URL https://opencores.org/ocsvn/fpga-cf/fpga-cf/trunk
Subversion Repositories fpga-cf

[/] [fpga-cf/] [trunk/] [hdl/] [boardsupport/] [v4/] [tx_client_fifo_8.v] - Blame information for rev 2

Details | Compare with Previous | View Log

//-----------------------------------------------------------------------------
// Title      : 8-bit Client to Local-link Transmitter FIFO
// Project    : Virtex-4 Embedded Tri-Mode Ethernet MAC Wrapper
// File       : tx_client_fifo_8.v
// Version    : 4.8
//-----------------------------------------------------------------------------
//
// (c) Copyright 2004-2010 Xilinx, Inc. All rights reserved.
//
// This file contains confidential and proprietary information
// of Xilinx, Inc. and is protected under U.S. and
// international copyright and other intellectual property
// laws.
//
// DISCLAIMER
// This disclaimer is not a license and does not grant any
// rights to the materials distributed herewith. Except as
// otherwise provided in a valid license issued to you by
// Xilinx, and to the maximum extent permitted by applicable
// law: (1) THESE MATERIALS ARE MADE AVAILABLE "AS IS" AND
// WITH ALL FAULTS, AND XILINX HEREBY DISCLAIMS ALL WARRANTIES
// AND CONDITIONS, EXPRESS, IMPLIED, OR STATUTORY, INCLUDING
// BUT NOT LIMITED TO WARRANTIES OF MERCHANTABILITY, NON-
// INFRINGEMENT, OR FITNESS FOR ANY PARTICULAR PURPOSE; and
// (2) Xilinx shall not be liable (whether in contract or tort,
// including negligence, or under any other theory of
// liability) for any loss or damage of any kind or nature
// related to, arising under or in connection with these
// materials, including for any direct, or any indirect,
// special, incidental, or consequential loss or damage
// (including loss of data, profits, goodwill, or any type of
// loss or damage suffered as a result of any action brought
// by a third party) even if such damage or loss was
// reasonably foreseeable or Xilinx had been advised of the
// possibility of the same.
//
// CRITICAL APPLICATIONS
// Xilinx products are not designed or intended to be fail-
// safe, or for use in any application requiring fail-safe
// performance, such as life-support or safety devices or
// systems, Class III medical devices, nuclear facilities,
// applications related to the deployment of airbags, or any
// other applications that could lead to death, personal
// injury, or severe property or environmental damage
// (individually and collectively, "Critical
// Applications"). Customer assumes the sole risk and
// liability of any use of Xilinx products in Critical
// Applications, subject only to applicable laws and
// regulations governing limitations on product liability.
//
// THIS COPYRIGHT NOTICE AND DISCLAIMER MUST BE RETAINED AS
// PART OF THIS FILE AT ALL TIMES.
//
//-----------------------------------------------------------------------------
// Description: This is a transmitter side local link fifo implementation for
//              the design example of the Virtex-4 Ethernet MAC Wrapper
//              core.
//              
//              The transmit FIFO is created from 2 Block RAMs of size 2048
//              words of 8-bits per word, giving a total frame memory capacity
//              of 4096 bytes.
//
//              Valid frame data received from local link interface is written
//              into the Block RAM on the write clock.  The FIFO will store
//              frames upto 4kbytes in length.  If larger frames are written
//              to the FIFO the local-link interface will accept the rest of the
//              frame, but that frame will be dropped by the FIFO and
//              the overflow signal will be asserted.
//
//              The FIFO is designed to work with a minimum frame length of 14 bytes.
//              
//              When there is at least one complete frame in the FIFO,
//              the MAC transmitter client interface will be driven to
//              request frame transmission by placing the first byte of
//              the frame onto tx_data[7:0] and by asserting
//              tx_data_valid.  The MAC will later respond by asserting
//              tx_ack.  At this point the remaining frame data is read
//              out of the FIFO in a continuous burst. Data is read out
//              of the FIFO on the rd_clk.
//
//              If the generic FULL_DUPLEX_ONLY is set to false, the FIFO will
//              requeue and retransmit frames as requested by the MAC.  Once a
//              frame has been transmitted by the FIFO it is stored until the
//              possible retransmit window for that frame has expired.
//
//              The FIFO has been designed to operate with different clocks
//              on the write and read sides. The write clock (locallink clock)
//              can be an equal or faster frequency than the read clock (client clock).
//              The minimum write clock frequency is the read clock frequency divided
//              by 4.
//
//              The FIFO memory size can be increased by expanding the rd_addr
//              and wr_addr signal widths, to address further BRAMs.
//
//-----------------------------------------------------------------------------
 
`timescale 1ps / 1ps
 
 
module tx_client_fifo_8
    (
        // MAC Interface
        rd_clk,
        rd_sreset,
        rd_enable,
        tx_data,
        tx_data_valid,
        tx_ack,
        tx_collision,
        tx_retransmit,
        overflow,
 
        // Local-link Interface
        wr_clk,
        wr_sreset,
        wr_data,
        wr_sof_n,
        wr_eof_n,
        wr_src_rdy_n,
        wr_dst_rdy_n,
        wr_fifo_status
        );
 
  //---------------------------------------------------------------------------
  // Define Interface Signals
  //---------------------------------------------------------------------------
 
  // MAC Interface
  input        rd_clk;
  input        rd_sreset;
  input        rd_enable;
  output [7:0] tx_data;
  output       tx_data_valid;
  input        tx_ack;
  input        tx_collision;
  input        tx_retransmit;
  output       overflow;
 
  // Local-link Interface
  input        wr_clk;
  input        wr_sreset;
  input  [7:0] wr_data;
  input        wr_sof_n;
  input        wr_eof_n;
  input        wr_src_rdy_n;
  output       wr_dst_rdy_n;
  output [3:0] wr_fifo_status;
 
  // If FULL_DUPLEX_ONLY is 1 then all the half duplex logic in the FIFO is removed.
  // The default for the fifo is to include the half duplex functionality 
  parameter    FULL_DUPLEX_ONLY = 0;
 
  reg [7:0]    tx_data;
  reg          tx_data_valid;
  reg [3:0]    wr_fifo_status;
 
