Messages in this thread | | | Date | Fri, 19 May 2023 12:02:44 +0200 | From | Simon Horman <> | Subject | Re: [PATCH net-next v3] octeontx2-pf: Add support for page pool |
| |
On Fri, May 19, 2023 at 12:43:52PM +0530, Ratheesh Kannoth wrote: > Page pool for each rx queue enhance rx side performance > by reclaiming buffers back to each queue specific pool. DMA > mapping is done only for first allocation of buffers. > As subsequent buffers allocation avoid DMA mapping, > it results in performance improvement. > > Image | Performance > ------------ | ------------ > Vannila | 3Mpps > | > with this | 42Mpps > change | > --------------------------- > > Signed-off-by: Ratheesh Kannoth <rkannoth@marvell.com>
...
> @@ -1205,10 +1226,28 @@ void otx2_sq_free_sqbs(struct otx2_nic *pfvf) > } > } > > +void otx2_free_bufs(struct otx2_nic *pfvf, struct otx2_pool *pool, > + u64 iova, int size) > +{ > + u64 pa = otx2_iova_to_phys(pfvf->iommu_domain, iova); > + struct page *page = virt_to_head_page(phys_to_virt(pa));
nit: please arrange local variables in networking code in reverse xmas tree order - longest line to shortest.
u64 pa = otx2_iova_to_phys(pfvf->iommu_domain, iova); struct page *page;
page = virt_to_head_page(phys_to_virt(pa));
The following tool can check this:
https://github.com/ecree-solarflare/xmastree
...
> @@ -1186,11 +1185,13 @@ bool otx2_sq_append_skb(struct net_device *netdev, struct otx2_snd_queue *sq, > } > EXPORT_SYMBOL(otx2_sq_append_skb); > > -void otx2_cleanup_rx_cqes(struct otx2_nic *pfvf, struct otx2_cq_queue *cq) > +void otx2_cleanup_rx_cqes(struct otx2_nic *pfvf, struct otx2_cq_queue *cq, int qidx) > { > struct nix_cqe_rx_s *cqe; > int processed_cqe = 0; > - u64 iova, pa; > + struct otx2_pool *pool; > + u16 pool_id; > + u64 iova;
Likewise here.
> > if (pfvf->xdp_prog) > xdp_rxq_info_unreg(&cq->xdp_rxq);
...
| |