core/state: implement optional BAL construction in statedb #31959

jwasinger · 2025-06-04T08:55:56Z

This PR implements changes to Geth which allow for the creation of block-level access lists, generated when executing blocks for import.

Ultimately, block access lists will be created from within the miner and validated as part of block execution. However, being able to generate them from historical blocks is useful for collecting access lists across large ranges of blocks. So the changes in this PR aim to facilitate data collection which will inform the continued development of the spec.

MariusVanDerWijden · 2025-06-08T07:50:18Z

Oh wow, I'm surprised that it takes that little code to built this. I guess we have a pretty good abstraction with the statedb

jwasinger · 2025-07-01T09:12:33Z

accounts/abi/unpack_test.go

@@ -1014,128 +1014,134 @@ func TestPackAndUnpackIncompatibleNumber(t *testing.T) {
 	cases := []struct {


I have no clue why this file is showing diffs. I've compared locally with master and there is no diff. Might be a github bug.

jwasinger · 2025-07-01T09:18:03Z

core/genesis.go

@@ -681,7 +681,7 @@ func DeveloperGenesisBlock(gasLimit uint64, faucet *common.Address) *Genesis {
 		},
 	}
 	if faucet != nil {
-		genesis.Alloc[*faucet] = types.Account{Balance: new(big.Int).Sub(new(big.Int).Lsh(big.NewInt(1), 256), big.NewInt(9))}


BAL max balance is limited to 12 bytes. Geth will panic if an attempt is made to record a larger balance in the BAl.

I use dev mode to test changes here so I reduced the allocation here to allow for that.

It's now 16 bytes in the spec.

gballet

I wonder why you need to add a tracer, and simply not use the witness to build the BAL?

gballet · 2025-07-11T09:16:53Z

core/vm/interpreter.go

@@ -35,6 +35,7 @@ type Config struct {
 	ExtraEips               []int // Additional EIPS that are to be enabled

 	StatelessSelfValidation bool // Generate execution witnesses and self-check against them (testing purpose)
+	BALConstruction         bool // Generate BAL for executed blocks (testing purposes)


can't you define an eip field in the rules (see eips.go) and use isEIPWhatever instead of adding another new field here?

BAL construction is only performed when creating a block, not verifying. So it's not possible to activate based on fork alone.

gballet · 2025-07-11T09:19:29Z

go.sum

@@ -116,6 +118,8 @@ github.com/fatih/color v1.16.0 h1:zmkK9Ngbjj+K0yRhTVONQh1p/HknKYSlNT+vZCzyokM=
 github.com/fatih/color v1.16.0/go.mod h1:fL2Sau1YI5c0pdGEVCbKQbLXB6edEj1ZgiY4NijnWvE=
 github.com/ferranbt/fastssz v0.1.2 h1:Dky6dXlngF6Qjc+EfDipAkE83N5I5DE68bY6O0VLNPk=
 github.com/ferranbt/fastssz v0.1.2/go.mod h1:X5UPrE2u1UJjxHA8X54u04SBwdAQjG2sFtWs39YxyWs=
+github.com/ferranbt/fastssz v0.1.4 h1:OCDB+dYDEQDvAgtAGnTSidK1Pe2tW3nFV40XyMkTeDY=


Is Peter's lib not working?

I would really like to keep fastszz out, it's really tricky to use

We've tentatively removed SSZ from the spec in favor of RLP. This branch hasn't been updated against #31948 . I will rebase this as soon as that's merged.

gballet · 2025-07-11T09:20:24Z

params/config.go

+	CancunTime      *uint64 `json:"cancunTime,omitempty"`      // Cancun switch time (nil = no fork, 0 = already on cancun)
+	PragueTime      *uint64 `json:"pragueTime,omitempty"`      // Prague switch time (nil = no fork, 0 = already on prague)
+	OsakaTime       *uint64 `json:"osakaTime,omitempty"`       // Osaka switch time (nil = no fork, 0 = already on osaka)
+	GlamsterdamTime *uint64 `json:"glamsterdamTime,omitempty"` // Glamsterdam switch time (nil = no fork, 0 = already on glamsterdam)


make this another PR

gballet · 2025-07-11T09:24:21Z

core/types/bal/bal.go

@@ -0,0 +1,227 @@
+package bal


FWIW, and for the sake of future maintainers, I think it's worth spelling the whole "block access list" out. yes it's long, but bal is just too non-descriptive for people who are not involved in BALs.