 
  //---------------------------------------------------------------------------
  // Define Internal Signals
  //---------------------------------------------------------------------------
 
  wire        GND;
  wire        VCC;
  wire [7:0]  GND_BUS;
 
  // Encode rd_state_machine states   
  parameter  IDLE_s = 4'b0000;      parameter  QUEUE1_s = 4'b0001;
  parameter  QUEUE2_s = 4'b0010;    parameter  QUEUE3_s = 4'b0011;
  parameter  QUEUE_ACK_s = 4'b0100; parameter  WAIT_ACK_s = 4'b0101;
  parameter  FRAME_s = 4'b0110;     parameter  DROP_s = 4'b0111;
  parameter  RETRANSMIT_s = 4'b1000;
 
  reg  [3:0]  rd_state;
  reg  [3:0]  rd_nxt_state;
 
  // Encode wr_state_machine states 
  parameter WAIT_s = 2'b00;  parameter DATA_s = 2'b01;
  parameter EOF_s = 2'b10;   parameter OVFLOW_s = 2'b11;
 
  reg  [1:0]  wr_state;
  reg  [1:0]  wr_nxt_state;
 
  reg  [7:0]  wr_data_bram;
  reg  [7:0]  wr_data_pipe[0:1];
  reg         wr_sof_pipe[0:1];
  reg         wr_eof_pipe[0:1];
  reg         wr_accept_pipe[0:1];
  reg         wr_accept_bram;
  reg  [0:0]  wr_eof_bram;
  reg  [11:0] wr_addr;
  wire        wr_addr_inc;
  wire        wr_start_addr_load;
  wire        wr_addr_reload;
  reg  [11:0] wr_start_addr;
  reg         wr_fifo_full;
  wire        wr_en;
  wire        wr_en_u;
  wire        wr_en_l;
  reg         wr_ovflow_dst_rdy;
  wire        wr_dst_rdy_int_n;
 
  reg         frame_in_fifo_sync;
  reg         frame_in_fifo;
  reg         rd_eof;
  reg         rd_eof_reg;
  reg         rd_eof_pipe;
  reg  [11:0] rd_addr;
  wire        rd_addr_inc;
  wire        rd_addr_reload;
  wire [7:0]  rd_data_bram_u;
  wire [7:0]  rd_data_bram_l;
  reg  [7:0]  rd_data_pipe_u;
  reg  [7:0]  rd_data_pipe_l;
  reg  [7:0]  rd_data_pipe;
  wire [0:0]  rd_eof_bram_u;
  wire [0:0]  rd_eof_bram_l;
  wire        rd_en;
  wire        rd_en_bram;
  reg         rd_bram_u;
  reg         rd_bram_u_reg;
 
  reg         rd_tran_frame_tog;
  reg         wr_tran_frame_tog;
  reg         wr_tran_frame_sync;
  reg         wr_tran_frame_delay;
  reg         rd_retran_frame_tog;
  reg         wr_retran_frame_tog;
  reg         wr_retran_frame_sync;
  reg         wr_retran_frame_delay;
  wire        wr_store_frame;
  reg         wr_transmit_frame;
  reg         wr_retransmit_frame;
  reg  [8:0]  wr_frames;
  reg         wr_frame_in_fifo;
 
  reg   [3:0] rd_16_count;
  wire        rd_txfer_en;
  reg  [11:0] rd_addr_txfer;
  reg         rd_txfer_tog;
  reg         wr_txfer_tog;
  reg         wr_txfer_tog_sync;
  reg         wr_txfer_tog_delay;
  wire        wr_txfer_en;
  reg  [11:0] wr_rd_addr;
  reg  [11:0] wr_addr_diff;
 
  reg         rd_drop_frame;
  reg         rd_retransmit;
  reg  [11:0] rd_start_addr;
  wire        rd_start_addr_load;
  wire        rd_start_addr_reload;
 
  reg  [11:0] rd_dec_addr;
 
  wire        rd_transmit_frame;
  wire        rd_retransmit_frame;
  reg         rd_col_window_expire;
  reg         rd_col_window_pipe[0:1];
  reg         wr_col_window_pipe[0:1];
 
  wire        wr_eof_state;
  reg         wr_eof_state_reg;
  wire        wr_fifo_overflow;
  reg  [9:0]  rd_slot_timer;
  reg         wr_col_window_expire;
  wire        rd_idle_state;
 
  reg         rd_enable_delay;
  reg         rd_enable_delay2;
 
 
  //---------------------------------------------------------------------------
  // Attributes for FIFO simulation and synthesis
  //---------------------------------------------------------------------------
  // ASYNC_REG attributes added to simulate actual behaviour under
  // asynchronous operating conditions.
  // synthesis attribute ASYNC_REG of wr_tran_frame_tog is "TRUE";
  // synthesis attribute ASYNC_REG of wr_retran_frame_tog is "TRUE";
  // synthesis attribute ASYNC_REG of frame_in_fifo_sync is "TRUE";
  // synthesis attribute ASYNC_REG of wr_rd_addr is "TRUE";
  // synthesis attribute ASYNC_REG of wr_txfer_tog is "TRUE";
  // synthesis attribute ASYNC_REG of wr_col_window_pipe[0] is "TRUE";
 
  // WRITE_MODE attributes added to Block RAM to mitigate port contention
  // synthesis attribute WRITE_MODE_A of ramgen_u is "READ_FIRST";
  // synthesis attribute WRITE_MODE_B of ramgen_u is "READ_FIRST";
  // synthesis attribute WRITE_MODE_A of ramgen_l is "READ_FIRST";
  // synthesis attribute WRITE_MODE_B of ramgen_l is "READ_FIRST";
 
 
 
  //---------------------------------------------------------------------------
  // Begin FIFO architecture
  //---------------------------------------------------------------------------
 
  assign GND = 1'b0;
  assign VCC = 1'b1;
  assign GND_BUS = 8'b0;
 
  // shouldn't have a functional impact - at 1G flop is high all the time
  // at other speeds it is high every other cycle with 2nd stage ensuring it is in phase with the normal
  // clk enable - THE ONLY PURPOSE OF THIS FLOP IS TO ENABLE THE rd_enable and rd_en FANOUT TO BE CONTROLLED
  always @(posedge rd_clk)
  begin
     rd_enable_delay <= rd_enable;
     rd_enable_delay2 <= rd_enable_delay;
  end
 
