| 1 | const std = @import("../../std.zig"); |
| 2 | const mem = std.mem; |
| 3 | const debug = std.debug; |
| 4 | |
| 5 | /// A single AES block. |
| 6 | pub const Block = struct { |
| 7 | const Repr = @Vector(2, u64); |
| 8 | |
| 9 | pub const block_length: usize = 16; |
| 10 | |
| 11 | /// Internal representation of a block. |
| 12 | repr: Repr, |
| 13 | |
| 14 | /// Convert a byte sequence into an internal representation. |
| 15 | pub fn fromBytes(bytes: *const [16]u8) Block { |
| 16 | const repr = mem.bytesToValue(Repr, bytes); |
| 17 | return Block{ .repr = repr }; |
| 18 | } |
| 19 | |
| 20 | /// Convert the internal representation of a block into a byte sequence. |
| 21 | pub fn toBytes(block: Block) [16]u8 { |
| 22 | return mem.toBytes(block.repr); |
| 23 | } |
| 24 | |
| 25 | /// XOR the block with a byte sequence. |
| 26 | pub fn xorBytes(block: Block, bytes: *const [16]u8) [16]u8 { |
| 27 | const x = block.repr ^ fromBytes(bytes).repr; |
| 28 | return mem.toBytes(x); |
| 29 | } |
| 30 | |
| 31 | const zero = @Vector(2, u64){ 0, 0 }; |
| 32 | |
| 33 | /// Encrypt a block with a round key. |
| 34 | pub fn encrypt(block: Block, round_key: Block) Block { |
| 35 | return Block{ |
| 36 | .repr = (asm ( |
| 37 | \\ mov %[out].16b, %[in].16b |
| 38 | \\ aese %[out].16b, %[zero].16b |
| 39 | \\ aesmc %[out].16b, %[out].16b |
| 40 | : [out] "=&x" (-> Repr), |
| 41 | : [in] "x" (block.repr), |
| 42 | [zero] "x" (zero), |
| 43 | )) ^ round_key.repr, |
| 44 | }; |
| 45 | } |
| 46 | |
| 47 | /// Encrypt a block with the last round key. |
| 48 | pub fn encryptLast(block: Block, round_key: Block) Block { |
| 49 | return Block{ |
| 50 | .repr = (asm ( |
| 51 | \\ mov %[out].16b, %[in].16b |
| 52 | \\ aese %[out].16b, %[zero].16b |
| 53 | : [out] "=&x" (-> Repr), |
| 54 | : [in] "x" (block.repr), |
| 55 | [zero] "x" (zero), |
| 56 | )) ^ round_key.repr, |
| 57 | }; |
| 58 | } |
| 59 | |
| 60 | /// Decrypt a block with a round key. |
| 61 | pub fn decrypt(block: Block, inv_round_key: Block) Block { |
| 62 | return Block{ |
| 63 | .repr = (asm ( |
| 64 | \\ mov %[out].16b, %[in].16b |
| 65 | \\ aesd %[out].16b, %[zero].16b |
| 66 | \\ aesimc %[out].16b, %[out].16b |
| 67 | : [out] "=&x" (-> Repr), |
| 68 | : [in] "x" (block.repr), |
| 69 | [zero] "x" (zero), |
| 70 | )) ^ inv_round_key.repr, |
| 71 | }; |
| 72 | } |
| 73 | |
| 74 | /// Decrypt a block with the last round key. |
| 75 | pub fn decryptLast(block: Block, inv_round_key: Block) Block { |
| 76 | return Block{ |
| 77 | .repr = (asm ( |
| 78 | \\ mov %[out].16b, %[in].16b |
| 79 | \\ aesd %[out].16b, %[zero].16b |
| 80 | : [out] "=&x" (-> Repr), |
| 81 | : [in] "x" (block.repr), |
| 82 | [zero] "x" (zero), |
| 83 | )) ^ inv_round_key.repr, |
| 84 | }; |
| 85 | } |
| 86 | |
| 87 | /// Apply the bitwise XOR operation to the content of two blocks. |
| 88 | pub fn xorBlocks(block1: Block, block2: Block) Block { |
| 89 | return Block{ .repr = block1.repr ^ block2.repr }; |
