[PATCH] riscv: Optimize hweight API with Zbb extension

[PATCH] riscv: Optimize hweight API with Zbb extension

[Xiao Wang]

The Hamming Weight of a number is the total number of bits set in it, so
the cpop/cpopw instruction from Zbb extension can be used to accelerate
hweight() API.

Signed-off-by: Xiao Wang <xiao.w.wang@intel.com>
---
 arch/riscv/include/asm/arch_hweight.h | 78 +++++++++++++++++++++++++++
 arch/riscv/include/asm/bitops.h       |  4 +-
 2 files changed, 81 insertions(+), 1 deletion(-)
 create mode 100644 arch/riscv/include/asm/arch_hweight.h

diff --git a/arch/riscv/include/asm/arch_hweight.h b/arch/riscv/include/asm/arch_hweight.h
new file mode 100644
index 000000000000..c20236a0725b
--- /dev/null
+++ b/arch/riscv/include/asm/arch_hweight.h
@@ -0,0 +1,78 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Based on arch/x86/include/asm/arch_hweight.h
+ */
+
+#ifndef _ASM_RISCV_HWEIGHT_H
+#define _ASM_RISCV_HWEIGHT_H
+
+#include <asm/alternative-macros.h>
+#include <asm/hwcap.h>
+
+#if (BITS_PER_LONG == 64)
+#define CPOPW	"cpopw "
+#elif (BITS_PER_LONG == 32)
+#define CPOPW	"cpop "
+#else
+#error "Unexpected BITS_PER_LONG"
+#endif
+
+static __always_inline unsigned int __arch_hweight32(unsigned int w)
+{
+#ifdef CONFIG_RISCV_ISA_ZBB
+	asm_volatile_goto(ALTERNATIVE("j %l[legacy]", "nop", 0,
+				      RISCV_ISA_EXT_ZBB, 1)
+			  : : : : legacy);
+
+	asm (".option push\n"
+	     ".option arch,+zbb\n"
+	     CPOPW "%0, %0\n"
+	     ".option pop\n"
+	     : "+r" (w) : :);
+
+	return w;
+
+legacy:
+#endif
+	return __sw_hweight32(w);
+}
+
+static inline unsigned int __arch_hweight16(unsigned int w)
+{
+	return __arch_hweight32(w & 0xffff);
+}
+
+static inline unsigned int __arch_hweight8(unsigned int w)
+{
+	return __arch_hweight32(w & 0xff);
+}
+
+#if BITS_PER_LONG == 64
+static __always_inline unsigned long __arch_hweight64(__u64 w)
+{
+# ifdef CONFIG_RISCV_ISA_ZBB
+	asm_volatile_goto(ALTERNATIVE("j %l[legacy]", "nop", 0,
+				      RISCV_ISA_EXT_ZBB, 1)
+			  : : : : legacy);
+
+	asm (".option push\n"
+	     ".option arch,+zbb\n"
+	     "cpop %0, %0\n"
+	     ".option pop\n"
+	     : "+r" (w) : :);
+
+	return w;
+
+legacy:
+# endif
+	return __sw_hweight64(w);
+}
+#else /* BITS_PER_LONG == 64 */
+static inline unsigned long __arch_hweight64(__u64 w)
+{
+	return  __arch_hweight32((u32)w) +
+		__arch_hweight32((u32)(w >> 32));
+}
+#endif /* !(BITS_PER_LONG == 64) */
+
+#endif /* _ASM_RISCV_HWEIGHT_H */
diff --git a/arch/riscv/include/asm/bitops.h b/arch/riscv/include/asm/bitops.h
index b212c2708cda..f7c167646460 100644
--- a/arch/riscv/include/asm/bitops.h
+++ b/arch/riscv/include/asm/bitops.h
@@ -271,7 +271,9 @@ static __always_inline int variable_fls(unsigned int x)
 #include <asm-generic/bitops/fls64.h>
 #include <asm-generic/bitops/sched.h>
 
-#include <asm-generic/bitops/hweight.h>
+#include <asm/arch_hweight.h>
+
+#include <asm-generic/bitops/const_hweight.h>
 
 #if (BITS_PER_LONG == 64)
 #define __AMO(op)	"amo" #op ".d"
-- 
2.25.1


_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv

Re: [PATCH] riscv: Optimize hweight API with Zbb extension

[Charlie Jenkins]

