EIPsInsight

Loading Experience

Preparing your insights...

Loading0%
ACDT2026-05-11_079

AllCoreDevs — Testing #079

2026-05-11 3 decisions 479 transcript lines

Transcript

Ask about this call

Answered from this call's summary and transcript only.

AI-generated — verify against the recording.

Call summary

Targets
  • Next week - Final repricing EIP numbers for 7904 — 00:09:23
Decisions
  • Close PR 5224 (SSZ limit raise); run DevNet <200M gas until progressive lists — 00:31:50
  • CL validates target adherence; EL validates EIP-1559 rules only — 00:53:36
  • Bid gas limit = target from preferences; payload gas limit = calculated — 01:00:24
Highlights
  • Ssz Limits:
    • ·PR 5224: Raise SSZ deposit limit to 1M to prevent deserialization failures — 00:29:15
    • ·Decision: Don't merge temporary fix; run DevNet at 150M gas until progressive lists — 00:31:05
  • Interop Recap:
    • ·Glamsterdam DevNet 3 unstable; Lighthouse sync issues, Teku envelope problems — 00:06:02
    • ·bal-devnet-6 near finalization; bal-devnet-7 will incorporate benchmarking optimizations — 00:08:08
    • ·Repricing: preliminary numbers for 7904/8038; optimizations ongoing, final EIPs next week — 00:08:59
    • ·FOCIL DevNet achieved interop during event; more Hegota testing planned — 00:10:37
  • Epbs Sync Issues:
    • ·Prysm downscoring peers serving incompatible block/payload ranges during sync — 00:11:45
    • ·Teku potentially serving payloads for non-canonical blocks; investigation needed — 00:13:29
    • ·Sync issues prevent DevNet 3 recovery; may require DevNet 4 — 00:14:39
  • Builder Deposit Spam:
    • ·PR 5223: Raise min builder withdrawability delay to 8192 epochs — 00:16:05
    • ·Testing needed: spam deposits to verify if delay extension solves verification DoS — 00:17:15
    • ·Lighthouse: 8192 signatures take ~700ms valid, 1s+ with invalids (M1) — 00:20:02
    • ·Attack cost estimate: $250k for 5-slot spam at 3% interest (2-month lock) — 00:23:17
    • ·Social slashing main deterrent; economic barrier considered low — 00:24:30
    • ·Repricing adds only 40% cost increase; not significant spam mitigation — 00:25:43
    • ·Alternative: EVM signature verification in new contract (caps by gas limit) — 00:27:26
  • Gas Limit Coordination:
    • ·Problem: EL/CL gas limit preferences not synchronized; affects ePBS bid validation — 00:36:28
    • ·Proposal: CL sends target gas limit via FCU; EL applies 1/1024 rule — 00:40:11
    • ·CL validates builder bids against target; EL validates EIP-1559 consensus rules — 00:42:49
    • ·Bid gas limit = target (from preferences), not calculated payload limit — 00:50:59
    • ·EL deprecates gas limit flag; accepts target via payload attributes only — 00:52:53
    • ·Alternative rejected: BPO-style hard-coded gas limits (loses emergency adjustment) — 00:56:06
Action Items
  • Justin Traglia — Validate signature verification benchmarks; present cost analysis at ACDC — 00:17:32
  • Nico Flaig — Update engine API PR 608 with target gas limit in payload attributes — 01:01:25
  • Potuz — Draft consensus spec PR for CL gas limit validation logic — 01:06:06
  • Bharath Vedartham — Update consensus spec: bid.gasLimit validation against target (not calculated) — 01:07:09

Meeting chat167

  • potuz

    gm

  • jochem-brouwer

    gm gm

  • nixo

    gm

  • Mario Vega

    gm 🙂

  • Enrico Del Fante (tbenr)

    gm

  • Mario Havel

    gMonday

  • Justin Traglia

    Gm

  • Toni Wahrstätter

    what an agenda today

  • Akash | ECH

    gm gm

  • potuz

    crap can't hear anything.

  • Mikhail Kalinin

    Me too

  • Dragan Rakita

    Is anyone talking? 🙂

  • Toni Wahrstätter

    not right now

  • Mikhail Kalinin

    Reacted to "Is anyone talking? 🙂" with 😃

  • Toni Wahrstätter

    but now

  • danceratopz

    sssh

  • Dragan Rakita

    Reacted to "not right now" with 😂

  • jochem-brouwer

    my sound has been tested and it has passed the test

  • Mikhail Kalinin

    Reacted to "my sound has been tested and it has passed the test" with 👏

  • Dragan Rakita

    Reacted to "my sound has been te..." with 👏

  • Parithosh Jayanthi

    Interop recap: https://blog.ethereum.org/2026/05/02/soldogn-interop-recap

  • Stefan Starflinger

    https://gist.github.com/qu0b/f3f905cadee4464a1a941838a5a5fadb

  • Potuz

    Just managed to log in here, @Barnabas I don’t think we finalized anything invalid

  • Potuz

    Client’s aren’t serving the right payloads though it seems

  • Mario Vega

    Reacted to "my sound has been te..." with 👏

  • Enrico Del Fante (tbenr)

    Replying to "Client’s aren’t serv..." On finalized slots or recent hot payloads?

