Changes between Initial Version and Version 1 of Component/Vci Master Nic


Ignore:
Timestamp:
Oct 9, 2019, 7:16:33 PM (5 years ago)
Author:
alain
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • Component/Vci Master Nic

    v1 v1  
     1[wiki:Component SocLib Components General Index]
     2
     3= !VciMasterNic =
     4
     5== 1) Functional Description ==
     6
     7The !VciMasterNic component, is a GMII compliant, network controller
     8for Gigabit Ethernet network, with a built-in DMA capability.
     9
     10It can support a throughput of 1 Gigabit/s, as long as the system clock frequency
     11is larger or equal to the GMII clock frequency (ie 125 MHz).
     12
     13To improve the throughput, this component supports up to 4 channels,
     14indexed by the source IP address for the received (RX) packets,
     15and indexed by the destination IP address for the sent (TX) packets.
     16The actual number of channels is an hardware parameter that cannot be larger than 8.
     17 
     18Regarding the GMII physical interface, this simulation model supports three modes
     19of operation, defined by a constructor parameter:
     20 * '''NIC_MODE_FILE''': Both the RX packets stream an the TX packets stream are read/written from/to dedicated files "nic_rx_file.txt" and "nic_tx_dile.txt", stored in the same directory as the top.cpp file.
     21 * '''NIC_MODE_SYNTHESIS''': The TX packet stream is still written to the "nic_tx_file.txt" file, but the RX packet stream is synthesised. The packet length (between 64 and 1538 bytes) and the source MAC address (8 possible values) are pseudo-random numbers.
     22 * '''NIC_MODE_TAP''': The TX and RX packet streams are send and received to and from the physical network controller of the workstation running the simulation.
     23
     24The packet length can have any value, from 64 to 1542 bytes.
     25
     26The minimal data transfer unit between software and the NIC is a 4K bytes '''container''',
     27containing an integer number of variable size packets.
     28The max number of packets in a container is 66 packets.
     29
     30The received packets (RX) and the sent packets (TX) are stored in
     31two memory mapped software FIFOs, implemented as chained buffers.
     32Each slot in these FIFOs is a 4 Kbytes container. The number of containers,
     33defining the queue depth, is a software defined parameter.
     34
     35The container format is defined below:
     36
     37The first 34 words define the fixed-format container header :
     38||   word0  ||  NB_WORDS || NB_PACKETS ||
     39||   word1  ||  PLEN[0]  || PLEN[1]    ||
     40||    ...   ||   ...     ||   ...      ||
     41||   word33 ||  PLEN[64] || PLEN[65]   ||
     42
     43 * NB_PACKETS is the actual number of packets in the container.
     44 * NB_WORDS is the number of useful words in the container.
     45 * PLEN[i] is the number of bytes for packet[i].
     46
     47The packets are stored in the (1024 - 34) following words,
     48The max number of packets in a container is 66 packets, and the packets are word-aligned.
     49
     50For the DMA engines, a container has only two states (full or empty), defined
     51by a single bit, called the container "status".
     52To access both the container status, and the data contained in the container, the DMA
     53engines use two physical addresses, that are packed in a 64 bits ''container descriptor'':
     54 * desc[25:0]  contain bits[31:6] of the "full" status physical address.
     55 * desc[51:26] contain bits[31:6] of the "buffer" physical address.
     56 * desc[63:52] contain the common 12 physical address extension bits.
     57
     58Inside the NIC controller, each channel implements a 2 slots chained buffer (two
     59containers) for RX, and another 2 slots chained buffer( two containers) for TX.
     60For each channel,  the build-in RX_DMA engine moves the RX containers from
     61the internal 2 slots chained buffer to the external chained buffer implementing
     62the RX queue in memory.
     63Another build-in TX-DMA engine moves the TX containers from the external
     64chained buffer implementing the TX queue in memory, to the internal TX 2 slots
     65chained buffer.
     66
     67To improve the throughput for one specific channel, the DMA engines use ''pipelined bursts'': The burst length cannot be larger than 64 bytes, but each channel send 4 pipelined VCI transactions to mask the round-trip latency. Therefore, thi NIC controller can control up to 32 parallel VCI transactions (4 channels * 4 bursts * 2 directions).
