713a7a3..eb9329741bad 100644
@@ -153,22 +153,6 @@ ATOMIC64_OPS(xor)
 
 #undef ATOMIC64_OPS
 
-static inline int atomic64_add_unless(atomic64_t *v, long i, long u)
-{
-	long c, old;
-
-	c = atomic64_read(v);
-	for (;;) {
-		if (unlikely(c == u))
-			break;
-		old = atomic64_cmpxchg(v, c, c + i);
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c != u;
-}
-
 static inline long atomic64_dec_if_positive(atomic64_t *v)
 {
 	long c, old, dec;
arch/sparc/include/asm/atomic_64.h +0 -15
index e4f1c93db31f..458783e99997 100644
@@ -93,21 +93,6 @@ static inline int atomic_xchg(atomic_t *v, int new)
 	((__typeof__((v)->counter))cmpxchg(&((v)->counter), (o), (n)))
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), new))
 
-static inline long atomic64_add_unless(atomic64_t *v, long a, long u)
-{
-	long c, old;
-	c = atomic64_read(v);
-	for (;;) {
-		if (unlikely(c == (u)))
-			break;
-		old = atomic64_cmpxchg((v), c, c + (a));
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return c != (u);
-}
-
 long atomic64_dec_if_positive(atomic64_t *v);
 
 #endif /* !(__ARCH_SPARC64_ATOMIC__) */
arch/x86/include/asm/atomic64_64.h +0 -19
index 6f95023894b7..7e04b294e6eb 100644
@@ -188,25 +188,6 @@ static inline long arch_atomic64_xchg(atomic64_t *v, long new)
 	return xchg(&v->counter, new);
 }
 
-/**
- * arch_atomic64_add_unless - add unless the number is a given value
- * @v: pointer of type atomic64_t
- * @a: the amount to add to v...
- * @u: ...unless v is equal to u.
- *
- * Atomically adds @a to @v, so long as it was not @u.
- * Returns the old value of @v.
- */
-static inline bool arch_atomic64_add_unless(atomic64_t *v, long a, long u)
-{
-	s64 c = arch_atomic64_read(v);
-	do {
-		if (unlikely(c == u))
-			return false;
-	} while (!arch_atomic64_try_cmpxchg(v, &c, c + a));
-	return true;
-}
-
 /*
  * arch_atomic64_dec_if_positive - decrement by 1 if old value positive
  * @v: pointer of type atomic_t
include/asm-generic/atomic-instrumented.h +0 -6
index e22d7e5f4ce7..719168de5c6e 100644
@@ -100,12 +100,6 @@ static __always_inline int atomic64_fetch_add_unless(atomic64_t *v, s64 a, s64 u
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic64_fetch_add_unless(v, a, u);
 }
-#else
-static __always_inline bool atomic64_add_unless(atomic64_t *v, s64 a, s64 u)
-{
-	kasan_check_write(v, sizeof(*v));
-	return arch_atomic64_add_unless(v, a, u);
-}
 #endif
 
 static __always_inline void atomic_inc(atomic_t *v)
include/linux/atomic.h +24 -2
index 8d93209052e1..ab534fad09e1 100644
@@ -1072,6 +1072,30 @@ static inline int atomic_dec_if_positive(atomic_t *v)
 #endif /* atomic64_try_cmpxchg */
 
 /**
+ * atomic64_fetch_add_unless - add unless the number is already a given value
+ * @v: pointer of type atomic64_t
+ * @a: the amount to add to v...
+ * @u: ...unless v is equal to u.
+ *
+ * Atomically adds @a to @v, so long as @v was not already @u.
+ * Returns original value of @v
+ */
+#ifndef atomic64_fetch_add_unless
+static inline long long atomic64_fetch_add_unless(atomic64_t *v, long long a,
+						  long long u)
+{
+	long long c = atomic64_read(v);
+
+	do {
+		if (unlikely(c == u))
+			break;
+	} while (!atomic64_try_cmpxchg(v, &c, c + a));
+
+	return c;
+}
+#endif
+
+/**
  * atomic64_add_unless - add unless the number is already a given value
  * @v: pointer of type atomic_t
  * @a: the amount to add to v...
@@ -1080,12 +1104,10 @@ static inline int atomic_dec_if_positive(atomic_t *v)
  * Atomically adds @a to @v, so long as @v was not already @u.
  * Returns non-zero if @v was not @u, and zero otherwise.
  */
-#ifdef atomic64_fetch_add_unless
 static inline int atomic64_add_unless(atomic64_t *v, long long a, long long u)
 {
 	return atomic64_fetch_add_unless(v, a, u) != u;
 }
-#endif
 
 /**
  * atomic64_inc_not_zero - increment unless the number is zero
Signature · 1 line
2.11.0
author reviewer
#24 lore raw

PATCH v2 13/16 atomics/treewide: make test ops optional

In reply to Mark Rutland

Some of the atomics return the result of a test applied after the atomic operation, and almost all architectures implement these as trivial wrappers around the underlying atomic. Specifically:

* <atomic>_inc_and_test(v) is (<atomic>_inc_return(v) == 0)

* <atomic>_dec_and_test(v) is (<atomic>_dec_return(v) == 0)

* <atomic>_sub_and_test(i, v) is (<atomic>_sub_return(i, v) == 0)

* <atomic>_add_negative(i, v) is (<atomic>_add_return(i, v) < 0)

Rather than have these definitions duplicated in all architectures, with minor inconsistencies in formatting and documentation, let's make these operations optional, with default fallbacks as above. Implementations must now provide a preprocessor symbol.

The instrumented atomics are updated accordingly.

Both x86 and m68k have custom implementations, which are left as-is, given preprocessor symbols to avoid being overridden.

There should be no functional change as a result of this patch.

---
 arch/alpha/include/asm/atomic.h           |  12 ---
 arch/arc/include/asm/atomic.h             |  10 ---
 arch/arm/include/asm/atomic.h             |   9 ---
 arch/arm64/include/asm/atomic.h           |   8 --
 arch/h8300/include/asm/atomic.h           |   5 --
 arch/hexagon/include/asm/atomic.h         |   5 --
 arch/ia64/include/asm/atomic.h            |  23 ------
 arch/m68k/include/asm/atomic.h            |   4 +
 arch/mips/include/asm/atomic.h            |  84 --------------------
 arch/parisc/include/asm/atomic.h          |  22 ------
 arch/powerpc/include/asm/atomic.h         |  30 --------
 arch/riscv/include/asm/atomic.h           |  46 -----------
 arch/s390/include/asm/atomic.h            |   8 --
 arch/sh/include/asm/atomic.h              |   4 -
 arch/sparc/include/asm/atomic_32.h        |  15 ----
 arch/sparc/include/asm/atomic_64.h        |  20 -----
 arch/x86/include/asm/atomic.h             |   4 +
 arch/x86/include/asm/atomic64_32.h        |  54 -------------
 arch/x86/include/asm/atomic64_64.h        |   4 +
 arch/xtensa/include/asm/atomic.h          |  42 ----------
 include/asm-generic/atomic-instrumented.h |  24 ++++++
 include/asm-generic/atomic.h              |   9 ---
 include/asm-generic/atomic64.h            |   4 -
 include/linux/atomic.h                    | 124 ++++++++++++++++++++++++++++++
 24 files changed, 160 insertions(+), 410 deletions(-)
arch/alpha/include/asm/atomic.h +0 -12
index dcb7bbeeae02..31a4904dc8fc 100644
@@ -297,24 +297,12 @@ static inline long atomic64_dec_if_positive(atomic64_t *v)
 	return old - 1;
 }
 
-#define atomic_add_negative(a, v) (atomic_add_return((a), (v)) < 0)
-#define atomic64_add_negative(a, v) (atomic64_add_return((a), (v)) < 0)
-
 #define atomic_dec_return(v) atomic_sub_return(1,(v))
 #define atomic64_dec_return(v) atomic64_sub_return(1,(v))
 
 #define atomic_inc_return(v) atomic_add_return(1,(v))
 #define atomic64_inc_return(v) atomic64_add_return(1,(v))
 
-#define atomic_sub_and_test(i,v) (atomic_sub_return((i), (v)) == 0)
-#define atomic64_sub_and_test(i,v) (atomic64_sub_return((i), (v)) == 0)
-
-#define atomic_inc_and_test(v) (atomic_add_return(1, (v)) == 0)
-#define atomic64_inc_and_test(v) (atomic64_add_return(1, (v)) == 0)
-
-#define atomic_dec_and_test(v) (atomic_sub_return(1, (v)) == 0)
-#define atomic64_dec_and_test(v) (atomic64_sub_return(1, (v)) == 0)
-
 #define atomic_inc(v) atomic_add(1,(v))
 #define atomic64_inc(v) atomic64_add(1,(v))
 
arch/arc/include/asm/atomic.h +0 -10
index 335ddc21ed03..555b0a94b643 100644
@@ -311,14 +311,8 @@ ATOMIC_OPS(xor, ^=, CTOP_INST_AXOR_DI_R2_R2_R3)
 #define atomic_inc(v)			atomic_add(1, v)
 #define atomic_dec(v)			atomic_sub(1, v)
 
-#define atomic_inc_and_test(v)		(atomic_add_return(1, v) == 0)
-#define atomic_dec_and_test(v)		(atomic_sub_return(1, v) == 0)
 #define atomic_inc_return(v)		atomic_add_return(1, (v))
 #define atomic_dec_return(v)		atomic_sub_return(1, (v))
-#define atomic_sub_and_test(i, v)	(atomic_sub_return(i, v) == 0)
-
-#define atomic_add_negative(i, v)	(atomic_add_return(i, v) < 0)
-
 
 #ifdef CONFIG_GENERIC_ATOMIC64
 
@@ -566,14 +560,10 @@ static inline long long atomic64_fetch_add_unless(atomic64_t *v, long long a,
 }
 #define atomic64_fetch_add_unless atomic64_fetch_add_unless
 
-#define atomic64_add_negative(a, v)	(atomic64_add_return((a), (v)) < 0)
 #define atomic64_inc(v)			atomic64_add(1LL, (v))
 #define atomic64_inc_return(v)		atomic64_add_return(1LL, (v))
-#define atomic64_inc_and_test(v)	(atomic64_inc_return(v) == 0)
-#define atomic64_sub_and_test(a, v)	(atomic64_sub_return((a), (v)) == 0)
 #define atomic64_dec(v)			atomic64_sub(1LL, (v))
 #define atomic64_dec_return(v)		atomic64_sub_return(1LL, (v))
-#define atomic64_dec_and_test(v)	(atomic64_dec_return((v)) == 0)
 
 #endif	/* !CONFIG_GENERIC_ATOMIC64 */
 
arch/arm/include/asm/atomic.h +0 -9
index 852e1fee72b0..35fb7f504daa 100644
@@ -248,13 +248,8 @@ ATOMIC_OPS(xor, ^=, eor)
 #define atomic_inc(v)		atomic_add(1, v)
 #define atomic_dec(v)		atomic_sub(1, v)
 
-#define atomic_inc_and_test(v)	(atomic_add_return(1, v) == 0)
-#define atomic_dec_and_test(v)	(atomic_sub_return(1, v) == 0)
 #define atomic_inc_return_relaxed(v)    (atomic_add_return_relaxed(1, v))
 #define atomic_dec_return_relaxed(v)    (atomic_sub_return_relaxed(1, v))
-#define atomic_sub_and_test(i, v) (atomic_sub_return(i, v) == 0)
-
-#define atomic_add_negative(i,v) (atomic_add_return(i, v) < 0)
 
 #ifndef CONFIG_GENERIC_ATOMIC64
 typedef struct {
@@ -517,14 +512,10 @@ static inline long long atomic64_fetch_add_unless(atomic64_t *v, long long a,
 }
 #define atomic64_fetch_add_unless atomic64_fetch_add_unless
 
-#define atomic64_add_negative(a, v)	(atomic64_add_return((a), (v)) < 0)
 #define atomic64_inc(v)			atomic64_add(1LL, (v))
 #define atomic64_inc_return_relaxed(v)	atomic64_add_return_relaxed(1LL, (v))
-#define atomic64_inc_and_test(v)	(atomic64_inc_return(v) == 0)
-#define atomic64_sub_and_test(a, v)	(atomic64_sub_return((a), (v)) == 0)
 #define atomic64_dec(v)			atomic64_sub(1LL, (v))
 #define atomic64_dec_return_relaxed(v)	atomic64_sub_return_relaxed(1LL, (v))
-#define atomic64_dec_and_test(v)	(atomic64_dec_return((v)) == 0)
 
 #endif /* !CONFIG_GENERIC_ATOMIC64 */
 #endif
arch/arm64/include/asm/atomic.h +0 -8
index 82db0e4febd4..edbe53fa3106 100644
@@ -110,10 +110,6 @@
 