  • Potuz

    Replying to "Client’s aren’t serv..." We’re unable to sync unfinalized

  • Jihoon

    I can share Hegota testing later

  • jochem-brouwer

    Heeft gereageerd op "I can share Hegota..." met 💙

  • Barnabas

    anyone can see a way to recover devnet 3?

  • Parithosh Jayanthi

    Fixing sync seems like it would do the trick?

  • Nico Flaig

    Replying to "Client’s aren’t serv..." still debugging on lodestar side trying to sync a node from last checkpoint right now was working on friday, now it's much harder

  • Potuz

    I think also Prysm has a bug trying to serve payloads it doesn’t have

  • Parithosh Jayanthi

    https://github.com/ethereum/consensus-specs/pull/5223

  • Mikhail Kalinin

    You can’t cap without a queue

  • Dustin

    the question is where the queue would be, yeah, CL or EL

  • Potuz

    The queue on the CL seems safe

  • Potuz

    But I just don’t think it solves the issue

  • Barnabas

    whats the worst case? 1 valid followed by 1 invalid?

  • Potuz

    A few invalid in random places

  • Potuz

    But that is a very expensive attack

  • Potuz

    So that’s fine

  • Potuz

    The problem is valid ones

  • Mikhail Kalinin

    If all CL devs are fine with potentially thousands sig verifications per slot then I am fine to have no cap

  • Justin Traglia

    We should just lower the gas limit 😅(jk)

  • Barnabas

    is there a way to have the evm execute the builder deposits sig verification without a new deposit contract?

  • Dustin

    Get a flash loan

  • Potuz

    Reacted to "Get a flash loan" with 🤣

  • Justin Traglia

    Reacted to "Get a flash loan" with 🤣

  • spencer

    Reacted to "Get a flash loan" with 🤣

  • Parithosh Jayanthi

    Reacted to "Get a flash loan" with 🤣

  • Parithosh Jayanthi

    https://github.com/ethereum/consensus-specs/pull/5224

  • Kevaundray Wedderburn

    Cost for pairing should be about 1ms

  • Barnabas

    could this pairing “job” not be paired somehow with the deposit itself?

  • Kevaundray Wedderburn

    Yeah if you batch, then you would essentially do a binary search to find the invalid signature, so logN number of pairings (worse case), where N is the size of the batch

  • jochem-brouwer

    Can we make a new contract which verifies this and then sends it to the original deposit contract?

  • jochem-brouwer

    Antwoord verzenden naar "Can we make a new ..." ah wait nvm if we still support sending to OG contract this defeats the purpose

  • Barnabas

    Replying to "Can we make a new co..." thats a big change

  • Potuz

    Replying to "Can we make a new co..." We can, but the original contract can still be reached.

  • Potuz

    Replying to "Can we make a new co..." A jinx

  • jochem-brouwer

    Antwoord verzenden naar "Can we make a new ..." yeah throwing around random ideas 😅

  • Potuz

    Replying to "Can we make a new co..." A new contract that verifies for the builder as Barnabas suggested works

  • jochem-brouwer

    Heeft gereageerd op "We can, but the or..." met 👍

  • jochem-brouwer

    Heeft gereageerd op "thats a big change" met 🤠

  • Potuz

    Replying to "Can we make a new co..." But it’s not an easy change

  • Potuz

    Just throwing it out there but raising the min deposit also solves this 🙂

  • jochem-brouwer

    Heeft gereageerd op "A new contract tha..." met 👍

  • jochem-brouwer

    Heeft gereageerd op "Just throwing it o..." met 👍

  • Barnabas

    Replying to "Just throwing it out..." not on devnets

  • Barnabas

    Replying to "Just throwing it out..." we are rich on devnets/testnets

  • Potuz

    Replying to "Just throwing it out..." lol

  • Parithosh Jayanthi

    https://github.com/ethereum/execution-apis/pull/794

  • Potuz

    @Parithosh Jayanthi are we talking about the gas limit issue?

  • Potuz

    That should be in the general part

  • Potuz

    Is Nico here?

