[dpdk-dev] [PATCH] net/thunderx: check data offset alignment requirement

Jerin Jacob jerin.jacob at caviumnetworks.com
Mon Mar 6 06:32:48 CET 2017


On Mon, Mar 06, 2017 at 11:00:01AM +0530, Hemant Agrawal wrote:
> On 3/4/2017 6:16 PM, Jerin Jacob wrote:
> > nicvf HW expects the DMA address of the packet data to be
> > aligned with cache line size.
> > 
> > Packet data offset is a function of struct mbuf size,
> > mbuf private size and headroom. mbuf private size can
> > be changed from the application in pool creation, this
> > check detects HW alignment requirement constraint in pmd
> > start function.
> > 
> > Signed-off-by: Jerin Jacob <jerin.jacob at caviumnetworks.com>
> > ---
> >  drivers/net/thunderx/nicvf_ethdev.c | 13 ++++++++++---
> >  1 file changed, 10 insertions(+), 3 deletions(-)
> > 
> > diff --git a/drivers/net/thunderx/nicvf_ethdev.c b/drivers/net/thunderx/nicvf_ethdev.c
> > index 1060319..ad7b5d6 100644
> > --- a/drivers/net/thunderx/nicvf_ethdev.c
> > +++ b/drivers/net/thunderx/nicvf_ethdev.c
> > @@ -1410,7 +1410,7 @@ static int
> >  nicvf_vf_start(struct rte_eth_dev *dev, struct nicvf *nic, uint32_t rbdrsz)
> >  {
> >  	int ret;
> > -	uint16_t qidx;
> > +	uint16_t qidx, data_off;
> >  	uint32_t total_rxq_desc, nb_rbdr_desc, exp_buffs;
> >  	uint64_t mbuf_phys_off = 0;
> >  	struct nicvf_rxq *rxq;
> > @@ -1451,10 +1451,17 @@ nicvf_vf_start(struct rte_eth_dev *dev, struct nicvf *nic, uint32_t rbdrsz)
> >  				     nic->vf_id, qidx, rxq->pool->name);
> >  			return -ENOMEM;
> >  		}
> > -		rxq->mbuf_phys_off -= nicvf_mbuff_meta_length(mbuf);
> > -		rxq->mbuf_phys_off -= RTE_PKTMBUF_HEADROOM;
> > +		data_off = nicvf_mbuff_meta_length(mbuf);
> > +		data_off += RTE_PKTMBUF_HEADROOM;
> >  		rte_pktmbuf_free(mbuf);
> > 
> > +		if (data_off % RTE_CACHE_LINE_SIZE) {
> > +			PMD_INIT_LOG(ERR, "unaligned data_offset=%d delta=%d\n",
> > +				data_off, data_off % RTE_CACHE_LINE_SIZE);
> 
> Do you also want to log about the particular pool having this issue?
> rxq->pool->name

Yes.I will add the pool name in v2.

Thanks.
> 
> > +			return -EINVAL;
> > +		}
> > +		rxq->mbuf_phys_off -= data_off;
> > +
> >  		if (mbuf_phys_off == 0)
> >  			mbuf_phys_off = rxq->mbuf_phys_off;
> >  		if (mbuf_phys_off != rxq->mbuf_phys_off) {
> > 
> otherwise, you may add:
> Acked-by: Hemant Agrawal <hemant.agrawal at nxp.com>
> 


More information about the dev mailing list