Initial implementation of Stateful Precompile Contracts #4
Conversation
evgeniy-scherbina
force-pushed
the
yevhenii/precompile-v2
branch
2 times, most recently
from
99d95a4 to
3b338ea
Compare
evgeniy-scherbina
force-pushed
the
yevhenii/precompile-v2
branch
from
06ccd5a to
c0e9f03
Compare
evgeniy-scherbina
changed the title
| @@ -231,6 +237,26 @@ func (c *sha256hash) Run(input []byte) ([]byte, error) { | |||
| return h[:], nil | |||
| } | |||
|
|
|||
| type sum3 struct{} | |||
|
|
|||
There was a problem hiding this comment.
Is it worth splitting the specific pre-compiles into a separate file per precompile (thinking yes if there are going to be 10's of pre-compiles, no if less)
There was a problem hiding this comment.
This is native stateless precompile, they all are placed in one file here: https://github.com/Kava-Labs/go-ethereum/blob/master/core/vm/contracts.go
I just extended it with one more example stateless smart contract.
New stateful smart contracts is placed one per golang package.
fyi: PR was moved to #10
| @@ -224,7 +243,8 @@ func (evm *EVM) Call(caller ContractRef, addr common.Address, input []byte, gas | |||
| } | |||
|
|
|||
| if isPrecompile { | |||
| ret, gas, err = RunPrecompiledContract(p, input, gas) | |||
| // TODO(yevhenii): deduct gas? | |||
There was a problem hiding this comment.
Worth adding test cases for this, I think the cosmos-sdk side of things will handle tracking and deducting actual gas (with a possible assist from statedb @drklee3 ?) but having unit and integration tests around what happens gas wise when a block with a failed precompile tx is executed are a useful foundation for testing the feature now and deeper as needed going forward
There was a problem hiding this comment.
fyi: PR was moved to #10
gas deduction changed a bit in a new PR, also I added some tests related to gas calculations
| } | ||
|
|
||
| // NewStatefulPrecompileContract generates new StatefulPrecompile using [functions] as the available functions and [fallback] | ||
| // as an optional fallback if there is no input data. Note: the selector of [fallback] will be ignored, so it is required to be left empty. |
There was a problem hiding this comment.
Guessing there's some higher level function we have to satisfy otherwise we could just drop this unused fallback variable?
There was a problem hiding this comment.
I already dropped fallback in a new version of PR, though I think we still need to discuss either we need fallback function or no
|
|
||
| func CalculateFunctionSelector(functionSignature string) []byte { | ||
| hash := crypto.Keccak256([]byte(functionSignature)) | ||
| return hash[:4] |
There was a problem hiding this comment.
Why just the last 4 bytes / worth a code comment explaining it for future readers?
There was a problem hiding this comment.
@galxy25 This function was improved in a new PR
|
Moved to #5 |
evgeniy-scherbina
added
duplicate
Moved to #5