Flake in eth/tracers test #1345
Description
There's a weird flake in the eth/tracer tests that seems to have appeared once the test was introduced: https://github.com/ava-labs/coreth/actions/runs/18568007433/job/52934285055
It happens on ubuntu with race detection about 0.5% of the time for me.
The ultimate source is that this error is being returned on the final tracer test:
{ // Override blocknumber with block n+1 and query a blockhash (resolves issue #32175)
blockNumber: rpc.LatestBlockNumber,
call: ethapi.TransactionArgs{
From: &accounts[0].addr,
Input: newRPCBytes([]byte{
byte(vm.PUSH1), byte(genBlocks),
byte(vm.BLOCKHASH),
byte(vm.PUSH1), 0x00,
byte(vm.MSTORE),
byte(vm.PUSH1), 0x20,
byte(vm.PUSH1), 0x00,
byte(vm.RETURN),
}),
},
config: &TraceCallConfig{
BlockOverrides: ðapi.BlockOverrides{Number: (*hexutil.Big)(big.NewInt(int64(genBlocks + 1)))},
},
want: fmt.Sprintf(`{"gas":59592,"failed":false,"returnValue":"%x"}`, backend.chain.GetHeaderByNumber(uint64(genBlocks)).Hash()),
}This test was added in #1301 and can be replicated with HashDB. I'm not sure what part of this behavior is non-deterministic.