 #define atomic_inc(v)			atomic_add(1, (v))
 #define atomic_dec(v)			atomic_sub(1, (v))
-#define atomic_inc_and_test(v)		(atomic_inc_return(v) == 0)
-#define atomic_dec_and_test(v)		(atomic_dec_return(v) == 0)
-#define atomic_sub_and_test(i, v)	(atomic_sub_return((i), (v)) == 0)
-#define atomic_add_negative(i, v)	(atomic_add_return((i), (v)) < 0)
 #define atomic_andnot			atomic_andnot
 
 /*
@@ -185,10 +181,6 @@
 
 #define atomic64_inc(v)			atomic64_add(1, (v))
 #define atomic64_dec(v)			atomic64_sub(1, (v))
-#define atomic64_inc_and_test(v)	(atomic64_inc_return(v) == 0)
-#define atomic64_dec_and_test(v)	(atomic64_dec_return(v) == 0)
-#define atomic64_sub_and_test(i, v)	(atomic64_sub_return((i), (v)) == 0)
-#define atomic64_add_negative(i, v)	(atomic64_add_return((i), (v)) < 0)
 #define atomic64_andnot			atomic64_andnot
 
 #endif
arch/h8300/include/asm/atomic.h +0 -5
index 250bb3baac23..b658b20d2232 100644
@@ -69,17 +69,12 @@ ATOMIC_OPS(sub, -=)
 #undef ATOMIC_OP_RETURN
 #undef ATOMIC_OP
 
-#define atomic_add_negative(a, v)	(atomic_add_return((a), (v)) < 0)
-#define atomic_sub_and_test(i, v)	(atomic_sub_return(i, v) == 0)
-
 #define atomic_inc_return(v)		atomic_add_return(1, v)
 #define atomic_dec_return(v)		atomic_sub_return(1, v)
 
 #define atomic_inc(v)			(void)atomic_inc_return(v)
-#define atomic_inc_and_test(v)		(atomic_inc_return(v) == 0)
 
 #define atomic_dec(v)			(void)atomic_dec_return(v)
-#define atomic_dec_and_test(v)		(atomic_dec_return(v) == 0)
 
 static inline int atomic_cmpxchg(atomic_t *v, int old, int new)
 {
arch/hexagon/include/asm/atomic.h +0 -5
index 86c67e9adbfa..31638f511674 100644
@@ -201,11 +201,6 @@ static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 #define atomic_inc(v) atomic_add(1, (v))
 #define atomic_dec(v) atomic_sub(1, (v))
 
-#define atomic_inc_and_test(v) (atomic_add_return(1, (v)) == 0)
-#define atomic_dec_and_test(v) (atomic_sub_return(1, (v)) == 0)
-#define atomic_sub_and_test(i, v) (atomic_sub_return(i, (v)) == 0)
-#define atomic_add_negative(i, v) (atomic_add_return(i, (v)) < 0)
-
 #define atomic_inc_return(v) (atomic_add_return(1, v))
 #define atomic_dec_return(v) (atomic_sub_return(1, v))
 
arch/ia64/include/asm/atomic.h +0 -23
index 0f80a3eafaba..e4143c462e65 100644
@@ -231,34 +231,11 @@ static __inline__ long atomic64_dec_if_positive(atomic64_t *v)
 	return dec;
 }
 
-/*
- * Atomically add I to V and return TRUE if the resulting value is
- * negative.
- */
-static __inline__ int
-atomic_add_negative (int i, atomic_t *v)
-{
-	return atomic_add_return(i, v) < 0;
-}
-
-static __inline__ long
-atomic64_add_negative (__s64 i, atomic64_t *v)
-{
-	return atomic64_add_return(i, v) < 0;
-}
-
 #define atomic_dec_return(v)		atomic_sub_return(1, (v))
 #define atomic_inc_return(v)		atomic_add_return(1, (v))
 #define atomic64_dec_return(v)		atomic64_sub_return(1, (v))
 #define atomic64_inc_return(v)		atomic64_add_return(1, (v))
 
-#define atomic_sub_and_test(i,v)	(atomic_sub_return((i), (v)) == 0)
-#define atomic_dec_and_test(v)		(atomic_sub_return(1, (v)) == 0)
-#define atomic_inc_and_test(v)		(atomic_add_return(1, (v)) == 0)
-#define atomic64_sub_and_test(i,v)	(atomic64_sub_return((i), (v)) == 0)
-#define atomic64_dec_and_test(v)	(atomic64_sub_return(1, (v)) == 0)
-#define atomic64_inc_and_test(v)	(atomic64_add_return(1, (v)) == 0)
-
 #define atomic_add(i,v)			(void)atomic_add_return((i), (v))
 #define atomic_sub(i,v)			(void)atomic_sub_return((i), (v))
 #define atomic_inc(v)			atomic_add(1, (v))
arch/m68k/include/asm/atomic.h +4 -0
index 596882cda224..9df09c876fa2 100644
@@ -138,6 +138,7 @@ static inline int atomic_dec_and_test(atomic_t *v)
 	__asm__ __volatile__("subql #1,%1; seq %0" : "=d" (c), "+m" (*v));
 	return c != 0;
 }
+#define atomic_dec_and_test atomic_dec_and_test
 
 static inline int atomic_dec_and_test_lt(atomic_t *v)
 {
@@ -155,6 +156,7 @@ static inline int atomic_inc_and_test(atomic_t *v)
 	__asm__ __volatile__("addql #1,%1; seq %0" : "=d" (c), "+m" (*v));
 	return c != 0;
 }
+#define atomic_inc_and_test atomic_inc_and_test
 
 #ifdef CONFIG_RMW_INSNS
 
@@ -201,6 +203,7 @@ static inline int atomic_sub_and_test(int i, atomic_t *v)
 			     : ASM_DI (i));
 	return c != 0;
 }
+#define atomic_sub_and_test atomic_sub_and_test
 
 static inline int atomic_add_negative(int i, atomic_t *v)
 {
@@ -210,5 +213,6 @@ static inline int atomic_add_negative(int i, atomic_t *v)
 			     : ASM_DI (i));
 	return c != 0;
 }
+#define atomic_add_negative atomic_add_negative
 
 #endif /* __ARCH_M68K_ATOMIC __ */
arch/mips/include/asm/atomic.h +0 -84

108 lines not on this page · raw message

arch/parisc/include/asm/atomic.h +0 -22
index f53ba2d6ff67..f85844ff6336 100644
@@ -142,22 +142,6 @@ ATOMIC_OPS(xor, ^=)
 #define atomic_inc_return(v)	(atomic_add_return(   1,(v)))
 #define atomic_dec_return(v)	(atomic_add_return(  -1,(v)))
 
-#define atomic_add_negative(a, v)	(atomic_add_return((a), (v)) < 0)
-
-/*
- * atomic_inc_and_test - increment and test
- * @v: pointer of type atomic_t
- *
- * Atomically increments @v by 1
- * and returns true if the result is zero, or false for all
- * other cases.
- */
-#define atomic_inc_and_test(v) (atomic_inc_return(v) == 0)
-
-#define atomic_dec_and_test(v)	(atomic_dec_return(v) == 0)
-
-#define atomic_sub_and_test(i,v)	(atomic_sub_return((i),(v)) == 0)
-
 #define ATOMIC_INIT(i)	{ (i) }
 
 #ifdef CONFIG_64BIT
@@ -246,12 +230,6 @@ atomic64_read(const atomic64_t *v)
 #define atomic64_inc_return(v)		(atomic64_add_return(   1,(v)))
 #define atomic64_dec_return(v)		(atomic64_add_return(  -1,(v)))
 
-#define atomic64_add_negative(a, v)	(atomic64_add_return((a), (v)) < 0)
-
-#define atomic64_inc_and_test(v) 	(atomic64_inc_return(v) == 0)
-#define atomic64_dec_and_test(v)	(atomic64_dec_return(v) == 0)
-#define atomic64_sub_and_test(i,v)	(atomic64_sub_return((i),(v)) == 0)
-
 /* exported interface */
 #define atomic64_cmpxchg(v, o, n) \
 	((__typeof__((v)->counter))cmpxchg(&((v)->counter), (o), (n)))
arch/powerpc/include/asm/atomic.h +0 -30
index 233dbf31911c..5d76f05d2be3 100644
@@ -129,8 +129,6 @@ ATOMIC_OPS(xor, xor)
 #undef ATOMIC_OP_RETURN_RELAXED
 #undef ATOMIC_OP
 
-#define atomic_add_negative(a, v)	(atomic_add_return((a), (v)) < 0)
-
 static __inline__ void atomic_inc(atomic_t *v)
 {
 	int t;
@@ -163,16 +161,6 @@ static __inline__ int atomic_inc_return_relaxed(atomic_t *v)
 	return t;
 }
 
-/*
- * atomic_inc_and_test - increment and test
- * @v: pointer of type atomic_t
- *
- * Atomically increments @v by 1
- * and returns true if the result is zero, or false for all
- * other cases.
- */
-#define atomic_inc_and_test(v) (atomic_inc_return(v) == 0)
-
 static __inline__ void atomic_dec(atomic_t *v)
 {
 	int t;
@@ -281,9 +269,6 @@ static __inline__ int atomic_inc_not_zero(atomic_t *v)
 }
 #define atomic_inc_not_zero(v) atomic_inc_not_zero((v))
 
-#define atomic_sub_and_test(a, v)	(atomic_sub_return((a), (v)) == 0)
-#define atomic_dec_and_test(v)		(atomic_dec_return((v)) == 0)
-
 /*
  * Atomically test *v and decrement if it is greater than 0.
  * The function returns the old value of *v minus 1, even if
@@ -413,8 +398,6 @@ ATOMIC64_OPS(xor, xor)
 #undef ATOMIC64_OP_RETURN_RELAXED
 #undef ATOMIC64_OP
 
-#define atomic64_add_negative(a, v)	(atomic64_add_return((a), (v)) < 0)
-
 static __inline__ void atomic64_inc(atomic64_t *v)
 {
 	long t;
@@ -445,16 +428,6 @@ static __inline__ long atomic64_inc_return_relaxed(atomic64_t *v)
 	return t;
 }
 
-/*
- * atomic64_inc_and_test - increment and test
- * @v: pointer of type atomic64_t
- *
- * Atomically increments @v by 1
- * and returns true if the result is zero, or false for all
- * other cases.
- */
-#define atomic64_inc_and_test(v) (atomic64_inc_return(v) == 0)
-
 static __inline__ void atomic64_dec(atomic64_t *v)
 {
 	long t;
@@ -488,9 +461,6 @@ static __inline__ long atomic64_dec_return_relaxed(atomic64_t *v)
 #define atomic64_inc_return_relaxed atomic64_inc_return_relaxed
 #define atomic64_dec_return_relaxed atomic64_dec_return_relaxed
 
-#define atomic64_sub_and_test(a, v)	(atomic64_sub_return((a), (v)) == 0)
-#define atomic64_dec_and_test(v)	(atomic64_dec_return((v)) == 0)
-
 /*
  * Atomically test *v and decrement if it is greater than 0.
  * The function returns the old value of *v minus 1.
arch/riscv/include/asm/atomic.h +0 -46

58 lines not on this page · raw message

arch/s390/include/asm/atomic.h +0 -8
index eb9329741bad..7f5fbd595f01 100644
@@ -55,17 +55,13 @@ static inline void atomic_add(int i, atomic_t *v)
 	__atomic_add(i, &v->counter);
 }
 
-#define atomic_add_negative(_i, _v)	(atomic_add_return(_i, _v) < 0)
 #define atomic_inc(_v)			atomic_add(1, _v)
 #define atomic_inc_return(_v)		atomic_add_return(1, _v)
-#define atomic_inc_and_test(_v)		(atomic_add_return(1, _v) == 0)
 #define atomic_sub(_i, _v)		atomic_add(-(int)(_i), _v)
 #define atomic_sub_return(_i, _v)	atomic_add_return(-(int)(_i), _v)
 #define atomic_fetch_sub(_i, _v)	atomic_fetch_add(-(int)(_i), _v)
-#define atomic_sub_and_test(_i, _v)	(atomic_sub_return(_i, _v) == 0)
 #define atomic_dec(_v)			atomic_sub(1, _v)
 #define atomic_dec_return(_v)		atomic_sub_return(1, _v)
-#define atomic_dec_and_test(_v)		(atomic_sub_return(1, _v) == 0)
 
 #define ATOMIC_OPS(op)							\
 static inline void atomic_##op(int i, atomic_t *v)			\
@@ -170,16 +166,12 @@ static inline long atomic64_dec_if_positive(atomic64_t *v)
 	return dec;
 }
 
