pezkuwichain/pezkuwi-subxt
1
0
Fork 0
mirror of https://github.com/pezkuwichain/pezkuwi-subxt.git synced 2026-06-14 08:41:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

contracts: Move Schedule from Storage to Config (#8773)

Browse Source 
* Move `Schedule` from Storage to Config

* Updated CHANGELOG

* Fix nits from review

* Fix migration

* Print the debug buffer as tracing message

* Use `debug` instead of `trace` and update README

* Add additional assert to test

* Rename `schedule_version` to `instruction_weights_version`

* Fixed typo

* Added more comments to wat fixtures

* Add clarification for the `debug_message` field
This commit is contained in:
Alexander Theißen
2021-05-13 21:56:11 +02:00
committed by GitHub
parent 3c0270fe57
commit 1ac95b6ba6
23 changed files with 1465 additions and 1056 deletions
Download Patch File Download Diff File Expand all files Collapse all files
substrate/frame/contracts/src/benchmarking/mod.rs
+55 -38
View File
@@ -290,7 +290,7 @@ benchmarks! {
on_initialize_per_trie_key {
let k in 0..1024;
let instance = ContractWithStorage::<T>::new(k, T::MaxValueSize::get())?;
let instance = ContractWithStorage::<T>::new(k, T::Schedule::get().limits.payload_len)?;
Storage::<T>::queue_trie_for_deletion(&instance.contract.alive_info()?)?;
}: {
Storage::<T>::process_deletion_queue_batch(Weight::max_value())
@@ -311,23 +311,15 @@ benchmarks! {
// first time after a new schedule was deployed: For every new schedule a contract needs
// to re-run the instrumentation once.
instrument {
let c in 0 .. T::MaxCodeSize::get() / 1024;
let c in 0 .. T::Schedule::get().limits.code_len / 1024;
let WasmModule { code, hash, .. } = WasmModule::<T>::sized(c * 1024);
Contracts::<T>::store_code_raw(code)?;
let mut module = PrefabWasmModule::from_storage_noinstr(hash)?;
let schedule = <CurrentSchedule<T>>::get();
let schedule = T::Schedule::get();
}: {
Contracts::<T>::reinstrument_module(&mut module, &schedule)?;
}
// This extrinsic is pretty much constant as it is only a simple setter.
update_schedule {
let schedule = Schedule {
version: 1,
.. Default::default()
};
}: _(RawOrigin::Root, schedule)
// This constructs a contract that is maximal expensive to instrument.
// It creates a maximum number of metering blocks per byte.
// The size of the salt influences the runtime because is is hashed in order to
@@ -340,7 +332,7 @@ benchmarks! {
// We cannot let `c` grow to the maximum code size because the code is not allowed
// to be larger than the maximum size **after instrumentation**.
instantiate_with_code {
let c in 0 .. Perbill::from_percent(50).mul_ceil(T::MaxCodeSize::get() / 1024);
let c in 0 .. Perbill::from_percent(50).mul_ceil(T::Schedule::get().limits.code_len / 1024);
let s in 0 .. code::max_pages::<T>() * 64;
let salt = vec![42u8; (s * 1024) as usize];
let endowment = caller_funding::<T>() / 3u32.into();
@@ -363,7 +355,7 @@ benchmarks! {
// `c`: Size of the code in kilobytes.
// `s`: Size of the salt in kilobytes.
instantiate {
let c in 0 .. T::MaxCodeSize::get() / 1024;
let c in 0 .. T::Schedule::get().limits.code_len / 1024;
let s in 0 .. code::max_pages::<T>() * 64;
let salt = vec![42u8; (s * 1024) as usize];
let endowment = caller_funding::<T>() / 3u32.into();
@@ -390,7 +382,7 @@ benchmarks! {
// part of `seal_input`.
