On Thu, Apr 05, 2018 at 08:36:18AM -0700, Stephen Hemminger wrote:
> I noticed Coverity defect in DPDK number 124563 was assigned to me.
>
> It looks like vmxnet3 driver code doesn't handle case where host incorrectly
> sends a frame with out setting SOP bit in first segment. This would lead to
> rxq->start_seq being NULL.
>
> Maybe something like this would fix it. I don't have VMware (or time) to
> test.
>
> diff --git a/drivers/net/vmxnet3/vmxnet3_rxtx.c
> b/drivers/net/vmxnet3/vmxnet3_rxtx.c
> index 57557492e0f5..3118d94add90 100644
> --- a/drivers/net/vmxnet3/vmxnet3_rxtx.c
> +++ b/drivers/net/vmxnet3/vmxnet3_rxtx.c
> @@ -813,6 +813,13 @@ vmxnet3_recv_pkts(void *rx_queue, struct rte_mbuf
> **rx_pkts, uint16_t nb_pkts)
>
> RTE_ASSERT(rxd->btype == VMXNET3_RXD_BTYPE_BODY);
>
> + if (unlikely(start == NULL)) {
> + PMD_RX_LOG(ERR, "Missing sop");
> +
> + rte_pktmbuf_free_seg(rxm);
> + goto rcd_done;
> + }
> +
> start->pkt_len += rxm->data_len;
> start->nb_segs++;
>
This looks reasonable to me, though my vmware instance is down at the moment, so
I'm unable to test
>
> PS: the email for the VMXNET3 maintainer Shrikrishna Khare
> <skh...@vmware.com> bounces.
>
I think it needs to be updated to point at pv-driv...@vmware.com, thats what
they use in the kernel these days
Neil
