kni: increase kernel version requirement for VA

[dpdk.git] / doc / guides / prog_guide / kernel_nic_interface.rst

diff --git a/doc/guides/prog_guide/kernel_nic_interface.rst b/doc/guides/prog_guide/kernel_nic_interface.rst
index 5afd05c..848b002 100644 (file)
--- a/doc/guides/prog_guide/kernel_nic_interface.rst
+++ b/doc/guides/prog_guide/kernel_nic_interface.rst
@@ -235,6 +235,15 @@ application functions:
     will be called which calls ``rte_eth_promiscuous_<enable|disable>()``
     on the specified ``port_id``.
 
+``config_allmulticast``:
+
+    Called when the user changes the allmulticast state of the KNI interface.
+    For example, when the user runs ``ifconfig <ifaceX> [-]allmulti``. If the
+    user sets this callback function to NULL, but sets the ``port_id`` field to
+    a value other than -1, a default callback handler in the rte_kni library
+    ``kni_config_allmulticast()`` will be called which calls
+    ``rte_eth_allmulticast_<enable|disable>()`` on the specified ``port_id``.
+
 In order to run these callbacks, the application must periodically call
 the ``rte_kni_handle_request()`` function.  Any user callback function
 registered will be called directly from ``rte_kni_handle_request()`` so
@@ -245,7 +254,7 @@ to create a separate thread or secondary process to periodically call
 
 The KNI interfaces can be deleted by a DPDK application with
 ``rte_kni_release()``.  All KNI interfaces not explicitly deleted will be
-deleted when the the ``/dev/kni`` device is closed, either explicitly with
+deleted when the ``/dev/kni`` device is closed, either explicitly with
 ``rte_kni_close()`` or when the DPDK application is closed.
 
 DPDK mbuf Flow
@@ -268,12 +277,16 @@ Use Case: Ingress
 -----------------
 
 On the DPDK RX side, the mbuf is allocated by the PMD in the RX thread context.
-This thread will enqueue the mbuf in the rx_q FIFO.
+This thread will enqueue the mbuf in the rx_q FIFO,
+and the next pointers in mbuf-chain will convert to physical address.
 The KNI thread will poll all KNI active devices for the rx_q.
 If an mbuf is dequeued, it will be converted to a sk_buff and sent to the net stack via netif_rx().
-The dequeued mbuf must be freed, so the same pointer is sent back in the free_q FIFO.
+The dequeued mbuf must be freed, so the same pointer is sent back in the free_q FIFO,
+and next pointers must convert back to virtual address if exists before put in the free_q FIFO.
 
 The RX thread, in the same main loop, polls this FIFO and frees the mbuf after dequeuing it.
+The address conversion of the next pointer is to prevent the chained mbuf
+in different hugepage segments from causing kernel crash.
 
 Use Case: Egress
 ----------------
@@ -287,6 +300,20 @@ The sk_buff is then freed and the mbuf sent in the tx_q FIFO.
 The DPDK TX thread dequeues the mbuf and sends it to the PMD via ``rte_eth_tx_burst()``.
 It then puts the mbuf back in the cache.
 
+IOVA = VA: Support
+------------------
+
+KNI operates in IOVA_VA scheme when
+
+- LINUX_VERSION_CODE > KERNEL_VERSION(4, 9, 0) and
+- EAL option `iova-mode=va` is passed or bus IOVA scheme in the DPDK is selected
+  as RTE_IOVA_VA.
+
+Due to IOVA to KVA address translations, based on the KNI use case there
+can be a performance impact. For mitigation, forcing IOVA to PA via EAL
+"--iova-mode=pa" option can be used, IOVA_DC bus iommu scheme can also
+result in IOVA as PA.
+
 Ethtool
 -------