refactor(bench): reuse allocator in parser + lexer benchmarks (#3053)

Re-use allocator in parser + lexer benchmarks.

I believe this is the recommended usage when parsing a bunch of files -
to re-use one allocator rather than create a fresh one for each run, so
it makes sense to me that this is what the benchmark should measure.

Doesn't show much difference on CodSpeed because it only runs the
benchmark once, and it treats allocations as free anyway. But I imagine
the difference may show up a bit more in a standard criterion benchmark.

crates/oxc_allocator/src/lib.rs

@@ -1,4 +1,7 @@
-use std::{convert::From, ops::Deref};
+use std::{
+    convert::From,
+    ops::{Deref, DerefMut},
+};
 
 mod arena;
 
@@ -24,6 +27,12 @@ impl Deref for Allocator {
     }
 }
 
+impl DerefMut for Allocator {
+    fn deref_mut(&mut self) -> &mut Self::Target {
+        &mut self.bump
+    }
+}
+
 #[cfg(test)]
 mod test {
     use std::ops::Deref;

tasks/benchmark/benches/lexer.rs

@@ -27,14 +27,14 @@ fn bench_lexer(criterion: &mut Criterion) {
             BenchmarkId::from_parameter(&file.file_name),
             &file.source_text,
             |b, source_text| {
-                b.iter_with_large_drop(|| {
+                // Do not include initializing allocator in benchmark.
-                    // Include the allocator drop time to make time measurement consistent.
+                // User code would likely reuse the same allocator over and over to parse multiple files,
-                    // Otherwise the allocator will allocate huge memory chunks (by power of two) from the
+                // so we do the same here.
-                    // system allocator, which makes time measurement unequal during long runs.
+                let mut allocator = Allocator::default();
-                    let allocator = Allocator::default();
+                b.iter(|| {
                     let mut lexer = Lexer::new_for_benchmarks(&allocator, source_text, source_type);
                     while lexer.next_token().kind != Kind::Eof {}
-                    allocator
+                    allocator.reset();
                 });
             },
         );

tasks/benchmark/benches/parser.rs

@@ -12,13 +12,13 @@ fn bench_parser(criterion: &mut Criterion) {
             BenchmarkId::from_parameter(&file.file_name),
             &file.source_text,
             |b, source_text| {
-                b.iter_with_large_drop(|| {
+                // Do not include initializing allocator in benchmark.
-                    // Include the allocator drop time to make time measurement consistent.
+                // User code would likely reuse the same allocator over and over to parse multiple files,
-                    // Otherwise the allocator will allocate huge memory chunks (by power of two) from the
+                // so we do the same here.
-                    // system allocator, which makes time measurement unequal during long runs.
+                let mut allocator = Allocator::default();
-                    let allocator = Allocator::default();
+                b.iter(|| {
-                    _ = Parser::new(&allocator, source_text, source_type).parse();
+                    Parser::new(&allocator, source_text, source_type).parse();
-                    allocator
+                    allocator.reset();
                 });
             },
         );