-#define atomic64_add_negative(_i, _v)	(atomic64_add_return(_i, _v) < 0)
 #define atomic64_inc(_v)		atomic64_add(1, _v)
 #define atomic64_inc_return(_v)		atomic64_add_return(1, _v)
-#define atomic64_inc_and_test(_v)	(atomic64_add_return(1, _v) == 0)
 #define atomic64_sub_return(_i, _v)	atomic64_add_return(-(long)(_i), _v)
 #define atomic64_fetch_sub(_i, _v)	atomic64_fetch_add(-(long)(_i), _v)
 #define atomic64_sub(_i, _v)		atomic64_add(-(long)(_i), _v)
-#define atomic64_sub_and_test(_i, _v)	(atomic64_sub_return(_i, _v) == 0)
 #define atomic64_dec(_v)		atomic64_sub(1, _v)
 #define atomic64_dec_return(_v)		atomic64_sub_return(1, _v)
-#define atomic64_dec_and_test(_v)	(atomic64_sub_return(1, _v) == 0)
 
 #endif /* __ARCH_S390_ATOMIC__  */
arch/sh/include/asm/atomic.h +0 -4
index 422fac764ca1..d438494fa112 100644
@@ -32,12 +32,8 @@
 #include <asm/atomic-irq.h>
 #endif
 
-#define atomic_add_negative(a, v)	(atomic_add_return((a), (v)) < 0)
 #define atomic_dec_return(v)		atomic_sub_return(1, (v))
 #define atomic_inc_return(v)		atomic_add_return(1, (v))
-#define atomic_inc_and_test(v)		(atomic_inc_return(v) == 0)
-#define atomic_sub_and_test(i,v)	(atomic_sub_return((i), (v)) == 0)
-#define atomic_dec_and_test(v)		(atomic_sub_return(1, (v)) == 0)
 
 #define atomic_inc(v)			atomic_add(1, (v))
 #define atomic_dec(v)			atomic_sub(1, (v))
arch/sparc/include/asm/atomic_32.h +0 -15
index 9d7a15acc0c5..3a26573790c6 100644
@@ -51,19 +51,4 @@ void atomic_set(atomic_t *, int);
 #define atomic_inc_return(v)	(atomic_add_return(        1, (v)))
 #define atomic_dec_return(v)	(atomic_add_return(       -1, (v)))
 
-#define atomic_add_negative(a, v)	(atomic_add_return((a), (v)) < 0)
-
-/*
- * atomic_inc_and_test - increment and test
- * @v: pointer of type atomic_t
- *
- * Atomically increments @v by 1
- * and returns true if the result is zero, or false for all
- * other cases.
- */
-#define atomic_inc_and_test(v) (atomic_inc_return(v) == 0)
-
-#define atomic_dec_and_test(v) (atomic_dec_return(v) == 0)
-#define atomic_sub_and_test(i, v) (atomic_sub_return(i, v) == 0)
-
 #endif /* !(__ARCH_SPARC_ATOMIC__) */
arch/sparc/include/asm/atomic_64.h +0 -20
index 458783e99997..634508282aea 100644
@@ -56,32 +56,12 @@ ATOMIC_OPS(xor)
 #define atomic_inc_return(v)   atomic_add_return(1, v)
 #define atomic64_inc_return(v) atomic64_add_return(1, v)
 
-/*
- * atomic_inc_and_test - increment and test
- * @v: pointer of type atomic_t
- *
- * Atomically increments @v by 1
- * and returns true if the result is zero, or false for all
- * other cases.
- */
-#define atomic_inc_and_test(v) (atomic_inc_return(v) == 0)
-#define atomic64_inc_and_test(v) (atomic64_inc_return(v) == 0)
-
-#define atomic_sub_and_test(i, v) (atomic_sub_return(i, v) == 0)
-#define atomic64_sub_and_test(i, v) (atomic64_sub_return(i, v) == 0)
-
-#define atomic_dec_and_test(v) (atomic_sub_return(1, v) == 0)
-#define atomic64_dec_and_test(v) (atomic64_sub_return(1, v) == 0)
-
 #define atomic_inc(v) atomic_add(1, v)
 #define atomic64_inc(v) atomic64_add(1, v)
 
 #define atomic_dec(v) atomic_sub(1, v)
 #define atomic64_dec(v) atomic64_sub(1, v)
 
-#define atomic_add_negative(i, v) (atomic_add_return(i, v) < 0)
-#define atomic64_add_negative(i, v) (atomic64_add_return(i, v) < 0)
-
 #define atomic_cmpxchg(v, o, n) (cmpxchg(&((v)->counter), (o), (n)))
 
 static inline int atomic_xchg(atomic_t *v, int new)
arch/x86/include/asm/atomic.h +4 -0
index 616327ac9d39..73bda4abe180 100644
@@ -80,6 +80,7 @@ static __always_inline void arch_atomic_sub(int i, atomic_t *v)
  * true if the result is zero, or false for all
  * other cases.
  */
+#define arch_atomic_sub_and_test arch_atomic_sub_and_test
 static __always_inline bool arch_atomic_sub_and_test(int i, atomic_t *v)
 {
 	GEN_BINARY_RMWcc(LOCK_PREFIX "subl", v->counter, "er", i, "%0", e);
@@ -117,6 +118,7 @@ static __always_inline void arch_atomic_dec(atomic_t *v)
  * returns true if the result is 0, or false for all other
  * cases.
  */
+#define arch_atomic_dec_and_test arch_atomic_dec_and_test
 static __always_inline bool arch_atomic_dec_and_test(atomic_t *v)
 {
 	GEN_UNARY_RMWcc(LOCK_PREFIX "decl", v->counter, "%0", e);
@@ -130,6 +132,7 @@ static __always_inline bool arch_atomic_dec_and_test(atomic_t *v)
  * and returns true if the result is zero, or false for all
  * other cases.
  */
+#define arch_atomic_inc_and_test arch_atomic_inc_and_test
 static __always_inline bool arch_atomic_inc_and_test(atomic_t *v)
 {
 	GEN_UNARY_RMWcc(LOCK_PREFIX "incl", v->counter, "%0", e);
@@ -144,6 +147,7 @@ static __always_inline bool arch_atomic_inc_and_test(atomic_t *v)
  * if the result is negative, or false when
  * result is greater than or equal to zero.
  */
+#define arch_atomic_add_negative arch_atomic_add_negative
 static __always_inline bool arch_atomic_add_negative(int i, atomic_t *v)
 {
 	GEN_BINARY_RMWcc(LOCK_PREFIX "addl", v->counter, "er", i, "%0", s);
arch/x86/include/asm/atomic64_32.h +0 -54

66 lines not on this page · raw message

arch/x86/include/asm/atomic64_64.h +4 -0
index 7e04b294e6eb..6a65228a3db6 100644
@@ -71,6 +71,7 @@ static inline void arch_atomic64_sub(long i, atomic64_t *v)
  * true if the result is zero, or false for all
  * other cases.
  */
+#define arch_atomic64_sub_and_test arch_atomic64_sub_and_test
 static inline bool arch_atomic64_sub_and_test(long i, atomic64_t *v)
 {
 	GEN_BINARY_RMWcc(LOCK_PREFIX "subq", v->counter, "er", i, "%0", e);
@@ -110,6 +111,7 @@ static __always_inline void arch_atomic64_dec(atomic64_t *v)
  * returns true if the result is 0, or false for all other
  * cases.
  */
+#define arch_atomic64_dec_and_test arch_atomic64_dec_and_test
 static inline bool arch_atomic64_dec_and_test(atomic64_t *v)
 {
 	GEN_UNARY_RMWcc(LOCK_PREFIX "decq", v->counter, "%0", e);
@@ -123,6 +125,7 @@ static inline bool arch_atomic64_dec_and_test(atomic64_t *v)
  * and returns true if the result is zero, or false for all
  * other cases.
  */
+#define arch_atomic64_inc_and_test arch_atomic64_inc_and_test
 static inline bool arch_atomic64_inc_and_test(atomic64_t *v)
 {
 	GEN_UNARY_RMWcc(LOCK_PREFIX "incq", v->counter, "%0", e);
@@ -137,6 +140,7 @@ static inline bool arch_atomic64_inc_and_test(atomic64_t *v)
  * if the result is negative, or false when
  * result is greater than or equal to zero.
  */
+#define arch_atomic64_add_negative arch_atomic64_add_negative
 static inline bool arch_atomic64_add_negative(long i, atomic64_t *v)
 {
 	GEN_BINARY_RMWcc(LOCK_PREFIX "addq", v->counter, "er", i, "%0", s);
arch/xtensa/include/asm/atomic.h +0 -42

54 lines not on this page · raw message

include/asm-generic/atomic-instrumented.h +24 -0
index 719168de5c6e..b59edd520e39 100644
@@ -225,29 +225,41 @@ static __always_inline s64 atomic64_dec_if_positive(atomic64_t *v)
 	return arch_atomic64_dec_if_positive(v);
 }
 
+#ifdef arch_atomic_dec_and_test
+#define atomic_dec_and_test atomic_dec_and_test
 static __always_inline bool atomic_dec_and_test(atomic_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic_dec_and_test(v);
 }
+#endif
 
+#ifdef arch_atomic64_dec_and_test
+#define atomic64_dec_and_test atomic64_dec_and_test
 static __always_inline bool atomic64_dec_and_test(atomic64_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic64_dec_and_test(v);
 }
+#endif
 
+#ifdef arch_atomic_inc_and_test
+#define atomic_inc_and_test atomic_inc_and_test
 static __always_inline bool atomic_inc_and_test(atomic_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic_inc_and_test(v);
 }
+#endif
 
+#ifdef arch_atomic64_inc_and_test
+#define atomic64_inc_and_test atomic64_inc_and_test
 static __always_inline bool atomic64_inc_and_test(atomic64_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic64_inc_and_test(v);
 }
+#endif
 
 static __always_inline int atomic_add_return(int i, atomic_t *v)
 {
@@ -333,29 +345,41 @@ static __always_inline s64 atomic64_fetch_xor(s64 i, atomic64_t *v)
 	return arch_atomic64_fetch_xor(i, v);
 }
 
+#ifdef arch_atomic_sub_and_test
+#define atomic_sub_and_test atomic_sub_and_test
 static __always_inline bool atomic_sub_and_test(int i, atomic_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic_sub_and_test(i, v);
 }
+#endif
 
+#ifdef arch_atomic64_sub_and_test
+#define atomic64_sub_and_test atomic64_sub_and_test
 static __always_inline bool atomic64_sub_and_test(s64 i, atomic64_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic64_sub_and_test(i, v);
 }
+#endif
 
+#ifdef arch_atomic_add_negative
+#define atomic_add_negative atomic_add_negative
 static __always_inline bool atomic_add_negative(int i, atomic_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic_add_negative(i, v);
 }
+#endif
 
+#ifdef arch_atomic64_add_negative
+#define atomic64_add_negative atomic64_add_negative
 static __always_inline bool atomic64_add_negative(s64 i, atomic64_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic64_add_negative(i, v);
 }
+#endif
 
 static __always_inline unsigned long
 cmpxchg_size(volatile void *ptr, unsigned long old, unsigned long new, int size)
include/asm-generic/atomic.h +0 -9
index 757e45821220..40cab858aaaa 100644
@@ -186,11 +186,6 @@ ATOMIC_OP(xor, ^)
 
 #include <linux/irqflags.h>
 
-static inline int atomic_add_negative(int i, atomic_t *v)
-{
-	return atomic_add_return(i, v) < 0;
-}
-
 static inline void atomic_add(int i, atomic_t *v)
 {
 	atomic_add_return(i, v);
@@ -214,10 +209,6 @@ static inline void atomic_dec(atomic_t *v)
 #define atomic_dec_return(v)		atomic_sub_return(1, (v))
 #define atomic_inc_return(v)		atomic_add_return(1, (v))
 
-#define atomic_sub_and_test(i, v)	(atomic_sub_return((i), (v)) == 0)
-#define atomic_dec_and_test(v)		(atomic_dec_return(v) == 0)
-#define atomic_inc_and_test(v)		(atomic_inc_return(v) == 0)
-
 #define atomic_xchg(ptr, v)		(xchg(&(ptr)->counter, (v)))
 #define atomic_cmpxchg(v, old, new)	(cmpxchg(&((v)->counter), (old), (new)))
 
include/asm-generic/atomic64.h +0 -4
index b0aa425361ad..15fe7cca29ab 100644
@@ -55,13 +55,9 @@ extern long long atomic64_xchg(atomic64_t *v, long long new);
 extern long long atomic64_fetch_add_unless(atomic64_t *v, long long a, long long u);
 #define atomic64_fetch_add_unless atomic64_fetch_add_unless
 
-#define atomic64_add_negative(a, v)	(atomic64_add_return((a), (v)) < 0)
 #define atomic64_inc(v)			atomic64_add(1LL, (v))
 #define atomic64_inc_return(v)		atomic64_add_return(1LL, (v))
-#define atomic64_inc_and_test(v) 	(atomic64_inc_return(v) == 0)
-#define atomic64_sub_and_test(a, v)	(atomic64_sub_return((a), (v)) == 0)
 #define atomic64_dec(v)			atomic64_sub(1LL, (v))
 #define atomic64_dec_return(v)		atomic64_sub_return(1LL, (v))
-#define atomic64_dec_and_test(v)	(atomic64_dec_return((v)) == 0)
 
 #endif  /*  _ASM_GENERIC_ATOMIC64_H  */
include/linux/atomic.h +124 -0

136 lines not on this page · raw message

Signature · 1 line
2.11.0
Acked-by on v2 13/16
#25 lore raw

Re: [PATCHv2 13/16] atomics/treewide: make test ops optional

Quoted 110 lines · On Tue, 29 May 2018 08:43:43 PDT (-0700), [email protected] wrote:

Some of the atomics return the result of a test applied after the atomic operation, and almost all architectures implement these as trivial wrappers around the underlying atomic. Specifically:

* <atomic>_inc_and_test(v) is (<atomic>_inc_return(v) == 0)

* <atomic>_dec_and_test(v) is (<atomic>_dec_return(v) == 0)

* <atomic>_sub_and_test(i, v) is (<atomic>_sub_return(i, v) == 0)

* <atomic>_add_negative(i, v) is (<atomic>_add_return(i, v) < 0)

Rather than have these definitions duplicated in all architectures, with minor inconsistencies in formatting and documentation, let's make these operations optional, with default fallbacks as above. Implementations must now provide a preprocessor symbol.

