bug: check for address(0) from ecrecover in sig lib. also cleanup lib. #410
Conversation
| @@ -12,7 +12,8 @@ library Sig { | |||
|
|
|||
| error S(); | |||
| error V(); | |||
| error SIG_LENGTH(); | |||
There was a problem hiding this comment.
no other errors are CONST_STYLE() so remoing in favor of Length. The library same is Sig so i think context lets us know its the signature length
| /// @dev splitAndRecover should only be used if it is known that the resulting | ||
| /// verifying bit (V) will be 27 || 28. Otherwise use recover, possibly calling split first. | ||
| /// @return The recovered address | ||
| function splitAndRecover(bytes32 h, bytes memory sig) internal pure returns (address) { |
There was a problem hiding this comment.
this whole method is deleted. it was a bit of legacy that got written and never used. nothing calls this.. the contracts just use recover with split being used by the Fakes and Unit tests.
|
managed to force |
|
|
||
| _, err = s.Dep.SigFake.RecoverTest(nil, hash, vrs) | ||
|
|
||
| assert.NotNil(err) |
There was a problem hiding this comment.
if we comment out the new line that reverts on address(0) this spec will fail.
i.e -- it just returns 0x0... and err is nil here
|
|
||
| return ecrecover(h, v, r, s); | ||
| if (recovered == address(0)) { |
There was a problem hiding this comment.
if this check is not present the ...failOn0Address spec fails.
No description provided.