triton_vm/example_programs.rs
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759
use isa::program::Program;
use isa::triton_program;
use lazy_static::lazy_static;
lazy_static! {
pub static ref FIBONACCI_SEQUENCE: Program = fibonacci_sequence();
pub static ref GREATEST_COMMON_DIVISOR: Program = greatest_common_divisor();
pub static ref PROGRAM_WITH_MANY_U32_INSTRUCTIONS: Program =
program_with_many_u32_instructions();
pub static ref VERIFY_SUDOKU: Program = verify_sudoku();
pub static ref CALCULATE_NEW_MMR_PEAKS_FROM_APPEND_WITH_SAFE_LISTS: Program =
calculate_new_mmr_peaks_from_append_with_safe_lists();
pub static ref MERKLE_TREE_AUTHENTICATION_PATH_VERIFY: Program =
merkle_tree_authentication_path_verify();
pub static ref MERKLE_TREE_UPDATE: Program = merkle_tree_update();
}
fn fibonacci_sequence() -> Program {
triton_program!(
// initialize stack: ⊥ 0 1 i
push 0
push 1
read_io 1
// is any looping necessary?
dup 0
skiz
call fib_loop
// pop zero, write result
pop 1
write_io 1
halt
// before: ⊥ 0 1 i
// after: ⊥ fib(i-1) fib(i) 0
fib_loop:
push -1 // ⊥ a b j -1
add // ⊥ a b (j-1)
swap 2 // ⊥ (j-1) b a
dup 1 // ⊥ (j-1) b a b
add // ⊥ (j-1) b (a+b)
swap 1 // ⊥ (j-1) (a+b) b
swap 2 // ⊥ b (a+b) (j-1)
dup 0 // ⊥ b (a+b) (j-1) (j-1)
skiz // ⊥ b (a+b) (j-1)
recurse
return
)
}
fn greatest_common_divisor() -> Program {
triton_program!(
read_io 2 // _ a b
dup 1 // _ a b a
dup 1 // _ a b a b
lt // _ a b b<a
skiz // _ a b
swap 1 // _ d n where n > d
loop_cond:
dup 1
push 0
eq
skiz
call terminate // _ d n where d != 0
dup 1 // _ d n d
dup 1 // _ d n d n
div_mod // _ d n q r
swap 2 // _ d r q n
pop 2 // _ d r
swap 1 // _ r d
call loop_cond
terminate:
// _ d n where d == 0
write_io 1 // _ d
halt
)
}
fn program_with_many_u32_instructions() -> Program {
triton_program!(
push 1311768464867721216 split
push 13387 push 78810 lt
push 5 push 7 pow
push 69584 push 6796 xor
push 64972 push 3915 and
push 98668 push 15787 div_mod
push 15787 push 98668 div_mod
push 98141 push 7397 and
push 67749 push 60797 lt
push 49528 split
push 53483 call lsb
push 79655 call is_u32
push 60615 log_2_floor
push 13 push 5 pow
push 86323 push 37607 xor
push 32374 push 20636 pow
push 97416 log_2_floor
push 14392 push 31589 div_mod
halt
lsb:
push 2 swap 1 div_mod return
is_u32:
split pop 1 push 0 eq return
)
}
/// Triton program to verify Merkle authentication paths.