On Sun, Nov 12, 2023 at 05:52:44PM +0800, Xiao Wang wrote:
> The Hamming Weight of a number is the total number of bits set in it, so
> the cpop/cpopw instruction from Zbb extension can be used to accelerate
> hweight() API.
> 
> Signed-off-by: Xiao Wang <xiao.w.wang@intel.com>
> ---
>  arch/riscv/include/asm/arch_hweight.h | 78 +++++++++++++++++++++++++++
>  arch/riscv/include/asm/bitops.h       |  4 +-
>  2 files changed, 81 insertions(+), 1 deletion(-)
>  create mode 100644 arch/riscv/include/asm/arch_hweight.h
> 
> diff --git a/arch/riscv/include/asm/arch_hweight.h b/arch/riscv/include/asm/arch_hweight.h
> new file mode 100644
> index 000000000000..c20236a0725b
> --- /dev/null
> +++ b/arch/riscv/include/asm/arch_hweight.h
> @@ -0,0 +1,78 @@
> +/* SPDX-License-Identifier: GPL-2.0 */
> +/*
> + * Based on arch/x86/include/asm/arch_hweight.h
> + */
> +
> +#ifndef _ASM_RISCV_HWEIGHT_H
> +#define _ASM_RISCV_HWEIGHT_H
> +
> +#include <asm/alternative-macros.h>
> +#include <asm/hwcap.h>
> +
> +#if (BITS_PER_LONG == 64)
> +#define CPOPW	"cpopw "
> +#elif (BITS_PER_LONG == 32)
> +#define CPOPW	"cpop "
> +#else
> +#error "Unexpected BITS_PER_LONG"
> +#endif
> +
> +static __always_inline unsigned int __arch_hweight32(unsigned int w)
> +{
> +#ifdef CONFIG_RISCV_ISA_ZBB
> +	asm_volatile_goto(ALTERNATIVE("j %l[legacy]", "nop", 0,
> +				      RISCV_ISA_EXT_ZBB, 1)
> +			  : : : : legacy);
> +
> +	asm (".option push\n"
> +	     ".option arch,+zbb\n"
> +	     CPOPW "%0, %0\n"
> +	     ".option pop\n"
> +	     : "+r" (w) : :);
> +
> +	return w;
> +
> +legacy:
> +#endif
> +	return __sw_hweight32(w);
> +}
> +
> +static inline unsigned int __arch_hweight16(unsigned int w)
> +{
> +	return __arch_hweight32(w & 0xffff);
> +}
> +
> +static inline unsigned int __arch_hweight8(unsigned int w)
> +{
> +	return __arch_hweight32(w & 0xff);
> +}
> +
> +#if BITS_PER_LONG == 64
> +static __always_inline unsigned long __arch_hweight64(__u64 w)
> +{
> +# ifdef CONFIG_RISCV_ISA_ZBB
> +	asm_volatile_goto(ALTERNATIVE("j %l[legacy]", "nop", 0,
> +				      RISCV_ISA_EXT_ZBB, 1)
> +			  : : : : legacy);
> +
> +	asm (".option push\n"
> +	     ".option arch,+zbb\n"
> +	     "cpop %0, %0\n"
> +	     ".option pop\n"
> +	     : "+r" (w) : :);
> +
> +	return w;
> +
> +legacy:
> +# endif
> +	return __sw_hweight64(w);
> +}
> +#else /* BITS_PER_LONG == 64 */
> +static inline unsigned long __arch_hweight64(__u64 w)
> +{
> +	return  __arch_hweight32((u32)w) +
> +		__arch_hweight32((u32)(w >> 32));
> +}
> +#endif /* !(BITS_PER_LONG == 64) */
> +
> +#endif /* _ASM_RISCV_HWEIGHT_H */
> diff --git a/arch/riscv/include/asm/bitops.h b/arch/riscv/include/asm/bitops.h
> index b212c2708cda..f7c167646460 100644
> --- a/arch/riscv/include/asm/bitops.h
> +++ b/arch/riscv/include/asm/bitops.h
> @@ -271,7 +271,9 @@ static __always_inline int variable_fls(unsigned int x)
>  #include <asm-generic/bitops/fls64.h>
>  #include <asm-generic/bitops/sched.h>
>  
> -#include <asm-generic/bitops/hweight.h>
> +#include <asm/arch_hweight.h>
> +
> +#include <asm-generic/bitops/const_hweight.h>
>  
>  #if (BITS_PER_LONG == 64)
>  #define __AMO(op)	"amo" #op ".d"
> -- 
> 2.25.1
> 
> 
> _______________________________________________
> linux-riscv mailing list
> linux-riscv@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-riscv

Reviewed-by: Charlie Jenkins <charlie@rivosinc.com>


_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv

Re: [PATCH] riscv: Optimize hweight API with Zbb extension

[patchwork-bot+linux-riscv]

Hello:

This patch was applied to riscv/linux.git (fixes)
by Palmer Dabbelt <palmer@rivosinc.com>:

On Sun, 12 Nov 2023 17:52:44 +0800 you wrote:
> The Hamming Weight of a number is the total number of bits set in it, so
> the cpop/cpopw instruction from Zbb extension can be used to accelerate
> hweight() API.
> 
> Signed-off-by: Xiao Wang <xiao.w.wang@intel.com>
> ---
>  arch/riscv/include/asm/arch_hweight.h | 78 +++++++++++++++++++++++++++
>  arch/riscv/include/asm/bitops.h       |  4 +-
>  2 files changed, 81 insertions(+), 1 deletion(-)
>  create mode 100644 arch/riscv/include/asm/arch_hweight.h

Here is the summary with links:
  - riscv: Optimize hweight API with Zbb extension
    https://git.kernel.org/riscv/c/55ca8d7aa2af

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html



_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv