FD: remove big surrounding RefCell, simplify socketpair
A while ago, I added the big implicit RefCell for all file descriptions since it avoided interior mutability in `eventfd`. However, this requires us to hold the RefCell "lock" around the entire invocation of the `read`/`write` methods on an FD, which is not great. For instance, if an FD wants to update epoll notifications from inside its `read`/`write`, it is very crucial that the notification check does not end up accessing the FD itself. Such cycles, however, occur naturally:
- eventfd wants to update notifications for itself
- socketfd wants to update notifications on its "peer", which will in turn check *its* peer to see whether that buffer is empty -- and my peer's peer is myself.
This then also lets us simplify socketpair, which currently holds a weak reference to its peer *and* a weak reference to the peer's buffer -- that was previously needed precisely to avoid this issue.
Make unused states of Reserved unrepresentable
In the [previous TB update](https://github.com/rust-lang/miri/pull/3742) we discovered that the existence of `Reserved + !ty_is_freeze + protected` is undesirable.
This has the side effect of making `Reserved { conflicted: true, ty_is_freeze: false }` unreachable.
As such it is desirable that this state would also be unrepresentable.
This PR eliminates the unused configuration by changing
```rs
enum PermissionPriv {
Reserved { ty_is_freeze: bool, conflicted: bool },
...
}
```
into
```rs
enum PermissionPriv {
ReservedFrz { conflicted: bool },
ReservedIM,
...
}
```
but this is not the only solution and `Reserved(Activable | Conflicted | InteriorMut)` could be discussed.
In addition to making the unreachable state not representable anymore, this change has the nice side effect of enabling `foreign_read` to no longer depend explicitly on the `protected` flag.
Currently waiting for
- `@JoJoDeveloping` to confirm that this is the same representation of `Reserved` as what is being implemented in simuliris,
- `@RalfJung` to approve that this does not introduce too much overhead in the trusted codebase.
detect incompatible CI rustc options more precisely
Previously, the logic here was simply checking whether the option was set in `config.toml`. This approach was not manageable in our CI runners as we set so many options in config.toml. In reality, those values are not incompatible since they are usually the same value used to generate the CI rustc. Now, the new logic compares the configuration values with the values used to generate the CI rustc, so we get more precise results and make the process more manageable.
r? Kobzol
Blocker for https://github.com/rust-lang/rust/pull/122709
Only try to modify file times of a writable file on Windows
Introduces a `set_file_times` helper which opens a given path as a file in r+w mode on Windows and then sets file times. Previously the file was open as read-only for Windows which caused permission errors locally.
This should hopefully make setting file times less error prone, since trying to set file times on read-only file on Windows also happened in #127850.
try-job: dist-loongarch64-musl
try-job: x86_64-msvc
Enable debuginfo tests that have been "temporarily disabled" for the past 6 years
The PR history is a bit of a mess because I had to test this a lot with try-jobs, so I'll try to summarize the non-obvious changes here.
A number of tests now have `min-lldb-version: 1800`. Those tests should have gotten an lldb version jump either in https://github.com/rust-lang/rust/pull/124781 or long ago. Note that all such tests with that lldb version requirement do not run in Apple CI.
`tests/debuginfo/drop-locations.rs` is staying disabled for now because gdb doesn't know to stop on the drop calls produced by a `}`: https://github.com/rust-lang/rust/issues/128971
`tests/debuginfo/function-arg-initialization.rs` now has `-Zmir-enable-passes=-SingleUseConsts`; without that we initialize the const before the function prelude: https://github.com/rust-lang/rust/issues/128945
`tests/debuginfo/by-value-non-immediate-argument.rs` fails because we don't generate a function prelude for unused non-immediate arguments, even with all optimizations disabled, and this seems to confuse debuggers on aarch64: https://github.com/rust-lang/rust/issues/128973
`tests/debuginfo/pretty-std.rs` is staying disabled on windows-gnu because our test harness doesn't know how to load our pretty-printers on that target: https://github.com/rust-lang/rust/issues/128981
`tests/debuginfo/method-on-enum.rs` and `tests/debuginfo/option-like-enum.rs` encounter some kind of gdb bug on i686-pc-windows-gnu. I don't know enough about that situation to write a good issue.
I plan on doing more work on this test suite. There's clearly a lot more basic cleanup work to do here.
Rollup of 9 pull requests
Successful merges:
- #128064 (Improve docs for Waker::noop and LocalWaker::noop)
- #128922 (rust-analyzer: use in-tree `pattern_analysis` crate)
- #128965 (Remove `print::Pat` from the printing of `WitnessPat`)
- #129018 (Migrate `rlib-format-packed-bundled-libs` and `native-link-modifier-bundle` `run-make` tests to rmake)
- #129037 (Port `run-make/libtest-json` and `run-make/libtest-junit` to rmake)
- #129078 (`ParamEnvAnd::fully_perform`: we have an `ocx`, use it)
- #129110 (Add a comment explaining the return type of `Ty::kind`.)