The instrumented atomics are updated accordingly.

Both x86 and m68k have custom implementations, which are left as-is, given preprocessor symbols to avoid being overridden.

There should be no functional change as a result of this patch.

---
 arch/alpha/include/asm/atomic.h           |  12 ---
 arch/arc/include/asm/atomic.h             |  10 ---
 arch/arm/include/asm/atomic.h             |   9 ---
 arch/arm64/include/asm/atomic.h           |   8 --
 arch/h8300/include/asm/atomic.h           |   5 --
 arch/hexagon/include/asm/atomic.h         |   5 --
 arch/ia64/include/asm/atomic.h            |  23 ------
 arch/m68k/include/asm/atomic.h            |   4 +
 arch/mips/include/asm/atomic.h            |  84 --------------------
 arch/parisc/include/asm/atomic.h          |  22 ------
 arch/powerpc/include/asm/atomic.h         |  30 --------
 arch/riscv/include/asm/atomic.h           |  46 -----------
 arch/s390/include/asm/atomic.h            |   8 --
 arch/sh/include/asm/atomic.h              |   4 -
 arch/sparc/include/asm/atomic_32.h        |  15 ----
 arch/sparc/include/asm/atomic_64.h        |  20 -----
 arch/x86/include/asm/atomic.h             |   4 +
 arch/x86/include/asm/atomic64_32.h        |  54 -------------
 arch/x86/include/asm/atomic64_64.h        |   4 +
 arch/xtensa/include/asm/atomic.h          |  42 ----------
 include/asm-generic/atomic-instrumented.h |  24 ++++++
 include/asm-generic/atomic.h              |   9 ---
 include/asm-generic/atomic64.h            |   4 -
 include/linux/atomic.h                    | 124 ++++++++++++++++++++++++++++++
 24 files changed, 160 insertions(+), 410 deletions(-)

[...]

arch/riscv/include/asm/atomic.h +0 -39
index d959bbaaad41..68eef0a805ca 100644
@@ -209,36 +209,6 @@ ATOMIC_OPS(xor, xor, i)
 #undef ATOMIC_FETCH_OP
 #undef ATOMIC_OP_RETURN
 
-/*
- * The extra atomic operations that are constructed from one of the core
- * AMO-based operations above (aside from sub, which is easier to fit above).
- * These are required to perform a full barrier, but they're OK this way
- * because atomic_*_return is also required to perform a full barrier.
- *
- */
-#define ATOMIC_OP(op, func_op, comp_op, I, c_type, prefix)		\
-static __always_inline							\
-bool atomic##prefix##_##op(c_type i, atomic##prefix##_t *v)		\
-{									\
-	return atomic##prefix##_##func_op##_return(i, v) comp_op I;	\
-}
-
-#ifdef CONFIG_GENERIC_ATOMIC64
-#define ATOMIC_OPS(op, func_op, comp_op, I)				\
-        ATOMIC_OP(op, func_op, comp_op, I,  int,   )
-#else
-#define ATOMIC_OPS(op, func_op, comp_op, I)				\
-        ATOMIC_OP(op, func_op, comp_op, I,  int,   )			\
-        ATOMIC_OP(op, func_op, comp_op, I, long, 64)
-#endif
-
-ATOMIC_OPS(add_and_test, add, ==, 0)
-ATOMIC_OPS(sub_and_test, sub, ==, 0)
-ATOMIC_OPS(add_negative, add,  <, 0)
-
-#undef ATOMIC_OP
-#undef ATOMIC_OPS
-
 #define ATOMIC_OP(op, func_op, I, c_type, prefix)			\
 static __always_inline							\
 void atomic##prefix##_##op(atomic##prefix##_t *v)			\
@@ -315,22 +285,6 @@ ATOMIC_OPS(dec, add, +, -1)
 #undef ATOMIC_FETCH_OP
 #undef ATOMIC_OP_RETURN
 
-#define ATOMIC_OP(op, func_op, comp_op, I, prefix)			\
-static __always_inline							\
-bool atomic##prefix##_##op(atomic##prefix##_t *v)			\
-{									\
-	return atomic##prefix##_##func_op##_return(v) comp_op I;	\
-}
-
-ATOMIC_OP(inc_and_test, inc, ==, 0,   )
-ATOMIC_OP(dec_and_test, dec, ==, 0,   )

arch/riscv/include/asm/atomic.h ATOMIC_OPS(dec, add, +, -1) · line 290

Quoted 10 lines
arch/riscv/include/asm/atomic.h +0 -7
-#ifndef CONFIG_GENERIC_ATOMIC64
-ATOMIC_OP(inc_and_test, inc, ==, 0, 64)
-ATOMIC_OP(dec_and_test, dec, ==, 0, 64)
-#endif
-
-#undef ATOMIC_OP
-
 /* This is required to provide a full barrier on success. */
 static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {
author reviewer
#26 lore raw

PATCH v2 14/16 atomics/treewide: remove atomic_inc_not_zero_hint()

In reply to Mark Rutland

While documentation suggests atomic_inc_not_zero_hint() will perform better than atomic_inc_not_zero(), this is unlikely to be the case. No architectures implement atomic_inc_not_zero_hint() directly, and thus it either falls back to atomic_inc_not_zero(), or a loop using atomic_cmpxchg().

Whenever the hint does not match the value in memory, the repeated use of atomic_cmpxchg() will be more expensive than the read that atomic_inc_not_zero_hint() attempts to avoid. For architectures with LL/SC atomics, a read cannot be avoided, and it would always be better to use atomic_inc_not_zero() directly. For other architectures, their own atomic_inc_not_zero() is likely to be more optimal than an atomic_cmpxchg() loop regardless.

Generally, atomic_inc_not_zero_hint() is liable to perform worse than atomic_inc_not_zero(). Further, atomic_inc_not_zero_hint() only exists for atomic_t, and not atomic64_t or atomic_long_t, and there is only one user in the kernel tree.

Given all this, let's remove atomic_inc_not_zero_hint(), and migrate the existing user over to atomic_inc_not_zero().

There should be no functional change as a result of this patch.

Cc Boqun Feng <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Will Deacon <[email protected]>
---
 include/linux/atomic.h | 32 --------------------------------
 net/atm/pppoatm.c      |  2 +-
 2 files changed, 1 insertion(+), 33 deletions(-)
include/linux/atomic.h +0 -32
index 7d41b6fffab6..8ec4d120b264 100644
@@ -656,38 +656,6 @@ static inline int atomic_fetch_andnot_release(int i, atomic_t *v)
 }
 #endif
 
-/**
- * atomic_inc_not_zero_hint - increment if not null
- * @v: pointer of type atomic_t
- * @hint: probable value of the atomic before the increment
- *
- * This version of atomic_inc_not_zero() gives a hint of probable
- * value of the atomic. This helps processor to not read the memory
- * before doing the atomic read/modify/write cycle, lowering
- * number of bus transactions on some arches.
- *
- * Returns: 0 if increment was not done, 1 otherwise.
- */
-#ifndef atomic_inc_not_zero_hint
-static inline int atomic_inc_not_zero_hint(atomic_t *v, int hint)
-{
-	int val, c = hint;
-
-	/* sanity test, should be removed by compiler if hint is a constant */
-	if (!hint)
-		return atomic_inc_not_zero(v);
-
-	do {
-		val = atomic_cmpxchg(v, c, c + 1);
-		if (val == c)
-			return 1;
-		c = val;
-	} while (c);
-
-	return 0;
-}
-#endif
-
 #ifndef atomic_inc_unless_negative
 static inline int atomic_inc_unless_negative(atomic_t *p)
 {
net/atm/pppoatm.c +1 -1
index 21d9d341a619..a4d82afd4586 100644
@@ -244,7 +244,7 @@ static int pppoatm_may_send(struct pppoatm_vcc *pvcc, int size)
 	 * the packet count limit, so...
 	 */
 	if (atm_may_send(pvcc->atmvcc, size) &&
-	    atomic_inc_not_zero_hint(&pvcc->inflight, NONE_INFLIGHT))
+	    atomic_inc_not_zero(&pvcc->inflight))
 		return 1;
 
 	/*
Signature · 1 line
2.11.0
author reviewer
#27 lore raw

PATCH v2 15/16 atomics/treewide: make unconditional inc/dec ops optional

In reply to Mark Rutland

Many of the inc/dec ops are mandatory, but for most architectures inc/dec are simply trivial wrappers around their corresponding add/sub ops.

Let's make all the inc/dec ops optional, so that we can get rid of these boilerplate wrappers.

The instrumented atomics are updated accordingly.

There should be no functional change as a result of this patch.

Cc Boqun Feng <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Will Deacon <[email protected]>
---
 arch/alpha/include/asm/atomic.h           | 12 -----
 arch/arc/include/asm/atomic.h             | 11 -----
 arch/arm/include/asm/atomic.h             | 11 -----
 arch/arm64/include/asm/atomic.h           | 24 ----------
 arch/h8300/include/asm/atomic.h           |  7 ---
 arch/hexagon/include/asm/atomic.h         |  6 ---
 arch/ia64/include/asm/atomic.h            |  9 ----
 arch/m68k/include/asm/atomic.h            |  5 +-
 arch/mips/include/asm/atomic.h            | 38 ----------------
 arch/parisc/include/asm/atomic.h          | 12 -----
 arch/powerpc/include/asm/atomic.h         |  4 ++
 arch/riscv/include/asm/atomic.h           | 76 -------------------------------
 arch/s390/include/asm/atomic.h            |  8 ----
 arch/sh/include/asm/atomic.h              |  6 ---
 arch/sparc/include/asm/atomic_32.h        |  5 --
 arch/sparc/include/asm/atomic_64.h        | 12 -----
 arch/x86/include/asm/atomic.h             |  5 +-
 arch/x86/include/asm/atomic64_32.h        |  4 ++
 arch/x86/include/asm/atomic64_64.h        |  5 +-
 arch/xtensa/include/asm/atomic.h          | 32 -------------
 include/asm-generic/atomic-instrumented.h | 24 ++++++++++
 include/asm-generic/atomic.h              | 13 ------
 include/asm-generic/atomic64.h            |  5 --
 include/linux/atomic.h                    | 48 +++++++++++++++++++
 24 files changed, 86 insertions(+), 296 deletions(-)
arch/alpha/include/asm/atomic.h +0 -12
index 31a4904dc8fc..9c50cc512a78 100644
@@ -297,16 +297,4 @@ static inline long atomic64_dec_if_positive(atomic64_t *v)
 	return old - 1;
 }
 
-#define atomic_dec_return(v) atomic_sub_return(1,(v))
-#define atomic64_dec_return(v) atomic64_sub_return(1,(v))
-
-#define atomic_inc_return(v) atomic_add_return(1,(v))
-#define atomic64_inc_return(v) atomic64_add_return(1,(v))
-
-#define atomic_inc(v) atomic_add(1,(v))
-#define atomic64_inc(v) atomic64_add(1,(v))
-
-#define atomic_dec(v) atomic_sub(1,(v))
-#define atomic64_dec(v) atomic64_sub(1,(v))
-
 #endif /* _ALPHA_ATOMIC_H */
arch/arc/include/asm/atomic.h +0 -11
index 555b0a94b643..abffdc525b46 100644
@@ -308,12 +308,6 @@ ATOMIC_OPS(xor, ^=, CTOP_INST_AXOR_DI_R2_R2_R3)
 #undef ATOMIC_OP_RETURN
 #undef ATOMIC_OP
 
