Interframe Gap and Deficit Idle Count
Explain the 96-bit gap. Then see how a 10 Gb/s sublayer bends it to put every Start on lane 0 at full rate.
You Will Be Able To
- State the minimum gap between packets and convert it to time at any rate.
- Explain why the XGMII needs Start on lane 0.
- Step the deficit idle count by hand for a run of frames.
- Tell a legal short gap from an RS that sends too fast.
- Lesson
- P11.6
- Time
- 30 min
- Runs on
- Your browser
- Check
- 5 questions
A MAC sends 65-octet frames, each with a 12-octet gap. On a four-lane XGMII the next Start would land on lane 1. The sublayer must move it.
| RS method | Gap after each 65-octet frame | Average gap | Frame octets per character |
|---|---|---|---|
| Ideal: 12 characters every time | 12 | 12 | 0.7647 |
| Always insert idle | 15 | 15.000 | 0.7386 |
| Deficit idle count | 11 to 15 | 12.000 | 0.7647 |
Bench run: 1,000 frames of 65 octets back to back on a 4-lane XGMII. A gap counts the Terminate and the idles after it.
The Gap
Between two packets the MAC leaves at least 96 bit times: 12 octets (4.4.2). The gap runs from the last bit of the FCS to the first bit of the next preamble.
| MAC rate | 96 bit times |
|---|---|
| 10 Mb/s | 9600 ns |
| 100 Mb/s | 960 ns |
| 1 Gb/s | 96 ns |
| 10 Gb/s | 9.6 ns |
| 25 Gb/s | 3.84 ns |
| 100 Gb/s | 0.96 ns |
| 400 Gb/s | 0.24 ns |
| 800 Gb/s | 0.12 ns |
The gap is 96 bit times at every rate (4.4.2, Table 4-2). It gets shorter in time as the rate rises. Bench arithmetic.
A MAC may leave more than 96 bit times, never less. On the XGMII the gap is the Terminate character plus idle characters.
One rate has an extra knob. A 10 Gb/s MAC for the WAN PHY stretches its gap to match a SONET rate (Table 4-2, ipgStretchRatio). This course does not use that PHY.
Start Must Sit on Lane 0
The XGMII carries four characters per transfer, one per lane. A Start character may only use lane 0 (46.2, Table 46-3). That keeps the PCS simple: a block always begins at a column edge.
A frame can end in any lane, so 12 more characters can leave the next Start on lane 1, 2 or 3.
- Cycle 1
The frame ends in lane 1. The Terminate takes lane 2, and an idle follows in lane 3.
- Cycle 2
Twelve characters from the Terminate would put Start on lane 2. The RS deletes two idles instead.
- Cycle 4
Start sits on lane 0 after a gap of 10 characters. The deficit idle count is now 2.
Bench run, columns 17 to 22 of the stream. The 46.3.1.4 rule allows this short gap because the count was 0 before it.
The reconciliation sublayer (RS) fixes that by bending the gap. 46.3.1.4 gives it two ways to do so.
Always insert
Add idles until Start reaches lane 0. Simple, but every frame pays up to three extra characters.
Deficit idle count
Sometimes delete idles and sometimes insert them, so that the average gap stays at 12.
The Deficit Idle Count
The deficit idle count (DIC) counts idles deleted minus idles inserted. It must stay between 0 and 3 (46.3.1.4). It is reset only at start-up.
- Control / FSM
- Control
- Where would Start land?
Count 12 characters from the Terminate. If Start falls on lane 0, nothing changes.
- Delete while there is room
Otherwise shorten the gap by m, as long as the count stays at 3 or below.
- Else insert
Lengthen the gap to the next lane 0 instead. That pays back 4 − m of the deficit.
Read the diagram as text
- Frame ends: Terminate + 11 idles
- m = 0?: Start already on lane 0
- Send 12: count unchanged
- DIC + m <= 3?: room to delete
- Delete m idles: DIC += m
- Insert 4 - m idles: DIC -= 4 - m
- Frame ends → m = 0?
- m = 0? → Send 12: yes
- m = 0? → DIC + m <= 3?: no
- DIC + m <= 3? → Delete m idles: yes
- DIC + m <= 3? → Insert 4 - m idles: no
The bounds 0 and 3 are those of 46.3.1.4. With them only one of the two choices is ever legal, so the bench computes it.
| Frame | Start would land on lane | Idles deleted | Idles inserted | Gap | DIC after |
|---|---|---|---|---|---|
| 64 octets | 0 | 0 | 0 | 12 | 0 |
| 65 octets | 1 | 1 | 0 | 11 | 1 |
| 66 octets | 2 | 2 | 0 | 10 | 3 |
| 67 octets | 3 | 0 | 1 | 13 | 2 |
| 65 octets | 1 | 1 | 0 | 11 | 3 |
| 65 octets | 1 | 0 | 3 | 15 | 0 |
| 65 octets | 1 | 1 | 0 | 11 | 1 |
| 65 octets | 1 | 1 | 0 | 11 | 2 |
Bench run of eth_gap from a count of 0. The eight gaps average 11.75 characters; over 1,000 frames the average is 12.
A single gap may therefore be 9 characters: three short of 12. The next short gaps must wait until an insertion has paid the deficit back.
What Each Method Costs
| RS | Average gap | Shortest | Longest | Frame octets per character |
|---|---|---|---|---|
| Deficit idle count | 11.997 | 9 | 15 | 0.9754 |
| Always insert | 13.477 | 12 | 15 | 0.9736 |
g.ipg_min_only = 1: always delete | 10.561 | 9 | 12 | 0.9771 |
Bench run: 1,000 frames of random size, 64 to 1,518 octets, back to back with a MAC gap of 12.
For mixed sizes the price of always inserting is small. For 65-octet frames it is large: 73.9 % of the characters carry frame octets, against 76.5 % with the DIC.
The DIC gives the same rate as an ideal 12-character gap, at the cost of a counter and some short gaps.
When the RS Is Wrong
The twin g.ipg_min_only = 1 always deletes and never inserts. Every gap is still 9 characters or more. Its average is 10.561.