- #129111 (Port the `sysroot-crates-are-unstable` Python script to rmake)
- #129135 (crashes: more tests)
r? `@ghost`
`@rustbot` modify labels: rollup
Port the `sysroot-crates-are-unstable` Python script to rmake
New version of #126231, and a follow-up to #129071.
One major difference is that the new version no longer tries to report *all* accidentally-stable crates, because the `run_make_support` helpers tend to halt the test as soon as something goes wrong. That's unfortunate, but I think it won't matter much in practice, and preserving the old behaviour doesn't seem worth the extra effort.
---
Part of #110479 (Python purge), with this being one of the non-trivial Python scripts that actually seems feasible and worthwhile to remove.
This is *not* part of #121876 (Makefile purge), because the underlying test is already using rmake; this PR just modifies the existing rmake recipe to do all the work itself instead of delegating to Python. So there's no particular urgency here.
r? ````@jieyouxu````
try-job: aarch64-gnu
try-job: aarch64-apple
try-job: test-various
try-job: armhf-gnu
try-job: x86_64-msvc
try-job: i686-mingw
Port `run-make/libtest-json` and `run-make/libtest-junit` to rmake
Unlike #126773, this is just a straightforward port to `rmake`, without attempting to switch to compiletest or get rid of the (trivial) Python scripts.
Part of #121876.
r? ````@jieyouxu````
try-job: x86_64-msvc
try-job: i686-mingw
try-job: test-various
try-job: aarch64-gnu
try-job: aarch64-apple
Remove `print::Pat` from the printing of `WitnessPat`
After the preliminary work done in #128536, we can now get rid of `print::Pat` entirely.
- First, we introduce a variant `PatKind::Print(String)`.
- Then we incrementally remove each other variant of `PatKind`, by having the relevant code produce `PatKind::Print` instead.
- Once `PatKind::Print` is the only remaining variant, it becomes easy to remove `print::Pat` and replace it with `String`.
There is more cleanup that I have in mind, but this seemed like a natural stopping point for one PR.
r? ```@Nadrieril```
Improve docs for Waker::noop and LocalWaker::noop
* Add a warning about a likely misuse. (See my commit message for longer rationale.)
* Apply some probably-accidentally-omitted changes to `LocalWaker`'s docs
* Add a comment about the clone-and-hack of the docs
I have used [semantic linefeeds](https://rhodesmill.org/brandon/2012/one-sentence-per-line/) for the docs formatting.
Fix problems with assoc expr token collection
There are several cases involving assoc exprs and attributes where the current code does the wrong thing. This PR adds some tests that demonstrate the problems and then fixes them.
r? `@petrochenkov`
This commit does the following.
- Renames `collect_tokens_trailing_token` as `collect_tokens`, because
(a) it's annoying long, and (b) the `_trailing_token` bit is less
accurate now that its types have changed.
- In `collect_tokens`, adds a `Option<CollectPos>` argument and a
`UsePreAttrPos` in the return type of `f`. These are used in
`parse_expr_force_collect` (for vanilla expressions) and in
`parse_stmt_without_recovery` (for two different cases of expression
statements). Together these ensure are enough to fix all the problems
with token collection and assoc expressions. The changes to the
`stringify.rs` test demonstrate some of these.
- Adds a new test. The code in this test was causing an assertion
failure prior to this commit, due to an invalid `NodeRange`.
The extra complexity is annoying, but necessary to fix the existing
problems.
This pre-existing type is suitable for use with the return value of the
`f` parameter in `collect_tokens_trailing_token`. The more descriptive
name will be useful because the next commit will add another boolean
value to the return value of `f`.
Coalesce `dep-info`, `dep-info-spaces` and `dep-info-doesnt-run-much` `run-make` tests into `dep-info` rmake.rs
Part of #121876 and the associated [Google Summer of Code project](https://blog.rust-lang.org/2024/05/01/gsoc-2024-selected-projects.html).
This is one of the most ancient tests in the `run-make` directory and its Makefile does some unexpected things, like creating and deleting a `done` directory over and over, sleeping at certain times (this is the [commit](0d9fd8e2a1) that added the `sleep`).
I tried to preserve the intent of the test, which is smoke-testing that `dep-info` works.
try-job: x86_64-msvc
try-job: i686-mingw
try-job: aarch64-gnu
try-job: aarch64-apple
try-job: test-various
try-job: armhf-gnu
try-job: dist-various-1
Rollup of 8 pull requests
Successful merges:
- #128348 (Unconditionally allow shadow call-stack sanitizer for AArch64)
- #129065 (Use `impl PartialEq<TokenKind> for Token` more.)
