search for: xen_netif_rx_ring_size

...; -} - int xen_netbk_rx_ring_full(struct xenvif *vif) { - RING_IDX peek = vif->rx_req_cons_peek; - RING_IDX needed = max_required_rx_slots(vif); + RING_IDX peek = vif->rx_req_cons_peek; - return ((vif->rx.sring->req_prod - peek) < needed) || - ((vif->rx.rsp_prod_pvt + XEN_NETIF_RX_RING_SIZE - peek) < needed); + return ((vif->rx.sring->req_prod < peek) || + (vif->rx.rsp_prod_pvt + XEN_NETIF_RX_RING_SIZE < peek)); } int xen_netbk_must_stop_queue(struct xenvif *vif) @@ -314,8 +302,7 @@ int xen_netbk_must_stop_queue(struct xenvif *vif) if (!xen_netbk_rx_ring_full(...

Hi, I ran into an issue where netback driver is crashing with BUG_ON(npo.meta_prod > ARRAY_SIZE(netbk->meta)). It is happening in Intel 10Gbps network when larger mtu values are used. The problem seems to be the way the slots are counted. After applying this patch things ran fine in my environment. I request to validate my changes. Thanks Siva

When counting required slots for skb, netback directly uses DIV_ROUND_UP to get slots required by header data. This is wrong when offset in the page of header data is not zero, and is also inconsistent with following calculation for required slot in netbk_gop_skb. In netbk_gop_skb, required slots are calculated based on offset and len in page of header data. It is possible that required slots