net/sfc: correct Rx checksum offloads for tunnel packets

[dpdk.git] / doc / guides / nics / sfc_efx.rst

diff --git a/doc/guides/nics/sfc_efx.rst b/doc/guides/nics/sfc_efx.rst
index 973a4a0..813cb6f 100644 (file)
--- a/doc/guides/nics/sfc_efx.rst
+++ b/doc/guides/nics/sfc_efx.rst
@@ -69,6 +69,8 @@ SFC EFX PMD has support for:
 
 - IPv4/IPv6 TCP/UDP receive checksum offload
 
+- Inner IPv4/IPv6 TCP/UDP receive checksum offload
+
 - Received packet type information
 
 - Receive side scaling (RSS)
@@ -116,6 +118,26 @@ required in the receive buffer.
 It should be taken into account when mbuf pool for receive is created.
 
 
+Tunnels support
+---------------
+
+NVGRE, VXLAN and GENEVE tunnels are supported on SFN8xxx family adapters
+with full-feature firmware variant running.
+**sfboot** should be used to configure NIC to run full-feature firmware variant.
+See Solarflare Server Adapter User's Guide for details.
+
+SFN8xxx family adapters provide either inner or outer packet classes.
+If adapter firmware advertises support for tunnels then the PMD
+configures the hardware to report inner classes, and outer classes are
+not reported in received packets.
+However, for VXLAN and GENEVE tunnels the PMD does report UDP as the
+outer layer 4 packet type.
+
+SFN8xxx family adapters report GENEVE packets as VXLAN.
+If UDP ports are configured for only one tunnel type then it is safe to
+treat VXLAN packet type indication as the corresponding UDP tunnel type.
+
+
 Flow API support
 ----------------
 
@@ -245,12 +267,16 @@ boolean parameters value.
   features available and required by the datapath implementation.
   **efx** chooses libefx-based datapath which supports VLAN insertion
   (full-feature firmware variant only), TSO and multi-segment mbufs.
+  Mbuf segments may come from different mempools, and mbuf reference
+  counters are treated responsibly.
   **ef10** chooses EF10 (SFN7xxx, SFN8xxx) native datapath which is
   more efficient than libefx-based but has no VLAN insertion and TSO
   support yet.
+  Mbuf segments may come from different mempools, and mbuf reference
+  counters are treated responsibly.
   **ef10_simple** chooses EF10 (SFN7xxx, SFN8xxx) native datapath which
   is even more faster then **ef10** but does not support multi-segment
-  mbufs.
+  mbufs, disallows multiple mempools and neglects mbuf reference counters.
 
 - ``perf_profile`` [auto|throughput|low-latency] (default **throughput**)