| 90 | } |
| 91 | |
| 92 | /// Apply the bitwise AND operation to the content of two blocks. |
| 93 | pub fn andBlocks(block1: Block, block2: Block) Block { |
| 94 | return Block{ .repr = block1.repr & block2.repr }; |
| 95 | } |
| 96 | |
| 97 | /// Apply the bitwise OR operation to the content of two blocks. |
| 98 | pub fn orBlocks(block1: Block, block2: Block) Block { |
| 99 | return Block{ .repr = block1.repr | block2.repr }; |
| 100 | } |
| 101 | |
| 102 | /// Apply the inverse MixColumns operation to a block. |
| 103 | pub fn invMixColumns(block: Block) Block { |
| 104 | return Block{ |
| 105 | .repr = asm ( |
| 106 | \\ aesimc %[out].16b, %[in].16b |
| 107 | : [out] "=x" (-> Repr), |
| 108 | : [in] "x" (block.repr), |
| 109 | ), |
| 110 | }; |
| 111 | } |
| 112 | |
| 113 | /// Perform operations on multiple blocks in parallel. |
| 114 | pub const parallel = struct { |
| 115 | /// The recommended number of AES encryption/decryption to perform in parallel for the chosen implementation. |
| 116 | pub const optimal_parallel_blocks = 6; |
| 117 | |
| 118 | /// Encrypt multiple blocks in parallel, each their own round key. |
| 119 | pub fn encryptParallel(comptime count: usize, blocks: [count]Block, round_keys: [count]Block) [count]Block { |
| 120 | comptime var i = 0; |
| 121 | var out: [count]Block = undefined; |
| 122 | inline while (i < count) : (i += 1) { |
| 123 | out[i] = blocks[i].encrypt(round_keys[i]); |
| 124 | } |
| 125 | return out; |
| 126 | } |
| 127 | |
| 128 | /// Decrypt multiple blocks in parallel, each their own round key. |
| 129 | pub fn decryptParallel(comptime count: usize, blocks: [count]Block, round_keys: [count]Block) [count]Block { |
| 130 | comptime var i = 0; |
| 131 | var out: [count]Block = undefined; |
| 132 | inline while (i < count) : (i += 1) { |
| 133 | out[i] = blocks[i].decrypt(round_keys[i]); |
| 134 | } |
| 135 | return out; |
| 136 | } |
| 137 | |
| 138 | /// Encrypt multiple blocks in parallel with the same round key. |
| 139 | pub fn encryptWide(comptime count: usize, blocks: [count]Block, round_key: Block) [count]Block { |
| 140 | comptime var i = 0; |
| 141 | var out: [count]Block = undefined; |
| 142 | inline while (i < count) : (i += 1) { |
| 143 | out[i] = blocks[i].encrypt(round_key); |
| 144 | } |
| 145 | return out; |
| 146 | } |
| 147 | |
| 148 | /// Decrypt multiple blocks in parallel with the same round key. |
| 149 | pub fn decryptWide(comptime count: usize, blocks: [count]Block, round_key: Block) [count]Block { |
| 150 | comptime var i = 0; |
| 151 | var out: [count]Block = undefined; |
| 152 | inline while (i < count) : (i += 1) { |
| 153 | out[i] = blocks[i].decrypt(round_key); |
| 154 | } |
| 155 | return out; |
| 156 | } |
| 157 | |
| 158 | /// Encrypt multiple blocks in parallel with the same last round key. |
| 159 | pub fn encryptLastWide(comptime count: usize, blocks: [count]Block, round_key: Block) [count]Block { |
| 160 | comptime var i = 0; |
