Dry run xcms on bridge hubs to get forwarded xcms across the W<>R bridge

[franciscoaguirre]

Addresses #5708
Closes: #5878

The DryRunApi did not return any forwarded xcms when used on bridge hubs.
This PR completes the story of dry running an XCM across the W<>R bridge.

The tests dry_run_transfer_to_westend_sends_xcm_to_bridge_hub and dry_run_transfer_to_rococo_sends_xcm_to_bridge_hub show how you can start from Asset Hub on one side of the bridge and get the message that will be executed on the Asset Hub on the other side.
These two tests use the test_dry_run_transfer_across_pk_bridge macro.

The most important implementation is InspectMessageQueues for pallet bridge messages.
With this new implementation we can implement this on bridge hub polkadot and kusama for the P<>K bridge.

I also added a call to clear_messages in dry_run_xcm which was missing (it was previously only added to dry_run_call).

TODO

• remove DescendOrigin when dispatching on BH and validate UniversalOrigin in the XCM message when doing dispatch on BH
• dry-run on the source BH should return final XCM as executed on destination AH + destination should be target AH

[franciscoaguirre]

bot fmt

[command-bot]

@franciscoaguirre https://gitlab.parity.io/parity/mirrors/polkadot-sdk/-/jobs/7548088 was started for your command "$PIPELINE_SCRIPTS_DIR/commands/fmt/fmt.sh". Check out https://gitlab.parity.io/parity/mirrors/polkadot-sdk/-/pipelines?page=1&scope=all&username=group_605_bot to know what else is being executed currently.

Comment bot cancel 29-a382d81c-ceb6-40ad-87d8-982dbec44101 to cancel this command or bot cancel to cancel all commands in this pull request.

[command-bot]

@franciscoaguirre Command "$PIPELINE_SCRIPTS_DIR/commands/fmt/fmt.sh" has finished. Result: https://gitlab.parity.io/parity/mirrors/polkadot-sdk/-/jobs/7548088 has finished. If any artifacts were generated, you can download them from https://gitlab.parity.io/parity/mirrors/polkadot-sdk/-/jobs/7548088/artifacts/download.

[acatangiu]

nice!

[bkontur]

@franciscoaguirre
I think we should be able to test or dry-run the full route: AHK -> BHK ... BHP -> AHP.

This way, in the end, we'll be able to write or run a periodic TypeScript/PAPI/whatever script that tests/simulates/dry-runs an asset transfer from AHK to AHP live, without the need to manually intervene (no chopsticks required).

I think the only missing piece is the DescendOrigin I mentioned here. We need to trigger some pre-dispatch, fake-message-proof-delivery, or whatever runtime API on the bridged BridgeHub, that will ultimately append the DescendOrigin. Let me think about it more.

[acatangiu]

doesn't the dispatch (and DescendOrigin appending) happen before sending to outbound HRMP queue? I mean outbound HRMP Q of BH will contain full XCM exactly as it goes to AH, so I don't see the problem

[bkontur]

well, not exactly, bridged/target BridgeHub appends to the xcm DescendOrigin when dispatching: https://github.com/paritytech/polkadot-sdk/blob/master/polkadot/xcm/xcm-builder/src/universal_exports.rs#L433-L439

so the XCM in outbound Q on sender BridgeHub is not exactly the same as it comes to the target AH

[bkontur]

but the appending DescendOrigin we can do "offchain" (in the test, PAPI call), so this PR should be enough probably

[franciscoaguirre]

I'm thinking of adding chopsticks tests for this under integration-tests/chopsticks/ but in another PR

[acatangiu]

The mechanism here is actually incomplete, and the tests only test the source side of the bridge.

We need to support the full path across all hops all the way to final destination.

[bkontur]

@franciscoaguirre

1. part - easy changes should be like:

• just to replace BridgeRococoToWestendMessagesPalletInstance with () for BridgeBlobDispatcher here:
  https://github.com/paritytech/polkadot-sdk/blob/master/cumulus/parachains/runtimes/bridge-hubs/bridge-hub-rococo/src/bridge_to_westend_config.rs#L85C53-L85C96 (for both BHs)
• remove the corespondent PalletInstance from AH's UniversalAliases check: https://github.com/paritytech/polkadot-sdk/blob/master/cumulus/parachains/runtimes/assets/asset-hub-rococo/src/xcm_config.rs#L618 (so just to check sibling BH location) (for both BHs)

2. part - important changes:

• now we need to validate and check decoded XCM somewhere on the BridgeHub in the BridgeBlobDispatcher, which does two things: 1. decoding from BridgeMessage and 2. send_xcm::<Router>, that xcm "contains" expected UniversalOrigin instruction.

Proposed solutions

This is how it looks now:

/// Dispatches received XCM messages from other bridge
type FromWestendMessageBlobDispatcher =
	BridgeBlobDispatcher<XcmRouter, UniversalLocation, BridgeRococoToWestendMessagesPalletInstance>;

1. Add filter capabilities to the *BridgeBlobDispatcher:

One possible solution would be to add some barrier/filter-like configurable capabilities to the BridgeBlobDispatcher interface/signature to allow filtering decoded XCM, e.g. Filter: Contains<Xcm<()> or maybe ShouldExecute. Something like:

pub struct BridgeBlobDispatcher<Router, OurPlace, OurPlaceBridgeInstance, Filter: Contains<Xcm<()>>(
or 
pub struct BridgeBlobDispatcher<Router, OurPlace, OurPlaceBridgeInstance, Filter: ShouldExecute>(

2. Proposed wrapper implementation with checking XCM:

There is also an easy solution, we can easily add another SendXcm wrapper implementation over Router. Something like this:

/// Dispatches received XCM messages from other bridge
type FromWestendMessageBlobDispatcher =
	BridgeBlobDispatcher<
		EnsureUniversalOrigin<XcmRouter, WestendGlobalConsensusWithNetworId>,
		UniversalLocation,
		()
	>;

// TODO: place somewhere in the xcm-builder:
pub struct EnsureUniversalOrigin;
impl SendXcm<XcmRouter: SendXcm, ExpectedUniversalOrigin: Get<Junction>> for EnsureUniversalOrigin {
	type Ticket = XcmRouter::Ticket;

	fn validate(destination: &mut Option<Location>, message: &mut Option<Xcm<()>>) -> SendResult<Self::Ticket> {

		// TODO: some validation for: ExpectedUniversalOrigin::get() and message
                // TODO: not sure how to validate `UniversalOrigin`:
                //  - check if message starts with required first instruction with expected UO?
                //  - or check recursively all the XCM messages? This could goes tricky, probably not the best solution at all.

		XcmRouter::valdiate(...)
	}

	fn deliver(ticket: Self::Ticket) -> Result<XcmHash, SendError> {
		XcmRouter::deliver(ticket)
	}
}

or maybe EnsureUniversalOrigin could be replace with some generic stuff:

pub struct FilteredSendXcm;
impl SendXcm<XcmRouter: SendXcm, Filter: Contains<Xcm>> for FilteredSendXcm {
	type Ticket = XcmRouter::Ticket;

	fn validate(destination: &mut Option<Location>, message: &mut Option<Xcm<()>>) -> SendResult<Self::Ticket> {
		
                // TODO: do some filtering
                ensure!(Filter::contains(message));

		XcmRouter::valdiate(...)
	}

        fn deliver(ticket: Self::Ticket) -> Result<XcmHash, SendError> {
		XcmRouter::deliver(ticket)
	}
}

3. other solution?