fix: Global min Self Delegation

[AlpinYukseloglu]

Closes: #113

What is the purpose of the change

Our current implementation of Superfluid Staking exposes the chain to a specific vulnerability relating to the validator creation process. This PR adds the necessary guardrails to make sure that an attacker cannot halt the chain by generating an arbitrary number of empty validators.

This was not an issue prior to Superfluid Staking because the full validator set was not used in the ante anywhere, but our specific implementation of Superfluid introduced a linear-time search of the validator set at epoch end that made the issue fixed in this PR a vulnerability.

[mattverse]

Would it be able to add tests for this change?(genesis_test, params_test in sim etc)

Also, how does this param enforce validators to have min_self_delegation?

[ValarDragon]

@AlpinYukseloglu you have to update the GetParams queries for the new field in the param struct before merge.

To test locally, inside of x/staking do go test ./..., the query tests should fail.

[ValarDragon]

https://github.com/osmosis-labs/cosmos-sdk/runs/5564114018?check_suite_focus=true

still failing

[AlpinYukseloglu]

https://github.com/osmosis-labs/cosmos-sdk/runs/5564114018?check_suite_focus=true

still failing

Interesting... local tests seem to be passing (go test ./... in x/staking and then also in its parent folders) except for when I do it in the highest folder (cosmos-sdk), in which case the handful of tests that fail are all ledger-related ("support for ledger devices is not available in this executable"). Am I testing incorrectly @ValarDragon?

[mattverse]

FYI: The Rosetta test was broken in the main sdk branch as well

[mattverse]

Actually, it seems that something in this PR does seem to break the rosetta-test that needs further investigation on fixes. Also, by a brief look, it seems that the "build" action might be failing due to changes in go.sum, can you revert go.sum into the original state?

[ValarDragon]

TODO: Ask in cosmos SDK discord how rosetta tests work / how to debug this failure.

If theres a genesis file somewhere for Rosetta, we did break it / would have to update it for this PR

[alexanderbez]

Frojdi told me to look at this function:

// ToRosetta attempts to converting an error into a rosetta
// error, if the error cannot be converted it will be parsed as unknown
func ToRosetta(err error) *types.Error {
    if err == nil {
        return nil
    }
    // if it's null or not known
    rosErr, ok := err.(*Error)
    if !ok {
        return ToRosetta(WrapError(ErrUnknown, err.Error()))
    }
    return rosErr.rosErr
}

I think we need to log err in the beginning of the function to get better insight into what's going on.

[alexanderbez]

🚀 🚀 🚀