-#define atomic_inc(v)			atomic_add(1, v)
-#define atomic_dec(v)			atomic_sub(1, v)
-
-#define atomic_inc_return(v)		atomic_add_return(1, (v))
-#define atomic_dec_return(v)		atomic_sub_return(1, (v))
-
 #ifdef CONFIG_GENERIC_ATOMIC64
 
 #include <asm-generic/atomic64.h>
@@ -560,11 +554,6 @@ static inline long long atomic64_fetch_add_unless(atomic64_t *v, long long a,
 }
 #define atomic64_fetch_add_unless atomic64_fetch_add_unless
 
-#define atomic64_inc(v)			atomic64_add(1LL, (v))
-#define atomic64_inc_return(v)		atomic64_add_return(1LL, (v))
-#define atomic64_dec(v)			atomic64_sub(1LL, (v))
-#define atomic64_dec_return(v)		atomic64_sub_return(1LL, (v))
-
 #endif	/* !CONFIG_GENERIC_ATOMIC64 */
 
 #endif	/* !__ASSEMBLY__ */
arch/arm/include/asm/atomic.h +0 -11
index 35fb7f504daa..5a58d061d3d2 100644
@@ -245,12 +245,6 @@ ATOMIC_OPS(xor, ^=, eor)
 
 #define atomic_xchg(v, new) (xchg(&((v)->counter), new))
 
-#define atomic_inc(v)		atomic_add(1, v)
-#define atomic_dec(v)		atomic_sub(1, v)
-
-#define atomic_inc_return_relaxed(v)    (atomic_add_return_relaxed(1, v))
-#define atomic_dec_return_relaxed(v)    (atomic_sub_return_relaxed(1, v))
-
 #ifndef CONFIG_GENERIC_ATOMIC64
 typedef struct {
 	long long counter;
@@ -512,11 +506,6 @@ static inline long long atomic64_fetch_add_unless(atomic64_t *v, long long a,
 }
 #define atomic64_fetch_add_unless atomic64_fetch_add_unless
 
-#define atomic64_inc(v)			atomic64_add(1LL, (v))
-#define atomic64_inc_return_relaxed(v)	atomic64_add_return_relaxed(1LL, (v))
-#define atomic64_dec(v)			atomic64_sub(1LL, (v))
-#define atomic64_dec_return_relaxed(v)	atomic64_sub_return_relaxed(1LL, (v))
-
 #endif /* !CONFIG_GENERIC_ATOMIC64 */
 #endif
 #endif
arch/arm64/include/asm/atomic.h +0 -24
index edbe53fa3106..078f785cd97f 100644
@@ -50,21 +50,11 @@
 #define atomic_add_return_release	atomic_add_return_release
 #define atomic_add_return		atomic_add_return
 
-#define atomic_inc_return_relaxed(v)	atomic_add_return_relaxed(1, (v))
-#define atomic_inc_return_acquire(v)	atomic_add_return_acquire(1, (v))
-#define atomic_inc_return_release(v)	atomic_add_return_release(1, (v))
-#define atomic_inc_return(v)		atomic_add_return(1, (v))
-
 #define atomic_sub_return_relaxed	atomic_sub_return_relaxed
 #define atomic_sub_return_acquire	atomic_sub_return_acquire
 #define atomic_sub_return_release	atomic_sub_return_release
 #define atomic_sub_return		atomic_sub_return
 
-#define atomic_dec_return_relaxed(v)	atomic_sub_return_relaxed(1, (v))
-#define atomic_dec_return_acquire(v)	atomic_sub_return_acquire(1, (v))
-#define atomic_dec_return_release(v)	atomic_sub_return_release(1, (v))
-#define atomic_dec_return(v)		atomic_sub_return(1, (v))
-
 #define atomic_fetch_add_relaxed	atomic_fetch_add_relaxed
 #define atomic_fetch_add_acquire	atomic_fetch_add_acquire
 #define atomic_fetch_add_release	atomic_fetch_add_release
@@ -108,8 +98,6 @@
 	cmpxchg_release(&((v)->counter), (old), (new))
 #define atomic_cmpxchg(v, old, new)	cmpxchg(&((v)->counter), (old), (new))
 
-#define atomic_inc(v)			atomic_add(1, (v))
-#define atomic_dec(v)			atomic_sub(1, (v))
 #define atomic_andnot			atomic_andnot
 
 /*
@@ -124,21 +112,11 @@
 #define atomic64_add_return_release	atomic64_add_return_release
 #define atomic64_add_return		atomic64_add_return
 
-#define atomic64_inc_return_relaxed(v)	atomic64_add_return_relaxed(1, (v))
-#define atomic64_inc_return_acquire(v)	atomic64_add_return_acquire(1, (v))
-#define atomic64_inc_return_release(v)	atomic64_add_return_release(1, (v))
-#define atomic64_inc_return(v)		atomic64_add_return(1, (v))
-
 #define atomic64_sub_return_relaxed	atomic64_sub_return_relaxed
 #define atomic64_sub_return_acquire	atomic64_sub_return_acquire
 #define atomic64_sub_return_release	atomic64_sub_return_release
 #define atomic64_sub_return		atomic64_sub_return
 
-#define atomic64_dec_return_relaxed(v)	atomic64_sub_return_relaxed(1, (v))
-#define atomic64_dec_return_acquire(v)	atomic64_sub_return_acquire(1, (v))
-#define atomic64_dec_return_release(v)	atomic64_sub_return_release(1, (v))
-#define atomic64_dec_return(v)		atomic64_sub_return(1, (v))
-
 #define atomic64_fetch_add_relaxed	atomic64_fetch_add_relaxed
 #define atomic64_fetch_add_acquire	atomic64_fetch_add_acquire
 #define atomic64_fetch_add_release	atomic64_fetch_add_release
@@ -179,8 +157,6 @@
 #define atomic64_cmpxchg_release	atomic_cmpxchg_release
 #define atomic64_cmpxchg		atomic_cmpxchg
 
-#define atomic64_inc(v)			atomic64_add(1, (v))
-#define atomic64_dec(v)			atomic64_sub(1, (v))
 #define atomic64_andnot			atomic64_andnot
 
 #endif
arch/h8300/include/asm/atomic.h +0 -7
index b658b20d2232..532f3409f751 100644
@@ -69,13 +69,6 @@ ATOMIC_OPS(sub, -=)
 #undef ATOMIC_OP_RETURN
 #undef ATOMIC_OP
 
-#define atomic_inc_return(v)		atomic_add_return(1, v)
-#define atomic_dec_return(v)		atomic_sub_return(1, v)
-
-#define atomic_inc(v)			(void)atomic_inc_return(v)
-
-#define atomic_dec(v)			(void)atomic_dec_return(v)
-
 static inline int atomic_cmpxchg(atomic_t *v, int old, int new)
 {
 	int ret;
arch/hexagon/include/asm/atomic.h +0 -6
index 31638f511674..311b9894ccc8 100644
@@ -198,10 +198,4 @@ static inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 }
 #define atomic_fetch_add_unless atomic_fetch_add_unless
 
-#define atomic_inc(v) atomic_add(1, (v))
-#define atomic_dec(v) atomic_sub(1, (v))
-
-#define atomic_inc_return(v) (atomic_add_return(1, v))
-#define atomic_dec_return(v) (atomic_sub_return(1, v))
-
 #endif
arch/ia64/include/asm/atomic.h +0 -9
index e4143c462e65..46a15a974bed 100644
@@ -231,19 +231,10 @@ static __inline__ long atomic64_dec_if_positive(atomic64_t *v)
 	return dec;
 }
 
-#define atomic_dec_return(v)		atomic_sub_return(1, (v))
-#define atomic_inc_return(v)		atomic_add_return(1, (v))
-#define atomic64_dec_return(v)		atomic64_sub_return(1, (v))
-#define atomic64_inc_return(v)		atomic64_add_return(1, (v))
-
 #define atomic_add(i,v)			(void)atomic_add_return((i), (v))
 #define atomic_sub(i,v)			(void)atomic_sub_return((i), (v))
-#define atomic_inc(v)			atomic_add(1, (v))
-#define atomic_dec(v)			atomic_sub(1, (v))
 
 #define atomic64_add(i,v)		(void)atomic64_add_return((i), (v))
 #define atomic64_sub(i,v)		(void)atomic64_sub_return((i), (v))
-#define atomic64_inc(v)			atomic64_add(1, (v))
-#define atomic64_dec(v)			atomic64_sub(1, (v))
 
 #endif /* _ASM_IA64_ATOMIC_H */
arch/m68k/include/asm/atomic.h +2 -3
index 9df09c876fa2..47228b0d4163 100644
@@ -126,11 +126,13 @@ static inline void atomic_inc(atomic_t *v)
 {
 	__asm__ __volatile__("addql #1,%0" : "+m" (*v));
 }
+#define atomic_inc atomic_inc
 
 static inline void atomic_dec(atomic_t *v)
 {
 	__asm__ __volatile__("subql #1,%0" : "+m" (*v));
 }
+#define atomic_dec atomic_dec
 
 static inline int atomic_dec_and_test(atomic_t *v)
 {
@@ -192,9 +194,6 @@ static inline int atomic_xchg(atomic_t *v, int new)
 
 #endif /* !CONFIG_RMW_INSNS */
 
-#define atomic_dec_return(v)	atomic_sub_return(1, (v))
-#define atomic_inc_return(v)	atomic_add_return(1, (v))
-
 static inline int atomic_sub_and_test(int i, atomic_t *v)
 {
 	char c;
arch/mips/include/asm/atomic.h +0 -38
index fd3008ae164c..79be687de4ab 100644
@@ -274,31 +274,12 @@ static __inline__ int atomic_sub_if_positive(int i, atomic_t * v)
 #define atomic_cmpxchg(v, o, n) (cmpxchg(&((v)->counter), (o), (n)))
 #define atomic_xchg(v, new) (xchg(&((v)->counter), (new)))
 
-#define atomic_dec_return(v) atomic_sub_return(1, (v))
-#define atomic_inc_return(v) atomic_add_return(1, (v))
-
 /*
  * atomic_dec_if_positive - decrement by 1 if old value positive
  * @v: pointer of type atomic_t
  */
 #define atomic_dec_if_positive(v)	atomic_sub_if_positive(1, v)
 
-/*
- * atomic_inc - increment atomic variable
- * @v: pointer of type atomic_t
- *
- * Atomically increments @v by 1.
- */
-#define atomic_inc(v) atomic_add(1, (v))
-
-/*
- * atomic_dec - decrement and test
- * @v: pointer of type atomic_t
- *
- * Atomically decrements @v by 1.
- */
-#define atomic_dec(v) atomic_sub(1, (v))
-
 #ifdef CONFIG_64BIT
 
 #define ATOMIC64_INIT(i)    { (i) }
@@ -554,31 +535,12 @@ static __inline__ long atomic64_sub_if_positive(long i, atomic64_t * v)
 	((__typeof__((v)->counter))cmpxchg(&((v)->counter), (o), (n)))
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), (new)))
 
-#define atomic64_dec_return(v) atomic64_sub_return(1, (v))
-#define atomic64_inc_return(v) atomic64_add_return(1, (v))
-
 /*
  * atomic64_dec_if_positive - decrement by 1 if old value positive
  * @v: pointer of type atomic64_t
  */
 #define atomic64_dec_if_positive(v)	atomic64_sub_if_positive(1, v)
 
-/*
- * atomic64_inc - increment atomic variable
- * @v: pointer of type atomic64_t
- *
- * Atomically increments @v by 1.
- */
-#define atomic64_inc(v) atomic64_add(1, (v))
-
-/*
- * atomic64_dec - decrement and test
- * @v: pointer of type atomic64_t
- *
- * Atomically decrements @v by 1.
- */
-#define atomic64_dec(v) atomic64_sub(1, (v))
-
 #endif /* CONFIG_64BIT */
 
 #endif /* _ASM_ATOMIC_H */
arch/parisc/include/asm/atomic.h +0 -12
index f85844ff6336..10bc490327c1 100644
@@ -136,12 +136,6 @@ ATOMIC_OPS(xor, ^=)
 #undef ATOMIC_OP_RETURN
 #undef ATOMIC_OP
 
-#define atomic_inc(v)	(atomic_add(   1,(v)))
-#define atomic_dec(v)	(atomic_add(  -1,(v)))
-
-#define atomic_inc_return(v)	(atomic_add_return(   1,(v)))
-#define atomic_dec_return(v)	(atomic_add_return(  -1,(v)))
-
 #define ATOMIC_INIT(i)	{ (i) }
 
 #ifdef CONFIG_64BIT
@@ -224,12 +218,6 @@ atomic64_read(const atomic64_t *v)
 	return READ_ONCE((v)->counter);
 }
 
