feat(avm): trace contract class and contract instance

[IlyasRidhuan]

This PR is centred around tracing and passing contract class & instance during simulator execution and passing it to circuit. We store each contract class & instance whenever the simulator calls getBytecode.

This changes the input interface to the bb binary - we no longer take in a specific bytecode to execute. Instead we get a vector of {contract_class, contract_instance, bytecode} which define all the (deduplicated) contract bytecode that will be executed during this "one-enqueued call" (actual implementation of 1-enqueued call tbd).

This doesnt do any derivation of id or address yet

[IlyasRidhuan]

• feat(avm): merkle tree gadget #9205
• fix(avm): address bytecode hashing comments #9436
• feat: derive address and class id in avm #8897
• feat(avm): trace contract class and contract instance #8840 👈
• feat: bytecode hashing init #8535
• refactor(avm): type aliasing for VmPublicInputs #8884
• master

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

Join @IlyasRidhuan and the rest of your teammates on Graphite

[IlyasRidhuan]

We have this constructor because some of the execution tests just use bytecode

[IlyasRidhuan]

We are definitely "over-hinting" since we also transmit the address, bytecode commitment and class id.

But for now these are useful in checking that the values calculated in witgen match the simulator

[dbanks12]

Nice work! I have a few comments, but nothing that couldn't be handled in a follow-up PR.

[dbanks12]

Can we do Buffer.alloc(0) here?

[IlyasRidhuan]

So I'm not really sure what the behavior should be if we can't find the bytecode - do we throw (because we can't execute the instructions in the simulator anyways)?

If not we should hint some representation of "empty". I think the circuit can't put nothing in the column because then it wouldn't be provable.

[dbanks12]

I think we decided that if you call a contract that doesn't exist, the call instruction should just do nothing and return success=0 with empty returndata.

So do whatever's easiest for the bytecode hashing, because then we'll just throw out the results and prove that the contract address passed to the call instruction doesn't exist in the nullifier tree.

[github-actions]

Changes to public function bytecode sizes

Generated at commit: a23cc9511f509863513a21b37341b841c4ce89da, compared to commit: 7766c8e714185d6e8b9fa392d7f371fb30da8f1a

🧾 Summary (100% most significant diffs)

Program	Bytecode size in bytes (+/-)	%
AvmTest::public_dispatch	-713 ✅	-1.32%
AvmTest::bulk_testing	-1,355 ✅	-5.72%

---

Full diff report 👇

Program	Bytecode size in bytes (+/-)	%
AvmTest::public_dispatch	53,252 (-713)	-1.32%
AvmTest::bulk_testing	22,336 (-1,355)	-5.72%