/// - input: merkle root, number of leafs, leaf values, APs
/// - output: Result<(), VMFail>
fn merkle_tree_authentication_path_verify() -> Program {
triton_program!(
read_io 1 // number of authentication paths to test
// stack: [num]
mt_ap_verify: // proper program starts here
push 0 write_mem 1 pop 1 // store number of APs at RAM address 0
// stack: []
read_io 5 // read Merkle root
// stack: [r4 r3 r2 r1 r0]
call check_aps
pop 5 // leave clean stack: Merkle root
// stack: []
halt // done – should be “return”
// subroutine: check AP one at a time
// stack before: [* r4 r3 r2 r1 r0]
// stack after: [* r4 r3 r2 r1 r0]
check_aps:
push 0 read_mem 1 pop 1 dup 0 // get number of APs left to check
// stack: [* r4 r3 r2 r1 r0 num_left num_left]
push 0 eq // see if there are authentication paths left
// stack: [* r4 r3 r2 r1 r0 0 num_left num_left==0]
skiz return // return if no authentication paths left
push -1 add // decrease number of authentication paths left to check
// stack: [* r4 r3 r2 r1 r0 num_left-1]
push 0 write_mem 1 pop 1 // write decreased number to address 0
// stack: [* r4 r3 r2 r1 r0]
call get_idx_and_leaf
// stack: [* r4 r3 r2 r1 r0 idx l4 l3 l2 l1 l0]
call traverse_tree
// stack: [* r4 r3 r2 r1 r0 1 d4 d3 d2 d1 d0]
call assert_tree_top
// stack: [* r4 r3 r2 r1 r0]
recurse // check next AP
// subroutine: read index & hash leaf
// stack before: [*]
// stack after: [* idx l4 l3 l2 l1 l0]
get_idx_and_leaf:
read_io 1 // read node index
read_io 5 // read leaf's value
return
// subroutine: go up tree
// stack before: [* r4 r3 r2 r1 r0 idx l4 l3 l2 l1 l0]
// stack after: [* r4 r3 r2 r1 r0 1 d4 d3 d2 d1 d0]
traverse_tree:
dup 5 push 1 eq skiz return // break loop if node index is 1
merkle_step recurse // move up one level in the Merkle tree
// subroutine: compare digests
// stack before: [* r4 r3 r2 r1 r0 1 d4 d3 d2 d1 d0]
// stack after: [* r4 r3 r2 r1 r0]
assert_tree_top:
// stack: [* r4 r3 r2 r1 r0 1 d4 d3 d2 d1 d0]
swap 1 swap 2 swap 3 swap 4 swap 5
// stack: [* r4 r3 r2 r1 r0 d4 d3 d2 d1 d0 1]
assert // ensure the entire path was traversed
// stack: [* r4 r3 r2 r1 r0 d4 d3 d2 d1 d0]
assert_vector // actually compare to root of tree
return
)
}
/// Triton program to verifiably change a Merkle tree's leaf. That is:
/// 1. verify that the supplied `old_leaf` is indeed a leaf in the Merkle tree
/// defined by the `merkle_root` and the `tree_height`,
/// 2. update the leaf at the specified `leaf_index` with the `new_leaf`, and
/// 3. return the new Merkle root.
///
/// The authentication path for the leaf to update has to be supplied via RAM.
///
/// - input:
/// - RAM address of leaf's authentication path
/// - leaf index to update
/// - Merkle tree's height
/// - old leaf
/// - (current) merkle root
/// - new leaf
/// - output:
/// - new root
fn merkle_tree_update() -> Program {
triton_program! {
read_io 3 // _ *ap leaf_index tree_height
push 2 pow add // _ *ap node_index
dup 1 push 1 dup 2 // _ *ap node_index *ap 1 node_index
read_io 5 // _ *ap node_index *ap 1 node_index [old_leaf; 5]
call compute_root // _ *ap node_index *ap' 1 1 [root; 5]
read_io 5 // _ *ap node_index *ap' 1 1 [root; 5] [presumed_root; 5]
assert_vector // _ *ap node_index *ap' 1 1 [root; 5]
pop 5 pop 3 // _ *ap node_index
push 1 swap 1 // _ *ap 1 node_index
read_io 5 // _ *ap 1 node_index [new_leaf; 5]
call compute_root // _ *ap' 1 1 [new_root; 5]
write_io 5 // _ *ap' 1 1
pop 3 halt // _
// BEFORE: _ *ap 1 node_index [leaf; 5]
// AFTER: _ (*ap + 5 * tree_height) 1 1 [root; 5]
compute_root:
merkle_step_mem
recurse_or_return
}
}
fn verify_sudoku() -> Program {
// RAM layout:
// 0..=8: primes for mapping digits 1..=9
// 9: flag for whether the Sudoku is valid
// 10..=90: the Sudoku grid
//
// 10 11 12 13 14 15 16 17 18
// 19 20 21 22 23 24 25 26 27
// 28 29 30 31 32 33 34 35 36
//
// 37 38 39 40 41 42 43 44 45
// 46 47 48 49 50 51 52 53 54
// 55 56 57 58 59 60 61 62 63
//
// 64 65 66 67 68 69 70 71 72
// 73 74 75 76 77 78 79 80 81
// 82 83 84 85 86 87 88 89 90
triton_program!(
call initialize_flag
call initialize_primes
call read_sudoku
call write_sudoku_and_check_rows
call check_columns
call check_squares
call assert_flag
// For checking whether the Sudoku is valid. Initially `true`, set to `false` if any
// inconsistency is found.
