* [PATCH v2 0/4] locking/atomic: arch_cmpxchg[64]_local undefined
@ 2023-11-04 9:16 wuqiang.matt
2023-11-04 9:16 ` [PATCH v2 1/4] locking/atomic: arc: data size checking in arch_cmpxchg wuqiang.matt
` (3 more replies)
0 siblings, 4 replies; 9+ messages in thread
From: wuqiang.matt @ 2023-11-04 9:16 UTC (permalink / raw)
To: vgupta, bcain, jonas, stefan.kristiansson, shorne, geert,
andi.shyti, mingo, palmer, andrzej.hajda, arnd, peterz, mhiramat
Cc: linux-snps-arc, linux-kernel, linux-hexagon, linux-openrisc,
linux-trace-kernel, mattwu, linux, wuqiang.matt
This patch series implement arch_cmpxchg[64]_local for arc,
openrisc and hexagon.
For architectures that support native cmpxchg, we'd like to
implement arch_cmpxchg[64]_local with the native variants of
supported data size. If not, the generci_cmpxchg[64]_local
will be used.
wuqiang.matt (4):
locking/atomic: arc: data size checking in arch_cmpxchg
locking/atomic: arc: arch_cmpxchg[64]_local undefined
locking/atomic: openrisc: arch_cmpxchg[64]_local undefined
locking/atomic: hexagon: arch_cmpxchg[64]_local undefined
arch/arc/include/asm/cmpxchg.h | 40 ++++++++++++++++++----
arch/hexagon/include/asm/cmpxchg.h | 51 ++++++++++++++++++++++++++++-
arch/openrisc/include/asm/cmpxchg.h | 6 ++++
3 files changed, 90 insertions(+), 7 deletions(-)
--
2.40.1
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH v2 1/4] locking/atomic: arc: data size checking in arch_cmpxchg
2023-11-04 9:16 [PATCH v2 0/4] locking/atomic: arch_cmpxchg[64]_local undefined wuqiang.matt
@ 2023-11-04 9:16 ` wuqiang.matt
2023-11-08 1:02 ` Masami Hiramatsu
2023-11-04 9:16 ` [PATCH v2 2/4] locking/atomic: arc: arch_cmpxchg[64]_local undefined wuqiang.matt
` (2 subsequent siblings)
3 siblings, 1 reply; 9+ messages in thread
From: wuqiang.matt @ 2023-11-04 9:16 UTC (permalink / raw)
To: vgupta, bcain, jonas, stefan.kristiansson, shorne, geert,
andi.shyti, mingo, palmer, andrzej.hajda, arnd, peterz, mhiramat
Cc: linux-snps-arc, linux-kernel, linux-hexagon, linux-openrisc,
linux-trace-kernel, mattwu, linux, wuqiang.matt
Macro __cmpxchg() renamed to __cmpxchg_32() to emphasize it's explicit
support of 32bit data size, BUILD_BUG_ON() added to avoid any possible
misuses with unsupported data types.
In case CONFIG_ARC_HAS_LLSC is undefined, arch_cmpxchg() uses spinlock
to accomplish SMP-safety, so the BUILD_BUG_ON checking is uncecessary.