That is a fault even though no single gap looks short. The average gap is now below 96 bit times, so the link carries frames faster than its MAC rate.
A check of each gap alone cannot see this. Only the running count shows the average drifting.
Gaps at the Receiver
A receiver cannot count on 12 octets. The gap shrinks on the way, and 4.4.2 says how far.
| MAC rate | Shortest gap a receiver may see | Note in 4.4.2 |
|---|---|---|
| 10 Mb/s | 47 bit times | Note 1 |
| 1 Gb/s | 64 bit times | Note 3 |
| 2.5, 5, 10 and 25 Gb/s | 40 bit times (5 octets) | Note 4 |
| 40 to 400 Gb/s | 8 bit times (1 octet) | Note 7 |
Read in IEEE Std 802.3-2022 4.4.2. The gap shrinks on the way: network delays, clock tolerance and, from 40 Gb/s, lane alignment.
A PHY may delete idles to absorb a clock difference between the two ends. A receive MAC must therefore accept a gap as short as the table says.
Try It
The bench asks the RS model for the gap after eight frames, from a count of 0.
Each line shows where Start would land, what the RS did, the gap and the count.
Add g.method = 1; after g.reset();. The RS then always inserts, and the average rises to 14.25.
frame 64 octets: Start would land on lane 0; deleted 0, inserted 0; gap 12; DIC 0 frame 65 octets: Start would land on lane 1; deleted 1, inserted 0; gap 11; DIC 1 frame 66 octets: Start would land on lane 2; deleted 2, inserted 0; gap 10; DIC 3 frame 67 octets: Start would land on lane 3; deleted 0, inserted 1; gap 13; DIC 2 frame 65 octets: Start would land on lane 1; deleted 1, inserted 0; gap 11; DIC 3 frame 65 octets: Start would land on lane 1; deleted 0, inserted 3; gap 15; DIC 0 frame 65 octets: Start would land on lane 1; deleted 1, inserted 0; gap 11; DIC 1 frame 65 octets: Start would land on lane 1; deleted 1, inserted 0; gap 11; DIC 2 average gap over 8 frames: 11.75 charactersRead included file: hs_pkg.sv
`timescale 1ns/1ns
// hs_pkg.sv: shared pure functions, and the bench random generator (read-only).
package hs_pkg;
// xorshift32 (Marsaglia 2003): a linear generator over GF(2), period 2^32 - 1. A zero state is
// replaced by 1, because zero maps to itself.
function automatic [31:0] xs32(input [31:0] s);
reg [31:0] x;
x = (s == 32'd0) ? 32'd1 : s;
x = x ^ (x << 13); x = x ^ (x >> 17); x = x ^ (x << 5);
return x;
endfunction
// A Fibonacci PRBS generator x^n + x^m + 1 (PRBS7: 7,6; PRBS9: 9,5; PRBS15: 15,14; PRBS23: 23,18;
// PRBS31: 31,28). The output bit is s[n-1]; the next state shifts the feedback into bit 0.
function automatic [31:0] prbs_next(input [31:0] s, input int n, input int m);
reg fb;
fb = s[n-1] ^ s[m-1];
return ((s << 1) | {31'd0, fb}) & ((n == 32) ? 32'hffffffff : ((32'd1 << n) - 32'd1));
endfunction
function automatic int ones(input [63:0] v);
int n; n = 0;
for (int i = 0; i < 64; i++) n += v[i];
return n;
endfunction
// GF(16) with x^4 + x + 1, and GF(1024) with x^10 + x^3 + 1: shift-and-add multiply, no tables.
function automatic [3:0] gf16_mul(input [3:0] a, input [3:0] b);
reg [3:0] p, x; p = 4'd0; x = a;
for (int i = 0; i < 4; i++) begin
if (b[i]) p = p ^ x;
x = x[3] ? ({x[2:0], 1'b0} ^ 4'b0011) : {x[2:0], 1'b0};
end
return p;
endfunction
function automatic [3:0] gf16_pow(input int e); // alpha^e, alpha = x
reg [3:0] p; int k; p = 4'd1; k = e % 15; if (k < 0) k += 15;
for (int i = 0; i < k; i++) p = gf16_mul(p, 4'd2);
return p;
endfunction
function automatic int gf16_log(input [3:0] a); // -1 for 0
reg [3:0] p; p = 4'd1;
if (a == 4'd0) return -1;
for (int i = 0; i < 15; i++) begin if (p == a) return i; p = gf16_mul(p, 4'd2); end
return -1;
endfunction
function automatic [9:0] gf1024_mul(input [9:0] a, input [9:0] b);
reg [9:0] p, x; p = 10'd0; x = a;
for (int i = 0; i < 10; i++) begin
if (b[i]) p = p ^ x;
x = x[9] ? ({x[8:0], 1'b0} ^ 10'b0000001001) : {x[8:0], 1'b0};
end
return p;
endfunction
function automatic [9:0] gf1024_pow(input int e);
reg [9:0] p; int k; p = 10'd1; k = e % 1023; if (k < 0) k += 1023;
for (int i = 0; i < k; i++) p = gf1024_mul(p, 10'd2);
return p;
endfunction
endpackage
// hs_rng: the bench's random numbers. Same seed, same numbers, on every engine.
module hs_rng;
int unsigned s = 32'h2545f491;
function automatic void seed(input int unsigned v); s = (v == 0) ? 32'd1 : v; endfunction
function automatic int unsigned next(); s = hs_pkg::xs32(s); return s; endfunction
function automatic int below(input int n); s = hs_pkg::xs32(s); return (n <= 0) ? 0 : int'(s % n); endfunction
function automatic bit bit1(); s = hs_pkg::xs32(s); return s[16]; endfunction
function automatic bit chance_ppm(input int ppm); s = hs_pkg::xs32(s); return (s % 1000000) < ppm; endfunction
function automatic [7:0] byte8(); s = hs_pkg::xs32(s); return s[23:16]; endfunction
endmodule
Read included file: hs_code.sv
`timescale 1ns/1ns
// hs_code.sv: line codes, CRCs and scramblers (read-only).
// ---------------------------------------------------------------------------------------------
// hs_8b10b: 8b/10b encoder and decoder with running disparity and the 12 K codes.
// code[9:0] = {a,b,c,d,e,i,f,g,h,j}; bit a (code[9]) goes on the wire first. rd: 0 = RD-, 1 = RD+.
// encode(d, k, rd_in) -> {rd_out, code[9:0]} (11 bits)
// decode(code, rd_in) -> {rd_out, disp_err, code_err, k, d[7:0]} (12 bits)
// enc(d, k) / dec(code): the same, with the running disparity kept in rd_tx / rd_rx
// A K request for a byte that is not one of the 12 K codes is encoded as data and counted in bad_k.
// Decoder: code_err = the code is in neither column of the table; disp_err = it is a valid code
// only in the other column. rd_out follows the sub-block rule on the received bits.
// Faults: no_rd (always the RD- column), a7_never (never the alternate D.x.A7),
// keep_rd_on_unbalanced (the 4b sub-block is chosen with the disparity from before the 6b one),
// no_balanced_flip (D.07 and D.x.3 keep their RD- form at RD+), dec_no_disp_check.
module hs_8b10b;
bit no_rd = 0, a7_never = 0, keep_rd_on_unbalanced = 0, no_balanced_flip = 0, dec_no_disp_check = 0;
bit trace = 0;
bit rd_tx = 0, rd_rx = 0;
int bad_k = 0, symbols = 0, code_errs = 0, disp_errs = 0;
bit last_k = 0, last_code_err = 0, last_disp_err = 0;
bit valid_m [0:1023]; bit valid_p [0:1023]; bit [8:0] kd [0:1023]; bit built = 0;
function automatic [5:0] t6(input [4:0] x); // 5b/6b, RD- column, abcdei
case (x)
5'd0: return 6'b100111; 5'd1: return 6'b011101; 5'd2: return 6'b101101; 5'd3: return 6'b110001;
5'd4: return 6'b110101; 5'd5: return 6'b101001; 5'd6: return 6'b011001; 5'd7: return 6'b111000;
5'd8: return 6'b111001; 5'd9: return 6'b100101; 5'd10: return 6'b010101; 5'd11: return 6'b110100;
5'd12: return 6'b001101; 5'd13: return 6'b101100; 5'd14: return 6'b011100; 5'd15: return 6'b010111;
5'd16: return 6'b011011; 5'd17: return 6'b100011; 5'd18: return 6'b010011; 5'd19: return 6'b110010;
5'd20: return 6'b001011; 5'd21: return 6'b101010; 5'd22: return 6'b011010; 5'd23: return 6'b111010;
5'd24: return 6'b110011; 5'd25: return 6'b100110; 5'd26: return 6'b010110; 5'd27: return 6'b110110;
5'd28: return 6'b001110; 5'd29: return 6'b101110; 5'd30: return 6'b011110; default: return 6'b101011;
endcase
endfunction
function automatic [3:0] t4d(input [2:0] y); // 3b/4b data, RD- column, fghj (y = 7 is P7)
case (y)
3'd0: return 4'b1011; 3'd1: return 4'b1001; 3'd2: return 4'b0101; 3'd3: return 4'b1100;
3'd4: return 4'b1101; 3'd5: return 4'b1010; 3'd6: return 4'b0110; default: return 4'b1110;
endcase
endfunction
function automatic [3:0] t4k(input [2:0] y); // 3b/4b for K codes, RD- form before the flip
case (y)
3'd0: return 4'b1011; 3'd1: return 4'b0110; 3'd2: return 4'b1010; 3'd3: return 4'b1100;
3'd4: return 4'b1101; 3'd5: return 4'b0101; 3'd6: return 4'b1001; default: return 4'b0111;
endcase
endfunction
function automatic bit k_ok(input [7:0] d);
return d[4:0] == 5'd28 || d == 8'hf7 || d == 8'hfb || d == 8'hfd || d == 8'hfe;
endfunction
function automatic int n6(input [5:0] v); return v[0] + v[1] + v[2] + v[3] + v[4] + v[5]; endfunction
function automatic int n4(input [3:0] v); return v[0] + v[1] + v[2] + v[3]; endfunction
// running disparity at the end of a sub-block (more ones: +; more zeros: -; 000111 and 0011: +; 111000 and 1100: -)
function automatic bit rd6(input [5:0] v, input bit rd);
return (n6(v) > 3) ? 1'b1 : (n6(v) < 3) ? 1'b0 : (v == 6'b000111) ? 1'b1 : (v == 6'b111000) ? 1'b0 : rd;
endfunction
function automatic bit rd4(input [3:0] v, input bit rd);
return (n4(v) > 2) ? 1'b1 : (n4(v) < 2) ? 1'b0 : (v == 4'b0011) ? 1'b1 : (v == 4'b1100) ? 1'b0 : rd;
endfunction
// faults: [0] no_rd [1] a7_never [2] keep_rd_on_unbalanced [3] no_balanced_flip
function automatic [10:0] core(input [7:0] d, input bit k, input bit rd_in, input [3:0] f);
reg [5:0] six; reg [3:0] four; reg rd0, r6, rd1, a7, kk; reg [4:0] x; reg [2:0] y; int o6, o4;
x = d[4:0]; y = d[7:5]; kk = k && k_ok(d); rd0 = f[0] ? 1'b0 : rd_in;
six = (kk && x == 5'd28) ? 6'b001111 : t6(x); o6 = n6(six);
if (rd0 && (o6 != 3 || (six == 6'b111000 && !f[3]))) begin six = ~six; o6 = 6 - o6; end
r6 = (o6 > 3) ? 1'b1 : (o6 < 3) ? 1'b0 : (six == 6'b000111) ? 1'b1 : (six == 6'b111000) ? 1'b0 : rd0;
rd1 = f[2] ? rd0 : r6;
if (kk) begin
four = t4k(y); if (rd1) four = ~four;
end else begin
a7 = (y == 3'd7) && !f[1] && ((!rd1 && (x == 5'd17 || x == 5'd18 || x == 5'd20)) || (rd1 && (x == 5'd11 || x == 5'd13 || x == 5'd14)));
four = a7 ? 4'b0111 : t4d(y);
if (rd1 && (n4(four) != 2 || (four == 4'b1100 && !f[3]))) four = ~four;
end
o4 = n4(four);
return {(o4 > 2) ? 1'b1 : (o4 < 2) ? 1'b0 : (four == 4'b0011) ? 1'b1 : (four == 4'b1100) ? 1'b0 : r6, six, four};
endfunction
function automatic [10:0] encode(input [7:0] d, input bit k, input bit rd_in);