initialize_flag:
push 1 // _ 1
push 0 // _ 1 0
write_mem 1 // _ 1
pop 1 // _
return
invalidate_flag:
push 0 // _ 0
push 0 // _ 0 0
write_mem 1 // _ 1
pop 1 // _
return
assert_flag:
push 0 // _ 0
read_mem 1 // _ flag -1
pop 1 // _ flag
assert // _
halt
// For mapping legal Sudoku digits to distinct primes. Helps with checking consistency of
// rows, columns, and boxes.
initialize_primes:
push 23 push 19 push 17
push 13 push 11 push 7
push 5 push 3 push 2
push 1 write_mem 5 write_mem 4
pop 1
return
read_sudoku:
call read9 call read9 call read9
call read9 call read9 call read9
call read9 call read9 call read9
return
read9:
call read1 call read1 call read1
call read1 call read1 call read1
call read1 call read1 call read1
return
// Applies the mapping from legal Sudoku digits to distinct primes.
read1: // _
read_io 1 // _ d
read_mem 1 // _ p d-1
pop 1 // _ p
return
write_sudoku_and_check_rows: // row0 row1 row2 row3 row4 row5 row6 row7 row8
push 10 // row0 row1 row2 row3 row4 row5 row6 row7 row8 10
call write_and_check_one_row // row0 row1 row2 row3 row4 row5 row6 row7 19
call write_and_check_one_row // row0 row1 row2 row3 row4 row5 row6 27
call write_and_check_one_row // row0 row1 row2 row3 row4 row5 36
call write_and_check_one_row // row0 row1 row2 row3 row4 45
call write_and_check_one_row // row0 row1 row2 row3 54
call write_and_check_one_row // row0 row1 row2 63
call write_and_check_one_row // row0 row1 72
call write_and_check_one_row // row0 81
call write_and_check_one_row // 90
pop 1 // ⊥
return
write_and_check_one_row: // row addr
dup 9 dup 9 dup 9
dup 9 dup 9 dup 9
dup 9 dup 9 dup 9 // row addr row
call check_9_numbers // row addr
write_mem 5 write_mem 4 // addr+9
return
check_columns:
push 82 call check_one_column
push 83 call check_one_column
push 84 call check_one_column
push 85 call check_one_column
push 86 call check_one_column
push 87 call check_one_column
push 88 call check_one_column
push 89 call check_one_column
push 90 call check_one_column
return
check_one_column:
read_mem 1 push -8 add read_mem 1 push -8 add read_mem 1 push -8 add
read_mem 1 push -8 add read_mem 1 push -8 add read_mem 1 push -8 add
read_mem 1 push -8 add read_mem 1 push -8 add read_mem 1 pop 1
call check_9_numbers
return
check_squares:
push 30 call check_one_square
push 33 call check_one_square
push 36 call check_one_square
push 57 call check_one_square
push 60 call check_one_square
push 63 call check_one_square
push 84 call check_one_square
push 87 call check_one_square
push 90 call check_one_square
return
check_one_square:
read_mem 3 push -6 add
read_mem 3 push -6 add
read_mem 3 pop 1
call check_9_numbers
return
check_9_numbers:
mul mul mul
mul mul mul
mul mul
// 223092870 = 2·3·5·7·11·13·17·19·23
push 223092870 eq
skiz return
call invalidate_flag
return
)
}
pub(crate) fn calculate_new_mmr_peaks_from_append_with_safe_lists() -> Program {
triton_program!(
// Stack and memory setup
push 0 // _ 0
push 3 // _ 0 3
push 1 // _ 0 3 1
push 00457470286889025784
push 04071246825597671119
push 17834064596403781463
push 17484910066710486708
push 06700794775299091393 // _ 0 3 1 [digest]
push 06595477061838874830
push 10897391716490043893
