Merge pull request #1231 from rylev/add-regex-1.5.5-benchmark

Add `regex-1.5.5` benchmark

Author: rylev

ci/check-profiling.sh

@@ -129,7 +129,7 @@ test   -f results/eprintln-Test-helloworld-Check-Full
 test ! -s results/eprintln-Test-helloworld-Check-Full
 
 # llvm-lines. `Debug` not `Check` because it doesn't support `Check` profiles.
-# Including both `helloworld` and `regex` benchmarks, as they exercise the
+# Including both `helloworld` and `regex-1.5.5` benchmarks, as they exercise the
 # zero dependency and the greater than zero dependency cases, respectively, the
 # latter of which has broken before.
 RUST_BACKTRACE=1 RUST_LOG=raw_cargo_messages=trace,collector=debug,rust_sysroot=debug \
@@ -138,12 +138,12 @@ RUST_BACKTRACE=1 RUST_LOG=raw_cargo_messages=trace,collector=debug,rust_sysroot=
         --id Test \
         --profiles Debug \
         --cargo $bindir/cargo \
-        --include helloworld,regex \
+        --include helloworld,regex-1.5.5 \
         --scenarios Full
 test -f results/ll-Test-helloworld-Debug-Full
 grep -q "Lines.*Copies.*Function name" results/ll-Test-helloworld-Debug-Full
-test -f results/ll-Test-regex-Debug-Full
-grep -q "Lines.*Copies.*Function name" results/ll-Test-regex-Debug-Full
+test -f results/ll-Test-regex-1.5.5-Debug-Full
+grep -q "Lines.*Copies.*Function name" results/ll-Test-regex-1.5.5-Debug-Full
 
 
 #----------------------------------------------------------------------------

collector/README.md

@@ -189,14 +189,14 @@ Finally, while most of the options you can pass to the collector are supported,
 the profilers used in the `profile_local` command are not. In Windows, the only currently supported
 profiler is the `self-profiler`.
 
-As a complete example, let's run just the `regex` benchmark in the `Debug`
+As a complete example, let's run just the `regex-1.5.5` benchmark in the `Debug`
 profile with self-profiling results available:
 
 ```pwsh
 $env:XPERF="C:\Program Files (x86)\Windows Kits\10\Windows Performance Toolkit\xperf.exe"
 $env:TRACELOG="C:\Program Files (x86)\Windows Kits\10\bin\10.0.19041.0\x64\tracelog.exe"
-.\target\release\collector.exe bench_local $env:RUST_ORIGINAL --id Original --profiles Debug --include regex --self-profile
-.\target\release\collector.exe bench_local $env:RUST_MODIFIED --id Modified --profiles Debug --include regex --self-profile
+.\target\release\collector.exe bench_local $env:RUST_ORIGINAL --id Original --profiles Debug --include regex-1.5.5 --self-profile
+.\target\release\collector.exe bench_local $env:RUST_MODIFIED --id Modified --profiles Debug --include regex-1.5.5 --self-profile
 .\target\release\site.exe .\results.db
 ```
 

collector/benchmarks/README.md

@@ -37,7 +37,7 @@ They mostly consist of real-world crates.
 - **hyper-2**: A fairly large crate. Utilizes async/await, and used by
   many Rust programs.
 - **piston-image**: A modular game engine. An interesting Rust program.
-- **regex**: A regular expression parser. Used by many Rust programs.
+- **regex-1.5.5**: A regular expression parser. Used by many Rust programs.
 - **ripgrep**: A line-oriented search tool. A widely-used utility.
 - **ripgrep-13.0.0**: A line-oriented search tool. A widely-used utility.
 - **serde**: A serialization/deserialization crate. Used by many other
@@ -140,7 +140,7 @@ Rust code being written today.
 - **inflate**: An old implementation of the DEFLATE algorithm. Contains
   a very large function containing many locals and basic blocks, similar to
   `keccak` but less extreme.
-- **regex**: See above.
+- **regex**: See above. This is an older version of the crate.
 - **piston-image**: See above.
 - **style-servo**: An old version of Servo's `style` crate. A large crate, and
   one used by old versions of Firefox.

