[FRAME] Runtime Omni Bencher (paritytech#3512)

This MR contains two major changes and some maintenance cleanup.  

## 1. Free Standing Pallet Benchmark Runner

Closes paritytech#3045, depends
on your runtime exposing the `GenesisBuilderApi` (like
paritytech#1492).

Introduces a new binary crate: `frame-omni-bencher`.  
It allows to directly benchmark a WASM blob - without needing a node or
chain spec.

This makes it much easier to generate pallet weights and should allow us
to remove bloaty code from the node.
It should work for all FRAME runtimes that dont use 3rd party host calls
or non `BlakeTwo256` block hashing (basically all polkadot parachains
should work).

It is 100% backwards compatible with the old CLI args, when the `v1`
compatibility command is used. This is done to allow for forwards
compatible addition of new commands.

### Example (full example in the Rust docs)

Installing the CLI:
```sh
cargo install --locked --path substrate/utils/frame/omni-bencher
frame-omni-bencher --help
```

Building the Westend runtime:
```sh
cargo build -p westend-runtime --release --features runtime-benchmarks
```

Benchmarking the runtime:
```sh
frame-omni-bencher v1 benchmark pallet --runtime target/release/wbuild/westend-runtime/westend_runtime.compact.compressed.wasm --all
```

## 2. Building the Benchmark Genesis State in the Runtime

Closes paritytech#2664

This adds `--runtime` and `--genesis-builder=none|runtime|spec`
arguments to the `benchmark pallet` command to make it possible to
generate the genesis storage by the runtime. This can be used with both
the node and the freestanding benchmark runners. It utilizes the new
`GenesisBuilder` RA and depends on having
paritytech#3412 deployed.

## 3. Simpler args for `PalletCmd::run`

You can do three things here to integrate the changes into your node:
- nothing: old code keeps working as before but emits a deprecated
warning
- delete: remove the pallet benchmarking code from your node and use the
omni-bencher instead
- patch: apply the patch below and keep using as currently. This emits a
deprecated warning at runtime, since it uses the old way to generate a
genesis state, but is the smallest change.

```patch
runner.sync_run(|config| cmd
-    .run::<HashingFor<Block>, ReclaimHostFunctions>(config)
+    .run_with_spec::<HashingFor<Block>, ReclaimHostFunctions>(Some(config.chain_spec))
)
```

## 4. Maintenance Change
- `pallet-nis` get a `BenchmarkSetup` config item to prepare its
counterparty asset.
- Add percent progress print when running benchmarks.
- Dont immediately exit on benchmark error but try to run as many as
possible and print errors last.

---------

Signed-off-by: Oliver Tale-Yazdi <oliver.tale-yazdi@parity.io>
Co-authored-by: Liam Aharon <liam.aharon@hotmail.com>

.gitlab/pipeline/check.yml

@@ -7,8 +7,8 @@ cargo-clippy:
   variables:
     RUSTFLAGS: "-D warnings"
   script:
-    - SKIP_WASM_BUILD=1 cargo clippy --all-targets --locked --workspace
-    - SKIP_WASM_BUILD=1 cargo clippy --all-targets --all-features --locked --workspace
+    - SKIP_WASM_BUILD=1 cargo clippy --all-targets --locked --workspace --quiet
+    - SKIP_WASM_BUILD=1 cargo clippy --all-targets --all-features --locked --workspace --quiet
 
 check-try-runtime:
   stage: check

.gitlab/pipeline/test.yml

@@ -322,7 +322,24 @@ quick-benchmarks:
     WASM_BUILD_NO_COLOR: 1
     WASM_BUILD_RUSTFLAGS: "-C debug-assertions -D warnings"
   script:
-    - time cargo run --locked --release -p staging-node-cli --bin substrate-node --features runtime-benchmarks -- benchmark pallet --execution wasm --wasm-execution compiled --chain dev --pallet "*" --extrinsic "*" --steps 2 --repeat 1
+    - time cargo run --locked --release -p staging-node-cli --bin substrate-node --features runtime-benchmarks --quiet -- benchmark pallet --chain dev --pallet "*" --extrinsic "*" --steps 2 --repeat 1 --quiet
+
+quick-benchmarks-omni:
+  stage: test
+  extends:
+    - .docker-env
+    - .common-refs
+    - .run-immediately
+  variables:
+    # Enable debug assertions since we are running optimized builds for testing
+    # but still want to have debug assertions.
+    RUSTFLAGS: "-C debug-assertions"
+    RUST_BACKTRACE: "full"
+    WASM_BUILD_NO_COLOR: 1
+    WASM_BUILD_RUSTFLAGS: "-C debug-assertions"
+  script:
+    - time cargo build --locked --quiet --release -p asset-hub-westend-runtime --features runtime-benchmarks
+    - time cargo run --locked --release -p frame-omni-bencher --quiet -- v1 benchmark pallet --runtime target/release/wbuild/asset-hub-westend-runtime/asset_hub_westend_runtime.compact.compressed.wasm --all --steps 2 --repeat 1 --quiet
 
