Dan Gohman [Sat, 17 Jul 2021 19:33:31 +0000 (12:33 -0700)]
Remove an unnecessary `Mutex` around argument initialization.
In the command-line argument initialization code, remove the Mutex
around the `ARGV` and `ARGC` variables, and simply check whether
ARGV is non-null before dereferencing it. This way, if either of
ARGV or ARGC is not initialized, we'll get an empty argument list.
This allows simple cdylibs to avoid having
`pthread_mutex_lock`/`pthread_mutex_unlock` appear in their symbol
tables if they don't otherwise use threads.
Auto merge of #87124 - Andy-Python-Programmer:code_model_uefi_patch, r=petrochenkov
Use small code model for UEFI targets
* Since the code model only applies to the code and not the data and the code model
only applies to functions you call through using `call`, `jmp` and data with `lea`, etc…
If you are calling functions using the function pointers from the UEFI structures the code
model does not apply in that case. It’s just related to the address space size of your own binary.
Since UEFI (uefi is all relocatable) uses relocatable PEs (relocatable code does not care about the
code model) so, we use the small code model here.
* Since applications don't usually take gigabytes of memory, setting the
target to use the small code model should result in better codegen (comparable
with majority of other targets).
Large code models are also known for generating horrible code, for
example 16 bytes of code to load a single 8-byte value.
Auto merge of #86062 - nagisa:nagisa/what-a-lie, r=estebank
Do not allow JSON targets to set is-builtin: true
Note that this will affect (and make builds fail for) all of the projects out there that have target files invalid in this way. Crater, however, does not really cover these kinds of the codebases, so it is quite difficult to measure the impact. That said, the target files invalid in this way can start causing build failures each time LLVM is upgraded, anyway, so it is probably a good opportunity to disallow this property, entirely.
Another approach considered was to simply not parse this field anymore, which would avoid making the builds explicitly fail, but it wasn't clear to me if `is-builtin` was always set unintentionally… In case this was the case, I'd expect people to file a feature request stating specifically for what purpose they were using `is-builtin`.
* Since the code model only applies to the code and not the data and the code model
only applies to functions you call through using `call`, `jmp` and data with `lea`, etc…
If you are calling functions using the function pointers from the UEFI structures the code
model does not apply in that case. It’s just related to the address space size of your own binary.
Since UEFI (uefi is all relocatable) uses relocatable PEs (relocatable code does not care about the
code model) so, we use the small code model here.
* Since applications don't usually take gigabytes of memory, setting the
target to use the small code model should result in better codegen (comparable
with majority of other targets).
Large code models are also known for generating horrible code, for
example 16 bytes of code to load a single 8-byte value.
* Use the LLVM default code model for the architecture for the
x86_64-unknown-uefi targets. For reference small is the default
code model on x86 in LLVM: <https://github.com/llvm/llvm-project/blob/7de2173c2a4c45711831cfee3ccf53690c76ff07/llvm/lib/Target/X86/X86TargetMachine.cpp#L204>
* Remove the comments too as they are not UEFI-specific and applies
to pretty much any target. I added them before as I was explicitily
setting the code model to small.
Auto merge of #83898 - Aaron1011:feature/hir-wf, r=estebank
Add initial implementation of HIR-based WF checking for diagnostics
During well-formed checking, we walk through all types 'nested' in
generic arguments. For example, WF-checking `Option<MyStruct<u8>>`
will cause us to check `MyStruct<u8>` and `u8`. However, this is done
on a `rustc_middle::ty::Ty`, which has no span information. As a result,
any errors that occur will have a very general span (e.g. the
definintion of an associated item).
This becomes a problem when macros are involved. In general, an
associated type like `type MyType = Option<MyStruct<u8>>;` may
have completely different spans for each nested type in the HIR. Using
the span of the entire associated item might end up pointing to a macro
invocation, even though a user-provided span is available in one of the
nested types.
This PR adds a framework for HIR-based well formed checking. This check
is only run during error reporting, and is used to obtain a more precise
span for an existing error. This is accomplished by individually
checking each 'nested' type in the HIR for the type, allowing us to
find the most-specific type (and span) that produces a given error.
The majority of the changes are to the error-reporting code. However,
some of the general trait code is modified to pass through more
information.
