diff options
Diffstat (limited to 'Documentation/networking')
| -rw-r--r-- | Documentation/networking/00-INDEX | 6 | ||||
| -rw-r--r-- | Documentation/networking/dpaa2/ethernet-driver.rst | 185 | ||||
| -rw-r--r-- | Documentation/networking/dpaa2/index.rst | 1 | ||||
| -rw-r--r-- | Documentation/networking/iavf.txt (renamed from Documentation/networking/i40evf.txt) | 16 | ||||
| -rw-r--r-- | Documentation/networking/netvsc.txt | 9 | ||||
| -rw-r--r-- | Documentation/networking/tcp.txt | 101 |
6 files changed, 206 insertions, 112 deletions
diff --git a/Documentation/networking/00-INDEX b/Documentation/networking/00-INDEX index 02a323c43261..f4f2b5d6c8d8 100644 --- a/Documentation/networking/00-INDEX +++ b/Documentation/networking/00-INDEX @@ -94,8 +94,8 @@ gianfar.txt - Gianfar Ethernet Driver. i40e.txt - README for the Intel Ethernet Controller XL710 Driver (i40e). -i40evf.txt - - Short note on the Driver for the Intel(R) XL710 X710 Virtual Function +iavf.txt + - README for the Intel Ethernet Adaptive Virtual Function Driver (iavf). ieee802154.txt - Linux IEEE 802.15.4 implementation, API and drivers igb.txt @@ -198,8 +198,6 @@ tc-actions-env-rules.txt - rules for traffic control (tc) actions. timestamping.txt - overview of network packet timestamping variants. -tcp.txt - - short blurb on how TCP output takes place. tcp-thin.txt - kernel tuning options for low rate 'thin' TCP streams. team.txt diff --git a/Documentation/networking/dpaa2/ethernet-driver.rst b/Documentation/networking/dpaa2/ethernet-driver.rst new file mode 100644 index 000000000000..90ec940749e8 --- /dev/null +++ b/Documentation/networking/dpaa2/ethernet-driver.rst @@ -0,0 +1,185 @@ +.. SPDX-License-Identifier: GPL-2.0 +.. include:: <isonum.txt> + +=============================== +DPAA2 Ethernet driver +=============================== + +:Copyright: |copy| 2017-2018 NXP + +This file provides documentation for the Freescale DPAA2 Ethernet driver. + +Supported Platforms +=================== +This driver provides networking support for Freescale DPAA2 SoCs, e.g. +LS2080A, LS2088A, LS1088A. + + +Architecture Overview +===================== +Unlike regular NICs, in the DPAA2 architecture there is no single hardware block +representing network interfaces; instead, several separate hardware resources +concur to provide the networking functionality: + +- network interfaces +- queues, channels +- buffer pools +- MAC/PHY + +All hardware resources are allocated and configured through the Management +Complex (MC) portals. MC abstracts most of these resources as DPAA2 objects +and exposes ABIs through which they can be configured and controlled. A few +hardware resources, like queues, do not have a corresponding MC object and +are treated as internal resources of other objects. + +For a more detailed description of the DPAA2 architecture and its object +abstractions see *Documentation/networking/dpaa2/overview.rst*. + +Each Linux net device is built on top of a Datapath Network Interface (DPNI) +object and uses Buffer Pools (DPBPs), I/O Portals (DPIOs) and Concentrators +(DPCONs). + +Configuration interface:: + + ----------------------- + | DPAA2 Ethernet Driver | + ----------------------- + . . . + . . . + . . . . . . . . . . . . + . . . + . . . + ---------- ---------- ----------- + | DPBP API | | DPNI API | | DPCON API | + ---------- ---------- ----------- + . . . software + ======= . ========== . ============ . =================== + . . . hardware + ------------------------------------------ + | MC hardware portals | + ------------------------------------------ + . . . + . . . + ------ ------ ------- + | DPBP | | DPNI | | DPCON | + ------ ------ ------- + +The DPNIs are network interfaces without a direct one-on-one mapping to PHYs. +DPBPs represent hardware buffer pools. Packet I/O is performed in the context +of DPCON objects, using DPIO portals for managing and communicating with the +hardware resources. + +Datapath (I/O) interface:: + + ----------------------------------------------- + | DPAA2 Ethernet Driver | + ----------------------------------------------- + | ^ ^ | | + | | | | | + enqueue| dequeue| data | dequeue| seed | + (Tx) | (Rx, TxC)| avail.| request| buffers| + | | notify| | | + | | | | | + V | | V V + ----------------------------------------------- + | DPIO Driver | + ----------------------------------------------- + | | | | | software + | | | | | ================ + | | | | | hardware + ----------------------------------------------- + | I/O hardware portals | + ----------------------------------------------- + | ^ ^ | | + | | | | | + | | | V | + V | ================ V + ---------------------- | ------------- + queues ---------------------- | | Buffer pool | + ---------------------- | ------------- + ======================= + Channel + +Datapath I/O (DPIO) portals provide enqueue and dequeue services, data +availability notifications and buffer pool management. DPIOs are shared between +all DPAA2 objects (and implicitly all DPAA2 kernel drivers) that work with data +frames, but must be affine to the CPUs for the purpose of traffic distribution. + +Frames are transmitted and received through hardware frame