if (k && !k_ok(d)) bad_k++;
return core(d, k, rd_in, {no_balanced_flip, keep_rd_on_unbalanced, a7_never, no_rd});
endfunction
function automatic void build();
reg [10:0] c;
for (int i = 0; i < 1024; i++) begin valid_m[i] = 0; valid_p[i] = 0; kd[i] = 9'd0; end
for (int i = 0; i < 268; i++) begin
reg [7:0] d; reg k;
if (i < 256) begin d = i[7:0]; k = 0; end
else begin k = 1; d = (i < 264) ? {3'(i - 256), 5'd28} : (i == 264) ? 8'hf7 : (i == 265) ? 8'hfb : (i == 266) ? 8'hfd : 8'hfe; end
c = core(d, k, 1'b0, 4'd0); valid_m[c[9:0]] = 1; kd[c[9:0]] = {k, d};
c = core(d, k, 1'b1, 4'd0); valid_p[c[9:0]] = 1; kd[c[9:0]] = {k, d};
end
built = 1;
endfunction
function automatic [11:0] decode(input [9:0] code, input bit rd_in);
reg here, there, ce, de, r;
if (!built) build();
here = rd_in ? valid_p[code] : valid_m[code];
there = rd_in ? valid_m[code] : valid_p[code];
ce = !(here || there); de = !ce && !here && !dec_no_disp_check;
r = rd4(code[3:0], rd6(code[9:4], rd_in));
return {r, de, ce, ce ? 9'd0 : kd[code]};
endfunction
function automatic [9:0] enc(input [7:0] d, input bit k);
reg [10:0] r; r = encode(d, k, rd_tx); rd_tx = r[10];
if (trace) $display("s%0d TX %s%0d.%0d code=%b rd=%s", symbols, (k && k_ok(d)) ? "K" : "D", d[4:0], d[7:5], r[9:0], r[10] ? "+" : "-");
symbols++;
return r[9:0];
endfunction
function automatic [8:0] dec(input [9:0] code);
reg [11:0] r; r = decode(code, rd_rx); rd_rx = r[11];
last_disp_err = r[10]; last_code_err = r[9]; last_k = r[8];
if (r[9]) code_errs++; if (r[10]) disp_errs++;
if (trace && (r[9] || r[10])) $display("RX %b %s", code, r[9] ? "code error" : "disparity error");
return r[8:0];
endfunction
function automatic bit comma7(input [6:0] w); return w == 7'b0011111 || w == 7'b1100000; endfunction
endmodule
// ---------------------------------------------------------------------------------------------
// hs_crc: a CRC of 1 to 32 bits with any polynomial, seed, bit order and final inversion.
// The register shifts left; the polynomial omits the x^width term. The field is sent register bit
// width-1 first (USB 2.0, USB 3.2 and Ethernet all send it this way).
// use_crc32(), use_crc16_usb2(), use_crc5_usb2(), use_crc16_usb3(): parameter sets
// reset(), crc_bit(b), crc_byte(v), crc_bits(v, n), field(), field_bit(i), check_value(), crc_residue_ok()
// Faults: msb_first (bytes enter MSB first), no_invert (the field is not complemented), seed_zero.
module hs_crc;
int width = 32;
bit [31:0] poly = 32'h04c11db7, seed = 32'hffffffff, residue = 32'hc704dd7b;
bit lsb_first = 1, invert = 1;
bit msb_first = 0, no_invert = 0, seed_zero = 0;
bit [31:0] r = 32'hffffffff;
function automatic [31:0] mask(); return (width >= 32) ? 32'hffffffff : ((32'd1 << width) - 32'd1); endfunction
function automatic void reset(); r = seed_zero ? 32'd0 : (seed & mask()); endfunction
function automatic void set(input int w, input [31:0] p, input [31:0] s, input [31:0] res, input bit lsb, input bit inv);
width = w; poly = p; seed = s; residue = res; lsb_first = lsb; invert = inv; reset();
endfunction
function automatic void use_crc32(); set(32, 32'h04c11db7, 32'hffffffff, 32'hc704dd7b, 1, 1); endfunction
function automatic void use_crc16_usb2(); set(16, 32'h8005, 32'hffff, 32'h800d, 1, 1); endfunction
function automatic void use_crc5_usb2(); set(5, 32'h05, 32'h1f, 32'h0c, 1, 1); endfunction
function automatic void use_crc16_usb3(); set(16, 32'h100b, 32'hffff, 32'hf6aa, 1, 1); endfunction
function automatic void crc_bit(input bit b);
reg fb; fb = b ^ r[width - 1];
r = ((r << 1) ^ (fb ? poly : 32'd0)) & mask();
endfunction
function automatic void crc_byte(input [7:0] v);
for (int i = 0; i < 8; i++) crc_bit(v[(lsb_first ^ msb_first) ? i : 7 - i]);
endfunction
function automatic void crc_bits(input [31:0] v, input int n); // n bits of v, bit 0 first
for (int i = 0; i < n; i++) crc_bit(v[i]);
endfunction
function automatic [31:0] field(); return (invert && !no_invert) ? (~r & mask()) : r; endfunction
function automatic bit field_bit(input int i); reg [31:0] f; f = field(); return f[width - 1 - i]; endfunction
// the usual published "check" number: the field with its first-sent bit read as bit 0
function automatic [31:0] check_value();
reg [31:0] f, o; f = field(); o = 32'd0;
for (int i = 0; i < width; i++) o[i] = f[width - 1 - i];
return lsb_first ? o : f;
endfunction
function automatic bit crc_residue_ok(); return r == (residue & mask()); endfunction
endmodule
// ---------------------------------------------------------------------------------------------
// hs_scr: additive (frame-synchronous) and multiplicative (self-synchronising) scramblers.