  //---------------------------------------------------------------------------
  // Write State machine and control
  //---------------------------------------------------------------------------
  // Write state machine
  // states are WAIT, DATA, EOF, OVFLOW
  // clock through next state of sm
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
         wr_state <= WAIT_s;
     else
         wr_state <= wr_nxt_state;
  end
 
  // decode next state, combinitorial
  // should never be able to overflow whilst not in the data state.
  always @(wr_state or wr_sof_pipe[1] or wr_eof_pipe[0] or wr_eof_pipe[1] or wr_eof_bram[0] or wr_fifo_overflow)
  begin
  case (wr_state)
     WAIT_s : begin
        if (wr_sof_pipe[1] == 1'b1)
           wr_nxt_state <= DATA_s;
        else
           wr_nxt_state <= WAIT_s;
        end
     DATA_s : begin
        // wait for the end of frame to be detected
        if (wr_fifo_overflow == 1'b1 && wr_eof_pipe[0] == 1'b0 && wr_eof_pipe[1] == 1'b0)
           wr_nxt_state <= OVFLOW_s;
        else if (wr_eof_pipe[1] == 1'b1)
           wr_nxt_state <= EOF_s;
        else
           wr_nxt_state <= DATA_s;
        end
     EOF_s : begin
        // if the start of frame is already in the pipe, a back to back frame
        // transmission has occured.  move straight back to frame state
        if (wr_sof_pipe[1] == 1'b1)
           wr_nxt_state <= DATA_s;
        else if (wr_eof_bram[0] == 1'b1)
           wr_nxt_state <= WAIT_s;
        else
           wr_nxt_state <= EOF_s;
        end
     OVFLOW_s : begin
        // wait until the end of frame is reached before clearing the overflow
        if (wr_eof_bram[0] == 1'b1)
           wr_nxt_state <= WAIT_s;
        else
           wr_nxt_state <= OVFLOW_s;
        end
     default : begin
        wr_nxt_state <= WAIT_s;
        end
  endcase
  end
 
 
  // decode output signals.
  assign wr_en = (wr_state == OVFLOW_s) ? 1'b0 : wr_accept_bram;
  assign wr_en_l = wr_en & !wr_addr[11];
  assign wr_en_u = wr_en & wr_addr[11];
 
  assign wr_addr_inc = wr_en;
 
  assign wr_addr_reload = (wr_state == OVFLOW_s) ? 1'b1 : 1'b0;
  assign wr_start_addr_load = (wr_state == EOF_s && wr_nxt_state == WAIT_s) ? 1'b1 :
                              (wr_state == EOF_s && wr_nxt_state == DATA_s) ? 1'b1 : 1'b0;
 
 
  // pause the local link flow when the fifo is full.
  assign wr_dst_rdy_int_n = (wr_state == OVFLOW_s) ? wr_ovflow_dst_rdy : wr_fifo_full;
  assign wr_dst_rdy_n = wr_dst_rdy_int_n;
 
  // when in overflow and have captured ovflow eof send dst rdy high again.
  assign overflow = (wr_state == OVFLOW_s) ? 1'b1 : 1'b0;
 
  // when in overflow and have captured ovflow eof send dst rdy high again.
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        wr_ovflow_dst_rdy <= 1'b0;
     else
        begin
        if (wr_fifo_overflow == 1'b1 && wr_state == DATA_s)
            wr_ovflow_dst_rdy <= 1'b0;
        else if (wr_eof_n == 1'b0 && wr_src_rdy_n == 1'b0)
            wr_ovflow_dst_rdy <= 1'b1;
        end
  end
 
  // eof signals for use in overflow logic
  assign wr_eof_state = (wr_state == EOF_s) ? 1'b1 : 1'b0;
 
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        wr_eof_state_reg <= 1'b0;
     else
        wr_eof_state_reg <= wr_eof_state;
  end
 
  //---------------------------------------------------------------------------
  // Read state machine and control
  //---------------------------------------------------------------------------
 
  // clock through the read state machine
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
        rd_state <= IDLE_s;
     else if (rd_enable == 1'b1)
        rd_state <= rd_nxt_state;
  end
 
  //---------------------------------------------------------------------------
  // Full Duplex Only State Machine
generate if (FULL_DUPLEX_ONLY == 1) begin : gen_fd_sm
  // decode the next state
  always @(rd_state or frame_in_fifo or rd_eof or tx_ack)
  begin
  case (rd_state)
           IDLE_s : begin
              // if there is a frame in the fifo start to queue the new frame
              // to the output
              if (frame_in_fifo == 1'b1)
                 rd_nxt_state <= QUEUE1_s;
              else
                 rd_nxt_state <= IDLE_s;
              end
           QUEUE1_s : begin
                 rd_nxt_state <= QUEUE2_s;
              end
           QUEUE2_s : begin
                 rd_nxt_state <= QUEUE3_s;
              end
           QUEUE3_s : begin
                 rd_nxt_state <= QUEUE_ACK_s;
              end
           QUEUE_ACK_s : begin
                 rd_nxt_state <= WAIT_ACK_s;
              end
           WAIT_ACK_s : begin
              // the output pipe line is fully loaded, so wait for ack from mac
              // before moving on
              if (tx_ack == 1'b1)
                 rd_nxt_state <= FRAME_s;
              else
                 rd_nxt_state <= WAIT_ACK_s;
              end
           FRAME_s : begin
              // when the end of frame has been reached wait another frame in
              // the fifo
              if (rd_eof == 1'b1)
                 rd_nxt_state <= IDLE_s;
              else
                 rd_nxt_state <= FRAME_s;
              end
           default : begin
                 rd_nxt_state <= IDLE_s;
              end
        endcase
  end
                                // full duplex state machine
 
end // gen_fd_sm
endgenerate
 
 
  //---------------------------------------------------------------------------
  // Full and Half Duplex State Machine
generate if (FULL_DUPLEX_ONLY != 1) begin : gen_hd_sm
  // decode the next state
  // should never receive a rd_drop_frame pulse outside of the Frame state
  always @(rd_state or frame_in_fifo or rd_eof_reg or tx_ack or rd_drop_frame or rd_retransmit)
  begin
  case (rd_state)
           IDLE_s : begin
              // if a retransmit request is detected go to retransmit state
              if (rd_retransmit == 1'b1)
                 rd_nxt_state <= RETRANSMIT_s;
              // if there is a frame in the fifo then queue the new frame to
              // the output
              else if (frame_in_fifo == 1'b1)
                 rd_nxt_state <= QUEUE1_s;
              else
                 rd_nxt_state <= IDLE_s;
              end
           QUEUE1_s : begin
              if (rd_retransmit == 1'b1)
                 rd_nxt_state <= RETRANSMIT_s;
              else
                rd_nxt_state <= QUEUE2_s;
              end
           QUEUE2_s : begin
              if (rd_retransmit == 1'b1)
                 rd_nxt_state <= RETRANSMIT_s;
              else
                 rd_nxt_state <= QUEUE3_s;
              end
           QUEUE3_s : begin
              if (rd_retransmit == 1'b1)
                 rd_nxt_state <= RETRANSMIT_s;
              else
                 rd_nxt_state <= QUEUE_ACK_s;
              end
           QUEUE_ACK_s : begin
              if (rd_retransmit == 1'b1)
                 rd_nxt_state <= RETRANSMIT_s;
              else
                 rd_nxt_state <= WAIT_ACK_s;
              end
           WAIT_ACK_s : begin
              // the output pipeline is now fully loaded so wait for ack from
              // mac before moving on.
              if (rd_retransmit == 1'b1)
                 rd_nxt_state <= RETRANSMIT_s;
              else if (tx_ack == 1'b1)
                 rd_nxt_state <= FRAME_s;
              else
                 rd_nxt_state <= WAIT_ACK_s;
              end
           FRAME_s : begin
              // if a collision only request, then must drop the rest of the
              // current frame, move to drop state
              if (rd_drop_frame == 1'b1)
                 rd_nxt_state <= DROP_s;
              else if (rd_retransmit == 1'b1)
                 rd_nxt_state <= RETRANSMIT_s;
              // continue transmitting frame until the end of the frame is
              // detected, then wait for a new frame to be sent.
              else if (rd_eof_reg == 1'b1)
                 rd_nxt_state <= IDLE_s;
              else
                 rd_nxt_state <= FRAME_s;
              end
           DROP_s : begin
              // wait until rest of frame has been cleared.
              if (rd_eof_reg == 1'b1)
                 rd_nxt_state <= IDLE_s;
              else
                 rd_nxt_state <= DROP_s;
              end
           RETRANSMIT_s : begin
              // reload the data pipe from the start of the frame
                 rd_nxt_state <= QUEUE1_s;
              end
           default : begin
                 rd_nxt_state <= IDLE_s;
              end
        endcase
  end
 
end // gen_hd_sm                               // half duplex state machine
endgenerate
 
  //---------------------------------------------------------------------------
  // decode output signals
  // decode output data
  always @(posedge rd_clk)
  begin
     if (rd_enable == 1'b1)
        begin
        if (rd_nxt_state == FRAME_s)
           tx_data <= rd_data_pipe;
        else
           begin
           case (rd_state)
              QUEUE_ACK_s :
                 tx_data <= rd_data_pipe;
              WAIT_ACK_s :
                 tx_data <= tx_data;
              FRAME_s :
                 tx_data <= rd_data_pipe;
              default :
                 tx_data <= 8'b0;
           endcase
           end
        end
  end
 
  // decode output data valid
  always @(posedge rd_clk)
  begin
     if (rd_enable == 1'b1)
        begin
        if (rd_nxt_state == FRAME_s)
           tx_data_valid <= ~(tx_collision && ~(tx_retransmit));
        else
           begin
           case (rd_state)
              QUEUE_ACK_s :
                 tx_data_valid <= 1'b1;
              WAIT_ACK_s :
                 tx_data_valid <= 1'b1;
              FRAME_s :
                 tx_data_valid <= ~(rd_nxt_state == DROP_s);
              default :
                 tx_data_valid <= 1'b0;
           endcase
           end
        end
  end
 
  //---------------------------------------------------------------------------
  // decode full duplex only control signals
generate if (FULL_DUPLEX_ONLY == 1) begin : gen_fd_decode
 
  assign rd_en = (rd_state == IDLE_s) ? 1'b0 :
                 (rd_nxt_state == FRAME_s) ? 1'b1 :
                 (rd_state == WAIT_ACK_s) ? 1'b0 : 1'b1;
 
  assign rd_addr_inc = rd_en;
 
  assign rd_addr_reload = (rd_state == FRAME_s && rd_nxt_state == IDLE_s) ? 1'b1 : 1'b0;
 
  // Transmit frame pulse is only 1 clock enabled pulse long.
  // Transmit frame pulse must never be more frequent than 64 clocks to allow toggle to cross clock domain
  assign rd_transmit_frame = (rd_state == WAIT_ACK_s && rd_nxt_state == FRAME_s) ? 1'b1 : 1'b0;
 
  // unused for full duplex only
  assign rd_start_addr_reload = 1'b0;
  assign rd_start_addr_load   = 1'b0;
  assign rd_retransmit_frame  = 1'b0;
 
end // gen_fd_decode                              // full duplex control signals
endgenerate
 
  //---------------------------------------------------------------------------
  // decode half duplex control signals
generate if (FULL_DUPLEX_ONLY != 1) begin : gen_hd_decode
 
  assign rd_en = (rd_state == IDLE_s) ? 1'b0 :
                 (rd_nxt_state == DROP_s && rd_eof == 1'b1) ? 1'b0 :
                 (rd_nxt_state == FRAME_s) ? 1'b1 :
                 (rd_state == RETRANSMIT_s) ? 1'b0 :
                 (rd_state == WAIT_ACK_s) ? 1'b0 : 1'b1;
 
  assign rd_addr_inc = rd_en;
 
  assign rd_addr_reload = (rd_state == FRAME_s && rd_nxt_state == IDLE_s) ? 1'b1 :
                          (rd_state == DROP_s && rd_nxt_state == IDLE_s) ? 1'b1 : 1'b0;
 
  assign rd_start_addr_reload = (rd_state == RETRANSMIT_s) ? 1'b1 : 1'b0;
 
  assign rd_start_addr_load = (rd_state == WAIT_ACK_s && rd_nxt_state == FRAME_s) ? 1'b1 :
                              (rd_col_window_expire == 1'b1) ? 1'b1 : 1'b0;
 
