bors [Mon, 29 Mar 2021 11:20:25 +0000 (11:20 +0000)]
Auto merge of #83637 - bjorn3:sync_cg_clif-2021-03-29, r=bjorn3
Sync rustc_codegen_cranelift
The main highlight of this sync is support for cross-compiling to Windows using MinGW. Native compilation with MinGW would also work I think, but using the MSVC toolchain is not yet supported as PE TLS is not yet implemented. Another nice improvement is that crate metadata is now loaded using mmap instead of by reading files. This improves compilation time a bit.
bors [Mon, 29 Mar 2021 08:27:59 +0000 (08:27 +0000)]
Auto merge of #83565 - RalfJung:miri, r=oli-obk
update Miri, and also run test suite with mir-opt-level=4
In the Miri repo, we run the Miri test suite once with default flags and once with `-O -Zmir-opt-level=4`. This helps identify and document situations where MIR optimizations mask UB -- it is okay for that to happen, but it might be god to look into it when it does happen. Recently these tests failed fairly frequently as new MIR optimizations were added, and since we only run them on the Miri side, it is not even clear which rustc PR introduced the change. So I propose we also run these tests in the rustc repo, such that toolstate tracking will tell us the exact PR (or at least the rollup) that caused the change.
bors [Mon, 29 Mar 2021 00:17:23 +0000 (00:17 +0000)]
Auto merge of #83605 - RalfJung:unaligned, r=petrochenkov
unaligned_references: align(N) fields in packed(N) structs are fine
This removes some false positives from the unaligned_references lint: in a `repr(packed(2))` struct, fields of alignment 2 (and less) are guaranteed to be properly aligned, so we do not have to consider them "disaligned".
Tom Eccles [Tue, 16 Mar 2021 18:22:21 +0000 (18:22 +0000)]
test: run-make: skip tests on unsupported platforms
The tests issue-36710 and incr-prev-body-beyond-eof were changed in a
previous commit so that the correct target was passed to rustc
(previously rustc was building for the host not for the specific
target).
Since that change it turns out that these platforms never worked (they
only appeared to work because rustc was actually building for the host
architecture).
The wasm architectures fall over trying to build the C++ file in
issue-36710. They look for clang (which isn't installed in the
test-various docker container). If clang is installed, they can't find
a wasm c++ standard library to link to.
nvtptx64-nvidia-cuda fails in rustc saying it can't find std. The rust
platforms support page says that std is supported on cuda so this is
surprising.
dist-i586-gnu-i586-i686-musl can't find the C++ compiler. There is only
a musl-gcc and no musl-g++ in /musl-i586/bin/. The Docker image probably
needs tweaking.
Tom Eccles [Sun, 21 Feb 2021 11:19:50 +0000 (11:19 +0000)]
bootstrap: don't run linkcheck when crosscompiling
When we cross compile, some things (and their documentation) are built
for the host (e.g. rustc), while others (and their documentation) are built
for the target. This generated documentation will have broken links
between documentation for different platforms e.g. between rustc and
cargo.
Tom Eccles [Sat, 9 Jan 2021 22:54:21 +0000 (22:54 +0000)]
run-make: Specify --target to rustc
Resolves #78911
The target's linker was used but rustc wasn't told to build for that
target (instead defaulting to the host). This led to the host instead of
the target getting tested and to the linker getting inappropriate
arguments.
bors [Sun, 28 Mar 2021 14:16:03 +0000 (14:16 +0000)]
Auto merge of #83582 - jyn514:might-not, r=joshtriplett
may not -> might not
may not -> might not
"may not" has two possible meanings:
1. A command: "You may not stay up past your bedtime."
2. A fact that's only sometimes true: "Some cities may not have bike lanes."
In some cases, the meaning is ambiguous: "Some cars may not have snow
tires." (do the cars *happen* to not have snow tires, or is it
physically impossible for them to have snow tires?)
This changes places where the standard library uses the "description of
fact" meaning to say "might not" instead.