// `c`: Size of the code in kilobytes.
call {
let c in 0 .. T::MaxCodeSize::get() / 1024;
let c in 0 .. T::Schedule::get().limits.code_len / 1024;
let data = vec![42u8; 1024];
let instance = Contract::<T>::with_caller(
whitelisted_caller(), WasmModule::dummy_with_bytes(c * 1024), vec![], Endow::CollectRent
@@ -423,7 +415,7 @@ benchmarks! {
// the reward for removing them.
// `c`: Size of the code of the contract that should be evicted.
claim_surcharge {
let c in 0 .. T::MaxCodeSize::get() / 1024;
let c in 0 .. T::Schedule::get().limits.code_len / 1024;
let instance = Contract::<T>::with_caller(
whitelisted_caller(), WasmModule::dummy_with_bytes(c * 1024), vec![], Endow::CollectRent
)?;
@@ -730,7 +722,7 @@ benchmarks! {
}
seal_terminate_per_code_kb {
let c in 0 .. T::MaxCodeSize::get() / 1024;
let c in 0 .. T::Schedule::get().limits.code_len / 1024;
let beneficiary = account::<T::AccountId>("beneficiary", 0, 0);
let beneficiary_bytes = beneficiary.encode();
let beneficiary_len = beneficiary_bytes.len();
@@ -771,7 +763,7 @@ benchmarks! {
// Restore just moves the trie id from origin to destination and therefore
// does not depend on the size of the destination contract. However, to not
// trigger any edge case we won't use an empty contract as destination.
let mut tombstone = ContractWithStorage::<T>::new(10, T::MaxValueSize::get())?;
let mut tombstone = ContractWithStorage::<T>::new(10, T::Schedule::get().limits.payload_len)?;
tombstone.evict()?;
let dest = tombstone.contract.account_id.encode();
@@ -847,14 +839,14 @@ benchmarks! {
// `t`: Code size of tombstone contract
// `d`: Number of supplied delta keys
seal_restore_to_per_code_kb_delta {
let c in 0 .. T::MaxCodeSize::get() / 1024;
let t in 0 .. T::MaxCodeSize::get() / 1024;
let c in 0 .. T::Schedule::get().limits.code_len / 1024;
let t in 0 .. T::Schedule::get().limits.code_len / 1024;
let d in 0 .. API_BENCHMARK_BATCHES;
let mut tombstone = ContractWithStorage::<T>::with_code(
WasmModule::<T>::dummy_with_bytes(t * 1024), 0, 0
)?;
tombstone.evict()?;
let delta = create_storage::<T>(d * API_BENCHMARK_BATCH_SIZE, T::MaxValueSize::get())?;
let delta = create_storage::<T>(d * API_BENCHMARK_BATCH_SIZE, T::Schedule::get().limits.payload_len)?;
let dest = tombstone.contract.account_id.encode();
let dest_len = dest.len();
@@ -938,7 +930,7 @@ benchmarks! {
seal_random {
let r in 0 .. API_BENCHMARK_BATCHES;
let pages = code::max_pages::<T>();
let subject_len = <CurrentSchedule<T>>::get().limits.subject_len;
let subject_len = T::Schedule::get().limits.subject_len;
assert!(subject_len < 1024);
let code = WasmModule::<T>::from(ModuleDefinition {
memory: Some(ImportedMemory::max::<T>()),
@@ -994,8 +986,8 @@ benchmarks! {
// `t`: Number of topics
// `n`: Size of event payload in kb
seal_deposit_event_per_topic_and_kb {
let t in 0 .. <CurrentSchedule<T>>::get().limits.event_topics;
let n in 0 .. T::MaxValueSize::get() / 1024;
let t in 0 .. T::Schedule::get().limits.event_topics;
let n in 0 .. T::Schedule::get().limits.payload_len / 1024;
let mut topics = (0..API_BENCHMARK_BATCH_SIZE)
.map(|n| (n * t..n * t + t).map(|i| T::Hashing::hash_of(&i)).collect::<Vec<_>>().encode())
.peekable();
@@ -1055,6 +1047,31 @@ benchmarks! {
let origin = RawOrigin::Signed(instance.caller.clone());
}: call(origin, instance.addr, 0u32.into(), Weight::max_value(), vec![])
