From 003a3f8cd3ea4594e2c2bc89c05562a202868ce0 Mon Sep 17 00:00:00 2001 From: Nicholas Nethercote Date: Thu, 20 Oct 2022 18:59:07 +1100 Subject: 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). --- src/test/codegen/match-optimized.rs | 60 +++++++++++++++++++++++++++++++++++ src/test/codegen/match-unoptimized.rs | 23 ++++++++++++++ src/test/codegen/match.rs | 29 ----------------- 3 files changed, 83 insertions(+), 29 deletions(-) create mode 100644 src/test/codegen/match-optimized.rs create mode 100644 src/test/codegen/match-unoptimized.rs delete mode 100644 src/test/codegen/match.rs (limited to 'src/test/codegen') diff --git a/src/test/codegen/match-optimized.rs b/src/test/codegen/match-optimized.rs new file mode 100644 index 00000000000..36402cc7353 --- /dev/null +++ b/src/test/codegen/match-optimized.rs @@ -0,0 +1,60 @@ +// 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, {{i8\*|ptr}} %1, align 1 +// CHECK-NEXT: br label %[[EXIT:[a-zA-Z0-9_]+]] +// CHECK: [[B]]: +// CHECK-NEXT: store i8 1, {{i8\*|ptr}} %1, align 1 +// CHECK-NEXT: br label %[[EXIT]] +// CHECK: [[C]]: +// CHECK-NEXT: store i8 2, {{i8\*|ptr}} %1, align 1 +// CHECK-NEXT: br label %[[EXIT]] + match e { + E::A => 0, + E::B => 1, + E::C => 2, + } +} + +#[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 `src/test/codegen/match-unoptimized.rs`. + +// 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, + } +} diff --git a/src/test/codegen/match-unoptimized.rs b/src/test/codegen/match-unoptimized.rs new file mode 100644 index 00000000000..be40b29e3d3 --- /dev/null +++ b/src/test/codegen/match-unoptimized.rs @@ -0,0 +1,23 @@ +// compile-flags: -C no-prepopulate-passes -Copt-level=0 + +#![crate_type = "lib"] + +#[repr(u16)] +pub enum E2 { + A = 13, + B = 42, +} + +// For unoptimized code we produce a `br` instead of a `switch`. Compare with +// `src/test/codegen/match-optimized.rs` + +// CHECK-LABEL: @exhaustive_match_2 +#[no_mangle] +pub fn exhaustive_match_2(e: E2) -> u8 { + // CHECK: %[[CMP:.+]] = icmp eq i16 %{{.+}}, 13 + // CHECK-NEXT: br i1 %[[CMP:.+]], + match e { + E2::A => 0, + E2::B => 1, + } +} diff --git a/src/test/codegen/match.rs b/src/test/codegen/match.rs deleted file mode 100644 index b203641fddb..00000000000 --- a/src/test/codegen/match.rs +++ /dev/null @@ -1,29 +0,0 @@ -// compile-flags: -C no-prepopulate-passes - -#![crate_type = "lib"] - -pub enum E { - A, - B, -} - -// 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: ] -// CHECK: [[OTHERWISE]]: -// CHECK-NEXT: unreachable -// CHECK: [[A]]: -// CHECK-NEXT: store i8 0, {{i8\*|ptr}} %1, align 1 -// CHECK-NEXT: br label %[[EXIT:[a-zA-Z0-9_]+]] -// CHECK: [[B]]: -// CHECK-NEXT: store i8 1, {{i8\*|ptr}} %1, align 1 -// CHECK-NEXT: br label %[[EXIT:[a-zA-Z0-9_]+]] - match e { - E::A => 0, - E::B => 1, - } -} -- cgit 1.4.1-3-g733a5