I just hate the way blockaccesslist looks...

gballet · 2025-07-11T09:24:52Z

params/config.go

+	IsEIP2929, IsEIP4762                                            bool
+	IsByzantium, IsConstantinople, IsPetersburg, IsIstanbul         bool
+	IsBerlin, IsLondon                                              bool
+	IsMerge, IsShanghai, IsCancun, IsPrague, IsOsaka, IsGlamsterdam bool


you did create that field, so I understand your need to do BALConstruction

gballet · 2025-07-11T09:25:34Z

go.mod

@@ -24,7 +24,7 @@ require (
 	github.com/ethereum/c-kzg-4844/v2 v2.1.0
 	github.com/ethereum/go-verkle v0.2.2
 	github.com/fatih/color v1.16.0
-	github.com/ferranbt/fastssz v0.1.2
+	github.com/ferranbt/fastssz v0.1.4


oh crap we already had it...

gballet · 2025-07-11T09:27:14Z

tests/block_test_util.go

@@ -111,7 +111,7 @@ type btHeaderMarshaling struct {
 	ExcessBlobGas *math.HexOrDecimal64
 }

-func (t *BlockTest) Run(snapshotter bool, scheme string, witness bool, tracer *tracing.Hooks, postCheck func(error, *core.BlockChain)) (result error) {
+func (t *BlockTest) Run(snapshotter bool, scheme string, witness bool, bal bool, tracer *tracing.Hooks, postCheck func(error, *core.BlockChain)) (result error) {


that's the issue I have with this BALConstruction: we will have to add a flag all over the place, when it's a glamsterdam thing (actually maybe not, but it's a well-identified eip which activation can be checked with rules)

See my comment above: bal construction is not strictly dependent on fork activation, it will be activated when we are creating a block (and glamsterdam is active)

jwasinger · 2025-07-14T03:00:41Z

Beacon root contract update and withdrawals are not properly handled here. However, the BAL spec does not specify how state mutations which occur before/after execution of block txs are handled. Trying to get this spec'd and then will fix in this branch.

…y a flag

jwasinger · 2025-07-23T10:23:23Z

closing this as it has become stale. I'm not sure where things will land after #32263 is completed and refactored.

jwasinger force-pushed the bal-construction branch 3 times, most recently from 1e6266f to cebf143 Compare June 18, 2025 14:38

jwasinger commented Jul 1, 2025

View reviewed changes

jwasinger force-pushed the bal-construction branch 8 times, most recently from b13d49d to e42ed68 Compare July 10, 2025 10:34

gballet reviewed Jul 11, 2025

View reviewed changes

jwasinger force-pushed the bal-construction branch 2 times, most recently from 34a7a2b to 9ef8dcb Compare July 14, 2025 02:07

core/types: add optional bal construction in statedb when triggered b…

b5948ee

…y a flag

jwasinger force-pushed the bal-construction branch from f5d6d01 to b5948ee Compare July 14, 2025 10:05

jwasinger closed this Jul 23, 2025

		@@ -1014,128 +1014,134 @@ func TestPackAndUnpackIncompatibleNumber(t *testing.T) {
		cases := []struct {

core/state: implement optional BAL construction in statedb #31959

core/state: implement optional BAL construction in statedb #31959

Uh oh!

Conversation

jwasinger commented Jun 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MariusVanDerWijden commented Jun 8, 2025

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!

gballet left a comment

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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jwasinger commented Jul 14, 2025

Uh oh!

jwasinger commented Jul 23, 2025

Uh oh!

Uh oh!

jwasinger commented Jun 4, 2025 •

edited

Loading