Re: [dpdk-dev] [PATCH v2 2/3] eal/acl: enable acl for armv7-a

[Jerin Jacob <jerin.jacob@caviumnetworks.com>]

On Thu, Dec 03, 2015 at 11:02:55PM +0800, Jianbo Liu wrote:
> Implement vqtbl1q_u8 intrinsic function, which is not support in armv7-a.
> 
> Signed-off-by: Jianbo Liu <jianbo.liu@linaro.org>

Acked-by: Jerin Jacob <jerin.jacob@caviumnetworks.com>

> ---
>  config/defconfig_arm-armv7a-linuxapp-gcc          |  1 -
>  lib/librte_acl/Makefile                           |  2 +-
>  lib/librte_acl/rte_acl.c                          |  5 ++++-
>  lib/librte_eal/common/include/arch/arm/rte_vect.h | 23 +++++++++++++++++++++++
>  4 files changed, 28 insertions(+), 3 deletions(-)
> 
> diff --git a/config/defconfig_arm-armv7a-linuxapp-gcc b/config/defconfig_arm-armv7a-linuxapp-gcc
> index 9924ff9..cbebd64 100644
> --- a/config/defconfig_arm-armv7a-linuxapp-gcc
> +++ b/config/defconfig_arm-armv7a-linuxapp-gcc
> @@ -53,7 +53,6 @@ CONFIG_RTE_LIBRTE_KNI=n
>  CONFIG_RTE_EAL_IGB_UIO=n
>  
>  # fails to compile on ARM
> -CONFIG_RTE_LIBRTE_ACL=n
>  CONFIG_RTE_LIBRTE_LPM=n
>  CONFIG_RTE_LIBRTE_TABLE=n
>  CONFIG_RTE_LIBRTE_PIPELINE=n
> diff --git a/lib/librte_acl/Makefile b/lib/librte_acl/Makefile
> index 897237d..2e394c9 100644
> --- a/lib/librte_acl/Makefile
> +++ b/lib/librte_acl/Makefile
> @@ -49,7 +49,7 @@ SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_bld.c
>  SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_gen.c
>  SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_run_scalar.c
>  
> -ifeq ($(CONFIG_RTE_ARCH_ARM64),y)
> +ifneq ($(filter y,$(CONFIG_RTE_ARCH_ARM) $(CONFIG_RTE_ARCH_ARM64)),)
>  SRCS-$(CONFIG_RTE_LIBRTE_ACL) += acl_run_neon.c
>  CFLAGS_acl_run_neon.o += -flax-vector-conversions -Wno-maybe-uninitialized
>  else
> diff --git a/lib/librte_acl/rte_acl.c b/lib/librte_acl/rte_acl.c
> index e2fdebd..4ba9786 100644
> --- a/lib/librte_acl/rte_acl.c
> +++ b/lib/librte_acl/rte_acl.c
> @@ -114,8 +114,11 @@ rte_acl_init(void)
>  {
>  	enum rte_acl_classify_alg alg = RTE_ACL_CLASSIFY_DEFAULT;
>  
> -#ifdef RTE_ARCH_ARM64
> +#if defined(RTE_ARCH_ARM64)
>  	alg =  RTE_ACL_CLASSIFY_NEON;
> +#elif defined(RTE_ARCH_ARM)
> +	if (rte_cpu_get_flag_enabled(RTE_CPUFLAG_NEON))
> +		alg =  RTE_ACL_CLASSIFY_NEON;
>  #else
>  #ifdef CC_AVX2_SUPPORT
>  	if (rte_cpu_get_flag_enabled(RTE_CPUFLAG_AVX2))
> diff --git a/lib/librte_eal/common/include/arch/arm/rte_vect.h b/lib/librte_eal/common/include/arch/arm/rte_vect.h
> index 21cdb4d..a33c054 100644
> --- a/lib/librte_eal/common/include/arch/arm/rte_vect.h
> +++ b/lib/librte_eal/common/include/arch/arm/rte_vect.h
> @@ -53,6 +53,29 @@ typedef union rte_xmm {
>  	double   pd[XMM_SIZE / sizeof(double)];
>  } __attribute__((aligned(16))) rte_xmm_t;
>  
> +#ifdef RTE_ARCH_ARM
> +/* NEON intrinsic vqtbl1q_u8() is not supported in ARMv7-A(AArch32) */
> +static __inline uint8x16_t
> +vqtbl1q_u8(uint8x16_t a, uint8x16_t b)
> +{
> +	uint8_t i, pos;
> +	rte_xmm_t rte_a, rte_b, rte_ret;
> +
> +	vst1q_u8(rte_a.u8, a);
> +	vst1q_u8(rte_b.u8, b);
> +
> +	for (i = 0; i < 16; i++) {
> +		pos = rte_b.u8[i];
> +		if (pos < 16)
> +			rte_ret.u8[i] = rte_a.u8[pos];
> +		else
> +			rte_ret.u8[i] = 0;
> +	}
> +
> +	return vld1q_u8(rte_ret.u8);
> +}
> +#endif
> +
>  #ifdef __cplusplus
>  }
>  #endif
> -- 
> 1.8.3.1
>