feat(ethrex): Use EIP-8025 feature to delegate as much logic into their guest program by jsign · Pull Request #39 · eth-act/ere-guests

jsign · 2026-04-21T22:09:25Z

Summary

This PR rewires stateless-validator-ethrex to use Ethrex's native EIP-8025 execution path instead of maintaining a local NewPayloadRequest -> Block conversion layer.

The main goal is to keep the guest thinner and align our input format with the upstream Ethrex interface, while reducing duplicate payload/parsing logic on our side.

What changed

updated Ethrex dependencies to the jsign/ethrex fork at a485d97, which includes the EIP-8025 guest entrypoint we need. I'll wait until feat(l1): fix EIP-8025 compliance lambdaclass/ethrex#6516 gets merged so we can change to ethrex repo again before merging.
replaced the old StatelessValidatorEthrexInput rkyv-based guest input with raw EIP-8025 wire bytes
added a small wire module that encodes the input as [ssz_len: u32 LE][ssz_bytes][rkyv_witness_bytes] -- this isn't perfectly EIP-8025 compliant since their ExecutionWitness with EIP-8025 feature still uses rkyv. The next step for Ethrex is moving the ExecutionWitness to be spec compliant (thus use SSZ). But we're getting closer.
changed the guest to call execution_program_eip8025_bytes(...) directly and return the upstream new_payload_request_root / valid result
moved host-side input construction to a dedicated build_eip8025_input(...) helper
derived SSZ encode/decode for the shared NewPayloadRequest types needed to serialize Electra/Fulu payloads
removed the local execution_payload.rs and new_payload_request.rs conversion/validation code, which is now redundant with the Ethrex implementation
updated debug and integration-test call sites to build the new wire input
limited the Ethrex path to Electra/Fulu fixtures, which matches the current EIP-8025-supported flow -- this is mostly because the NewRequestPayload expected in Ethrex today only supports this fork.

Why

This reduces maintenance in a sensitive part of the stack:

less duplicated block/payload translation logic in this repo
closer alignment with the upstream Ethrex guest interface
clearer separation between wire encoding on the host side and validation/execution inside Ethrex
smaller surface area for fork-specific payload handling bugs

…s side

…alidator-ethrex

refactor: clean up guest.rs and wire.rs by removing unused code

jsign · 2026-04-21T22:17:03Z

+        if !matches!(fork, ForkName::Electra | ForkName::Fulu) {
+            return None;
+        }


Mainly because now we rely on whatever input is supported by Ethrex which is Electra/Fulu and prob forward from now on.

Should prob be okay until we do runs for older forks some day.
Amsterdam support should be added soon for Ethrex to continue with forward forks.

jsign · 2026-04-21T22:17:30Z

Nice all this is gone.

jsign · 2026-04-21T22:19:18Z

-#[derive(rkyv::Serialize, rkyv::Deserialize, rkyv::Archive)]
-pub struct StatelessValidatorEthrexInput {
-    /// New payload request data.
-    pub new_payload_request: NewPayloadRequest,
-    /// database containing all the data necessary to execute
-    pub execution_witness: ExecutionWitness,
-}


This is gone, since now the guest program receives raw bytes which are directly passed to Ethreum guest program entrypoint. See L21.

Does it mean that the deserialize_input cost is not merged into stf?

The generic deserialize_input scope is still separate, but it only covers copying the raw input bytes now. The actual EIP-8025 decode/deserialization happens inside Ethrex’s execution_program_eip8025_bytes, so that part is now accounted under stf.

