Implement SELFBALANCE opcode from EIP-1884 #24
Conversation
Codecov Report
@@ Coverage Diff @@
## master #24 +/- ##
========================================
+ Coverage 99.88% 100% +0.11%
========================================
Files 5 5
Lines 842 845 +3
Branches 109 109
========================================
+ Hits 841 845 +4
+ Misses 1 0 -1 |
|
The invalid opcode test is failing because |
Add |
|
Will make this one depend on ethereum/evmc#372 |
|
Rebased and fixed the previous issues, however now for some reason |
|
It just checks stack requirements earlier (at the beginning of the block) than instruction returns |
|
Rebased. |
| TEST_F(evm_state, selfbalance) | ||
| { | ||
| accounts[msg.destination].set_balance(0x0504030201); | ||
| auto code = bytecode{} + push(1) + OP_SELFBALANCE + mstore(0) + ret(32 - 6, 6); |
There was a problem hiding this comment.
There's no argument to selfbalanace.
There was a problem hiding this comment.
But when OP_SELFBALANCE is undefined, there is an argument missing to mstore(0).
There was a problem hiding this comment.
But shouldn't it just abort because of invalid instruction?
There was a problem hiding this comment.
Only in naive implementation of EVM. Here we check stack requirements for the whole block of instructions before executing them.
There was a problem hiding this comment.
I get that, but the concept is wrong then. evmone fails with a strange error code on invalid input then?
There was a problem hiding this comment.
Not sure this concept you mean, but yes, it can return other error code.
There was a problem hiding this comment.
This suggest to me that evmone cannot handle badly formatted EVM input, which has a stack underflow.
|
Please add CHANGELOG entry. |
|
@chfast updated |
Ref https://eips.ethereum.org/EIPS/eip-1884
Need to actually add all the changes to EVMC apparently.