I'm trying to make a pipeline processor using Verilog HDL. I realized that there are maybe some race conditions somewhere in my code. So I'm going to write a sudo code and would like to ask you about if there is a race condition within and how to avoid it :
module A(input wire reset, input wire clock, output reg a_reg_o);
always #(posedge clock)
begin
if(reset == 1'h1)
begin
a_reg_o = 1'h0;
end
else
begin
a_reg_o = 1'h1;
end
end
endmodule
module B(input wire reset, input wire clock, input a_i);
reg b;
always #(posedge clock)
begin
if(reset == 1'h1)
begin
b = 1'h0;
end
else
begin
if(a_i == 1'h1)
begin
b = 1'h1;
end
else
begin
b = 1'h0;
end
end
end
endmodule
module Main(input wire reset, input wire clock);
wire a_o;
A a(reset, clock, a_o);
B b(reset, clock, a_o)
endmodule
So imagine I trigger reset signal. After the first positive edge of clock the register a_reg_o is going to be 0 and register b from module B also is going to be 0 (No race conditions yet). Now I release the reset button and let it to be negative. On the next positive edge of the clock the register a_reg_o is going to be 1, but what about register b from module B ? Is it going to be :
1. Zero, Because it didn't see the a_i changes yet.
2. It depends on the modules (A and B) total delay ( i.e. a race condition).
Thank you.
Yes there can be a race condition, because you won't know whether the net a_o is first driven by module A and then captured by module B or vice versa.
So you should use the Non Blocking Assignment for this, as that will ensure that no matter which module gets executed, the module B will always have the previous value of net a_o.
You can find more about this Non Blocking Assignments with following link. http://www.sunburst-design.com/papers/CummingsSNUG2000SJ_NBA.pdf
This is why there a non-blocking (NBA) assignments in Verilog. The coding rule is whenever there are multiple processes (in this case, multiple always blocks) accessing the same signal (a_o) synchronized to the same event (#posdege clock) where one process writes and another process reads, you need to use and NBA <= assignment to write to the signal.
Related
I understood the basic difference between blocking and non-blocking statements in Verilog. But still it is not possible for me to understand what's happening & when and where to use blocking and non-blocking statements. For example, consider simple d ff code:
module dff (clk, reset,d, q, qb);
input clk;
input reset;
input d;
output q;
output qb;
reg q;
assign qb = ~q;
always #(posedge clk or posedge reset)
begin
if (reset) begin
// Asynchronous reset when reset goes high
q <= 1'b0;
end else begin
// Assign D to Q on positive clock edge
q <= d;
end
end
endmodule
But if I write the very same logic using two-segment coding technique:
module dff(input wire d,
clk,
reset,
en,
output wire q);
reg q;
reg r_reg, r_next;
always #(posedge clk, posedge reset)
if(reset)
r_reg<=1'b0;
else
r_reg<=r_next;
always #*
if(en)
r_reg=d;
else
r_reg=r_next;
assign q<=r_reg;
endmodule
Now, in this code, I just didn't understand why are using <= in the first always block and why they are using = in 2nd always block. I also know that in combinational logic circuit = is advised to use & in sequential <= this is advised to use. But still, I couldn't be able to find out the answer to the usage of blocking and non-blocking statements. Can you please help me!?
Blocking/non-blocking assignments is a simulation artifact only. Contrary to the believe, verilog does not describe hardware. Verilog describes desired behavior of the hardware trying to fit it into an event-driven simulation scheme.
Here is a simple example of a shift register which employs 2 flops:
always #(posedge clk)
out1 = in;
always #(posedge clk)
out2 = out1;
Now, what would the output of the out2 be? Since we are dealing with simulation, then it depends on the order in which these 2 statements are executed. Either it will be the old value of out1, or the new one (actually the value of in);.
In hardware there is no such mess. It will flop the value which existed at the posedge time, the old value of out1 (well, unless there are unusual delays in clocks).
In order to match this behavior, the non-blocking assignment was introduced. Verilog simulation is done in simulation ticks. Every tick is as long as there are events which could cause re-evaluation of other blocks. The non-blocking assignments are scheduled to be executed at the end of such a tick with current rhs values ( in reality there are several scheduling zones). So, consider the following:
always #(posedge clk)
out1 <= in;
always #(posedge clk)
out2 <= out1;
In the above example the all assignments will happen at the end of the tick. 'out2` will be assigned a value which existed at the time of the <=, so, it will be the old value of out1. Now, it does not matter in which order they are executed.
So, the top-level recommendation is to use blocking assignments (=) for combinational logic and use non-blocking assignments (<=) for all outputs of state devices, flops and latches. Note that some temporary variables inside state devices, which are only used there internally should also be assigned with blocking. Also, never use non-blocking assignments in clock trees.