-#define atomic64_inc(v)		(atomic64_add(   1,(v)))
-#define atomic64_dec(v)		(atomic64_add(  -1,(v)))
-
-#define atomic64_inc_return(v)		(atomic64_add_return(   1,(v)))
-#define atomic64_dec_return(v)		(atomic64_add_return(  -1,(v)))
-
 /* exported interface */
 #define atomic64_cmpxchg(v, o, n) \
 	((__typeof__((v)->counter))cmpxchg(&((v)->counter), (o), (n)))
arch/powerpc/include/asm/atomic.h +4 -0
index 5d76f05d2be3..ebaefdee4a57 100644
@@ -143,6 +143,7 @@ static __inline__ void atomic_inc(atomic_t *v)
 	: "r" (&v->counter)
 	: "cc", "xer");
 }
+#define atomic_inc atomic_inc
 
 static __inline__ int atomic_inc_return_relaxed(atomic_t *v)
 {
@@ -175,6 +176,7 @@ static __inline__ void atomic_dec(atomic_t *v)
 	: "r" (&v->counter)
 	: "cc", "xer");
 }
+#define atomic_dec atomic_dec
 
 static __inline__ int atomic_dec_return_relaxed(atomic_t *v)
 {
@@ -411,6 +413,7 @@ static __inline__ void atomic64_inc(atomic64_t *v)
 	: "r" (&v->counter)
 	: "cc", "xer");
 }
+#define atomic64_inc atomic64_inc
 
 static __inline__ long atomic64_inc_return_relaxed(atomic64_t *v)
 {
@@ -441,6 +444,7 @@ static __inline__ void atomic64_dec(atomic64_t *v)
 	: "r" (&v->counter)
 	: "cc", "xer");
 }
+#define atomic64_dec atomic64_dec
 
 static __inline__ long atomic64_dec_return_relaxed(atomic64_t *v)
 {
arch/riscv/include/asm/atomic.h +0 -76

82 lines not on this page · raw message

arch/s390/include/asm/atomic.h +0 -8
index 7f5fbd595f01..376e64af951f 100644
@@ -55,13 +55,9 @@ static inline void atomic_add(int i, atomic_t *v)
 	__atomic_add(i, &v->counter);
 }
 
-#define atomic_inc(_v)			atomic_add(1, _v)
-#define atomic_inc_return(_v)		atomic_add_return(1, _v)
 #define atomic_sub(_i, _v)		atomic_add(-(int)(_i), _v)
 #define atomic_sub_return(_i, _v)	atomic_add_return(-(int)(_i), _v)
 #define atomic_fetch_sub(_i, _v)	atomic_fetch_add(-(int)(_i), _v)
-#define atomic_dec(_v)			atomic_sub(1, _v)
-#define atomic_dec_return(_v)		atomic_sub_return(1, _v)
 
 #define ATOMIC_OPS(op)							\
 static inline void atomic_##op(int i, atomic_t *v)			\
@@ -166,12 +162,8 @@ static inline long atomic64_dec_if_positive(atomic64_t *v)
 	return dec;
 }
 
-#define atomic64_inc(_v)		atomic64_add(1, _v)
-#define atomic64_inc_return(_v)		atomic64_add_return(1, _v)
 #define atomic64_sub_return(_i, _v)	atomic64_add_return(-(long)(_i), _v)
 #define atomic64_fetch_sub(_i, _v)	atomic64_fetch_add(-(long)(_i), _v)
 #define atomic64_sub(_i, _v)		atomic64_add(-(long)(_i), _v)
-#define atomic64_dec(_v)		atomic64_sub(1, _v)
-#define atomic64_dec_return(_v)		atomic64_sub_return(1, _v)
 
 #endif /* __ARCH_S390_ATOMIC__  */
arch/sh/include/asm/atomic.h +0 -6
index d438494fa112..f37b95a80232 100644
@@ -32,12 +32,6 @@
 #include <asm/atomic-irq.h>
 #endif
 
-#define atomic_dec_return(v)		atomic_sub_return(1, (v))
-#define atomic_inc_return(v)		atomic_add_return(1, (v))
-
-#define atomic_inc(v)			atomic_add(1, (v))
-#define atomic_dec(v)			atomic_sub(1, (v))
-
 #define atomic_xchg(v, new)		(xchg(&((v)->counter), new))
 #define atomic_cmpxchg(v, o, n)		(cmpxchg(&((v)->counter), (o), (n)))
 
arch/sparc/include/asm/atomic_32.h +0 -5
index 3a26573790c6..94c930f0bc62 100644
@@ -38,8 +38,6 @@ void atomic_set(atomic_t *, int);
 
 #define atomic_add(i, v)	((void)atomic_add_return( (int)(i), (v)))
 #define atomic_sub(i, v)	((void)atomic_add_return(-(int)(i), (v)))
-#define atomic_inc(v)		((void)atomic_add_return(        1, (v)))
-#define atomic_dec(v)		((void)atomic_add_return(       -1, (v)))
 
 #define atomic_and(i, v)	((void)atomic_fetch_and((i), (v)))
 #define atomic_or(i, v)		((void)atomic_fetch_or((i), (v)))
@@ -48,7 +46,4 @@ void atomic_set(atomic_t *, int);
 #define atomic_sub_return(i, v)	(atomic_add_return(-(int)(i), (v)))
 #define atomic_fetch_sub(i, v)  (atomic_fetch_add (-(int)(i), (v)))
 
-#define atomic_inc_return(v)	(atomic_add_return(        1, (v)))
-#define atomic_dec_return(v)	(atomic_add_return(       -1, (v)))
-
 #endif /* !(__ARCH_SPARC_ATOMIC__) */
arch/sparc/include/asm/atomic_64.h +0 -12
index 634508282aea..304865c7cdbb 100644
@@ -50,18 +50,6 @@ ATOMIC_OPS(xor)
 #undef ATOMIC_OP_RETURN
 #undef ATOMIC_OP
 
-#define atomic_dec_return(v)   atomic_sub_return(1, v)
-#define atomic64_dec_return(v) atomic64_sub_return(1, v)
-
-#define atomic_inc_return(v)   atomic_add_return(1, v)
-#define atomic64_inc_return(v) atomic64_add_return(1, v)
-
-#define atomic_inc(v) atomic_add(1, v)
-#define atomic64_inc(v) atomic64_add(1, v)
-
-#define atomic_dec(v) atomic_sub(1, v)
-#define atomic64_dec(v) atomic64_sub(1, v)
-
 #define atomic_cmpxchg(v, o, n) (cmpxchg(&((v)->counter), (o), (n)))
 
 static inline int atomic_xchg(atomic_t *v, int new)
arch/x86/include/asm/atomic.h +2 -3
index 73bda4abe180..823fd2f320cf 100644
@@ -92,6 +92,7 @@ static __always_inline bool arch_atomic_sub_and_test(int i, atomic_t *v)
  *
  * Atomically increments @v by 1.
  */
+#define arch_atomic_inc arch_atomic_inc
 static __always_inline void arch_atomic_inc(atomic_t *v)
 {
 	asm volatile(LOCK_PREFIX "incl %0"
@@ -104,6 +105,7 @@ static __always_inline void arch_atomic_inc(atomic_t *v)
  *
  * Atomically decrements @v by 1.
  */
+#define arch_atomic_dec arch_atomic_dec
 static __always_inline void arch_atomic_dec(atomic_t *v)
 {
 	asm volatile(LOCK_PREFIX "decl %0"
@@ -177,9 +179,6 @@ static __always_inline int arch_atomic_sub_return(int i, atomic_t *v)
 	return arch_atomic_add_return(-i, v);
 }
 
-#define arch_atomic_inc_return(v)  (arch_atomic_add_return(1, v))
-#define arch_atomic_dec_return(v)  (arch_atomic_sub_return(1, v))
-
 static __always_inline int arch_atomic_fetch_add(int i, atomic_t *v)
 {
 	return xadd(&v->counter, i);
arch/x86/include/asm/atomic64_32.h +4 -0
index a26810d005e0..472c7af0ed48 100644
@@ -158,6 +158,7 @@ static inline long long arch_atomic64_inc_return(atomic64_t *v)
 			     "S" (v) : "memory", "ecx");
 	return a;
 }
+#define arch_atomic64_inc_return arch_atomic64_inc_return
 
 static inline long long arch_atomic64_dec_return(atomic64_t *v)
 {
@@ -166,6 +167,7 @@ static inline long long arch_atomic64_dec_return(atomic64_t *v)
 			     "S" (v) : "memory", "ecx");
 	return a;
 }
+#define arch_atomic64_dec_return arch_atomic64_dec_return
 
 /**
  * arch_atomic64_add - add integer to atomic64 variable
@@ -203,6 +205,7 @@ static inline long long arch_atomic64_sub(long long i, atomic64_t *v)
  *
  * Atomically increments @v by 1.
  */
+#define arch_atomic64_inc arch_atomic64_inc
 static inline void arch_atomic64_inc(atomic64_t *v)
 {
 	__alternative_atomic64(inc, inc_return, /* no output */,
@@ -215,6 +218,7 @@ static inline void arch_atomic64_inc(atomic64_t *v)
  *
  * Atomically decrements @v by 1.
  */
+#define arch_atomic64_dec arch_atomic64_dec
 static inline void arch_atomic64_dec(atomic64_t *v)
 {
 	__alternative_atomic64(dec, dec_return, /* no output */,
arch/x86/include/asm/atomic64_64.h +2 -3
index 6a65228a3db6..1b282272a801 100644
@@ -83,6 +83,7 @@ static inline bool arch_atomic64_sub_and_test(long i, atomic64_t *v)
  *
  * Atomically increments @v by 1.
  */
+#define arch_atomic64_inc arch_atomic64_inc
 static __always_inline void arch_atomic64_inc(atomic64_t *v)
 {
 	asm volatile(LOCK_PREFIX "incq %0"
@@ -96,6 +97,7 @@ static __always_inline void arch_atomic64_inc(atomic64_t *v)
  *
  * Atomically decrements @v by 1.
  */
+#define arch_atomic64_dec arch_atomic64_dec
 static __always_inline void arch_atomic64_dec(atomic64_t *v)
 {
 	asm volatile(LOCK_PREFIX "decq %0"
@@ -173,9 +175,6 @@ static inline long arch_atomic64_fetch_sub(long i, atomic64_t *v)
 	return xadd(&v->counter, -i);
 }
 
-#define arch_atomic64_inc_return(v)  (arch_atomic64_add_return(1, (v)))
-#define arch_atomic64_dec_return(v)  (arch_atomic64_sub_return(1, (v)))
-
 static inline long arch_atomic64_cmpxchg(atomic64_t *v, long old, long new)
 {
 	return arch_cmpxchg(&v->counter, old, new);
arch/xtensa/include/asm/atomic.h +0 -32
index 332ae4eca737..7de0149e1cf7 100644
@@ -197,38 +197,6 @@ ATOMIC_OPS(xor)
 #undef ATOMIC_OP_RETURN
 #undef ATOMIC_OP
 
-/**
- * atomic_inc - increment atomic variable
- * @v: pointer of type atomic_t
- *
- * Atomically increments @v by 1.
- */
-#define atomic_inc(v) atomic_add(1,(v))
-
-/**
- * atomic_inc - increment atomic variable
- * @v: pointer of type atomic_t
- *
- * Atomically increments @v by 1.
- */
-#define atomic_inc_return(v) atomic_add_return(1,(v))
-
-/**
- * atomic_dec - decrement atomic variable
- * @v: pointer of type atomic_t
- *
- * Atomically decrements @v by 1.
- */
-#define atomic_dec(v) atomic_sub(1,(v))
-
-/**
- * atomic_dec_return - decrement atomic variable
- * @v: pointer of type atomic_t
- *
- * Atomically decrements @v by 1.
- */
-#define atomic_dec_return(v) atomic_sub_return(1,(v))
-
 #define atomic_cmpxchg(v, o, n) ((int)cmpxchg(&((v)->counter), (o), (n)))
 #define atomic_xchg(v, new) (xchg(&((v)->counter), new))
 
include/asm-generic/atomic-instrumented.h +24 -0
index b59edd520e39..8c9887a0cb1d 100644
@@ -102,29 +102,41 @@ static __always_inline int atomic64_fetch_add_unless(atomic64_t *v, s64 a, s64 u
 }
 #endif
 
+#ifdef arch_atomic_inc
+#define atomic_inc atomic_inc
 static __always_inline void atomic_inc(atomic_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	arch_atomic_inc(v);
 }
+#endif
 
+#ifdef arch_atomic64_inc
+#define atomic64_inc atomic64_inc
 static __always_inline void atomic64_inc(atomic64_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	arch_atomic64_inc(v);
 }
+#endif
 