  // Transmit frame pulse must never be more frequent than 64 clocks to allow toggle to cross clock domain
  assign rd_transmit_frame = (rd_state == WAIT_ACK_s && rd_nxt_state == FRAME_s) ? 1'b1 : 1'b0;
 
  // Retransmit frame pulse must never be more frequent than 16 clocks to allow toggle to cross clock domain
  assign rd_retransmit_frame = (rd_state == RETRANSMIT_s) ? 1'b1 : 1'b0;
 
end // gen_hd_decode                           // half duplex control signals
endgenerate
 
  //---------------------------------------------------------------------------
  // Frame Count
  // We need to maintain a count of frames in the fifo, so that we know when a
  // frame is available for transmission.  The counter must be held on the
  // write clock domain as this is the faster clock.
  //---------------------------------------------------------------------------
 
  // A frame has been written to the fifo
  assign wr_store_frame = (wr_state == EOF_s && wr_nxt_state != EOF_s) ? 1'b1 : 1'b0;
 
  // generate a toggle to indicate when a frame has been transmitted from the fifo
  always @(posedge rd_clk)
  begin  // process
     if (rd_sreset == 1'b1)
         rd_tran_frame_tog <= 1'b0;
     else if (rd_enable == 1'b1)
        if (rd_transmit_frame == 1'b1)     // assumes EOF_s is valid for one clock
                                        // cycle only ever!  check
              rd_tran_frame_tog <= !rd_tran_frame_tog;
  end
 
  // move the read transmit frame signal onto the write clock domain
  always @(posedge wr_clk)
  begin
      if (wr_sreset == 1'b1)
         begin
            wr_tran_frame_tog  <= 1'b0;
            wr_tran_frame_sync <= 1'b0;
            wr_tran_frame_delay <= 1'b0;
            wr_transmit_frame   <= 1'b0;
         end
      else
        begin
           wr_tran_frame_tog  <= rd_tran_frame_tog;
           wr_tran_frame_sync <= wr_tran_frame_tog;
           wr_tran_frame_delay <= wr_tran_frame_sync;
           // edge detector
           if ((wr_tran_frame_delay ^ wr_tran_frame_sync) == 1'b1)
             wr_transmit_frame    <= 1'b1;
           else
             wr_transmit_frame    <= 1'b0;
        end
  end
 
  //---------------------------------------------------------------------------  
generate if (FULL_DUPLEX_ONLY == 1) begin : gen_fd_count
 
  // count the number of frames in the fifo.  the counter is incremented when a
  // frame is stored and decremented when a frame is transmitted.  Need to keep
  // the counter on the write clock as this is the fastest clock.
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        wr_frames <= 9'b0;
     else
        if ((wr_store_frame & !wr_transmit_frame) == 1'b1)
           wr_frames <= wr_frames + 1;
        else if ((!wr_store_frame & wr_transmit_frame) == 1'b1)
           wr_frames <= wr_frames - 1;
  end
 
end // gen_fd_count
endgenerate
 
  //---------------------------------------------------------------------------
generate if (FULL_DUPLEX_ONLY != 1) begin : gen_hd_count
 
  // generate a toggle to indicate when a frame has been transmitted from the fifo
  always @(posedge rd_clk)
  begin  // process
     if (rd_sreset == 1'b1)
        rd_retran_frame_tog <= 1'b0;
     else if (rd_enable == 1'b1)
        if (rd_retransmit_frame == 1'b1)     // assumes EOF_s is valid for one clock
                                   // cycle only ever!  check
           rd_retran_frame_tog <= !rd_retran_frame_tog;
  end
 
  // move the read transmit frame signal onto the write clock domain
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        begin
           wr_retran_frame_tog  <= 1'b0;
           wr_retran_frame_sync <= 1'b0;
           wr_retran_frame_delay <= 1'b0;
           wr_retransmit_frame  <= 1'b0;
        end
     else
        begin
           wr_retran_frame_tog  <= rd_retran_frame_tog;
           wr_retran_frame_sync <= wr_retran_frame_tog;
           wr_retran_frame_delay <= wr_retran_frame_sync;
           // edge detector
           if ((wr_retran_frame_delay ^ wr_retran_frame_sync) == 1'b1)
              wr_retransmit_frame    <= 1'b1;
           else
              wr_retransmit_frame    <= 1'b0;
        end
  end
 
  // count the number of frames in the fifo.  the counter is incremented when a
  // frame is stored or retransmitted and decremented when a frame is transmitted.  Need to keep
  // the counter on the write clock as this is the fastest clock.
  // Assumes transmit and retransmit cannot happen at same time
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        wr_frames <= 9'b0;
     else
        if ((wr_store_frame & wr_retransmit_frame) == 1'b1)
           wr_frames <= wr_frames + 2;
        else if (((wr_store_frame | wr_retransmit_frame) & !wr_transmit_frame) == 1'b1)
           wr_frames <= wr_frames + 1;
        else if (wr_transmit_frame == 1'b1 & !wr_store_frame)
           wr_frames <= wr_frames - 1;
  end
 
end // gen_hd_count
endgenerate
 
 
  //---------------------------------------------------------------------------
  // generate a frame in fifo signal for use in control logic
  always @(posedge wr_clk)
  begin
      if (wr_sreset == 1'b1)
         wr_frame_in_fifo <= 1'b0;
      else
         if (wr_frames != 9'b0)
            wr_frame_in_fifo <= 1'b1;
         else
            wr_frame_in_fifo <= 1'b0;
  end
 
  // register back onto read domain for use in the read logic
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
        begin
           frame_in_fifo_sync <= 1'b0;
           frame_in_fifo <= 1'b0;
        end
     else if (rd_enable == 1'b1)
        begin
           frame_in_fifo_sync <= wr_frame_in_fifo;
           frame_in_fifo <= frame_in_fifo_sync;
        end
  end
 