This is just `std::vec` for now - if you think this is a good idea I can
convert the rest of the standard library.
bors [Sun, 28 Mar 2021 08:53:51 +0000 (08:53 +0000)]
Auto merge of #83593 - petrochenkov:nounwrap, r=nagisa
rustc_target: Avoid unwraps when adding linker flags
These `unwrap`s assume that some linker flags were already added by `*_base::opts()` methods, but that's doesn't necessarily remain the case when we are reducing the number of flags hardcoded in targets, as https://github.com/rust-lang/rust/pull/83587 shows.
bors [Sun, 28 Mar 2021 03:51:22 +0000 (03:51 +0000)]
Auto merge of #81354 - SkiFire13:binary-search-assume, r=nagisa
Instruct LLVM that binary_search returns a valid index
This allows removing bound checks when the return value of `binary_search` is used to index into the slice it was call on. I also added a codegen test for this, not sure if it's the right thing to do (I didn't find anything on the dev guide), but it felt so.
Joshua Nelson [Sun, 28 Mar 2021 02:16:17 +0000 (22:16 -0400)]
Address more review comments
- Add back various diagnostic methods on `Session`.
It seems unfortunate to duplicate these in so many places, but in the
meantime, making the API inconsistent between `Session` and `Diagnostic`
also seems unfortunate.
- Add back TyCtxtAt methods
These will hopefully be used in the near future.
- Add back `with_const`, it would need to be added soon after anyway.
- Add back `split()` and `get_mut()`, they're useful.
Joshua Nelson [Tue, 16 Mar 2021 16:57:31 +0000 (12:57 -0400)]
Address review comments
- Add back `HirIdVec`, with a comment that it will soon be used.
- Add back `*_region` functions, with a comment they may soon be used.
- Remove `-Z borrowck_stats` completely. It didn't do anything.
- Remove `make_nop` completely.
- Add back `current_loc`, which is used by an out-of-tree tool.
- Fix style nits
- Remove `AtomicCell` with `cfg(parallel_compiler)` for consistency.
Joshua Nelson [Tue, 16 Mar 2021 05:50:34 +0000 (01:50 -0400)]
Remove (lots of) dead code
Found with https://github.com/est31/warnalyzer.
Dubious changes:
- Is anyone else using rustc_apfloat? I feel weird completely deleting
x87 support.
- Maybe some of the dead code in rustc_data_structures, in case someone
wants to use it in the future?
- Don't change rustc_serialize
I plan to scrap most of the json module in the near future (see
https://github.com/rust-lang/compiler-team/issues/418) and fixing the
tests needed more work than I expected.
TODO: check if any of the comments on the deleted code should be kept.
bors [Sat, 27 Mar 2021 22:19:17 +0000 (22:19 +0000)]
Auto merge of #83103 - petrochenkov:unilex, r=Aaron1011
resolve: Partially unify early and late scope-relative identifier resolution
Reuse `early_resolve_ident_in_lexical_scope` instead of a chunk of code in `resolve_ident_in_lexical_scope` doing the same job.
`early_resolve_ident_in_lexical_scope`/`visit_scopes` had to be slightly extended to be able to 1) start from a specific module instead of the current parent scope and 2) report one deprecation lint.
`early_resolve_ident_in_lexical_scope` still doesn't support walking through "ribs", that part is left in `resolve_ident_in_lexical_scope` (moreover, I'm pretty sure it's buggy, but that's a separate issue, cc https://github.com/rust-lang/rust/issues/52389 at least).
Joshua Nelson [Sat, 27 Mar 2021 19:56:07 +0000 (15:56 -0400)]
may not -> might not
"may not" has two possible meanings:
1. A command: "You may not stay up past your bedtime."
2. A fact that's only sometimes true: "Some cities may not have bike lanes."
In some cases, the meaning is ambiguous: "Some cars may not have snow
tires." (do the cars *happen* to not have snow tires, or is it
physically impossible for them to have snow tires?)