Signed-off-by: wuqiang.matt <wuqiang.matt@bytedance.com>
---
arch/arc/include/asm/cmpxchg.h | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/arch/arc/include/asm/cmpxchg.h b/arch/arc/include/asm/cmpxchg.h
index e138fde067de..bf46514f6f12 100644
--- a/arch/arc/include/asm/cmpxchg.h
+++ b/arch/arc/include/asm/cmpxchg.h
@@ -18,14 +18,16 @@
* if (*ptr == @old)
* *ptr = @new
*/
-#define __cmpxchg(ptr, old, new) \
+#define __cmpxchg_32(ptr, old, new) \
({ \
__typeof__(*(ptr)) _prev; \
\
+ BUILD_BUG_ON(sizeof(*(ptr)) != 4); \
+ \
__asm__ __volatile__( \
- "1: llock %0, [%1] \n" \
+ "1: llock %0, [%1] \n" \
" brne %0, %2, 2f \n" \
- " scond %3, [%1] \n" \
+ " scond %3, [%1] \n" \
" bnz 1b \n" \
"2: \n" \
: "=&r"(_prev) /* Early clobber prevent reg reuse */ \
@@ -47,7 +49,7 @@
\
switch(sizeof((_p_))) { \
case 4: \
- _prev_ = __cmpxchg(_p_, _o_, _n_); \
+ _prev_ = __cmpxchg_32(_p_, _o_, _n_); \
break; \
default: \
BUILD_BUG(); \
@@ -65,8 +67,6 @@
__typeof__(*(ptr)) _prev_; \
unsigned long __flags; \
\
- BUILD_BUG_ON(sizeof(_p_) != 4); \
- \
/* \
* spin lock/unlock provide the needed smp_mb() before/after \
*/ \
--
2.40.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v2 2/4] locking/atomic: arc: arch_cmpxchg[64]_local undefined
2023-11-04 9:16 [PATCH v2 0/4] locking/atomic: arch_cmpxchg[64]_local undefined wuqiang.matt
2023-11-04 9:16 ` [PATCH v2 1/4] locking/atomic: arc: data size checking in arch_cmpxchg wuqiang.matt
@ 2023-11-04 9:16 ` wuqiang.matt
2023-11-08 1:02 ` Masami Hiramatsu
2023-11-04 9:16 ` [PATCH v2 3/4] locking/atomic: openrisc: " wuqiang.matt
2023-11-04 9:16 ` [PATCH v2 4/4] locking/atomic: hexagon: " wuqiang.matt
3 siblings, 1 reply; 9+ messages in thread
From: wuqiang.matt @ 2023-11-04 9:16 UTC (permalink / raw)
To: vgupta, bcain, jonas, stefan.kristiansson, shorne, geert,
andi.shyti, mingo, palmer, andrzej.hajda, arnd, peterz, mhiramat
Cc: linux-snps-arc, linux-kernel, linux-hexagon, linux-openrisc,
linux-trace-kernel, mattwu, linux, wuqiang.matt
For architectures that support native cmpxchg, we'd like to
implement arch_cmpxchg[64]_local with the native variants of
supported data size. If not, the generci_cmpxchg[64]_local
will be used.
Signed-off-by: wuqiang.matt <wuqiang.matt@bytedance.com>
---
arch/arc/include/asm/cmpxchg.h | 28 ++++++++++++++++++++++++++++
1 file changed, 28 insertions(+)
diff --git a/arch/arc/include/asm/cmpxchg.h b/arch/arc/include/asm/cmpxchg.h
index bf46514f6f12..91429f2350df 100644
--- a/arch/arc/include/asm/cmpxchg.h
+++ b/arch/arc/include/asm/cmpxchg.h
@@ -80,6 +80,34 @@
#endif
+/*
+ * always make arch_cmpxchg[64]_local available, native cmpxchg
+ * will be used if available, then generic_cmpxchg[64]_local
+ */
+#include <asm-generic/cmpxchg-local.h>
+static inline unsigned long __cmpxchg_local(volatile void *ptr,
+ unsigned long old,
+ unsigned long new, int size)
+{
+ switch (size) {
+#ifdef CONFIG_ARC_HAS_LLSC
+ case 4:
+ return __cmpxchg_32((int32_t *)ptr, old, new);
+#endif
+ default:
+ return __generic_cmpxchg_local(ptr, old, new, size);
+ }
+
+ return old;
+}
+#define arch_cmpxchg_local(ptr, o, n) ({ \
+ (__typeof__(*ptr))__cmpxchg_local((ptr), \
+ (unsigned long)(o), \
+ (unsigned long)(n), \
+ sizeof(*(ptr))); \
+})
+#define arch_cmpxchg64_local(ptr, o, n) __generic_cmpxchg64_local((ptr), (o), (n))
+
/*
* xchg
*/
--
2.40.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v2 3/4] locking/atomic: openrisc: arch_cmpxchg[64]_local undefined
2023-11-04 9:16 [PATCH v2 0/4] locking/atomic: arch_cmpxchg[64]_local undefined wuqiang.matt
2023-11-04 9:16 ` [PATCH v2 1/4] locking/atomic: arc: data size checking in arch_cmpxchg wuqiang.matt
2023-11-04 9:16 ` [PATCH v2 2/4] locking/atomic: arc: arch_cmpxchg[64]_local undefined wuqiang.matt
@ 2023-11-04 9:16 ` wuqiang.matt
2023-11-08 1:03 ` Masami Hiramatsu
2023-11-04 9:16 ` [PATCH v2 4/4] locking/atomic: hexagon: " wuqiang.matt
3 siblings, 1 reply; 9+ messages in thread
From: wuqiang.matt @ 2023-11-04 9:16 UTC (permalink / raw)
To: vgupta, bcain, jonas, stefan.kristiansson, shorne, geert,
andi.shyti, mingo, palmer, andrzej.hajda, arnd, peterz, mhiramat
Cc: linux-snps-arc, linux-kernel, linux-hexagon, linux-openrisc,
linux-trace-kernel, mattwu, linux, wuqiang.matt
For architectures that support native cmpxchg, we'd like to
implement arch_cmpxchg[64]_local with the native variants of
supported data size. If not, the generci_cmpxchg[64]_local
will be used.