| 161 | var out: [count]Block = undefined; |
| 162 | inline while (i < count) : (i += 1) { |
| 163 | out[i] = blocks[i].encryptLast(round_key); |
| 164 | } |
| 165 | return out; |
| 166 | } |
| 167 | |
| 168 | /// Decrypt multiple blocks in parallel with the same last round key. |
| 169 | pub fn decryptLastWide(comptime count: usize, blocks: [count]Block, round_key: Block) [count]Block { |
| 170 | comptime var i = 0; |
| 171 | var out: [count]Block = undefined; |
| 172 | inline while (i < count) : (i += 1) { |
| 173 | out[i] = blocks[i].decryptLast(round_key); |
| 174 | } |
| 175 | return out; |
| 176 | } |
| 177 | }; |
| 178 | }; |
| 179 | |
| 180 | /// A fixed-size vector of AES blocks. |
| 181 | /// All operations are performed in parallel, using SIMD instructions when available. |
| 182 | pub fn BlockVec(comptime blocks_count: comptime_int) type { |
| 183 | return struct { |
| 184 | const Self = @This(); |
| 185 | |
| 186 | /// The number of AES blocks the target architecture can process with a single instruction. |
| 187 | pub const native_vector_size = 1; |
| 188 | |
| 189 | /// The size of the AES block vector that the target architecture can process with a single instruction, in bytes. |
| 190 | pub const native_word_size = native_vector_size * 16; |
| 191 | |
| 192 | const native_words = blocks_count; |
| 193 | |
| 194 | /// Internal representation of a block vector. |
| 195 | repr: [native_words]Block, |
| 196 | |
| 197 | /// Length of the block vector in bytes. |
| 198 | pub const block_length: usize = blocks_count * 16; |
| 199 | |
| 200 | /// Convert a byte sequence into an internal representation. |
| 201 | pub fn fromBytes(bytes: *const [blocks_count * 16]u8) Self { |
| 202 | var out: Self = undefined; |
| 203 | inline for (0..native_words) |i| { |
| 204 | out.repr[i] = Block.fromBytes(bytes[i * native_word_size ..][0..native_word_size]); |
| 205 | } |
| 206 | return out; |
| 207 | } |
| 208 | |
| 209 | /// Convert the internal representation of a block vector into a byte sequence. |
| 210 | pub fn toBytes(block_vec: Self) [blocks_count * 16]u8 { |
| 211 | var out: [blocks_count * 16]u8 = undefined; |
| 212 | inline for (0..native_words) |i| { |
| 213 | out[i * native_word_size ..][0..native_word_size].* = block_vec.repr[i].toBytes(); |
| 214 | } |
| 215 | return out; |
| 216 | } |
| 217 | |
| 218 | /// XOR the block vector with a byte sequence. |
| 219 | pub fn xorBytes(block_vec: Self, bytes: *const [blocks_count * 16]u8) [blocks_count * 16]u8 { |
| 220 | var out: [blocks_count * 16]u8 = undefined; |
| 221 | inline for (0..native_words) |i| { |
| 222 | out[i * native_word_size ..][0..native_word_size].* = block_vec.repr[i].xorBytes(bytes[i * native_word_size ..][0..native_word_size]); |
| 223 | } |
| 224 | return out; |
| 225 | } |
| 226 | |
| 227 | /// Apply the forward AES operation to the block vector with a vector of round keys. |
| 228 | pub fn encrypt(block_vec: Self, round_key_vec: Self) Self { |