 test-frame-examples-compile-to-wasm:
   # into one job

Cargo.toml

@@ -492,6 +492,7 @@ members = [
 	"substrate/utils/frame/frame-utilities-cli",
 	"substrate/utils/frame/generate-bags",
 	"substrate/utils/frame/generate-bags/node-runtime",
+	"substrate/utils/frame/omni-bencher",
 	"substrate/utils/frame/remote-externalities",
 	"substrate/utils/frame/rpc/client",
 	"substrate/utils/frame/rpc/state-trie-migration-rpc",

cumulus/polkadot-parachain/src/command.rs

@@ -568,7 +568,7 @@ pub fn run() -> Result<()> {
 			match cmd {
 				BenchmarkCmd::Pallet(cmd) =>
 					if cfg!(feature = "runtime-benchmarks") {
-						runner.sync_run(|config| cmd.run::<sp_runtime::traits::HashingFor<Block>, ReclaimHostFunctions>(config))
+						runner.sync_run(|config| cmd.run_with_spec::<sp_runtime::traits::HashingFor<Block>, ReclaimHostFunctions>(Some(config.chain_spec)))
 					} else {
 						Err("Benchmarking wasn't enabled when building the node. \
 				You can enable it with `--features runtime-benchmarks`."

polkadot/cli/src/command.rs

@@ -451,8 +451,10 @@ pub fn run() -> Result<()> {
 
 					if cfg!(feature = "runtime-benchmarks") {
 						runner.sync_run(|config| {
-							cmd.run::<sp_runtime::traits::HashingFor<service::Block>, ()>(config)
-								.map_err(|e| Error::SubstrateCli(e))
+							cmd.run_with_spec::<sp_runtime::traits::HashingFor<service::Block>, ()>(
+								Some(config.chain_spec),
+							)
+							.map_err(|e| Error::SubstrateCli(e))
 						})
 					} else {
 						Err(sc_cli::Error::Input(

polkadot/runtime/common/src/auctions.rs

@@ -1899,10 +1899,13 @@ mod benchmarking {
 
 			// Trigger epoch change for new random number value:
 			{
+				pallet_babe::EpochStart::<T>::set((Zero::zero(), u32::MAX.into()));
 				pallet_babe::Pallet::<T>::on_initialize(duration + now + T::EndingPeriod::get());
 				let authorities = pallet_babe::Pallet::<T>::authorities();
-				let next_authorities = authorities.clone();
-				pallet_babe::Pallet::<T>::enact_epoch_change(authorities, next_authorities, None);
+				// Check for non empty authority set since it otherwise emits a No-OP warning.
+				if !authorities.is_empty() {
+					pallet_babe::Pallet::<T>::enact_epoch_change(authorities.clone(), authorities, None);
+				}
 			}
 
