rust/tests/codegen/match-optimized.rs

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

61 lines
1.5 KiB
Rust
Raw Normal View History

Use `br` instead of `switch` in more cases. `codegen_switchint_terminator` already uses `br` instead of `switch` when there is one normal target plus the `otherwise` target. But there's another common case with two normal targets and an `otherwise` target that points to an empty unreachable BB. This comes up a lot when switching on the tags of enums that use niches. The pattern looks like this: ``` bb1: ; preds = %bb6 %3 = load i8, ptr %_2, align 1, !range !9, !noundef !4 %4 = sub i8 %3, 2 %5 = icmp eq i8 %4, 0 %_6 = select i1 %5, i64 0, i64 1 switch i64 %_6, label %bb3 [ i64 0, label %bb4 i64 1, label %bb2 ] bb3: ; preds = %bb1 unreachable ``` This commit adds code to convert the `switch` to a `br`: ``` bb1: ; preds = %bb6 %3 = load i8, ptr %_2, align 1, !range !9, !noundef !4 %4 = sub i8 %3, 2 %5 = icmp eq i8 %4, 0 %_6 = select i1 %5, i64 0, i64 1 %6 = icmp eq i64 %_6, 0 br i1 %6, label %bb4, label %bb2 bb3: ; No predecessors! unreachable ``` This has a surprisingly large effect on compile times, with reductions of 5% on debug builds of some crates. The reduction is all due to LLVM taking less time. Maybe LLVM is just much better at handling `br` than `switch`. The resulting code is still suboptimal. - The `icmp`, `select`, `icmp` sequence is silly, converting an `i1` to an `i64` and back to an `i1`. But with the current code structure it's hard to avoid, and LLVM will easily clean it up, in opt builds at least. - `bb3` is usually now truly dead code (though not always, so it can't be removed universally).
2022-10-20 07:59:07 +00:00
//@ compile-flags: -C no-prepopulate-passes -O
#![crate_type = "lib"]
pub enum E {
A,
B,
C,
}
// CHECK-LABEL: @exhaustive_match
#[no_mangle]
pub fn exhaustive_match(e: E) -> u8 {
// CHECK: switch{{.*}}, label %[[OTHERWISE:[a-zA-Z0-9_]+]] [
// CHECK-NEXT: i[[TY:[0-9]+]] [[DISCR:[0-9]+]], label %[[A:[a-zA-Z0-9_]+]]
// CHECK-NEXT: i[[TY:[0-9]+]] [[DISCR:[0-9]+]], label %[[B:[a-zA-Z0-9_]+]]
// CHECK-NEXT: i[[TY:[0-9]+]] [[DISCR:[0-9]+]], label %[[C:[a-zA-Z0-9_]+]]
// CHECK-NEXT: ]
// CHECK: [[OTHERWISE]]:
// CHECK-NEXT: unreachable
//
// CHECK: [[A]]:
// CHECK-NEXT: store i8 0, ptr %_0, align 1
// CHECK-NEXT: br label %[[EXIT:[a-zA-Z0-9_]+]]
// CHECK: [[B]]:
// CHECK-NEXT: store i8 1, ptr %_0, align 1
// CHECK-NEXT: br label %[[EXIT]]
// CHECK: [[C]]:
// CHECK-NEXT: store i8 3, ptr %_0, align 1
// CHECK-NEXT: br label %[[EXIT]]
Use `br` instead of `switch` in more cases. `codegen_switchint_terminator` already uses `br` instead of `switch` when there is one normal target plus the `otherwise` target. But there's another common case with two normal targets and an `otherwise` target that points to an empty unreachable BB. This comes up a lot when switching on the tags of enums that use niches. The pattern looks like this: ``` bb1: ; preds = %bb6 %3 = load i8, ptr %_2, align 1, !range !9, !noundef !4 %4 = sub i8 %3, 2 %5 = icmp eq i8 %4, 0 %_6 = select i1 %5, i64 0, i64 1 switch i64 %_6, label %bb3 [ i64 0, label %bb4 i64 1, label %bb2 ] bb3: ; preds = %bb1 unreachable ``` This commit adds code to convert the `switch` to a `br`: ``` bb1: ; preds = %bb6 %3 = load i8, ptr %_2, align 1, !range !9, !noundef !4 %4 = sub i8 %3, 2 %5 = icmp eq i8 %4, 0 %_6 = select i1 %5, i64 0, i64 1 %6 = icmp eq i64 %_6, 0 br i1 %6, label %bb4, label %bb2 bb3: ; No predecessors! unreachable ``` This has a surprisingly large effect on compile times, with reductions of 5% on debug builds of some crates. The reduction is all due to LLVM taking less time. Maybe LLVM is just much better at handling `br` than `switch`. The resulting code is still suboptimal. - The `icmp`, `select`, `icmp` sequence is silly, converting an `i1` to an `i64` and back to an `i1`. But with the current code structure it's hard to avoid, and LLVM will easily clean it up, in opt builds at least. - `bb3` is usually now truly dead code (though not always, so it can't be removed universally).
