src/math/random/chacha8/block.bend source
src/math/random/chacha8/block.bend on the hub · documented module
# Generated by tools/generators/chacha8rand_gen.py; do not edit.# The ChaCha8 core of ChaCha8Rand (C2SP chacha8rand; Go's# internal/chacha8rand block_generic). One double round is fully unrolled# (four column and four diagonal quarter rounds, RFC 8439 section 2.1-2.3,# constant rotations); a block is four double rounds (eight rounds). As# C2SP specifies, only the key words 4..11 are added back after the rounds:# the constants and the counter are not (the standard ChaCha output minus# them). A group is four consecutive blocks interleaved word by word, read# as 32 little-endian 64-bit words: word 2w of the group is word w of the# first block (low half) and of the second (high half), word 2w + 1 the# same for the third and fourth block.## Proved equal to spec/math/random/chacha8rand.bend for every key and# counter (proofs/math/random/chacha8/).import Baseimport ../../u64.bend as W# the 256-bit key as eight little-endian wordstype Key is Data: K{k0: U32, k1: U32, k2: U32, k3: U32, k4: U32, k5: U32, k6: U32, k7: U32}# the ChaCha state x[0..15]type Vector is Data: V{v0: U32, v1: U32, v2: U32, v3: U32, v4: U32, v5: U32, v6: U32, v7: U32, v8: U32, v9: U32, v10: U32, v11: U32, v12: U32, v13: U32, v14: U32, v15: U32}# the initial state: constants, key, 32-bit block counter, zero noncedef init(k: Key, +ctr: U32) -> Vector: match k: case K{k0,k1,k2,k3,k4,k5,k6,k7}: V{1634760805,857760878,2036477234,1797285236,k0,k1,k2,k3,k4,k5,k6,k7,ctr,0,0,0}# one double round: four column quarter rounds, then four diagonal onesdef dr(v: Vector) -> Vector: match v: case V{+a0,+a1,+a2,+a3,+a4,+a5,+a6,+a7,+a8,+a9,+a10,+a11,+a12,+a13,+a14,+a15}: +q0a1 = U32.add(a0,a4) +q0d1x = U32.xor(a12,q0a1) +q0d1 = U32.or(U32.shln(q0d1x,16n),U32.shrn(q0d1x,16n)) +q0c1 = U32.add(a8,q0d1) +q0b1x = U32.xor(a4,q0c1) +q0b1 = U32.or(U32.shln(q0b1x,12n),U32.shrn(q0b1x,20n)) +q0a2 = U32.add(q0a1,q0b1) +q0d2x = U32.xor(q0d1,q0a2) +q0d2 = U32.or(U32.shln(q0d2x,8n),U32.shrn(q0d2x,24n)) +q0c2 = U32.add(q0c1,q0d2) +q0b2x = U32.xor(q0b1,q0c2) +q0b2 = U32.or(U32.shln(q0b2x,7n),U32.shrn(q0b2x,25n)) +q1a1 = U32.add(a1,a5) +q1d1x = U32.xor(a13,q1a1) +q1d1 = U32.or(U32.shln(q1d1x,16n),U32.shrn(q1d1x,16n)) +q1c1 = U32.add(a9,q1d1) +q1b1x = U32.xor(a5,q1c1) +q1b1 = U32.or(U32.shln(q1b1x,12n),U32.shrn(q1b1x,20n)) +q1a2 = U32.add(q1a1,q1b1) +q1d2x = U32.xor(q1d1,q1a2) +q1d2 = U32.or(U32.shln(q1d2x,8n),U32.shrn(q1d2x,24n)) +q1c2 = U32.add(q1c1,q1d2) +q1b2x = U32.xor(q1b1,q1c2) +q1b2 = U32.or(U32.shln(q1b2x,7n),U32.shrn(q1b2x,25n)) +q2a1 = U32.add(a2,a6) +q2d1x = U32.xor(a14,q2a1) +q2d1 = U32.or(U32.shln(q2d1x,16n),U32.shrn(q2d1x,16n)) +q2c1 = U32.add(a10,q2d1) +q2b1x = U32.xor(a6,q2c1) +q2b1 = U32.or(U32.shln(q2b1x,12n),U32.shrn(q2b1x,20n)) +q2a2 = U32.add(q2a1,q2b1) +q2d2x = U32.xor(q2d1,q2a2) +q2d2 = U32.or(U32.shln(q2d2x,8n),U32.shrn(q2d2x,24n)) +q2c2 = U32.add(q2c1,q2d2) +q2b2x = U32.xor(q2b1,q2c2) +q2b2 = U32.or(U32.shln(q2b2x,7n),U32.shrn(q2b2x,25n)) +q3a1 = U32.add(a3,a7) +q3d1x = U32.xor(a15,q3a1) +q3d1 = U32.or(U32.shln(q3d1x,16n),U32.shrn(q3d1x,16n)) +q3c1 = U32.add(a11,q3d1) +q3b1x = U32.xor(a7,q3c1) +q3b1 = U32.or(U32.shln(q3b1x,12n),U32.shrn(q3b1x,20n)) +q3a2 = U32.add(q3a1,q3b1) +q3d2x = U32.xor(q3d1,q3a2) +q3d2 = U32.or(U32.shln(q3d2x,8n),U32.shrn(q3d2x,24n)) +q3c2 = U32.add(q3c1,q3d2) +q3b2x = U32.xor(q3b1,q3c2) +q3b2 = U32.or(U32.shln(q3b2x,7n),U32.shrn(q3b2x,25n)) +q4a1 = U32.add(q0a2,q1b2) +q4d1x = U32.xor(q3d2,q4a1) +q4d1 = U32.or(U32.shln(q4d1x,16n),U32.shrn(q4d1x,16n)) +q4c1 = U32.add(q2c2,q4d1) +q4b1x = U32.xor(q1b2,q4c1) +q4b1 = U32.or(U32.shln(q4b1x,12n),U32.shrn(q4b1x,20n)) +q4a2 = U32.add(q4a1,q4b1) +q4d2x = U32.xor(q4d1,q4a2) +q4d2 = U32.or(U32.shln(q4d2x,8n),U32.shrn(q4d2x,24n)) +q4c2 = U32.add(q4c1,q4d2) +q4b2x = U32.xor(q4b1,q4c2) +q4b2 = U32.or(U32.shln(q4b2x,7n),U32.shrn(q4b2x,25n)) +q5a1 = U32.add(q1a2,q2b2) +q5d1x = U32.xor(q0d2,q5a1) +q5d1 = U32.or(U32.shln(q5d1x,16n),U32.shrn(q5d1x,16n)) +q5c1 = U32.add(q3c2,q5d1) +q5b1x = U32.xor(q2b2,q5c1) +q5b1 = U32.or(U32.shln(q5b1x,12n),U32.shrn(q5b1x,20n)) +q5a2 = U32.add(q5a1,q5b1) +q5d2x = U32.xor(q5d1,q5a2) +q5d2 = U32.or(U32.shln(q5d2x,8n),U32.shrn(q5d2x,24n)) +q5c2 = U32.add(q5c1,q5d2) +q5b2x = U32.xor(q5b1,q5c2) +q5b2 = U32.or(U32.shln(q5b2x,7n),U32.shrn(q5b2x,25n)) +q6a1 = U32.add(q2a2,q3b2) +q6d1x = U32.xor(q1d2,q6a1) +q6d1 = U32.or(U32.shln(q6d1x,16n),U32.shrn(q6d1x,16n)) +q6c1 = U32.add(q0c2,q6d1) +q6b1x = U32.xor(q3b2,q6c1) +q6b1 = U32.or(U32.shln(q6b1x,12n),U32.shrn(q6b1x,20n)) +q6a2 = U32.add(q6a1,q6b1) +q6d2x = U32.xor(q6d1,q6a2) +q6d2 = U32.or(U32.shln(q6d2x,8n),U32.shrn(q6d2x,24n)) +q6c2 = U32.add(q6c1,q6d2) +q6b2x = U32.xor(q6b1,q6c2) +q6b2 = U32.or(U32.shln(q6b2x,7n),U32.shrn(q6b2x,25n)) +q7a1 = U32.add(q3a2,q0b2) +q7d1x = U32.xor(q2d2,q7a1) +q7d1 = U32.or(U32.shln(q7d1x,16n),U32.shrn(q7d1x,16n)) +q7c1 = U32.add(q1c2,q7d1) +q7b1x = U32.xor(q0b2,q7c1) +q7b1 = U32.or(U32.shln(q7b1x,12n),U32.shrn(q7b1x,20n)) +q7a2 = U32.add(q7a1,q7b1) +q7d2x = U32.xor(q7d1,q7a2) +q7d2 = U32.or(U32.shln(q7d2x,8n),U32.shrn(q7d2x,24n)) +q7c2 = U32.add(q7c1,q7d2) +q7b2x = U32.xor(q7b1,q7c2) +q7b2 = U32.or(U32.shln(q7b2x,7n),U32.shrn(q7b2x,25n)) V{q4a2,q5a2,q6a2,q7a2,q7b2,q4b2,q5b2,q6b2,q6c2,q7c2,q4c2,q5c2,q5d2,q6d2,q7d2,q4d2}# the key words added back to words 4..11def finish(k: Key, v: Vector) -> Vector: match k v: case K{k0,k1,k2,k3,k4,k5,k6,k7} V{v0,v1,v2,v3,v4,v5,v6,v7,v8,v9,v10,v11,v12,v13,v14,v15}: V{v0,v1,v2,v3,U32.add(v4,k0),U32.add(v5,k1),U32.add(v6,k2),U32.add(v7,k3),U32.add(v8,k4),U32.add(v9,k5),U32.add(v10,k6),U32.add(v11,k7),v12,v13,v14,v15}# the ChaCha8Rand block with counter ctr: eight roundsdef block(+k: Key, +ctr: U32) -> Vector: finish(k,dr(dr(dr(dr(init(k,ctr))))))# four blocks interleaved word by word, as 32 little-endian 64-bit wordsdef interleave(a: Vector, b: Vector, c: Vector, d: Vector) -> List<&2,W.U64>: match a b c d: case V{a0,a1,a2,a3,a4,a5,a6,a7,a8,a9,a10,a11,a12,a13,a14,a15} V{b0,b1,b2,b3,b4,b5,b6,b7,b8,b9,b10,b11,b12,b13,b14,b15} V{c0,c1,c2,c3,c4,c5,c6,c7,c8,c9,c10,c11,c12,c13,c14,c15} V{d0,d1,d2,d3,d4,d5,d6,d7,d8,d9,d10,d11,d12,d13,d14,d15}: [W.U64{a0,b0},W.U64{c0,d0},W.U64{a1,b1},W.U64{c1,d1},W.U64{a2,b2},W.U64{c2,d2},W.U64{a3,b3},W.U64{c3,d3},W.U64{a4,b4},W.U64{c4,d4},W.U64{a5,b5},W.U64{c5,d5},W.U64{a6,b6},W.U64{c6,d6},W.U64{a7,b7},W.U64{c7,d7},W.U64{a8,b8},W.U64{c8,d8},W.U64{a9,b9},W.U64{c9,d9},W.U64{a10,b10},W.U64{c10,d10},W.U64{a11,b11},W.U64{c11,d11},W.U64{a12,b12},W.U64{c12,d12},W.U64{a13,b13},W.U64{c13,d13},W.U64{a14,b14},W.U64{c14,d14},W.U64{a15,b15},W.U64{c15,d15}]# the group of the blocks ctr, ctr + 1, ctr + 2, ctr + 3def group(+k: Key, +ctr: U32) -> List<&2,W.U64>: interleave(block(k,ctr),block(k,U32.add(ctr,1)),block(k,U32.add(ctr,2)),block(k,U32.add(ctr,3)))