// The size of the supplied message does not influence the weight because as it is never
// processed during on-chain execution: It is only ever read during debugging which happens
// when the contract is called as RPC where weights do not matter.
seal_debug_message {
let r in 0 .. API_BENCHMARK_BATCHES;
let max_bytes = code::max_pages::<T>() * 64 * 1024;
let code = WasmModule::<T>::from(ModuleDefinition {
memory: Some(ImportedMemory { min_pages: 1, max_pages: 1 }),
imported_functions: vec![ImportedFunction {
name: "seal_debug_message",
params: vec![ValueType::I32, ValueType::I32],
return_type: Some(ValueType::I32),
}],
call_body: Some(body::repeated(r * API_BENCHMARK_BATCH_SIZE, &[
Instruction::I32Const(0), // value_ptr
Instruction::I32Const(max_bytes as i32), // value_len
Instruction::Call(0),
Instruction::Drop,
])),
.. Default::default()
});
let instance = Contract::<T>::new(code, vec![], Endow::Max)?;
let origin = RawOrigin::Signed(instance.caller.clone());
}: call(origin, instance.addr, 0u32.into(), Weight::max_value(), vec![])
// Only the overhead of calling the function itself with minimal arguments.
// The contract is a bit more complex because I needs to use different keys in order
// to generate unique storage accesses. However, it is still dominated by the storage
@@ -1091,7 +1108,7 @@ benchmarks! {
}: call(origin, instance.addr, 0u32.into(), Weight::max_value(), vec![])
seal_set_storage_per_kb {
let n in 0 .. T::MaxValueSize::get() / 1024;
let n in 0 .. T::Schedule::get().limits.payload_len / 1024;
let key = T::Hashing::hash_of(&1u32).as_ref().to_vec();
let key_len = key.len();
let code = WasmModule::<T>::from(ModuleDefinition {
@@ -1155,7 +1172,7 @@ benchmarks! {
<System<T>>::block_number(),
&mut info,
key.as_slice().try_into().map_err(|e| "Key has wrong length")?,
Some(vec![42; T::MaxValueSize::get() as usize])
Some(vec![42; T::Schedule::get().limits.payload_len as usize])
)
.map_err(|_| "Failed to write to storage during setup.")?;
}
@@ -1210,7 +1227,7 @@ benchmarks! {
}: call(origin, instance.addr, 0u32.into(), Weight::max_value(), vec![])
seal_get_storage_per_kb {
let n in 0 .. T::MaxValueSize::get() / 1024;
let n in 0 .. T::Schedule::get().limits.payload_len / 1024;
let key = T::Hashing::hash_of(&1u32).as_ref().to_vec();
let key_len = key.len();
let code = WasmModule::<T>::from(ModuleDefinition {
@@ -1227,7 +1244,7 @@ benchmarks! {
},
DataSegment {
offset: key_len as u32,
value: T::MaxValueSize::get().to_le_bytes().into(),
value: T::Schedule::get().limits.payload_len.to_le_bytes().into(),
},
],
call_body: Some(body::repeated(API_BENCHMARK_BATCH_SIZE, &[
@@ -1363,7 +1380,7 @@ benchmarks! {
}: call(origin, instance.addr, 0u32.into(), Weight::max_value(), vec![])
seal_call_per_code_transfer_input_output_kb {
let c in 0 .. T::MaxCodeSize::get() / 1024;
let c in 0 .. T::Schedule::get().limits.code_len / 1024;
let t in 0 .. 1;
let i in 0 .. code::max_pages::<T>() * 64;
let o in 0 .. (code::max_pages::<T>() - 1) * 64;
@@ -1560,7 +1577,7 @@ benchmarks! {
}
seal_instantiate_per_code_input_output_salt_kb {
let c in 0 .. T::MaxCodeSize::get() / 1024;
let c in 0 .. T::Schedule::get().limits.code_len / 1024;
let i in 0 .. (code::max_pages::<T>() - 1) * 64;
let o in 0 .. (code::max_pages::<T>() - 1) * 64;
let s in 0 .. (code::max_pages::<T>() - 1) * 64;