// poly holds the coefficients: bit k is x^k; the +1 term is always used (taps() sets bit 0); len is the degree.
// Additive: a Galois LFSR. The key bit is s[len-1]; each step shifts left and adds the low terms of
// the polynomial when the key bit is 1. The data never enters the register.
// Multiplicative: the line bit is the data bit XOR the register bits at the polynomial's terms;
// the scrambler shifts its own output in, the descrambler shifts the received line bit in.
// scr_add(len, poly, seed), scr_mul(len, poly), use_usb3_gen1(), use_x58(), use_x23_usb3() (see its note)
// scr_bit(b), dscr_bit(b), scr_byte(v), dscr_byte(v) (bit 0 first), scr_reset(), scr_hold() (a skip
// symbol: no step), scr_unscrambled() (a symbol sent in the clear: 8 steps, data untouched)
// Faults: advance_on_skip (scr_hold steps the LFSR 8 times), no_reset (scr_reset does nothing),
// wrong_tap (the lowest term x^k becomes x^(k+1)), seed_zero (scr_reset loads 0).
module hs_scr;
bit mult = 0; int len = 16;
bit [63:0] poly = 64'h10039, seed = 64'hffff;
bit advance_on_skip = 0, no_reset = 0, wrong_tap = 0, seed_zero = 0;
bit [63:0] s = 64'hffff;
int steps = 0;
function automatic [63:0] lmask(); return (len >= 64) ? 64'hffffffffffffffff : ((64'd1 << len) - 64'd1); endfunction
function automatic [63:0] taps();
reg [63:0] p; int k; p = poly | 64'd1; k = 1;
if (wrong_tap) begin
while (k < len && !p[k]) k++;
if (k < len) begin p[k] = 1'b0; p[k + 1] = 1'b1; end
end
return p;
endfunction
function automatic void load(); s = (seed_zero ? 64'd0 : seed) & lmask(); endfunction
function automatic void scr_add(input int n, input [63:0] p, input [63:0] sd); mult = 0; len = n; poly = p; seed = sd; s = sd & lmask(); endfunction
function automatic void scr_mul(input int n, input [63:0] p); mult = 1; len = n; poly = p; seed = 64'd0; s = 64'd0; endfunction
function automatic void use_usb3_gen1(); scr_add(16, 64'h10039, 64'hffff); endfunction // x^16 + x^5 + x^4 + x^3 + 1, seed FFFFh
function automatic void use_x58(); scr_mul(58, (64'd1 << 58) | (64'd1 << 39) | 64'd1); endfunction // x^58 + x^39 + 1
// USB 3.2 Gen 2 (§6.3.2.3) polynomial x^23 + x^21 + x^16 + x^8 + x^5 + x^2 + 1 with seed 1DBFBCh. Only the
// polynomial and seed are the specification's: the register form here is this library's Galois form, and
// the reset and bypass rules of §6.3.2.3 are not modelled, so do not quote its key stream as USB's.
function automatic void use_x23_usb3(); scr_add(23, (64'd1 << 23) | (64'd1 << 21) | (64'd1 << 16) | (64'd1 << 8) | (64'd1 << 5) | (64'd1 << 2) | 64'd1, 64'h1dbfbc); endfunction
function automatic void scr_reset(); if (!no_reset) load(); endfunction
function automatic bit key(); return s[len - 1]; endfunction
function automatic void step();
reg kb; kb = s[len - 1];
s = ((s << 1) & lmask()) ^ (kb ? (taps() & lmask()) : 64'd0);
steps++;
endfunction
// the XOR of the register bits at the polynomial's terms x^1..x^len (bit k-1 of the register for x^k)
function automatic bit fbx(input [63:0] st); return ^(st & (taps() >> 1) & lmask()); endfunction
function automatic bit scr_bit(input bit b);
reg o;
if (!mult) begin o = b ^ s[len - 1]; step(); end
else begin o = b ^ fbx(s); s = ((s << 1) | {63'd0, o}) & lmask(); steps++; end
return o;
endfunction
function automatic bit dscr_bit(input bit b);
reg o;
if (!mult) begin o = b ^ s[len - 1]; step(); end
else begin o = b ^ fbx(s); s = ((s << 1) | {63'd0, b}) & lmask(); steps++; end
return o;
endfunction
function automatic [7:0] scr_byte(input [7:0] v);
reg [7:0] o; for (int i = 0; i < 8; i++) o[i] = scr_bit(v[i]); return o;
endfunction
function automatic [7:0] dscr_byte(input [7:0] v);
reg [7:0] o; for (int i = 0; i < 8; i++) o[i] = dscr_bit(v[i]); return o;
endfunction
// a skip symbol passes unscrambled and the LFSR holds
function automatic void scr_hold(); if (advance_on_skip && !mult) for (int i = 0; i < 8; i++) step(); endfunction
// another symbol that is sent unscrambled (a K code other than COM and SKP): the LFSR still steps 8 times
function automatic void scr_unscrambled(); if (!mult) for (int i = 0; i < 8; i++) step(); endfunction
endmodule
Read included file: eth_mac.sv
`timescale 1ns/1ns
// eth_frame: one MAC frame as octets, its packet as wire bits, and a bit-serial receiver.
// pay[i] client data, set by the bench before build()
// build(da, sa, lt, len) destination, source, Length/Type, len data octets, then the FCS -> oct[0..n-1].
// It adds no pad: give it 46 data octets or more. (Padding is P11.2.1's task; the
// pages show the padding MAC's frames as measured tables.)
