Mark Rousskov [Wed, 11 Aug 2021 17:17:21 +0000 (13:17 -0400)]
PGO for LLVM builds on x86_64-unknown-linux-gnu in CI
This shows up to 5% less instruction counts on multiple benchmarks, and up to
19% wins on the -j1 wall times for rustc self-compilation.
We can afford to spend the extra cycles building LLVM essentially once more for
the x86_64-unknown-linux-gnu CI build today. The builder finishes in around 50
minutes on average, and this adds just 10 more minutes. Given the sizeable
improvements in compiler performance, this is definitely worth it.
bors [Tue, 24 Aug 2021 17:41:24 +0000 (17:41 +0000)]
Auto merge of #87472 - inquisitivecrystal:stabilize-force-warn, r=Mark-Simulacrum
Stabilize and document `--force-warn`
This PR will stabilize and document the `--force-warn` command line option. It is currently a draft, pending an FCP.
I've taken the liberty of tidying up the lint level command line options a bit as part of this. The changes are quite minor and should only affect rustc's help output. I'm making them here because they're trivial and, in one case, necessary to unify the way `--force-warn` with the way the other options are displayed.
I also want to mention that `@rylev` has done a ton of work on moving this along and deserves most of the credit. I'm just the one who landed up writing this particular PR.
bors [Tue, 24 Aug 2021 12:13:37 +0000 (12:13 +0000)]
Auto merge of #87699 - ubamrein:use-iphone-deployment-target-for-llvm, r=petrochenkov
Allow specifying an deployment target version for all iOS llvm targets
Closes: https://github.com/rust-lang/rust/issues/79408
This pull requests adds the same procedure to define the iOS-version for the LLVM-target as was used for the simulator target and the desktop target.
This then closes the original problem mentioned in the above issue. The problem with incompatible bitcode remains, but is probably not easy fixable.
I realised that something is still not right. Try to fix that.
bors [Tue, 24 Aug 2021 03:58:22 +0000 (03:58 +0000)]
Auto merge of #87739 - Aaron1011:remove-used-attrs, r=wesleywiser
Remove `Session.used_attrs` and move logic to `CheckAttrVisitor`
Instead of updating global state to mark attributes as used,
we now explicitly emit a warning when an attribute is used in
an unsupported position. As a side effect, we are to emit more
detailed warning messages (instead of just a generic "unused" message).
`Session.check_name` is removed, since its only purpose was to mark
the attribute as used. All of the callers are modified to use
`Attribute.has_name`
Additionally, `AttributeType::AssumedUsed` is removed - an 'assumed
used' attribute is implemented by simply not performing any checks
in `CheckAttrVisitor` for a particular attribute.
We no longer emit unused attribute warnings for the `#[rustc_dummy]`
attribute - it's an internal attribute used for tests, so it doesn't
mark sense to treat it as 'unused'.
With this commit, a large source of global untracked state is removed.
bors [Tue, 24 Aug 2021 01:36:09 +0000 (01:36 +0000)]
Auto merge of #85556 - FabianWolff:issue-85071, r=estebank,jackh726
Warn about unreachable code following an expression with an uninhabited type
This pull request fixes #85071. The issue is that liveness analysis currently is "smarter" than reachability analysis when it comes to detecting uninhabited types: Unreachable code is detected during type checking, where full type information is not yet available. Therefore, the check for type inhabitedness is quite crude:
https://github.com/rust-lang/rust/blob/fc81ad22c453776de16acf9938976930cf8c9401/compiler/rustc_typeck/src/check/expr.rs#L202-L205
i.e. it only checks for `!`, but not other, non-trivially uninhabited types, such as empty enums, structs containing an uninhabited type, etc. By contrast, liveness analysis, which runs after type checking, can benefit from the more sophisticated `tcx.is_ty_uninhabited_from()`:
https://github.com/rust-lang/rust/blob/fc81ad22c453776de16acf9938976930cf8c9401/compiler/rustc_passes/src/liveness.rs#L981
https://github.com/rust-lang/rust/blob/fc81ad22c453776de16acf9938976930cf8c9401/compiler/rustc_passes/src/liveness.rs#L996
This can lead to confusing warnings when a variable is reported as unused, but the use of the variable is not reported as unreachable. For instance:
```rust
enum Foo {}
fn f() -> Foo {todo!()}
fn main() {
let x = f();
let _ = x;
}
```
currently leads to
```
warning: unused variable: `x`
--> t1.rs:5:9
|
5 | let x = f();
| ^ help: if this is intentional, prefix it with an underscore: `_x`
|
= note: `#[warn(unused_variables)]` on by default
warning: 1 warning emitted
```
which is confusing, because `x` _appears_ to be used in line 6. With my changes, I get:
```
warning: unreachable expression
--> t1.rs:6:13
|
5 | let x = f();
| --- any code following this expression is unreachable
6 | let _ = x;
| ^ unreachable expression
|
= note: `#[warn(unreachable_code)]` on by default
note: this expression has type `Foo`, which is uninhabited
--> t1.rs:5:13
|
5 | let x = f();
| ^^^
warning: unused variable: `x`
--> t1.rs:5:9
|
5 | let x = f();
| ^ help: if this is intentional, prefix it with an underscore: `_x`
|
= note: `#[warn(unused_variables)]` on by default
warning: 2 warnings emitted
```
My implementation is slightly inelegant because unreachable code warnings can now be issued in two different places (during type checking and during liveness analysis), but I think it is the solution with the least amount of unnecessary code duplication, given that the new warning integrates nicely with liveness analysis, where unreachable code is already implicitly detected for the purpose of finding unused variables.