Since this has no soundness implications, I've implemented a minimal
version to begin with, which can be extended over time. In particular,
this only works for HIR items with a corresponding `DefId` (e.g. it will
not work for WF-checking performed within function bodies).
Aaron Hill [Sun, 4 Apr 2021 20:55:39 +0000 (16:55 -0400)]
Add initial implementation of HIR-based WF checking for diagnostics
During well-formed checking, we walk through all types 'nested' in
generic arguments. For example, WF-checking `Option<MyStruct<u8>>`
will cause us to check `MyStruct<u8>` and `u8`. However, this is done
on a `rustc_middle::ty::Ty`, which has no span information. As a result,
any errors that occur will have a very general span (e.g. the
definintion of an associated item).
This becomes a problem when macros are involved. In general, an
associated type like `type MyType = Option<MyStruct<u8>>;` may
have completely different spans for each nested type in the HIR. Using
the span of the entire associated item might end up pointing to a macro
invocation, even though a user-provided span is available in one of the
nested types.
This PR adds a framework for HIR-based well formed checking. This check
is only run during error reporting, and is used to obtain a more precise
span for an existing error. This is accomplished by individually
checking each 'nested' type in the HIR for the type, allowing us to
find the most-specific type (and span) that produces a given error.
The majority of the changes are to the error-reporting code. However,
some of the general trait code is modified to pass through more
information.
Since this has no soundness implications, I've implemented a minimal
version to begin with, which can be extended over time. In particular,
this only works for HIR items with a corresponding `DefId` (e.g. it will
not work for WF-checking performed within function bodies).
Auto merge of #87201 - GuillaumeGomez:rollup-4loi2q9, r=GuillaumeGomez
Rollup of 7 pull requests
Successful merges:
- #87107 (Loop over all opaque types instead of looking at just the first one with the same DefId)
- #87158 (Suggest full enum variant for local modules)
- #87174 (Stabilize `[T; N]::map()`)
- #87179 (Mark `const_trait_impl` as active)
- #87180 (feat(rustdoc): open sidebar menu when links inside it are focused)
- #87188 (Add GUI test for auto-hide-trait-implementations setting)
- #87200 (TAIT: Infer all inference variables in opaque type substitutions via InferCx)
Rollup merge of #87200 - oli-obk:fixup_fixup_opaque_types, r=nikomatsakis
TAIT: Infer all inference variables in opaque type substitutions via InferCx
The previous algorithm was correct for the example given in its
documentation, but when the TAIT was declared as a free item
instead of an associated item, the generic parameters were the
wrong ones.
Rollup merge of #87179 - fee1-dead:active-const-impl, r=oli-obk
Mark `const_trait_impl` as active
See [this zulip thread](https://rust-lang.zulipchat.com/#narrow/stream/146212-t-compiler.2Fconst-eval/topic/implementation.20path.20for.20const.20trait.20impls).
Rollup merge of #87174 - inquisitivecrystal:array-map, r=kennytm
Stabilize `[T; N]::map()`
This stabilizes the `[T; N]::map()` function, gated by the `array_map` feature. The FCP has [already completed.](https://github.com/rust-lang/rust/issues/75243#issuecomment-878448138)
Oli Scherer [Fri, 16 Jul 2021 17:34:23 +0000 (17:34 +0000)]
Infer all inference variables via InferCx
The previous algorithm was correct for the example given in its
documentation, but when the TAIT was declared as a free item
instead of an associated item, the generic parameters were the
wrong ones.
Auto merge of #87140 - camsteffen:pat-slice-refs, r=oli-obk
Remove refs from Pat slices
Changes `PatKind::Or(&'hir [&'hir Pat<'hir>])` to `PatKind::Or(&'hir [Pat<'hir>])` and others. This is more consistent with `ExprKind`, saves a little memory, and is a little easier to use.
Auto merge of #87182 - GuillaumeGomez:rollup-whwohua, r=GuillaumeGomez
Rollup of 7 pull requests
Successful merges:
- #86983 (Add or improve natvis definitions for common standard library types)
- #87069 (ExprUseVisitor: Treat ByValue use of Copy types as ImmBorrow)
- #87138 (Correct invariant documentation for `steps_between`)
- #87145 (Make --cap-lints and related options leave crate hash alone)
- #87161 (RFC2229: Use the correct place type)
- #87162 (Fix type decl layout "overflow")
- #87167 (Fix sidebar display on small devices)
Rollup merge of #87167 - GuillaumeGomez:sidebar-display-mobile, r=notriddle
Fix sidebar display on small devices
Part of #87059.
