编写benchmarking分两种情况，如下：
- 对函数进行性能测试时需要的构造条件不会涉及到本pallet以外的其它pallet；
- 在对函数进行性能测试时需要先使用其它的pallet构造测试的先决条件。
第一种情况相对来说比较简单，这个也比较好找到例子。第二种情况则比较复杂，写起来也比较麻烦。不过在我们的开发中，大部分都是第一种情况。
针对这两种情况，我们分别来讲解如何编写代码。本节，主要讲第一种情况。
我们创建一个pallet，名字叫做use-benchmarking，对应的use-benchmarking/src/lib.rs的代码如下：
#![cfg_attr(not(feature = "std"), no_std)]
pub use pallet::*;
#[frame_support::pallet]
pub mod pallet {
use codec::Codec;
use frame_support::{
pallet_prelude::*, sp_runtime::traits::AtLeast32BitUnsigned, sp_std::fmt::Debug,
};
use frame_system::pallet_prelude::*;
use scale_info::prelude::vec::Vec;
use sp_io::hashing::{blake2_128, twox_128};
#[pallet::pallet]
#[pallet::generate_store(pub(super) trait Store)]
pub struct Pallet<T>(_);
// 3. Runtime Configuration Trait
#[pallet::config]
pub trait Config: frame_system::Config {
type Event: From<Event<Self>> + IsType<<Self as frame_system::Config>::Event>;
//声明StudentNumber类型
type StudentNumberType: Member
+ Parameter
+ AtLeast32BitUnsigned
+ Codec
+ Copy
+ Debug
+ Default
+ MaxEncodedLen
+ MaybeSerializeDeserialize;
//声明StudentName类型
type StudentNameType: Parameter
+ Member
+ AtLeast32BitUnsigned
+ Codec
+ Default
+ From<u128>
+ Into<u128>
+ Copy
+ MaxEncodedLen
+ MaybeSerializeDeserialize
+ Debug;
}
// 4. Runtime Storage
// 用storageMap存储学生信息，（key， value）分别对应的是学号和姓名.
#[pallet::storage]
#[pallet::getter(fn students_info)]
pub type StudentsInfo<T: Config> =
StorageMap<_, Blake2_128Concat, T::StudentNumberType, T::StudentNameType, ValueQuery>;
// 5. Runtime Events
// Can stringify event types to metadata.
#[pallet::event]
#[pallet::generate_deposit(pub(super) fn deposit_event)]
pub enum Event<T: Config> {
SetStudentInfo(T::StudentNumberType, T::StudentNameType),
}
// 8. Runtime Errors
#[pallet::error]
pub enum Error<T> {
// 相同学号的只允许设置一次名字
SetStudentsInfoDuplicate,
}
// 7. Extrinsics
// Functions that are callable from outside the runtime.
#[pallet::call]
impl<T: Config> Pallet<T> {
#[pallet::weight(100)]
pub fn set_student_info(
origin: OriginFor<T>,
student_number: T::StudentNumberType,
student_name: T::StudentNameType,
) -> DispatchResultWithPostInfo {
ensure_signed(origin)?;
if StudentsInfo::<T>::contains_key(student_number) {
return Err(Error::<T>::SetStudentsInfoDuplicate.into())
}
StudentsInfo::<T>::insert(&student_number, &student_name);
Self::deposit_event(Event::SetStudentInfo(student_number, student_name));
Self::generate_key();
Ok(().into())
}
}
}
编写mock，编写benchmarking时需要的mock和之前在use-tests中编写的差不多，我们这里直接贴代码，如下：
use crate as pallet_template;
use frame_support::traits::{ConstU16, ConstU64};
use frame_system as system;
use sp_core::H256;
use sp_runtime::{
testing::Header,
traits::{BlakeTwo256, IdentityLookup},
};
type UncheckedExtrinsic = frame_system::mocking::MockUncheckedExtrinsic<Test>;
type Block = frame_system::mocking::MockBlock<Test>;
// Configure a mock runtime to test the pallet.