push 01807330184488272967
push 05415221245149797169
push 05057320540678713304 // _ 0 3 1 [digest] [digest]
push 01838589939278841373
push 02628975953172153832
push 06845409670928290394
push 00880730500905369322
push 04594396536654736100 // _ 0 3 1 [digest] [digest] [digest]
push 64 // _ 0 3 1 [digest] [digest] [digest] 64
push 2 // _ 0 3 1 [digest] [digest] [digest] 64 2
push 323 // _ 0 3 1 [digest] [digest] [digest] 64 2 323
push 0 // _ 0 3 1 [digest] [digest] [digest] 64 2 323 0
write_mem 3 // _ 0 3 1 [digest] [digest] [digest] 3
write_mem 5 // _ 0 3 1 [digest] [digest] 8
write_mem 5 // _ 0 3 1 [digest] 13
pop 1 // _ 0 3 1 [digest]
call tasm_mmr_calculate_new_peaks_from_append_safe
halt
// Main function
// BEFORE: _ old_leaf_count_hi old_leaf_count_lo *peaks [digest]
// AFTER: _ *new_peaks *auth_path
tasm_mmr_calculate_new_peaks_from_append_safe:
dup 5 dup 5 dup 5 dup 5 dup 5 dup 5
call tasm_list_safe_u32_push_digest
pop 5 // _ old_leaf_count_hi old_leaf_count_lo *peaks
// Create auth_path return value (vector living in RAM)
// All MMR auth paths have capacity for 64 digests
push 64 // _ old_leaf_count_hi old_leaf_count_lo *peaks 64
call tasm_list_safe_u32_new_digest
swap 1
// stack: _ old_leaf_count_hi old_leaf_count_lo *auth_path *peaks
dup 3 dup 3
// stack: _ old_leaf_count_hi old_leaf_count_lo *auth_path *peaks old_leaf_count_hi old_leaf_count_lo
call tasm_arithmetic_u64_incr
call tasm_arithmetic_u64_index_of_last_nonzero_bit
call tasm_mmr_calculate_new_peaks_from_append_safe_while
// stack: _ old_leaf_count_hi old_leaf_count_lo *auth_path *peaks (rll = 0)
pop 1
swap 3 pop 1 swap 1 pop 1
// stack: _ *peaks *auth_path
return
// Stack start and end: _ old_leaf_count_hi old_leaf_count_lo *auth_path *peaks rll
tasm_mmr_calculate_new_peaks_from_append_safe_while:
dup 0
push 0
eq
skiz
return
// Stack: _ old_leaf_count_hi old_leaf_count_lo *auth_path *peaks rll
swap 2 swap 1
// Stack: _ old_leaf_count_hi old_leaf_count_lo rll *auth_path *peaks
dup 0
dup 0
call tasm_list_safe_u32_pop_digest
// Stack: _ old_leaf_count_hi old_leaf_count_lo rll *auth_path *peaks *peaks [digest (new_hash)]
dup 5
// Stack: _ old_leaf_count_hi old_leaf_count_lo rll *auth_path *peaks *peaks [digest (new_hash)] *peaks
call tasm_list_safe_u32_pop_digest
// Stack: _ old_leaf_count_hi old_leaf_count_lo rll *auth_path *peaks *peaks [digest (new_hash)] [digests (previous_peak)]
// Update authentication path with latest previous_peak
dup 12
dup 5 dup 5 dup 5 dup 5 dup 5
// Stack: _ old_leaf_count_hi old_leaf_count_lo rll *auth_path *peaks *peaks [digest (new_hash)] [digests (previous_peak)] *auth_path [digests (previous_peak)]
call tasm_list_safe_u32_push_digest
// Stack: _ old_leaf_count_hi old_leaf_count_lo rll *auth_path *peaks *peaks [digest (new_hash)] [digests (previous_peak)]
hash
// Stack: _ old_leaf_count_hi old_leaf_count_lo rll *auth_path *peaks *peaks [digests (new_peak)]
call tasm_list_safe_u32_push_digest
// Stack: _ old_leaf_count_hi old_leaf_count_lo rll *auth_path *peaks
swap 1 swap 2
// Stack: _ old_leaf_count_hi old_leaf_count_lo *auth_path *peaks rll
push -1
add
// Stack: _ old_leaf_count_hi old_leaf_count_lo *auth_path *peaks (rll - 1)