bors [Mon, 23 Aug 2021 20:10:29 +0000 (20:10 +0000)]
Auto merge of #88265 - m-ou-se:rollup-soymv20, r=m-ou-se
Rollup of 6 pull requests
Successful merges:
- #87976 (Account for tabs when highlighting multiline code suggestions)
- #88174 (Clarify some wording in Rust 2021 lint docs)
- #88188 (Greatly improve limitation handling on parallel rustdoc GUI test run)
- #88230 (Fix typos “a”→“an”)
- #88232 (Add notes to macro-not-found diagnostics to point out how things with the same name were not a match.)
- #88259 (Do not mark `-Z thir-unsafeck` as unsound anymore)
Mara Bos [Mon, 23 Aug 2021 18:45:50 +0000 (20:45 +0200)]
Rollup merge of #88232 - m-ou-se:macro-name-imported-but-not-macro, r=estebank
Add notes to macro-not-found diagnostics to point out how things with the same name were not a match.
This adds notes like:
```
error: cannot find derive macro `Serialize` in this scope
--> $DIR/issue-88206.rs:22:10
|
LL | #[derive(Serialize)]
| ^^^^^^^^^
|
note: `Serialize` is imported here, but it is not a derive macro
--> $DIR/issue-88206.rs:17:11
|
LL | use hey::{Serialize, Deserialize};
| ^^^^^^^^^
```
Mara Bos [Mon, 23 Aug 2021 18:45:49 +0000 (20:45 +0200)]
Rollup merge of #88230 - steffahn:a_an, r=oli-obk
Fix typos “a”→“an”
Fix typos in comments; found using a regex to find some easy instance of incorrect usage of a vs. an.
While automation was used to find these, every change was checked manually.
Changes in submodules get separate PRs:
* https://github.com/rust-lang/stdarch/pull/1201
* https://github.com/rust-lang/cargo/pull/9821
* https://github.com/rust-lang/miri/pull/1874
* https://github.com/rust-lang/rls/pull/1746
* https://github.com/rust-analyzer/rust-analyzer/pull/9984
_folks @ rust-analyzer are fast at merging…_
* https://github.com/rust-analyzer/rust-analyzer/pull/9985
* https://github.com/rust-analyzer/rust-analyzer/pull/9987
* https://github.com/rust-analyzer/rust-analyzer/pull/9989
_For `clippy`, I don’t know if the changes should better better be moved to a PR to the original repo._
<hr>
This has some overlap with #88226, but neither is a strict superset of the other.
If you want multiple commits, I can split it up; in that case, make sure to suggest a criterion for splitting.
Even if `dso_local` were properly supported in this way on macOS, it seems
incorrect to add this annotation as liberally as we did. The `dso_local`
annotation is for symbols that ultimately end up in the same linkage unit, but
we were adding this annotation even for `static` values inside `extern` blocks
marked with `#[link(type="framework")]`, which should be considered dynamically
linked. Note that Clang likewise avoids emitting `dso_local` for `dllimport`
symbols:
For this symbol, the use of `dso_local` causes LLVM to emit a relocation of
type `X86_64_RELOC_SIGNED`, which is a 32-bit signed PC-relative offset. If the
binary is large enough, `0x0` might be out of range, and the link will fail.