 		}: {

polkadot/runtime/parachains/src/configuration.rs

@@ -231,7 +231,7 @@ pub struct HostConfiguration<BlockNumber> {
 
 impl<BlockNumber: Default + From<u32>> Default for HostConfiguration<BlockNumber> {
 	fn default() -> Self {
-		Self {
+		let ret = Self {
 			async_backing_params: AsyncBackingParams {
 				max_candidate_depth: 0,
 				allowed_ancestry_len: 0,
@@ -270,7 +270,30 @@ impl<BlockNumber: Default + From<u32>> Default for HostConfiguration<BlockNumber
 			minimum_backing_votes: LEGACY_MIN_BACKING_VOTES,
 			node_features: NodeFeatures::EMPTY,
 			scheduler_params: Default::default(),
-		}
+		};
+
+		#[cfg(feature = "runtime-benchmarks")]
+		let ret = ret.with_benchmarking_default();
+		ret
+	}
+}
+
+#[cfg(feature = "runtime-benchmarks")]
+impl<BlockNumber: Default + From<u32>> HostConfiguration<BlockNumber> {
+	/// Mutate the values of self to be good estimates for benchmarking.
+	///
+	/// The values do not need to be worst-case, since the benchmarking logic extrapolates. They
+	/// should be a bit more than usually expected.
+	fn with_benchmarking_default(mut self) -> Self {
+		self.max_head_data_size = self.max_head_data_size.max(1 << 20);
+		self.max_downward_message_size = self.max_downward_message_size.max(1 << 16);
+		self.hrmp_channel_max_capacity = self.hrmp_channel_max_capacity.max(1000);
+		self.hrmp_channel_max_message_size = self.hrmp_channel_max_message_size.max(1 << 16);
+		self.hrmp_max_parachain_inbound_channels =
+			self.hrmp_max_parachain_inbound_channels.max(100);
+		self.hrmp_max_parachain_outbound_channels =
+			self.hrmp_max_parachain_outbound_channels.max(100);
+		self
 	}
 }
 

polkadot/runtime/parachains/src/hrmp/benchmarking.rs

@@ -76,6 +76,8 @@ where
 	let deposit: BalanceOf<T> = config.hrmp_sender_deposit.unique_saturated_into();
 	let capacity = config.hrmp_channel_max_capacity;
 	let message_size = config.hrmp_channel_max_message_size;
+	assert!(message_size > 0, "Invalid genesis for benchmarking");
+	assert!(capacity > 0, "Invalid genesis for benchmarking");
 
 	let sender: ParaId = from.into();
 	let sender_origin: crate::Origin = from.into();

polkadot/runtime/parachains/src/paras/benchmarking.rs

@@ -82,7 +82,7 @@ fn generate_disordered_actions_queue<T: Config>() {
 
 benchmarks! {
 	force_set_current_code {
-		let c in 1 .. MAX_CODE_SIZE;
+		let c in MIN_CODE_SIZE .. MAX_CODE_SIZE;
 		let new_code = ValidationCode(vec![0; c as usize]);
 		let para_id = ParaId::from(c as u32);
 		CurrentCodeHash::<T>::insert(&para_id, new_code.hash());
@@ -92,7 +92,7 @@ benchmarks! {
 		assert_last_event::<T>(Event::CurrentCodeUpdated(para_id).into());
 	}
 	force_set_current_head {
-		let s in 1 .. MAX_HEAD_DATA_SIZE;
+		let s in MIN_CODE_SIZE .. MAX_HEAD_DATA_SIZE;
 		let new_head = HeadData(vec![0; s as usize]);
 		let para_id = ParaId::from(1000);
 	}: _(RawOrigin::Root, para_id, new_head)
@@ -104,7 +104,7 @@ benchmarks! {
 		let context = BlockNumberFor::<T>::from(1000u32);
 	}: _(RawOrigin::Root, para_id, context)
 	force_schedule_code_upgrade {
-		let c in 1 .. MAX_CODE_SIZE;
+		let c in MIN_CODE_SIZE .. MAX_CODE_SIZE;
 		let new_code = ValidationCode(vec![0; c as usize]);
 		let para_id = ParaId::from(c as u32);
 		let block = BlockNumberFor::<T>::from(c);
@@ -114,7 +114,7 @@ benchmarks! {
 		assert_last_event::<T>(Event::CodeUpgradeScheduled(para_id).into());
 	}
 	force_note_new_head {
-		let s in 1 .. MAX_HEAD_DATA_SIZE;
+		let s in MIN_CODE_SIZE .. MAX_HEAD_DATA_SIZE;
 		let para_id = ParaId::from(1000);
 		let new_head = HeadData(vec![0; s as usize]);
 		let old_code_hash = ValidationCode(vec![0]).hash();
@@ -126,7 +126,7 @@ benchmarks! {
 		generate_disordered_pruning::<T>();
 		Pallet::<T>::schedule_code_upgrade(
 			para_id,
-			ValidationCode(vec![0]),
+			ValidationCode(vec![0u8; MIN_CODE_SIZE as usize]),
 			expired,
 			&config,
 			UpgradeStrategy::SetGoAheadSignal,
@@ -145,7 +145,7 @@ benchmarks! {
 	}
 