+#ifdef arch_atomic_dec
+#define atomic_dec atomic_dec
 static __always_inline void atomic_dec(atomic_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	arch_atomic_dec(v);
 }
+#endif
 
+#ifdef atch_atomic64_dec
+#define atomic64_dec
 static __always_inline void atomic64_dec(atomic64_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	arch_atomic64_dec(v);
 }
+#endif
 
 static __always_inline void atomic_add(int i, atomic_t *v)
 {
@@ -186,29 +198,41 @@ static __always_inline void atomic64_xor(s64 i, atomic64_t *v)
 	arch_atomic64_xor(i, v);
 }
 
+#ifdef arch_atomic_inc_return
+#define atomic_inc_return atomic_inc_return
 static __always_inline int atomic_inc_return(atomic_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic_inc_return(v);
 }
+#endif
 
+#ifdef arch_atomic64_in_return
+#define atomic64_inc_return atomic64_inc_return
 static __always_inline s64 atomic64_inc_return(atomic64_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic64_inc_return(v);
 }
+#endif
 
+#ifdef arch_atomic_dec_return
+#define atomic_dec_return atomic_dec_return
 static __always_inline int atomic_dec_return(atomic_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic_dec_return(v);
 }
+#endif
 
+#ifdef arch_atomic64_dec_return
+#define atomic64_dec_return atomic64_dec_return
 static __always_inline s64 atomic64_dec_return(atomic64_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic64_dec_return(v);
 }
+#endif
 
 #ifdef arch_atomic64_inc_not_zero
 #define atomic64_inc_not_zero atomic64_inc_not_zero
include/asm-generic/atomic.h +0 -13
index 40cab858aaaa..13324aa828eb 100644
@@ -196,19 +196,6 @@ static inline void atomic_sub(int i, atomic_t *v)
 	atomic_sub_return(i, v);
 }
 
-static inline void atomic_inc(atomic_t *v)
-{
-	atomic_add_return(1, v);
-}
-
-static inline void atomic_dec(atomic_t *v)
-{
-	atomic_sub_return(1, v);
-}
-
-#define atomic_dec_return(v)		atomic_sub_return(1, (v))
-#define atomic_inc_return(v)		atomic_add_return(1, (v))
-
 #define atomic_xchg(ptr, v)		(xchg(&(ptr)->counter, (v)))
 #define atomic_cmpxchg(v, old, new)	(cmpxchg(&((v)->counter), (old), (new)))
 
include/asm-generic/atomic64.h +0 -5
index 15fe7cca29ab..1810a0ca96ec 100644
@@ -55,9 +55,4 @@ extern long long atomic64_xchg(atomic64_t *v, long long new);
 extern long long atomic64_fetch_add_unless(atomic64_t *v, long long a, long long u);
 #define atomic64_fetch_add_unless atomic64_fetch_add_unless
 
-#define atomic64_inc(v)			atomic64_add(1LL, (v))
-#define atomic64_inc_return(v)		atomic64_add_return(1LL, (v))
-#define atomic64_dec(v)			atomic64_sub(1LL, (v))
-#define atomic64_dec_return(v)		atomic64_sub_return(1LL, (v))
-
 #endif  /*  _ASM_GENERIC_ATOMIC64_H  */
include/linux/atomic.h +48 -0

92 lines not on this page · raw message

Signature · 1 line
2.11.0
Acked-by on v2 15/16
#28 lore raw

Re: [PATCHv2 15/16] atomics/treewide: make unconditional inc/dec ops optional

Quoted 118 lines · On Tue, 29 May 2018 08:43:45 PDT (-0700), [email protected] wrote:

Many of the inc/dec ops are mandatory, but for most architectures inc/dec are simply trivial wrappers around their corresponding add/sub ops.

Let's make all the inc/dec ops optional, so that we can get rid of these boilerplate wrappers.

The instrumented atomics are updated accordingly.

There should be no functional change as a result of this patch.

Cc Boqun Feng <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Will Deacon <[email protected]>
---
 arch/alpha/include/asm/atomic.h           | 12 -----
 arch/arc/include/asm/atomic.h             | 11 -----
 arch/arm/include/asm/atomic.h             | 11 -----
 arch/arm64/include/asm/atomic.h           | 24 ----------
 arch/h8300/include/asm/atomic.h           |  7 ---
 arch/hexagon/include/asm/atomic.h         |  6 ---
 arch/ia64/include/asm/atomic.h            |  9 ----
 arch/m68k/include/asm/atomic.h            |  5 +-
 arch/mips/include/asm/atomic.h            | 38 ----------------
 arch/parisc/include/asm/atomic.h          | 12 -----
 arch/powerpc/include/asm/atomic.h         |  4 ++
 arch/riscv/include/asm/atomic.h           | 76 -------------------------------
 arch/s390/include/asm/atomic.h            |  8 ----
 arch/sh/include/asm/atomic.h              |  6 ---
 arch/sparc/include/asm/atomic_32.h        |  5 --
 arch/sparc/include/asm/atomic_64.h        | 12 -----
 arch/x86/include/asm/atomic.h             |  5 +-
 arch/x86/include/asm/atomic64_32.h        |  4 ++
 arch/x86/include/asm/atomic64_64.h        |  5 +-
 arch/xtensa/include/asm/atomic.h          | 32 -------------
 include/asm-generic/atomic-instrumented.h | 24 ++++++++++
 include/asm-generic/atomic.h              | 13 ------
 include/asm-generic/atomic64.h            |  5 --
 include/linux/atomic.h                    | 48 +++++++++++++++++++
 24 files changed, 86 insertions(+), 296 deletions(-)

[...]

arch/riscv/include/asm/atomic.h +0 -69
index 68eef0a805ca..512b89485790 100644
@@ -209,82 +209,6 @@ ATOMIC_OPS(xor, xor, i)
 #undef ATOMIC_FETCH_OP
 #undef ATOMIC_OP_RETURN
 
-#define ATOMIC_OP(op, func_op, I, c_type, prefix)			\
-static __always_inline							\
-void atomic##prefix##_##op(atomic##prefix##_t *v)			\
-{									\
-	atomic##prefix##_##func_op(I, v);				\
-}
-
-#define ATOMIC_FETCH_OP(op, func_op, I, c_type, prefix)			\
-static __always_inline							\
-c_type atomic##prefix##_fetch_##op##_relaxed(atomic##prefix##_t *v)	\
-{									\
-	return atomic##prefix##_fetch_##func_op##_relaxed(I, v);	\
-}									\
-static __always_inline							\
-c_type atomic##prefix##_fetch_##op(atomic##prefix##_t *v)		\
-{									\
-	return atomic##prefix##_fetch_##func_op(I, v);			\
-}
-
-#define ATOMIC_OP_RETURN(op, asm_op, c_op, I, c_type, prefix)		\
-static __always_inline							\
-c_type atomic##prefix##_##op##_return_relaxed(atomic##prefix##_t *v)	\
-{									\
-        return atomic##prefix##_fetch_##op##_relaxed(v) c_op I;		\
-}									\
-static __always_inline							\
-c_type atomic##prefix##_##op##_return(atomic##prefix##_t *v)		\
-{									\
-        return atomic##prefix##_fetch_##op(v) c_op I;			\
-}
-
-#ifdef CONFIG_GENERIC_ATOMIC64
-#define ATOMIC_OPS(op, asm_op, c_op, I)					\
-        ATOMIC_OP(       op, asm_op,       I,  int,   )			\
-        ATOMIC_FETCH_OP( op, asm_op,       I,  int,   )			\
-        ATOMIC_OP_RETURN(op, asm_op, c_op, I,  int,   )
-#else
-#define ATOMIC_OPS(op, asm_op, c_op, I)					\
-        ATOMIC_OP(       op, asm_op,       I,  int,   )			\
-        ATOMIC_FETCH_OP( op, asm_op,       I,  int,   )			\
-        ATOMIC_OP_RETURN(op, asm_op, c_op, I,  int,   )			\
-        ATOMIC_OP(       op, asm_op,       I, long, 64)			\
-        ATOMIC_FETCH_OP( op, asm_op,       I, long, 64)			\
-        ATOMIC_OP_RETURN(op, asm_op, c_op, I, long, 64)
-#endif
-
-ATOMIC_OPS(inc, add, +,  1)
-ATOMIC_OPS(dec, add, +, -1)
-
-#define atomic_inc_return_relaxed	atomic_inc_return_relaxed
-#define atomic_dec_return_relaxed	atomic_dec_return_relaxed
-#define atomic_inc_return		atomic_inc_return
-#define atomic_dec_return		atomic_dec_return
-
-#define atomic_fetch_inc_relaxed	atomic_fetch_inc_relaxed
-#define atomic_fetch_dec_relaxed	atomic_fetch_dec_relaxed
-#define atomic_fetch_inc		atomic_fetch_inc
-#define atomic_fetch_dec		atomic_fetch_dec
-
-#ifndef CONFIG_GENERIC_ATOMIC64
-#define atomic64_inc_return_relaxed	atomic64_inc_return_relaxed
-#define atomic64_dec_return_relaxed	atomic64_dec_return_relaxed
-#define atomic64_inc_return		atomic64_inc_return
-#define atomic64_dec_return		atomic64_dec_return
-
-#define atomic64_fetch_inc_relaxed	atomic64_fetch_inc_relaxed
-#define atomic64_fetch_dec_relaxed	atomic64_fetch_dec_relaxed
-#define atomic64_fetch_inc		atomic64_fetch_inc
-#define atomic64_fetch_dec		atomic64_fetch_dec

arch/riscv/include/asm/atomic.h ATOMIC_OPS(xor, xor, i) · line 214

Quoted 10 lines
arch/riscv/include/asm/atomic.h +0 -7
-#endif
-
-#undef ATOMIC_OPS
-#undef ATOMIC_OP
-#undef ATOMIC_FETCH_OP
-#undef ATOMIC_OP_RETURN
-
 /* This is required to provide a full barrier on success. */
 static __always_inline int atomic_fetch_add_unless(atomic_t *v, int a, int u)
 {

As far as the RISC-V stuff goes: this works for me, as I always like to have less code in the arch port.

Thanks!

author reviewer
#29 lore raw

PATCH v2 16/16 atomics/treewide: make conditional inc/dec ops optional

In reply to Mark Rutland

The conditional inc/dec ops differ for atomic_t and atomic64_t:

* atomic_inc_unless_positive() is optional for atomic_t, and doesn't exist for atomic64_t.

* atomic_dec_unless_negative() is optional for atomic_t, and doesn't exist for atomic64_t.

* atomic_dec_if_positive is optional for atomic_t, and is mandatory for atomic64_t.

Let's make these consistently optional for both. At the same time, let's clean up the exisiting fallbacks to use atomic_try_cmpxchg().

The instrumented atomics are updated accordingly.

There should be no functional change as a result of this patch.

Cc Boqun Feng <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Will Deacon <[email protected]>
---
 arch/alpha/include/asm/atomic.h           |  1 +
 arch/arc/include/asm/atomic.h             |  1 +
 arch/arm/include/asm/atomic.h             |  1 +
 arch/arm64/include/asm/atomic.h           |  2 +
 arch/ia64/include/asm/atomic.h            | 16 -----
 arch/parisc/include/asm/atomic.h          | 23 --------
 arch/powerpc/include/asm/atomic.h         |  1 +
 arch/s390/include/asm/atomic.h            | 17 ------
 arch/sparc/include/asm/atomic_64.h        |  1 +
 arch/x86/include/asm/atomic64_32.h        |  1 +
 arch/x86/include/asm/atomic64_64.h        | 18 ------
 include/asm-generic/atomic-instrumented.h |  3 +
 include/asm-generic/atomic64.h            |  1 +
 include/linux/atomic.h                    | 97 +++++++++++++++++++++++--------
 14 files changed, 85 insertions(+), 98 deletions(-)
arch/alpha/include/asm/atomic.h +1 -0
index 9c50cc512a78..81bf9d15dcc1 100644
@@ -296,5 +296,6 @@ static inline long atomic64_dec_if_positive(atomic64_t *v)
 	smp_mb();
 	return old - 1;
 }
+#define atomic64_dec_if_positive atomic64_dec_if_positive
 
 #endif /* _ALPHA_ATOMIC_H */
arch/arc/include/asm/atomic.h +1 -0
index abffdc525b46..261c6a4d3f05 100644
@@ -517,6 +517,7 @@ static inline long long atomic64_dec_if_positive(atomic64_t *v)
 
 	return val;
 }
+#define atomic64_dec_if_positive atomic64_dec_if_positive
 
 /**
  * atomic64_fetch_add_unless - add unless the number is a given value
arch/arm/include/asm/atomic.h +1 -0
index 5a58d061d3d2..884c241424fe 100644
@@ -474,6 +474,7 @@ static inline long long atomic64_dec_if_positive(atomic64_t *v)
 
 	return result;
 }
+#define atomic64_dec_if_positive atomic64_dec_if_positive
 
 static inline long long atomic64_fetch_add_unless(atomic64_t *v, long long a,
 						  long long u)
arch/arm64/include/asm/atomic.h +2 -0
index 078f785cd97f..9bca54dda75c 100644
@@ -159,5 +159,7 @@
 
 #define atomic64_andnot			atomic64_andnot
 
+#define atomic64_dec_if_positive	atomic64_dec_if_positive
+
 #endif
 #endif
arch/ia64/include/asm/atomic.h +0 -16
index 46a15a974bed..206530d0751b 100644
@@ -215,22 +215,6 @@ ATOMIC64_FETCH_OP(xor, ^)
 	(cmpxchg(&((v)->counter), old, new))
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), new))
 
-static __inline__ long atomic64_dec_if_positive(atomic64_t *v)
-{
-	long c, old, dec;
-	c = atomic64_read(v);
-	for (;;) {
-		dec = c - 1;
-		if (unlikely(dec < 0))
-			break;
-		old = atomic64_cmpxchg((v), c, dec);
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return dec;
-}
-
 #define atomic_add(i,v)			(void)atomic_add_return((i), (v))
 #define atomic_sub(i,v)			(void)atomic_sub_return((i), (v))
 
arch/parisc/include/asm/atomic.h +0 -23
index 10bc490327c1..118953d41763 100644
@@ -223,29 +223,6 @@ atomic64_read(const atomic64_t *v)
 	((__typeof__((v)->counter))cmpxchg(&((v)->counter), (o), (n)))
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), new))
 
-/*
- * atomic64_dec_if_positive - decrement by 1 if old value positive
- * @v: pointer of type atomic_t
- *
- * The function returns the old value of *v minus 1, even if
- * the atomic variable, v, was not decremented.
- */
-static inline long atomic64_dec_if_positive(atomic64_t *v)
-{
-	long c, old, dec;
-	c = atomic64_read(v);
-	for (;;) {
-		dec = c - 1;
-		if (unlikely(dec < 0))
-			break;
-		old = atomic64_cmpxchg((v), c, dec);
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return dec;
-}
-
 #endif /* !CONFIG_64BIT */
 
 
arch/powerpc/include/asm/atomic.h +1 -0
index ebaefdee4a57..a0156cb43d1f 100644
@@ -488,6 +488,7 @@ static __inline__ long atomic64_dec_if_positive(atomic64_t *v)
 
 	return t;
 }
+#define atomic64_dec_if_positive atomic64_dec_if_positive
 
 #define atomic64_cmpxchg(v, o, n) (cmpxchg(&((v)->counter), (o), (n)))
 #define atomic64_cmpxchg_relaxed(v, o, n) \
arch/s390/include/asm/atomic.h +0 -17
index 376e64af951f..fd20ab5d4cf7 100644
@@ -145,23 +145,6 @@ ATOMIC64_OPS(xor)
 
 #undef ATOMIC64_OPS
 
-static inline long atomic64_dec_if_positive(atomic64_t *v)
-{
-	long c, old, dec;
-
-	c = atomic64_read(v);
-	for (;;) {
-		dec = c - 1;
-		if (unlikely(dec < 0))
-			break;
-		old = atomic64_cmpxchg((v), c, dec);
-		if (likely(old == c))
-			break;
-		c = old;
-	}
-	return dec;
-}
-
 #define atomic64_sub_return(_i, _v)	atomic64_add_return(-(long)(_i), _v)
 #define atomic64_fetch_sub(_i, _v)	atomic64_fetch_add(-(long)(_i), _v)
 #define atomic64_sub(_i, _v)		atomic64_add(-(long)(_i), _v)
arch/sparc/include/asm/atomic_64.h +1 -0
index 304865c7cdbb..6963482c81d8 100644
@@ -62,5 +62,6 @@ static inline int atomic_xchg(atomic_t *v, int new)
 #define atomic64_xchg(v, new) (xchg(&((v)->counter), new))
 
 long atomic64_dec_if_positive(atomic64_t *v);
+#define atomic64_dec_if_positive atomic64_dec_if_positive
 
 #endif /* !(__ARCH_SPARC64_ATOMIC__) */
arch/x86/include/asm/atomic64_32.h +1 -0
index 472c7af0ed48..ef959f02d070 100644
@@ -254,6 +254,7 @@ static inline int arch_atomic64_inc_not_zero(atomic64_t *v)
 	return r;
 }
 
+#define arch_atomic64_dec_if_positive arch_atomic64_dec_if_positive
 static inline long long arch_atomic64_dec_if_positive(atomic64_t *v)
 {
 	long long r;
arch/x86/include/asm/atomic64_64.h +0 -18
index 1b282272a801..849f1c566a11 100644
@@ -191,24 +191,6 @@ static inline long arch_atomic64_xchg(atomic64_t *v, long new)
 	return xchg(&v->counter, new);
 }
 
-/*
- * arch_atomic64_dec_if_positive - decrement by 1 if old value positive
- * @v: pointer of type atomic_t
- *
- * The function returns the old value of *v minus 1, even if
- * the atomic variable, v, was not decremented.
- */
-static inline long arch_atomic64_dec_if_positive(atomic64_t *v)
-{
-	s64 dec, c = arch_atomic64_read(v);
-	do {
-		dec = c - 1;
-		if (unlikely(dec < 0))
-			break;
-	} while (!arch_atomic64_try_cmpxchg(v, &c, dec));
-	return dec;
-}
-
 static inline void arch_atomic64_and(long i, atomic64_t *v)
 {
 	asm volatile(LOCK_PREFIX "andq %1,%0"
include/asm-generic/atomic-instrumented.h +3 -0
index 8c9887a0cb1d..1458de7ece87 100644
@@ -243,11 +243,14 @@ static __always_inline s64 atomic64_inc_not_zero(atomic64_t *v)
 }
 #endif
 
+#ifdef arch_atomic64_dec_if_positive
+#define atomic64_dec_if_positive atomic64_dec_if_positive
 static __always_inline s64 atomic64_dec_if_positive(atomic64_t *v)
 {
 	kasan_check_write(v, sizeof(*v));
 	return arch_atomic64_dec_if_positive(v);
 }
+#endif
 
 #ifdef arch_atomic_dec_and_test
 #define atomic_dec_and_test atomic_dec_and_test
include/asm-generic/atomic64.h +1 -0
index 1810a0ca96ec..296f1b3c6661 100644
@@ -50,6 +50,7 @@ ATOMIC64_OPS(xor)
 #undef ATOMIC64_OP
 
 extern long long atomic64_dec_if_positive(atomic64_t *v);
+#define atomic64_dec_if_positive atomic64_dec_if_positive
 extern long long atomic64_cmpxchg(atomic64_t *v, long long o, long long n);
 extern long long atomic64_xchg(atomic64_t *v, long long new);
 extern long long atomic64_fetch_add_unless(atomic64_t *v, long long a, long long u);
include/linux/atomic.h +73 -24

124 lines not on this page · raw message

Signature · 1 line
2.11.0
author reviewer
#30 lore raw
In reply to Mark Rutland
Quoted 14 lines · Mark Rutland

This series contains a few cleanups of the atomic API, fixing inconsistencies between atomic_* and atomic64_*, and minimizing repetition in arch code. This is nicer for arch code, and the improved regularity will help when generating the atomic headers in future.

Mark Rutland (16):
atomics/treewide: s/__atomic_add_unless/atomic_fetch_add_unless/ atomics/treewide: remove redundant atomic_inc_not_zero() definitions atomics/treewide: make atomic64_inc_not_zero() optional atomics/treewide: make atomic_fetch_add_unless() optional
atomics: prepare for atomic64_fetch_add_unless() atomics/generic: define atomic64_fetch_add_unless() atomics/alpha: define atomic64_fetch_add_unless() atomics/arc: define atomic64_fetch_add_unless()

Quoted 8 lines

atomics/arm: define atomic64_fetch_add_unless() atomics/powerpc: define atomic64_fetch_add_unless() atomics/riscv: define atomic64_fetch_add_unless() atomics/treewide: make atomic64_fetch_add_unless() optional atomics/treewide: make test ops optional atomics/treewide: remove atomic_inc_not_zero_hint() atomics/treewide: make unconditional inc/dec ops optional atomics/treewide: make conditional inc/dec ops optional

While testing the subsqeuent generated header patches, I discovered that the andnot ifdeffery was inconsistent to that for all other atomics, so I'm going to append the below patch to this series, if it needs a v3.

Otherwise, that's pushed out to my atomics/api-cleanup branch.

Thanks,
Mark,

---->8----

Quoted 1 line

From da1cef7ec7c1ad143a06723164263ef308bfb569 Mon Sep 17 00:00:00 2001

From: Mark Rutland <[email protected]>
Date: Wed, 30 May 2018 11:55:28 +0100
Subject: [PATCH] atomics/treewide: clean up andnot ifdeffery

The ifdeffery for atomic*_{fetch_,}andnot() is unlike that for all the other atomics. If atomic*_andnot() is not defined, the corresponding atomic*_fetch_andnot() is assumed to not be defined.

Additionally, the fallbacks for the various orering cases are written much later in atomic.h as static inlines.

This isn't problematic today, but gets in the way of scripting the generation of atomics. To prepare for scripting, this patch:

* Switches to separate ifdefs for atomic*_andnot() and atomic*_fetch_andnot(), updating implementations as appropriate.

* Moves the fallbacks into the standards ifdefs, as macro expansions rather than static inlines.

* Removes trivial andnot implementations from architectures, where these are superceded by core code.

There should be no functional change as a result of this patch.

Cc Boqun Feng <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Will Deacon <[email protected]>
---
 arch/arc/include/asm/atomic.h |  8 ++--
 arch/arm/include/asm/atomic.h |  2 +
 include/linux/atomic.h        | 96 ++++++++++++++-----------------------------
 3 files changed, 36 insertions(+), 70 deletions(-)
arch/arc/include/asm/atomic.h +4 -4
index 261c6a4d3f05..b706781825ac 100644
@@ -187,7 +187,8 @@ static inline int atomic_fetch_##op(int i, atomic_t *v)			\
 ATOMIC_OPS(add, +=, add)
 ATOMIC_OPS(sub, -=, sub)
 
-#define atomic_andnot atomic_andnot
+#define atomic_andnot		atomic_andnot
+#define atomic_fetch_andnot	atomic_fetch_andnot
 
 #undef ATOMIC_OPS
 #define ATOMIC_OPS(op, c_op, asm_op)					\
@@ -296,8 +297,6 @@ ATOMIC_OPS(add, +=, CTOP_INST_AADD_DI_R2_R2_R3)
 	ATOMIC_FETCH_OP(op, c_op, asm_op)
 
 ATOMIC_OPS(and, &=, CTOP_INST_AAND_DI_R2_R2_R3)
-#define atomic_andnot(mask, v) atomic_and(~(mask), (v))
-#define atomic_fetch_andnot(mask, v) atomic_fetch_and(~(mask), (v))
 ATOMIC_OPS(or, |=, CTOP_INST_AOR_DI_R2_R2_R3)
 ATOMIC_OPS(xor, ^=, CTOP_INST_AXOR_DI_R2_R2_R3)
 
@@ -430,7 +429,8 @@ static inline long long atomic64_fetch_##op(long long a, atomic64_t *v)	\
 	ATOMIC64_OP_RETURN(op, op1, op2)				\
 	ATOMIC64_FETCH_OP(op, op1, op2)
 
-#define atomic64_andnot atomic64_andnot
+#define atomic64_andnot		atomic64_andnot
+#define atomic64_fetch_andnot	atomic64_fetch_andnot
 
 ATOMIC64_OPS(add, add.f, adc)
 ATOMIC64_OPS(sub, sub.f, sbc)
arch/arm/include/asm/atomic.h +2 -0
index 884c241424fe..f74756641410 100644
@@ -216,6 +216,8 @@ static inline int atomic_cmpxchg(atomic_t *v, int old, int new)
 	return ret;
 }
 
+#define atomic_fetch_andnot		atomic_fetch_andnot
+
 #endif /* __LINUX_ARM_ARCH__ */
 
 #define ATOMIC_OPS(op, c_op, asm_op)					\
include/linux/atomic.h +30 -66

134 lines not on this page · raw message

Signature · 1 line
2.11.0
author reviewer
#31 lore raw

include/linux/atomic.h static inline int atomic_dec_if_positive(atomic_t *v) · line 1013

Quoted 5 lines · Mark Rutland
include/linux/atomic.h +3 -2
-#ifdef atomic64_andnot
-/* atomic64_fetch_andnot_relaxed */
+#ifndef atomic_andnot
+#define atomic_andnot(i, v)		atomic_and(~(int)(i), (v))
+#endif

Ugh; s/atomic/atomic64 here.

I've pushed out a corrected branch.

Mark.