2022-10-20 07:59:07 +00:00
match e {
E::A => 0,
E::B => 1,
E::C => 3,
Use `br` instead of `switch` in more cases. `codegen_switchint_terminator` already uses `br` instead of `switch` when there is one normal target plus the `otherwise` target. But there's another common case with two normal targets and an `otherwise` target that points to an empty unreachable BB. This comes up a lot when switching on the tags of enums that use niches. The pattern looks like this: ``` bb1: ; preds = %bb6 %3 = load i8, ptr %_2, align 1, !range !9, !noundef !4 %4 = sub i8 %3, 2 %5 = icmp eq i8 %4, 0 %_6 = select i1 %5, i64 0, i64 1 switch i64 %_6, label %bb3 [ i64 0, label %bb4 i64 1, label %bb2 ] bb3: ; preds = %bb1 unreachable ``` This commit adds code to convert the `switch` to a `br`: ``` bb1: ; preds = %bb6 %3 = load i8, ptr %_2, align 1, !range !9, !noundef !4 %4 = sub i8 %3, 2 %5 = icmp eq i8 %4, 0 %_6 = select i1 %5, i64 0, i64 1 %6 = icmp eq i64 %_6, 0 br i1 %6, label %bb4, label %bb2 bb3: ; No predecessors! unreachable ``` This has a surprisingly large effect on compile times, with reductions of 5% on debug builds of some crates. The reduction is all due to LLVM taking less time. Maybe LLVM is just much better at handling `br` than `switch`. The resulting code is still suboptimal. - The `icmp`, `select`, `icmp` sequence is silly, converting an `i1` to an `i64` and back to an `i1`. But with the current code structure it's hard to avoid, and LLVM will easily clean it up, in opt builds at least. - `bb3` is usually now truly dead code (though not always, so it can't be removed universally).
2022-10-20 07:59:07 +00:00
}
}
#[repr(u16)]
pub enum E2 {
A = 13,
B = 42,
}
// For optimized code we produce a switch with an unreachable target as the `otherwise` so LLVM
// knows the possible values. Compare with `tests/codegen/match-unoptimized.rs`.
Use `br` instead of `switch` in more cases. `codegen_switchint_terminator` already uses `br` instead of `switch` when there is one normal target plus the `otherwise` target. But there's another common case with two normal targets and an `otherwise` target that points to an empty unreachable BB. This comes up a lot when switching on the tags of enums that use niches. The pattern looks like this: ``` bb1: ; preds = %bb6 %3 = load i8, ptr %_2, align 1, !range !9, !noundef !4 %4 = sub i8 %3, 2 %5 = icmp eq i8 %4, 0 %_6 = select i1 %5, i64 0, i64 1 switch i64 %_6, label %bb3 [ i64 0, label %bb4 i64 1, label %bb2 ] bb3: ; preds = %bb1 unreachable ``` This commit adds code to convert the `switch` to a `br`: ``` bb1: ; preds = %bb6 %3 = load i8, ptr %_2, align 1, !range !9, !noundef !4 %4 = sub i8 %3, 2 %5 = icmp eq i8 %4, 0 %_6 = select i1 %5, i64 0, i64 1 %6 = icmp eq i64 %_6, 0 br i1 %6, label %bb4, label %bb2 bb3: ; No predecessors! unreachable ``` This has a surprisingly large effect on compile times, with reductions of 5% on debug builds of some crates. The reduction is all due to LLVM taking less time. Maybe LLVM is just much better at handling `br` than `switch`. The resulting code is still suboptimal. - The `icmp`, `select`, `icmp` sequence is silly, converting an `i1` to an `i64` and back to an `i1`. But with the current code structure it's hard to avoid, and LLVM will easily clean it up, in opt builds at least. - `bb3` is usually now truly dead code (though not always, so it can't be removed universally).
2022-10-20 07:59:07 +00:00
// CHECK-LABEL: @exhaustive_match_2
#[no_mangle]
pub fn exhaustive_match_2(e: E2) -> u8 {
// CHECK: switch i16 %{{.+}}, label %[[UNREACH:.+]] [
// CHECK-NEXT: i16 13,
// CHECK-NEXT: i16 42,
// CHECK-NEXT: ]
// CHECK: [[UNREACH]]:
// CHECK-NEXT: unreachable
match e {
E2::A => 0,
E2::B => 1,
}
}