collector/benchmarks/regex-1.5.5/.cargo_vcs_info.json

@@ -0,0 +1,6 @@
+{
+  "git": {
+    "sha1": "d130381b150756ba7e5940efdc6ebdf47f4febc0"
+  },
+  "path_in_vcs": ""
+}

collector/benchmarks/regex-1.5.5/.gitignore

@@ -0,0 +1,7 @@
+target
+bench-log
+.*.swp
+wiki
+tags
+examples/debug.rs
+tmp/

collector/benchmarks/regex-1.5.5/0-compile-one.patch

@@ -0,0 +1,13 @@
+diff --git a/src/compile.rs b/src/compile.rs
+index 9db743f..ef1948e 100644
+--- a/src/compile.rs
++++ b/src/compile.rs
+@@ -137,6 +137,8 @@ impl Compiler {
+     }
+
+     fn compile_one(mut self, expr: &Hir) -> result::Result<Program, Error> {
++        {} // @030
++
+         // If we're compiling a forward DFA and we aren't anchored, then
+         // add a `.*?` before the first capture group.
+         // Other matching engines handle this by baking the logic into the

collector/benchmarks/regex-1.5.5/1-is-valid-cap-letter.patch

@@ -0,0 +1,13 @@
+diff --git a/src/expand.rs b/src/expand.rs
+index 9bea703..3b6ae94 100644
+--- a/src/expand.rs
++++ b/src/expand.rs
+@@ -128,6 +128,7 @@ fn find_cap_ref(replacement: &[u8]) -> Option<CaptureRef<'_>> {
+ }
+
+ /// Returns true if and only if the given byte is allowed in a capture name.
+ fn is_valid_cap_letter(b: &u8) -> bool {
++    { }
+     match *b {
+         b'0'..=b'9' | b'a'..=b'z' | b'A'..=b'Z' | b'_' => true,
+         _ => false,

collector/benchmarks/regex-1.5.5/2-Compiler-new.patch

@@ -0,0 +1,12 @@
+diff --git a/src/compile.rs b/src/compile.rs
+index 9db743f..fb812ae 100644
+--- a/src/compile.rs
++++ b/src/compile.rs
+@@ -54,6 +54,7 @@ impl Compiler {
+     ///
+     /// Various options can be set before calling `compile` on an expression.
+     pub fn new() -> Self {
++        {}
+         Compiler {
+             insts: vec![],
+             compiled: Program::new(),

collector/benchmarks/regex-1.5.5/3-reverse.patch

@@ -0,0 +1,12 @@
+diff --git a/src/compile.rs b/src/compile.rs
+index 9db743f..4e56c2d 100644
+--- a/src/compile.rs
++++ b/src/compile.rs
+@@ -114,6 +114,7 @@ impl Compiler {
+     /// When set, the machine returned is suitable for matching text in
+     /// reverse. In particular, all concatenations are flipped.
+     pub fn reverse(mut self, yes: bool) -> Self {
++        {}
+         self.compiled.is_reverse = yes;
+         self
+     }

collector/benchmarks/regex-1.5.5/4-byte-frequencies.patch

@@ -0,0 +1,13 @@
+diff --git a/src/freqs.rs b/src/freqs.rs
+index 92bafc1..6eb5799 100644
+--- a/src/freqs.rs
++++ b/src/freqs.rs
+@@ -2,7 +2,7 @@
+ // edit directly
+
+ pub const BYTE_FREQUENCIES: [u8; 256] = [
+-    55,  // '\x00'
++    54+1,  // '\x00'
+     52,  // '\x01'
+     51,  // '\x02'
+     50,  // '\x03'

collector/benchmarks/regex-1.5.5/5-Job.patch

@@ -0,0 +1,14 @@
+diff --git a/src/backtrack.rs b/src/backtrack.rs
+index 3c06254..4b72fd4 100644
+--- a/src/backtrack.rs
++++ b/src/backtrack.rs
+@@ -82,8 +82,8 @@ impl Cache {
+ /// stack to do it.
+ #[derive(Clone, Copy, Debug)]
+ enum Job {
+-    Inst { ip: InstPtr, at: InputAt },
+     SaveRestore { slot: usize, old_pos: Option<usize> },
++    Inst { ip: InstPtr, at: InputAt },
+ }
+
+ impl<'a, 'm, 'r, 's, I: Input> Bounded<'a, 'm, 'r, 's, I> {

collector/benchmarks/regex-1.5.5/6-println.patch

@@ -0,0 +1,12 @@
+diff --git a/src/re_set.rs b/src/re_set.rs
+index 95c4306..eff56b0 100644
+--- a/src/re_set.rs
++++ b/src/re_set.rs
+@@ -216,6 +216,7 @@ pub struct SetMatches {
+ impl SetMatches {
+     /// Whether this set contains any matches.
+     pub fn matched_any(&self) -> bool {
++        println!("testing");
+         self.matched_any
+     }
+