Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Remove no_integrated_as mode. #70345

Merged
merged 1 commit into from
Mar 28, 2020
Merged

Remove no_integrated_as mode. #70345

merged 1 commit into from
Mar 28, 2020

Conversation

nnethercote
Copy link
Contributor

Specifically, remove both -Z no_integrated_as and
TargetOptions::no_integrated_as. The latter was only used for the
msp430_none_elf platform, for which it's no longer required.

r? @alexcrichton

@rust-highfive rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Mar 24, 2020
@nnethercote
Copy link
Contributor Author

Note that I'm not 100% certain that msp430_none_elf no longer needs no_integrated_as. #43099 was the PR that added that target, saying:

Once the LLVM backend gains the ability to directly produce MSP430 object files
we can stop depending on msp430-elf-gcc for producing object files; when that
occurs the asm related fields will change. This change won't break existing
user code.

That was written in July 2017, and I'm hoping that the LLVM backend now has that ability. @japaric, are you able to confirm if it does?

@petrochenkov
Copy link
Contributor

The latter was only used for the
msp430_none_elf platform, for which it's no longer required.

Targets are not limited to those that are built-in and exist now though.
It is useful to be able to connect rustc to new LLVM backends in various stages of development (including proprietary ones), that may mean the lack of complete integrated assembler.

@bors
Copy link
Contributor

bors commented Mar 24, 2020

☔ The latest upstream changes (presumably #70351) made this pull request unmergeable. Please resolve the merge conflicts.

@jonas-schievink
Copy link
Contributor

MSP430 no longer needs this, but like @petrochenkov said, external targets might. Is there any specific reason for removing support for this? Has the code been a big maintenance burden or slowed down development in other ways? If not I'd like to kindly request not to remove it.

@alexcrichton
Copy link
Member

Code-wise this looks fine by me, thanks @nnethercote!

@jonas-schievink the codegen backend is currently a large and sprawling maze of lots of conditionals, so having extra ones "just because" is a burden on maintaining this because everyone new who looks at the backend has to understand themselves why there's all these odd cases and booleans for what to do.

@petrochenkov personally I don't disagree that it's great to work wherever possible, but there's a very real cost on maintenance here. If there's no clear benefit my opinion is that it should be removed, and at this time I think there's unclear benefit because there's no known target using this.

@Dylan-DPC-zz
Copy link

@nnethercote @alexcrichton any updates on this? do we want to merge it ?

@alexcrichton
Copy link
Member

@Dylan-DPC er it seems a bit early to be pinging about this? This PR itself is only 14 hours old and the last discussion was 4 hours ago. I'm not sure @nnethercote is even awake yet to continue discussion on this...

@Dylan-DPC-zz
Copy link

@alexcrichton oops too many tabs open. Sorry :D

@nnethercote
Copy link
Contributor Author

I have started looking at this codegen stuff because I am working on #66961. As Alex says, the codegen backend is complex and messy. I've already filed a couple of clean-up PRs as preliminaries: #70297, #70289. Alex suggested that no_integrated_as was unnecessary and as a newcomer to the codebase who is trying to understand it and clean it up, I am sympathetic to that view. But I've never worked on adding a new target platform and I am definitely not an expert on this stuff.

@alexcrichton
Copy link
Member

I personally feel that if we need this functionality it should come back in a less-complicated form. For example we shouldn't have to plumb many different options around the compiler. Instead there should be a desire for an object file with one piece of configuration saying "instead of LLVM generating an object file, use this Command to generate an object file".

Having a small local if statement is much more easily maintained and understood than something which ends up being extremely rarely used yet sprawling over the codebase.

@hanna-kruppe
Copy link
Contributor

FWIW, I have combined rustc with out-of-tree LLVM backends and also worked on LLVM support for new ISA extensions, and I have never had a need for this flag. Teaching LLVM to emit basic object files is usually simple compared to everything else that goes into backend development, and usually quite valuable. Certainly there are targets out there which don't prioritize it, but in my experience, being complete enough to be used with rustc but unable to produce object files is an anomaly rather than a normal stage in an LLVM backend's lifecycle.