@@ -1927,7 +1944,7 @@ benchmarks! {
// w_br_table_per_entry = w_bench
instr_br_table_per_entry {
let e in 1 .. <CurrentSchedule<T>>::get().limits.br_table_size;
let e in 1 .. T::Schedule::get().limits.br_table_size;
let entry: Vec<u32> = [0, 1].iter()
.cloned()
.cycle()
@@ -1983,7 +2000,7 @@ benchmarks! {
// w_call_indrect = w_bench - 3 * w_param
instr_call_indirect {
let r in 0 .. INSTR_BENCHMARK_BATCHES;
let num_elements = <CurrentSchedule<T>>::get().limits.table_size;
let num_elements = T::Schedule::get().limits.table_size;
use self::code::TableSegment;
let mut sbox = Sandbox::from(&WasmModule::<T>::from(ModuleDefinition {
// We need to make use of the stack here in order to trigger stack height
@@ -2013,8 +2030,8 @@ benchmarks! {
// linearly depend on the amount of parameters to this function.
// Please note that this is not necessary with a direct call.
instr_call_indirect_per_param {
let p in 0 .. <CurrentSchedule<T>>::get().limits.parameters;
let num_elements = <CurrentSchedule<T>>::get().limits.table_size;
let p in 0 .. T::Schedule::get().limits.parameters;
let num_elements = T::Schedule::get().limits.table_size;
use self::code::TableSegment;
let mut sbox = Sandbox::from(&WasmModule::<T>::from(ModuleDefinition {
// We need to make use of the stack here in order to trigger stack height
@@ -2044,7 +2061,7 @@ benchmarks! {
// w_local_get = w_bench - 1 * w_param
instr_local_get {
let r in 0 .. INSTR_BENCHMARK_BATCHES;
let max_locals = <CurrentSchedule<T>>::get().limits.stack_height;
let max_locals = T::Schedule::get().limits.stack_height;
let mut call_body = body::repeated_dyn(r * INSTR_BENCHMARK_BATCH_SIZE, vec![
RandomGetLocal(0, max_locals),
Regular(Instruction::Drop),
@@ -2061,7 +2078,7 @@ benchmarks! {
// w_local_set = w_bench - 1 * w_param
instr_local_set {
let r in 0 .. INSTR_BENCHMARK_BATCHES;
let max_locals = <CurrentSchedule<T>>::get().limits.stack_height;
let max_locals = T::Schedule::get().limits.stack_height;
let mut call_body = body::repeated_dyn(r * INSTR_BENCHMARK_BATCH_SIZE, vec![
RandomI64Repeated(1),
RandomSetLocal(0, max_locals),
@@ -2078,7 +2095,7 @@ benchmarks! {
// w_local_tee = w_bench - 2 * w_param
instr_local_tee {
let r in 0 .. INSTR_BENCHMARK_BATCHES;
let max_locals = <CurrentSchedule<T>>::get().limits.stack_height;
let max_locals = T::Schedule::get().limits.stack_height;
let mut call_body = body::repeated_dyn(r * INSTR_BENCHMARK_BATCH_SIZE, vec![
RandomI64Repeated(1),
RandomTeeLocal(0, max_locals),
@@ -2096,7 +2113,7 @@ benchmarks! {
// w_global_get = w_bench - 1 * w_param
instr_global_get {
let r in 0 .. INSTR_BENCHMARK_BATCHES;
let max_globals = <CurrentSchedule<T>>::get().limits.globals;
let max_globals = T::Schedule::get().limits.globals;
let mut sbox = Sandbox::from(&WasmModule::<T>::from(ModuleDefinition {
call_body: Some(body::repeated_dyn(r * INSTR_BENCHMARK_BATCH_SIZE, vec![
RandomGetGlobal(0, max_globals),
@@ -2112,7 +2129,7 @@ benchmarks! {
// w_global_set = w_bench - 1 * w_param
instr_global_set {
let r in 0 .. INSTR_BENCHMARK_BATCHES;
let max_globals = <CurrentSchedule<T>>::get().limits.globals;
let max_globals = T::Schedule::get().limits.globals;
let mut sbox = Sandbox::from(&WasmModule::<T>::from(ModuleDefinition {
call_body: Some(body::repeated_dyn(r * INSTR_BENCHMARK_BATCH_SIZE, vec![
RandomI64Repeated(1),