feat(brillig): Handle Brillig overflow checks inside the VM

[vezenovm]

Description

Problem*

Part of a general effort to reduce Brillig bytecode size and execution traces

Summary*

This PR simply removes the add_overflow_check from brillig_gen and leaves it up to the VM to implement their own overflow checks if that is what is desired. Handling for the overflow checks done during Brillig gen was included in the Brillig VM.

Additional Context

In the following flamegraphs I search for jump and jump if opcodes. The total samples went down from 126k to 99k samples.

sha256_regression on master (126k samples, 29.5% of the program execution is jumps and jump ifs):

sha256_regression after this PR (99k samples, 25.9% of the program execution is jumps and jump ifs on a much more reduced trace):

Documentation*

Check one:

• No documentation needed.
• Documentation included in this PR.
• [For Experimental Features] Documentation to be submitted in a separate PR.

PR Checklist*

• I have tested the changes locally.
• I have formatted the changes with Prettier and/or cargo fmt on default settings.

[github-actions]

Changes to Brillig bytecode sizes

Generated at commit: 802678e6f1ba9f3615e192e1e8147124fbda1c0f, compared to commit: b473d99b2b70b595596b8392617256dbaf5d5642

🧾 Summary (10% most significant diffs)

Program	Brillig opcodes (+/-)	%
loop	-16 ✅	-21.33%
brillig_cow_regression	-524 ✅	-21.70%
6_array	-96 ✅	-21.92%
6	-308 ✅	-21.92%
sha256_var_witness_const_regression	-320 ✅	-22.19%
sha256	-476 ✅	-22.55%
brillig_acir_as_brillig	-8 ✅	-23.53%
brillig_calls	-8 ✅	-23.53%
brillig_loop	-24 ✅	-23.53%
2_div	-12 ✅	-25.00%
3_add	-12 ✅	-25.53%
simple_bitwise	-16 ✅	-28.57%
1_mul	-32 ✅	-51.61%