 
  //---------------------------------------------------------------------------
  // Address counters
  //---------------------------------------------------------------------------
  // Address counters
  // write address is incremented when write enable signal has been asserted
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        wr_addr <= 12'b0;
     else if (wr_addr_reload == 1'b1)
        wr_addr <= wr_start_addr;
     else if (wr_addr_inc == 1'b1)
        wr_addr <= wr_addr + 1;
  end
 
  // store the start address incase the address must be reset
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        wr_start_addr <= 12'b0;
     else if (wr_start_addr_load == 1'b1)
        wr_start_addr <= wr_addr + 1;
  end
 
  //---------------------------------------------------------------------------
generate if (FULL_DUPLEX_ONLY == 1) begin : gen_fd_addr
  // read address is incremented when read enable signal has been asserted
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
        rd_addr <= 12'b0;
     else if (rd_enable == 1'b1)
        if (rd_addr_reload == 1'b1)
           rd_addr <= rd_dec_addr;
        else if (rd_addr_inc == 1'b1)
           rd_addr <= rd_addr + 1;
  end
 
  // do not need to keep a start address, but the address is needed to
  // calculate fifo occupancy.
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
        rd_start_addr <= 12'b0;
     else if (rd_enable == 1'b1)
        rd_start_addr <= rd_addr;
  end
 
 
 
end // gen_fd_addr                           // full duplex address counters
endgenerate
 
  //---------------------------------------------------------------------------
generate if (FULL_DUPLEX_ONLY != 1) begin : gen_hd_addr
  // read address is incremented when read enable signal has been asserted
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
        rd_addr <= 12'b0;
     else if (rd_enable == 1'b1)
        if (rd_addr_reload == 1'b1)
           rd_addr <= rd_dec_addr;
        else if (rd_start_addr_reload == 1'b1)
           rd_addr <= rd_start_addr;
        else if (rd_addr_inc == 1'b1)
           rd_addr <= rd_addr + 1;
  end
 
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
        rd_start_addr <= 12'b0;
     else if (rd_enable == 1'b1)
        if (rd_start_addr_load == 1'b1)
           rd_start_addr <= rd_addr - 4;
  end
 
  // Collision window expires after MAC has been transmitting for required slot
  // time.  This is 512 clock cycles at 1G.  Also if the end of frame has fully
  // been transmitted by the mac then a collision cannot occur.
  // this collision expire signal goes high at 768 cycles from the start of the
  // frame.
  // inefficient for short frames, however should be enough to prevent fifo
  // locking up.
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
        rd_col_window_expire <= 1'b0;
     else if (rd_enable == 1'b1)
        if (rd_transmit_frame == 1'b1)
           rd_col_window_expire <= 1'b0;
        else if (rd_slot_timer[9:8] == 2'b11)
           rd_col_window_expire <= 1'b1;
  end
 
  assign rd_idle_state = (rd_state == IDLE_s) ? 1'b1 : 1'b0;
 
  always @(posedge rd_clk)
  begin
     if (rd_enable == 1'b1)
        begin
           rd_col_window_pipe[0] <= rd_col_window_expire & rd_idle_state;
           if (rd_txfer_en == 1'b1)
              rd_col_window_pipe[1] <= rd_col_window_pipe[0];
        end
  end
 
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)         // will not count until after first
                                    // frame is sent.
        rd_slot_timer <= 10'b0;
     else if (rd_enable == 1'b1)
        if (rd_transmit_frame == 1'b1)  // reset counter
           rd_slot_timer <= 10'b0;
        // do not allow counter to role over.
        // only count when frame is being transmitted.
        else if (rd_slot_timer != 10'b1111111111)
           rd_slot_timer <= rd_slot_timer + 1;
  end
 
 
end // gen_hd_addr                           // half duplex address counters
endgenerate
 
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
           rd_dec_addr <= 12'b0;
     else if (rd_enable == 1'b1)
        if (rd_addr_inc == 1'b1)
           rd_dec_addr <= rd_addr - 1;
  end
 
  //---------------------------------------------------------------------------
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
        begin
           rd_bram_u <= 1'b0;
           rd_bram_u_reg <= 1'b0;
        end
     else if (rd_enable == 1'b1)
        if (rd_addr_inc == 1'b1)
           begin
              rd_bram_u <= rd_addr[11];
              rd_bram_u_reg <= rd_bram_u;
           end
  end
 
  //---------------------------------------------------------------------------
  // Data Pipelines
  //---------------------------------------------------------------------------
  // register input signals to fifo
  // no reset to allow srl16 target
  always @(posedge wr_clk)
  begin
     wr_data_pipe[0] <= wr_data;
     if (wr_accept_pipe[0] == 1'b1)
        wr_data_pipe[1] <= wr_data_pipe[0];
     if (wr_accept_pipe[1] == 1'b1)
        wr_data_bram    <= wr_data_pipe[1];
  end
 
  // no reset to allow srl16 target
  always @(posedge wr_clk)
  begin
     wr_sof_pipe[0] <= !wr_sof_n;
     if (wr_accept_pipe[0] == 1'b1)
        wr_sof_pipe[1] <= wr_sof_pipe[0];
  end
 
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        begin
           wr_accept_pipe[0] <= 1'b0;
           wr_accept_pipe[1] <= 1'b0;
           wr_accept_bram    <= 1'b0;
        end
     else
        begin
           wr_accept_pipe[0] <= !wr_src_rdy_n & !wr_dst_rdy_int_n;
           wr_accept_pipe[1] <= wr_accept_pipe[0];
           wr_accept_bram    <= wr_accept_pipe[1];
        end
  end
 
  always @(posedge wr_clk)
  begin
     wr_eof_pipe[0] <= !wr_eof_n;
     if (wr_accept_pipe[0] == 1'b1)
        wr_eof_pipe[1] <= wr_eof_pipe[0];
     if (wr_accept_pipe[1] == 1'b1)
        wr_eof_bram[0] <= wr_eof_pipe[1];
  end
 