Instead of hiding the sidebar on small devices, we instead move it out of the viewport so that it remains "visible" to our text only users.
Could you confirm it works for you `@ahicks92` and `@DataTriny` please? You can give it a try at [this URL](https://guillaume-gomez.fr/rustdoc-test/test_docs/index.html).
The ICE occurred because instead of looking at the type of the place after all the projections are applied, we instead looked at the `base_ty` of the Place to decide whether a discriminant should be read of not. This lead to two issues:
1. the kind of the type is not necessarily `Adt` since we only look at the `base_ty`, it could be instead `Ref` for example
2. if the kind of the type is `Adt` you could still be looking at the wrong variant to make a decision on whether the discriminant should be read or not
Rollup merge of #87138 - dhwthompson:fix-range-invariant, r=JohnTitor
Correct invariant documentation for `steps_between`
Given that the previous example involves stepping forward from A to B, the equivalent example on this line would make most sense as stepping backward from B to A.
I should probably add a caveat here that I’m fairly new to Rust, and this is my first contribution to this repo, so it’s very possible that I’ve misunderstood how this is supposed to work (either on a technical level or a social one). If this is the case, please do let me know.
Rollup merge of #86983 - wesleywiser:natvis_std_types, r=michaelwoerister
Add or improve natvis definitions for common standard library types
Natvis definitions are used by Windows debuggers to provide a better experience when inspecting a value for types with natvis definitions. Many of our standard library types and intrinsic Rust types like slices and `str` already have natvis definitions.
This PR adds natvis definitions for missing types (like all of the `Atomic*` types) and improves some of the existing ones (such as showing the ref count on `Arc<T>` and `Rc<T>` and showing the borrow state of `RefCell<T>`). I've also added cdb tests to cover these definitions and updated existing tests with the new visualizations.
With this PR, the following types now visualize in a much more intuitive way:
### Type: `NonZero{I,U}{8,16,32,64,128,size}`, `Atomic{I,U}{8,16,32,64,size}`, `AtomicBool` and `Wrapping<T>`
Auto merge of #86662 - mockersf:fix-86620-link-unknown-location, r=jyn514
fix dead link for method in trait of blanket impl from third party crate
fix #86620
* changes `href` method to raise the actual error it had instead of an `Option`
* set the href link correctly in case of an error
I did not manage to make a small reproducer, I think it happens in a situation where
* crate A expose a trait with a blanket impl
* crate B use the trait from crate A
* crate C use types from crate B
* building docs for crate C without dependencies
Auto merge of #87177 - ehuss:update-cargo, r=ehuss
Update cargo
6 commits in 66a6737a0c9f3a974af2dd032a65d3e409c77aac..27277d966b3cfa454d6dea7f724cb961c036251c
2021-07-14 20:54:28 +0000 to 2021-07-16 00:50:39 +0000
- Flag another curl error as possibly spurious (rust-lang/cargo#9695)
- Add `d` as an alias for `doc` (rust-lang/cargo#9680)
- `cargo fix --edition`: extend warning when on latest edition (rust-lang/cargo#9694)
- Update env_logger requirement from 0.8.1 to 0.9.0 (rust-lang/cargo#9688)
- Document cargo limitation w/ workspaces & configs (rust-lang/cargo#9674)
- Change some warnings to errors (rust-lang/cargo#9689)
Auto merge of #86993 - jackh726:project-gat-binders, r=nikomatsakis
Replace associated item bound vars with placeholders when projecting
Fixes #76407
Fixes #76826
Similar, but more limited, to #85499. This allows us to handle things like `for<'a> <T as Trait>::Assoc<'a>` but not `for<'a> <T as Trait<'a>>::Assoc`, unblocking GATs.
Auto merge of #87156 - JohnTitor:rollup-osuhe53, r=JohnTitor
Rollup of 8 pull requests
Successful merges:
- #85579 (Added Arc::try_pin)
- #86478 (Add -Zfuture-incompat-test to assist with testing future-incompat reports.)