---

Full diff report 👇

Program	Brillig opcodes (+/-)	%
regression_4709	133,725 (-8)	-0.01%
slice_dynamic_index	2,840 (-29)	-1.01%
brillig_slices	532 (-8)	-1.48%
modulus	1,745 (-32)	-1.80%
slices	2,255 (-49)	-2.13%
nested_dyn_array_regression_5782	141 (-4)	-2.76%
simple_2d_array	136 (-4)	-2.86%
brillig_blake2s	135 (-4)	-2.88%
blake3	135 (-4)	-2.88%
7	135 (-4)	-2.88%
brillig_blake3	135 (-4)	-2.88%
brillig_cow_assign	130 (-4)	-2.99%
databus_two_calldata	259 (-8)	-3.00%
brillig_constant_reference_regression	128 (-4)	-3.03%
to_bytes_consistent	126 (-4)	-3.08%
strings	1,007 (-32)	-3.08%
signed_arithmetic	247 (-8)	-3.14%
to_le_bytes	123 (-4)	-3.15%
regression_4202	114 (-4)	-3.39%
databus_two_calldata_simple	113 (-4)	-3.42%
fold_call_witness_condition	110 (-4)	-3.51%
array_to_slice	983 (-36)	-3.53%
array_if_cond_simple	103 (-4)	-3.74%
bigint	2,160 (-84)	-3.74%
binary_operator_overloading	405 (-16)	-3.80%
signed_cmp	93 (-4)	-4.12%
array_neq	92 (-4)	-4.17%
array_eq	90 (-4)	-4.26%
poseidon2	328 (-16)	-4.65%
bench_2_to_17	321 (-16)	-4.75%
poseidon_bn254_hash	5,410 (-280)	-4.92%
poseidon_bn254_hash_width_3	5,410 (-280)	-4.92%
slice_regex	2,449 (-128)	-4.97%
fold_numeric_generic_poseidon	729 (-40)	-5.20%
no_predicates_numeric_generic_poseidon	729 (-40)	-5.20%
array_sort	290 (-16)	-5.23%
struct_inputs	288 (-16)	-5.26%
brillig_loop_size_regression	70 (-4)	-5.41%
fold_2_to_17	555 (-32)	-5.45%
check_large_field_bits	271 (-16)	-5.57%
wildcard_type	269 (-16)	-5.61%
brillig_oracle	334 (-20)	-5.65%
to_be_bytes	194 (-12)	-5.83%
eddsa	10,502 (-660)	-5.91%
regression_5252	4,582 (-296)	-6.07%
poseidonsponge_x5_254	4,224 (-276)	-6.13%
side_effects_constrain_array	115 (-8)	-6.50%
databus_composite_calldata	511 (-36)	-6.58%
brillig_rc_regression_6123	161 (-12)	-6.94%
array_dynamic	304 (-24)	-7.32%
7_function	591 (-48)	-7.51%
brillig_cow	392 (-32)	-7.55%
higher_order_functions	683 (-56)	-7.58%
databus_in_fn_with_empty_arr	48 (-4)	-7.69%
regression	664 (-56)	-7.78%
slice_loop	324 (-28)	-7.95%
to_bits	182 (-16)	-8.08%
pedersen_commitment	181 (-16)	-8.12%
references	268 (-24)	-8.22%
databus	87 (-8)	-8.42%
global_consts	209 (-20)	-8.73%
generics	192 (-20)	-9.43%
regression_struct_array_conditional	494 (-52)	-9.52%
pedersen_hash	322 (-36)	-10.06%
simple_shield	782 (-88)	-10.11%
conditional_regression_547	35 (-4)	-10.26%
merkle_insert	698 (-80)	-10.28%
brillig_hash_to_field	138 (-16)	-10.39%
hash_to_field	138 (-16)	-10.39%
import	130 (-16)	-10.96%
schnorr	1,300 (-160)	-10.96%
cast_and_shift_global	63 (-8)	-11.27%
reference_only_used_as_alias	220 (-28)	-11.29%
aes128_encrypt	493 (-64)	-11.49%
signed_div	154 (-20)	-11.49%
tuple_inputs	333 (-44)	-11.67%
pedersen_check	657 (-88)	-11.81%
brillig_pedersen	657 (-88)	-11.81%
fold_complex_outputs	498 (-68)	-12.01%
fold_distinct_return	27 (-4)	-12.90%
break_and_continue	53 (-8)	-13.11%
u128	2,533 (-384)	-13.16%
uhashmap	14,268 (-2,340)	-14.09%
ecdsa_secp256k1	876 (-156)	-15.12%
conditional_regression_661	110 (-20)	-15.38%
conditional_1	1,142 (-208)	-15.41%
acir_inside_brillig_recursion	65 (-12)	-15.58%
brillig_keccak	1,486 (-284)	-16.05%
keccak256	1,486 (-284)	-16.05%
sha2_byte	2,807 (-540)	-16.13%
brillig_nested_arrays	144 (-28)	-16.28%
regression_3607	41 (-8)	-16.33%
sha256_regression	6,191 (-1,220)	-16.46%
fold_fibonacci	60 (-12)	-16.67%
brillig_recursion	60 (-12)	-16.67%
array_dynamic_nested_blackbox_input	831 (-176)	-17.48%
hashmap	22,331 (-4,752)	-17.55%
nested_array_in_slice	1,108 (-236)	-17.56%
array_dynamic_blackbox_input	981 (-212)	-17.77%
brillig_sha256	659 (-156)	-19.14%
regression_4449	702 (-176)	-20.05%
ram_blowup_regression	818 (-212)	-20.58%
brillig_calls_array	77 (-20)	-20.62%
conditional_regression_short_circuit	1,167 (-308)	-20.88%
sha256_var_padding_regression	4,242 (-1,132)	-21.06%
sha256_var_size_regression	1,645 (-444)	-21.25%
nested_array_dynamic	2,008 (-544)	-21.32%
loop	59 (-16)	-21.33%
brillig_cow_regression	1,891 (-524)	-21.70%
6_array	342 (-96)	-21.92%
6	1,097 (-308)	-21.92%
sha256_var_witness_const_regression	1,122 (-320)	-22.19%
sha256	1,635 (-476)	-22.55%
brillig_acir_as_brillig	26 (-8)	-23.53%
brillig_calls	26 (-8)	-23.53%
brillig_loop	78 (-24)	-23.53%
2_div	36 (-12)	-25.00%
3_add	35 (-12)	-25.53%
simple_bitwise	40 (-16)	-28.57%
1_mul	30 (-32)	-51.61%

