Ohadn/program segment as ppt

[ohad-nir-starkware]

This change is 

---

Note

Medium Risk
Touches prover/verifier commitment inputs and preprocessed-trace sizing/ids, so any mismatch in program_in_ppt handling can break proof generation/verification. Uses a global program table which adds concurrency/statefulness risk if misused across proofs.

Overview
Adds optional inclusion of the Cairo program segment into the preprocessed trace by introducing new program preprocessed columns and wiring a program_in_ppt flag through ProverParameters and both proof structs.

When program_in_ppt is enabled, the prover/verifier build a PreProcessedTrace::new_with_program(...) (using a global program table) so the preprocessed commitment/log sizes/column ids match the extra columns; helper APIs for generating preprocessed commitment roots now accept optional program data. Minor follow-ups update serialization destructuring and test input data accordingly.

^{Written by Cursor Bugbot for commit 2406cde. This will update automatically on new commits. Configure here.}

[anatgstarkware]

@anatgstarkware made 1 comment.
Reviewable status: 0 of 16 files reviewed, 1 unresolved discussion (waiting on ohad-nir-starkware).

---

stwo_cairo_prover/crates/cairo-air/src/air.rs line 68 at r2 (raw file):

    pub channel_salt: u32,
    pub preprocessed_trace_variant: PreProcessedTraceVariant,
    pub program: Vec<(u32, [u32; 8])>,

Why did you add it to the proof?
It's already in the claim
but I think we can remove it from there now

Code quote:

pub program: Vec<(u32, [u32; 8])>,

[ohad-nir-starkware]

@ohad-nir-starkware made 1 comment.
Reviewable status: 0 of 16 files reviewed, 1 unresolved discussion (waiting on anatgstarkware and ohad-nir-starkware).

---

stwo_cairo_prover/crates/cairo-air/src/air.rs line 68 at r2 (raw file):

Previously, anatgstarkware (anatg) wrote…

Why did you add it to the proof?
It's already in the claim
but I think we can remove it from there now

Done.

[anatgstarkware]

@anatgstarkware partially reviewed 7 files and made 2 comments.
Reviewable status: 2 of 16 files reviewed, 2 unresolved discussions (waiting on ohad-nir-starkware).

---

stwo_cairo_prover/crates/cairo-air/src/air.rs line 68 at r2 (raw file):

Previously, ohad-nir-starkware (Ohad Nir) wrote…

Done.

I think the program shouldn't be in the proof at all (currently it's in the public data), and the verifier should know a specific ppt with a specific program

---

stwo_cairo_prover/crates/common/src/preprocessed_columns/program.rs line 12 at r6 (raw file):

/// Extracts a 9-bit limb from a 256-bit value stored as `[u32; 8]` little-endian limbs.
fn extract_9bit_limb(value: &[u32; 8], limb_index: usize) -> u32 {

I think it's better to create all columns once using get_limbs from prover_typers/cpu.rs and return the correct value in packed_at (i.e. create a ProgramTable struct)

Code quote:

fn extract_9bit_limb(value: &[u32; 8], limb_index: usize) -> u32 {

[ohad-nir-starkware]

@ohad-nir-starkware made 2 comments.
Reviewable status: 2 of 16 files reviewed, 2 unresolved discussions (waiting on anatgstarkware and ohad-nir-starkware).

---

stwo_cairo_prover/crates/cairo-air/src/air.rs line 68 at r2 (raw file):

Previously, anatgstarkware (anatg) wrote…

I think the program shouldn't be in the proof at all (currently it's in the public data), and the verifier should know a specific ppt with a specific program

maybe we should do it in a future PR?

---

stwo_cairo_prover/crates/common/src/preprocessed_columns/program.rs line 12 at r6 (raw file):

Previously, anatgstarkware (anatg) wrote…

I think it's better to create all columns once using get_limbs from prover_typers/cpu.rs and return the correct value in packed_at (i.e. create a ProgramTable struct)

Done.

[ohad-nir-starkware]

@ohad-nir-starkware made 1 comment.
Reviewable status: 2 of 20 files reviewed, 2 unresolved discussions (waiting on anatgstarkware and ohad-nir-starkware).

---

stwo_cairo_prover/crates/cairo-air/src/air.rs line 68 at r2 (raw file):

Previously, ohad-nir-starkware (Ohad Nir) wrote…

maybe we should do it in a future PR?

I implemented some "middle of the road" approach, let me know what you think.
It will make more sense once we remove the program from the proof, which I think could be the next step.

[anatgstarkware]

@anatgstarkware reviewed 9 files and all commit messages, and made 1 comment.
Reviewable status: 9 of 20 files reviewed, 5 unresolved discussions (waiting on ohad-nir-starkware).

---

stwo_cairo_prover/crates/cairo-air/src/air.rs line 68 at r2 (raw file):

Previously, ohad-nir-starkware (Ohad Nir) wrote…

I implemented some "middle of the road" approach, let me know what you think.
It will make more sense once we remove the program from the proof, which I think could be the next step.

I think we shouldn't add the program_in_ppt bool to the proof

[ohad-nir-starkware]

@ohad-nir-starkware made 1 comment and resolved 1 discussion.
Reviewable status: 9 of 20 files reviewed, 4 unresolved discussions (waiting on anatgstarkware).

---

stwo_cairo_prover/crates/cairo-air/src/air.rs line 68 at r2 (raw file):

Previously, anatgstarkware (anatg) wrote…

I think we shouldn't add the program_in_ppt bool to the proof

but preprocessed_trace_variant is in the proof, how else do we pass that information?

[ohad-nir-starkware]

@ohad-nir-starkware made 1 comment and resolved 1 discussion.
Reviewable status: 6 of 20 files reviewed, 2 unresolved discussions (waiting on anatgstarkware).

[ohad-nir-starkware]

@ohad-nir-starkware made 1 comment and resolved 1 discussion.
Reviewable status: 6 of 20 files reviewed, 2 unresolved discussions (waiting on anatgstarkware).

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, have a team admin enable autofix in the Cursor dashboard.}

[cursor]

Global mutable state creates race condition between callers

Medium Severity

PROGRAM_TABLE is a global LazyLock<RwLock<...>> that's written by set_program_table and then read by ProgramColumn::new inside new_with_program. These two steps are not atomic as a unit — if two threads concurrently call new_with_program with different program data, one thread's set_program_table can overwrite the other's data before the other thread's ProgramColumn::new reads it, silently producing columns with the wrong program. Passing the program data directly into ProgramColumn::new (or holding the lock across both operations) would eliminate this hidden coupling.

Additional Locations (1)

• stwo_cairo_prover/crates/common/src/preprocessed_columns/preprocessed_trace.rs#L223-L233

 

[ohad-nir-starkware]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• pad the ppt program columns to allow lengths that arent powers of 2 #1736
• remove the program from PublicMemory when program_in_ppt=true #1725
• Ohadn/verify program builtin without extra component #1724
• deploy the verify_program builtin and pad the program segment #1719
• Ohadn/program segment as ppt #1672 👈 (View in Graphite)
• compute program_segment in the adapter and add it to ProverInput (#1671) #1740
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.