experiment(amsterdam): prove exact SSTORE write prepayment

[chugarchugarr]

Experimental execution-semantics proof for glamsterdam-write-prepayment, pinned to forks/amsterdam@20f7f6271a720091e5fea0a82e7bc802866ae36a.

The branch exposes four independently named executable assertions:

1. OSAKA_PASS — one parent gives a child a fixed CALL budget; the child changes an existing nonzero slot.
2. AMSTERDAM_FAIL — identical child bytecode and budget run out of gas after EIP-8038 reprices the warm first-change SSTORE from 2,900 to 10,100.
3. AMSTERDAM_VOUCHER_PASS — a transaction-boundary prepayment for the exact 7,200 write delta restores the child-frame requirement to 2,900.
4. AMSTERDAM_TOTAL_CHARGE_PRESERVED — the normal and prepaid storage-resource paths both charge 12,100:
   • normal: 2,100 cold access + 10,000 write;
   • prepaid: 2,000 access-list prepayment + 7,200 write delta + 100 warm access + 2,800 legacy write.

Prototype representation:

• three domain-separated access-list markers contribute 6,000 of the write delta;
• a 1,200 intrinsic remainder is charged at the transaction boundary;
• Amsterdam records one (address, storage key) prepayment in the transaction environment;
• SSTORE consumes it once, including across an OOG/reverted child attempt.

The marker representation is only a proof vehicle. It is not a proposed final transaction wire format. Encoding overhead is isolated from the conserved 12,100 storage-resource charge.

Containment tests also prove partial markers do not activate a discount, an OOG attempt cannot replay a voucher, and ordinary Amsterdam SSTORE behavior remains unchanged.

Mainnet scanning remains out of scope for this PR.

RLP-1 resolution tracking

This experiment is now the first live engineering claim tracked through the external RLP-1 resolution-lineage profile.

Pinned evidence snapshot:
https://github.com/chugarchugarr/resolution-receipt-technocore/blob/proposal/resolution-lineage/public/claims/glamsterdam-write-prepayment/evidence-pr-head.json

Signed resolution record:
https://github.com/chugarchugarr/resolution-receipt-technocore/blob/proposal/resolution-lineage/public/claims/glamsterdam-write-prepayment/resolution-0001.json

Current derived state: UNRESOLVED.

The four internal executable assertions are recorded as PASS against pinned PR head e16d4096a95ad1409fe4b26af0dbeb8de7dd1fe9. A fifth required check — independent technically credible falsification/review — remains UNRESOLVED, so RLP-1 mechanically prevents this experiment from being represented as SURVIVED before that external check exists.

This RLP-1 state is a bounded resolution claim only. It is not an Ethereum acceptance, adoption, mainnet, or wire-format claim.

[chugarchugarr]

Focused proof is green: https://github.com/chugarchugarr/execution-specs/actions/runs/32996887596

@misilva73 @weiihann — requesting a narrow falsification pass because this experiment is driven by EIP-8038's SSTORE repricing. The two review questions are:

1. Does the identical fixed child budget correctly cross Osaka PASS → Amsterdam OOG?
2. Does relocating exactly the 7,200 STORAGE_WRITE delta to transaction scope restore the child while preserving the full 12,100 Amsterdam storage-resource charge?

This is not a proposed wire format or an adoption claim. Mainnet-value scanning is out of scope. Canonical evidence boundary: chugarchugarr/-x402-resolution-receipt#2

[chugarchugarr]

This experiment is now the first real engineering claim tracked through RLP-1. The pinned evidence object and signed resolution record are public in chugarchugarr/resolution-receipt-technocore on proposal/resolution-lineage.

Current derived state is intentionally UNRESOLVED: the four executable assertions are recorded as PASS, while a fifth required external-falsification check remains UNRESOLVED until a technically credible reviewer accepts, narrows, or falsifies the two review questions already posted here.

The machine check for that public RLP-1 claim is green on RLP head dd63ce241d6ceec25db447b7dc33c435664d7b11. This is not an Ethereum acceptance or adoption claim.