// wire_bits(), wire_bit(i) the packet in transmit order: preamble, SFD, frame
// rx_start(), rx_bit(b), rx_end() hunt for the SFD, collect octets, then parse()
// parse() da, sa, lt, is_length, is_type, data_len, fcs_rx, fcs_calc, fcs_ok
// Faults: fcs_msb_first, fcs_no_invert, fcs_covers_preamble, sfd_wrong, msb_first_octets.
module eth_frame;
localparam int MIN_FRAME = 64, MAX_BASIC = 1518, MAX_QTAG = 1522, MAX_ENVELOPE = 2000;
bit fcs_msb_first = 0, fcs_no_invert = 0, fcs_covers_preamble = 0, sfd_wrong = 0, msb_first_octets = 0;
hs_crc crc();
bit [7:0] pay [0:2047];
bit [7:0] oct [0:2047];
int n = 0;
bit [47:0] da = 0, sa = 0; bit [15:0] lt = 0;
bit is_length = 0, is_type = 0, fcs_ok = 0;
int data_len = 0;
bit [31:0] fcs_rx = 0, fcs_calc = 0;
// receiver state
bit in_frame = 0; bit [7:0] hist = 0, cur = 0; int nb = 0, excess = 0, sfd_at = -1, bits_seen = 0;
// The FCS of oct[0..len-1] as four octets: the first octet sent is in [7:0].
function automatic [31:0] fcs_of(input int len);
crc.use_crc32(); crc.msb_first = fcs_msb_first; crc.no_invert = fcs_no_invert; crc.reset();
if (fcs_covers_preamble) begin
for (int i = 0; i < 7; i++) crc.crc_byte(8'h55);
crc.crc_byte(8'hd5);
end
for (int i = 0; i < len; i++) crc.crc_byte(oct[i]);
return crc.check_value();
endfunction
function automatic void build(input [47:0] d, input [47:0] s, input [15:0] t, input int len);
int k; reg [31:0] f;
k = 0;
for (int i = 0; i < 6; i++) begin oct[k] = d[47 - 8*i -: 8]; k++; end
for (int i = 0; i < 6; i++) begin oct[k] = s[47 - 8*i -: 8]; k++; end
oct[k] = t[15:8]; k++;
oct[k] = t[7:0]; k++;
for (int i = 0; i < len; i++) begin oct[k] = pay[i]; k++; end
f = fcs_of(k);
for (int i = 0; i < 4; i++) begin oct[k] = f[8*i +: 8]; k++; end
n = k; excess = 0;
parse();
endfunction
function automatic void parse();
da = 0; sa = 0; lt = 0; is_length = 0; is_type = 0; fcs_ok = 0; data_len = 0; fcs_rx = 0; fcs_calc = 0;
if (n >= 14) begin
da = {oct[0], oct[1], oct[2], oct[3], oct[4], oct[5]};
sa = {oct[6], oct[7], oct[8], oct[9], oct[10], oct[11]};
lt = {oct[12], oct[13]};
is_length = (lt <= 16'd1500);
is_type = (lt >= 16'd1536);
end
if (n >= 18) begin
fcs_rx = {oct[n-1], oct[n-2], oct[n-3], oct[n-4]};
fcs_calc = fcs_of(n - 4);
fcs_ok = (fcs_rx == fcs_calc);
data_len = is_length ? int'(lt) : n - 18;
end
endfunction
function automatic int wire_bits(); return 8 * (8 + n); endfunction
function automatic bit wire_bit(input int i);
int o, b; reg [7:0] v;
o = i / 8; b = i % 8;
if (o < 7) v = 8'h55;
else if (o == 7) v = sfd_wrong ? 8'h55 : 8'hd5;
else v = oct[o - 8];
return msb_first_octets ? v[7 - b] : v[b];
endfunction
function automatic void rx_start();
in_frame = 0; hist = 0; cur = 0; nb = 0; n = 0; excess = 0; sfd_at = -1; bits_seen = 0;
endfunction
function automatic void rx_bit(input bit b);
bits_seen++;
if (!in_frame) begin
hist = {hist[6:0], b}; // the oldest bit is hist[7]
if (hist == 8'b10101011) begin in_frame = 1; sfd_at = bits_seen; end
end else begin
if (msb_first_octets) cur[7 - nb] = b; else cur[nb] = b;
nb++;
if (nb == 8) begin oct[n] = cur; n++; nb = 0; end
end
endfunction
function automatic void rx_end(); excess = nb; parse(); endfunction
endmodule
// eth_axis: a MAC client bus in the style of AXI4-Stream (IHI 0051B 2.2 handshake, 2.5 TKEEP, 2.6 TLAST, 2.9 TUSER),
// 64 bits wide. Byte 0 of a beat is tdata[7:0] and is first in the frame (AXI4-Stream 2.4.1). Bench conventions,
// not AXI4-Stream's or 802.3's: tkeep is all ones except on the last beat, where it covers bytes 0 to k-1; one tuser
// bit (transfer-based, 2.9.1), read on the last beat only, marks a bad frame (receive) or asks the MAC to abort it
// (transmit).
// load(n) beats for octets oct[0..n-1]: bd[], bk[], bl[], nbeats; bv[] = valid per cycle, ncyc
// tx_cycles() the transmit MAC: one beat per cycle once a frame has started. Returns 0 sent,
// 1 underrun (tvalid fell inside the frame), 2 aborted by tuser. sent_octets, fail_cycle.
// Fault: tvalid_gap (load() leaves cycle 3 of the frame empty: the client was late with a beat).
module eth_axis;
bit tvalid_gap = 0;
bit [7:0] oct [0:2047];
bit [63:0] bd [0:300]; bit [7:0] bk [0:300]; bit bl [0:300]; bit bu [0:300];
int nbeats = 0;
bit bv [0:301]; int bix [0:301]; int ncyc = 0;
int sent_octets = 0, fail_cycle = -1;
function automatic void load(input int n);
int b, k;
nbeats = (n + 7) / 8;
for (b = 0; b < nbeats; b++) begin
bd[b] = 64'd0; bk[b] = 8'd0; bl[b] = (b == nbeats - 1); bu[b] = 0;
for (k = 0; k < 8; k++) if (8*b + k < n) begin bd[b][8*k +: 8] = oct[8*b + k]; bk[b][k] = 1'b1; end
end
ncyc = 0;
for (b = 0; b < nbeats; b++) begin
if (tvalid_gap && b == 3) begin bv[ncyc] = 0; bix[ncyc] = -1; ncyc++; end
bv[ncyc] = 1; bix[ncyc] = b; ncyc++;
end
endfunction
function automatic int tx_cycles();
bit started; started = 0; sent_octets = 0; fail_cycle = -1;
for (int c = 0; c < ncyc; c++) begin
if (!bv[c]) begin
if (started) begin fail_cycle = c; return 1; end
end else begin
started = 1;
if (bl[bix[c]] && bu[bix[c]]) begin fail_cycle = c; return 2; end
for (int k = 0; k < 8; k++) if (bk[bix[c]][k]) sent_octets++;
end
end
return 0;
endfunction
endmodule
// eth_pause: the PAUSE function of MAC Control (Annex 31B) at the station that receives PAUSE frames, in octet
// times (one octet time is 8 bit times; a pause quantum is 512 bit times = 64 octet times, 31B.2).
