Const prop aggregates even if partially or fully modified
r? @wesleywiser
cc @rust-lang/wg-mir-opt I'm moderately scared of this change, but I'm confident in having reviewed all the cases.
Rollup of 4 pull requests
Successful merges:
- #71840 (Rework MIR drop tree lowering)
- #71882 (Update the `cc` crate)
- #71945 (Sort "implementations on foreign types" section in the sidebar)
- #72043 (Add missing backtick in E0569 explanation)
Failed merges:
r? @ghost
Sort "implementations on foreign types" section in the sidebar
Fixes#71926.
We were sorting by the ID instead of sorting by the name. They're not in the same order as the implementations but I think it makes more sense this way considering this is what we do for the methods as well.
r? @kinnison
cc @rust-lang/rustdoc
Rework MIR drop tree lowering
This PR changes how drops are generated in MIR construction. This is the first half of the fix for #47949.
Rather than generating the drops for a given unwind/break/continue/return/generator drop path as soon as they are needed, the required drops are recorded and get generated later.
The motivation for this is
* It simplifies the caching scheme, because it's now possible to walk up the currently scheduled drop tree to recover state.
* The basic block order for MIR more closely resembles execution order.
This PR also:
* Highlights cleanup blocks in the graphviz MIR output.
* Removes some unnecessary drop flag assignments.
Fix disagreeement about CGU reuse and LTO
This commit fixes an issue where the codegen backend's selection of LTO
disagreed with what the codegen later thought was being done. Discovered
in #72006 we have a longstanding issue where if `-Clinker-plugin-lto` in
optimized mode is compiled incrementally it will always panic on the
second compilation. The underlying issue turned out to be that the
production of the original artifact determined that LTO should not be
done (because it's being postponed to the linker) but the CGU reuse
selection thought that LTO was done so it was trying to load pre-LTO
artifacts which were never generated.
The fix here is to ensure that the logic when generating code which
determines what kind of LTO is being done is shared amongst the CGU
reuse decision and the backend actually doing LTO. This means that
they'll both be in agreement about whether the previous compilation did
indeed produce incremental pre-LTO artifacts.
Closes#72006
This commit fixes an issue where the codegen backend's selection of LTO
disagreed with what the codegen later thought was being done. Discovered
in #72006 we have a longstanding issue where if `-Clinker-plugin-lto` in
optimized mode is compiled incrementally it will always panic on the
second compilation. The underlying issue turned out to be that the
production of the original artifact determined that LTO should not be
done (because it's being postponed to the linker) but the CGU reuse
selection thought that LTO was done so it was trying to load pre-LTO
artifacts which were never generated.
The fix here is to ensure that the logic when generating code which
determines what kind of LTO is being done is shared amongst the CGU
reuse decision and the backend actually doing LTO. This means that
they'll both be in agreement about whether the previous compilation did
indeed produce incremental pre-LTO artifacts.
Closes#72006
Simplify the `tcx.alloc_map` API
This PR changes all functions that require manually locking the `alloc_map` to functions on `TyCtxt` that lock the map internally. In the same step we make the `TyCtxt::alloc_map` field private.
r? @RalfJung
rustllvm: Use .init_array rather than .ctors
LLVM TargetMachines default to using the (now-legacy) .ctors
representation of init functions. Mixing .ctors and .init_array
representations can cause issues when linking with lld.
This happens in practice for:
* Our profiling runtime which is currently implicitly built with
.init_array since it is built by clang, which sets this field.
* External C/C++ code that may be linked into the same process.
Fixes: #71233
submodules: update cargo from f534844c2 to cb06cb269
Changes:
````
more clippy fixes
Document that bench is unstable in the man page.
Update assertions in LTO calculations
Updated comments in resolve.rs to reflect actual data strcture used.
Try to remove secrets from http.debug.
Revert always computing filename Metadata.
clean -p: call `get_many` once.
Implement new `clean -p` using globs.
Rework how Cargo computes the rustc file outputs.
Add CrateType to replace LibKind.
````
I'd like to get the fix for https://github.com/rust-lang/cargo/issues/8223 into nightly asap.
r? @ehuss
x.py: allow configuring the build directory
This allows configuring the directory for build artifacts, instead of having it always be `./build`. This means you can set it to a constant location, letting you reuse the same cache while working in several different directories.
The configuration lives in `config.toml` under `build.build-dir`. By default, it keeps the existing default of `./build`, but it can be configured to any relative or absolute path. Additionally, it allows making outputs relative to the root of the git repository using `$ROOT`.
r? @Mark-Simulacrum
Shrink `LocalDecl`
`LocalDecl` contributes 4-8% of peak heap memory usage on a range of benchmarks. This PR reduces its size from 128 bytes to 56 bytes on 64-bit, and does some clean-ups as well.
r? @matthewjasper
Add core::future::{pending,ready}
Adds two future constructors to `core`: `future::ready` and `future::pending`. These functions enable constructing futures of any type that either immediately resolve, or never resolve which is an incredible useful tool when writing documentation.
These functions have prior art in both the `futures` and `async-std` crates. This implementation has been adapted from the `futures` crate.
## Examples
In https://github.com/rust-lang/rust/pull/70817 we propose adding the `ready!` macro. In the example we use an `async fn` which does not return a future that implements `Unpin`, which leads to the use of `unsafe`. Instead had we had `future::ready` available, we could've written the same example without using `unsafe`:
```rust
use core::task::{Context, Poll};
use core::future::{self, Future};
use core::pin::Pin;
pub fn do_poll(cx: &mut Context<'_>) -> Poll<()> {
let mut fut = future::ready(42_u8);
let num = ready!(Pin::new(fut).poll(cx));
// ... use num
Poll::Ready(())
}
```
## Why future::ready?
Arguably `future::ready` and `async {}` can be considered equivalent. The main differences are that `future::ready` returns a future that implements `Unpin`, and the returned future is a concrete type. This is useful for traits that require a future as an associated type that can sometimes be a no-op ([example](https://docs.rs/http-service/0.4.0/http_service/trait.HttpService.html#associatedtype.ConnectionFuture)).
The final, minor argument is that `future::ready` and `future::pending` form a counterpart to the enum members of `Poll`: `Ready` and `Pending`. These functions form a conceptual bridge between `Poll` and `Future`, and can be used as a useful teaching device.
## References
- [`futures::future::ready`](https://docs.rs/futures/0.3.4/futures/future/fn.ready.html)
- [`futures::future::pending`](https://docs.rs/futures/0.3.4/futures/future/fn.pending.html)
- [`async_std::future::pending`](https://docs.rs/async-std/1.5.0/async_std/future/fn.pending.html)
- [`async_std::future::ready`](https://docs.rs/async-std/1.5.0/async_std/future/fn.ready.html)
This allows configuring the directory for build artifacts, instead of having it always be ./build. This means you can set it to a constant location, letting you reuse the same cache while working in several different directories.
The configuration lives in config.toml under build.build-dir. By default, it keeps the existing default of ./build, but it can be configured to any relative or absolute path. Additionally, it allows making outputs relative to the root of the git repository using $ROOT.
Changes:
````
more clippy fixes
Document that bench is unstable in the man page.
Update assertions in LTO calculations
Updated comments in resolve.rs to reflect actual data strcture used.
Try to remove secrets from http.debug.
Revert always computing filename Metadata.
clean -p: call `get_many` once.
Implement new `clean -p` using globs.
Rework how Cargo computes the rustc file outputs.
Add CrateType to replace LibKind.
````
Work around ICEs during cross-compilation for target, ast, & attr
This applies the fix for #72003 to work around #56935 to three more libraries. With these additional fixes, I'm able to use rustfmt_lib from wasm (https://github.com/rust-lang/rustfmt/issues/4132#issuecomment-616587989), which was my goal.
To get it working locally and to test, I copied the `.cargo/registry/src` and applied the fix and replaced the reference in my project:
``` toml
[replace]
"rustc-ap-rustc_span:656.0.0" = { path = "../rustc-ap-rustc_span" }
"rustc-ap-rustc_target:656.0.0" = { path = "../rustc-ap-rustc_target" }
"rustc-ap-rustc_ast:656.0.0" = { path = "../rustc-ap-rustc_ast" }
"rustc-ap-rustc_attr:656.0.0" = { path = "../rustc-ap-rustc_attr" }
```
allow wasm target for rustc-ap-rustc_span
This fixes#71998 by applying the work-a-round. The root cause is probably #56935, as @petrochenkov pointed out.
I reproduced the bug by:
```
cd ~/.cargo/registry/src/github.com-1ecc6299db9ec823/rustc-ap-rustc_span-657.0.0/
cargo build --target wasm32-unknown-unknown
```
Adding this line fixes it.
Reduce `TypedArena` creations in `check_match`.
`check_match` creates a new `TypedArena` for every call to
`create_and_enter`. DHAT tells me that each `TypedArena` typically is
barely used, with typically a single allocation per arena.
This commit moves the `TypedArena` creation outwards a bit, into
`check_match`, and then passes it into `create_and_enter`. This reduces
the number of arenas created by about 4-5x, for a very small perf win.
(Moving the arena creation further outwards is hard because
`check_match` is a query.)
r? @oli-obk
Improve bitcode generation for Apple platforms
Some improvements for iOS bitcode support suggested by Alex over at https://github.com/getditto/rust-bitcode/issues/9. r? @alexcrichton
This improves Rust's bitcode generation so that provided you have a compatible LLVM version, Rust targeting iOS should work out of the box when compiled into bitcode-enabled apps, and when submitted to the App Store. I've tested these changes using Xcode 11.4.1 and Apple's vendored LLVM, [tag `swift-5.2.3-RELEASE`](https://github.com/apple/llvm-project/releases/tag/swift-5.2.3-RELEASE).
1. Force `aarch64-apple-ios` and `aarch64-apple-tvos` targets to always emit full bitcode sections, even when cargo is trying to optimise by invoking `rustc` with `-Cembed-bitcode=no`. Since Apple recommends bitcode on iOS and requires it on tvOS it is likely that this is what developers intend. Currently you need to override the codegen options with `RUSTFLAGS`, which is far from obvious.
2. Provide an LLVM cmdline in the target spec. Apple's bitcode verification process looks for some arguments. For Rust modules to be accepted we must pretend they were produced similarly. A suitable default is provided in `TargetOptions` for iOS, copied directly from the a clang equivalent section.
In the context of Apple platforms, the predominant purpose of bitcode is App Store submissions, so simulator and 32-bit targets are not relevant. I'm hoping that the cmdline strings will not be a maintenance burden to keep up-to-date. If the event of any future incompatibilities, hopefully a custom target config would offer enough flexibility to work around it. It's impossible to say for sure.
Due to unrelated build errors I haven't been able to build and test a full tvOS toolchain. I've stopped short of providing a similar `bitcode_llvm_cmdline` until I can actually test it.
Unify lints handling in rustdoc
This is a small cleanup. The goal is to unify a bit things to make the reading simpler.
r? @kinnison
cc @rust-lang/rustdoc