     68The CMD/RSP matching uses both the VCI TRDID and PKTID fields:
     69 * the channel index is sent in TRDID[3:2]
     70 * the burst   index is sent in TRDID[1:0]
     71 * the is_rx   bit   is sent in SRCID     
     72
     73== 2) Addressable registers and buffers ==
     74
     75In a virtualized environment each channel segment will be mapped in the address space of a different virtual machine.
     76Each channel takes a segment of 32 Kbytes in the address space, to simplify the address decoding, but only 20K bytes are used.
     77
     78 * The first 4 Kbytes contain the RX_0 container data
     79 * The next  4 Kbytes contain the RX_1 container data
     80 * The next  4 Kbytes contain the TX_0 container data
     81 * The next  4 Kbytes contain the TX_1 container data
     82 * The next  4 Kbytes contain the channel addressable registers:
     83
     84|| NIC_RX_STS_0      || RX_0 status (full or empty) || read/write ||
     85|| NIC_RX_STS_1      || RX_1 status (full or empty) || read/write ||
     86|| NIC_TX_STS_0      || TX_0 status (full or empty) || read/write ||
     87|| NIC_TX_STS_1      || TX_1 status (full or empty) || read/write ||
     88|| NIC_RX_DESC_LO_0  || RX_0 descriptor low word    || read/write ||
     89|| NIC_RX_DESC_HI_0  || RX_0 descriptor high word   || read/write ||
     90|| NIC_RX_DESC_LO_1  || RX_1 descriptor low word    || read/write ||
     91|| NIC_RX_DESC_HI_1  || RX_1 descriptor high word   || read/write ||
     92|| NIC_TX_DESC_LO_0  || TX_0 descriptor low word    || read/write ||
     93|| NIC_TX_DESC_HI_0  || TX_0 descriptor high word   || read/write ||
     94|| NIC_TX_DESC_LO_1  || TX_1 descriptor low word    || read/write ||
     95|| NIC_TX_DESC_HI_1  || TX_1 descriptor high word   || read/write ||
     96|| NIC_MAC_4         || MAC address 32 LSB bits     || read_only  ||
     97|| NIC_MAC_2         || MAC address 16 MSB bits     || read_only  ||
     98|| NIC_RX_RUN        || RX channel activated        || write_only ||
     99|| NIC_TX_RUN        || TX channel  activated       || write_only ||
     100
     101
     102On top of the channels segments is the hypervisor segment, taking 4 Kbytes,
     103and containing the global configuration registers: (all read/write).
     104In a virtualized environment, the corresponding page should not be mapped
     105in the virtual machines address spaces, as it should not accessed by the virtual machines.
     106|| Register name         || function                                       || Reset value  ||
     107|| NIC_G_VIS             || bitfield / bit N = 0 -> channel N is disabled  || all inactive ||
     108|| NIC_G_ON              || NIC active if non zero (inactive at reset)     || inactive     ||
     109|| NIC_G_BC_ENABLE       || boolean / broadcast enabled if true            || disabled     ||
     110|| NIC_G_TDM_ENABLE      || boolean / enable TDM dor TX if true            || disabled     ||
     111|| NIC_G_TDM_PERIOD      || value of TDM time slot                         ||              ||
     112|| NIC_G_PYPASS_ENABLE   || boolean / enable bypass for TX if true         || enabled      ||
     113|| NIC_G_MAC_4[8]        || default MAC address 32 LSB bits for channel[i] ||              ||
     114|| NIC_G_MAC_2[8]        || default MAC address 16 LSB bits for channel[i] ||              ||
     115
     116The Hypervisor segment contains also various event counters for statistics (read/write)
     117
     118|| NIC_G_NPKT_RX_G2S_RECEIVED       || number of packets received on GMII RX port      ||
     119|| NIC_G_NPKT_RX_G2S_DISCARDED      || number of RX packets discarded by RX_G2S FSM    ||
     120||                                  ||                                                 ||