// reset(), rx_pause(t, q) a valid PAUSE frame with pause_time q ended at time t: the timer restarts (31B.3.3)
// may_start(t) 1 when a new data frame may start at time t. A frame already started is never cut
// (31B.3.7). react is the bench's reaction time: frames may still start until t + react.
// paused_until the time the timer runs out
// pause_octets(q, o[]) not here: build a PAUSE frame with eth_frame (DA 01-80-C2-00-00-01, 8808h, 0001h, q)
// Faults: pause_ignored, pause_adds_not_overrides (a new frame adds to the timer, and zero does not cancel),
// slow_reaction (react = 70 quanta, over the 60 allowed at 10 Gb/s), timer_in_bits (q counts bit times).
module eth_pause;
localparam int QUANTUM = 64; // octet times
bit pause_ignored = 0, pause_adds_not_overrides = 0, slow_reaction = 0, timer_in_bits = 0;
int react = 0, t_rx = -1, paused_until = 0;
function automatic void reset(); t_rx = -1; paused_until = 0; endfunction
function automatic void rx_pause(input int t, input int q);
int len;
if (pause_ignored) return;
len = timer_in_bits ? (q + 7) / 8 : q * QUANTUM;
if (pause_adds_not_overrides && paused_until > t) paused_until = paused_until + len;
else paused_until = t + len;
t_rx = t;
endfunction
function automatic bit may_start(input int t);
int r;
r = slow_reaction ? 70 * QUANTUM : react;
if (t_rx < 0) return 1;
if (t < t_rx + r) return 1;
return t >= paused_until;
endfunction
endmodule
// eth_pfc: priority-based flow control (Annex 31D). The frame is 802.3's; holding a priority's queue is the MAC
// Control client's job (IEEE 802.1Q, not held), so the eight timers here are a bench client, in octet times.
// build(f_oct[], pev, times[]) not here: the bench writes opcode 01-01, the vector and eight times into eth_frame
// reset(), rx(t, pev, tm[0..7]) for each e[n] = 1 (bit n of the vector's low octet), priority n stops until
// t + 64 * time[n] (hold_end[n]); e[n] = 0 leaves priority n alone; time[n] = 0 releases it
// blocked(n, t) 1 while priority n is held
// Fault: pfc_wrong_priority (reads e[n] from bit 7 - n).
module eth_pfc;
bit pfc_wrong_priority = 0;
int hold_end [0:7];
function automatic void reset(); for (int i = 0; i < 8; i++) hold_end[i] = 0; endfunction
function automatic void rx(input int t, input [15:0] pev, input [127:0] tm); // tm[16*n +: 16] = time[n]
bit e;
for (int n = 0; n < 8; n++) begin
e = pfc_wrong_priority ? pev[7 - n] : pev[n];
if (e) hold_end[n] = t + 64 * int'(tm[16*n +: 16]);
end
endfunction
function automatic bit blocked(input int n, input int t); return t < hold_end[n]; endfunction
endmodule
// eth_csma_tx: one half-duplex transmitter (4.2.3.2), one call of tick() per octet time.
// reset(), load(frame_octets) a frame of frame_octets (DA to FCS) waits to be sent; it is padded to min_frame
// tick(carrier) carrier: another station's signal is on the wire here now. Returns 1 when the
// station puts an octet (packet or jam) on the wire.
// plan[i] the wait, in slots, after collision i + 1. The bench sets it (default 0). This
// readable station draws nothing and has no attempt limit.
// state 0 idle, 1 defer, 2 send, 3 jam, 4 backoff, 5 sent (as far as the sender knows)
// pkt, sent, quiet, left, attempts, late, last_start, last_r
// min_frame is 64 octets; a bench may lower it to show why it is 64. Fault: no_backoff (every wait is 0 slots).
// Deference is a bench simplification: the 12-octet quiet count restarts on any carrier, own or received, and the
// station never starts while carrier is present. 4.2.8 restarts only the first part of the gap, and then transmits
// whatever the carrier.
module eth_csma_tx;
localparam int SLOT = 64, JAM = 4, GAP = 12, PRE = 8; // octet times
bit no_backoff = 0;
int min_frame = 64;
int plan [0:63];
int state = 0, pkt = 0, sent = 0, quiet = GAP, left = 0, attempts = 0, late = 0, last_start = -1, last_r = 0, now = 0;
function automatic void reset();
state = 0; pkt = 0; sent = 0; quiet = GAP; left = 0; attempts = 0; late = 0; last_start = -1; last_r = 0; now = 0;
endfunction
function automatic void load(input int frame_octets);
pkt = PRE + frame_octets;
if (frame_octets < min_frame) pkt = PRE + min_frame;
state = 1; attempts = 0; sent = 0; late = 0;
endfunction
function automatic bit tick(input bit carrier);
bit on;
on = 0;
if (state == 1 && !carrier && quiet >= GAP) begin state = 2; sent = 0; last_start = now; end
if (state == 2 && carrier) begin
attempts++; if (sent > SLOT) late++;
state = 3; left = JAM;
end
case (state)
2: begin sent++; on = 1; if (sent == pkt) state = 5; end
3: begin
on = 1; left--;
if (left == 0) begin
last_r = no_backoff ? 0 : plan[(attempts - 1) % 64];
left = SLOT * last_r;
state = (left == 0) ? 1 : 4;
end
end
4: begin left--; if (left == 0) state = 1; end
default: ;
endcase
quiet = (carrier || on) ? 0 : ((quiet < GAP) ? quiet + 1 : GAP);
now++;
return on;
endfunction
endmodule
// eth_csma: two eth_csma_tx stations, A and B, at the two ends of a wire with a one-way delay of d octet times.
