net/bnxt: limit Rx representor packets per poll

Without some limit on the number of packets transferred from the
HW ring to the representor ring per burst receive call, an entire ring's
worth of packets can be transferred. This can break assumptions
about ring indices (index on return could be identical to the index
on entry, which is assumed to mean that no packets were processed),
and can result in representor packets being dropped unnecessarily
due to representor ring overflow.

Fix by limiting the number of representor packets transferred per
poll to requested burst size.

Fixes: 6dc83230b43b ("net/bnxt: support port representor data path")
Cc: stable@dpdk.org
Signed-off-by: Lance Richardson <lance.richardson@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>

diff --git a/drivers/net/bnxt/bnxt_rxr.c b/drivers/net/bnxt/bnxt_rxr.c
index 445e6150d1dbf9471d89242485fd2f9ac6f4a4ce..e2d388e69fcc29fc3a9b16ccc9cfe566633871d3 100644 (file)
--- a/drivers/net/bnxt/bnxt_rxr.c
+++ b/drivers/net/bnxt/bnxt_rxr.c
@@ -901,7 +901,7 @@ uint16_t bnxt_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
                }
 
                raw_cons = NEXT_RAW_CMP(raw_cons);
-               if (nb_rx_pkts == nb_pkts || evt)
+               if (nb_rx_pkts == nb_pkts || nb_rep_rx_pkts == nb_pkts || evt)
                        break;
                /* Post some Rx buf early in case of larger burst processing */
                if (nb_rx_pkts == BNXT_RX_POST_THRESH)