     121|| NIC_G_NPKT_RX_DES_SUCCESS        || number of RX packets transmited by RX_DES FSM   ||
     122|| NIC_G_NPKT_RX_DES_TOO_SMALL      || number of discarded too small RX packets        ||
     123|| NIC_G_NPKT_RX_DES_TOO_BIG        || number of discarded too big RX packets          ||
     124|| NIC_G_NPKT_RX_DES_MFIFO_FULL     || number of discarded RX packets for fifo full    ||
     125|| NIC_G_NPKT_RX_DES_CRC_FAIL       || number of discarded RX packets for checksum     ||
     126||                                  ||                                                 ||
     127|| NIC_G_NPKT_RX_DISPATCH_RECEIVED  || number of packets received by RX_DISPATCH FSM   ||
     128|| NIC_G_NPKT_RX_DISPATCH_BROADCAST || number of broadcast RX packets received         ||
     129|| NIC_G_NPKT_RX_DISPATCH_DST_FAIL  || number of discarded RX packets for DST MAC      ||
     130|| NIC_G_NPKT_RX_DISPATCH_CH_FULL   || number of discarded RX packets for channel full ||
     131||                                  ||                                                 ||
     132|| NIC_G_NPKT_TX_DISPATCH_RECEIVED  || number of packets received by TX_DISPATCH FSM   ||
     133|| NIC_G_NPKT_TX_DISPATCH_TOO_SMALL || number of discarded too small TX packets        ||
     134|| NIC_G_NPKT_TX_DISPATCH_TOO_BIG   || number of discarded too big TX packets          ||
     135|| NIC_G_NPKT_TX_DISPATCH_SRC_FAIL  || number of discarded TX packets because SRC MAC  ||
     136|| NIC_G_NPKT_TX_DISPATCH_BROADCAST || number of broadcast TX packets received         ||
     137|| NIC_G_NPKT_TX_DISPATCH_BYPASS    || number of bypassed TX->RX packets               ||
     138|| NIC_G_NPKT_TX_DISPATCH_TRANSMIT  || number of transmit TX packets                   ||
     139
     140For extensibility issues, you should access all these registers using the globally-defined offsets in file
     141 source:trunk/soclib/soclib/module/connectivity_component/vci_multi_nic/include/soclib/multi_nic.h
     142
     143This hardware component checks for segmentation violation, and can be used as a default target.
     144
     145== 3) Component definition & usage ==
     146
     147source:trunk/soclib/soclib/module/connectivity_component/vci_multi_nic/caba/metadata/vci_multi_nic.sd
     148
     149{{{
     150Uses( 'vci_multi_nic' )
     151}}}
     152
     153== 4) CABA  Implementation ==
     154
     155=== CABA sources ===
     156 
     157 * interface : source:trunk/soclib/soclib/module/connectivity_component/vci_multi_nic/caba/source/include/vci_multi_nic.h
     158 * implementation : source:trunk/soclib/soclib/module/connectivity_component/vci_multi_nic/caba/source/src/vci_multi_nic.cpp
     159
     160=== CABA Constructor parameters ===
     161{{{
     162VciMultiNic(
     163     sc_module_name name,   //  Component Name
     164     const soclib::common::IntTab &tgtid,  // Target index
     165     const soclib::common::MappingTable &mt,   // MappingTable
     166     const size_t channels,   // Number of channels
     167     const uint32_t mac4,   //  MAC address 32 LSB bits
     168     const uint32_t mac2,   //  MAC address 16 MSB bits
     169     const int mode,   // GMII physical interface modeling
     170     const uint32_t inter_frame_gap);   // delay between two packets
     171
     172}}}
     173
     174===  CABA Ports ===
     175
     176 * '''p_resetn''' : Global system reset
     177 * '''p_clk''' : Global system clock
     178 * '''p_vci''' : The VCI target port
     179 * '''p_rx_irq[k]''' : As many RX IRQ ports as the number of channels
     180 * '''p_tx_irq[k]''' : As many TX IRQ ports as the number of channels
     181
     182== 4) TLM-DT implementation ==
     183
     184The TLM-DT implementation is not available yet.