This changes places where the standard library uses the "description of
fact" meaning to say "might not" instead.
This is just `std::vec` for now - if you think this is a good idea I can
convert the rest of the standard library.
Dylan DPC [Sat, 27 Mar 2021 19:37:12 +0000 (20:37 +0100)]
Rollup merge of #83548 - Aaron1011:capture-none-delims, r=petrochenkov
Always preserve `None`-delimited groups in a captured `TokenStream`
Previously, we would silently remove any `None`-delimiters when
capturing a `TokenStream`, 'flattenting' them to their inner tokens.
This was not normally visible, since we usually have
`TokenKind::Interpolated` (which gets converted to a `None`-delimited
group during macro invocation) instead of an actual `None`-delimited
group.
However, there are a couple of cases where this becomes visible to
proc-macros:
1. A cross-crate `macro_rules!` macro has a `None`-delimited group
stored in its body (as a result of being produced by another
`macro_rules!` macro). The cross-crate `macro_rules!` invocation
can then expand to an attribute macro invocation, which needs
to be able to see the `None`-delimited group.
2. A proc-macro can invoke an attribute proc-macro with its re-collected
input. If there are any nonterminals present in the input, they will
get re-collected to `None`-delimited groups, which will then get
captured as part of the attribute macro invocation.
Both of these cases are incredibly obscure, so there hopefully won't be
any breakage. This change will allow more agressive 'flattenting' of
nonterminals in #82608 without losing `None`-delimited groups.
Dylan DPC [Sat, 27 Mar 2021 19:37:09 +0000 (20:37 +0100)]
Rollup merge of #82993 - camelid:source-use-diag, r=jyn514
rustdoc: Use diagnostics for error when including sources
This error probably almost never happens, but we should still use the
diagnostic infrastructure. My guess is that the error was added back
before rustdoc used the rustc diagnostic infrastructure (it was all
`println!` and `eprintln!` back then!) and since it likely rarely occurs
and this code doesn't change that much, no one thought to transition it
to using diagnostics.
Note that the old error was actually a warning (it didn't stop the rest
of doc building). It seems very unlikely that this would fail without
the rest of the doc build failing, so it makes more sense for it to be a
hard error.
The error looks like this:
error: failed to render source code for `src/test/rustdoc/smart-punct.rs`: "bar": foo
--> src/test/rustdoc/smart-punct.rs:3:1
|
3 | / #![crate_name = "foo"]
4 | |
5 | | //! This is the "start" of the 'document'! How'd you know that "it's" ...
6 | | //!
... |
22 | | //! I say "don't smart-punct me -- please!"
23 | | //! ```
| |_______^
I wasn't sure how to trigger the error, so to create that message I
temporarily made rustdoc always emit it. That's also why it says "bar"
and "foo" instead of a real error message.
Note that the span of the diagnostic starts at line 3 because line 1 of
that file is a (non-doc) comment and line 2 is a blank line.
Dylan DPC [Sat, 27 Mar 2021 19:37:07 +0000 (20:37 +0100)]
Rollup merge of #82917 - cuviper:iter-zip, r=m-ou-se
Add function core::iter::zip
This makes it a little easier to `zip` iterators:
```rust
for (x, y) in zip(xs, ys) {}
// vs.
for (x, y) in xs.into_iter().zip(ys) {}
```
You can `zip(&mut xs, &ys)` for the conventional `iter_mut()` and
`iter()`, respectively. This can also support arbitrary nesting, where
it's easier to see the item layout than with arbitrary `zip` chains:
```rust
for ((x, y), z) in zip(zip(xs, ys), zs) {}
for (x, (y, z)) in zip(xs, zip(ys, zs)) {}
// vs.
for ((x, y), z) in xs.into_iter().zip(ys).zip(xz) {}
for (x, (y, z)) in xs.into_iter().zip((ys.into_iter().zip(xz)) {}
```
It may also format more nicely, especially when the first iterator is a
longer chain of methods -- for example:
Dylan DPC [Sat, 27 Mar 2021 19:37:05 +0000 (20:37 +0100)]
Rollup merge of #82525 - RalfJung:unaligned-ref-warn, r=petrochenkov
make unaligned_references future-incompat lint warn-by-default
and also remove the safe_packed_borrows lint that it replaces.
`std::ptr::addr_of!` has hit beta now and will hit stable in a month, so I propose we start fixing https://github.com/rust-lang/rust/issues/27060 for real: creating a reference to a field of a packed struct needs to eventually become a hard error; this PR makes it a warn-by-default future-incompat lint. (The lint already existed, this just raises its default level.) At the same time I removed the corresponding code from unsafety checking; really there's no reason an `unsafe` block should make any difference here.
For references to packed fields outside `unsafe` blocks, this means `unaligned_refereces` replaces the previous `safe_packed_borrows` warning with a link to https://github.com/rust-lang/rust/issues/82523 (and no more talk about unsafe blocks making any difference). So behavior barely changes, the warning is just worded differently. For references to packed fields inside `unsafe` blocks, this PR shows a new future-incompat warning.
Closes https://github.com/rust-lang/rust/issues/46043 because that lint no longer exists.
Dylan DPC [Sat, 27 Mar 2021 19:37:04 +0000 (20:37 +0100)]
Rollup merge of #81351 - lcnr:big-money-big-prices, r=oli-obk
combine: stop eagerly evaluating consts
`super_relate_consts` eagerly evaluates constants which doesn't seem too great.
I now also finally understand why all of the unused substs test passed. The reason being
that we just evaluated the constants in `super_relate_consts` :laughing:
While this change isn't strictly necessary as evaluating consts here doesn't hurt, it still feels a lot cleaner to do it this way
Yuki Okushi [Sat, 27 Mar 2021 16:33:19 +0000 (01:33 +0900)]
Rollup merge of #83560 - m-ou-se:io-chain-debug, r=sfackler
Derive Debug for io::Chain instead of manually implementing it.
This derives Debug for io::Chain instead of manually implementing it.
The manual implementation has the same bounds, so I don't think there's any reason for a manual implementation. The names used in the derive implementation are even nicer (`first`/`second`) than the manual implementation (`t`/`u`), and include the `done_first` field too.
Yuki Okushi [Sat, 27 Mar 2021 16:33:18 +0000 (01:33 +0900)]
Rollup merge of #83559 - m-ou-se:rwlock-guard-debug-fix, r=jackh726
Fix Debug implementation for RwLock{Read,Write}Guard.
This would attempt to print the Debug representation of the lock that the guard has locked, which will try to lock again, fail, and just print `"<locked>"` unhelpfully.
After this change, this just prints the contents of the mutex, like the other smart pointers (and MutexGuard) do.
MutexGuard had this problem too: https://github.com/rust-lang/rust/issues/57702
Yuki Okushi [Sat, 27 Mar 2021 16:33:15 +0000 (01:33 +0900)]
Rollup merge of #83462 - ijackson:exitstatus-message-wording, r=joshtriplett
ExitStatus: print "exit status: {}" rather than "exit code: {}" on unix
Proper Unix terminology is "exit status" (vs "wait status"). "exit
code" is imprecise on Unix and therefore unclear. (As far as I can
tell, "exit code" is correct terminology on Windows.)
This new wording is unfortunately inconsistent with the identifier
names in the Rust stdlib.
It is the identifier names that are wrong, as discussed at length in eg
https://doc.rust-lang.org/nightly/std/process/struct.ExitStatus.html
https://doc.rust-lang.org/nightly/std/os/unix/process/trait.ExitStatusExt.html
Unfortunately for API stability reasons it would be a lot of work, and
a lot of disruption, to change the names in the stdlib (eg to rename
`std::process::ExitStatus` to `std::process::ChildStatus` or
something), but we should fix the message output. Many (probably
most) readers of these messages about exit statuses will be users and
system administrators, not programmers, who won't even know that Rust
has this wrong terminology.
So I think the right thing is to fix the documentation (as I have
already done) and, now, the terminology in the implementation.
This is a user-visible change to the behaviour of all Rust programs
which run Unix subprocesses. Hopefully no-one is matching against the
exit status string, except perhaps in tests.
Yuki Okushi [Sat, 27 Mar 2021 16:33:13 +0000 (01:33 +0900)]
Rollup merge of #83348 - osa1:issue83344, r=jackh726
format macro argument parsing fix
When the character next to `{}` is "shifted" (when mapping a byte index
in the format string to span) we should avoid shifting the span end
index, so first map the index of `}` to span, then bump the span,
instead of first mapping the next byte index to a span (which causes
bumping the end span too much).
bors [Sat, 27 Mar 2021 13:25:16 +0000 (13:25 +0000)]
Auto merge of #83245 - the8472:generalize-slice-fill, r=m-ou-se
Generalize and inline slice::fill specializations
This makes the memset specialization applicable to more types. And since the code now lives in a generic method it is also eligible for cross-crate inlining which should fix #83235
Mara Bos [Sat, 27 Mar 2021 12:36:07 +0000 (13:36 +0100)]
Derive Debug for io::Chain instead of manually implementing it.
The manual implementation has the same bounds, so I don't think there's
any reason for a manual implementation. The names used in the derive
implementation are even nicer (`first`/`second`) than the manual
implementation (`t`/`u`), and include the `done_first` field too.
Mara Bos [Sat, 27 Mar 2021 12:31:17 +0000 (13:31 +0100)]
Fix Debug implementation for RwLock{Read,Write}Guard.
This would attempt to print the Debug representation of the lock that
the guard has locked, which will try to lock again, fail, and just print
"<locked>" unhelpfully.
After this change, this just prints the contents of the mutex, like the
other smart pointers (and MutexGuard) do.
bors [Sat, 27 Mar 2021 10:40:16 +0000 (10:40 +0000)]
Auto merge of #78618 - workingjubilee:ieee754-fmt, r=m-ou-se
Add IEEE 754 compliant fmt/parse of -0, infinity, NaN
This pull request improves the Rust float formatting/parsing libraries to comply with IEEE 754's formatting expectations around certain special values, namely signed zero, the infinities, and NaN. It also adds IEEE 754 compliance tests that, while less stringent in certain places than many of the existing flt2dec/dec2flt capability tests, are intended to serve as the beginning of a roadmap to future compliance with the standard. Some relevant documentation is also adjusted with clarifying remarks.
This PR follows from discussion in https://github.com/rust-lang/rfcs/issues/1074, and closes #24623.
The most controversial change here is likely to be that -0 is now printed as -0. Allow me to explain: While there appears to be community support for an opt-in toggle of printing floats as if they exist in the naively expected domain of numbers, i.e. not the extended reals (where floats live), IEEE 754-2019 is clear that a float converted to a string should be capable of being transformed into the original floating point bit-pattern when it satisfies certain conditions (namely, when it is an actual numeric value i.e. not a NaN and the original and destination float width are the same). -0 is given special attention here as a value that should have its sign preserved. In addition, the vast majority of other programming languages not only output `-0` but output `-0.0` here.
While IEEE 754 offers a broad leeway in how to handle producing what it calls a "decimal character sequence", it is clear that the operations a language provides should be capable of round tripping, and it is confusing to advertise the f32 and f64 types as binary32 and binary64 yet have the most basic way of producing a string and then reading it back into a floating point number be non-conformant with the standard. Further, existing documentation suggested that e.g. -0 would be printed with -0 regardless of the presence of the `+` fmt character, but it prints "+0" instead if given such (which was what led to the opening of #24623).
There are other parsing and formatting issues for floating point numbers which prevent Rust from complying with the standard, as well as other well-documented challenges on the arithmetic level, but I hope that this can be the beginning of motion towards solving those challenges.
When the character next to `{}` is "shifted" (when mapping a byte index
in the format string to span) we should avoid shifting the span end
index, so first map the index of `}` to span, then bump the span,
instead of first mapping the next byte index to a span (which causes
bumping the end span too much).
bors [Sat, 27 Mar 2021 07:59:24 +0000 (07:59 +0000)]
Auto merge of #83547 - JohnTitor:rollup-qh7j6hg, r=JohnTitor
Rollup of 9 pull requests
Successful merges:
- #83239 (Remove/replace some outdated crates from the dependency tree)
- #83328 (Fixes to inline assmebly tests)
- #83343 (Simplify and fix byte skipping in format! string parser)
- #83388 (Make # pretty print format easier to discover)
- #83431 (Tell GitHub to highlight `config.toml.example` as TOML)
- #83508 (Use the direct link to the platform support page)
- #83511 (compiletest: handle llvm_version with suffix like "12.0.0libcxx")
- #83524 (Document that the SocketAddr memory representation is not stable)
- #83525 (fix doc comment for `ty::Dynamic`)
Yuki Okushi [Sat, 27 Mar 2021 03:37:24 +0000 (12:37 +0900)]
Rollup merge of #83524 - faern:document-socketaddr-mem-layout, r=sfackler
Document that the SocketAddr memory representation is not stable
Intended to help out with #78802. Work has been put into finding and fixing code that assumes the memory layout of `SocketAddrV4` and `SocketAddrV6`. But it turns out there are cases where new code continues to make the same assumption ([example](https://github.com/spacejam/seaslug/commit/96927dc2b7b918860a79c4eb6336051e52c6137a#diff-917db3d8ca6f862ebf42726b23c72a12b35e584e497ebdb24e474348d7c6ffb6R610-R621)).
The memory layout of a type in `std` is never part of the public API. Unless explicitly stated I guess. But since that is invalidly relied upon by a considerable amount of code for these particular types, it might make sense to explicitly document this. This can be temporary. Once #78802 lands it does not make sense to rely on the layout any longer, and this documentation can also be removed.
Yuki Okushi [Sat, 27 Mar 2021 03:37:23 +0000 (12:37 +0900)]
Rollup merge of #83511 - 12101111:fix-llvm-version-suffix, r=Mark-Simulacrum
compiletest: handle llvm_version with suffix like "12.0.0libcxx"
The previous code only remove the suffix begin with `-`, but Gentoo Linux [define `LLVM_VERSION_SUFFIX="libcxx"`](https://github.com/gentoo/gentoo/blob/604d79f327176eecb05293d7154e24231229cb31/sys-devel/llvm/llvm-11.1.0.ebuild#L378) when llvm is linked to libc++ and lead to a panic:
```
thread 'main' panicked at 'Malformed version component: ParseIntError { kind: InvalidDigit }', src/tools/compiletest/src/header.rs:968:28
```
This new code will handle all suffix not beginning with digit or dot.
Yuki Okushi [Sat, 27 Mar 2021 03:37:20 +0000 (12:37 +0900)]
Rollup merge of #83388 - alamb:alamb/fmt-dcs, r=Mark-Simulacrum
Make # pretty print format easier to discover
# Rationale:
I use (cargo cult?) three formats in rust: `{}`, debug `{:?}`, and pretty-print debug `{:#?}`. I discovered `{:#?}` in some blog post or guide when I started working in Rust. While `#` is documented I think it is hard to discover. So taking the good advice of ```@carols10cents``` I am trying to improve the docs with a PR
As a reminder "pretty print" means that where `{:?}` will print something like
```
foo: { b1: 1, b2: 2}
```
`{:#?}` will prints something like
```
foo {
b1: 1
b2: 3
}
```
# Changes
Add an example to `fmt` to try and make it easier to discover `#`