On 1/10/2018 1:38 AM, Lu, Wenzhuo wrote: > Hi Ferruh, > >> -----Original Message----- >> From: Yigit, Ferruh >> Sent: Wednesday, January 10, 2018 1:58 AM >> To: Lu, Wenzhuo <wenzhuo...@intel.com>; dev@dpdk.org >> Cc: Wu, Jingjing <jingjing...@intel.com> >> Subject: Re: [dpdk-dev] [PATCH v5 12/14] net/avf: enable sse vector Rx Tx >> func >> >> On 1/8/2018 5:13 AM, Wenzhuo Lu wrote: >>> From: Jingjing Wu <jingjing...@intel.com> >>> >>> Signed-off-by: Jingjing Wu <jingjing...@intel.com> >> >> <...> >> >>> @@ -31,5 +31,6 @@ SRCS-$(CONFIG_RTE_LIBRTE_AVF_PMD) += >> avf_common.c >>> SRCS-$(CONFIG_RTE_LIBRTE_AVF_PMD) += avf_ethdev.c >>> SRCS-$(CONFIG_RTE_LIBRTE_AVF_PMD) += avf_vchnl.c >>> SRCS-$(CONFIG_RTE_LIBRTE_AVF_PMD) += avf_rxtx.c >>> +SRCS-$(CONFIG_RTE_LIBRTE_AVF_INC_VECTOR) += avf_rxtx_vec_sse.c >> >> You may need to wrap this with an arch check to not break other >> architecture builds. > I think we don't need to wrap it because I remember it's said SSE is > supported by default. > For example, before support ARM64 by this patch, > 'b20971b6cca0d01c41ff06e161581754810bfeb7 net/ixgbe: implement vector driver > for ARM' > Ixgbe makefile just looks like this.
By default CONFIG_RTE_LIBRTE_AVF_PMD=y and CONFIG_RTE_LIBRTE_AVF_INC_VECTOR=y, so when you compile DPDK in an arm box, won't is cause a build error? I think either this PMD or its vector mode should be disabled for other architecture configs, or wrap the file with architecture check.