Avoiding `dso_local` causes LLVM to use the GOT instead, emitting a relocation
of type `X86_64_RELOC_GOT_LOAD`, which will properly handle the large offset
and cause the link to succeed.
As a side note, the static relocation model is effectively deprecated for
security reasons on macOS, as it prohibits PIE. It's also completely
unsupported on Apple Silicon, so I don't think it's worth going to the effort
of properly supporting this model on that platform.
bors [Sun, 22 Aug 2021 20:23:37 +0000 (20:23 +0000)]
Auto merge of #88240 - GuillaumeGomez:rollup-wdom91m, r=GuillaumeGomez
Rollup of 7 pull requests
Successful merges:
- #86747 (Improve wording of the `drop_bounds` lint)
- #87166 (Show discriminant before overflow in diagnostic for duplicate values.)
- #88077 (Generate an iOS LLVM target with a specific version)
- #88164 (PassWrapper: adapt for LLVM 14 changes)
- #88211 (cleanup: `Span::new` -> `Span::with_lo`)
- #88229 (Suggest importing the right kind of macro.)
- #88238 (Stop tracking namespace in used_imports.)
Guillaume Gomez [Sun, 22 Aug 2021 18:52:54 +0000 (20:52 +0200)]
Rollup merge of #88211 - petrochenkov:withhilo, r=jyn514
cleanup: `Span::new` -> `Span::with_lo`
Extracted from https://github.com/rust-lang/rust/pull/84373 as suggested in https://github.com/rust-lang/rust/pull/84373#issuecomment-857773867.
It turned out less useful then I expected, but anyway.
r? `@cjgillot`
`@bors` rollup
Guillaume Gomez [Sun, 22 Aug 2021 18:52:53 +0000 (20:52 +0200)]
Rollup merge of #88164 - durin42:llvm-14-san-opts, r=nikic
PassWrapper: adapt for LLVM 14 changes
These API changes appear to have all taken place in
https://reviews.llvm.org/D105007, which moved HWAddressSanitizerPass and
AddressSanitizerPass to only accept their options type as a ctor
argument instead of the sequence of bools etc. This required a couple of
parameter additions, which I made match the default prior to the
mentioned upstream LLVM change.
This patch restores rustc to building (though not quite passing all
tests, I've mailed other patches for those issues) against LLVM HEAD.
Guillaume Gomez [Sun, 22 Aug 2021 18:52:52 +0000 (20:52 +0200)]
Rollup merge of #88077 - kit-981:feature/fix-minimum-os-version-in-header, r=petrochenkov
Generate an iOS LLVM target with a specific version
This commit adds the `LC_VERSION_MIN_IPHONEOS` load command to the Mach-O header generated for `aarch64-apple-ios` binaries. The operating system will look for this load command to determine the minimum supported operating system version and will not allow the binary to run if it's absent. This logic already exists for the simulator toolchain.
I've been using `otool` from a [cctools](https://github.com/tpoechtrager/cctools-port) toolchain to parse the header and validate that this change adds the required load command.
This change appears to be enough to build Rust binaries that can run on a jailbroken iPhone.
Guillaume Gomez [Sun, 22 Aug 2021 18:52:51 +0000 (20:52 +0200)]
Rollup merge of #87166 - de-vri-es:show-discriminant-before-overflow, r=jackh726
Show discriminant before overflow in diagnostic for duplicate values.
This PR adds the value before overflow for explicit discriminant values in the error for duplicate discriminant values.
I found it rather confusing to see only the overflowed value.
It only does this for literals, since overflows in const evaluated arithmetic are already a hard error.
This is my first PR to the compiler, so please let me know if the implementation can be improved :)
Guillaume Gomez [Sun, 22 Aug 2021 18:52:50 +0000 (20:52 +0200)]
Rollup merge of #86747 - FabianWolff:issue-86653, r=GuillaumeGomez
Improve wording of the `drop_bounds` lint
This PR addresses #86653. The issue is sort of a false positive of the `drop_bounds` lint, but I would argue that the best solution for #86653 is simply a rewording of the warning message and lint description, because even if the lint is _technically_ wrong, it still forces the programmer to think about what they are doing, and they can always use `#[allow(drop_bounds)]` if they think that they really need the `Drop` bound.
There are two issues with the current warning message and lint description:
- First, it says that `Drop` bounds are "useless", which is technically incorrect because they actually do have the effect of allowing you e.g. to call methods that also have a `Drop` bound on their generic arguments for some reason. I have changed the wording to emphasize not that the bound is "useless", but that it is most likely not what was intended.