- #86947 (Move assert_matches to an inner module)
- #87081 (Add tracking issue number to `wasi_ext`)
- #87127 (Add safety comments in private core::slice::rotate::ptr_rotate function)
- #87134 (Make SelfInTyParamDefault wording not be specific to type defaults)
- #87147 (Update cargo)
- #87154 (Fix misuse of rev attribute on <a> tag)
Rollup merge of #87154 - GuillaumeGomez:rev-attr-a, r=JohnTitor
Fix misuse of rev attribute on <a> tag
The `rev` attribute is supposed to talk about "ownership" as far as I could found out. This attribute seems not very well defined in the HTML spec and its usage in rustdoc is suboptimal.
It was found out in https://github.com/rust-lang/rust/pull/87149.
Rollup merge of #87147 - ehuss:update-cargo, r=ehuss
Update cargo
13 commits in 3ebb5f15a940810f250b68821149387af583a79e..66a6737a0c9f3a974af2dd032a65d3e409c77aac
2021-07-02 20:35:38 +0000 to 2021-07-14 20:54:28 +0000
- Add format option to `cargo tree` to print the lib_name (rust-lang/cargo#9663)
- Prefer patched versions of dependencies (rust-lang/cargo#9639)
- When a dependency does not have a version, git or path, fails directly (rust-lang/cargo#9686)
- Spot the crate typo easily (rust-lang/cargo#9665)
- remove unnecessary 'collect' (rust-lang/cargo#9616)
- Make it easier to run testsuite with a custom toolchain. (rust-lang/cargo#9679)
- Serialize `cargo fix` (rust-lang/cargo#9677)
- Don't recommend filing issues on rust-lang/cargo for Cargo.toml errors. (rust-lang/cargo#9658)
- Update nightly failure notification. (rust-lang/cargo#9657)
- Update Windows env uppercase key check. (rust-lang/cargo#9654)
- Unignore fix_edition_2021. (rust-lang/cargo#9662)
- Warning when using features in patch (rust-lang/cargo#9666)
- Unify cargo and rustc's error reporting (rust-lang/cargo#9655)
Rollup merge of #86478 - ehuss:future-incompat-test, r=oli-obk
Add -Zfuture-incompat-test to assist with testing future-incompat reports.
This adds a `-Zfuture-incompat-test` cli flag to assist with testing future-incompatible reports. This flag causes all lints to be treated as a future-incompatible lint, and will emit a report for them. This is being added so that Cargo's testsuite can reliably test the reporting infrastructure. Right now, Cargo relies on using array_into_iter as a test subject. Since the breaking "future incompatible" lints are never intended to last forever, this means Cargo's testsuite would always need to keep changing to choose different lints (for example, #86330 proposed dropping that moniker for array_into_iter). With this flag, Cargo's tests can trigger any lint and check for the report.
Auto merge of #86876 - jyn514:56935-target-crate-num, r=petrochenkov
Reuse CrateNum for proc-macro crates even when cross-compiling
Proc-macros are always compiled for the host, so this should be the same
in every way as recompiling the crate.
I am not sure why the previous code special-cased the target, since the
compiler properly gives an error when trying to load a crate for a
different host:
```
error[E0461]: couldn't find crate `dependency` with expected target triple x86_64-unknown-linux-gnu
--> /home/joshua/rustc4/src/test/ui/cfg-dependent.rs:8:2
|
LL | dependency::is_64();
| ^^^^^^^^^^
|
= note: the following crate versions were found:
crate `dependency`, target triple i686-unknown-linux-gnu: /home/joshua/rustc4/build/x86_64-unknown-linux-gnu/test/ui/cfg-dependent/auxiliary/libdependency.so
```
I think another possible fix is to remove the check altogether. But I'm
not sure, and this fix works, so I'm not making the larger change here.
Joshua Nelson [Sun, 4 Jul 2021 03:35:24 +0000 (23:35 -0400)]
Reuse CrateNum for proc-macro crates even when cross-compiling
Proc-macros are always compiled for the host, so this should be the same
in every way as recompiling the crate.
I am not sure why the previous code special-cased the target, since the
compiler properly gives an error when trying to load a crate for a
different host:
```
error[E0461]: couldn't find crate `dependency` with expected target triple x86_64-unknown-linux-gnu
--> /home/joshua/rustc4/src/test/ui/cfg-dependent.rs:8:2
|
LL | dependency::is_64();
| ^^^^^^^^^^
|
= note: the following crate versions were found:
crate `dependency`, target triple i686-unknown-linux-gnu: /home/joshua/rustc4/build/x86_64-unknown-linux-gnu/test/ui/cfg-dependent/auxiliary/libdependency.so
```
I think another possible fix is to remove the check altogether. But I'm
not sure, and this fix works, so I'm not making the larger change here.
