Hi, > -----Original Message----- > From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Zhang, Helin > Sent: Friday, March 31, 2017 3:34 PM > To: Thomas Monjalon; Yongseok Koh; Yigit, Ferruh; Ananyev, Konstantin > Cc: dev@dpdk.org; adrien.mazarg...@6wind.com; > nelio.laranje...@6wind.com > Subject: Re: [dpdk-dev] [PATCH v2 1/3] lib/librte_ether: remove requirement > of aligned RETA size > > > > > -----Original Message----- > > From: Thomas Monjalon [mailto:thomas.monja...@6wind.com] > > Sent: Thursday, March 30, 2017 8:38 AM > > To: Yongseok Koh; Yigit, Ferruh; Zhang, Helin; Ananyev, Konstantin > > Cc: dev@dpdk.org; adrien.mazarg...@6wind.com; > > nelio.laranje...@6wind.com > > Subject: Re: [PATCH v2 1/3] lib/librte_ether: remove requirement of > > aligned RETA size > > > > 2017-03-20 16:04, Yongseok Koh: > > > In rte_eth_check_reta_mask(), it is required to align the size of > > > the RETA table to RTE_RETA_GROUP_SIZE but as the size can be less > > > than the limit, this should be removed. The change is also applied > > > to a command of > > testpmd. > > > > > > Signed-off-by: Yongseok Koh <ys...@mellanox.com> > > [...] > > > - if (reta_size != RTE_ALIGN(reta_size, RTE_RETA_GROUP_SIZE)) { > > > - RTE_PMD_DEBUG_TRACE("Invalid reta size, should be %u > > aligned\n", > > > - > > RTE_RETA_GROUP_SIZE); > > > - return -EINVAL; > > > - } > > > - > > > - num = reta_size / RTE_RETA_GROUP_SIZE; > > > + num = (reta_size + RTE_RETA_GROUP_SIZE - 1) / > > RTE_RETA_GROUP_SIZE; > > > > There is no comment for this constraint neither in the code nor in the > > commit: http://dpdk.org/commit/66c594904 So, I guess it can be removed. > > If a check is needed, it could be added in the relevant drivers. > > > > Helin, Konstantin, please check for Intel drivers. > Hi Thomas > > Thank you very much for the reminder! > We will check that and see if there is any impacts to Intel drivers. I don't think it has any impact to the drivers. To my opinion, it's a good fix as it makes the name ' reta_size' more reasonable.
> > Regards, > Helin > > > > Ferruh, please take care of this series.