frame_support::construct_runtime!(
pub enum Test where
Block = Block,
NodeBlock = Block,
UncheckedExtrinsic = UncheckedExtrinsic,
{
System: frame_system::{Pallet, Call, Config, Storage, Event<T>},
TemplateModule: pallet_template::{Pallet, Call, Storage, Event<T>},
}
);
impl system::Config for Test {
type BaseCallFilter = frame_support::traits::Everything;
type BlockWeights = ();
type BlockLength = ();
type DbWeight = ();
type Origin = Origin;
type Call = Call;
type Index = u64;
type BlockNumber = u64;
type Hash = H256;
type Hashing = BlakeTwo256;
type AccountId = u64;
type Lookup = IdentityLookup<Self::AccountId>;
type Header = Header;
type Event = Event;
type BlockHashCount = ConstU64<250>;
type Version = ();
type PalletInfo = PalletInfo;
type AccountData = ();
type OnNewAccount = ();
type OnKilledAccount = ();
type SystemWeightInfo = ();
type SS58Prefix = ConstU16<42>;
type OnSetCode = ();
type MaxConsumers = frame_support::traits::ConstU32<16>;
}
impl pallet_template::Config for Test {
type Event = Event;
}
// Build genesis storage according to the mock runtime.
pub fn new_test_ext() -> sp_io::TestExternalities {
system::GenesisConfig::default().build_storage::<Test>().unwrap().into()
}
这里需要注意的是，编写好mock后，我们对其导出，也就是在use-benchmarking/src/lib.rs添加如下代码：
#[cfg(test)]
mod mock;
编写benchmarking的目的主要是为调度函数生成对应的权重计算函数，对应到本例子中就主要是use-benchmarking的set_student_info函数生成对应的权重计算函数。我们首先在use-benchmarking/src目录下创建一个文件，名为benchmarking.rs，编写内容如下：
use super::*;
#[allow(unused)]
use crate::Pallet as UseBenchmarkingDemo;
use frame_benchmarking::{benchmarks, whitelisted_caller};
use frame_system::RawOrigin;
benchmarks! {
set_student_info { //1、准备条件
let s in 0 .. 100;
let caller: T::AccountId = whitelisted_caller();
}:{ //2、调用调度函数
let _ = UseBenchmarkingDemo::<T>::set_student_info(RawOrigin::Signed(caller).into(), s.into(), Default::default());
}
verify {//3、进行验证
assert_eq!(<StudentsInfo<T>>::get::<<T as pallet::Config>::StudentNumberType>(s.into()), Default::default());
}
// 使用mock中的new_test_ext
impl_benchmark_test_suite!(UseBenchmarkingDemo, crate::mock::new_test_ext(), crate::mock::Test);
}
下面我们就来逐步讲解。
编写benchmark以宏benchmarks!包含，里面的内容主要分三部分，做的事情分别是准备条件、调用调度函数、对执行结果验证。对于第一步，其实就是对我们调度函数中的变量进行赋值，详情可以参考https://docs.substrate.io/reference/how-to-guides/weights/add-benchmarks/
首先，还是需要把use-benchmarking这个pallet加到runtime的依赖中，所以在runtime/Cargo.toml中添加如下：
[dependencies]
...
pallet-use-benchmarking = { version = "4.0.0-dev", default-features = false, path = "../pallets/use-benchmarking" }
...
[features]
default = ["std"]
std = [
...
"pallet-use-benchmarking/std",
...
]
runtime-benchmarks = [
...
"pallet-use-benchmarking/runtime-benchmarks",
...