recurse
// Before: _ value_hi value_lo
// After: _ (value + 1)_hi (value + 1)_lo
tasm_arithmetic_u64_incr_carry:
pop 1
push 1
add
dup 0
push 4294967296
eq
push 0
eq
assert
push 0
return
tasm_arithmetic_u64_incr:
push 1
add
dup 0
push 4294967296
eq
skiz
call tasm_arithmetic_u64_incr_carry
return
// Before: _ *list, elem[4], elem[3], elem[2], elem[1], elem[0]
// After: _
tasm_list_safe_u32_push_digest:
dup 5 // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] *list
push 1 add // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] *list+1
read_mem 2 // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] capacity len *list-1
// Verify that length < capacity
swap 2 // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] *list-1 len capacity
dup 1 // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] *list-1 len capacity len
lt // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] *list-1 len capacity>len
assert // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] *list-1 len
// Adjust ram pointer
push 5 // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] *list-1 len 5
mul // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] *list-1 5·len
add // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] *list+5·len-1
push 3 // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] *list+5·len 3
add // _ *list elem[4] elem[3] elem[2] elem[1] elem[0] *list+5·len+2
// Write all elements
write_mem 5 // _ *list *list+5·len+7
// Remove ram pointer
pop 1 // _ *list
// Increase length indicator by one
read_mem 1 // _ len *list-1
push 1 add // _ len *list
swap 1 // _ *list len
push 1 // _ *list len 1
add // _ *list len+1
swap 1 // _ len+1 *list
write_mem 1 // _ *list+1
pop 1 // _
return
// BEFORE: _ capacity
// AFTER:
tasm_list_safe_u32_new_digest:
// Convert capacity in number of elements to number of VM words required for that list
dup 0 // _ capacity capacity
push 5 // _ capacity capacity 5
mul // _ capacity 5·capacity
// _ capacity capacity_in_bfes
push 2 // _ capacity capacity_in_bfes 2
add // _ capacity capacity_in_bfes+2
// _ capacity words_to_allocate
call tasm_memory_dyn_malloc // _ capacity *list
// Write initial length = 0 to `*list`, and capacity to `*list + 1`
push 0 // _ capacity *list 0
swap 1 // _ capacity 0 *list
write_mem 2 // _ (*list+2)
push -2 // _ (*list+2) -2
add // _ *list
return
tasm_arithmetic_u64_decr:
push -1
add
dup 0
push -1
eq
skiz
call tasm_arithmetic_u64_decr_carry
return
tasm_arithmetic_u64_decr_carry:
pop 1
push -1
add
dup 0
push -1
eq
push 0
eq
assert
push 4294967295
return
// BEFORE: _ value_hi value_lo
// AFTER: _ log2_floor(value)
tasm_arithmetic_u64_log_2_floor:
swap 1
push 1
dup 1
// stack: _ value_lo value_hi 1 value_hi
skiz call tasm_arithmetic_u64_log_2_floor_then
skiz call tasm_arithmetic_u64_log_2_floor_else
// stack: _ log2_floor(value)
return
tasm_arithmetic_u64_log_2_floor_then:
// value_hi != 0
// stack: _ value_lo value_hi 1
swap 1
swap 2
pop 2
// stack: _ value_hi
log_2_floor
push 32
add
// stack: _ (log2_floor(value_hi) + 32)
push 0
// stack: _ (log2_floor(value_hi) + 32) 0
return
tasm_arithmetic_u64_log_2_floor_else:
