On 11/11/2021 2:29 AM, Xu, Wei1 wrote:
Running in 'csum' mode, the 'from' and 'to' parameters are not
in the correct order when calling rte_ether_addr_copy() which
means the 'src/dst' mac addresses in the mbuf will be overwriten.


Hi Wei,

Original code looks good. What are you trying to fix exactly?

API order is, 'rte_ether_addr_copy(from, to)'.

I assume your expectation is packet to keep original src & dst MAC address,
but it is not working that way, dest addrs is written by user configured
'peer address'.

With your change for each packet, packet mac address written to testpmd
configuration, which doesn't make sense.

As a result, the packets will not be recognized and received
by the receiver(s).

Test CLI:
./app/dpdk-testpmd -n 1 -l 1-2 -a 09:00.0 -- -i --forward-mode=csum

Fixes: 10f4620(app/testpmd: modify mac in csum forwarding)

v2:
        - fixed indentation and long line warnings.

Signed-off-by: Wei Xu <wei1...@intel.com>
---
  app/test-pmd/csumonly.c | 8 ++++----
  1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/app/test-pmd/csumonly.c b/app/test-pmd/csumonly.c
index 8526d9158a..08484fcda2 100644
--- a/app/test-pmd/csumonly.c
+++ b/app/test-pmd/csumonly.c
@@ -872,10 +872,10 @@ pkt_burst_checksum_forward(struct fwd_stream *fs)
                 * and inner headers */
eth_hdr = rte_pktmbuf_mtod(m, struct rte_ether_hdr *);
-               rte_ether_addr_copy(&peer_eth_addrs[fs->peer_addr],
-                               &eth_hdr->dst_addr);
-               rte_ether_addr_copy(&ports[fs->tx_port].eth_addr,
-                               &eth_hdr->src_addr);
+               rte_ether_addr_copy(&eth_hdr->dst_addr,
+                               &peer_eth_addrs[fs->peer_addr]);
+               rte_ether_addr_copy(&eth_hdr->src_addr,
+                               &ports[fs->tx_port].eth_addr);
                parse_ethernet(eth_hdr, &info);
                l3_hdr = (char *)eth_hdr + info.l2_len;

Reply via email to