-
Notifications
You must be signed in to change notification settings - Fork 1.6k
Decrease max_code_size
on chain
#7108
Comments
I snug a change into #6271 since it already schedules a config upgrade in the runtime upgrade. About the value: it was 4MiB previously, but i can do a sweep across the parachains to check the current max. The migration output looks like this
|
3 MiB would also be fine it seems, although the previous was 4. Current max across all paras is 1.70MiB. We should probably announce this in the Forum before proceeding. Parachain Code Size
(Chains with |
Would you like me to make a Polkadot forum post about this? It seems pretty straightforward. |
What you mean by previously?
Not sure we need a forum post for this, mainly requires a governance proposal to set the value. IMO that should be enough. |
Ah nvm. I was a local change of mine 🤦♂️
Just wanted to ensure that all the para-chain teams see it - just in case that they have a pending upgrade that would get invalidated. But a proposal should probably be enough. Can we still get feedback from the @paritytech/parachains-core team on this? |
Ahh yeah, good point. Anything above 5MiB would lead to issues right now any way. |
We (@eskimor and I) are already tagged here :) |
The 3MB limit is also enforced at the network level. Hence any runtime upgrade greater than 3MB could hardly make it anyway already. (It could technically if the backing validator is the block producer at the same time.) |
Re-addig these checks is blocked on #7108 Signed-off-by: Oliver Tale-Yazdi <[email protected]>
Proposal is here https://kusama.polkassembly.io/referenda/173 |
Kusama executed successfully. For Polkadot we can probably wait til OpenGov, to not block the current Referendum queue since it is not time-critical. |
I have always imagined some extrinsic which would append data to some storage over multiple blocks, allowing some single file stored on chain greater than the block limit. In this case, a chain could make multiple upload extrinsics to the preimage pallet, and then trigger an upgrade from the larger file, bypassing the per-block limit. |
I think we can do this simply with We could also try to change the format of the proof, to kind of do a matching against known values. Then it could realize that the written data is actually already in storage and can reference it instead of copying. But that would require some reverse lookup from value->key, or some matching of read+write ops. Sounds very hacky. |
* Add ProcessXcmMessage struct * Migrate away from weights in host config * New well-known key to report UMPQ capacity * Add missing file * Fixes * Remove original UMP files * Docs * Update runtime/parachains/src/inclusion/mod.rs Co-authored-by: asynchronous rob <[email protected]> * Add benchmarking * Benchmarks * Mock example of using the QueueChangeHandler to update the WKK * Use master Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Merge remote-tracking branch 'origin/master' into gav-message-queue * Update Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update remove-weight migration The migration got touched on master; just resolving conflicts here. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add message- to dispatch-origin conversion for XCM processing Just using the `impl Into<MultiLocation>` was a bit inflexible. Like this, the Relaychain can convert `UMP(para)` to a MultiLocation `para`. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * DNM: Temporarily comment code since XCMv3 is not merged yet Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use u64 for queue-wide limits on UmpAcceptanceCheckErr Using u32 here was one audit finding for the queue pallet. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Define one sub-queue per *MP queue Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Harden check_upward_messages Using safe math and casts. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add type-safe well_known_keys Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add message-queue weights Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Deploy MessageQueue to Polkadot Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update Cargo.toml Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Migrate to parachain config V5 Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update UMP tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert messed up merge 🤦 Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update remove-weight migration The migration got touched on master; just resolving conflicts here. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add message- to dispatch-origin conversion for XCM processing Just using the `impl Into<MultiLocation>` was a bit inflexible. Like this, the Relaychain can convert `UMP(para)` to a MultiLocation `para`. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * DNM: Temporarily comment code since XCMv3 is not merged yet Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use u64 for queue-wide limits on UmpAcceptanceCheckErr Using u32 here was one audit finding for the queue pallet. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Define one sub-queue per *MP queue Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Harden check_upward_messages Using safe math and casts. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add type-safe well_known_keys Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add message-queue weights Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Deploy MessageQueue to Polkadot Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update Cargo.toml Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Migrate to parachain config V5 Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update UMP tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Make stuff compile Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Move DMP and HRMP messages to the MessageQueue It currently does not compile in the CIbecause of some local tweaks to Substrate. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Diener for CI Signed-off-by: Oliver Tale-Yazdi <[email protected]> * diener update cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert wrong changes Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "DNM: Temporarily comment code since XCMv3 is not merged yet" This reverts commit 130a48a. * Make compile Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixup runtimes Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Define benchmarks Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use master Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Lockfile Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix test Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add AggregateMessageOrigin This enum currently only holds one value, but having it will make it easier in the future to extend. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Forbid UMP for off-boarding paras - Reject candidates with UMP messages for off-boarding paras - Forbid scheduling off-boarding when a para has unprocess UMPs Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Delete stupid test Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use BoundedVec for upward messages Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add weights and fix MessageProcessor Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Bound receive_upward_messages and check bound in configuration pallet Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Bound Debug impl Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * clippy Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix test runtime Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix xcm-simulator Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Properly fix xcm-simulator and fuzzer Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * cargo update -p sp-io Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Adapt to upstream Substrate changes Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix ProcesseMessage impls Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Some tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use master Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * cargo update -p sp-io Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use new MQ API Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix test Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add UMP while Para offboarding tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use Mocked message processor for benchmarking Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use variables for constants Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add MQ pallet weights Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use MQ pallet weights Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Configure QueueChangeHandler Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add config test Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix MQ serive weight Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup outgoing UMP dispatch queues Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use Master Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Weight mul is not const Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Clippy Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove merge marker Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update runtime/parachains/src/inclusion/mod.rs Co-authored-by: Gavin Wood <[email protected]> * Update runtime/kusama/src/lib.rs Co-authored-by: Gavin Wood <[email protected]> * Use lowercase UMP Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Clarify comment Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use Weight::from_parts Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix test Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix doc Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Emit event after the fact Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add defensive_proof to receive_upward_messages Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Reapply "Remove original UMP files" Looks like they came back from the dead. Re-apply commit 34b8a40 Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove old files Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Rename MaxUmpMessageLen -> MaxUmpMessageLenOf Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Test defensive message dropping of receive_upward_messages Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixup imports Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update implementors guide Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove FAIL-CI mark Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Delete unused code Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add another test for MQ change hook Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Imports Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Keep Kusama runtime formatting Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "Delete unused code" This reverts commit dd76bca. * Feature gate mock functions Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Review: Use saturating_add Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Test RelayDispatchQueueSize storage key Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Move migration to own file to avoid merge conflicts Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Migration in own file Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixup migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Rococo: configure MQ pallet Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixup tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "Rococo: configure MQ pallet" Going to do this as follow up, since it needs Substrate changes and i dont want to stall this MR any longer. This reverts commit b9c15e8. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "Fixup tests" This reverts commit 88f1cbe. * Fixup migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix CI Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix other migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Bump MAX_CODE_SIZE to 10MiB Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add ForceUpdateUmpLimits migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * clippy Signed-off-by: Oliver Tale-Yazdi <[email protected]> * clippy Signed-off-by: Oliver Tale-Yazdi <[email protected]> * imports Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use defensive instead of defensive_proof 'defensive_proof' also prints the 'self', which spams the console too much when running the tests. Just the length is enough. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Rename to ScheduleConfigUpdate Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixup migration checks Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add MAX_CODE_SIZE to ScheduleConfigUpdate Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Set MAX_CODE_SIZE to 4MiB Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix benchmark Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix formatting Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "Add MAX_CODE_SIZE to ScheduleConfigUpdate" This reverts commit 7caffb0. * Revert "Set MAX_CODE_SIZE to 4MiB" This reverts commit 103ffba. * Revert "Bump MAX_CODE_SIZE to 10MiB" This reverts commit 530734b. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove consistency check from migration Re-addig these checks is blocked on #7108 Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix constants Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Bump MAX_UPWARD_MESSAGE_SIZE_BOUND for Westend Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix migrations Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use old nightly for fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixes * cargo fmt * Fix tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixes * Add MQ pallet to fuzzer Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix XMC simulator example Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove runtime-benchmarks from fuzzers Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "Remove runtime-benchmarks from fuzzers" This reverts commit e1f2bb0. * Fix example simulator Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add V6 migration and remove old ones Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Actually make old migrations reusable Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Dont delete old migrations Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Future proof AggregateMessageOrigin and review fixes There are indications that Loopback and Bridged will be needed soon. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * More cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fix benchmarks Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix fuzzer build Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Review Co-authored-by: muharem <[email protected]> Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove old migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Set MQ service weight to 20% Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix tabs in Markdown Signed-off-by: Oliver Tale-Yazdi <[email protected]> --------- Signed-off-by: Oliver Tale-Yazdi <[email protected]> Co-authored-by: asynchronous rob <[email protected]> Co-authored-by: Oliver Tale-Yazdi <[email protected]> Co-authored-by: Keith Yeung <[email protected]>