I think this prob means we should rename stf for Ethrex now to something a bit more generic, since doesn't have the same meaning as stf before. I'll think some name and change it :)

Changed to run_validation -- not sure if fully convinced, but seems to be a fairer name.

jsign · 2026-04-21T22:20:11Z


 /// [`Guest`] implementation for Ethrex stateless validator.
 #[derive(Debug, Clone)]
 pub struct StatelessValidatorEthrexGuest;


This got massively simplified, since now we rely on Ethrex logic doing all this stuff.

jsign · 2026-04-21T22:22:59Z

-        #[cfg(not(feature = "std"))]
-        {
-            Self::compute_inner::<P>(input, new_payload_request_root, crypto)
-        }


Trying to assume no std, tracked by lambdaclass/ethrex#6339

jsign · 2026-04-21T22:24:59Z

Another nice deletion, also done by Ethrex guest program now.

jsign · 2026-04-21T22:25:46Z

This is the encoding defined by Ethrex today which has an hybrid of using SSZ for the NewPayloadRequest and rkyv for ExecutionWitness.

Eventually this should be gone whenever it complies to the spec (which uses all SSZ).

jsign · 2026-04-21T22:26:00Z

+ethrex-common = { git = "https://github.com/jsign/ethrex.git", rev = "a485d9717e9ce5da6901361b9a04c8026d62fa68", default-features = false }
+ethrex-crypto = { git = "https://github.com/jsign/ethrex.git", rev = "a485d9717e9ce5da6901361b9a04c8026d62fa68", default-features = false }
+ethrex-guest-program = { git = "https://github.com/jsign/ethrex.git", rev = "a485d9717e9ce5da6901361b9a04c8026d62fa68", default-features = false }
+ethrex-rlp = { git = "https://github.com/jsign/ethrex.git", rev = "a485d9717e9ce5da6901361b9a04c8026d62fa68", default-features = false }
+ethrex-rpc = { git = "https://github.com/jsign/ethrex.git", rev = "a485d9717e9ce5da6901361b9a04c8026d62fa68", default-features = false }
+ethrex-vm = { git = "https://github.com/jsign/ethrex.git", rev = "a485d9717e9ce5da6901361b9a04c8026d62fa68", default-features = false }


To be updated whenever lambdaclass/ethrex#6516 gets merged.

jsign · 2026-04-21T22:42:30Z

@han0110, despite is better to wait until Ethrex PR is merged, I think from this side should be okay to have your eyes for a review.

han0110

Nice simplifcation, looks much cleaner! Just got a small question

han0110 · 2026-04-22T03:02:46Z

-#[derive(rkyv::Serialize, rkyv::Deserialize, rkyv::Archive)]
-pub struct StatelessValidatorEthrexInput {
-    /// New payload request data.
-    pub new_payload_request: NewPayloadRequest,
-    /// database containing all the data necessary to execute
-    pub execution_witness: ExecutionWitness,
-}


Does it mean that the deserialize_input cost is not merged into stf?

Signed-off-by: jsign <jsign.uy@gmail.com>

# Conflicts: # Cargo.lock # bin/stateless-validator-ethrex/risc0/Cargo.lock # bin/stateless-validator-ethrex/sp1/Cargo.lock # bin/stateless-validator-ethrex/zisk/Cargo.lock # crates/stateless-validator-ethrex/Cargo.toml # crates/stateless-validator-ethrex/src/execution_payload.rs # crates/stateless-validator-ethrex/src/guest.rs # crates/stateless-validator-ethrex/src/host.rs

…ion program import

Co-authored-by: Copilot <copilot@github.com>

jsign · 2026-04-27T13:18:05Z

@han0110, the lambdaclass/ethrex#6516 pr was merged. Now switched Cargo.toml to point to ethrex main repo and branch. Reg code, nothing interesting has changed since your last comment (only this nit change during the ethrex pr reviewing).