| 229 | var out: Self = undefined; |
| 230 | inline for (0..native_words) |i| { |
| 231 | out.repr[i] = block_vec.repr[i].encrypt(round_key_vec.repr[i]); |
| 232 | } |
| 233 | return out; |
| 234 | } |
| 235 | |
| 236 | /// Apply the forward AES operation to the block vector with a vector of last round keys. |
| 237 | pub fn encryptLast(block_vec: Self, round_key_vec: Self) Self { |
| 238 | var out: Self = undefined; |
| 239 | inline for (0..native_words) |i| { |
| 240 | out.repr[i] = block_vec.repr[i].encryptLast(round_key_vec.repr[i]); |
| 241 | } |
| 242 | return out; |
| 243 | } |
| 244 | |
| 245 | /// Apply the inverse AES operation to the block vector with a vector of round keys. |
| 246 | pub fn decrypt(block_vec: Self, inv_round_key_vec: Self) Self { |
| 247 | var out: Self = undefined; |
| 248 | inline for (0..native_words) |i| { |
| 249 | out.repr[i] = block_vec.repr[i].decrypt(inv_round_key_vec.repr[i]); |
| 250 | } |
| 251 | return out; |
| 252 | } |
| 253 | |
| 254 | /// Apply the inverse AES operation to the block vector with a vector of last round keys. |
| 255 | pub fn decryptLast(block_vec: Self, inv_round_key_vec: Self) Self { |
| 256 | var out: Self = undefined; |
| 257 | inline for (0..native_words) |i| { |
| 258 | out.repr[i] = block_vec.repr[i].decryptLast(inv_round_key_vec.repr[i]); |
| 259 | } |
| 260 | return out; |
| 261 | } |
| 262 | |
| 263 | /// Apply the bitwise XOR operation to the content of two block vectors. |
| 264 | pub fn xorBlocks(block_vec1: Self, block_vec2: Self) Self { |
| 265 | var out: Self = undefined; |
| 266 | inline for (0..native_words) |i| { |
| 267 | out.repr[i] = block_vec1.repr[i].xorBlocks(block_vec2.repr[i]); |
| 268 | } |
| 269 | return out; |
| 270 | } |
| 271 | |
| 272 | /// Apply the bitwise AND operation to the content of two block vectors. |
| 273 | pub fn andBlocks(block_vec1: Self, block_vec2: Self) Self { |
| 274 | var out: Self = undefined; |
| 275 | inline for (0..native_words) |i| { |
| 276 | out.repr[i] = block_vec1.repr[i].andBlocks(block_vec2.repr[i]); |
| 277 | } |
| 278 | return out; |
| 279 | } |
| 280 | |
| 281 | /// Apply the bitwise OR operation to the content of two block vectors. |
| 282 | pub fn orBlocks(block_vec1: Self, block_vec2: Self) Self { |
| 283 | var out: Self = undefined; |
| 284 | inline for (0..native_words) |i| { |
| 285 | out.repr[i] = block_vec1.repr[i].orBlocks(block_vec2.repr[i]); |
| 286 | } |
| 287 | return out; |
| 288 | } |
| 289 | |
| 290 | /// Apply the inverse MixColumns operation to each block in the vector. |
| 291 | pub fn invMixColumns(block_vec: Self) Self { |
| 292 | var out: Self = undefined; |
| 293 | inline for (0..native_words) |i| { |
| 294 | out.repr[i] = block_vec.repr[i].invMixColumns(); |
| 295 | } |
| 296 | return out; |
| 297 | } |
| 298 | }; |
| 299 | } |
| 300 | |
| 301 | fn KeySchedule(comptime Aes: type) type { |
| 302 | std.debug.assert(Aes.rounds == 10 or Aes.rounds == 14); |