In this case, the limit is not the block size, but the PoV size. For large items in PoV, we have this issue open which can solve that also using the preimage pallet: paritytech/polkadot-sdk#811 Yes, some engineering required, but I don't see a fundamental limitation here yet. |
Polkadot proposal is here https://polkadot.polkassembly.io/referenda/9 |
* Add ProcessXcmMessage struct * Migrate away from weights in host config * New well-known key to report UMPQ capacity * Add missing file * Fixes * Remove original UMP files * Docs * Update runtime/parachains/src/inclusion/mod.rs Co-authored-by: asynchronous rob <[email protected]> * Add benchmarking * Benchmarks * Mock example of using the QueueChangeHandler to update the WKK * Use master Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Merge remote-tracking branch 'origin/master' into gav-message-queue * Update Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update remove-weight migration The migration got touched on master; just resolving conflicts here. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add message- to dispatch-origin conversion for XCM processing Just using the `impl Into<MultiLocation>` was a bit inflexible. Like this, the Relaychain can convert `UMP(para)` to a MultiLocation `para`. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * DNM: Temporarily comment code since XCMv3 is not merged yet Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use u64 for queue-wide limits on UmpAcceptanceCheckErr Using u32 here was one audit finding for the queue pallet. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Define one sub-queue per *MP queue Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Harden check_upward_messages Using safe math and casts. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add type-safe well_known_keys Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add message-queue weights Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Deploy MessageQueue to Polkadot Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update Cargo.toml Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Migrate to parachain config V5 Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update UMP tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert messed up merge 🤦 Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update remove-weight migration The migration got touched on master; just resolving conflicts here. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add message- to dispatch-origin conversion for XCM processing Just using the `impl Into<MultiLocation>` was a bit inflexible. Like this, the Relaychain can convert `UMP(para)` to a MultiLocation `para`. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * DNM: Temporarily comment code since XCMv3 is not merged yet Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use u64 for queue-wide limits on UmpAcceptanceCheckErr Using u32 here was one audit finding for the queue pallet. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Define one sub-queue per *MP queue Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Harden check_upward_messages Using safe math and casts. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add type-safe well_known_keys Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add message-queue weights Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Deploy MessageQueue to Polkadot Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update Cargo.toml Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Migrate to parachain config V5 Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update UMP tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Make stuff compile Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Move DMP and HRMP messages to the MessageQueue It currently does not compile in the CIbecause of some local tweaks to Substrate. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Diener for CI Signed-off-by: Oliver Tale-Yazdi <[email protected]> * diener update cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert wrong changes Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "DNM: Temporarily comment code since XCMv3 is not merged yet" This reverts commit 820aa235cb21dd1d2621843607f7682bf035434e. * Make compile Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixup runtimes Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Define benchmarks Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use master Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Lockfile Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix test Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add AggregateMessageOrigin This enum currently only holds one value, but having it will make it easier in the future to extend. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Forbid UMP for off-boarding paras - Reject candidates with UMP messages for off-boarding paras - Forbid scheduling off-boarding when a para has unprocess UMPs Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Delete stupid test Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use BoundedVec for upward messages Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add weights and fix MessageProcessor Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Bound receive_upward_messages and check bound in configuration pallet Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Bound Debug impl Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * clippy Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix test runtime Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix xcm-simulator Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Properly fix xcm-simulator and fuzzer Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * cargo update -p sp-io Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Adapt to upstream Substrate changes Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix ProcesseMessage impls Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Some tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use master Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * cargo update -p sp-io Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use new MQ API Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix test Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add UMP while Para offboarding tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use Mocked message processor for benchmarking Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use variables for constants Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add MQ pallet weights Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use MQ pallet weights Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Configure QueueChangeHandler Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add config test Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix MQ serive weight Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Cleanup outgoing UMP dispatch queues Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use Master Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update