- #129072 (Infer async closure args from `Fn` bound even if there is no corresponding `Future` bound on return)
- #129096 (Print more verbose error for commands that capture output)
- #129101 (Fix projections when parent capture is by-ref but child capture is by-value in the `ByMoveBody` pass)
- #129106 (Remove redundant type ops: `Eq`/`Subtype`)
- #129122 (Remove duplicated `Rustdoc::output` method from `run-make-support` lib)
- #129124 (rustdoc-json: Use FxHashMap from rustdoc_json_types)
r? `@ghost`
`@rustbot` modify labels: rollup
rustdoc-json: Use FxHashMap from rustdoc_json_types
Alternative to #110051 and #127456.
This lets us avoid rehashing the json index after building it by using the same hashmap type in construction and in rustdoc_json_types. The above PR's tried to do this by having rustdoc_json_types get the same hashmap that rustdoc has via rustc_data_structures. However, this can be made simpler if we change rustdoc instead.
For the rustdoc-type republish on crates.io, It will filter out the `pub use`, and not change source at all. https://github.com/aDotInTheVoid/rustdoc-types/pull/30. That code [already replaces the hashmap](8d6528669e/update.sh (L11)) to use the one in `std::collections` (instead of `FxHashMap`)
try-job: dist-arm-linux
r? ``@GuillaumeGomez``
Remove duplicated `Rustdoc::output` method from `run-make-support` lib
I discovered recently that `--output` is deprecated in rustdoc and that `--out-dir` is doing the exact same thing. To keep things along with the current rustdoc status, I removed the `Rustdoc::output` method.
cc `@jieyouxu`
r? `@Kobzol`
Fix projections when parent capture is by-ref but child capture is by-value in the `ByMoveBody` pass
This fixes a somewhat strange bug where we build the incorrect MIR in #129074. This one is weird, but I don't expect it to actually matter in practice since it almost certainly results in a move error in borrowck. However, let's not ICE.
Given the code:
```
#![feature(async_closure)]
// NOT copy.
struct Ty;
fn hello(x: &Ty) {
let c = async || {
*x;
//~^ ERROR cannot move out of `*x` which is behind a shared reference
};
}
fn main() {}
```
The parent coroutine-closure captures `x: &Ty` by-ref, resulting in an upvar of `&&Ty`. The child coroutine captures `x` by-value, resulting in an upvar of `&Ty`. When constructing the by-move body for the coroutine-closure, we weren't applying an additional deref projection to convert the parent capture into the child capture, resulting in an type error in assignment, which is a validation ICE.
As I said above, this only occurs (AFAICT) in code that eventually results in an error, because it is only triggered by HIR that attempts to move a non-copy value out of a ref. This doesn't occur if `Ty` is `Copy`, since we'd instead capture `x` by-ref in the child coroutine.
Fixes#129074
Print more verbose error for commands that capture output
https://github.com/rust-lang/rust/pull/128874 made bootstrap command errors less verbose without `-v`. However, in some cases it's too extreme. If a command fails, it now outputs just `Command has failed. Rerun with -v to see more details.`, without providing any context.
I think that I found a reasonable heuristic to figure out when we should print a more verbose error. When the command doesn't capture output, its stdout/stderr is printed, therefore the user sees context about the error. However, when the command captures its output, the user won't see any error message in the output, which is not great. So only in that case, bootstrap now prints a slightly more verbose output (and also prints the captured output).
r? `@onur-ozkan`
Infer async closure args from `Fn` bound even if there is no corresponding `Future` bound on return
In #127482, I implemented the functionality to infer an async closure signature when passed into a function that has `Fn` + `Future` where clauses that look like:
```
fn whatever(callback: F)
where
F: Fn(Arg) -> Fut,
Fut: Future<Output = Out>,
```
However, #127781 demonstrates that this is still incomplete to address the cases users care about. So let's not bail when we fail to find a `Future` bound, and try our best to just use the args from the `Fn` bound if we find it. This is *fine* since most users of closures only really care about the *argument* types for inference guidance, since we require the receiver of a `.` method call to be known in order to probe methods.
When I experimented with programmatically rewriting `|| async {}` to `async || {}` in #127827, this also seems to have fixed ~5000 regressions (probably all coming from usages `TryFuture`/`TryStream` from futures-rs): the [before](https://github.com/rust-lang/rust/pull/127827#issuecomment-2254061733) and [after](https://github.com/rust-lang/rust/pull/127827#issuecomment-2255470176) crater runs.
Fixes#127781.