| 303 | const rounds = Aes.rounds; |
| 304 | |
| 305 | return struct { |
| 306 | const Self = @This(); |
| 307 | |
| 308 | const Repr = Aes.block.Repr; |
| 309 | |
| 310 | const zero = @Vector(2, u64){ 0, 0 }; |
| 311 | const mask1 = @Vector(16, u8){ 13, 14, 15, 12, 13, 14, 15, 12, 13, 14, 15, 12, 13, 14, 15, 12 }; |
| 312 | const mask2 = @Vector(16, u8){ 12, 13, 14, 15, 12, 13, 14, 15, 12, 13, 14, 15, 12, 13, 14, 15 }; |
| 313 | |
| 314 | round_keys: [rounds + 1]Block, |
| 315 | |
| 316 | fn drc128(comptime rc: u8, t: Repr) Repr { |
| 317 | var v1: Repr = undefined; |
| 318 | var v2: Repr = undefined; |
| 319 | var v3: Repr = undefined; |
| 320 | var v4: Repr = undefined; |
| 321 | |
| 322 | return asm ( |
| 323 | \\ movi %[v2].4s, %[rc] |
| 324 | \\ tbl %[v4].16b, {%[t].16b}, %[mask].16b |
| 325 | \\ ext %[r].16b, %[zero].16b, %[t].16b, #12 |
| 326 | \\ aese %[v4].16b, %[zero].16b |
| 327 | \\ eor %[v2].16b, %[r].16b, %[v2].16b |
| 328 | \\ ext %[r].16b, %[zero].16b, %[r].16b, #12 |
| 329 | \\ eor %[v1].16b, %[v2].16b, %[t].16b |
| 330 | \\ ext %[v3].16b, %[zero].16b, %[r].16b, #12 |
| 331 | \\ eor %[v1].16b, %[v1].16b, %[r].16b |
| 332 | \\ eor %[r].16b, %[v1].16b, %[v3].16b |
| 333 | \\ eor %[r].16b, %[r].16b, %[v4].16b |
| 334 | : [r] "=&x" (-> Repr), |
| 335 | [v1] "=&x" (v1), |
| 336 | [v2] "=&x" (v2), |
| 337 | [v3] "=&x" (v3), |
| 338 | [v4] "=&x" (v4), |
| 339 | : [rc] "N" (rc), |
| 340 | [t] "x" (t), |
| 341 | [zero] "x" (zero), |
| 342 | [mask] "x" (mask1), |
| 343 | ); |
| 344 | } |
| 345 | |
| 346 | fn drc256(comptime second: bool, comptime rc: u8, t: Repr, tx: Repr) Repr { |
| 347 | var v1: Repr = undefined; |
| 348 | var v2: Repr = undefined; |
| 349 | var v3: Repr = undefined; |
| 350 | var v4: Repr = undefined; |
| 351 | |
| 352 | return asm ( |
| 353 | \\ movi %[v2].4s, %[rc] |
| 354 | \\ tbl %[v4].16b, {%[t].16b}, %[mask].16b |
| 355 | \\ ext %[r].16b, %[zero].16b, %[tx].16b, #12 |
| 356 | \\ aese %[v4].16b, %[zero].16b |
| 357 | \\ eor %[v1].16b, %[tx].16b, %[r].16b |
| 358 | \\ ext %[r].16b, %[zero].16b, %[r].16b, #12 |
| 359 | \\ eor %[v1].16b, %[v1].16b, %[r].16b |
| 360 | \\ ext %[v3].16b, %[zero].16b, %[r].16b, #12 |
| 361 | \\ eor %[v1].16b, %[v1].16b, %[v2].16b |
| 362 | \\ eor %[v1].16b, %[v1].16b, %[v3].16b |
| 363 | \\ eor %[r].16b, %[v1].16b, %[v4].16b |
| 364 | : [r] "=&x" (-> Repr), |
| 365 | [v1] "=&x" (v1), |
| 366 | [v2] "=&x" (v2), |
| 367 | [v3] "=&x" (v3), |
| 368 | [v4] "=&x" (v4), |
| 369 | : [rc] "N" (if (second) @as(u8, 0) else rc), |
| 370 | [t] "x" (t), |
| 371 | [tx] "x" (tx), |
| 372 | [zero] "x" (zero), |
| 373 | [mask] "x" (if (second) mask2 else mask1), |
| 374 | ); |
| 375 | } |
| 376 | |
| 377 | fn expand128(t1: *Block) Self { |
| 378 | var round_keys: [11]Block = undefined; |
| 379 | const rcs = [_]u8{ 1, 2, 4, 8, 16, 32, 64, 128, 27, 54 }; |
| 380 | inline for (rcs, 0..) |rc, round| { |
| 381 | round_keys[round] = t1.*; |
| 382 | t1.repr = drc128(rc, t1.repr); |
| 383 | } |