- Second, it claims that `std::mem::needs_drop` detects whether a type has a destructor. But I think this is also technically wrong: The `Drop` bound says whether the type has a destructor or not, whereas `std::mem::needs_drop` also takes nested types with destructors into account, even if the top-level type does not itself have one (although I'm not 100% sure about the exact terminology here, i.e. whether the "drop glue" of the top-level type counts as a destructor or not).
bors [Sun, 22 Aug 2021 13:26:32 +0000 (13:26 +0000)]
Auto merge of #88163 - camsteffen:collapsible-match-fix, r=Manishearth
Fix clippy::collapsible_match with let expressions
This fixes rust-lang/rust-clippy#7575 which is a regression from #80357. I am fixing the bug here instead of in the clippy repo (if that's okay) because a) the regression has not been synced yet and b) I would like to land the fix on nightly asap.
The fix is basically to re-generalize `match` and `if let` for the lint implementation (they were split because `if let` no longer desugars to `match` in the HIR).
Also fixes rust-lang/rust-clippy#7586 and fixes rust-lang/rust-clippy#7591
cc `@rust-lang/clippy`
`@xFrednet` do you want to review this?
bors [Sun, 22 Aug 2021 10:44:38 +0000 (10:44 +0000)]
Auto merge of #88139 - lcnr:marker-trait-attr, r=nikomatsakis
marker_traits: require `EvaluatedToOk` during winnowing
closes #84955, while it doesn't really fix it in a way that makes me happy it should prevent the issue for now and this
test can't be reproduced anyways, so it doesn't make much sense to keep it open.
fixes #84917 as only one of the impls depends on regions, so we now drop the ambiguous one instead of the correct one.
cc https://rust-lang.zulipchat.com/#narrow/stream/144729-wg-traits/topic/winnowing.20soundly/near/247899832
bors [Sun, 22 Aug 2021 08:03:47 +0000 (08:03 +0000)]
Auto merge of #88122 - Seppel3210:master, r=dtolnay
Fix example in `Extend<(A, B)>` impl
After looking over the examples in my last PR (#85835) on doc.rust-lang.org/nightly I realized that the example didn't actually show what I wanted it to show 😅
So here's the better example
bors [Sun, 22 Aug 2021 05:19:48 +0000 (05:19 +0000)]
Auto merge of #85166 - mbhall88:file-prefix, r=dtolnay
add file_prefix method to std::path
This is an initial implementation of `std::path::Path::file_prefix`. It is effectively a "left" variant of the existing [`file_stem`](https://doc.rust-lang.org/std/path/struct.Path.html#method.file_stem) method. An illustration of the difference is
```rust
use std::path::Path;
let path = Path::new("foo.tar.gz");
assert_eq!(path.file_stem(), Some("foo.tar"));
assert_eq!(path.file_prefix(), Some("foo"));
```
In my own development, I generally find I almost always want the prefix, rather than the stem, so I thought it might be best to suggest it's addition to libstd.
Of course, as this is my first contribution, I expect there is probably more work that needs to be done. Additionally, if the libstd team feel this isn't appropriate then so be it.
There has been some [discussion about this on Zulip](https://rust-lang.zulipchat.com/#narrow/stream/219381-t-libs/topic/file_lstem/near/238076313) and a user there suggested I open a PR to see whether someone in the libstd team thinks it is worth pursuing.
bors [Sun, 22 Aug 2021 02:16:35 +0000 (02:16 +0000)]
Auto merge of #88217 - jackh726:rollup-3k74o2m, r=jackh726
Rollup of 13 pull requests
Successful merges:
- #87604 (CI: Verify commits in beta & stable are in upstream branches.)
- #88057 (Update RELEASES to clarify attribute macro values.)
- #88072 (Allow the iOS toolchain to be built on Linux)
- #88170 (Update release note for 1.55.0.)