// run(fa, ready_a, fb, ready_b, tmax) each station loads its frame (fa, fb octets) at its ready time
// on_a[t], on_b[t] what each station put on the wire at time t
// ev_t[], ev_who[] (0 A, 1 B), ev_kind[], ev_v[], n_ev the event log:
// kind 0 starts attempt v; 1 sees a collision after v octets; 2 jam over, backs off v slots;
// 3 last octet sent (v = octets); 4 gives up after v attempts
// delivered(i) 1 when station i's last attempt reached the far end with no overlap there
// a.plan[], b.plan[], a.no_backoff, b.no_backoff, a.min_frame, b.min_frame: the stations' settings
module eth_csma;
eth_csma_tx a(); eth_csma_tx b();
int d = 10, n_ev = 0, tend = 0;
bit on_a [0:16383]; bit on_b [0:16383];
int ev_t [0:255]; int ev_who [0:255]; int ev_kind [0:255]; int ev_v [0:255];
function automatic void log_ev(input int t, input int who, input int kind, input int v);
if (n_ev < 256) begin ev_t[n_ev] = t; ev_who[n_ev] = who; ev_kind[n_ev] = kind; ev_v[n_ev] = v; n_ev++; end
endfunction
function automatic void note(input int t, input int who, input int s0, input int s1, input int sent0, input int att, input int r, input int sent1);
if (s0 != 2 && s1 == 2) log_ev(t, who, 0, att + 1);
if (s0 == 2 && (s1 == 3 || s1 == 4)) log_ev(t, who, 1, sent0);
if (s0 == 3 && (s1 == 4 || s1 == 1)) log_ev(t, who, 2, r);
if (s0 != 5 && s1 == 5) log_ev(t, who, 3, sent1);
if (s0 != 6 && s1 == 6) log_ev(t, who, 4, att);
endfunction
function automatic void run(input int fa, input int ready_a, input int fb, input int ready_b, input int tmax);
bit ca, cb; int sa, sb, xa, xb;
a.reset(); b.reset(); n_ev = 0; tend = tmax;
for (int t = 0; t < tmax; t++) begin
if (t == ready_a) a.load(fa);
if (t == ready_b) b.load(fb);
ca = (t >= d) ? on_b[t - d] : 1'b0;
cb = (t >= d) ? on_a[t - d] : 1'b0;
sa = a.state; sb = b.state; xa = a.sent; xb = b.sent;
on_a[t] = a.tick(ca);
on_b[t] = b.tick(cb);
note(t, 0, sa, a.state, xa, a.attempts, a.last_r, a.sent);
note(t, 1, sb, b.state, xb, b.attempts, b.last_r, b.sent);
end
endfunction
function automatic bit delivered(input int i);
int s, p, u;
if (i == 0 && a.state != 5) return 0;
if (i == 1 && b.state != 5) return 0;
s = (i == 0) ? a.last_start : b.last_start;
p = (i == 0) ? a.pkt : b.pkt;
for (int t = s; t < s + p; t++) begin
u = t + d;
if (u < tend && ((i == 0) ? on_b[u] : on_a[u])) return 0;
end
return 1;
endfunction
endmodule
// eth_gap: how long the gap before the next Start is on a 4-lane XGMII, where Start must sit on lane 0
// (46.3.1.4). A gap counts every character from the Terminate to the last idle before Start. A packet is
// 8 + n characters (Start, six preamble octets, SFD, an n-octet frame) and starts on lane 0.
// reset() DIC = 0 (the count is reset only at initialisation)
// method 0 deficit idle count, 1 always insert (both allowed by 46.3.1.4)
// next_gap(n, mac_gap) the gap after an n-octet frame when the MAC asks for mac_gap (12 or more)
// deleted, inserted, dic what the last call did, and the count after it
// Faults: ipg_min_only (always deletes, ignoring the bound), dic_max4 (bounds the count at 4, not 3),
// start_any_lane (no alignment: the gap is always mac_gap).
module eth_gap;
bit ipg_min_only = 0, dic_max4 = 0, start_any_lane = 0;
int method = 0, dic = 0, deleted = 0, inserted = 0;
function automatic void reset(); dic = 0; deleted = 0; inserted = 0; endfunction
function automatic int next_gap(input int n, input int mac_gap);
int m, top;
deleted = 0; inserted = 0;
m = (n + mac_gap) % 4; // lanes past lane 0 the next Start would land on
if (start_any_lane || m == 0) return mac_gap;
top = dic_max4 ? 4 : 3;
if (method == 0 && (ipg_min_only || dic + m <= top)) begin deleted = m; dic = dic + m; end
else begin inserted = 4 - m; if (method == 0) dic = dic - (4 - m); end
return mac_gap - deleted + inserted;
endfunction
endmodule
Check Yourself
Common Mistakes
Wrong Measure the gap from the end of the Terminate character.
On the XGMII the Terminate takes the first character of the gap.
Fix Count the Terminate and the idles that follow it.
Wrong Treat a 9-character gap at 10 Gb/s as a transmit error.
The DIC allows up to three deleted idles, as long as the count stays in bounds.
Fix Judge short gaps by the running count, not one by one.
Wrong Reset the DIC at the start of every frame.
The count must carry the deficit from one gap to the next. It is reset only at start-up.
Fix Keep it as long-lived state in the RS.
Wrong Require 12 octets of gap at the receive MAC.
Gaps shrink on the way: down to 5 octets at 10 Gb/s, and 1 octet from 40 Gb/s.
Fix Accept the shortest gaps of 4.4.2, notes 4 and 7.
Remember
- The minimum gap is 96 bit times, 12 octets, at every rate.
- On the XGMII, Start may only use lane 0, so the RS bends the gap.
- It may always insert idles, or keep a deficit idle count from 0 to 3.
- With the DIC one gap may be 9 characters, but the average stays 12.
- Receivers must accept gaps shrunk to 40 bit times at 10 Gb/s and 8 from 40 Gb/s.