Jump to conversation
Unresolved conversations (0)
Nice work!

Nice work!

All of your conversations have been resolved.

Resolved conversations (6)
@petrochenkov petrochenkov Apr 18, 2022
The check fails on Windows due to a forward slash.
Outdated
...ake-fulldeps/reproducible-build/linker.rs
@petrochenkov petrochenkov Apr 9, 2022
`fn exported_symbols` and `fn linked_symbols` are nearly identical. Maybe the common part can be moved to a separate function taking the `info.level.is_below_threshold(export_threshold) || info.used` / `info.level.is_below_threshold(export_threshold)` piece as a closure parameter. UPD: Or just a boolean parameter telling to consider `info.used`.
...iler/rustc_codegen_ssa/src/back/linker.rs
petrochenkov nbdd0121
@petrochenkov petrochenkov Apr 9, 2022
```suggestion // due to their `SymbolExportLevel::Rust` export level but may end up being exported from dylibs. ```
Outdated
compiler/rustc_passes/src/reachable.rs
@bjorn3 bjorn3 Apr 3, 2022
Is there any reason why `#[thread_local]` would be combined with `#[used]`?
Outdated
...stc_codegen_ssa/src/back/symbol_export.rs
nbdd0121
@bjorn3 bjorn3 Apr 3, 2022
Why should symbols that aren't exported be preserved?
...iler/rustc_codegen_ssa/src/back/linker.rs
bjorn3 carbotaniuman
nbdd0121
@bjorn3 bjorn3 Apr 3, 2022
This is technically not valid for TLS symbols. They need a specific platform dependent relocation kind AFAIK. Some linkers may allow this in discarded sections, but I'm not sure all of them do.
Outdated
compiler/rustc_codegen_ssa/src/back/link.rs
nbdd0121 bjorn3