- #88172 (Test that type alias impl trait happens in a submodule)
- #88179 (Mailmap entry for myself)
- #88182 (We meant to use a trait instead of lifetime here)
- #88183 (test TAIT in different positions)
- #88189 (Add TAIT struct test)
- #88192 (Use of impl trait in an impl as the value for an associated type in a dyn)
- #88194 (Test use of impl Trait in an impl as the value for an associated type in an impl trait)
- #88197 (Test tait use in a fn type)
- #88201 (Test that incomplete inference for TAITs fail)
Jack Huey [Sun, 22 Aug 2021 00:56:38 +0000 (20:56 -0400)]
Rollup merge of #88197 - spastorino:tait-test-fn-type, r=oli-obk
Test tait use in a fn type
r? `@oli-obk`
I thought this was going to work but doesn't, quickly checked with Niko and he told me that we ruled this out for now. I'm not exactly sure why and how but here we have a test with a FIXME :)
Jack Huey [Sun, 22 Aug 2021 00:56:29 +0000 (20:56 -0400)]
Rollup merge of #88072 - kit-981:feature/build-ios-toolchain-on-linux, r=Mark-Simulacrum
Allow the iOS toolchain to be built on Linux
The iOS toolchain can be built on Linux with minor changes. The compilation will invoke `xcrun` to find the path to the iPhone SDK but a fake `xcrun` executable can be used.
```
#!/bin/sh
echo "/path/to/sdk"
```
The iOS toolchain can then be built and linked with rustup.
It's possible to take this toolchain and compile an iOS executable with it. This requires the ld64 linker and an iOS SDK. The ld64 linker can be taken from [cctools](https://github.com/tpoechtrager/cctools-port). A project's .cargo/config can then be edited to use the linker for this target.
Jack Huey [Sun, 22 Aug 2021 00:56:28 +0000 (20:56 -0400)]
Rollup merge of #88057 - ehuss:releases-doc-macros, r=Mark-Simulacrum
Update RELEASES to clarify attribute macro values.
As noted in #87681, macros do not work with the `#[path]` attribute. Since the places where macros *can* be used is very limited, I have changed this to just focus on `#[doc]` which is the only attribute where this is really useful.
bors [Sat, 21 Aug 2021 23:35:54 +0000 (23:35 +0000)]
Auto merge of #88075 - Xuanwo:vec_deque_retain, r=dtolnay
Optimize unnecessary check in VecDeque::retain
This pr is highly inspired by https://github.com/rust-lang/rust/pull/88060 which shared the same idea: we can split the `for` loop into stages so that we can remove unnecessary checks like `del > 0`.
## Benchmarks
Before
```rust
test collections::vec_deque::tests::bench_retain_half_10000 ... bench: 290,125 ns/iter (+/- 8,717)
test collections::vec_deque::tests::bench_retain_odd_10000 ... bench: 291,588 ns/iter (+/- 9,621)
test collections::vec_deque::tests::bench_retain_whole_10000 ... bench: 287,426 ns/iter (+/- 9,009)
```
After
```rust
test collections::vec_deque::tests::bench_retain_half_10000 ... bench: 243,940 ns/iter (+/- 8,563)
test collections::vec_deque::tests::bench_retain_odd_10000 ... bench: 242,768 ns/iter (+/- 3,903)
test collections::vec_deque::tests::bench_retain_whole_10000 ... bench: 202,926 ns/iter (+/- 6,332)
```
Based on the current benchmark, this PR will improve the perf of `VecDeque::retain` by around 16%. For special cases, the improvement will be up to 30%.
bors [Sat, 21 Aug 2021 18:32:37 +0000 (18:32 +0000)]
Auto merge of #82776 - jyn514:extern-url-fallback, r=GuillaumeGomez
Give precedence to `html_root_url` over `--extern-html-root-url` by default, but add a way to opt-in to the previous behavior
## What is an HTML root url?
It tells rustdoc where it should link when documentation for a crate is
not available locally; for example, when a crate is a dependency of a
crate documented with `cargo doc --no-deps`.
## What is the difference between `html_root_url` and `--extern-html-root-url`?
Both of these tell rustdoc what the HTML root should be set to.
`doc(html_root_url)` is set by the crate author, while
`--extern-html-root-url` is set by the person documenting the crate.
These are often different. For example, docs.rs uses
`--extern-html-root-url https://docs.rs/crate-name/version` to ensure
all crates have documentation, even if `html_root_url` is not set.
Conversely, crates such as Rocket set `doc(html_root_url =
"https://api.rocket.rs")`, because they prefer users to view the
documentation on their own site.
Crates also set `html_root_url` to ensure they have
documentation when building locally when offline. This is unfortunate to
require, because it's more work from the library author. It also makes
it impossible to distinguish between crates that want to be viewed on a
different site (e.g. Rocket) and crates that just want documentation to
be visible offline at all (e.g. Tokio). I have authored a separate
change to the API guidelines to no longer recommend doing this:
rust-lang/api-guidelines#230.
## Why change the default?
In the past, docs.rs has been the main user of `--extern-html-root-url`.
However, it's useful for other projects as well. In particular, Cargo
wants to pass it by default when running `--no-deps`
(rust-lang/cargo#8296).
Unfortunately, for these other use cases, the priority order is
inverted. They want to give *precedence* to the URL the crate picks, and
only fall back to the `--extern-html-root` if no `html_root_url` is
present. That allows passing `--extern-html-root` unconditionally,
without having to parse the source code to see what attributes are
present.
For docs.rs, however, we still want to keep the old behavior, so that
all links on docs.rs stay on the site.
Aaron Hill [Thu, 29 Jul 2021 17:00:41 +0000 (12:00 -0500)]
Remove `Session.used_attrs` and move logic to `CheckAttrVisitor`
Instead of updating global state to mark attributes as used,
we now explicitly emit a warning when an attribute is used in
an unsupported position. As a side effect, we are to emit more
detailed warning messages (instead of just a generic "unused" message).
`Session.check_name` is removed, since its only purpose was to mark
the attribute as used. All of the callers are modified to use
`Attribute.has_name`
Additionally, `AttributeType::AssumedUsed` is removed - an 'assumed
used' attribute is implemented by simply not performing any checks
in `CheckAttrVisitor` for a particular attribute.
We no longer emit unused attribute warnings for the `#[rustc_dummy]`
attribute - it's an internal attribute used for tests, so it doesn't
mark sense to treat it as 'unused'.
With this commit, a large source of global untracked state is removed.
bors [Sat, 21 Aug 2021 15:51:50 +0000 (15:51 +0000)]
Auto merge of #88134 - rylev:force-warn-improvements, r=nikomatsakis
Force warn improvements
As part of stablization of the `--force-warn` option (#86516) I've made the following changes:
* Error when the `warnings` lint group is based to the `--force-warn` option
* Tests have been updated to make it easier to understand the semantics of `--force-warn`
bors [Sat, 21 Aug 2021 09:25:28 +0000 (09:25 +0000)]
Auto merge of #87570 - nikic:llvm-13, r=nagisa
Upgrade to LLVM 13
Work in progress update to LLVM 13. Main changes:
* InlineAsm diagnostics reported using SrcMgr diagnostic kind are now handled. Previously these used a separate diag handler.
* Codegen tests are updated for additional attributes.
* Some data layouts have changed.
* Switch `#[used]` attribute from `llvm.used` to `llvm.compiler.used` to avoid SHF_GNU_RETAIN flag introduced in https://reviews.llvm.org/D97448, which appears to trigger a bug in older versions of gold.
* Set `LLVM_INCLUDE_TESTS=OFF` to avoid Python 3.6 requirement.
Upstream issues:
* ~~https://bugs.llvm.org/show_bug.cgi?id=51210 (InlineAsm diagnostic reporting for module asm)~~ Fixed by https://github.com/llvm/llvm-project/commit/1558bb80c01b695ce12642527cbfccf16cf54ece.
* ~~https://bugs.llvm.org/show_bug.cgi?id=51476 (Miscompile on AArch64 due to incorrect comparison elimination)~~ Fixed by https://github.com/llvm/llvm-project/commit/81b106584f2baf33e09be2362c35c1bf2f6bfe94.
* https://bugs.llvm.org/show_bug.cgi?id=51207 (Can't set custom section flags anymore). Problematic change reverted in our fork, https://reviews.llvm.org/D107216 posted for upstream revert.
* https://bugs.llvm.org/show_bug.cgi?id=51211 (Regression in codegen for #83623). This is an optimization regression that we may likely have to eat for this release. The fix for #83623 was based on an incorrect premise, and this needs to be properly addressed in the MergeICmps pass.
bors [Sat, 21 Aug 2021 01:29:12 +0000 (01:29 +0000)]
Auto merge of #88149 - Mark-Simulacrum:prep-never-type, r=jackh726
Refactor fallback code to prepare for never type
This PR contains cherry-picks of some of `@nikomatsakis's` work from #79366, and shouldn't (AFAICT) represent any change in behavior. However, the refactoring is good regardless of the never type work being landed, and will reduce the size of those eventual PR(s) (and rebase pain).
I am not personally an expert on this code, and the commits are essentially 100% `@nikomatsakis's,` but they do seem reasonable to me by my understanding. Happy to edit with review, of course. Commits are best reviewed in sequence rather than all together.