  • Nico Flaig

    Barnabas knows it too

  • Potuz

    Reacted to "Barnabas knows it too" with 👍

  • Nico Flaig

    https://github.com/ethereum/pm/issues/2038#issuecomment-4420608861

  • brechy

    Reacted to "Yeah if you batch, t..." with 👍

  • Toni Wahrstätter

    just for my understanding. The gas limit in the bid is the preference (what is configured using a flag) and not the +/- 1/1024 gas limit?

  • Dustin

    Well, ELs typically have flags for this

  • Nico Flaig

    Replying to "Well, ELs typically ..." but they are static

  • Dustin

    EL then CL

  • Dustin

    I think

  • Bharath Vedartham

    Replying to "just for my understanding. The gas limit in the bid is the preference (what is configured using a flag) and not the +/- 1/1024 gas limit?" The spec mentions that the limit in the bid is the same as the gas limit in the payload https://github.com/ethereum/consensus-specs/blob/031e521f0a82b0e475ecfc13f79dd251a6284fc2/specs/gloas/builder.md?plain=1#L125

  • Potuz

    Reacted to "EL then CL" with 👍

  • Cayman Nava

    Reacted to "EL then CL" with 👍

  • Nico Flaig

    https://github.com/ethereum/execution-apis/pull/608

  • Bharath Vedartham

    Replying to "just for my understanding. The gas limit in the bid is the preference (what is configured using a flag) and not the +/- 1/1024 gas limit?" So i believe would be the current gas limit

  • brechy

    Replying to "Cost for pairing sho..." This is in the case of single invalid signature but could be problematic in the scenario of many bad signatures

  • Toni Wahrstätter

    I assume the gas limit we pass over from the CL to the EL is still the preferred one and the EL takes care of the 1/1024 rule still

  • Giulio

    I agree with Marius

  • jochem-brouwer

    Heeft gereageerd op "I assume the gas l..." met 👍

  • Nico Flaig

    we need to separate these 2 topics, 1) is about building the payload, and 2) is about who validates it (as potuz is saying)

  • Barnabas

    why should the bid have the gas limit in it at all?

  • Toni Wahrstätter

    Replying to "why should the bid h..." you would want to select a payload that does what you configured your validator to do (increase or decrease it)

  • Barnabas

    Replying to "why should the bid h..." send the bid in the proposer preferences, any bid over that proposer preference should be invalid no?

  • Toni Wahrstätter

    Are the proposer preferences updated everytime when a validator sets a new gas limit using the flag?

  • Ben Adams

    So CL target is hard constraint now?

  • Potuz

    It is now a constraint as well

  • Potuz

    We reject the builder’s header if it doesn’t give us something compatible

  • Marius Van Der Wijden (M)

    I don't understand what the el need to change now?

  • Nico Flaig

    but who validates the calculated gas limit in gloas? @Potuz

  • Marius Van Der Wijden (M)

    Sgtm

  • Potuz

    Reacted to "Sgtm" with ❤️

  • Ben Adams

    Spicy

  • Potuz

    Barnabas the dictator

  • Justin Traglia

    Reacted to "Barnabas the dictato..." with 😅

  • Potuz

    Replying to "but who validates th..." What validation Nico? The validation that it is compatible with the previous block is done by the EL

  • Toni Wahrstätter

    The EL should never use the value received through fcu for making the +/- 1/1024 rule stricter than it is. It should just be used for local building. But for validating, ELs shouldn't care up the gas limit going up or down, only that the 1/1024 rule is obeyed.

  • Potuz

    Replying to "but who validates th..." The validation that it was compatible with the preference should be in the CL

  • Ben Adams

    I there an EIP for BPO style @Barnabas ?

  • jochem-brouwer

    If we have to schedule an emergency gas limit downgrade then this is harder to do, so I dont think we should do BPO-like forks for this

  • Nico Flaig

    Replying to "but who validates th..." but we are not passing the target gas limit to the EL during validation

  • Barnabas

    Replying to "I there an EIP for B..." can make one

  • Dustin

    BPO-like forks have significant costs on the deployment/testing/etc side, due to needing new client releases or similar. BPOs are hardforks.

  • Ben Adams

    Reacted to "can make one" with 👍

  • Potuz

    Replying to "but who validates th..." The EL only validates 1559

  • nixo

    Reacted to "Spicy" with 🌶️

  • Potuz

    Replying to "but who validates th..." We validate the compatibility with the target

  • jochem-brouwer

    Antwoord verzenden naar "If we have to sche..." I like simplifying the protocol but for now I do not think I like this idea

  • Nico Flaig

    Replying to "but who validates th..." how can they do that without knowing what the target gas limit is?