// value_hi == 0
// stack: _ value_lo value_hi
pop 1
log_2_floor
return
// Before: _ *list
// After: _ elem{N - 1}, elem{N - 2}, ..., elem{0}
tasm_list_safe_u32_pop_digest:
read_mem 1 // _ len *list-1
push 1 add // _ len *list
// Assert that length is not 0
dup 1 // _ len *list len
push 0 // _ len *list len 0
eq // _ len *list len==0
push 0 // _ len *list len==0 0
eq // _ len *list len!=0
assert // _ len *list
// Decrease length value by one and write back to memory
dup 1 // _ len *list len
push -1 // _ len *list len -1
add // _ len *list len-1
swap 1 // _ len len-1 *list
write_mem 1 // _ len *list+1
push -1 add // _ len *list
// Read elements
swap 1 // _ *list len
push 5 // _ *list len 5
mul // _ *list 5·len
// _ *list offset_for_last_element
add // _ *list+offset_for_last_element
// _ address_for_last_element
read_mem 5 // _ [elements] address_for_last_element-5
pop 1 // _ [elements]
return
// BEFORE: rhs_hi rhs_lo lhs_hi lhs_lo
// AFTER: (rhs & lhs)_hi (rhs & lhs)_lo
tasm_arithmetic_u64_and:
swap 3
and
// stack: _ lhs_lo rhs_lo (lhs_hi & rhs_hi)
swap 2
and
// stack: _ (lhs_hi & rhs_hi) (rhs_lo & lhs_lo)
return
// BEFORE: _ value_hi value_lo
// AFTER: _ index_of_last_non-zero_bit
tasm_arithmetic_u64_index_of_last_nonzero_bit:
dup 1
dup 1
// _ value_hi value_lo value_hi value_lo
call tasm_arithmetic_u64_decr
// _ value_hi value_lo (value - 1)_hi (value - 1)_lo
push 4294967295
push 4294967295
// _ value_hi value_lo (value - 1)_hi (value - 1)_lo 0xFFFFFFFF 0xFFFFFFFF
call tasm_arithmetic_u64_xor
// _ value_hi value_lo ~(value - 1)_hi ~(value - 1)_lo
call tasm_arithmetic_u64_and
// _ (value & ~(value - 1))_hi (value & ~(value - 1))_lo
// The above value is now a power of two in u64. Calling log2_floor on this
// value gives us the index we are looking for.
call tasm_arithmetic_u64_log_2_floor
return
// Return a pointer to a free address and allocate `size` words for this pointer
// Before: _ size
// After: _ *next_addr
tasm_memory_dyn_malloc:
push 0 // _ size *free_pointer
read_mem 1 // _ size *next_addr' *free_pointer-1
pop 1 // _ size *next_addr'
// add 1 iff `next_addr` was 0, i.e. uninitialized.
dup 0 // _ size *next_addr' *next_addr'
push 0 // _ size *next_addr' *next_addr' 0
eq // _ size *next_addr' (*next_addr' == 0)
add // _ size *next_addr
dup 0 // _ size *next_addr *next_addr
dup 2 // _ size *next_addr *next_addr size
// Ensure that `size` does not exceed 2^32
split
swap 1
push 0
eq
assert
add // _ size *free_pointer *next_addr *(next_addr + size)
// Ensure that no more than 2^32 words are allocated, because I don't want a wrap-around
// in the address space
split
swap 1
push 0
eq
assert
swap 1 // _ size *(next_addr + size) *next_addr
swap 2 // _ *next_addr *(next_addr + size) size
pop 1 // _ *next_addr *(next_addr + size)
push 0 // _ *next_addr *(next_addr + size) *free_pointer
write_mem 1 // _ *next_addr *free_pointer+1
pop 1 // _ *next_addr
return
// BEFORE: rhs_hi rhs_lo lhs_hi lhs_lo
// AFTER: (rhs ^ lhs)_hi (rhs ^ lhs)_lo
tasm_arithmetic_u64_xor:
swap 3
xor
// stack: _ lhs_lo rhs_lo (lhs_hi ^ rhs_hi)
swap 2
xor
// stack: _ (lhs_hi ^ rhs_hi) (rhs_lo ^ lhs_lo)
return
)
}