| 384 | round_keys[rcs.len] = t1.*; |
| 385 | return Self{ .round_keys = round_keys }; |
| 386 | } |
| 387 | |
| 388 | fn expand256(t1: *Block, t2: *Block) Self { |
| 389 | var round_keys: [15]Block = undefined; |
| 390 | const rcs = [_]u8{ 1, 2, 4, 8, 16, 32 }; |
| 391 | round_keys[0] = t1.*; |
| 392 | inline for (rcs, 0..) |rc, round| { |
| 393 | round_keys[round * 2 + 1] = t2.*; |
| 394 | t1.repr = drc256(false, rc, t2.repr, t1.repr); |
| 395 | round_keys[round * 2 + 2] = t1.*; |
| 396 | t2.repr = drc256(true, rc, t1.repr, t2.repr); |
| 397 | } |
| 398 | round_keys[rcs.len * 2 + 1] = t2.*; |
| 399 | t1.repr = drc256(false, 64, t2.repr, t1.repr); |
| 400 | round_keys[rcs.len * 2 + 2] = t1.*; |
| 401 | return Self{ .round_keys = round_keys }; |
| 402 | } |
| 403 | |
| 404 | /// Invert the key schedule. |
| 405 | pub fn invert(key_schedule: Self) Self { |
| 406 | const round_keys = &key_schedule.round_keys; |
| 407 | var inv_round_keys: [rounds + 1]Block = undefined; |
| 408 | inv_round_keys[0] = round_keys[rounds]; |
| 409 | comptime var i = 1; |
| 410 | inline while (i < rounds) : (i += 1) { |
| 411 | inv_round_keys[i] = Block{ |
| 412 | .repr = asm ( |
| 413 | \\ aesimc %[inv_rk].16b, %[rk].16b |
| 414 | : [inv_rk] "=x" (-> Repr), |
| 415 | : [rk] "x" (round_keys[rounds - i].repr), |
| 416 | ), |
| 417 | }; |
| 418 | } |
| 419 | inv_round_keys[rounds] = round_keys[0]; |
| 420 | return Self{ .round_keys = inv_round_keys }; |
| 421 | } |
| 422 | }; |
| 423 | } |
| 424 | |
| 425 | /// A context to perform encryption using the standard AES key schedule. |
| 426 | pub fn AesEncryptCtx(comptime Aes: type) type { |
| 427 | std.debug.assert(Aes.key_bits == 128 or Aes.key_bits == 256); |
| 428 | const rounds = Aes.rounds; |
| 429 | |
| 430 | return struct { |
| 431 | const Self = @This(); |
| 432 | pub const block = Aes.block; |
| 433 | pub const block_length = block.block_length; |
| 434 | key_schedule: KeySchedule(Aes), |
| 435 | |
| 436 | /// Create a new encryption context with the given key. |
| 437 | pub fn init(key: [Aes.key_bits / 8]u8) Self { |
| 438 | var t1 = Block.fromBytes(key[0..16]); |
| 439 | const key_schedule = if (Aes.key_bits == 128) ks: { |
| 440 | break :ks KeySchedule(Aes).expand128(&t1); |
| 441 | } else ks: { |
| 442 | var t2 = Block.fromBytes(key[16..32]); |
| 443 | break :ks KeySchedule(Aes).expand256(&t1, &t2); |
| 444 | }; |
| 445 | return Self{ |
| 446 | .key_schedule = key_schedule, |
| 447 | }; |
| 448 | } |
| 449 | |
| 450 | /// Encrypt a single block. |
| 451 | pub fn encrypt(ctx: Self, dst: *[16]u8, src: *const [16]u8) void { |
| 452 | const round_keys = ctx.key_schedule.round_keys; |
| 453 | var t = Block.fromBytes(src).xorBlocks(round_keys[0]); |
| 454 | comptime var i = 1; |
| 455 | inline while (i < rounds) : (i += 1) { |
| 456 | t = t.encrypt(round_keys[i]); |
| 457 | } |
| 458 | t = t.encryptLast(round_keys[rounds]); |
| 459 | dst.* = t.toBytes(); |
| 460 | } |
| 461 | |
