Document WebAssembly target feature expectations
This commit is a result of the discussion on #128475 and incorporates parts of #109807 as well. This is all done as a new page of documentation for the `wasm32-unknown-unknown` target which previously did not exist. This new page goes into details about the preexisting target and additionally documents the expectations for WebAssembly features and code generation.
The tl;dr is that LLVM will enable features over time after most engines have had support for awhile. Compiling without features requires `-Ctarget-cpu=mvp` to rustc plus `-Zbuild-std` to Cargo.
Closes#109807Closes#119811Closes#128475
Rollup of 8 pull requests
Successful merges:
- #127623 (fix: fs::remove_dir_all: treat internal ENOENT as success)
- #128876 (Ship MinGW-w64 runtime DLLs along with `rust-lld.exe` for `-pc-windows-gnu` targets)
- #129055 (Migrate `x86_64-fortanix-unknown-sgx-lvi` `run-make` test to rmake)
- #129386 (Use a LocalDefId in ResolvedArg.)
- #129400 (Update `compiler_builtins` to `0.1.120`)
- #129414 (Fix extern crates not being hidden with `doc(hidden)`)
- #129417 (Don't trigger refinement lint if predicates reference errors)
- #129433 (Fix a missing import in a doc in run-make-support)
r? `@ghost`
`@rustbot` modify labels: rollup
Fix extern crates not being hidden with `doc(hidden)`
Fixes#126796.
Only the current crate should never be stripped, any other crate should be strippable.
r? ``@notriddle``
Update `compiler_builtins` to `0.1.120`
Includes https://github.com/rust-lang/compiler-builtins/pull/672 which fixes regression issue with Apple and Windows compilers.
try-job: aarch64-apple
try-job: x86_64-apple-1
try-job: x86_64-msvc
Migrate `x86_64-fortanix-unknown-sgx-lvi` `run-make` test to rmake
Part of #121876 and the associated [Google Summer of Code project](https://blog.rust-lang.org/2024/05/01/gsoc-2024-selected-projects.html).
The final Makefile! Every Makefile test is now claimed.
This is difficult to test due to the uncommon architecture it is specific to. I don't think it is in the CI (I didn't find it in `jobs.yml`, but if there is a way to test it, please do.
Locally, on Linux, it compiles and panics at the `llvm_filecheck` part (if I replace the `x86_64-fortanix-unknown-sgx` with `x86_64-unknown-linux-gnu`, of course), which is expected.
For this reason, the Makefile and associated script have been kept, but with a leading underscore.
Ship MinGW-w64 runtime DLLs along with `rust-lld.exe` for `-pc-windows-gnu` targets
`rust-lld.exe` built for `x86_64-pc-windows-gnu` depends on `libgcc_s_seh-1.dll` and `libwinpthread-1.dll` from MinGW-w64. Until now, they were not shipped alongside `rust-lld.exe`, and you could not run `rust-lld.exe` on most systems.
This problem didn't surface until now because:
* Most targets don't use `rust-lld` by default.
* Some people had these DLLs in their `PATH` from some other MinGW binary.
* `rustup` used to add `bin` to the `PATH`, which contains these DLLs for `rustc.exe`. But it no longer does that: ce3c09a0cbFixes#125809
try-job: dist-x86_64-mingw
Revert #129187 and #129302
The two PRs naively switched to `std::fs::remove_dir_all`, but failed to gracefully handle the failure case where the top-level directory entry does not exist, causing https://github.com/rust-lang/rust/pull/129187#issuecomment-2304849757 `./x clean` to fail locally when `tmp` does not exist.
I plan to reland the two PRs with fixed top-level dir entry handling and more testing, but let's quickly revert to unblock people.
Reverts #129187.
Reverts #129302.
r? bootstrap
Allow rust staticlib to work with MSVC's /WHOLEARCHIVE
This fixes#129020 by renaming the `__NULL_IMPORT_DESCRIPTOR` to prevent conflicts.
try-job: dist-i686-msvc
Make Tree Borrows Provenance GC no longer produce stack overflows
Most functions operating on Tree Borrows' trees are carefully written to not cause stack overflows due to too much recursion. The one exception is [`Tree::keep_only_needed`](94f5588faf/src/borrow_tracker/tree_borrows/tree.rs (L724)), which just uses regular recursion.
This function is part of the provenance GC, so it is called regularly for every allocation in the program.
Tests show that this is a problem in practice. For example, the test `fill::horizontal_line` in crate `tiny-skia` (version 0.11.4) is such a test.
This PR changes this, this test no now longer crashes. Instead, it succeeds (after a _long_ time).