 	add_trusted_validation_code {
-		let c in 1 .. MAX_CODE_SIZE;
+		let c in MIN_CODE_SIZE .. MAX_CODE_SIZE;
 		let new_code = ValidationCode(vec![0; c as usize]);
 
 		pvf_check::prepare_bypassing_bench::<T>(new_code.clone());

polkadot/runtime/rococo/src/lib.rs

@@ -1154,6 +1154,8 @@ impl pallet_nis::Config for Runtime {
 	type MaxIntakeWeight = MaxIntakeWeight;
 	type ThawThrottle = ThawThrottle;
 	type RuntimeHoldReason = RuntimeHoldReason;
+	#[cfg(feature = "runtime-benchmarks")]
+	type BenchmarkSetup = ();
 }
 
 parameter_types! {

prdoc/pr_3512.prdoc

@@ -0,0 +1,47 @@
+title: "[FRAME] Introduce Runtime Omni Bencher"
+
+doc:
+  - audience: Node Dev
+    description: |
+      Introduces a new freestanding binary; the `frame-omni-bencher`. This can be used as alternative to the node-integrated `benchmark pallet` command. It currently only supports pallet benchmarking.
+
+      The optional change to integrate this MR is in the node. The `run` function is now deprecated in favour or `run_with_spec`. This should be rather easy to integrate:
+
+      ```patch
+      runner.sync_run(|config| cmd
+      -    .run::<HashingFor<Block>, ReclaimHostFunctions>(config)
+      +    .run_with_spec::<HashingFor<Block>, ReclaimHostFunctions>(Some(config.chain_spec))
+      )
+      ```
+
+      Additionally, a new `--runtime` CLI arg was introduced to the `benchmark pallet` command. It allows to generate the genesis state directly by the runtime, instead of using a spec file.
+
+crates:
+  - name: pallet-nis
+    bump: major
+  - name: frame-benchmarking-cli
+    bump: major
+  - name: polkadot-parachain-bin
+    bump: patch
+  - name: polkadot-cli
+    bump: patch
+  - name: polkadot-runtime-common
+    bump: patch
+  - name: polkadot-runtime-parachains
+    bump: patch
+  - name: rococo-runtime
+    bump: patch
+  - name: staging-node-cli
+    bump: patch
+  - name: kitchensink-runtime
+    bump: patch
+  - name: pallet-babe
+    bump: patch
+  - name: frame-benchmarking
+    bump: patch
+  - name: pallet-election-provider-multi-phase
+    bump: patch
+  - name: pallet-fast-unstake
+    bump: patch
+  - name: pallet-contracts
+    bump: patch

substrate/bin/node/cli/src/command.rs

@@ -107,7 +107,7 @@ pub fn run() -> Result<()> {
 							)
 						}
 
-						cmd.run::<HashingFor<Block>, sp_statement_store::runtime_api::HostFunctions>(config)
+						cmd.run_with_spec::<HashingFor<Block>, sp_statement_store::runtime_api::HostFunctions>(Some(config.chain_spec))
 					},
 					BenchmarkCmd::Block(cmd) => {
 						// ensure that we keep the task manager alive

substrate/bin/node/runtime/src/lib.rs

@@ -1759,6 +1759,25 @@ impl pallet_nis::Config for Runtime {
 	type MaxIntakeWeight = MaxIntakeWeight;
 	type ThawThrottle = ThawThrottle;
 	type RuntimeHoldReason = RuntimeHoldReason;
+	#[cfg(feature = "runtime-benchmarks")]
+	type BenchmarkSetup = SetupAsset;
+}
+
+#[cfg(feature = "runtime-benchmarks")]
+pub struct SetupAsset;
+#[cfg(feature = "runtime-benchmarks")]
+impl pallet_nis::BenchmarkSetup for SetupAsset {
+	fn create_counterpart_asset() {
+		let owner = AccountId::from([0u8; 32]);
+		// this may or may not fail depending on if the chain spec or runtime genesis is used.
+		let _ = Assets::force_create(
+			RuntimeOrigin::root(),
+			9u32.into(),
+			sp_runtime::MultiAddress::Id(owner),
+			true,
+			1,
+		);
+	}
 }
 