Signed-off-by: wuqiang.matt <wuqiang.matt@bytedance.com>
---
arch/openrisc/include/asm/cmpxchg.h | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/arch/openrisc/include/asm/cmpxchg.h b/arch/openrisc/include/asm/cmpxchg.h
index 8ee151c072e4..f1ffe8b6f5ef 100644
--- a/arch/openrisc/include/asm/cmpxchg.h
+++ b/arch/openrisc/include/asm/cmpxchg.h
@@ -139,6 +139,12 @@ static inline unsigned long __cmpxchg(volatile void *ptr, unsigned long old,
(unsigned long)(n), \
sizeof(*(ptr))); \
})
+#define arch_cmpxchg_local arch_cmpxchg
+
+/* always make arch_cmpxchg64_local available for openrisc */
+#include <asm-generic/cmpxchg-local.h>
+
+#define arch_cmpxchg64_local(ptr, o, n) __generic_cmpxchg64_local((ptr), (o), (n))
/*
* This function doesn't exist, so you'll get a linker error if
--
2.40.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v2 4/4] locking/atomic: hexagon: arch_cmpxchg[64]_local undefined
2023-11-04 9:16 [PATCH v2 0/4] locking/atomic: arch_cmpxchg[64]_local undefined wuqiang.matt
` (2 preceding siblings ...)
2023-11-04 9:16 ` [PATCH v2 3/4] locking/atomic: openrisc: " wuqiang.matt
@ 2023-11-04 9:16 ` wuqiang.matt
2023-11-08 1:04 ` Masami Hiramatsu
3 siblings, 1 reply; 9+ messages in thread
From: wuqiang.matt @ 2023-11-04 9:16 UTC (permalink / raw)
To: vgupta, bcain, jonas, stefan.kristiansson, shorne, geert,
andi.shyti, mingo, palmer, andrzej.hajda, arnd, peterz, mhiramat
Cc: linux-snps-arc, linux-kernel, linux-hexagon, linux-openrisc,
linux-trace-kernel, mattwu, linux, wuqiang.matt,
kernel test robot
For architectures that support native cmpxchg, we'd like to
implement arch_cmpxchg[64]_local with the native variants of
supported data size. If not, the generci_cmpxchg[64]_local
will be used.
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202310272207.tLPflya4-lkp@intel.com/
Signed-off-by: wuqiang.matt <wuqiang.matt@bytedance.com>
---
arch/hexagon/include/asm/cmpxchg.h | 51 +++++++++++++++++++++++++++++-
1 file changed, 50 insertions(+), 1 deletion(-)
diff --git a/arch/hexagon/include/asm/cmpxchg.h b/arch/hexagon/include/asm/cmpxchg.h
index bf6cf5579cf4..2b5e5bbaf807 100644
--- a/arch/hexagon/include/asm/cmpxchg.h
+++ b/arch/hexagon/include/asm/cmpxchg.h
@@ -8,6 +8,8 @@
#ifndef _ASM_CMPXCHG_H
#define _ASM_CMPXCHG_H
+#include <linux/build_bug.h>
+
/*
* __arch_xchg - atomically exchange a register and a memory location
* @x: value to swap
@@ -51,13 +53,15 @@ __arch_xchg(unsigned long x, volatile void *ptr, int size)
* variable casting.
*/
-#define arch_cmpxchg(ptr, old, new) \
+#define __cmpxchg_32(ptr, old, new) \
({ \
__typeof__(ptr) __ptr = (ptr); \
__typeof__(*(ptr)) __old = (old); \
__typeof__(*(ptr)) __new = (new); \
__typeof__(*(ptr)) __oldval = 0; \
\
+ BUILD_BUG_ON(sizeof(*(ptr)) != 4); \
+ \
asm volatile( \
"1: %0 = memw_locked(%1);\n" \
" { P0 = cmp.eq(%0,%2);\n" \
@@ -72,4 +76,49 @@ __arch_xchg(unsigned long x, volatile void *ptr, int size)
__oldval; \
})
+#define __cmpxchg(ptr, old, val, size) \
+({ \
+ __typeof__(*(ptr)) oldval; \
+ \
+ switch (size) { \
+ case 4: \
+ oldval = __cmpxchg_32(ptr, old, val); \
+ break; \
+ default: \
+ BUILD_BUG(); \
+ oldval = val; \
+ break; \
+ } \
+ \
+ oldval; \
+})
+
+#define arch_cmpxchg(ptr, o, n) __cmpxchg((ptr), (o), (n), sizeof(*(ptr)))
+
+/*
+ * always make arch_cmpxchg[64]_local available, native cmpxchg
+ * will be used if available, then generic_cmpxchg[64]_local
+ */
+#include <asm-generic/cmpxchg-local.h>
+
+#define arch_cmpxchg_local(ptr, old, val) \
+({ \
+ __typeof__(*(ptr)) retval; \
+ int size = sizeof(*(ptr)); \
+ \
+ switch (size) { \
+ case 4: \
+ retval = __cmpxchg_32(ptr, old, val); \
+ break; \
+ default: \
+ retval = __generic_cmpxchg_local(ptr, old, \
+ val, size); \
+ break; \
+ } \
+ \
+ retval; \
+})
+
+#define arch_cmpxchg64_local(ptr, o, n) __generic_cmpxchg64_local((ptr), (o), (n))
+
#endif /* _ASM_CMPXCHG_H */
--
2.40.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH v2 1/4] locking/atomic: arc: data size checking in arch_cmpxchg
2023-11-04 9:16 ` [PATCH v2 1/4] locking/atomic: arc: data size checking in arch_cmpxchg wuqiang.matt
@ 2023-11-08 1:02 ` Masami Hiramatsu
0 siblings, 0 replies; 9+ messages in thread
From: Masami Hiramatsu @ 2023-11-08 1:02 UTC (permalink / raw)
To: wuqiang.matt
Cc: vgupta, bcain, jonas, stefan.kristiansson, shorne, geert,
andi.shyti, mingo, palmer, andrzej.hajda, arnd, peterz,
linux-snps-arc, linux-kernel, linux-hexagon, linux-openrisc,
linux-trace-kernel, mattwu, linux
On Sat, 4 Nov 2023 17:16:12 +0800
"wuqiang.matt" <wuqiang.matt@bytedance.com> wrote:
> Macro __cmpxchg() renamed to __cmpxchg_32() to emphasize it's explicit
> support of 32bit data size, BUILD_BUG_ON() added to avoid any possible
> misuses with unsupported data types.
>
> In case CONFIG_ARC_HAS_LLSC is undefined, arch_cmpxchg() uses spinlock
> to accomplish SMP-safety, so the BUILD_BUG_ON checking is uncecessary.
>
Looks good to me.
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
> Signed-off-by: wuqiang.matt <wuqiang.matt@bytedance.com>
> ---
> arch/arc/include/asm/cmpxchg.h | 12 ++++++------
> 1 file changed, 6 insertions(+), 6 deletions(-)
>
> diff --git a/arch/arc/include/asm/cmpxchg.h b/arch/arc/include/asm/cmpxchg.h
> index e138fde067de..bf46514f6f12 100644
> --- a/arch/arc/include/asm/cmpxchg.h
> +++ b/arch/arc/include/asm/cmpxchg.h
> @@ -18,14 +18,16 @@
> * if (*ptr == @old)
> * *ptr = @new
> */
> -#define __cmpxchg(ptr, old, new) \
> +#define __cmpxchg_32(ptr, old, new) \
> ({ \
> __typeof__(*(ptr)) _prev; \
> \
> + BUILD_BUG_ON(sizeof(*(ptr)) != 4); \
> + \
> __asm__ __volatile__( \
> - "1: llock %0, [%1] \n" \
> + "1: llock %0, [%1] \n" \
> " brne %0, %2, 2f \n" \
> - " scond %3, [%1] \n" \
> + " scond %3, [%1] \n" \
> " bnz 1b \n" \
> "2: \n" \
> : "=&r"(_prev) /* Early clobber prevent reg reuse */ \
> @@ -47,7 +49,7 @@
> \
> switch(sizeof((_p_))) { \
> case 4: \
> - _prev_ = __cmpxchg(_p_, _o_, _n_); \
> + _prev_ = __cmpxchg_32(_p_, _o_, _n_); \
> break; \
> default: \
> BUILD_BUG(); \
> @@ -65,8 +67,6 @@
> __typeof__(*(ptr)) _prev_; \
> unsigned long __flags; \
> \
> - BUILD_BUG_ON(sizeof(_p_) != 4); \
> - \
> /* \
> * spin lock/unlock provide the needed smp_mb() before/after \
> */ \
> --
> 2.40.1
>
--
Masami Hiramatsu (Google) <mhiramat@kernel.org>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v2 2/4] locking/atomic: arc: arch_cmpxchg[64]_local undefined
2023-11-04 9:16 ` [PATCH v2 2/4] locking/atomic: arc: arch_cmpxchg[64]_local undefined wuqiang.matt
@ 2023-11-08 1:02 ` Masami Hiramatsu
0 siblings, 0 replies; 9+ messages in thread
From: Masami Hiramatsu @ 2023-11-08 1:02 UTC (permalink / raw)
To: wuqiang.matt
Cc: vgupta, bcain, jonas, stefan.kristiansson, shorne, geert,
andi.shyti, mingo, palmer, andrzej.hajda, arnd, peterz,
linux-snps-arc, linux-kernel, linux-hexagon, linux-openrisc,
linux-trace-kernel, mattwu, linux
On Sat, 4 Nov 2023 17:16:13 +0800
"wuqiang.matt" <wuqiang.matt@bytedance.com> wrote:
> For architectures that support native cmpxchg, we'd like to
> implement arch_cmpxchg[64]_local with the native variants of
> supported data size. If not, the generci_cmpxchg[64]_local
> will be used.
>
Looks good to me.
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
> Signed-off-by: wuqiang.matt <wuqiang.matt@bytedance.com>
> ---
> arch/arc/include/asm/cmpxchg.h | 28 ++++++++++++++++++++++++++++
> 1 file changed, 28 insertions(+)
>
> diff --git a/arch/arc/include/asm/cmpxchg.h b/arch/arc/include/asm/cmpxchg.h
> index bf46514f6f12..91429f2350df 100644
> --- a/arch/arc/include/asm/cmpxchg.h
> +++ b/arch/arc/include/asm/cmpxchg.h
> @@ -80,6 +80,34 @@
>
> #endif
>
> +/*
> + * always make arch_cmpxchg[64]_local available, native cmpxchg
> + * will be used if available, then generic_cmpxchg[64]_local
> + */
> +#include <asm-generic/cmpxchg-local.h>
> +static inline unsigned long __cmpxchg_local(volatile void *ptr,
> + unsigned long old,
> + unsigned long new, int size)
> +{
> + switch (size) {
> +#ifdef CONFIG_ARC_HAS_LLSC
> + case 4:
> + return __cmpxchg_32((int32_t *)ptr, old, new);
> +#endif
> + default:
> + return __generic_cmpxchg_local(ptr, old, new, size);
> + }
> +
> + return old;
> +}
> +#define arch_cmpxchg_local(ptr, o, n) ({ \
