[cranelift] Rejigger the compile API (#4540)

* Move `emit_to_memory` to `MachCompileResult`

This small refactoring makes it clearer to me that emitting to memory
doesn't require anything else from the compilation `Context`. While it's
a trivial change, it's a small public API change that shouldn't cause
too much trouble, and doesn't seem RFC-worthy. Happy to hear different
opinions about this, though!

* hide the MachCompileResult behind a method

* Add a `CompileError` wrapper type that references a `Function`

* Rename MachCompileResult to CompiledCode

* Additionally remove the last unsafe API in cranelift-codegen
This commit is contained in:
Benjamin Bouvier
2022-08-02 21:05:40 +02:00
committed by GitHub
parent 37cd96beff
commit ff37c9d8a4
17 changed files with 156 additions and 198 deletions

View File

@@ -49,7 +49,7 @@ use crate::flowgraph;
use crate::ir::{self, Function};
#[cfg(feature = "unwind")]
use crate::isa::unwind::systemv::RegisterMappingError;
use crate::machinst::{MachCompileResult, TextSectionBuilder, UnwindInfoKind};
use crate::machinst::{CompiledCode, TextSectionBuilder, UnwindInfoKind};
use crate::settings;
use crate::settings::SetResult;
use crate::CodegenResult;
@@ -230,11 +230,7 @@ pub trait TargetIsa: fmt::Display + Send + Sync {
fn dynamic_vector_bytes(&self, dynamic_ty: ir::Type) -> u32;
/// Compile the given function.
fn compile_function(
&self,
func: &Function,
want_disasm: bool,
) -> CodegenResult<MachCompileResult>;
fn compile_function(&self, func: &Function, want_disasm: bool) -> CodegenResult<CompiledCode>;
#[cfg(feature = "unwind")]
/// Map a regalloc::Reg to its corresponding DWARF register.
@@ -254,7 +250,7 @@ pub trait TargetIsa: fmt::Display + Send + Sync {
#[cfg(feature = "unwind")]
fn emit_unwind_info(
&self,
result: &MachCompileResult,
result: &CompiledCode,
kind: UnwindInfoKind,
) -> CodegenResult<Option<crate::isa::unwind::UnwindInfo>>;