]
接下来，则是在runtime/src/lib.rs的代码中添加，我们先将use-benchmarking添加到runtime中，添加如下代码：
impl pallet_use_benchmarking::Config for Runtime {
type Event = Event;
type StudentNumberType = u32;
type StudentNameType = u128;
}
construct_runtime!(
pub enum Runtime where
Block = Block,
NodeBlock = opaque::Block,
UncheckedExtrinsic = UncheckedExtrinsic
{
System: frame_system,
RandomnessCollectiveFlip: pallet_randomness_collective_flip,
Timestamp: pallet_timestamp,
Aura: pallet_aura,
Grandpa: pallet_grandpa,
Balances: pallet_balances,
TransactionPayment: pallet_transaction_payment,
Sudo: pallet_sudo,
...
//添加下面这行
UseBenchmarkingDemo: pallet_use_benchmarking,
}
);
然后就是将use-benchmarking pallet添加到对应的benchmark宏中，需要添加的代码如下：
#[cfg(feature = "runtime-benchmarks")]
mod benches {
define_benchmarks!(
[frame_benchmarking, BaselineBench::<Runtime>]
[frame_system, SystemBench::<Runtime>]
[pallet_balances, Balances]
[pallet_timestamp, Timestamp]
[pallet_template, TemplateModule]
//这行为添加的代码
[pallet_use_benchmarking, UseBenchmarkingDemo]
);
}
首先时编译，编译的命令需要带上--features runtime-benchmarks， 完整的编译命令如下：
cargo build --features runtime-benchmarks
然后就是生成对应的weights文件，生成之前，我们需要从substrate的repo中拷贝模板放到我们的substrate-node-template目录下：
mkdir .maintain
cd .maintain
cp substrate/.maintain/frame-weight-template.hbs . #此处的substrate同官方的repo仓库
然后就是生成weights文件的命令，如下：
./target/debug/node-template benchmark --chain dev --execution wasm --wasm-execution compiled --pallet pallet_use_benchmarking --extrinsic "*" --steps 20 --repeat 10 --output ./pallets/use-benchmarking/src/weights.rs --template ./.maintain/frame-weight-template.hbs
执行完后就会在./pallets/use-benchmarking/src/目录下生成对应的weights.rs文件。
至此，我们生成了weights.rs文件，但是这仅仅只是生成，还需要把生成的权重函数用到pallet中，演示demo对应的修改为这个提交（https://github.com/anonymousGiga/learn-substrate-easy-source/commit/8a4131c5f18c3bb769618265dddb41652d89a70b）
接下来我们将权重函数添加到pallet中，演示demo中对应的提交为(https://github.com/anonymousGiga/learn-substrate-easy-source/commit/ce698a63939a8ad1e004c5ced2a1c9222178fb93)。
首先，我们需要在pallets/use-benchmarking/Cargo.toml中添加如下：
...
[dependencies]
...
//添加下面这行
sp-std = { default-features = false, version = "4.0.0", git = "https://github.com/paritytech/substrate.git", branch = "polkadot-v0.9.18" }
[features]
default = ["std"]
std = [
...
//添加下面这行
"sp-std/std",
]
接着我们需要在pallets/use-benchmarking/src/lib.rs添加：
pub mod weights;
pub use weights::WeightInfo;
然后需要在mod内部添加引入：
#[frame_support::pallet]
pub mod pallet {
...
//添加这行
use crate::WeightInfo;
...
}
另外，我们还需要在pallets/use-benchmarking/src/lib.rs中的Config中添加WeightInfo类型，如下：
pub trait Config: frame_system::Config {
...
// 添加这里
type WeightInfo: WeightInfo;
}
然后我们需要把调度函数set_student_info上方的#[pallet::weight(100)]修改成#[pallet::weight(<T as Config>::WeightInfo::set_student_info((*student_number).into() ))]。
最后，我们还需要回到runtime/src/lib.rs文件，修改pallet_use_benchmarking的配置，如下：
impl pallet_use_benchmarking::Config for Runtime {
type Event = Event;
type StudentNumberType = u32;
type StudentNameType = u128;
//添加这一行
type WeightInfo = pallet_use_benchmarking::weights::SubstrateWeight<Runtime>;
}
重新编译，pallet的对应的权重就会在新的执行文件中生效，编译命令如下：
cargo build