> + (__typeof__(*ptr))__cmpxchg_local((ptr), \
> + (unsigned long)(o), \
> + (unsigned long)(n), \
> + sizeof(*(ptr))); \
> +})
> +#define arch_cmpxchg64_local(ptr, o, n) __generic_cmpxchg64_local((ptr), (o), (n))
> +
> /*
> * xchg
> */
> --
> 2.40.1
>
--
Masami Hiramatsu (Google) <mhiramat@kernel.org>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v2 3/4] locking/atomic: openrisc: arch_cmpxchg[64]_local undefined
2023-11-04 9:16 ` [PATCH v2 3/4] locking/atomic: openrisc: " wuqiang.matt
@ 2023-11-08 1:03 ` Masami Hiramatsu
0 siblings, 0 replies; 9+ messages in thread
From: Masami Hiramatsu @ 2023-11-08 1:03 UTC (permalink / raw)
To: wuqiang.matt
Cc: vgupta, bcain, jonas, stefan.kristiansson, shorne, geert,
andi.shyti, mingo, palmer, andrzej.hajda, arnd, peterz,
linux-snps-arc, linux-kernel, linux-hexagon, linux-openrisc,
linux-trace-kernel, mattwu, linux
On Sat, 4 Nov 2023 17:16:14 +0800
"wuqiang.matt" <wuqiang.matt@bytedance.com> wrote:
> For architectures that support native cmpxchg, we'd like to
> implement arch_cmpxchg[64]_local with the native variants of
> supported data size. If not, the generci_cmpxchg[64]_local
> will be used.
Looks good to me.
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
>
> Signed-off-by: wuqiang.matt <wuqiang.matt@bytedance.com>
> ---
> arch/openrisc/include/asm/cmpxchg.h | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/arch/openrisc/include/asm/cmpxchg.h b/arch/openrisc/include/asm/cmpxchg.h
> index 8ee151c072e4..f1ffe8b6f5ef 100644
> --- a/arch/openrisc/include/asm/cmpxchg.h
> +++ b/arch/openrisc/include/asm/cmpxchg.h
> @@ -139,6 +139,12 @@ static inline unsigned long __cmpxchg(volatile void *ptr, unsigned long old,
> (unsigned long)(n), \
> sizeof(*(ptr))); \
> })
> +#define arch_cmpxchg_local arch_cmpxchg
> +
> +/* always make arch_cmpxchg64_local available for openrisc */
> +#include <asm-generic/cmpxchg-local.h>
> +
> +#define arch_cmpxchg64_local(ptr, o, n) __generic_cmpxchg64_local((ptr), (o), (n))
>
> /*
> * This function doesn't exist, so you'll get a linker error if
> --
> 2.40.1
>
--
Masami Hiramatsu (Google) <mhiramat@kernel.org>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v2 4/4] locking/atomic: hexagon: arch_cmpxchg[64]_local undefined
2023-11-04 9:16 ` [PATCH v2 4/4] locking/atomic: hexagon: " wuqiang.matt
@ 2023-11-08 1:04 ` Masami Hiramatsu
0 siblings, 0 replies; 9+ messages in thread
From: Masami Hiramatsu @ 2023-11-08 1:04 UTC (permalink / raw)
To: wuqiang.matt
Cc: vgupta, bcain, jonas, stefan.kristiansson, shorne, geert,
andi.shyti, mingo, palmer, andrzej.hajda, arnd, peterz,
linux-snps-arc, linux-kernel, linux-hexagon, linux-openrisc,
linux-trace-kernel, mattwu, linux, kernel test robot
On Sat, 4 Nov 2023 17:16:15 +0800
"wuqiang.matt" <wuqiang.matt@bytedance.com> wrote:
> For architectures that support native cmpxchg, we'd like to
> implement arch_cmpxchg[64]_local with the native variants of
> supported data size. If not, the generci_cmpxchg[64]_local
> will be used.
>
> Reported-by: kernel test robot <lkp@intel.com>
> Closes: https://lore.kernel.org/oe-kbuild-all/202310272207.tLPflya4-lkp@intel.com/
>
Looks good to me.
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
But I need hexagon's maintainer's comment too.