| 462 | /// Encrypt+XOR a single block. |
| 463 | pub fn xor(ctx: Self, dst: *[16]u8, src: *const [16]u8, counter: [16]u8) void { |
| 464 | const round_keys = ctx.key_schedule.round_keys; |
| 465 | var t = Block.fromBytes(&counter).xorBlocks(round_keys[0]); |
| 466 | comptime var i = 1; |
| 467 | inline while (i < rounds) : (i += 1) { |
| 468 | t = t.encrypt(round_keys[i]); |
| 469 | } |
| 470 | t = t.encryptLast(round_keys[rounds]); |
| 471 | dst.* = t.xorBytes(src); |
| 472 | } |
| 473 | |
| 474 | /// Encrypt multiple blocks, possibly leveraging parallelization. |
| 475 | pub fn encryptWide(ctx: Self, comptime count: usize, dst: *[16 * count]u8, src: *const [16 * count]u8) void { |
| 476 | const round_keys = ctx.key_schedule.round_keys; |
| 477 | var ts: [count]Block = undefined; |
| 478 | comptime var j = 0; |
| 479 | inline while (j < count) : (j += 1) { |
| 480 | ts[j] = Block.fromBytes(src[j * 16 .. j * 16 + 16][0..16]).xorBlocks(round_keys[0]); |
| 481 | } |
| 482 | comptime var i = 1; |
| 483 | inline while (i < rounds) : (i += 1) { |
| 484 | ts = Block.parallel.encryptWide(count, ts, round_keys[i]); |
| 485 | } |
| 486 | ts = Block.parallel.encryptLastWide(count, ts, round_keys[i]); |
| 487 | j = 0; |
| 488 | inline while (j < count) : (j += 1) { |
| 489 | dst[16 * j .. 16 * j + 16].* = ts[j].toBytes(); |
| 490 | } |
| 491 | } |
| 492 | |
| 493 | /// Encrypt+XOR multiple blocks, possibly leveraging parallelization. |
| 494 | pub fn xorWide(ctx: Self, comptime count: usize, dst: *[16 * count]u8, src: *const [16 * count]u8, counters: [16 * count]u8) void { |
| 495 | const round_keys = ctx.key_schedule.round_keys; |
| 496 | var ts: [count]Block = undefined; |
| 497 | comptime var j = 0; |
| 498 | inline while (j < count) : (j += 1) { |
| 499 | ts[j] = Block.fromBytes(counters[j * 16 .. j * 16 + 16][0..16]).xorBlocks(round_keys[0]); |
| 500 | } |
| 501 | comptime var i = 1; |
| 502 | inline while (i < rounds) : (i += 1) { |
| 503 | ts = Block.parallel.encryptWide(count, ts, round_keys[i]); |
| 504 | } |
| 505 | ts = Block.parallel.encryptLastWide(count, ts, round_keys[i]); |
| 506 | j = 0; |
| 507 | inline while (j < count) : (j += 1) { |
| 508 | dst[16 * j .. 16 * j + 16].* = ts[j].xorBytes(src[16 * j .. 16 * j + 16]); |
| 509 | } |
| 510 | } |
| 511 | }; |
| 512 | } |
| 513 | |
| 514 | /// A context to perform decryption using the standard AES key schedule. |
| 515 | pub fn AesDecryptCtx(comptime Aes: type) type { |
| 516 | std.debug.assert(Aes.key_bits == 128 or Aes.key_bits == 256); |
| 517 | const rounds = Aes.rounds; |
| 518 | |
| 519 | return struct { |
| 520 | const Self = @This(); |
| 521 | pub const block = Aes.block; |
| 522 | pub const block_length = block.block_length; |
| 523 | key_schedule: KeySchedule(Aes), |
| 524 | |
| 525 | /// Create a decryption context from an existing encryption context. |
| 526 | pub fn initFromEnc(ctx: AesEncryptCtx(Aes)) Self { |
| 527 | return Self{ |
| 528 | .key_schedule = ctx.key_schedule.invert(), |