  // register data outputs from bram
  // no reset to allow srl16 target
  always @(posedge rd_clk)
  begin
     if (rd_enable == 1'b1)
        if (rd_en == 1'b1)
           begin
              rd_data_pipe_u <= rd_data_bram_u;
              rd_data_pipe_l <= rd_data_bram_l;
              if (rd_bram_u_reg == 1'b1)
                 rd_data_pipe <= rd_data_pipe_u;
              else
                 rd_data_pipe <= rd_data_pipe_l;
           end
  end
 
   // register data outputs from bram
  // no reset to allow srl16 target
  always @(posedge rd_clk)
  begin
     if (rd_enable == 1'b1)
        if (rd_en == 1'b1)
           begin
              if (rd_bram_u == 1'b1)
                 rd_eof_pipe <= rd_eof_bram_u[0];
              else
                 rd_eof_pipe <= rd_eof_bram_l[0];
              rd_eof <= rd_eof_pipe;
              rd_eof_reg <= rd_eof | rd_eof_pipe;
           end
  end
 
  //---------------------------------------------------------------------------
generate if (FULL_DUPLEX_ONLY != 1) begin : gen_hd_input
  // register the collision and retransmit signals
  always @(posedge rd_clk)
  begin
     if (rd_enable == 1'b1)
        rd_drop_frame <= tx_collision & !tx_retransmit;
  end
 
  always @(posedge rd_clk)
  begin
     if (rd_enable == 1'b1)
        rd_retransmit <= tx_collision & tx_retransmit;
  end
 
end // gen_hd_input                        // half duplex register input
endgenerate
 
  //---------------------------------------------------------------------------
  // Fifo full functionality
  //---------------------------------------------------------------------------
  // when full duplex full functionality is difference between read and write addresses.
  // when in half duplex is difference between read start and write addresses.
  // Cannot use gray code this time as the read address and read start addresses jump by more than 1
 
  // generate an enable pulse for the read side every 16 read clocks.  This provides for the worst case
  // situation where wr clk is 20Mhz and rd clk is 125 Mhz.
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
        rd_16_count <= 4'b0;
     else if (rd_enable == 1'b1)
        rd_16_count <= rd_16_count + 1;
  end
 
  assign rd_txfer_en = (rd_16_count == 4'b1111) ? 1'b1 : 1'b0;
 
  // register the start address on the enable pulse
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
        rd_addr_txfer <= 12'b0;
     else if (rd_enable == 1'b1)
        begin
        if (rd_txfer_en == 1'b1)
           rd_addr_txfer <= rd_start_addr;
        end
  end
 
  // generate a toggle to indicate that the address has been loaded.
  always @(posedge rd_clk)
  begin
     if (rd_sreset == 1'b1)
        rd_txfer_tog <= 1'b0;
     else if (rd_enable == 1'b1)
        begin
        if (rd_txfer_en == 1'b1)
           rd_txfer_tog <= !rd_txfer_tog;
        end
  end
 
  // pass the toggle to the write side
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        begin
           wr_txfer_tog <= 1'b0;
           wr_txfer_tog_sync <= 1'b0;
           wr_txfer_tog_delay <= 1'b0;
        end
     else
        begin
           wr_txfer_tog <= rd_txfer_tog;
           wr_txfer_tog_sync <= wr_txfer_tog;
           wr_txfer_tog_delay <= wr_txfer_tog_sync;
        end
  end
 
  // generate an enable pulse from the toggle, the address should have been steady on the wr clock input for at least one clock
  assign wr_txfer_en = wr_txfer_tog_delay ^ wr_txfer_tog_sync;
 
  // capture the address on the write clock when the enable pulse is high.
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        wr_rd_addr <= 12'b0;
     else if (wr_txfer_en == 1'b1)
        wr_rd_addr <= rd_addr_txfer;
  end
 
 
  // Obtain the difference between write and read pointers
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        wr_addr_diff <= 12'b0;
     else
        wr_addr_diff <= wr_rd_addr - wr_addr;
  end
 
 
  // Detect when the FIFO is full
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        wr_fifo_full <= 1'b0;
     else
        // The FIFO is considered to be full if the write address
        // pointer is within 1 to 3 of the read address pointer.
        if (wr_addr_diff[11:4] == 8'b0 && wr_addr_diff[3:2] != 2'b0)
           wr_fifo_full <= 1'b1;
        else
           wr_fifo_full <= 1'b0;
  end
 
  // memory overflow occurs when the fifo is full and there are no frames
  // available in the fifo for transmission.  If the collision window has
  // expired and there are no frames in the fifo and the fifo is full, then the
  // fifo is in an overflow state.  we must accept the rest of the incoming
  // frame in overflow condition.
 
generate if (FULL_DUPLEX_ONLY == 1) begin : gen_fd_ovflow
     // in full duplex mode, the fifo memory can only overflow if the fifo goes
     // full but there is no frame available to be retranmsitted
     // do not allow to go high when the frame count is being updated, ie wr_store_frame is asserted.
     assign wr_fifo_overflow = (wr_fifo_full == 1'b1 && wr_frame_in_fifo == 1'b0
                                   && wr_eof_state == 1'b0 && wr_eof_state_reg == 1'b0) ? 1'b1 : 1'b0;
end // gen_fd_ovflow
endgenerate
 
generate if (FULL_DUPLEX_ONLY != 1) begin : gen_hd_ovflow
    // register wr col window to give address counter sufficient time to update.
     // do not allow to go high when the frame count is being updated, ie wr_store_frame is asserted.
    assign wr_fifo_overflow = (wr_fifo_full == 1'b1 && wr_frame_in_fifo == 1'b0
                                  && wr_eof_state == 1'b0 && wr_eof_state_reg == 1'b0 && wr_col_window_expire == 1'b1) ? 1'b1 : 1'b0;
 
    // register rd_col_window signal
    // this signal is long, and will remain high until overflow functionality
    // has finished, so save just to register the once.
    always @(posedge wr_clk)
    begin  // process
       if (wr_sreset == 1'b1)
          begin
             wr_col_window_pipe[0] <= 1'b0;
             wr_col_window_pipe[1] <= 1'b0;
             wr_col_window_expire  <= 1'b0;
          end
       else
          begin
             if (wr_txfer_en == 1'b1)
                wr_col_window_pipe[0] <= rd_col_window_pipe[1];
             wr_col_window_pipe[1] <= wr_col_window_pipe[0];
             wr_col_window_expire <= wr_col_window_pipe[1];
          end
    end
 
end // gen_hd_ovflow
endgenerate
 
 
 
  //--------------------------------------------------------------------
  // Create FIFO Status Signals in the Write Domain
  //--------------------------------------------------------------------
 
  // The FIFO status signal is four bits which represents the occupancy
  // of the FIFO in 16'ths.  To generate this signal we therefore only
  // need to compare the 4 most significant bits of the write address
  // pointer with the 4 most significant bits of the read address 
  // pointer.
 
