Enum ctor fn + branch = missed optimization/inlining

Abierto
#166,153 0 comentarios 0 reacciones 0 asignados Ver en GitHub

Los mantenedores suelen responder en 1 día

Nadie ha tomado este issue todavía.

Evaluación

Dificultad
3/5
Tiempo estimado
1-2 días
Aptitud para principiantes
49/100
Tipo de issue
Error
Claridad
Bastante claro
Estado de actividad
Activo
Stack tecnológico
rust

Línea de trabajo

The report includes blockmaker and blockmaker2 examples and assembly output, but names no rustc source file or test. Start by reproducing the difference on the linked Godbolt setup with nightly and -C opt-level=3, then trace how enum variant constructors used as function values are optimized. Done when the constructor function branch produces equivalent optimized output to the direct conditional construction, with a regression test covering the case.

Escrito por el modelo de indexación a partir del texto del issue.

Descripción

A-LLVM C-optimization T-compiler

I tried this code:

pub enum Block {
    Outlined(i32),
    Hollow(i32),
}

#[unsafe(no_mangle)]
pub fn blockmaker(condition: bool, mass: i32) -> Block {
    (if condition { Block::Hollow } else { Block::Outlined }) (mass)
}

#[unsafe(no_mangle)]
pub fn blockmaker2(condition: bool, mass: i32) -> Block {
    if condition { Block::Hollow(mass) } else { Block::Outlined(mass) }
}

I expected to see this happen: blockmaker should result in the same assembly as blockmaker2

Instead, this happened: They compile to different assembly:

example[9d75cd1f3a21f3ed]::Block::Hollow:
  mov eax, 1
  mov edx, edi
  ret

example[9d75cd1f3a21f3ed]::Block::Outlined:
  xor eax, eax
  mov edx, edi
  ret

blockmaker:
  lea rax, [rip + example[9d75cd1f3a21f3ed]::Block::Hollow]
  lea rcx, [rip + example[9d75cd1f3a21f3ed]::Block::Outlined]
  test edi, edi
  cmovne rcx, rax
  mov edi, esi
  jmp rcx

blockmaker2:
  mov edx, esi
  mov eax, edi
  ret

blockmaker uses a weird way to do conditional jumps (instead of using the traditional jnz/jz + jmp). Meanwhile, blockmaker2 seems to be properly inlined and optimized.

Godbolt link

This missed optimization happens on all three channels (stable 1.99.0/beta/nightly) on godbolt, for amd64, with -C opt-level=3

Lenguaje dominante
Rust
Estrellas
122k
Forks
16.7k
Merge medio
2 d 16 h
PR fusionados (30 d)
515

Preparar el entorno

Primeros pasos

  1. Lee el issue completo y luego la guía de contribución del proyecto.
  2. Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
  3. Haz un fork del repositorio y trabaja en una rama.
  4. Abre un pull request que haga referencia al número del issue.

Más de rust-lang/rust

Todos los issues de rust-lang/rust

Issues similares

Más issues de Rust

Recibe los nuevos issues en tu correo

Un resumen breve de issues de GitHub para principiantes.