Auto merge of #87137 - richkadel:compiler-builtins-0.1.47, r=tmandry
Update compiler-builtins to 0.1.47
Bumped to `0.1.47` to resolve missing symbols on `aarch` when linking
`cargo`. This was due to a recent change in a `cargo` dependency on
`curl` (upstream C library added code that uses the uncommon `long
double` type).
Auto merge of #86765 - cuviper:fuse-less-specialized, r=joshtriplett
Make the specialized Fuse still deal with None
Fixes #85863 by removing the assumption that we'll never see a cleared iterator in the `I: FusedIterator` specialization. Now all `Fuse` methods check for the possibility that `self.iter` is `None`, and the specialization only avoids _setting_ that to `None` in `&mut self` methods.
David Thompson [Wed, 14 Jul 2021 20:48:18 +0000 (13:48 -0700)]
Correct invariant documentation for `steps_between`
Given that the previous example involves stepping forward from A to B,
the equivalent example on this line would make most sense as stepping
backward from B to A.
Bumped to `0.1.47` to resolve missing symbols on `aarch` when linking
`cargo`. This was due to a recent change in a `cargo` dependency on
`curl` (upstream C library added code that uses the uncommon `long
double` type).
Auto merge of #7462 - xFrednet:7369-branches-sharing-code-else-expr-fp, r=camsteffen
FP fix and documentation for `branches_sharing_code` lint
Closes rust-lang/rust-clippy#7369
Related rust-lang/rust-clippy#7452 I'm still thinking about the best way to fix this. I could simply add another visitor to ensure that the moved expressions don't modify values being used in the condition, but I'm not totally happy with this due to the complexity. I therefore only documented it for now
changelog: [`branches_sharing_code`] fixed false positive where block expressions would sometimes be ignored.
Rollup merge of #87056 - GuillaumeGomez:fix-codeblocks-overflow, r=notriddle
Fix codeblocks overflow
Fixes #87043.
Instead of completely relying on `pulldown-cmark` (and its potential changes), I decided to move the generation of codeblocks HTML directly in rustdoc so we can unify the DOM and the CSS classes.
Rollup merge of #87027 - petrochenkov:builderhelp, r=oli-obk
expand: Support helper attributes for built-in derive macros
This is needed for https://github.com/rust-lang/rust/pull/86735 (derive macro `Default` should have a helper attribute `default`).
With this PR we can specify helper attributes for built-in derives using syntax `#[rustc_builtin_macro(MacroName, attributes(attr1, attr2, ...))]` which mirrors equivalent syntax for proc macros `#[proc_macro_derive(MacroName, attributes(attr1, attr2, ...))]`.
Otherwise expansion infra was already ready for this.
The attribute parsing code is shared between proc macro derives and built-in macros (`fn parse_macro_name_and_helper_attrs`).
Auto merge of #87082 - michaelwoerister:const-in-debuginfo-type-names-fix, r=oli-obk,wesleywiser
Handle non-integer const generic parameters in debuginfo type names.
This PR fixes an ICE introduced by https://github.com/rust-lang/rust/pull/85269 which started emitting const generic arguments for debuginfo names but did not cover the case where such an argument could not be evaluated to a flat string of bits.
The fix implemented in this PR is very basic: If `try_eval_bits()` fails for the constant in question, we fall back to generating a stable hash of the constant and emit that instead. This way we get a (virtually) unique name and side step the problem of generating a string representation of a potentially complex value.
The downside is that the generated name will be rather opaque. E.g. the regression test adds a function `const_generic_fn_non_int<()>` which is then rendered as `const_generic_fn_non_int<{CONST#fe3cfa0214ac55c7}>`. I think it's an open question how to deal with this more gracefully.
I'd be interested in ideas on how to do this better.
r? `@wesleywiser`
cc `@dpaoliello` (do you see any problems with this approach?)
cc `@Mark-Simulacrum` & `@nagisa` (who I've seen comment on debuginfo issues recently -- anyone else?)