about summary refs log tree commit diff
path: root/compiler/rustc_codegen_ssa/src/traits
diff options
context:
space:
mode:
authorbors <bors@rust-lang.org>2025-02-18 07:49:09 +0000
committerbors <bors@rust-lang.org>2025-02-18 07:49:09 +0000
commit3b022d8ceea570db9730be34d964f0cc663a567f (patch)
tree98b0584387859ce2ebcc221b9b86997cce17ba7b /compiler/rustc_codegen_ssa/src/traits
parentaaa861493456e8a10e552dd208f85486de772007 (diff)
parent7bb5f4dd78a7a45729ab503805ba1b3065cb0de9 (diff)
Auto merge of #133852 - x17jiri:cold_path, r=saethlin
improve cold_path()

#120370 added a new instrinsic `cold_path()` and used it to fix `likely` and `unlikely`

However, in order to limit scope, the information about cold code paths is only used in 2-target switch instructions. This is sufficient for `likely` and `unlikely`, but limits usefulness of `cold_path` for idiomatic rust. For example, code like this:

```
if let Some(x) = y { ... }
```

may generate 3-target switch:

```
switch y.discriminator:
0 => true branch
1 = > false branch
_ => unreachable
```

and therefore marking a branch as cold will have no effect.

This PR improves `cold_path()` to work with arbitrary switch instructions.

Note that for 2-target switches, we can use `llvm.expect`, but for multiple targets we need to manually emit branch weights. I checked Clang and it also emits weights in this situation. The Clang's weight calculation is more complex that this PR, which I believe is mainly because `switch` in `C/C++` can have multiple cases going to the same target.
Diffstat (limited to 'compiler/rustc_codegen_ssa/src/traits')
-rw-r--r--compiler/rustc_codegen_ssa/src/traits/builder.rs14
1 files changed, 14 insertions, 0 deletions
diff --git a/compiler/rustc_codegen_ssa/src/traits/builder.rs b/compiler/rustc_codegen_ssa/src/traits/builder.rs
index 2c843e2f5e4..48ae000f2c6 100644
--- a/compiler/rustc_codegen_ssa/src/traits/builder.rs
+++ b/compiler/rustc_codegen_ssa/src/traits/builder.rs
@@ -110,6 +110,20 @@ pub trait BuilderMethods<'a, 'tcx>:
         else_llbb: Self::BasicBlock,
         cases: impl ExactSizeIterator<Item = (u128, Self::BasicBlock)>,
     );
+
+    // This is like `switch()`, but every case has a bool flag indicating whether it's cold.
+    //
+    // Default implementation throws away the cold flags and calls `switch()`.
+    fn switch_with_weights(
+        &mut self,
+        v: Self::Value,
+        else_llbb: Self::BasicBlock,
+        _else_is_cold: bool,
+        cases: impl ExactSizeIterator<Item = (u128, Self::BasicBlock, bool)>,
+    ) {
+        self.switch(v, else_llbb, cases.map(|(val, bb, _)| (val, bb)))
+    }
+
     fn invoke(
         &mut self,
         llty: Self::Type,