fix(ssa): Validate field to integer cast

[vezenovm]

Description

Problem*

Resolves #8087

Builds upon #8798 as I was originally worried about merge conflicts for building off of #8765.

Summary*

This simply adds validation for Field -> integer casting. Casting involving signed integer requires a much more involved validation check.

Additional Context

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]

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'ACVM Benchmarks'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: ec007e6	Previous: 9d44ab7	Ratio
perfectly_parallel_batch_inversion_opcodes	3573493 ns/iter (± 1553)	2778522 ns/iter (± 7503)	1.29

This comment was automatically generated by workflow using github-action-benchmark.

CC: @TomAFrench

[github-actions]

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Execution Memory'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: 1625b68	Previous: 6f1b46f	Ratio
private-kernel-inner	292.98 MB	202.09 MB	1.45

This comment was automatically generated by workflow using github-action-benchmark.

CC: @TomAFrench

[github-actions]

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Execution Time'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: c8a7b93	Previous: 9580a12	Ratio
rollup-merge	0.004 s	0.003 s	1.33

This comment was automatically generated by workflow using github-action-benchmark.

CC: @TomAFrench

[github-actions]

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Test Suite Duration'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: c8a7b93	Previous: 9580a12	Ratio
test_report_zkpassport_noir_rsa_	2 s	1 s	2

This comment was automatically generated by workflow using github-action-benchmark.

CC: @TomAFrench

[github-actions]

ACVM Benchmarks

Details

Benchmark suite	Current: bd45216	Previous: 12bf0db	Ratio
purely_sequential_opcodes	247842 ns/iter (± 520)	254138 ns/iter (± 1529)	0.98
perfectly_parallel_opcodes	221797 ns/iter (± 2635)	225984 ns/iter (± 1376)	0.98
perfectly_parallel_batch_inversion_opcodes	2784994 ns/iter (± 2173)	2791262 ns/iter (± 10832)	1.00

This comment was automatically generated by workflow using github-action-benchmark.

[github-actions]

Execution Time

Details

Benchmark suite	Current: bd45216	Previous: 12bf0db	Ratio
private-kernel-inner	0.027 s	0.027 s	1
private-kernel-reset	0.158 s	0.155 s	1.02
private-kernel-tail	0.011 s	0.011 s	1
rollup-base-private	0.29 s	0.293 s	0.99
rollup-base-public	0.186 s	0.191 s	0.97
rollup-block-root	10.8 s	10.7 s	1.01
rollup-merge	0.004 s	0.004 s	1
rollup-root	0.009 s	0.009 s	1
semaphore-depth-10	0.02 s	0.02 s	1

This comment was automatically generated by workflow using github-action-benchmark.

[github-actions]

Artifact Size

Details

Benchmark suite	Current: bd45216	Previous: 12bf0db	Ratio
private-kernel-inner	1134.4 KB	1134.4 KB	1
private-kernel-reset	2071.4 KB	2071.4 KB	1
private-kernel-tail	588.5 KB	588.5 KB	1
rollup-base-private	4944.6 KB	4944.6 KB	1
rollup-base-public	3986.2 KB	3986.2 KB	1
rollup-block-root-empty	240.3 KB	240.3 KB	1
rollup-block-root-single-tx	25691.2 KB	25691.2 KB	1
rollup-block-root	25717.7 KB	25717.7 KB	1
rollup-merge	184.7 KB	184.7 KB	1
rollup-root	420.8 KB	420.8 KB	1
semaphore-depth-10	636.4 KB	636.4 KB	1

This comment was automatically generated by workflow using github-action-benchmark.

[github-actions]

Compilation Time

Details

Benchmark suite	Current: bd45216	Previous: 12bf0db	Ratio
private-kernel-inner	2.386 s	2.354 s	1.01
private-kernel-reset	7.364 s	7.536 s	0.98
private-kernel-tail	1.09 s	1.108 s	0.98
rollup-base-private	15.52 s	16.14 s	0.96
rollup-base-public	12.98 s	13.28 s	0.98
rollup-block-root-empty	1.194 s	1.332 s	0.90
rollup-block-root-single-tx	124 s	127 s	0.98
rollup-block-root	131 s	127 s	1.03
rollup-merge	1.104 s	1.094 s	1.01
rollup-root	1.652 s	1.65 s	1.00
semaphore-depth-10	0.794 s	0.755 s	1.05

This comment was automatically generated by workflow using github-action-benchmark.

[github-actions]

Test Suite Duration

Details

Benchmark suite	Current: bd45216	Previous: 12bf0db	Ratio
test_report_AztecProtocol_aztec-packages_noir-projects_aztec-nr	68 s	70 s	0.97
test_report_AztecProtocol_aztec-packages_noir-projects_noir-contracts	109 s	110 s	0.99
test_report_AztecProtocol_aztec-packages_noir-projects_noir-protocol-circuits_crates_blob	43 s	42 s	1.02
test_report_AztecProtocol_aztec-packages_noir-projects_noir-protocol-circuits_crates_private-kernel-lib	210 s	203 s	1.03
test_report_AztecProtocol_aztec-packages_noir-projects_noir-protocol-circuits_crates_rollup-lib	263 s	272 s	0.97
test_report_AztecProtocol_aztec-packages_noir-projects_noir-protocol-circuits_crates_types	70 s	75 s	0.93
test_report_noir-lang_noir-bignum_	421 s	372 s	1.13
test_report_noir-lang_noir_bigcurve_	244 s	219 s	1.11
test_report_noir-lang_sha512_	31 s	28 s	1.11
test_report_zkpassport_noir_rsa_	1 s	1 s	1

This comment was automatically generated by workflow using github-action-benchmark.

[github-actions]

Compilation Memory

Details

Benchmark suite	Current: bd45216	Previous: 12bf0db	Ratio
private-kernel-inner	292.98 MB	292.94 MB	1.00
private-kernel-reset	534.02 MB	534.02 MB	1
private-kernel-tail	191.54 MB	191.56 MB	1.00
rollup-base-private	1380 MB	1380 MB	1
rollup-base-public	1510 MB	1510 MB	1
rollup-block-root-empty	339.39 MB	339.39 MB	1
rollup-block-root-single-tx	7830 MB	7830 MB	1
rollup-block-root	7830 MB	7830 MB	1
rollup-merge	327.16 MB	327.17 MB	1.00
rollup-root	381.46 MB	381.44 MB	1.00
semaphore_depth_10	106.39 MB	106.39 MB	1

This comment was automatically generated by workflow using github-action-benchmark.

[github-actions]

Execution Memory

Details

Benchmark suite	Current: bd45216	Previous: 12bf0db	Ratio
private-kernel-inner	202.09 MB	202.09 MB	1
private-kernel-reset	225.81 MB	225.81 MB	1
private-kernel-tail	177.35 MB	177.35 MB	1
rollup-base-private	489.89 MB	489.89 MB	1
rollup-base-public	423.19 MB	423.19 MB	1
rollup-block-root	1400 MB	1400 MB	1
rollup-merge	312.02 MB	312.02 MB	1
rollup-root	317.7 MB	317.7 MB	1
semaphore_depth_10	70.96 MB	70.96 MB	1

This comment was automatically generated by workflow using github-action-benchmark.

[guipublic]

I think the added truncates in remove_bit_shifts can be avoided.

[vezenovm]

I think the added truncates in remove_bit_shifts can be avoided.

I ultimately removed all of them @guipublic.

I have moved to only running validation after initial SSA gen for the time being. It used to live in FunctionBuilder::finish which was also used by the inlining passes. Making validation more generally aware of value ranges is overly complex for this PR and should be done as part of an entirely separate value range analysis work. The validator can then reference the map constructed by a separate value range analysis to determine whether a Cast is safe. As the validator is currently written it can only be run on the initial SSA (which in itself required adding the truncate simplification invariants).

[guipublic]

Could you also fix the wrong 'cast' in remove_bit_shits.rs?
Line 133, remove the casting: let pow = self.insert_cast(pow, typ);, and return a pow of type 'Field' instead, and implement the casting after:
if max_bit <= bit_size then do the casting here, we know it is safe
'else': the result is already 'cast' after the truncate, so there is nothing to do. (just remove the casting of 'pow' to Field, as it is already a Field now).

[vezenovm]

(just remove the casting of 'pow' to Field, as it is already a Field now).

I also ended up casting the overflow predicate to a Field so that the predicate * pow multiplication had matching types.

[guipublic]

LGTM