I'm newbie in ASIC design. I have a design with for example two inputs a ,b. I'm using the following code for initialize these two signals. But the Design compiler generating a warning that the register "a" is a constant and will be removed. When I'm trying to do post-synthesis simulation these two signals are all 'z'. So how can I apply initial signal assignment to avoid such a problem?
always #(posedge(clk) or posedge (rst)) begin
if (rst) begin
a<=4d'5;
b <=4'd10;
end
end
While describing hardware system, you need to consider that input signals to your module comes from another module/system and their values are decided by that signals. Inputs to any module can only be wire type.
You can think of a module as a box that has inputs and outputs. The values of output signals are decided by input signal + logic inside the box. However, the module cannot decide what its inputs should be. It is only possible if there is feedback, and even in that case it would depend on other signals that are outside of the module's control.
As a result, output signals can be declared as output reg but the same is not true for inputs. However there is solution to your problem, I think what you want can be designed using the following method:
module your_module(
input clk,
input rst,
//other inputs and outputs that you might need
input [3:0] a,
input [3:0] b
);
//define registers
reg [3:0] a_register;
reg [3:0] b_register;
/*
These registers are defined to make it possible to
to give any value to that logics when posedge rst
is detected, otherwise you can use them as your
input logics
*/
//use initial block if you need
always#(posedge clk or posedge rst) begin
if(rst) begin
a_register <= 4'd5;
b_register <= 4'd10;
end
else
begin
a_register <= a;
b_register <= b;
// and use a_register and b_register as you want to use a and b
end
end
endmodule
I made a basic example on eda playground of the issue I got.
Let s say I have two clocks 1x and 2x. 2x is divided from 1x using flop divider.
I have two registers a and b. a is clocked on 1x, b is clocked in 2x.
b is sampling value of a.
When we have rising edge of 1x and 2x clocks, b is not taking the expected value of a but it s taking the next cycle value.
This is because of this clock divider scheme, if we make division using icgs and en it works fine.
But is there a way to make it work using this clock divider scheme with flops ?
EDA playground link : https://www.edaplayground.com/x/map#
module race_test;
logic clk1x = 0;
logic clk2x = 0;
always
#5ns clk1x = !clk1x;
int a, b;
always #(posedge clk1x) begin
a <= a+1;
clk2x <= !clk2x;
end
// Problem here is that b will sample postpone value of a
// clk2x is not triggering at the same time than clk1x but a bit later
// This can be workaround by putting blocking assignment for clock divider
always #(posedge clk2x) begin
b <= a;
end
initial begin
$dumpfile("test.vcd");
$dumpvars;
#1us
$stop;
end
endmodule
Digital clock dividers present problems with both simulation and physical timing.
Verilog's non-blocking assignment operator assumes that everyone reading and writing the same variables are synchronized to the same clock event. By using an NBA writing to clk2x, you have shifted the reading of a to another delta time*, and as you discovered, a has already been updated.
In real hardware, there are considerable propagation delays that usually avoid this situation. However, you are using the same D-flop to assign to clk2x, so there will be propagation delays there as well. You last always block now represents a clock domain crossing issue. So depending on the skews between the two clocks, you could still have a race condition.
One way of correcting this is using a clock generator module with an even higher frequency clock
always #2.5ns clk = !clk;
always #(posedge clk) begin
clk1x <= !clk1x;
if (clk1x == 1)
clk2x = !clk2x;
Of course you have solved this problem, but I think there is a better way.
The book says one can use blocking assignment to avoid race. But blocking assignemnt causes errors in synopsys lint check. So, one way to avoid race problem without lint error is to use dummy logic, like this
wire [31:0] a_logic;
wire dummy_sel;
assign dummy_sel = 1'b0;
assign a_logic = dummy_sel ? ~a : a;
always #(posedge clk2x) begin
b <= a_logic;
end
I'm trying to write a roll shift/ ring counter that takes two switches as the clocks in verilog.
My code is as follows:
module roll(CLK1, CLK2, LEDS);
input CLK1;
input CLK2;
output [3:0] LEDS;
reg [3:0] LEDS;
initial
begin
LEDS = 4'b0001;
end
always#(posedge CLK1 or posedge CLK2)
begin
if(CLK1)
begin
LEDS[3]<=LEDS[2];
LEDS[2]<=LEDS[1];
LEDS[1]<=LEDS[0];
LEDS[0]<=LEDS[3];
end
// Roll Right
if(CLK2)
begin
LEDS[3]<=LEDS[0];
LEDS[2]<=LEDS[3];
LEDS[1]<=LEDS[2];
LEDS[0]<=LEDS[1];
end
end
endmodule
I tried using two always blocks, but then figured out that I cannot do that. and when I have the posedge CLK2 in the always statement, the leds on my FPGA all stay on.
Remember Verilog is not a programming language it is a hardware description language.
And when coding for synthesis, you will only be successful if you write code that can be instantiated with actual gates. So writing an always block with sensitivity to edges of two different signals can't be synthesized unless the response to one of the two signals has the effect of a RESET or PRESET operation.
Your code also logically doesn't do what it seems you want to. Consider what your code says will happen if there is a rising edge on CLK2 when CLK1 is already high (or vice versa). Your lights will roll left and then immediately roll right gain, resulting in no change.
A more usual approach would be to have a clock running much faster than the UP and DOWN inputs are expected to change, and use that to drive the logic. For example
module roller(input clk, input rst, input UP, input DOWN, output reg LEDS[3:0]);
reg UP1, DOWN1;
always #(posedge clk or posedge rst)
if (rst) begin
LEDS[3:0] <= 4'b0001;
end
else
begin
UP1 <= UP;
DOWN1 <= DOWN;
if (UP & ~UP1) begin
LEDS[3:0] <= {LEDS[2:0], LEDS[3]};
end
else if (DOWN & ~DOWN1) begin
LEDS[3:0] <= {LEDS[0], LEDS[3:1]};
end
end
endmodule;
Notice that this gives priority to UP. If both UP and DOWN are asserted, the pattern will roll "up" rather than down. If you want a different behavior, you'd have to modify the code to achieve it.
I am trying to write a program in Verilog that should "move" a light LED on an array of LEDs. With a button the light should move to the left, with another one it should move to the right. This is my code:
module led_shift(UP, DOWN, RES, CLK, LED);
input UP, DOWN, RES, CLK;
output reg [7:0] LED;
reg [7:0] STATE;
always#(negedge DOWN or negedge UP or negedge RES)
begin
if(!RES)
begin
STATE <= 8'b00010000;
end
else
begin
STATE <= UP ? STATE>>1 : STATE<<1;
end
end
always # (posedge CLK)
begin
LED <= STATE;
end
endmodule
The problem is in STATE <= UP ? STATE>>1 : STATE<<1; and the error the following:
Error (10200): Verilog HDL Conditional Statement error at led_shift.v(34): cannot match operand(s) in the condition to the corresponding edges in the enclosing event control of the always construct
I tried to modify the code without using that kind of if:
always#(negedge DOWN or negedge UP or negedge RES)
begin
if(!RES)
STATE <= 8'b00010000;
else
begin
if(!DOWN)
STATE <= STATE<<1;
else
begin
if(!UP)
STATE <= STATE>>1;
else
STATE <= STATE;
end
end
end
It compiles, but does not work: the LED "moves" only to the left, when I press the other button all LEDs shut down. Probably there is a problem in my code, but I cannot understand why my first code does not compile at all.
Thank you for any help!
harrym
It is not clear for the synthesizer to know how to control STATE with the 3 asynchronous control signals.
Most likely your your synthesizer is trying to map STATE to a D flip flop with asynchronous active low set and reset. For example it might be trying to synthesize to something like:
dff state_0_(.Q(STATE[0], .CLK(DOWN), .SET_N(UP), .RST_N(RES(, .D(/*...*/));
In a real flop with asynchronous set and reset, the default should be consent and would explain the error in your first code. In your second attempt, UP becomes part of the combination logic cloud along with DOWN. DOWN is also being used as a clock. Since UP is not a clock, shifting continuously happens while UP is low, completely shifting the on bit out instantly. Another error for the second case would actually be more appropriate.
For the synthesizer to do a better job, you first need to synchronize your asynchronous control signals. Use the same technique as CDC (clock domain crossing; A paper by Cliff Cummings goes into detains here). A basic example:
always #(posedge clk) begin
pre_sync_DOWN <= DOWN;
sync_DOWN <= pre_sync_DOWN;
end
Now that the controls signals are synchronized, make STATE the output of your combination logic. Example:
always #* begin
if(!sync_RES)
STATE = 8'b00010000;
else
case({sync_UP,sync_DOWN})
2'b01 : STATE = LED>>1;
2'b10 : STATE = LED<<1;
default: STATE = LED;
endcase
end
With everything running on one clock domain and explicitly defined combination logic, the synthesizer can construct equivalent logic using flops and basic gates.
FYI:
To shift only on a negedge event you need to keep the last sync value and check for the high to low transition. Remember to swap sync_ with do_ in the combination logic that drives STATE.
always #(posedge clk)
keep_DOWN <= sync_DOWN;
always #*
do_DOWN = (keep_DOWN && !sync_DOWN);