han0110

LGTM

## Summary This branch fixes the EIP-8025 guest-program execution flow and aligns the SSZ-side `NewPayloadRequest` model with the Electra/CL spec. At a high level, it adds a byte-oriented EIP-8025 execution entrypoint, moves execution requests into a typed SSZ container, and uses that typed representation to reconstruct the EL `requests_hash` correctly when converting a `NewPayloadRequest` into a block. ## What changed - Added `execution_program_eip8025_bytes`, which decodes and executes the EIP-8025 wire format directly: - `[ssz_len: u32 LE][ssz_bytes][rkyv_bytes]` - Refactored the EIP-8025 validation flow so both entrypoints share the same logic for: - `NewPayloadRequest -> Block` conversion - block hash validation - blob versioned hash validation - stateless execution via `execute_blocks` - Updated the EIP-8025 SSZ types in `ethrex-common` to better match the Electra consensus spec: - `ExecutionPayload` no longer carries deposit/withdrawal/consolidation request lists - `NewPayloadRequest.execution_requests` is now a typed `ExecutionRequests` container - spec constants were renamed/normalized for clarity - the consolidation request limit was corrected to the Electra value - Added `ExecutionRequests::to_encoded_requests()` to convert the SSZ typed request container into the EIP-7685 encoded request list expected by `compute_requests_hash` - Re-exported the new EIP-8025 byte entrypoint from `ethrex-guest-program` ## Behavior notes - Malformed EIP-8025 wire input still returns an error - For well-formed input, the byte-entrypoint now always computes the `new_payload_request_root` - If validation/execution fails after decoding, the byte-entrypoint returns `ProgramOutput { valid: false, ... }` instead of failing outright ## Why this matters This makes the EIP-8025 guest path closer to the consensus-side structure we actually need to prove against, while also giving us a direct raw-bytes execution entrypoint for the EIP-8025 wire format. It also removes the mismatch between the SSZ request representation and the EL request-hash computation path by deriving `requests_hash` from a properly typed `ExecutionRequests` container. ## External usage I'm already using this PR branch into an ere-guests PR here: eth-act/ere-guests#39 This allowed `ere-guests` to basically remove a lot of wrapper code before calling the guest program regarding execution-payload->Block, output calculation, etc -- making now Ethrex own all that logic as expected. --------- Signed-off-by: jsign <jsign.uy@gmail.com>

jsign added 6 commits April 21, 2026 14:01

fix: update ethrex dependencies to latest commit

cd8a8a9

fix: update ethrex dependencies to point to jsign fork

27f21ca

Refactor stateless-validator-ethrex to support EIP-8025 input encoding

7e27e30

feat: update Ethrex fork again which lets us simplify more things thi…

e1fdc60

…s side

refactor: remove unused dependencies and clean up code in stateless-v…

02a2335

…alidator-ethrex

fix: update ethrex dependencies to latest commit hash

0094337

refactor: clean up guest.rs and wire.rs by removing unused code

jsign mentioned this pull request Apr 21, 2026

feat(l1): fix EIP-8025 compliance lambdaclass/ethrex#6516

Merged

nits

ba15fce

jsign commented Apr 21, 2026

View reviewed changes

jsign marked this pull request as ready for review April 21, 2026 22:40

jsign requested a review from han0110 April 21, 2026 22:42

han0110 reviewed Apr 22, 2026

View reviewed changes

jsign added 6 commits April 22, 2026 10:24

ethrex: change scope name

df8109d

Signed-off-by: jsign <jsign.uy@gmail.com>

fmt

4f7992e

fix: update ethrex dependencies to latest revision and correct execut…

62ca670

…ion program import

Merge branch 'main' into jsign-ethrex-8025

763e778

nit

6203b0a

benbencik mentioned this pull request Apr 24, 2026

feat(l1): add ssz encoding for witness in eip-8025 lambdaclass/ethrex#6524

Closed

update ethrex dep to main repo

e1a9874

Co-authored-by: Copilot <copilot@github.com>

jsign requested a review from han0110 April 27, 2026 13:16

han0110 approved these changes Apr 27, 2026

View reviewed changes

jsign merged commit 3387b8f into eth-act:main Apr 27, 2026
41 checks passed

Conversation

jsign commented Apr 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

What changed

Why

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jsign commented Apr 21, 2026

Uh oh!

han0110 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jsign commented Apr 27, 2026

Uh oh!

han0110 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

jsign commented Apr 21, 2026 •

edited

Loading