[github-actions]

Changes to number of Brillig opcodes executed

Generated at commit: 802678e6f1ba9f3615e192e1e8147124fbda1c0f, compared to commit: b473d99b2b70b595596b8392617256dbaf5d5642

🧾 Summary (10% most significant diffs)

Program	Brillig opcodes (+/-)	%
regression_4449	-37,584 ✅	-17.09%
array_dynamic_blackbox_input	-3,322 ✅	-17.14%
simple_bitwise	-8 ✅	-17.39%
sha256_var_padding_regression	-43,284 ✅	-18.42%
sha256_var_size_regression	-3,776 ✅	-20.57%
2_div	-8 ✅	-21.05%
sha256_regression	-26,967 ✅	-21.34%
3_add	-8 ✅	-21.62%
brillig_cow_regression	-133,777 ✅	-22.92%
ram_blowup_regression	-203,013 ✅	-23.13%
loop	-32 ✅	-23.36%
brillig_loop	-48 ✅	-24.49%
1_mul	-24 ✅	-48.00%

---

Full diff report 👇

Program	Brillig opcodes (+/-)	%
bench_2_to_17	144,664,522 (-26,138)	-0.02%
fold_2_to_17	131,420,107 (-49,716)	-0.04%
slice_dynamic_index	4,299 (-21)	-0.49%
array_if_cond_simple	303 (-2)	-0.66%
databus_two_calldata_simple	163 (-2)	-1.21%
brillig_slices	709 (-10)	-1.39%
nested_dyn_array_regression_5782	124 (-2)	-1.59%
slices	3,820 (-62)	-1.60%
signed_arithmetic	199 (-4)	-1.97%
array_to_slice	2,308 (-50)	-2.12%
brillig_constant_reference_regression	178 (-4)	-2.20%
simple_2d_array	176 (-4)	-2.22%
databus_two_calldata	479 (-12)	-2.44%
signed_cmp	77 (-2)	-2.53%
slice_regex	3,749 (-100)	-2.60%
databus_composite_calldata	740 (-22)	-2.89%
array_sort	530 (-16)	-2.93%
brillig_loop_size_regression	66 (-2)	-2.94%
databus	119 (-4)	-3.25%
brillig_rc_regression_6123	272 (-10)	-3.55%
poseidon2	682 (-26)	-3.67%
brillig_cow_assign	512 (-20)	-3.76%
regression_4202	204 (-8)	-3.77%
regression_struct_array_conditional	1,241 (-50)	-3.87%
fold_numeric_generic_poseidon	5,048 (-212)	-4.03%
no_predicates_numeric_generic_poseidon	5,048 (-212)	-4.03%
struct_inputs	622 (-28)	-4.31%
array_dynamic	477 (-22)	-4.41%
strings	1,804 (-84)	-4.45%
fold_call_witness_condition	85 (-4)	-4.49%
slice_loop	1,218 (-62)	-4.84%
databus_in_fn_with_empty_arr	39 (-2)	-4.88%
brillig_cow	1,218 (-64)	-4.99%
binary_operator_overloading	187 (-10)	-5.08%
conditional_regression_661	112 (-6)	-5.08%
array_neq	1,186 (-64)	-5.12%
array_eq	1,184 (-64)	-5.13%
to_le_bytes	1,119 (-62)	-5.25%
higher_order_functions	1,438 (-80)	-5.27%
conditional_regression_547	33 (-2)	-5.71%
regression_3607	33 (-2)	-5.71%
7	1,040 (-64)	-5.80%
blake3	1,040 (-64)	-5.80%
brillig_blake2s	1,040 (-64)	-5.80%
brillig_blake3	1,040 (-64)	-5.80%
wildcard_type	646 (-40)	-5.83%
u128	24,549 (-1,608)	-6.15%
reference_only_used_as_alias	229 (-16)	-6.53%
uhashmap	145,074 (-10,146)	-6.54%
import	134 (-10)	-6.94%
pedersen_commitment	258 (-20)	-7.19%
7_function	2,608 (-204)	-7.25%
to_be_bytes	2,293 (-186)	-7.50%
generics	220 (-18)	-7.56%
brillig_oracle	2,416 (-200)	-7.65%
pedersen_hash	531 (-44)	-7.65%
global_consts	1,635 (-136)	-7.68%
brillig_pedersen	1,047 (-88)	-7.75%
pedersen_check	1,047 (-88)	-7.75%
fold_distinct_return	23 (-2)	-8.00%
simple_shield	2,371 (-208)	-8.07%
merkle_insert	3,080 (-276)	-8.22%
to_bytes_consistent	690 (-62)	-8.24%
regression	2,466 (-222)	-8.26%
tuple_inputs	583 (-54)	-8.48%
side_effects_constrain_array	64 (-6)	-8.57%
break_and_continue	101 (-10)	-9.01%
references	383 (-38)	-9.03%
aes128_encrypt	4,166 (-416)	-9.08%
fold_complex_outputs	778 (-78)	-9.11%
eddsa	670,438 (-67,460)	-9.14%
acir_inside_brillig_recursion	237 (-24)	-9.20%
modulus	17,930 (-1,844)	-9.33%
schnorr	9,204 (-980)	-9.62%
brillig_hash_to_field	857 (-98)	-10.26%
hash_to_field	857 (-98)	-10.26%
nested_array_dynamic	2,957 (-344)	-10.42%
nested_array_in_slice	1,369 (-162)	-10.58%
brillig_calls_array	83 (-10)	-10.75%
brillig_keccak	32,773 (-4,294)	-11.58%
keccak256	32,773 (-4,294)	-11.58%
brillig_recursion	3,999 (-528)	-11.66%
fold_fibonacci	3,999 (-528)	-11.66%
array_dynamic_nested_blackbox_input	3,965 (-543)	-12.05%
check_large_field_bits	2,400 (-338)	-12.34%
brillig_nested_arrays	126 (-18)	-12.50%
conditional_1	5,113 (-732)	-12.52%
poseidonsponge_x5_254	166,999 (-23,918)	-12.53%
regression_5252	831,420 (-119,826)	-12.60%
poseidon_bn254_hash	147,854 (-21,470)	-12.68%
poseidon_bn254_hash_width_3	147,854 (-21,470)	-12.68%
brillig_sha256	3,659 (-550)	-13.07%
to_bits	532 (-80)	-13.07%
sha2_byte	41,025 (-6,178)	-13.09%
hashmap	49,770 (-7,619)	-13.28%
ecdsa_secp256k1	6,102 (-982)	-13.86%
signed_div	1,112 (-180)	-13.93%
sha256	9,276 (-1,582)	-14.57%
sha256_var_witness_const_regression	6,009 (-1,032)	-14.66%
cast_and_shift_global	174 (-30)	-14.71%
conditional_regression_short_circuit	6,318 (-1,132)	-15.19%
6	6,243 (-1,132)	-15.35%
brillig_acir_as_brillig	22 (-4)	-15.38%
brillig_calls	22 (-4)	-15.38%
6_array	1,531 (-314)	-17.02%
regression_4449	182,318 (-37,584)	-17.09%
array_dynamic_blackbox_input	16,064 (-3,322)	-17.14%
simple_bitwise	38 (-8)	-17.39%
sha256_var_padding_regression	191,657 (-43,284)	-18.42%
sha256_var_size_regression	14,578 (-3,776)	-20.57%
2_div	30 (-8)	-21.05%
sha256_regression	99,386 (-26,967)	-21.34%
3_add	29 (-8)	-21.62%
brillig_cow_regression	449,996 (-133,777)	-22.92%
ram_blowup_regression	674,649 (-203,013)	-23.13%
loop	105 (-32)	-23.36%
brillig_loop	148 (-48)	-24.49%
1_mul	26 (-24)	-48.00%