queues, which can be +grouped in channels for the purpose of hardware scheduling. The Ethernet driver +enqueues TX frames on egress queues and after transmission is complete a TX +confirmation frame is sent back to the CPU. + +When frames are available on ingress queues, a data availability notification +is sent to the CPU; notifications are raised per channel, so even if multiple +queues in the same channel have available frames, only one notification is sent. +After a channel fires a notification, is must be explicitly rearmed. + +Each network interface can have multiple Rx, Tx and confirmation queues affined +to CPUs, and one channel (DPCON) for each CPU that services at least one queue. +DPCONs are used to distribute ingress traffic to different CPUs via the cores' +affine DPIOs. + +The role of hardware buffer pools is storage of ingress frame data. Each network +interface has a privately owned buffer pool which it seeds with kernel allocated +buffers. + + +DPNIs are decoupled from PHYs; a DPNI can be connected to a PHY through a DPMAC +object or to another DPNI through an internal link, but the connection is +managed by MC and completely transparent to the Ethernet driver. + +:: + + --------- --------- --------- + | eth if1 | | eth if2 | | eth ifn | + --------- --------- --------- + . . . + . . . + . . . + --------------------------- + | DPAA2 Ethernet Driver | + --------------------------- + . . . + . . . + . . . + ------ ------ ------ ------- + | DPNI | | DPNI | | DPNI | | DPMAC |----+ + ------ ------ ------ ------- | + | | | | | + | | | | ----- + =========== ================== | PHY | + ----- + +Creating a Network Interface +============================ +A net device is created for each DPNI object probed on the MC bus. Each DPNI has +a number of properties which determine the network interface configuration +options and associated hardware resources. + +DPNI objects (and the other DPAA2 objects needed for a network interface) can be +added to a container on the MC bus in one of two ways: statically, through a +Datapath Layout Binary file (DPL) that is parsed by MC at boot time; or created +dynamically at runtime, via the DPAA2 objects APIs. + + +Features & Offloads +=================== +Hardware checksum offloading is supported for TCP and UDP over IPv4/6 frames. +The checksum offloads can be independently configured on RX and TX through +ethtool. + +Hardware offload of unicast and multicast MAC filtering is supported on the +ingress path and permanently enabled. + +Scatter-gather frames are supported on both RX and TX paths. On TX, SG support +is configurable via ethtool; on RX it is always enabled. + +The DPAA2 hardware can process jumbo Ethernet frames of up to 10K bytes. + +The Ethernet driver defines a static flow hashing scheme that distributes +traffic based on a 5-tuple key: src IP, dst IP, IP proto, L4 src port, +L4 dst port. No user configuration is supported for now. + +Hardware specific statistics for the network interface as well as some +non-standard driver stats can be consulted through ethtool -S option. diff --git a/Documentation/networking/dpaa2/index.rst b/Documentation/networking/dpaa2/index.rst index 10bea113a7bc..67bd87fe6c53 100644 --- a/Documentation/networking/dpaa2/index.rst +++ b/Documentation/networking/dpaa2/index.rst @@ -7,3 +7,4 @@ DPAA2 Documentation overview dpio-driver + ethernet-driver diff --git a/Documentation/networking/i40evf.txt b/Documentation/networking/iavf.txt index e9b3035b95d0..cc902a2369d6 100644 --- a/Documentation/networking/i40evf.txt +++ b/Documentation/networking/iavf.txt @@ -2,7 +2,7 @@ Linux* Base Driver for Intel(R) Network Connection ================================================== Intel Ethernet Adaptive Virtual Function Linux driver. -Copyright(c) 2013-2017 Intel Corporation. +Copyright(c) 2013-2018 Intel Corporation. Contents ======== @@ -11,20 +11,21 @@ Contents - Known Issues/Troubleshooting - Support -This file describes the i40evf Linux* Base Driver. +This file describes the iavf Linux* Base Driver. This driver +was formerly called i40evf. -The i40evf driver supports the below mentioned virtual function +The iavf driver supports the below mentioned virtual function devices and can only be activated on kernels running the i40e or newer Physical Function (PF) driver compiled with CONFIG_PCI_IOV. -The i40evf driver requires CONFIG_PCI_MSI to be enabled. +The iavf driver requires CONFIG_PCI_MSI to be enabled. -The guest OS loading the i40evf driver must support MSI-X interrupts. +The guest OS loading the iavf driver must support MSI-X interrupts. Supported Hardware ================== Intel XL710 X710 Virtual Function -Intel Ethernet Adaptive Virtual Function Intel X722 Virtual Function +Intel Ethernet Adaptive Virtual Function Identifying Your Adapter ======================== @@ -32,7 +33,8 @@ Identifying Your Adapter For more information on how to identify your adapter, go to the Adapter & Driver ID Guide at: - http://support.intel.com/support/go/network/adapter/idguide.htm + https://www.intel.com/content/www/us/en/support/articles/000005584/network-and-i-o/ethernet-products.html + Known Issues/Troubleshooting ============================ diff --git a/Documentation/networking/netvsc.txt b/Documentation/networking/netvsc.txt index 92f5b31392fa..3bfa635bbbd5 100644 --- a/Documentation/networking/netvsc.txt +++ b/Documentation/networking/netvsc.txt @@ -45,6 +45,15 @@ Features like packets and significantly reduces CPU usage under heavy Rx load. + Large Receive Offload (LRO), or Receive Side Coalescing (RSC) + ------------------------------------------------------------- + The driver supports LRO/RSC in the vSwitch feature. It reduces the per packet + processing overhead by coalescing multiple TCP segments when possible. The + feature is enabled by default on VMs running on Windows Server 2019 and + later. It may be changed by ethtool command: + ethtool -K eth0 lro on + ethtool -K eth0 lro off + SR-IOV support -------------- Hyper-V supports SR-IOV as a hardware acceleration option. If SR-IOV diff --git a/Documentation/networking/tcp.txt b/Documentation/networking/tcp.txt deleted file mode 100644 index 9c7139d57e57..000000000000 --- a/Documentation/networking/tcp.txt +++ /dev/null @@ -1,101 +0,0 @@ -TCP protocol -============ - -Last updated: 3 June 2017 - -Contents -======== - -- Congestion control -- How the new TCP output machine [nyi] works - -Congestion control -================== - -The following variables are used in the tcp_sock for congestion control: -snd_cwnd The size of the congestion window -snd_ssthresh Slow start threshold. We are in slow start if - snd_cwnd is less than this. -snd_cwnd_cnt A counter used to slow down the rate of increase - once we exceed slow start threshold. -snd_cwnd_clamp This is the maximum size that snd_cwnd can grow to. -snd_cwnd_stamp Timestamp for when congestion window last validated. -snd_cwnd_used Used as a highwater mark for how much of the - congestion window is in use. It is used to adjust - snd_cwnd down when the link is limited by the - application rather than the network. - -As of 2.6.13, Linux supports pluggable congestion control algorithms. -A congestion control mechanism can be registered through functions in -tcp_cong.c. The functions used by the congestion control mechanism are -registered via passing a tcp_congestion_ops struct to -tcp_register_congestion_control. As a minimum, the congestion control -mechanism must provide a valid name and must implement either ssthresh, -cong_avoid and undo_cwnd hooks or the "omnipotent" cong_control hook. - -Private data for a congestion control mechanism is stored in tp->ca_priv. -tcp_ca(tp) returns a pointer to this space. This is preallocated space - it -is important to check the size of your private data will fit this space, or -alternatively, space could be allocated elsewhere and a pointer to it could -be stored here. - -There are three kinds of congestion control algorithms currently: The -simplest ones are derived from TCP reno (highspeed, scalable) and just -provide an alternative congestion window calculation. More complex -ones like BIC try to look at other events to provide better -heuristics. There are also round trip time based algorithms like -Vegas and Westwood+. - -Good TCP congestion control is a complex problem because the algorithm -needs to maintain fairness and performance. Please review current -research and RFC's before developing new modules. - -The default congestion control mechanism is chosen based on the -DEFAULT_TCP_CONG Kconfig parameter. If you really want a particular default -value then you can set it using sysctl net.ipv4.tcp_congestion_control. The -module will be autoloaded if needed and you will get the expected protocol. If -you ask for an unknown congestion method, then the sysctl attempt will fail. - -If you remove a TCP congestion control module, then you will get the next -available one. Since reno cannot be built as a module, and cannot be -removed, it will always be available. - -How the new TCP output machine [nyi] works. -=========================================== - -Data is kept on a single queue. The skb->users flag tells us if the frame is -one that has been queued already. To add a frame we throw it on the end. Ack -walks down the list from the start. - -We keep a set of control flags - - - sk->tcp_pend_event - - TCP_PEND_ACK Ack needed - TCP_ACK_NOW Needed now - TCP_WINDOW Window update check - TCP_WINZERO Zero probing - - - sk->transmit_queue The transmission frame begin - sk->transmit_new First new frame pointer - sk->transmit_end Where to add frames - - sk->tcp_last_tx_ack Last ack seen - sk->tcp_dup_ack Dup ack count for fast retransmit - - -Frames are queued for output by tcp_write. We do our best to send the frames -off immediately if possible, but otherwise queue and compute the body -checksum in the copy. - -When a write is done we try to clear any pending events and piggy back them. -If the window is full we queue full sized frames. On the first timeout in -zero window we split this. - -On a timer we walk the retransmit list to send any retransmits, update the -backoff timers etc. A change of route table stamp causes a change of header -and recompute. We add any new tcp level headers and refinish the checksum -before sending. - |