@nnethercote nnethercote mentioned this pull request Mar 26, 2020
Merged
@nnethercote
Remove no_integrated_as mode.
02840ca 
Specifically, remove both `-Z no_integrated_as` and
`TargetOptions::no_integrated_as`. The latter was only used for the
`msp430_none_elf` platform, for which it's no longer required.
@nnethercote
Copy link
Contributor Author

FWIW, I rebased this code. After #70384 it's looking even nicer.

@alexcrichton
Copy link
Member

@bors: r+

This looks like a great simplification to me in what's already pretty gnarly code. It's also relatively easy to revert if truly necessary and we can go back to the drawing board to figure out how to integrate this in a cleaner fashion. Because of that, I'm gonna r+ this.

@bors
Copy link
Contributor

bors commented Mar 27, 2020

📌 Commit 02840ca has been approved by alexcrichton

@bors bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Mar 27, 2020
@Centril Centril mentioned this pull request Mar 27, 2020
Merged
bors added a commit to rust-lang-ci/rust that referenced this pull request Mar 28, 2020
@bors
Auto merge of rust-lang#70483 - Centril:rollup-slli4yf, r=Centril
2acf32d 
Rollup of 5 pull requests

Successful merges:

 - rust-lang#70345 (Remove `no_integrated_as` mode.)
 - rust-lang#70434 (suggest `;` on expr `mac!()` which is good as stmt `mac!()`)
 - rust-lang#70457 (non-exhastive diagnostic: add note re. scrutinee type)
 - rust-lang#70478 (Refactor type_of for constants)
 - rust-lang#70480 (clarify hir_id <-> node_id method names)

Failed merges:

r? @ghost
@bors bors merged commit 08e867c into rust-lang:master Mar 28, 2020
@nnethercote nnethercote deleted the rm-no_integrated_as branch March 28, 2020 06:43
@nnethercote
Copy link
Contributor Author

@Centril: this PR removed a -C option, and so should be mentioned in release notes.

@Centril Centril added the relnotes Marks issues that should be documented in the release notes of the next release. label Apr 3, 2020
@Centril Centril added the T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. label Apr 3, 2020
@Centril Centril added this to the 1.44 milestone Apr 3, 2020
@cr1901
Copy link
Contributor

cr1901 commented Apr 10, 2020 •
edited
Loading

A bit "too-little too-late" for me, but I've been trying to diagnose some regressions that I suspect popped up in msp430 during the switch to an integrated assembler. It's taking a while because:

  1. Well, the state of the world's pretty bad at the moment.
  2. The best granularity I have, even after bisecting, is a dozen or so commits thanks to rollups.

Right now, I'm not sure if this'll be necessary, but @alexcrichton seems to imply it's easy to add this option back in in a less-complicated form as a workaround?

I personally feel that if we need this functionality it should come back in a less-complicated form. For example we shouldn't have to plumb many different options around the compiler. Instead there should be a desire for an object file with one piece of configuration saying "instead of LLVM generating an object file, use this Command to generate an object file".

I simply don't think I have the bandwidth to hack on LLVM, but would be interested in discussing re-enabling a no-integrated-as option cleanly if necessary.

EDIT: Additionally for the purposes of experimentation, it's possible to target MIR or Cranelift instead of LLVM. I know people besides me are concerned about alternate backends existing- the refactor wouldn't have happened otherwise. How does the removal of no_integrated_as interact with alternate backends?

@bjorn3
Copy link
Member

bjorn3 commented Apr 10, 2020

How does the removal of no_integrated_as interact with alternate backends?

At least rustc_codegen_cranelift didn't support it anyway. Cranelift only supports emitting machine code, not assembly. Also code that was removed in this PR was either in the LLVM backend, or was just some utils in the rustc_codegen_ssa. This means that it is possible to implement this in other backends without touching rustc.

@alexcrichton
Copy link
Member

@cr1901 I don't know if it'll be easy per se, but I think it would be best to try to integrate it today in an ideally cleaner form. For example it would ideally be one local decision along the lines of "either have LLVM produce the object or have LLVM produce the asm and we run an external assembler". This previously touched a number of parts in the codebase (as can be seen here).

