I doubt that whether the calculation of state root is incorrect.

[yangby-cryptape]

Current Behavior

I'm not an Ethereum expert.
In following explain, if I was wrong, please point that.

With my limited knowledge, I think the global state has no relation with the sender of a contract when the contract doesn't save any information about the sender into storage.

Let's check the Metadata contract:

axon/builtin-contract/metadata/contracts/metadata.sol

Lines 63 to 65 in e719adb

	function construct() public {
	highest_epoch = U64_MAX;
	}

Before 10 minutes, I have been told the solc contract is not used.
Check the following contract:

axon/core/executor/src/system_contract/metadata/mod.rs

Lines 42 to 47 in e719adb

	impl<Adapter: ExecutorAdapter + ApplyBackend> SystemContract<Adapter>
	for MetadataContract<Adapter>
	{
	const ADDRESS: H160 = METADATA_CONTRACT_ADDRESS;
	fn exec_(&self, adapter: &mut Adapter, tx: &SignedTransaction) -> TxResp {

It doesn't save the sender.

But, in the following unit test, the different senders will lead to different state root (CI checks is passed).

axon/core/run/src/tests.rs

Lines 37 to 54 in 1c6af01

	TestCase {
	chain_name: "single_node",
	config_file: "config.toml",
	chain_spec_file: "specs/single_node/chain-spec.toml",
	key_file: "debug-v1.key",
	input_genesis_hash: "0x5a4918d8cf9440b11f40047e184e4e2e0231cc449bd473d7dca1275b2ef938e1",
	genesis_state_root: "0x956e0bff5a59189447ae1bc8727ae18f7d01d0841fb9f6302c7b8675a6089763",
	genesis_receipts_root: "0x7e747618f612d08dfe54bcb67f58f13a49e8b1bafee9a8f19a3a0f7122f44d02",
	},
	TestCase {
	chain_name: "single_node",
	config_file: "config.toml",
	chain_spec_file: "specs/single_node/chain-spec.toml",
	key_file: "debug-v2.key",
	input_genesis_hash: "0x5a4918d8cf9440b11f40047e184e4e2e0231cc449bd473d7dca1275b2ef938e1",
	genesis_state_root: "0x9d7fa055987ed34353ec84465d98b9e1bd6f3af871c36bd2d0d3f44e45170d27",
	genesis_receipts_root: "0x7e747618f612d08dfe54bcb67f58f13a49e8b1bafee9a8f19a3a0f7122f44d02",
	},

So, I doubt that whether the calculation of state root is incorrect.

p.s. The accounts of both two senders used in the unit test are empty, so no state for their balance should be existed.

Expected Behavior

The state root should be the same if only the sender is different but the sander is not saved into state.

[yangby-cryptape]

The nonce for the sender will be increased, that why the state roots are different.

[yangby-cryptape]

There are still some strange logs that need to be investigated, but they are not part of this issue.
If I found any further results, I will create another issue.