Handle statics in MIR as const pointers
This is the first PR towards the goal of removing `PlaceBase::Static`. In this PR:
* Statics are lowered to dereferencing a const pointer.
* The temporaries holding such pointers are tracked in MIR, for the most part this is only used for diagnostics. There are two exceptions:
* The borrow checker has some checks for thread-locals that directly use this data.
* Const checking will suppress "cannot dereference raw pointer" diagnostics for pointers to `static mut`/`extern static`. This is to maintain the current behaviour (12 tests fail otherwise).
The following are left to future PRs (I think that @spastorino will be working on the first 3):
* Applying the same treatments to promoted statics.
* Removing `PlaceBase::Static`.
* Replacing `PlaceBase` with `Local`.
* Moving the ever growing collection of metadata that we have for diagnostics in MIR passes somewhere more appropriate.
r? @oli-obk
SourceMap is now in the root of all rustc-specific crates, syntax_pos,
so there's no need for the trait object to decouple the dependencies
between librustc_errors and libsyntax as was needed previously.
Split ConstValue into two enums
Hello,
Issue #59210 appeared abandoned, so I gave it a go.
Some further cleanup and refactoring may be mandated.
I did not test beyond `x.py check`, since my home computer dies compiling librustc.
Fixes#59210
Move self-profile infrastructure to data structures
The single dependency on queries (QueryName) can be fairly easily
abstracted via a trait and this further decouples Session from librustc
(the primary goal).
This is intended as a precursor to moving Session out of librustc, but since that involves lots of smaller steps that move around code I'm splitting it up into separate PRs.
Move Session fields to CrateStore
`allocator_kind` and `injected_panic_runtime` are both query-like, this moves them out of Session and into CrateStore, avoiding the `Once` they previously had by clearing separating initialization and de-initialization.
The single dependency on queries (QueryName) can be fairly easily
abstracted via a trait and this further decouples Session from librustc
(the primary goal).
rustc_metadata: don't let LLVM confuse rmeta blobs for COFF object files.
This has likely been a silent issue since 1.10 but only caused trouble recently (see https://github.com/rust-lang/rust/issues/65536#issuecomment-552018224), when recent changes to the `rmeta` schema introduced more opportunities for COFF parse errors.
To prevent any undesired interactions with old compilers, I've renamed the file inside `rlib`s from `rust.metadata.bin` to `lib.rmeta` (not strongly attached to it, suggestions welcome).
Fixes#65536.
<hr/>
Before:
```
$ llvm-objdump -all-headers build/*/stage1-std/*/release/deps/libcore-*.rmeta
build/x86_64-unknown-linux-gnu/stage1-std/x86_64-unknown-linux-gnu/release/deps/libcore-6b9e8b5a59b79a1d.rmeta: file format COFF-<unknown arch>
architecture: unknown
start address: 0x00000000
Sections:
Idx Name Size VMA Type
SYMBOL TABLE:
```
After:
```
$ llvm-objdump -all-headers build/*/stage1-std/*/release/deps/libcore-*.rmeta
llvm-objdump: error: 'build/x86_64-unknown-linux-gnu/stage1-std/x86_64-unknown-linux-gnu/release/deps/libcore-6b9e8b5a59b79a1d.rmeta':
The file was not recognized as a valid object file
```
syntax: ABI-oblivious grammar
This PR has the following effects:
1. `extern $lit` is now legal where `$lit:literal` and `$lit` is substituted for a string literal.
2. `extern "abi_that_does_not_exist"` is now *syntactically* legal whereas before, the set of ABI strings was hard-coded into the grammar of the language. With this PR, the set of ABIs are instead validated and translated during lowering. That seems more appropriate.
3. `ast::FloatTy` is now distinct from `rustc_target::abi::FloatTy`. The former is used substantially more and the translation between them is only necessary in a single place.
4. As a result of 2-3, libsyntax no longer depends on librustc_target, which should improve pipe-lining somewhat.
cc @rust-lang/lang -- the points 1-2 slightly change the definition of the language but in a way which seems consistent with our general principles (in particular wrt. the discussions of turning things into semantic errors). I expect this to be uncontroversial but it's worth letting y'all know. :)
r? @varkor
We also sever syntax's dependency on rustc_target as a result.
This should slightly improve pipe-lining.
Moreover, some cleanup is done in related code.
caller_location: point to macro invocation sites, like file!/line!, and use in core::panic!.
The main change here is to `core::panic!`, trying to fix this remaining regression: https://github.com/rust-lang/rust/pull/65927#issuecomment-547625147
However, in order for `caller_location` to be usable from macros the same way `file!()`/`line!()` are, it needs to have the same behavior (of extracting the macro invocation site `Span` and using that).
Arguably we would've had to do this at some point anyway, if we want to use `#[track_caller]` to replace the `file!()`/`line!()` uses from macros, but I'm not sure the RFC mentions this at all.
r? @petrochenkov cc @anp @nnethercote
Rename `LocalInternedString` and more
This PR renames `LocalInternedString` as `SymbolStr`, removes an unnecessary `impl` from it, improves comments, and cleans up some `SymbolStr` uses.
r? @estebank
Because it's highly magical, which goes against the goal of keeping
`SymbolStr` simple. Plus it's only used in a handful of places that
only require minor changes.
Revert PR 64324: dylibs export generics again (for now)
As discussed on PR #65781, this is a targeted attempt to undo the main semantic change from PR #64324, by putting `dylib` back in the set of crate types that export generic symbols.
The main reason to do this is that PR #64324 had unanticipated side-effects that caused bugs like #64872, and in the opinion of @alexcrichton and myself, the impact of #64872 is worse than #64319.
In other words, it is better for us, in the short term, to reopen#64319 as currently unfixed for now than to introduce new bugs like #64872.
Fix#64872Reopen#64319
Re-enable Emscripten's exception handling support
Passes LLVM codegen and Emscripten link-time flags for exception
handling if and only if the panic strategy is `unwind`. Sets the
default panic strategy for Emscripten targets to `unwind`. Re-enables
tests that depend on unwinding support for Emscripten, including
`should_panic` tests.
r? @alexcrichton
Fix#64153
This PR changes how the compiler detects if an object file from an upstream crate is a Rust object file or not. Instead of checking if the name starts with the crate name and ends with `.o` (which is not always the case, as described in #64153), it now just checks if the filename ends with `.rcgu.o`.
This fixes#64153. However, ideally we'd clean up the code around filename generation some more. Then this check could be made more robust.
r? @alexcrichton
Lint ignored `#[inline]` on function prototypes
Fixes https://github.com/rust-lang/rust/issues/51280.
- Adds a `unused_attribute` lint for `#[inline]` on function prototypes.
- As a consequence, foreign items, impl items and trait items now have their attributes checked, which could cause some code to no longer compile (it was previously erroneously ignored).
Before this commit, we had an event that would only track the compression step
for proc-macros and Rust dylibs. After the commit we measure the time for
acutally generating the crate metadata bytes.