@cr1901
Copy link
Contributor

cr1901 commented Apr 10, 2020

At least rustc_codegen_cranelift didn't support it anyway. Cranelift only supports emitting machine code, not assembly. Also code that was removed in this PR was either in the LLVM backend, or was just some utils in the rustc_codegen_ssa. This means that it is possible to implement this in other backends without touching rustc.

Ack.

@alexcrichton Right now, I'm not sure how I would add the option back in. But if you want it reintroduced in a cleaner form, then great! Let me figure out whether the change to an integrated assembler is to blame for my woes in the first place.

If it is, then I think I have some interest in reintroducing this option if someone can give me some guidance on how it can be done (under the assumption it's far less painful than hacking on llvm, and it gives me a means to compare and contrast if I find new regressions).

Also,

in my experience, being complete enough to be used with rustc but unable to produce object files is an anomaly rather than a normal stage in an LLVM backend's lifecycle.

Perhaps, but if the gnu assembler is perfectly good as-is for a target without an upstream llvm backend, not needing to implement the assembler means I can test Rust binaries in practice more quickly. :P

@alexcrichton
Copy link
Member

Unfortuantely I don't personally have time to review/guide how to re-add this. Others on the compiler team may be able to help out though!

nnethercote added a commit to nnethercote/rust that referenced this pull request Apr 19, 2020
@nnethercote
Improve codegen option docs.
dc06539 
This commit:
- Adds "following values" indicators for all the options that are
  missing them.
- Tweaks some wording and punctuation for consistency.
- Rewords some things for clarity.
- Removes the `no-integrated-as` entry, because that option was removed
  in rust-lang#70345.
netbsd-srcmastr pushed a commit to NetBSD/pkgsrc that referenced this pull request Jun 8, 2020
@he32
Update rust to version 1.44.0.
705c0b3 
Pkgsrc changes:
 * Remove a couple diffs which are now integrated upstream.
 * Adjust cargo checksums after upstream upgrades.
 * Belatedly bump the curl dependency
 * Unset DESTDIR during the build phase, to work around a mysterious
   build bug deep in the bowels of llvm.
 * Bump nearly all bootstraps to 1.43.1.

Upstream changes:

Version 1.44.0 (2020-06-04)
==========================

Language
--------
- [You can now use `async/.await` with `#[no_std]` enabled.][69033]
- [Added the `unused_braces` lint.][70081]

**Syntax-only changes**

- [Expansion-driven outline module parsing][69838]
```rust
#[cfg(FALSE)]
mod foo {
    mod bar {
        mod baz; // `foo/bar/baz.rs` doesn't exist, but no error!
    }
}
```

These are still rejected semantically, so you will likely receive an error but
these changes can be seen and parsed by macros and conditional compilation.

Compiler
--------
- [Rustc now respects the `-C codegen-units` flag in incremental mode.][70156]
  Additionally when in incremental mode rustc defaults to 256 codegen units.
- [Refactored `catch_unwind`, to have zero-cost unless unwinding is enabled and
  a panic is thrown.][67502]
- [Added tier 3\* support for the `aarch64-unknown-none` and
  `aarch64-unknown-none-softfloat` targets.][68334]
- [Added tier 3 support for `arm64-apple-tvos` and
  `x86_64-apple-tvos` targets.][68191]

Libraries
---------
- [Special cased `vec![]` to map directly to `Vec::new()`.][70632] This allows
  `vec![]` to be able to be used in `const` contexts.
- [`convert::Infallible` now implements `Hash`.][70281]
- [`OsString` now implements `DerefMut` and `IndexMut` returning
  a `&mut OsStr`.][70048]
- [Unicode 13 is now supported.][69929]
- [`String` now implements `From<&mut str>`.][69661]
- [`IoSlice` now implements `Copy`.][69403]
- [`Vec<T>` now implements `From<[T; N]>`.][68692] Where `N` is less than 32.
- [`proc_macro::LexError` now implements `fmt::Display` and `Error`.][68899]
- [`from_le_bytes`, `to_le_bytes`, `from_be_bytes`, `to_be_bytes`,
  `from_ne_bytes`, and `to_ne_bytes` methods are now `const` for all
  integer types.][69373]

Stabilized APIs
---------------
- [`PathBuf::with_capacity`]
- [`PathBuf::capacity`]
- [`PathBuf::clear`]
- [`PathBuf::reserve`]
- [`PathBuf::reserve_exact`]
- [`PathBuf::shrink_to_fit`]
- [`f32::to_int_unchecked`]
- [`f64::to_int_unchecked`]
- [`Layout::align_to`]
- [`Layout::pad_to_align`]
- [`Layout::array`]
- [`Layout::extend`]

Cargo
-----
- [Added the `cargo tree` command which will print a tree graph of
  your dependencies.][cargo/8062] E.g.
  ```
    mdbook v0.3.2 (/Users/src/rust/mdbook)
  +-- ammonia v3.0.0
  |   +-- html5ever v0.24.0
  |   |   +-- log v0.4.8
  |   |   |   +-- cfg-if v0.1.9
  |   |   +-- mac v0.1.1
  |   |   +-- markup5ever v0.9.0
  |   |       +-- log v0.4.8 (*)
  |   |       +-- phf v0.7.24
  |   |       |   +-- phf_shared v0.7.24
  |   |       |       +-- siphasher v0.2.3
  |   |       |       +-- unicase v1.4.2
  |   |       |           [build-dependencies]
  |   |       |           +-- version_check v0.1.5
  ...
  ```
  You can also display dependencies on multiple versions of the same crate with
  `cargo tree -d` (short for `cargo tree --duplicates`).

Misc
----
- [Rustdoc now allows you to specify `--crate-version` to have rustdoc include
  the version in the sidebar.][69494]

Compatibility Notes
-------------------
- [Rustc now correctly generates static libraries on Windows GNU targets with
  the `.a` extension, rather than the previous `.lib`.][70937]
- [Removed the `-C no_integrated_as` flag from rustc.][70345]
- [The `file_name` property in JSON output of macro errors now points the actual
  source file rather than the previous format of `<NAME macros>`.][70969]
  **Note:** this may not point a file that actually exists on the user's system.
- [The minimum required external LLVM version has been bumped to LLVM 8.][71147]
- [`mem::{zeroed, uninitialised}` will now panic when used with types that do
  not allow zero initialization such as `NonZeroU8`.][66059] This was
  previously a warning.
- [In 1.45.0 (the next release) converting a `f64` to `u32` using the `as`
  operator has been defined as a saturating operation.][71269] This was
  previously undefined behaviour, you can use the `{f64, f32}::to_int_unchecked`
  methods to continue using the current behaviour which may desirable in rare
  performance sensitive situations.

Internal Only
-------------
These changes provide no direct user facing benefits, but represent significant
improvements to the internals and overall performance of rustc and
related tools.

- [dep_graph Avoid allocating a set on when the number reads are small.][69778]
- [Replace big JS dict with JSON parsing.][71250]

[69373]: rust-lang/rust#69373
[66059]: rust-lang/rust#66059
[68191]: rust-lang/rust#68191
[68899]: rust-lang/rust#68899
[71147]: rust-lang/rust#71147
[71250]: rust-lang/rust#71250
[70937]: rust-lang/rust#70937
[70969]: rust-lang/rust#70969
[70632]: rust-lang/rust#70632
[70281]: rust-lang/rust#70281
[70345]: rust-lang/rust#70345
[70048]: rust-lang/rust#70048
[70081]: rust-lang/rust#70081
[70156]: rust-lang/rust#70156
[71269]: rust-lang/rust#71269
[69838]: rust-lang/rust#69838
[69929]: rust-lang/rust#69929
[69661]: rust-lang/rust#69661
[69778]: rust-lang/rust#69778
[69494]: rust-lang/rust#69494
[69403]: rust-lang/rust#69403
[69033]: rust-lang/rust#69033
[68692]: rust-lang/rust#68692
[68334]: rust-lang/rust#68334
[67502]: rust-lang/rust#67502
[cargo/8062]: rust-lang/cargo#8062
[`PathBuf::with_capacity`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.with_capacity
[`PathBuf::capacity`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.capacity
[`PathBuf::clear`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.clear
[`PathBuf::reserve`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.reserve
[`PathBuf::reserve_exact`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.reserve_exact
[`PathBuf::shrink_to_fit`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.shrink_to_fit
[`f32::to_int_unchecked`]: https://doc.rust-lang.org/std/primitive.f32.html#method.to_int_unchecked
[`f64::to_int_unchecked`]: https://doc.rust-lang.org/std/primitive.f64.html#method.to_int_unchecked
[`Layout::align_to`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.align_to
[`Layout::pad_to_align`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.pad_to_align
[`Layout::array`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.array
[`Layout::extend`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.extend
netbsd-srcmastr pushed a commit to NetBSD/pkgsrc that referenced this pull request Jun 9, 2020
@alarixnia
rustc: Update to 1.44.0
e76ac9f 
Version 1.44.0 (2020-06-04)
==========================

Language
--------
- [You can now use `async/.await` with `#[no_std]` enabled.][69033]
- [Added the `unused_braces` lint.][70081]

**Syntax-only changes**

- [Expansion-driven outline module parsing][69838]
```rust
#[cfg(FALSE)]
mod foo {
    mod bar {
        mod baz; // `foo/bar/baz.rs` doesn't exist, but no error!
    }
}
```

These are still rejected semantically, so you will likely receive an error but
these changes can be seen and parsed by macros and conditional compilation.

Compiler
--------
- [Rustc now respects the `-C codegen-units` flag in incremental mode.][70156]
  Additionally when in incremental mode rustc defaults to 256 codegen units.
- [Refactored `catch_unwind` to have zero-cost, unless unwinding is enabled and
  a panic is thrown.][67502]
- [Added tier 3\* support for the `aarch64-unknown-none` and
  `aarch64-unknown-none-softfloat` targets.][68334]
- [Added tier 3 support for `arm64-apple-tvos` and
  `x86_64-apple-tvos` targets.][68191]


Libraries
---------
- [Special cased `vec![]` to map directly to `Vec::new()`.][70632] This allows
  `vec![]` to be able to be used in `const` contexts.
- [`convert::Infallible` now implements `Hash`.][70281]
- [`OsString` now implements `DerefMut` and `IndexMut` returning
  a `&mut OsStr`.][70048]
- [Unicode 13 is now supported.][69929]
- [`String` now implements `From<&mut str>`.][69661]
- [`IoSlice` now implements `Copy`.][69403]
- [`Vec<T>` now implements `From<[T; N]>`.][68692] Where `N` is at most 32.
- [`proc_macro::LexError` now implements `fmt::Display` and `Error`.][68899]
- [`from_le_bytes`, `to_le_bytes`, `from_be_bytes`, `to_be_bytes`,
  `from_ne_bytes`, and `to_ne_bytes` methods are now `const` for all
  integer types.][69373]

Stabilized APIs
---------------
- [`PathBuf::with_capacity`]
- [`PathBuf::capacity`]
- [`PathBuf::clear`]
- [`PathBuf::reserve`]
- [`PathBuf::reserve_exact`]
- [`PathBuf::shrink_to_fit`]
- [`f32::to_int_unchecked`]
- [`f64::to_int_unchecked`]
- [`Layout::align_to`]
- [`Layout::pad_to_align`]
- [`Layout::array`]
- [`Layout::extend`]

Cargo
-----
- [Added the `cargo tree` command which will print a tree graph of
  your dependencies.][cargo/8062] E.g.
  ```
    mdbook v0.3.2 (/Users/src/rust/mdbook)
  ├── ammonia v3.0.0
  │   ├── html5ever v0.24.0
  │   │   ├── log v0.4.8
  │   │   │   └── cfg-if v0.1.9
  │   │   ├── mac v0.1.1
  │   │   └── markup5ever v0.9.0
  │   │       ├── log v0.4.8 (*)
  │   │       ├── phf v0.7.24
  │   │       │   └── phf_shared v0.7.24
  │   │       │       ├── siphasher v0.2.3
  │   │       │       └── unicase v1.4.2
  │   │       │           [build-dependencies]
  │   │       │           └── version_check v0.1.5
  ...
  ```
  You can also display dependencies on multiple versions of the same crate with
  `cargo tree -d` (short for `cargo tree --duplicates`).

Misc
----
- [Rustdoc now allows you to specify `--crate-version` to have rustdoc include
  the version in the sidebar.][69494]

Compatibility Notes
-------------------
- [Rustc now correctly generates static libraries on Windows GNU targets with
  the `.a` extension, rather than the previous `.lib`.][70937]
- [Removed the `-C no_integrated_as` flag from rustc.][70345]
- [The `file_name` property in JSON output of macro errors now points the actual
  source file rather than the previous format of `<NAME macros>`.][70969]
  **Note:** this may not point to a file that actually exists on the user's system.
- [The minimum required external LLVM version has been bumped to LLVM 8.][71147]
- [`mem::{zeroed, uninitialised}` will now panic when used with types that do
  not allow zero initialization such as `NonZeroU8`.][66059] This was
  previously a warning.
- [In 1.45.0 (the next release) converting a `f64` to `u32` using the `as`
  operator has been defined as a saturating operation.][71269] This was previously
  undefined behaviour, but you can use the `{f64, f32}::to_int_unchecked` methods to
  continue using the current behaviour, which may be desirable in rare performance
  sensitive situations.

Internal Only
-------------
These changes provide no direct user facing benefits, but represent significant
improvements to the internals and overall performance of rustc and
related tools.

- [dep_graph Avoid allocating a set on when the number reads are small.][69778]
- [Replace big JS dict with JSON parsing.][71250]

[69373]: rust-lang/rust#69373
[66059]: rust-lang/rust#66059
[68191]: rust-lang/rust#68191
[68899]: rust-lang/rust#68899
[71147]: rust-lang/rust#71147
[71250]: rust-lang/rust#71250
[70937]: rust-lang/rust#70937
[70969]: rust-lang/rust#70969
[70632]: rust-lang/rust#70632
[70281]: rust-lang/rust#70281
[70345]: rust-lang/rust#70345
[70048]: rust-lang/rust#70048
[70081]: rust-lang/rust#70081
[70156]: rust-lang/rust#70156
[71269]: rust-lang/rust#71269
[69838]: rust-lang/rust#69838
[69929]: rust-lang/rust#69929
[69661]: rust-lang/rust#69661
[69778]: rust-lang/rust#69778
[69494]: rust-lang/rust#69494
[69403]: rust-lang/rust#69403
[69033]: rust-lang/rust#69033
[68692]: rust-lang/rust#68692
[68334]: rust-lang/rust#68334
[67502]: rust-lang/rust#67502
[cargo/8062]: rust-lang/cargo#8062
[`PathBuf::with_capacity`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.with_capacity
[`PathBuf::capacity`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.capacity
[`PathBuf::clear`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.clear
[`PathBuf::reserve`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.reserve
[`PathBuf::reserve_exact`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.reserve_exact
[`PathBuf::shrink_to_fit`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.shrink_to_fit
[`f32::to_int_unchecked`]: https://doc.rust-lang.org/std/primitive.f32.html#method.to_int_unchecked
[`f64::to_int_unchecked`]: https://doc.rust-lang.org/std/primitive.f64.html#method.to_int_unchecked
[`Layout::align_to`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.align_to
[`Layout::pad_to_align`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.pad_to_align
[`Layout::array`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.array
[`Layout::extend`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.extend
netbsd-srcmastr pushed a commit to NetBSD/pkgsrc that referenced this pull request Jul 6, 2020
@he32
Re-update rust to version 1.44.0.
316dcfc 
Pkgsrc changes:
 * Remove the clutter caused by the cross-compile setup from Makefile
   (Now consigned to my own private cross.mk file.)
 * Remove a couple of patches which are now integrated upstream.
 * Minor adjustments to a couple of other patches.
 * Adjust cargo checksums after upstream upgrades.
 * Belatedly bump the curl dependency
 * If doing a "dist" build, unset DESTDIR during the build phase,
   to work around a mysterious build bug deep in the bowels of llvm,
   causing llvm tools to be installed to a directory unexpecetd by
   the rest of the rust build, ref.
   rust-lang/rust#73132
   A "dist" build is not expected to be followed by an "install".
 * Bump nearly all bootstraps to 1.43.1; NetBSD earmv7hf bootstrap
   bumped to 1.44.0, as that one now finally builds and works.

Upstream changes:

Version 1.44.0 (2020-06-04)
==========================

Language
--------
- [You can now use `async/.await` with `#[no_std]` enabled.][69033]
- [Added the `unused_braces` lint.][70081]

**Syntax-only changes**

- [Expansion-driven outline module parsing][69838]
```rust
#[cfg(FALSE)]
mod foo {
    mod bar {
        mod baz; // `foo/bar/baz.rs` doesn't exist, but no error!
    }
}
```

These are still rejected semantically, so you will likely receive an error but
these changes can be seen and parsed by macros and conditional compilation.

Compiler
--------
- [Rustc now respects the `-C codegen-units` flag in incremental mode.][70156]
  Additionally when in incremental mode rustc defaults to 256 codegen units.
- [Refactored `catch_unwind`, to have zero-cost unless unwinding is enabled and
  a panic is thrown.][67502]
- [Added tier 3\* support for the `aarch64-unknown-none` and
  `aarch64-unknown-none-softfloat` targets.][68334]
- [Added tier 3 support for `arm64-apple-tvos` and
  `x86_64-apple-tvos` targets.][68191]

Libraries
---------
- [Special cased `vec![]` to map directly to `Vec::new()`.][70632] This allows
  `vec![]` to be able to be used in `const` contexts.
- [`convert::Infallible` now implements `Hash`.][70281]
- [`OsString` now implements `DerefMut` and `IndexMut` returning
  a `&mut OsStr`.][70048]
- [Unicode 13 is now supported.][69929]
- [`String` now implements `From<&mut str>`.][69661]
- [`IoSlice` now implements `Copy`.][69403]
- [`Vec<T>` now implements `From<[T; N]>`.][68692] Where `N` is less than 32.
- [`proc_macro::LexError` now implements `fmt::Display` and `Error`.][68899]
- [`from_le_bytes`, `to_le_bytes`, `from_be_bytes`, `to_be_bytes`,
  `from_ne_bytes`, and `to_ne_bytes` methods are now `const` for all
  integer types.][69373]

Stabilized APIs
---------------
- [`PathBuf::with_capacity`]
- [`PathBuf::capacity`]
- [`PathBuf::clear`]
- [`PathBuf::reserve`]
- [`PathBuf::reserve_exact`]
- [`PathBuf::shrink_to_fit`]
- [`f32::to_int_unchecked`]
- [`f64::to_int_unchecked`]
- [`Layout::align_to`]
- [`Layout::pad_to_align`]
- [`Layout::array`]
- [`Layout::extend`]

Cargo
-----
- [Added the `cargo tree` command which will print a tree graph of
  your dependencies.][cargo/8062] E.g.
  ```
    mdbook v0.3.2 (/Users/src/rust/mdbook)
  +-- ammonia v3.0.0
  |   +-- html5ever v0.24.0
  |   |   +-- log v0.4.8
  |   |   |   +-- cfg-if v0.1.9
  |   |   +-- mac v0.1.1
  |   |   +-- markup5ever v0.9.0
  |   |       +-- log v0.4.8 (*)
  |   |       +-- phf v0.7.24
  |   |       |   +-- phf_shared v0.7.24
  |   |       |       +-- siphasher v0.2.3
  |   |       |       +-- unicase v1.4.2
  |   |       |           [build-dependencies]
  |   |       |           +-- version_check v0.1.5
  ...
  ```
  You can also display dependencies on multiple versions of the same crate with
  `cargo tree -d` (short for `cargo tree --duplicates`).

Misc
----
- [Rustdoc now allows you to specify `--crate-version` to have rustdoc include
  the version in the sidebar.][69494]

Compatibility Notes
-------------------
- [Rustc now correctly generates static libraries on Windows GNU targets with
  the `.a` extension, rather than the previous `.lib`.][70937]
- [Removed the `-C no_integrated_as` flag from rustc.][70345]
- [The `file_name` property in JSON output of macro errors now points the actual
  source file rather than the previous format of `<NAME macros>`.][70969]
  **Note:** this may not point a file that actually exists on the user's system.
- [The minimum required external LLVM version has been bumped to LLVM 8.][71147]
- [`mem::{zeroed, uninitialised}` will now panic when used with types that do
  not allow zero initialization such as `NonZeroU8`.][66059] This was
  previously a warning.
- [In 1.45.0 (the next release) converting a `f64` to `u32` using the `as`
  operator has been defined as a saturating operation.][71269] This was
  previously undefined behaviour, you can use the `{f64, f32}::to_int_unchecked`
  methods to continue using the current behaviour which may desirable in rare
  performance sensitive situations.

Internal Only
-------------
These changes provide no direct user facing benefits, but represent significant
improvements to the internals and overall performance of rustc and
related tools.

- [dep_graph Avoid allocating a set on when the number reads are small.][69778]
- [Replace big JS dict with JSON parsing.][71250]

[69373]: rust-lang/rust#69373
[66059]: rust-lang/rust#66059
[68191]: rust-lang/rust#68191
[68899]: rust-lang/rust#68899
[71147]: rust-lang/rust#71147
[71250]: rust-lang/rust#71250
[70937]: rust-lang/rust#70937
[70969]: rust-lang/rust#70969
[70632]: rust-lang/rust#70632
[70281]: rust-lang/rust#70281
[70345]: rust-lang/rust#70345
[70048]: rust-lang/rust#70048
[70081]: rust-lang/rust#70081
[70156]: rust-lang/rust#70156
[71269]: rust-lang/rust#71269
[69838]: rust-lang/rust#69838
[69929]: rust-lang/rust#69929
[69661]: rust-lang/rust#69661
[69778]: rust-lang/rust#69778
[69494]: rust-lang/rust#69494
[69403]: rust-lang/rust#69403
[69033]: rust-lang/rust#69033
[68692]: rust-lang/rust#68692
[68334]: rust-lang/rust#68334
[67502]: rust-lang/rust#67502
[cargo/8062]: rust-lang/cargo#8062
[`PathBuf::with_capacity`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.with_capacity
[`PathBuf::capacity`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.capacity
[`PathBuf::clear`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.clear
[`PathBuf::reserve`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.reserve
[`PathBuf::reserve_exact`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.reserve_exact
[`PathBuf::shrink_to_fit`]: https://doc.rust-lang.org/std/path/struct.PathBuf.html#method.shrink_to_fit
[`f32::to_int_unchecked`]: https://doc.rust-lang.org/std/primitive.f32.html#method.to_int_unchecked
[`f64::to_int_unchecked`]: https://doc.rust-lang.org/std/primitive.f64.html#method.to_int_unchecked
[`Layout::align_to`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.align_to
[`Layout::pad_to_align`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.pad_to_align
[`Layout::array`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.array
[`Layout::extend`]: https://doc.rust-lang.org/std/alloc/struct.Layout.html#method.extend
@petrochenkov
Copy link
Contributor

Another example of a target without integrated assembler - https://rust-lang.zulipchat.com/#narrow/stream/182449-t-compiler.2Fhelp/topic/Report.20patch.20for.20a.20new.20OS

@cr1901 cr1901 mentioned this pull request Nov 23, 2021
Closed
@bjorn3 bjorn3 mentioned this pull request Nov 17, 2022
Closed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees

@alexcrichton alexcrichton

Labels
relnotes Marks issues that should be documented in the release notes of the next release. S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue.
Projects
None yet
Milestone
1.44
Development

Successfully merging this pull request may close these issues.