Fix abort-on-eprintln during process shutdown
This commit fixes an issue where if `eprintln!` is used in a TLS
destructor it can accidentally cause the process to abort. TLS
destructors are executed after `main` returns on the main thread, and at
this point we've also deinitialized global `Lazy` values like those
which store the `Stderr` and `Stdout` internals. This means that despite
handling TLS not being accessible in `eprintln!`, we will fail due to
not being able to call `stderr()`. This means that we'll double-panic
quickly because panicking also attempt to write to stderr.
The fix here is to reimplement the global stderr handle to avoid the
need for destruction. This avoids the need for `Lazy` as well as the
hidden panic inside of the `stderr` function.
Overall this should improve the robustness of printing errors and/or
panics in weird situations, since the `stderr` accessor should be
infallible in more situations.
add #[rustc_layout(debug)]
@eddyb recently told me about the `#[rustc_layout]` attribute, and I think it would be very useful if it could be used to print all the layout information Rust has about a type. When working with layouts (e.g. in Miri), it is often not clear how certain surface language features get represented internally. I have some awful hacks locally to be able to dump this debug information; with this attribute I could get it on the playground which is so much better. :)
hir: replace "items" terminology with "nodes" where appropriate.
The newly added `HirOwnerItems` confused me before I realized that "items" there actually referred to HIR nodes, not `hir:Item` or "item-like" (which we should IMO replace with "owner").
I suspect the naming had something to do with `ItemLocalId`'s use of "item".
That is, `ItemLocalId` could be interpreted to mean one of two things:
* `IntraItemNodeId` i.e. `IntraOwnerNodeId`
* this is IMO correct, and I'd even like to rename it, but I didn't want to throw that into this PR
* `IntraOwnerItemId`
* this is what `HirOwnerItems` would seem to imply
r? @Zoxc cc @michaelwoerister @nikomatsakis
Refactorings to get rid of rustc_codegen_utils
r? @eddyb
cc #45276
After this, the only modules left in `rustc_codegen_utils` are
- `link`: a bunch of linking-related functions (many dealing with file names). These are mostly consumed by save analysis, rustc_driver, rustc_interface, and of course codegen. I assume they live here because we don't want a dependency of save analysis on codegen... Perhaps they can be moved to librustc?
- ~`symbol_names` and `symbol_names_test`: honestly it seems odd that `symbol_names_test` is not a submodule of `symbol_names`. It seems like these could honestly live in their own crate or move to librustc. Already name mangling is exported as the `symbol_name` query.~ (move it to its own crate)
I don't mind doing either of the above as part of this PR or a followup if you want.
BTreeMap: remove shared root
This replaces the shared root with `Option`s in the BTreeMap code, and then slightly cleans up the node manipulation code taking advantage of the removal of the shared root. I expect that further simplification is possible, but wanted to get this posted for initial review.
Note that `BTreeMap::new()` continues to not allocate.
Benchmarks seem within the margin of error/unaffected, as expected for an entirely predictable branch.
```
name alloc-bench-a ns/iter alloc-bench-b ns/iter diff ns/iter diff % speedup
btree::map::iter_mut_20 20 21 1 5.00% x 0.95
btree::set::clone_100 1,360 1,439 79 5.81% x 0.95
btree::set::clone_100_and_into_iter 1,319 1,434 115 8.72% x 0.92
btree::set::clone_10k 143,515 150,991 7,476 5.21% x 0.95
btree::set::clone_10k_and_clear 142,792 152,916 10,124 7.09% x 0.93
btree::set::clone_10k_and_into_iter 146,019 154,561 8,542 5.85% x 0.94
```
can_begin_literal_maybe_minus: `true` on `"-"? lit` NTs.
Make `can_begin_literal_or_bool` (renamed to `can_begin_literal_maybe_minus`) accept `NtLiteral(e) | NtExpr(e)` where `e` is either a literal or a negated literal.
Fixes https://github.com/rust-lang/rust/issues/70050.
r? @petrochenkov
Use generator resume arguments in the async/await lowering
This removes the TLS requirement from async/await and enables it in `#![no_std]` crates.
Closes https://github.com/rust-lang/rust/issues/56974
I'm not confident the HIR lowering is completely correct, there seem to be quite a few undocumented invariants in there. The `async-std` and tokio test suites are passing with these changes though.
This commit fixes an issue where if `eprintln!` is used in a TLS
destructor it can accidentally cause the process to abort. TLS
destructors are executed after `main` returns on the main thread, and at
this point we've also deinitialized global `Lazy` values like those
which store the `Stderr` and `Stdout` internals. This means that despite
handling TLS not being accessible in `eprintln!`, we will fail due to
not being able to call `stderr()`. This means that we'll double-panic
quickly because panicking also attempt to write to stderr.
The fix here is to reimplement the global stderr handle to avoid the
need for destruction. This avoids the need for `Lazy` as well as the
hidden panic inside of the `stderr` function.
Overall this should improve the robustness of printing errors and/or
panics in weird situations, since the `stderr` accessor should be
infallible in more situations.
Add regression test for TAIT lifetime inference (issue #55099)
Fixes#55099
The minimized reproducer in issue #55099 now compiles successfully.
This commit adds a regression test for it.
Extend search
I realized that when looking for "struct:String" in the rustdoc search for example, the "in arguments" and "returned" tabs were always empty. After some investigation, I realized it was because we only provided the name, and not the type, making it impossible to pass the "type filtering" check.
To resolve this, I added the type alongside the name. Note for the future: we could improve this by instead only registering the path id and use the path dictionary directly. The only problem with that solution (which I already tested) is that it becomes complicated for types in other crates. It'd force us to handle both case with an id and a case with `(name, type)`. I found the current PR big enough to not want to provide it directly. However, I think this is definitely worth it to make it work this way in the future.
About the two tests I added: they don't have much interest except checking that we actually have something returned in the search in the cases of a type filtering with and without literal search.
I also had to update a bit the test script to add the new locally global (haha) variable I created (`NO_TYPE_FILTER`). I added this variable to make the code easier to read than just "-1".
r? @kinnison
cc @ollie27
std: Don't abort process when printing panics in tests
This commit fixes an issue when using `set_print` and friends, notably
used by libtest, to avoid aborting the process if printing panics. This
previously panicked due to borrowing a mutable `RefCell` twice, and this
is worked around by borrowing these cells for less time, instead
taking out and removing contents temporarily.
Closes#69558
Smaller and more correct generator codegen
This removes unnecessary panicking branches in the resume function when the generator can not return or unwind, respectively.
Closes https://github.com/rust-lang/rust/issues/66100
It also addresses the correctness concerns wrt poisoning on unwind. These are not currently a soundness issue because any operation *inside* a generator that could possibly unwind will result in a cleanup path for dropping it, ultimately reaching a `Resume` terminator, which we already handled correctly. Future MIR optimizations might optimize that out, though.
r? @Zoxc
Previously, metadata encoding used DUMMY_SP to represent any spans that
referenced an 'imported' SourceFile - e.g. a SourceFile from an upstream
dependency. These leads to sub-optimal error messages in certain cases
(see the included test).
This PR changes how we encode and decode spans in crate metadata. We
encode spans in one of two ways:
* 'Local' spans, which reference non-imported SourceFiles, are encoded
exactly as before.
* 'Foreign' spans, which reference imported SourceFiles, are encoded
with the CrateNum of their 'originating' crate. Additionally, their
'lo' and 'high' values are rebased on top of the 'originating' crate,
which allows them to be used with the SourceMap data encoded for that
crate.
The `ExternalSource` enum is renamed to `ExternalSourceKind`. There is
now a struct called `ExternalSource`, which holds an
`ExternalSourceKind` along with the original line number information for
the file. This is used during `Span` serialization to rebase spans onto
their 'owning' crate.
Remove some imports to the rustc crate
- When we have `NestedVisitorMap::None`, we use `type Map = dyn intravisit::Map<'v>;` instead of the actual map. This doesn't actually result in dynamic dispatch (in the future we may want to use an associated type default to simplify the code).
- Use `rustc_session::` imports instead of `rustc::{session, lint}`.
r? @Zoxc
Expansion-driven outline module parsing
After this PR, the parser will not do any conditional compilation or loading of external module files when `mod foo;` is encountered. Instead, the parser only leaves `mod foo;` in place in the AST, with no items filled in. Expansion later kicks in and will load the actual files and do the parsing. This entails that the following is now valid:
```rust
#[cfg(FALSE)]
mod foo {
mod bar {
mod baz; // `foo/bar/baz.rs` doesn't exist, but no error!
}
}
```
Fixes https://github.com/rust-lang/rust/issues/64197.
r? @petrochenkov
Use smaller discriminants for generators
Closes https://github.com/rust-lang/rust/issues/69815
I'm not yet sure about the runtime performance impact of this, so I'll try running this on some benchmarks (if I can find any). (Update: No impact on the benchmarks I've measured on)
* [x] Add test with a generator that has exactly 256 total states
* [x] Add test with a generator that has more than 256 states so that it needs to use a u16 discriminant
* [x] Add tests for the size of `Option<[generator]>`
* [x] Add tests for the `discriminant_value` intrinsic in all cases
Erase regions in writeback
Regions in `TypeckTables` (except canonicalized user annotations) are now erased. Further, we no longer do lexical region solving on item bodies with `-Zborrowck=mir`.
cc #68261
r? @nikomatsakis
