bootstrap: Memoize the LLVM rebuild hash to avoid very slow `x check`
Recently I've encountered a massive regression in the performance of re-running `x check` after making no changes.
It used to take around 2 seconds, and now it takes 20-30 seconds. That's quite a hassle when r-a runs it every time I save.
After some poking around, what I've found is that each individual call to `generate_smart_stamp_hash` doesn't take a particularly long time (around 0.5 sec), but it gets called dozens of times during `x check`, and that seems to be what's adding up to 20-30 seconds.
---
https://rust-lang.zulipchat.com/#narrow/stream/326414-t-infra.2Fbootstrap/topic/Massive.20regression.20in.20no-op.20.60x.20check.60
cc `@onur-ozkan`
skip {tidy,compiletest,rustdoc-gui} based tests for `DocTests::Only`
As use of `--doc` with `x test` is intended for running doc-tests only, executing compiletest, tidy or rustdoc-gui based tests considered as an incorrect behavior from bootstrap. This change fixes that.
rustdoc: Remove space from fake-variadic fn ptr impls
before: `for fn (T₁, T₂, …, Tₙ) -> Ret`
after: `for fn(T₁, T₂, …, Tₙ) -> Ret`
I don't think we usually have spaces there, so it looks weird.
cc `@notriddle` since you added the space in https://github.com/rust-lang/rust/pull/98180 (or rather, added the feature with a space included).
Eagerly compute output_filenames
It can be computed before creating TyCtxt. Previously the query would also write the dep info file, which meant that the output filenames couldn't be accessed before macro expansion is done. The dep info file writing is now done as a separate non-query function. The old query was always executed again anyways due to depending on the HIR.
Also encode the output_filenames in rlink files to ensure `#![crate_name]` affects the linking stage when doing separate compiling and linking using `-Zno-link`/`-Zlink-only`.
Non null convenience ops
Based on https://github.com/rust-lang/libs-team/issues/251.
I went through all of the methods on `*mut` and added every method, which does not require additional safety conditions, to `NonNull`. (exceptions: `guaranteed_eq`, `guaranteed_ne`, `with_metadata_of`, it's unclear if they are useful here...)
I'm also not sure what types should the "second pointer parameter" be. `*mut`/`*const` might be more permissible, but given that `NonNull` doesn't coerce to them, it might also be annoying. For now I chose the "use `NonNull` everywhere" path, but I'm not sure it's the correct one...
<sub>I'm eepy, so I probably messed up somewhere while copying...</sub>
cc `@scottmcm`
r? libs-api
As use of `--doc` with `x test` is intended for running doc-tests only, executing
compiletest, tidy or rustdoc-gui based tests considered as an incorrect behavior
from bootstrap. This change fixes that.
Signed-off-by: onur-ozkan <work@onurozkan.dev>
Remove myself from users on vacation
I think I have some capacity for reviews now and recently there were a few PRs with explicit `r? `@WaffleLapkin`` which I'm actually capable to review, so
Call FileEncoder::finish in rmeta encoding
Fixes https://github.com/rust-lang/rust/issues/117254
The bug here was that rmeta encoding never called FileEncoder::finish. Now it does. Most of the changes here are needed to support that, since rmeta encoding wants to finish _then_ access the File in the encoder, so finish can't move out.
I tried adding a `cfg(debug_assertions)` exploding Drop impl to FileEncoder that checked for finish being called before dropping, but fatal errors cause unwinding so this isn't really possible. If we encounter a fatal error with a dirty FileEncoder, the Drop impl ICEs even though the implementation is correct. If we try to paper over that by wrapping FileEncoder in ManuallyDrop then that just erases the fact that Drop automatically checks that we call finish on all paths.
I also changed the name of DepGraph::encode to DepGraph::finish_encoding, because that's what it does and it makes the fact that it is the path to FileEncoder::finish less confusing.
r? `@WaffleLapkin`
`AmbiguityCause` should not eagerly format strings
Minor tweak found when working on some coherence diagnostics stuff (towards `-Ztrait-solver=next-coherence` stabilization)
Refactor borrowck liveness values
This PR starts cleaning up `rustc_borrowck`, in particular around liveness values:
- refactors simple names that make no sense anymore: either referring to older structures using region elements, or to bitset containers and values.
- improves comments and fixes others
- removes unused return values and unneeded generic arguments
r? `@matthewjasper`
Add `debug_assert_nounwind` and convert `assert_unsafe_precondition`
`assert_unsafe_precondition` checks non-CTFE-evaluable conditions in runtime and performs no-op in compile time, while many of its current usage can be checked during const eval.
