Add independent regex_redux benchmark test case (#1109)

* add independent regex_redux benchmark test case

* exclude regex_redux benchmark crate in CI where needed

.github/workflows/rust.yml

@@ -42,6 +42,7 @@ jobs:
             --exclude wasmi_fuzz
             --exclude wasmi_benches_tiny_keccak
             --exclude wasmi_benches_reverse_complement
+            --exclude wasmi_benches_regex_redux
       - name: Build (wasm32)
         run: >-
           cargo build
@@ -56,6 +57,7 @@ jobs:
           --exclude wasmi_fuzz
           --exclude wasmi_benches_tiny_keccak
           --exclude wasmi_benches_reverse_complement
+          --exclude wasmi_benches_regex_redux
 
   test-asan:
     name: Test (Address Sanitizer)

Cargo.toml

@@ -7,6 +7,7 @@ members = [
     "crates/wasi",
     "crates/wasmi/benches/rust/tiny_keccak",
     "crates/wasmi/benches/rust/reverse_complement",
+    "crates/wasmi/benches/rust/regex_redux",
     "fuzz",
 ]
 exclude = []

crates/wasmi/benches/benches.rs

@@ -768,20 +768,20 @@ fn bench_execute_rev_comp(c: &mut Criterion) {
 
 fn bench_execute_regex_redux(c: &mut Criterion) {
     c.bench_function("execute/regex_redux", |b| {
-        let (mut store, instance) = load_instance_from_file(WASM_KERNEL);
+        let (mut store, instance) = load_instance_from_file("benches/rust/regex_redux.wasm");
 
         // Allocate buffers for the input and output.
-        let test_data_ptr = instance
-            .get_typed_func::<i32, i32>(&store, "prepare_regex_redux")
+        let data_ptr = instance
+            .get_typed_func::<i32, i32>(&store, "setup")
             .unwrap()
             .call(&mut store, REVCOMP_INPUT.len() as i32)
             .unwrap();
 
         // Get the pointer to the input buffer.
         let input_data_mem_offset = instance
-            .get_typed_func::<i32, i32>(&store, "regex_redux_input_ptr")
+            .get_typed_func::<i32, i32>(&store, "input_ptr")
             .unwrap()
-            .call(&mut store, test_data_ptr)
+            .call(&mut store, data_ptr)
             .unwrap();
 
         // Copy test data inside the wasm memory.
@@ -792,12 +792,25 @@ fn bench_execute_regex_redux(c: &mut Criterion) {
             .unwrap();
 
         // Actually run the benchmark:
-        let run = instance
-            .get_typed_func::<i32, ()>(&store, "bench_regex_redux")
-            .unwrap();
+        let run = instance.get_typed_func::<i32, ()>(&store, "run").unwrap();
         b.iter(|| {
-            run.call(&mut store, test_data_ptr).unwrap();
-        })
+            run.call(&mut store, data_ptr).unwrap();
+        });
+
+        // Check the result of the regex find.
+        let result = instance
+            .get_typed_func::<i32, i32>(&store, "output")
+            .unwrap()
+            .call(&mut store, data_ptr)
+            .unwrap();
+        assert_eq!(result, 2);
+
+        // Teardown benchmark data.
+        instance
+            .get_typed_func::<i32, ()>(&store, "teardown")
+            .unwrap()
+            .call(&mut store, data_ptr)
+            .unwrap();
     });
 }
 

crates/wasmi/benches/rust/regex_redux/Cargo.toml

@@ -0,0 +1,13 @@
+[package]
+name = "wasmi_benches_regex_redux"
+version = "0.0.0"
+publish = false
+authors.workspace = true
+edition.workspace = true
+
+[lib]
+path = "lib.rs"
+crate-type = ["cdylib"]
+
+[dependencies]
+regex = "1.10.5"

crates/wasmi/benches/rust/regex_redux/README.md

@@ -0,0 +1,35 @@
+## Benchmark - Regex Redux
+
+Rust based `regex_redux` benchmark test case.
+
+Build with
+
+```
+cargo build \
+    --package wasmi_benches_regex_redux \
+    --profile wasm \
+    --target wasm32-unknown-unknown
+```
+
+Post-optimize with `wasm-opt` using:
+
+```
+wasm-opt \
+    -O3 \
+    target/wasm32-unknown-unknown/wasm/wasmi_benches_regex_redux.wasm \
+    -o target/wasm32-unknown-unknown/wasm/wasmi_benches_regex_redux.opt.wasm
+```
+
+Finally move the post-optimized Wasm binary into the proper place:
+
+```
+cp \
+    target/wasm32-unknown-unknown/wasm/wasmi_benches_regex_redux.opt.wasm \
+    crates/wasmi/benches/rust/regex_redux.wasm
+```
+
+Benchmark with:
+
+```
+cargo bench --bench benches execute/regex_redux
+```

crates/wasmi/benches/rust/regex_redux/lib.rs

@@ -0,0 +1,44 @@
+//! Initially it was supposed to be like [1]. However it turned out
+//! that executing this code in Wasmi way too slow, unfortunately.
+//!
+//! [1]: https://benchmarksgame-team.pages.debian.net/benchmarksgame/program/regexredux-rust-2.html
+
+use regex::bytes::Regex;
+
+#[repr(C)]
+pub struct RegexReduxData {
+    regex: Regex,
+    input: Box<[u8]>,
+    output: Option<usize>,
+}
+
+#[no_mangle]
+pub extern "C" fn setup(size: usize) -> Box<RegexReduxData> {
+    Box::new(RegexReduxData {
+        regex: Regex::new("agggtaa[cgt]|[acg]ttaccct").unwrap(),
+        input: vec![0; size].into(),
+        output: None,
+    })
+}
+
+#[no_mangle]
+pub extern "C" fn teardown(_: Box<RegexReduxData>) {}
+
+#[no_mangle]
+pub extern "C" fn input_ptr(data: &mut RegexReduxData) -> *mut u8 {
+    data.input.as_mut_ptr()
+}
+
+#[no_mangle]
+pub extern "C" fn output(data: &mut RegexReduxData) -> u32 {
+    match data.output {
+        Some(output) => output.try_into().unwrap(),
+        None => u32::MAX,
+    }
+}
+
+#[no_mangle]
+pub extern "C" fn run(data: &mut RegexReduxData) {
+    let result: usize = data.regex.find_iter(&data.input[..]).count();
+    data.output = Some(result);
+}