  // The 4 most significant bits of the write pointer minus the 4 msb of
  // the read pointer gives us our FIFO status.
  always @(posedge wr_clk)
  begin
     if (wr_sreset == 1'b1)
        wr_fifo_status <= 4'b0;
     else
        if (wr_addr_diff == 12'b0)
           wr_fifo_status <= 4'b0;
        else
           begin
              wr_fifo_status[3] <= !wr_addr_diff[11];
              wr_fifo_status[2] <= !wr_addr_diff[10];
              wr_fifo_status[1] <= !wr_addr_diff[9];
              wr_fifo_status[0] <= !wr_addr_diff[8];
           end
  end
 
  //---------------------------------------------------------------------------
  // Memory
  //---------------------------------------------------------------------------
  assign rd_en_bram = rd_en & rd_enable_delay2;
 
  // Block Ram for lower address space (rx_addr(11) = 1'b0)
  defparam ramgen_l.WRITE_MODE_A = "READ_FIRST";
  defparam ramgen_l.WRITE_MODE_B = "READ_FIRST";
  RAMB16_S9_S9 ramgen_l (
      .WEA  (wr_en_l),
      .ENA  (VCC),
      .SSRA (wr_sreset),
      .CLKA (wr_clk),
      .ADDRA(wr_addr[10:0]),
      .DIA  (wr_data_bram),
      .DIPA (wr_eof_bram),
      .DOA  (),
      .DOPA (),
      .WEB  (GND),
      .ENB  (rd_en_bram),
      .SSRB (rd_sreset),
      .CLKB (rd_clk),
      .ADDRB(rd_addr[10:0]),
      .DIB  (GND_BUS[7:0]),
      .DIPB (GND_BUS[0:0]),
      .DOB  (rd_data_bram_l),
      .DOPB (rd_eof_bram_l));
 
    // Block Ram for lower address space (rx_addr(11) = 1'b0)
  defparam ramgen_u.WRITE_MODE_A = "READ_FIRST";
  defparam ramgen_u.WRITE_MODE_B = "READ_FIRST";
  RAMB16_S9_S9 ramgen_u (
      .WEA  (wr_en_u),
      .ENA  (VCC),
      .SSRA (wr_sreset),
      .CLKA (wr_clk),
      .ADDRA(wr_addr[10:0]),
      .DIA  (wr_data_bram),
      .DIPA (wr_eof_bram),
      .DOA  (),
      .DOPA (),
      .WEB  (GND),
      .ENB  (rd_en_bram),
      .SSRB (rd_sreset),
      .CLKB (rd_clk),
      .ADDRB(rd_addr[10:0]),
      .DIB  (GND_BUS[7:0]),
      .DIPB (GND_BUS[0:0]),
      .DOB  (rd_data_bram_u),
      .DOPB (rd_eof_bram_u));
 
 
 
endmodule
Browse

Tools

Subversion Repositories fpga-cf

[/] [fpga-cf/] [trunk/] [hdl/] [boardsupport/] [v4/] [tx_client_fifo_8.v] - Blame information for rev 2

Line No.	Rev	Author	Line
1	2	peteralieb	`//-----------------------------------------------------------------------------`
2			`// Title : 8-bit Client to Local-link Transmitter FIFO`
3			`// Project : Virtex-4 Embedded Tri-Mode Ethernet MAC Wrapper`
4			`// File : tx_client_fifo_8.v`
5			`// Version : 4.8`
6			`//-----------------------------------------------------------------------------`
7			`//`
8			`// (c) Copyright 2004-2010 Xilinx, Inc. All rights reserved.`
9			`//`
10			`// This file contains confidential and proprietary information`
11			`// of Xilinx, Inc. and is protected under U.S. and`
12			`// international copyright and other intellectual property`
13			`// laws.`
14			`//`
15			`// DISCLAIMER`
16			`// This disclaimer is not a license and does not grant any`
17			`// rights to the materials distributed herewith. Except as`
18			`// otherwise provided in a valid license issued to you by`
19			`// Xilinx, and to the maximum extent permitted by applicable`
20			`// law: (1) THESE MATERIALS ARE MADE AVAILABLE "AS IS" AND`
21			`// WITH ALL FAULTS, AND XILINX HEREBY DISCLAIMS ALL WARRANTIES`
22			`// AND CONDITIONS, EXPRESS, IMPLIED, OR STATUTORY, INCLUDING`
23			`// BUT NOT LIMITED TO WARRANTIES OF MERCHANTABILITY, NON-`
24			`// INFRINGEMENT, OR FITNESS FOR ANY PARTICULAR PURPOSE; and`
25			`// (2) Xilinx shall not be liable (whether in contract or tort,`
26			`// including negligence, or under any other theory of`
27			`// liability) for any loss or damage of any kind or nature`
28			`// related to, arising under or in connection with these`
29			`// materials, including for any direct, or any indirect,`
30			`// special, incidental, or consequential loss or damage`
31			`// (including loss of data, profits, goodwill, or any type of`
32			`// loss or damage suffered as a result of any action brought`
33			`// by a third party) even if such damage or loss was`
34			`// reasonably foreseeable or Xilinx had been advised of the`
35			`// possibility of the same.`
36			`//`
37			`// CRITICAL APPLICATIONS`
38			`// Xilinx products are not designed or intended to be fail-`
39			`// safe, or for use in any application requiring fail-safe`
40			`// performance, such as life-support or safety devices or`