define block hash provider and default impl using frame_system

[vedhavyas]

This PR introduces BlockHashProvider into pallet_mmr::Config
This type is used to get block_hash for a given block_number rather than directly using frame_system::Pallet::block_hash

The DefaultBlockHashProvider uses frame_system::Pallet::block_hash to get the block_hash

Closes: #4062

[paritytech-cicd-pr]

The CI pipeline was cancelled due to failure one of the required jobs.
Job name: cargo-clippy
Logs: https://gitlab.parity.io/parity/mirrors/polkadot-sdk/-/jobs/5881092

[serban300]

The PR makes sense for MMR. I was also wondering if in order to make it more generic it would make sense to change frame_system::Config::BlockHashCount to a method or something more generic. Not sure. But we could explore this in a separate issue. And related to this I was curious how your BlockHashProvider implementation looks like. Are you keeping a separate item in the storage with more block_num -> block_hash items then the frame_system ? And how do you decide when to prune that ?

[vedhavyas]

I was also wondering if in order to make it more generic it would make sense to change frame_system::Config::BlockHashCount to a method or something more generic.

I'm not quiet sure how this would work for chains that do not have quick finality. Subspace for example, Runtime would not know at which point finality comes so our BlockHashCount method should ideally be able to Host function or something else for this. This seems to introduce more tricky situations on the first thought but we will need to think more on this.

But we could explore this in a separate issue.

That is fair. Thanks!

And related to this I was curious how your BlockHashProvider implementation looks like. Are you keeping a separate item in the storage with more block_num -> block_hash items then the frame_system ? And how do you decide when to prune that ?

At the moment, we have not decided on the approach of how to fetch yet but we have some options.

• Our domains, enshrined rollups, would need to know consensus hash so we have storage for these hashes and we are planning to use these but there are potential concerns that came up with this usage
• Other option is to use a Host function to fetch the canonical hash using the block_number since, I believe, we will have access to those until the finality occurs.

Overall, this PR should allow us to impl BlockHashProvider depending on which approach we take

[vedhavyas]

@serban300 seems like check is failing, IIUC, is unrelated to this PR. Is this expected ?

[serban300]

@serban300 seems like check is failing, IIUC, is unrelated to this PR. Is this expected ?

Yes, the failing checks are unrelated to the PR so far

[vedhavyas]

@acatangiu @serban300 I'm not sure why the the required test failed. Seems like the change is not related to it. I would really appreciate your help on this. Thank you!

[serban300]

test-linux-oldkernel-stable 1/3 has been having issues lately. I'll retry it