Instrument the CALL, CALLCODE, DELEGATECALL and STATICCALL opcodes #486

MishkaRogachev · 2025-07-11T13:48:10Z

For NIT-3483 and NIT-3480

Instrument CALL and CALLCODE opcode with multigas calculations:

Base gasCall func
Base gasCallCode func
makeCallVariantGasCallEIP2929 for EIP-2929 (access lists)
makeCallVariantGasEIP4762 for EIP-4762 (Verkle trees)
makeCallVariantGasCallEIP7702 for EIP-7702 (set code transaction)
Base gasDelegateCall func
Base gasStaticCall func

MishkaRogachev · 2025-07-11T14:57:37Z

core/vm/gas_table.go

 	if evm.chainRules.IsEIP4762 && !contract.IsSystemCall {
 		if transfersValue {
-			gas, overflow = math.SafeAdd(gas, evm.AccessEvents.ValueTransferGas(contract.Address(), address))
+			valueTransferGas := evm.AccessEvents.ValueTransferGas(contract.Address(), address)
+			overflow := multiGas.SafeIncrement(multigas.ResourceKindStorageAccess, valueTransferGas)


Note: In EIP-4762 (Verkle trees), value transfers require accessing the account's balance leaf in the tree, which is considered a storage access operation

MishkaRogachev · 2025-07-15T11:26:25Z

core/vm/gas_table.go

 				return multigas.ZeroGas(), 0, ErrGasUintOverflow
 			}
 		}
 	}
-	evm.callGasTemp, err = callGas(evm.chainRules.IsEIP150, contract.Gas, gas, stack.Back(0))
+
+	// NOTE(NIT-3483): using only the computation gas for callGas


Needs to be double-checked

I don't really understand why.

I think the third parameter to callGas below (called base in that function) should be the sum of all dimensions, not just the computation gas, and with my earlier comment on how CallNewAccountGas needs to be storage growth, that will actually matter

Thank you, now I figured out. Initially I thought that memoryGasCost only makes sense for computation gas because I was getting inconsistent results in tests, which was due to the wrong type for CallNewAccountGas. Fixed both issues with latest commit

eljobe · 2025-07-15T11:41:04Z

core/vm/gas_table.go

@@ -52,7 +52,7 @@ func memoryGasCost(mem *Memory, newMemSize uint64) (*multigas.MultiGas, uint64,
 		fee := newTotalFee - mem.lastGasCost
 		mem.lastGasCost = newTotalFee

-		return multigas.ZeroGas(), fee, nil
+		return multigas.ComputationGas(fee), fee, nil


How did you decide to allocate this to computation gas?
Is that what Tyler's tracer did?
I would have guessed (possibly, incorrectly) that gas for expanding memory would be state growth gas.

My assumption was based on this statement, and here and in all following places I have treated memory operations as computation cost since there are no appeals on statedb. Please correct me if I'm wrong

Okay. So, this is, at least, consistent with what Tyler has done in the tracer. Good.

Now, though, I do need to understand why that's the right behavior.

Maybe I don't understand what the difference is between memory expansion and the storage in the statedb.

By way of analogy, are we saying that the "memory" of the EVM is similar to RAM on a real machine. And, that entries in the statedb are more like "disk"? And, so, when we talk about "state growth" and "state access" we are only talking about the equivalent of disk growth and i/o? And that "memory" is only what has to be paged into the working memory of the VM to support the computations which are happening while processing the transactions and even though the values of the allocated memory are part of what makes up the global state root, it's not really part of the "state" we're talking about when we talk about "state growth" or "state access?"

@tsahee, @relyt29 and @magicxyyz am I understanding this right?

yes that is exactly correct

Okay. Thanks. I feel like we should put some abbreviated explanation of this reasoning in the inline documentation so that we remember why we're treating memory expansion gas as computation gas. Keep it brief, but, let's do explain it.

I don't see the documentation for this. Did you decide not to add something?

core/vm/gas_table.go

eljobe · 2025-07-15T11:42:32Z

core/vm/gas_table.go

 				return multigas.ZeroGas(), 0, ErrGasUintOverflow
 			}
 		}
 	}
-	evm.callGasTemp, err = callGas(evm.chainRules.IsEIP150, contract.Gas, gas, stack.Back(0))
+
+	// NOTE(NIT-3483): using only the computation gas for callGas


I don't really understand why.

core/vm/gas_table.go

eljobe · 2025-07-18T15:30:08Z

core/vm/gas_table.go

@@ -52,7 +52,7 @@ func memoryGasCost(mem *Memory, newMemSize uint64) (*multigas.MultiGas, uint64,
 		fee := newTotalFee - mem.lastGasCost
 		mem.lastGasCost = newTotalFee

-		return multigas.ZeroGas(), fee, nil
+		return multigas.ComputationGas(fee), fee, nil


I don't see the documentation for this. Did you decide not to add something?

cla-bot bot added the s CLA signed label Jul 11, 2025

MishkaRogachev changed the title ~~Instrument gasCall with multi gas~~ Instrument the CALL and CALLCODE opcodes Jul 11, 2025

MishkaRogachev force-pushed the instrument_call_gas_func branch from e66c1e6 to 3e5a9fd Compare July 11, 2025 14:54

MishkaRogachev commented Jul 11, 2025

View reviewed changes

MishkaRogachev force-pushed the instrument_call_gas_func branch 2 times, most recently from e99cca4 to 487493b Compare July 14, 2025 17:54

MishkaRogachev changed the title ~~Instrument the CALL and CALLCODE opcodes~~ Instrument the CALL, CALLCODE, DELEGATECALL and STATICCALL opcodes Jul 15, 2025

MishkaRogachev commented Jul 15, 2025

View reviewed changes

MishkaRogachev marked this pull request as ready for review July 15, 2025 11:49

MishkaRogachev requested review from eljobe and tsahee July 15, 2025 11:50

MishkaRogachev assigned eljobe Jul 15, 2025

eljobe requested changes Jul 15, 2025

View reviewed changes

eljobe assigned MishkaRogachev and unassigned eljobe Jul 15, 2025

MishkaRogachev force-pushed the instrument_call_gas_func branch from 0f1ce7a to ca34cf3 Compare July 15, 2025 14:01

MishkaRogachev assigned eljobe and MishkaRogachev and unassigned MishkaRogachev and eljobe Jul 15, 2025

MishkaRogachev added 8 commits July 17, 2025 12:28

Instrument gasCall with multi gas

970caeb

Instrument gasCallCode with multi gas and unify the test scenario

01bcf64

Instrument makeCallVariantGasCallEIP2929 with multi gas

c0bc014

Fixup gas operations test name

983b1fa

Instrument makeCallVariantGasEIP4762 with multi gas

d694447

Instrument gasDelegateCall and gasStaticCall with multi gas

5e78561

Instrument makeCallVariantGasCallEIP7702 with multi gas

2661c91

Fix resource kind and callGas parameter in gasCall and gasCallCode

0665b84

MishkaRogachev force-pushed the instrument_call_gas_func branch from ca34cf3 to 0665b84 Compare July 17, 2025 10:29

MishkaRogachev mentioned this pull request Jul 17, 2025

Instrument CREATE and CREATE2 gas func #489

Open

2 tasks

MishkaRogachev changed the base branch from master to cbp-feature July 18, 2025 11:17

MishkaRogachev mentioned this pull request Jul 18, 2025

Instrument LOG0-LOG4 gas function #492

Open

1 task

MishkaRogachev requested review from eljobe and gligneul July 18, 2025 13:45

MishkaRogachev assigned eljobe and unassigned MishkaRogachev Jul 18, 2025

eljobe requested changes Jul 18, 2025

View reviewed changes

eljobe assigned MishkaRogachev and unassigned eljobe Jul 18, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Instrument the CALL, CALLCODE, DELEGATECALL and STATICCALL opcodes #486

Instrument the CALL, CALLCODE, DELEGATECALL and STATICCALL opcodes #486

MishkaRogachev commented Jul 11, 2025 •

edited

Loading

Uh oh!

MishkaRogachev Jul 11, 2025

Uh oh!

MishkaRogachev Jul 15, 2025

Uh oh!

eljobe Jul 15, 2025

Uh oh!

relyt29 Jul 15, 2025

Uh oh!

MishkaRogachev Jul 15, 2025

Uh oh!

eljobe Jul 15, 2025

Uh oh!

MishkaRogachev Jul 15, 2025

Uh oh!

eljobe Jul 15, 2025 •

edited

Loading

Uh oh!

relyt29 Jul 15, 2025

Uh oh!

eljobe Jul 15, 2025

Uh oh!

eljobe Jul 18, 2025

Uh oh!

Uh oh!

eljobe Jul 15, 2025

Uh oh!

Uh oh!

Uh oh!

eljobe Jul 18, 2025

Uh oh!

Uh oh!

Instrument the CALL, CALLCODE, DELEGATECALL and STATICCALL opcodes #486

Are you sure you want to change the base?

Instrument the CALL, CALLCODE, DELEGATECALL and STATICCALL opcodes #486

Conversation

MishkaRogachev commented Jul 11, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

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!

eljobe Jul 15, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

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!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

MishkaRogachev commented Jul 11, 2025 •

edited

Loading

eljobe Jul 15, 2025 •

edited

Loading