self-profiling: Add events for metadata loading (plus a small dep-tracking optimization)
This PR
- adds self-profiling events related to loading things from crate metadata
- makes the compiler cache the `DepNodeIndex` of upstream crates, so that they don't have to be looked up over and over.
The commits are best reviewed in isolation.
Self-profiling tracking issue: https://github.com/rust-lang/rust/issues/58967
r? @Mark-Simulacrum
cc @wesleywiser
It means an allocation is required to create an empty `TokenStream`, but
all other operations are simpler and marginally faster due to not having
to check for `None`. Overall it simplifies the code for a negligible
performance effect.
The commit also removes `TokenStream::empty` by implementing `Default`,
which is now possible.
Rollup of 10 pull requests
Successful merges:
- #65214 (Split non-CAS atomic support off into target_has_atomic_load_store)
- #65246 (vxWorks: implement get_path() and get_mode() for File fmt::Debug)
- #65312 (improve performance of signed saturating_mul)
- #65336 (Fix typo in task::Waker)
- #65346 (nounwind tests and cleanup)
- #65347 (Fix #[unwind(abort)] with Rust ABI)
- #65366 (Implement Error::source on IntoStringError + Remove superfluous cause impls)
- #65369 (Don't discard value names when using address or memory sanitizer)
- #65370 (Add `dyn` to `Any` documentation)
- #65373 (Fix typo in docs for `Rc`)
Failed merges:
r? @ghost
Add `dyn` to `Any` documentation
I noticed that in documentation to `Any` trait the old trait object syntax is used, which could be confusing for newcomers, since we generally recommend using `dyn Trait` instead of just `Trait`. This PR changes the documentation comment, so that it uses `&dyn Any`, `&mut dyn Any` and `Box<dyn Any>`, correspondingly.
Don't discard value names when using address or memory sanitizer
The value names are used when reporting issues found by address
sanitizer or memory sanitizer. Avoid discarding names when those
sanitizers are enabled, unless explicitly requested to do otherwise.
Implement Error::source on IntoStringError + Remove superfluous cause impls
IntoStringError only implemented `Error::cause`, which is deprecated. This implemements `Error::source` instead.
`Error::cause` will still work as before, thanks to the default implementation.
I think this was the only/last `Error` impl in the standard library to have a cause, but not a source.
nounwind tests and cleanup
This is a follow-up to @pnkfelix' https://github.com/rust-lang/rust/pull/65020. In particular it adds some tests as @nagisa asked. It also does a cleanup that the original PR omitted to reduce backporting risks.
I hope I finally managed to write an uncontroversial PR in this area. ;) This should not change any behavior, just test it better.
Split non-CAS atomic support off into target_has_atomic_load_store
This PR implements my proposed changes in https://github.com/rust-lang/rust/issues/32976#issuecomment-518542029 by removing `target_has_atomic = "cas"` and splitting `target_has_atomic` into two separate `cfg`s:
* `target_has_atomic = 8/16/32/64/128`: This indicates the largest width that the target can atomically CAS (which implies support for all atomic operations).
* ` target_has_atomic_load_store = 8/16/32/64/128`: This indicates the largest width that the target can support loading or storing atomically (but may not support CAS).
cc #32976
r? @alexcrichton
I noticed that in documentation to `Any` trait the old trait object syntax is used, which could be confusing for newcomers, since we generally recommend using `dyn Trait` instead of just `Trait`. This PR changes the documentation comment, so that it uses `&dyn Any`, `&mut dyn Any` and `Box<dyn Any>`, correspondingly.
