[dpdk-dev] net/enic: fix possible Rx corruption
Commit Message
Initialize the mbuf data offset to RTE_PKTMBUF_HEADROOM as the
enic takes ownership of them. If allocated mbufs had some offset
other than RTE_PKTMBUF_HEADROOM, the application would read mbuf
data starting at the wrong place and misinterpret the packet.
Fixes: 856d7ba7ed22 ("net/enic: support scattered Rx")
Reviewed-by: Nelson Escobar <neescoba@cisco.com>
Signed-off-by: John Daley <johndale@cisco.com>
---
Please include this patch in 16.07 since the bug found in late testing
can cause data corruption.
drivers/net/enic/enic_main.c | 1 +
drivers/net/enic/enic_rxtx.c | 1 +
2 files changed, 2 insertions(+)
Comments
2016-07-19 15:42, John Daley:
> Initialize the mbuf data offset to RTE_PKTMBUF_HEADROOM as the
> enic takes ownership of them. If allocated mbufs had some offset
> other than RTE_PKTMBUF_HEADROOM, the application would read mbuf
> data starting at the wrong place and misinterpret the packet.
>
> Fixes: 856d7ba7ed22 ("net/enic: support scattered Rx")
>
> Reviewed-by: Nelson Escobar <neescoba@cisco.com>
> Signed-off-by: John Daley <johndale@cisco.com>
Applied, thanks
@@ -317,6 +317,7 @@ enic_alloc_rx_queue_mbufs(struct enic *enic, struct vnic_rq *rq)
return -ENOMEM;
}
+ mb->data_off = RTE_PKTMBUF_HEADROOM;
dma_addr = (dma_addr_t)(mb->buf_physaddr
+ RTE_PKTMBUF_HEADROOM);
rq_enet_desc_enc(rqd, dma_addr,
@@ -300,6 +300,7 @@ enic_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
(struct cq_enet_rq_desc *)&cqd);
/* Push descriptor for newly allocated mbuf */
+ nmb->data_off = RTE_PKTMBUF_HEADROOM;
dma_addr = (dma_addr_t)(nmb->buf_physaddr +
RTE_PKTMBUF_HEADROOM);
rq_enet_desc_enc(rqd_ptr, dma_addr,