| 529 | }; |
| 530 | } |
| 531 | |
| 532 | /// Create a new decryption context with the given key. |
| 533 | pub fn init(key: [Aes.key_bits / 8]u8) Self { |
| 534 | const enc_ctx = AesEncryptCtx(Aes).init(key); |
| 535 | return initFromEnc(enc_ctx); |
| 536 | } |
| 537 | |
| 538 | /// Decrypt a single block. |
| 539 | pub fn decrypt(ctx: Self, dst: *[16]u8, src: *const [16]u8) void { |
| 540 | const inv_round_keys = ctx.key_schedule.round_keys; |
| 541 | var t = Block.fromBytes(src).xorBlocks(inv_round_keys[0]); |
| 542 | comptime var i = 1; |
| 543 | inline while (i < rounds) : (i += 1) { |
| 544 | t = t.decrypt(inv_round_keys[i]); |
| 545 | } |
| 546 | t = t.decryptLast(inv_round_keys[rounds]); |
| 547 | dst.* = t.toBytes(); |
| 548 | } |
| 549 | |
| 550 | /// Decrypt multiple blocks, possibly leveraging parallelization. |
| 551 | pub fn decryptWide(ctx: Self, comptime count: usize, dst: *[16 * count]u8, src: *const [16 * count]u8) void { |
| 552 | const inv_round_keys = ctx.key_schedule.round_keys; |
| 553 | var ts: [count]Block = undefined; |
| 554 | comptime var j = 0; |
| 555 | inline while (j < count) : (j += 1) { |
| 556 | ts[j] = Block.fromBytes(src[j * 16 .. j * 16 + 16][0..16]).xorBlocks(inv_round_keys[0]); |
| 557 | } |
| 558 | comptime var i = 1; |
| 559 | inline while (i < rounds) : (i += 1) { |
| 560 | ts = Block.parallel.decryptWide(count, ts, inv_round_keys[i]); |
| 561 | } |
| 562 | ts = Block.parallel.decryptLastWide(count, ts, inv_round_keys[i]); |
| 563 | j = 0; |
| 564 | inline while (j < count) : (j += 1) { |
| 565 | dst[16 * j .. 16 * j + 16].* = ts[j].toBytes(); |
| 566 | } |
| 567 | } |
| 568 | }; |
| 569 | } |
| 570 | |
| 571 | /// AES-128 with the standard key schedule. |
| 572 | pub const Aes128 = struct { |
| 573 | pub const key_bits: usize = 128; |
| 574 | pub const rounds = ((key_bits - 64) / 32 + 8); |
| 575 | pub const block = Block; |
| 576 | |
| 577 | /// Create a new context for encryption. |
| 578 | pub fn initEnc(key: [key_bits / 8]u8) AesEncryptCtx(Aes128) { |
| 579 | return AesEncryptCtx(Aes128).init(key); |
| 580 | } |
| 581 | |
| 582 | /// Create a new context for decryption. |
| 583 | pub fn initDec(key: [key_bits / 8]u8) AesDecryptCtx(Aes128) { |
| 584 | return AesDecryptCtx(Aes128).init(key); |
| 585 | } |
| 586 | }; |
| 587 | |
| 588 | /// AES-256 with the standard key schedule. |
| 589 | pub const Aes256 = struct { |
| 590 | pub const key_bits: usize = 256; |
| 591 | pub const rounds = ((key_bits - 64) / 32 + 8); |
| 592 | pub const block = Block; |
| 593 | |
| 594 | /// Create a new context for encryption. |
| 595 | pub fn initEnc(key: [key_bits / 8]u8) AesEncryptCtx(Aes256) { |
| 596 | return AesEncryptCtx(Aes256).init(key); |
| 597 | } |
| 598 | |
| 599 | /// Create a new context for decryption. |
| 600 | pub fn initDec(key: [key_bits / 8]u8) AesDecryptCtx(Aes256) { |
| 601 | return AesDecryptCtx(Aes256).init(key); |
| 602 | } |
| 603 | }; |