Update cargo
12 commits in ba8b39413c74d08494f94a7542fe79aa636e1661..8f40fc59fb0c8df91c97405785197f3c630304ea
2024-08-16 22:48:57 +0000 to 2024-08-21 22:37:06 +0000
- Tests rely on absence of RUST_BACKTRACE (rust-lang/cargo#14441)
- fix: -Cmetadata includes whether extra rustflags is same as host (rust-lang/cargo#14432)
- [mdman] Normalize newlines when rendering options (rust-lang/cargo#14428)
- fix: doctest respects Cargo's color options (rust-lang/cargo#14425)
- Be more permissive while packaging unpublishable crates. (rust-lang/cargo#14408)
- fix: Limiting pre-release match semantics to use only on `OptVersionReq::Req` (rust-lang/cargo#14412)
- test: add a regression test for Issue 14409 (rust-lang/cargo#14430)
- chore: update label trigger for Command-info (rust-lang/cargo#14422)
- doc: add lockfile-path unstable doc section (rust-lang/cargo#14423)
- doc: update lockfile-path tracking issue (rust-lang/cargo#14424)
- fix: remove list owners feature of info subcommand (rust-lang/cargo#14418)
- Lockfile path tests (follow-up) (rust-lang/cargo#14417)
epoll test: avoid some subtly dangling pointers
Turns out `let data = MaybeUninit::<u64>::uninit().as_ptr();` is a dangling pointer, the memory gets freed at the end of that line. For these cases we don't care as we don't actually access the pointer, but let's not do such subtle things.
Rollup of 8 pull requests
Successful merges:
- #128432 (WASI: forbid `unsafe_op_in_unsafe_fn` for `std::{os, sys}`)
- #129373 (Add missing module flags for CFI and KCFI sanitizers)
- #129374 (Use `assert_unsafe_precondition!` in `AsciiChar::digit_unchecked`)
- #129376 (Change `assert_unsafe_precondition` docs to refer to `check_language_ub`)
- #129382 (Add `const_cell_into_inner` to `OnceCell`)
- #129387 (Advise against removing the remaining Python scripts from `tests/run-make`)
- #129388 (Do not rely on names to find lifetimes.)
- #129395 (Pretty-print own args of existential projections (dyn-Trait w/ GAT constraints))
r? `@ghost`
`@rustbot` modify labels: rollup
Pretty-print own args of existential projections (dyn-Trait w/ GAT constraints)
Previously we would just drop them. This bug isn't that significant as it can only be triggered by user code that constrains GATs inside trait object types which is currently gated under the interim feature `generic_associated_types_extended` (whose future is questionable) or on stable if the GATs are 'disabled' in dyn-Trait via `where Self: Sized` (in which case the assoc type bindings get ignored anyway (and trigger the warn-by-default lint `unused_associated_type_bounds`)), so yeah.
Affects diagnostic output and output of `std::any::type_name{_of_val}`.
Do not rely on names to find lifetimes.
For some reason, we were trying to find the lifetime parameter from its name, instead of using the def_id we have.
This PR uses it instead. This changes some ui tests, I think to be more sensible.
Advise against removing the remaining Python scripts from `tests/run-make`
After some recent PRs (e.g. #129185), there are only two Python scripts left in `tests/run-make`.
Having come so far, it's tempting to try to get rid of the remaining ones. But after trying that myself, I've come to the conclusion that it's not worth the extra hassle, especially if it means pulling in an XML-parsing crate just for one test.
This PR therefore leaves behind a few signpost comments to explain why getting rid of these particular scripts has low value.
Add `const_cell_into_inner` to `OnceCell`
`Cell` and `RefCell` have their `into_inner` methods const unstable. `OnceCell` has the same logic, so add it under the same gate.
Tracking issue: https://github.com/rust-lang/rust/issues/78729
Add missing module flags for CFI and KCFI sanitizers
Set the cfi-normalize-integers and kcfi-offset module flags when Control-Flow Integrity sanitizers are used, so functions generated by the LLVM backend use the same CFI/KCFI options as rustc.
cfi-normalize-integers tells LLVM to also use integer normalization for generated functions when -Zsanitizer-cfi-normalize-integers is used.
kcfi-offset specifies the number of prefix nops between the KCFI type hash and the function entry when -Z patchable-function-entry is used. Note that LLVM assumes all indirectly callable functions use the same number of prefix NOPs with -Zsanitizer=kcfi.
Trying to get rid of this Python script looks tempting, because it's currently
the only Python script in the whole `run-make` suite that we actually run.
But getting rid of it would require pulling in a Rust crate to parse XML
instead, and that's probably not worth the extra hassle for a relatively-minor
test.