Thank you,
> Signed-off-by: wuqiang.matt <wuqiang.matt@bytedance.com>
> ---
> arch/hexagon/include/asm/cmpxchg.h | 51 +++++++++++++++++++++++++++++-
> 1 file changed, 50 insertions(+), 1 deletion(-)
>
> diff --git a/arch/hexagon/include/asm/cmpxchg.h b/arch/hexagon/include/asm/cmpxchg.h
> index bf6cf5579cf4..2b5e5bbaf807 100644
> --- a/arch/hexagon/include/asm/cmpxchg.h
> +++ b/arch/hexagon/include/asm/cmpxchg.h
> @@ -8,6 +8,8 @@
> #ifndef _ASM_CMPXCHG_H
> #define _ASM_CMPXCHG_H
>
> +#include <linux/build_bug.h>
> +
> /*
> * __arch_xchg - atomically exchange a register and a memory location
> * @x: value to swap
> @@ -51,13 +53,15 @@ __arch_xchg(unsigned long x, volatile void *ptr, int size)
> * variable casting.
> */
>
> -#define arch_cmpxchg(ptr, old, new) \
> +#define __cmpxchg_32(ptr, old, new) \
> ({ \
> __typeof__(ptr) __ptr = (ptr); \
> __typeof__(*(ptr)) __old = (old); \
> __typeof__(*(ptr)) __new = (new); \
> __typeof__(*(ptr)) __oldval = 0; \
> \
> + BUILD_BUG_ON(sizeof(*(ptr)) != 4); \
> + \
> asm volatile( \
> "1: %0 = memw_locked(%1);\n" \
> " { P0 = cmp.eq(%0,%2);\n" \
> @@ -72,4 +76,49 @@ __arch_xchg(unsigned long x, volatile void *ptr, int size)
> __oldval; \
> })
>
> +#define __cmpxchg(ptr, old, val, size) \
> +({ \
> + __typeof__(*(ptr)) oldval; \
> + \
> + switch (size) { \
> + case 4: \
> + oldval = __cmpxchg_32(ptr, old, val); \
> + break; \
> + default: \
> + BUILD_BUG(); \
> + oldval = val; \
> + break; \
> + } \
> + \
> + oldval; \
> +})
> +
> +#define arch_cmpxchg(ptr, o, n) __cmpxchg((ptr), (o), (n), sizeof(*(ptr)))
> +
> +/*
> + * always make arch_cmpxchg[64]_local available, native cmpxchg
> + * will be used if available, then generic_cmpxchg[64]_local
> + */
> +#include <asm-generic/cmpxchg-local.h>
> +
> +#define arch_cmpxchg_local(ptr, old, val) \
> +({ \
> + __typeof__(*(ptr)) retval; \
> + int size = sizeof(*(ptr)); \
> + \
> + switch (size) { \
> + case 4: \
> + retval = __cmpxchg_32(ptr, old, val); \
> + break; \
> + default: \
> + retval = __generic_cmpxchg_local(ptr, old, \
> + val, size); \
> + break; \
> + } \
> + \
> + retval; \
> +})
> +
> +#define arch_cmpxchg64_local(ptr, o, n) __generic_cmpxchg64_local((ptr), (o), (n))
> +
> #endif /* _ASM_CMPXCHG_H */
> --
> 2.40.1
>
--
Masami Hiramatsu (Google) <mhiramat@kernel.org>
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2023-11-08 1:04 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-11-04 9:16 [PATCH v2 0/4] locking/atomic: arch_cmpxchg[64]_local undefined wuqiang.matt
2023-11-04 9:16 ` [PATCH v2 1/4] locking/atomic: arc: data size checking in arch_cmpxchg wuqiang.matt
2023-11-08 1:02 ` Masami Hiramatsu
2023-11-04 9:16 ` [PATCH v2 2/4] locking/atomic: arc: arch_cmpxchg[64]_local undefined wuqiang.matt
2023-11-08 1:02 ` Masami Hiramatsu
2023-11-04 9:16 ` [PATCH v2 3/4] locking/atomic: openrisc: " wuqiang.matt
2023-11-08 1:03 ` Masami Hiramatsu
2023-11-04 9:16 ` [PATCH v2 4/4] locking/atomic: hexagon: " wuqiang.matt
2023-11-08 1:04 ` Masami Hiramatsu
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).