 parameter_types! {

substrate/frame/babe/src/lib.rs

@@ -283,7 +283,7 @@ pub mod pallet {
 	/// entropy was fixed (i.e. it was known to chain observers). Since epochs are defined in
 	/// slots, which may be skipped, the block numbers may not line up with the slot numbers.
 	#[pallet::storage]
-	pub(super) type EpochStart<T: Config> =
+	pub type EpochStart<T: Config> =
 		StorageValue<_, (BlockNumberFor<T>, BlockNumberFor<T>), ValueQuery>;
 
 	/// How late the current block is compared to its parent.

substrate/frame/benchmarking/src/v1.rs

@@ -850,8 +850,8 @@ macro_rules! impl_bench_name_tests {
 					if !($extra) {
 						let disabled = $crate::__private::vec![ $( stringify!($names_extra).as_ref() ),* ];
 						if disabled.contains(&stringify!($name)) {
-							$crate::__private::log::error!(
-								"INFO: extra benchmark skipped - {}",
+							$crate::__private::log::debug!(
+								"extra benchmark skipped - {}",
 								stringify!($name),
 							);
 							return ();
@@ -874,21 +874,21 @@ macro_rules! impl_bench_name_tests {
 								$crate::BenchmarkError::Override(_) => {
 									// This is still considered a success condition.
 									$crate::__private::log::error!(
-										"WARNING: benchmark error overridden - {}",
+										"benchmark error overridden - {}",
 										stringify!($name),
 									);
 								},
 								$crate::BenchmarkError::Skip => {
 									// This is considered a success condition.
-									$crate::__private::log::error!(
-										"WARNING: benchmark error skipped - {}",
+									$crate::__private::log::debug!(
+										"benchmark skipped - {}",
 										stringify!($name),
 									);
 								},
 								$crate::BenchmarkError::Weightless => {
 									// This is considered a success condition.
-									$crate::__private::log::error!(
-										"WARNING: benchmark weightless skipped - {}",
+									$crate::__private::log::debug!(
+										"benchmark weightless skipped - {}",
 										stringify!($name),
 									);
 								}

substrate/frame/contracts/src/benchmarking/mod.rs

@@ -332,7 +332,7 @@ mod benchmarks {
 	#[benchmark(pov_mode = Measured)]
 	fn migration_noop() {
 		let version = LATEST_MIGRATION_VERSION;
-		assert_eq!(StorageVersion::get::<Pallet<T>>(), version);
+		StorageVersion::new(version).put::<Pallet<T>>();
 		#[block]
 		{
 			Migration::<T>::migrate(Weight::MAX);
@@ -358,7 +358,7 @@ mod benchmarks {
 	#[benchmark(pov_mode = Measured)]
 	fn on_runtime_upgrade_noop() {
 		let latest_version = LATEST_MIGRATION_VERSION;
-		assert_eq!(StorageVersion::get::<Pallet<T>>(), latest_version);
+		StorageVersion::new(latest_version).put::<Pallet<T>>();
 		#[block]
 		{
 			<Migration<T, false> as frame_support::traits::OnRuntimeUpgrade>::on_runtime_upgrade();

substrate/frame/election-provider-multi-phase/src/lib.rs

@@ -931,7 +931,7 @@ pub mod pallet {
 				.expect(error_message);
 
 			// Store the newly received solution.
-			log!(info, "queued unsigned solution with score {:?}", ready.score);
+			log!(debug, "queued unsigned solution with score {:?}", ready.score);
 			let ejected_a_solution = <QueuedSolution<T>>::exists();
 			<QueuedSolution<T>>::put(ready);
 			Self::deposit_event(Event::SolutionStored {

substrate/frame/fast-unstake/src/lib.rs

@@ -560,7 +560,7 @@ pub mod pallet {
 				if !remaining.is_zero() {
 					Self::halt("not enough balance to unreserve");
 				} else {
-					log!(info, "unstaked {:?}, outcome: {:?}", stash, result);
+					log!(debug, "unstaked {:?}, outcome: {:?}", stash, result);
 					Self::deposit_event(Event::<T>::Unstaked { stash, result });
 				}
 			};