Self Destruct's doesn't return the correct balance

[0xOmarA]

Some of the ML tests for self-destruct show that the instruction doesn't return the correct balance to the desired accounts.

Tests failing with this issue are:

• translated_semantic_tests/various/selfdestruct_post_cancun/test.json::0::E-
• translated_semantic_tests/various/selfdestruct_post_cancun_multiple_beneficiaries/test.json::0::E-

[0xRVE]

👀

[0xRVE]

For some reason the existential deposit is not refunded to the origin, but in stead it is transferred to beneficiary. Still trying to find out why this happens.

[0xRVE]

fixes #209

These three lines make it so that the existential deposit is transferred as regular balance (as opposed to deposit).
https://github.com/paritytech/polkadot-sdk/blob/ef07f24b0cd49a52d1a5a0f3a42ea8d714187f18/substrate/frame/revive/src/exec.rs#L1225

let ed = <Contracts<T>>::min_balance();
frame.nested_storage.record_charge(&StorageDeposit::Charge(ed))?;
<Contracts<T>>::charge_deposit(None, origin, account_id, ed, self.exec_config)
	.map_err(|_| <Error<T>>::StorageDepositNotEnoughFunds)?;

Because of this the beneficiary receives 1 ether + ED which fails our assertion (we assert that the beneficiary received 1 ether). NB: this only happens if the contract is destroyed in the same transaction as it was created.
So I simply changed the assertion to 1 ether + ED.

Another issue is these two tests:

    function test_create_and_terminate() public {
        deploy_create();
        assert(exists());
        test_balance_after_create();
        terminate();
        test_balance_after_selfdestruct();
    }

    function test_create2_and_terminate() public {
        deploy_create2();
        assert(exists());
        test_balance_after_create();
        terminate();
        test_balance_after_selfdestruct();
    }

At the time we call test_balance_after_selfdestruct the contract has not been destroyed yet because the transaction is still going.
I removed this type of test and check the balances in a separate transaction.

[0xRVE]

@athei will change behaviour of selfdestruct that will make these tests pass

[0xOmarA]

@0xRVE Why was it moved to WontFix? My understanding is that it will be fixed before the release but that it will be fixed with the changes alex will make?

[pgherveou]

it's kind of expected to receive ED more right, are we changing this behavior @0xRVE ?

[0xRVE]

I moved it to "wont fix" because @athei will fix it in his upcoming PR

paritytech/resolc-compiler-tests#5 (comment)