Rollup merge of #121533 - ratmice:wasm_init_fini_array, r=nnethercote

Handle .init_array link_section specially on wasm

Given that wasm-ld now has support for [.init_array](8f2bd8ae68/llvm/lib/MC/WasmObjectWriter.cpp (L1852)), it appears we can easily implement that section by falling through to the normal path rather than taking the typical custom_section path for wasm.

The wasm-ld appears to have a bunch of limitations. Only one static with the `link_section` in a crate or else you hit the fatal error in the link above "only one .init_array section fragment supported". They do not get merged.

You can still call multiple constructors by setting it to an array.

```
unsafe extern "C" fn ctor() {
    println!("foo");
}
#[used]
#[link_section = ".init_array"]
static FOO: [unsafe extern "C" fn(); 2] = [ctor, ctor];
```

Another issue appears to be that if crate *A* depends on crate *B*, but *A* doesn't call any symbols from *B* and *B* doesn't `#[export_name = ...]` any symbols, then crate *B*'s constructor will not be called.  The workaround to this is to provide an exported symbol in crate *B*.
This commit is contained in:
Trevor Gross 2024-07-19 03:27:46 -05:00 committed by GitHub
commit 986d6bf9fb
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 36 additions and 9 deletions

View File

@ -495,8 +495,14 @@ impl<'ll> CodegenCx<'ll, '_> {
}
// Wasm statics with custom link sections get special treatment as they
// go into custom sections of the wasm executable.
if self.tcx.sess.target.is_like_wasm {
// go into custom sections of the wasm executable. The exception to this
// is the `.init_array` section which are treated specially by the wasm linker.
if self.tcx.sess.target.is_like_wasm
&& attrs
.link_section
.map(|link_section| !link_section.as_str().starts_with(".init_array"))
.unwrap_or(true)
{
if let Some(section) = attrs.link_section {
let section = llvm::LLVMMDStringInContext2(
self.llcx,

View File

@ -166,21 +166,42 @@ fn maybe_check_static_with_link_section(tcx: TyCtxt<'_>, id: LocalDefId) {
return;
}
// For the wasm32 target statics with `#[link_section]` are placed into custom
// sections of the final output file, but this isn't link custom sections of
// other executable formats. Namely we can only embed a list of bytes,
// nothing with provenance (pointers to anything else). If any provenance
// show up, reject it here.
// For the wasm32 target statics with `#[link_section]` other than `.init_array`
// are placed into custom sections of the final output file, but this isn't like
// custom sections of other executable formats. Namely we can only embed a list
// of bytes, nothing with provenance (pointers to anything else). If any
// provenance show up, reject it here.
// `#[link_section]` may contain arbitrary, or even undefined bytes, but it is
// the consumer's responsibility to ensure all bytes that have been read
// have defined values.
//
// The `.init_array` section is left to go through the normal custom section code path.
// When dealing with `.init_array` wasm-ld currently has several limitations. This manifests
// in workarounds in user-code.
//
// * The linker fails to merge multiple items in a crate into the .init_array section.
// To work around this, a single array can be used placing multiple items in the array.
// #[link_section = ".init_array"]
// static FOO: [unsafe extern "C" fn(); 2] = [ctor, ctor];
// * Even symbols marked used get gc'd from dependant crates unless at least one symbol
// in the crate is marked with an `#[export_name]`
//
// Once `.init_array` support in wasm-ld is complete, the user code workarounds should
// continue to work, but would no longer be necessary.
if let Ok(alloc) = tcx.eval_static_initializer(id.to_def_id())
&& alloc.inner().provenance().ptrs().len() != 0
{
let msg = "statics with a custom `#[link_section]` must be a \
if attrs
.link_section
.map(|link_section| !link_section.as_str().starts_with(".init_array"))
.unwrap()
{
let msg = "statics with a custom `#[link_section]` must be a \
simple list of bytes on the wasm target with no \
extra levels of indirection such as references";
tcx.dcx().span_err(tcx.def_span(id), msg);
tcx.dcx().span_err(tcx.def_span(id), msg);
}
}
}