VerilogCodingStyle

Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 1 of 17 Verilog Coding Guidelines This document describes coding styles and guidelines for writing Verilog code for ASIC blocks and test benches. Reviewers Reviewer Name and Title ASIC Manager Bob Parker, Mgr, Hardware Engineering Modification History Rev Date Originator Comments A 10/30/00 Jane Smith Added test bench information on file name, reporting results B 11/2/00 Jane Smith Added document number and sign-off page Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 2 of 17 Table of Contents 1 Overview........................................................................................................................................... 3 2 Alignment/Space .............................................................................................................................. 4 2.1 Tabs ............................................................................................................................................ 4 2.2 Text file width ............................................................................................................................. 4 2.3 White space around operators .................................................................................................... 4 2.4 Nested indentation levels ............................................................................................................. 4 2.5 Alignment ................................................................................................................................... 6 3 Naming conventions ........................................................................................................................ 8 3.1 Modules/Variables ...................................................................................................................... 8 3.2 Special Case Variables................................................................................................................ 8 3.2.1 Clocks ................................................................................................................................................................................. 8 3.2.2 Flip-Flops & Latches ....................................................................................................................................................... 8 3.2.3 Using Signal Source in Name......................................................................................................................................... 9 3.2.4 Active Low signals ........................................................................................................................................................... 9 3.2.5 Active level globally defined........................................................................................................................................ 10 3.2.6 Resets ............................................................................................................................................................................... 10 3.2.7 Summary of prefixes & suffixes for naming signals ................................................................................................ 10 3.3 Signal widths ............................................................................................................................. 11 4 Module Coding............................................................................................................................... 12 4.1 Module declaration................................................................................................................... 12 4.2 Module instantiations ............................................................................................................... 13 4.3 Sequential Logic ....................................................................................................................... 13 4.4 Combinatorial Logic ................................................................................................................. 14 4.5 Defines ...................................................................................................................................... 14 4.6 Assignment statements .............................................................................................................. 14 5 Comments....................................................................................................................................... 15 5.1 End of line comments ................................................................................................................ 15 6 Test Benches .................................................................................................................................. 17 6.1 File Name .................................................................................................................................. 17 6.2 Self-Checking Test Bench Results ............................................................................................. 17 Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 3 of 17 1 Overview In the course of projects, we are all constantly reviewing, maintaining, updating, or inheriting each other’s code. To make these painful tasks a bit easier, the following coding guidelines have been identified. These conventions also allow us to create tools to automate mundane parts of the coding process such as exploding module declarations or creating hierarchical modules. One of the most important issues in reading code is style. Given this, the following style guidelines should be adhered to religiously. There is a template for modules: $AIRO/asic/verilog/templates/module.v. The header by itself is in: $AIRO/asic/verilog/templates/header.v. All of the parts are labeled. Code should be entered into the appropriate section. Comment headers are to be left in the code, even if there is no code entered in that section. It makes it easier for someone reading it to know there are no assign statements (as an example) if the header exists but has no code. Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 4 of 17 2 Alignment/Space 2.1 Tabs All alignment related white space should be 4 position tab characters. All popular UNIX text editors can be made to use four position tabs instead of eight position tabs. The following UNIX command can be used to create an alias that will print Verilog using enscript. alias printcode ‘/usr/5bin/pr -t -e4 \!$ | enscript -b\!$’ 2.2 Text file width Files must be restricted to 160 columns wide. These limits aid in creating readable printouts. The template is set up for 120 character width. 2.3 White space around operators White space between operators can make code more readable. The exact spacing to leave is left as individual preference. One example to improve readability is shown below. Incorrect Example: if((my_signal1==1’b0)&&(my_bus[3:0]==4’d5)) begin Correct Example: if ((my_signal == 1’b0) && (my_bus[3:0] == 4’d5)) begin 2.4 Nested indentation levels Indentation levels are to be used to show code nesting. Case structures should use multiple indentation levels to line up action statements. Blank lines may be used as desired to improve code readability.. The Verilog keyword begin should appear on a line by itself so it can line up with its end statement. The begin/end block should always be used, even if there is only one statement. This makes adding lines of code much easier with fewer errors. Incorrect Example: if ( this ) begin for ( i == 0; i < 10; i = i + 1 ) begin statement1; statement2; end statement3; statement4; end else statement5; Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 5 of 17 Correct Example: if ( this ) begin for ( i == 0; i < 10; i = i + 1 ) begin statement1; statement2; end statement3; statement4; end else begin statement5; end Case statements are a little more complex. The begin/end structure should always be used in a case definition, indentation levels should be used to offset the statements that are encapsulated, but the use of blank lines can be used or omitted to best show the statement groupings. Incorrect Example: case ( myBus[3:0] ) 4’b0000 : my_signal1 = TRUE; 4’b0001 : my_signal1 = FALSE; 4’b0010 : begin my_signal1 = TRUE; my_signal2 = FALSE; end 4’b0100 : my_signal2 = FALSE; default : my_signal1 = TRUE; endcase Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 6 of 17 Correct Example: case ( myBus[3:0] ) 4’b0000 : begin my_signal1 = TRUE; end 4’b0001 : begin my_signal1 = FALSE; end 4’b0010 : begin my_signal1 = TRUE; my_signal2 = FALSE; end 4’b0100 : begin my_signal2 = FALSE; end default : begin my_signal1 = TRUE; end endcase 2.5 Alignment Aligning code statements can seem like a menial task, but it significantly adds to the under- standing of code. Alignment should be used in declarations, assignments, multi-line statements, and end of line comments. Incorrect Example: reg[3:0] my_signal1; reg[31:0] myDecodedSignal1; reg[4:0] my_signal2, my_signal3, my_signal4; wire[2:0] mySelect; Correct Example: //---------------------------------------------------------- // Signal Declarations: reg //---------------------------------------------------------- reg [3:0] my_signal1; //description reg [31:0] my_decoded_signal1; //description reg [4:0] my_signal2, //description reg my_signal3, //description reg my_signal4; //description //---------------------------------------------------------- // Signal Declarations: wire //---------------------------------------------------------- wire [2:0] mySelect; //description Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 7 of 17 Incorrect Example: if ( myBoolean ) begin my_signal1 = TRUE; my_delayed_signal1 = !your_signal; end Correct Example: if ( myBoolean ) begin my_signal1 = TRUE; my_delayed_signal1 = !your_signal; end Some complex Boolean expressions make much more sense if they are expressed as multi- line aligned statements. The unaligned example is... Incorrect Example: if ( ( my_signal1 && your_signal1 ) || ( my_signal2 && your_signal2 ) ||( my_signal3 && your_signal3 ) ) begin Correct Example: if ( ( my_signal1 && your_signal1 ) || ( my_signal2 && your_signal2 ) || ( my_signal3 && your_signal3 ) ) begin Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 8 of 17 3 Naming conventions 3.1 Modules/Variables All modules and signal names will be lower case, delimited by underscores “_”. Correct Example: module my_module ( ena_fft, ena_mdi, fft_in, mdi_out, my_signal1, my_signal2 ); All constant names should be upper case with underscore delimiters (MY_TRUE), with underscore delimiters every 4 characters (7’b00X_10XZ). 3.2 Special Case Variables 3.2.1 Clocks Clocks have special naming requirements. These restrictions are placed to help with synthesis and back-end tools. A clock signal has a prefix “clk”, followed by a description, then a suffix.. The suffixes are “_drv”, and “_src”, If a signal starts with “clk”, it must end with either “_drv” or “_src” where “_drv” indicates the signal is the output of a clock fanout tree and drives flip-flop clock inputs ONLY; “_src” indicates the signal is the input to a clock fanout ONLY.” Correct Example: module my_module ( clk_a_drv, clk_mdi_src, clk_mdi_drv, ); Incorrect Example: module FftBlock clk_Fft, clkFft, fftClk_drv, my_Signal_src 3.2.2 Flip-Flops & Latches Flip-flops also have special naming requirements to help synthesis and back-end CAE tools. Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 9 of 17 There are no prefixes, only the four suffixes: “_ff”, “_nxt”, (normal ffs), “_meta”, and “_sync”. (for special synchronizing ffs) The output of a flip-flop is named: name_ff, the value the flip- flop will be loaded with on the next clock is name_nxt. The signal name_ff becomes the Q output of the flip-flop while name_nxt is the D input. It helps readability of the combinatorial block to know that a signal is a flip-flop output or is to be loaded into a flip-flop. The flip-flop instance is named “name_ff_reg” by the synthesis tool. (Add figure to show flip-flops) The special suffixes, “_meta”, and “_sync” are used when two flip-flops are put next to each other to synchronize a signal to a new clock. The first is named reg_meta, and its output can be metastable, the second flip-flop has reg_meta as an input and reg_sync as an output. Latches use “_lat” for the output, and “_nxt” for the input. Correct Example: always @ ( posedge clkFft_drv or `RESET_EDGE reset_x ) if ( reset_x = `RESET_ON ) begin my_signal0_ff <= 1’b0; my_signal1_ff <= 1’b0; my_signal1_dup_ff <= 1’b0; my_signal2_meta <= 1’b0; my_signal2_sync <= 1’b0; end else begin my_signal0_ff <= my_signal0_nxt; my_signal1_ff <= my_signal1_nxt; my_signal1_dup_ff <= my_signal1_nxt; my_signal2_meta <= async_input; my_signal2_sync <= my_signal2_meta; end end 3.2.3 Using Signal Source in Name Understanding and synthesis can be improved by coding the name of the block that sources a signal into its name. An example, described below is the configuration bits in the PHY chips. If the source is known, it is a good idea to use it as a prefix to the name The PHY ASICs have many registers that contain programmable setup information written by the MAC. These bits are static during the operation of the PHY. They are named with a prefix “cfg_ ”. When a bit is named with this prefix, it can be synthesized with no timing constraints to the rest of the PHY circuit. 3.2.4 Active Low signals All active low signals have the suffix “_n”. If a clock is active low, the “_n” is before the “_drv” or “_src”, such as clk_a_n_drv. Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 10 of 17 3.2.5 Active level globally defined In some cases, the active level of a signal is defined globally. This is useful for resets or pad enables whose active level may be different for different libraries. Or there may be a different active level for FPGAs and the ASIC. The suffix “_x” is used, and a compare is always made to a constant that is globally defined for the design. 3.2.6 Resets Resets are named special to allow for libraries with different senses (active high or active low). It is desirable to buffer the reset centrally rather than have it reset in every block. Otherwise, since the reset runs all over the chip, it could require a lot of buffering and therefore there could be a lot of skew on release of the reset from some flip-flops to others. This is controlled by buffering it in much the same way as a clock, in a centralized block. The problem is that different vendors have all of their flips-flops with either active high or active low reset (not both). We have had to use the same netlist for both FPGAs and ASICs and had to use different active levels for each. We handle this by naming the asynchronous reset “reset_x” and always comparing to a globally defined constant, “RESET_ON”. The active sense can then be changed for the whole design. Another define, called “RESET_EDGE” must be declared globally that is used in the sequential always block to indicate posedge or negedge. If an active low reset is used, then a global file for the ASIC will have the following: `define RESET_EDGE negedge `define RESET_ON 1’b0 For an active high reset, the global file will contain: `define RESET_EDGE posedge `define RESET_ON 1’b1 and the flip-flop will be defined as (for either case) always @ (posedge clk_drv or `RESET_EDGE reset_x) begin if (reset_x = `RESET_ON) begin reg1_ff <= 1’b0; reg2_ff <= 1’b0; end else begin reg1_ff <= reg1_nxt; reg2_ff <= reg2_nxt; end end 3.2.7 Summary of prefixes & suffixes for naming signals Prefix clk Clock signal, must either drive a clock fan-out tree or be the driven signal from a fan-out tree that runs only to flip-flop clock inputs. Used with either “_drv” or “_src” suffix Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 11 of 17 cfg Configuration bit, written by the MAC and static during operation of the PHY Suffix _drv Used only with “clk” prefix. Runs directly to flip-flop clock inputs only. _src Used only with “clk” prefix. It runs only to a clock fan-out tree whose output has suffix “_drv”.” _ff Output , Q, of a flip-flop _nxt Input, D, of a flip-flop _n Active low signal. _x Used when sense is library dependent and is defined globally for the design (as in resets or I/O pad enables) 3.3 Signal widths Never assign different width signals to each other. Incorrect Example: reg [3:0] my_ignal, reg [2:0] your_signal; my_signal = your_signal;. Correct Example: reg [3:0] my_signal reg [2:0] your_signal; my_signal[2:0] = your_signal; Along these lines, always completely specify the width of constants. For instance Incorrect Example: reg [3:0] my_signal my_signal= my_signal+ 1’d1; This will work, but a more specific declaration would be Correct Example: reg [3:0] my_signal my_signal = my_signal + 4’d1; Document Number ENG-85857 Revision B Author Jane Smith A printed version of this document is an uncontrolled copy. Cisco Systems, Inc. Page 12 of 17 4 Module Coding Each module has sections for parameters, ports, signal declarations, assign statements, assign statements, instantiations, sequential logic, and combinatorial logic. They are organized as in the template file $AIRO/asic/verilog/templates/module.v. Each module will have separate sections for sequential and combinatorial logic. Sequential and combinatorial logic will not be combined in the same always block. 4.1 Module declaration The module declaration consists of the module name, the IO list, and the input/output declarations. The module name should reside on a line by itself and the opening parenthesis. The IO list should be as inputs in alphabetical order, outputs in alphabetical order, and bidirects in alphabetical order, with a blank line between each group. Only one signal/vector should be listed on each line. Any changes to the module IO list should be integrated in the same fashion. Correct Example: module my_module ( //----------------------- my_input1, my_input2 [2:0], my_input3, my_output1 [31:0], my_output2, my_inOut ); The module input/output declaration is a simple cut and paste of the IO list in the module declaration. These declarations should also be one per line in alphabetical order. Notice the alignment in the following example. Make sure the parameter list is also alphabetical, especially since they can be re-declared in the instantiation by position only. Correct Example: //----------------------- // Parameters //----------------------- parameter CONST1 = 1’b0; parameter CONST2 = 3’b010; //----------------------- // Input Ports //----------------------- input my_input1; input [2:0] my_input2; input my_input3; //-