  • Potuz

    Replying to "but who validates th..." That is my proposal

  • Potuz

    Replying to "but who validates th..." 1559 doesn’t care about the target only cares that the gas limit hasn’t changed that much

  • Nico Flaig

    Replying to "but who validates th..." but how do you know if builder adhered to the proposers target gas limit then?

  • pawan

    Don’t CLs do this target check against builder block already?

  • Parithosh Jayanthi

    Choosing to continue on this topic in lieu of devnet scoping today. We can do devnet scoping async/on ACDC this week as a topic.

  • Potuz

    I think what Toni’s saying is my favorite, but it’s not clear if others in the CL want this

  • Potuz

    Replying to "Don’t CLs do this ta..." Yes we do

  • Potuz

    Replying to "Don’t CLs do this ta..." That’s why I think it should not be a contentious issue for us

  • pawan

    Reacted to "That’s why I think i..." with 👍

  • Nico Flaig

    there is a PR https://github.com/ethereum/execution-apis/pull/608

  • nixo

    lol

  • Marius Van Der Wijden (M)

    You stick your head out, you get whacked

  • Parithosh Jayanthi

    Reacted to "You stick your head out, you get whacked" with 😂

  • felipe

    Reacted to You stick your head ... with "😂"

  • Eitan Seri-Levi

    Reacted to "You stick your head out, you get whacked" with 😂

  • Parithosh Jayanthi

    “Your reward for good work is more work”

  • Jared Wasinger

    Reacted to "You stick your head ..." with 😂

  • jochem-brouwer

    volunteering is non-optional

  • nixo

    Reacted to "You stick your head out, you get whacked" with 😂

  • nixo

    Reacted to "“Your reward for good work is more work”" with 😅

  • Potuz

    It’s in the EIP ( at least that’s what I copied )

  • jochem-brouwer

    gas limit bound defined by yellow paper

  • emma (reth)

    Reacted to "You stick your head ..." with 😂

  • jochem-brouwer

    Antwoord verzenden naar "gas limit bound de..." (part of protocol since chainstart / before EIP-1559)

  • Nico Flaig

    I think this is also ok, but we shuld define the calculation in the CL spec in that case, it's a simple algorithm

  • Potuz

    Replying to "I think this is also..." Yeah, I can do that one if you want

  • Parithosh Jayanthi

    We’re overtime, but choosing to continue today in case anyone wants to drop off!

  • Toni Wahrstätter

    I'm for the CL checking the increase/decrease and keep the EL only caring about 1/1024 (no matter of the direction)

  • jochem-brouwer

    Heeft gereageerd op "I'm for the CL che..." met 👍

  • Potuz

    I can draft that PR

  • Nico Flaig

    Reacted to "I can draft that PR" with 👍

  • Barnabas

    https://github.com/ethereum/EIPs/pull/11644

  • Nico Flaig

    Reacted to "Yeah, I can do that ..." with 👍

  • Potuz

    Conditioned to Nico’s doing the engine API

  • Barnabas

    Replying to "https://github.com/e..." https://ethereum-magicians.org/t/eip-xxxxx-gas-limit-schedule-gpo/28494

  • jochem-brouwer

    Antwoord verzenden naar "https://github.com..." wew this is fast

  • Barnabas

    Replying to "https://github.com/e..." gpo ftw

  • Parithosh Jayanthi

    Replying to "https://github.com/e..." Claude works fast

  • Barnabas

    Replying to "https://github.com/e..." psst

  • Cayman Nava

    Reacted to "Conditioned to Nico’..." with 😂

  • Barnabas

    Reacted to "Conditioned to Nico’..." with 😂

  • Nico Flaig

    Replying to "Conditioned to Nico’..." will do, don't worry, can you fix the devnet please

  • Potuz

    Reacted to "will do, don't worry, can you fix the devnet please" with 🤣

  • Cayman Nava

    Reacted to "will do, don't worry..." with 🤣

  • Parithosh Jayanthi

    Reacted to "will do, don't worry, can you fix the devnet please" with 🤣

Key decisions

  • Close PR 5224 (SSZ limit raise); run DevNet <200M gas until progressive lists

    Due to sync issues, it was decided to run DevNet with a gas limit of less than 200M until progressive lists are implemented.
  • CL validates target adherence; EL validates EIP-1559 rules only

    CL will validate builder bids against target gas limits, while EL will only validate EIP-1559 consensus rules.
  • Bid gas limit = target from preferences; payload gas limit = calculated

    The bid gas limit will be set to the target gas limit from preferences, while the payload gas limit will be calculated.