Cargo.lock Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Weight mul is not const Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Clippy Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove merge marker Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update runtime/parachains/src/inclusion/mod.rs Co-authored-by: Gavin Wood <[email protected]> * Update runtime/kusama/src/lib.rs Co-authored-by: Gavin Wood <[email protected]> * Use lowercase UMP Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Clarify comment Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use Weight::from_parts Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix test Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix doc Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Emit event after the fact Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add defensive_proof to receive_upward_messages Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Reapply "Remove original UMP files" Looks like they came back from the dead. Re-apply commit cf6d316f0 Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove old files Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Rename MaxUmpMessageLen -> MaxUmpMessageLenOf Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Test defensive message dropping of receive_upward_messages Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixup imports Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Update implementors guide Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove FAIL-CI mark Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Delete unused code Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add another test for MQ change hook Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Imports Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Keep Kusama runtime formatting Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "Delete unused code" This reverts commit dd76bca5025b7e1ef846a9539c3607eed185f16a. * Feature gate mock functions Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Review: Use saturating_add Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Test RelayDispatchQueueSize storage key Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Move migration to own file to avoid merge conflicts Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Migration in own file Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixup migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Rococo: configure MQ pallet Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixup tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "Rococo: configure MQ pallet" Going to do this as follow up, since it needs Substrate changes and i dont want to stall this MR any longer. This reverts commit b9c15e8a8339c4e877d654ee3f09903af4210736. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "Fixup tests" This reverts commit 88f1cbe20774d20e5e1e554e798960ae39437af1. * Fixup migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix CI Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix other migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Bump MAX_CODE_SIZE to 10MiB Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add ForceUpdateUmpLimits migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * clippy Signed-off-by: Oliver Tale-Yazdi <[email protected]> * clippy Signed-off-by: Oliver Tale-Yazdi <[email protected]> * imports Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use defensive instead of defensive_proof 'defensive_proof' also prints the 'self', which spams the console too much when running the tests. Just the length is enough. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Rename to ScheduleConfigUpdate Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixup migration checks Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add MAX_CODE_SIZE to ScheduleConfigUpdate Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Set MAX_CODE_SIZE to 4MiB Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix benchmark Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix formatting Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "Add MAX_CODE_SIZE to ScheduleConfigUpdate" This reverts commit 7caffb09e83083b57affd548215e45b25c3d64dc. * Revert "Set MAX_CODE_SIZE to 4MiB" This reverts commit 103ffbaf686487d2fbe0082a16826af17cacc1a1. * Revert "Bump MAX_CODE_SIZE to 10MiB" This reverts commit 530734b7b0da5b7680054e0242348fcc79a666fe. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove consistency check from migration Re-addig these checks is blocked on paritytech/polkadot#7108 Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix constants Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Bump MAX_UPWARD_MESSAGE_SIZE_BOUND for Westend Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix migrations Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Use old nightly for fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixes * cargo fmt * Fix tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix tests Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fixes * Add MQ pallet to fuzzer Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix XMC simulator example Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove runtime-benchmarks from fuzzers Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Revert "Remove runtime-benchmarks from fuzzers" This reverts commit e1f2bb01b6dea2dd465539d3658719895b58b557. * Fix example simulator Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Add V6 migration and remove old ones Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Actually make old migrations reusable Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Dont delete old migrations Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Future proof AggregateMessageOrigin and review fixes There are indications that Loopback and Bridged will be needed soon. Signed-off-by: Oliver Tale-Yazdi <[email protected]> * More cleanup Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fmt Signed-off-by: Oliver Tale-Yazdi <[email protected]> * fix benchmarks Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix fuzzer build Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Review Co-authored-by: muharem <[email protected]> Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Remove old migration Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Set MQ service weight to 20% Signed-off-by: Oliver Tale-Yazdi <[email protected]> * Fix tabs in Markdown Signed-off-by: Oliver Tale-Yazdi <[email protected]> --------- Signed-off-by: Oliver Tale-Yazdi <[email protected]> Co-authored-by: asynchronous rob <[email protected]> Co-authored-by: Oliver Tale-Yazdi <[email protected]> Co-authored-by: Keith Yeung <[email protected]>
Polkadot executed https://polkadot.subsquare.io/referenda/referendum/9 successfully. |
The
max_code_size
is currently set to10MiB
in Kusama and Polkadot. In the node we are currently assume that theMAX_CODE_SIZE
is3MiB
.As discussed with @eskimor on Tuesday, the
max_code_size
can not be bigger than the maximum block length/size of Polkadot itself. Otherwise we could not build a block that includes a parachain upgrade which code is bigger than the maximum block size.We should check all the current active PVF and their size. I would propose we downgrade the
max_code_size
on Polkadot/Kusama to4MiB
or maybe3MiB
depending on what the biggest active PVF is.CC @Sophia-Gold
The text was updated successfully, but these errors were encountered: