Wesley Wiser [Wed, 7 Nov 2018 03:31:09 +0000 (22:31 -0500)]
Don't inline virtual calls (take 2)
When I fixed the previous mis-optimizations, I didn't realize there were
actually two different places where we mutate `callsites` and both of
them should have the same behavior.
As a result, if a function was inlined and that function contained
virtual function calls, they were incorrectly being inlined. I also
added a test case which covers this.
bors [Fri, 9 Nov 2018 06:56:25 +0000 (06:56 +0000)]
Auto merge of #55803 - Mark-Simulacrum:rollup, r=Mark-Simulacrum
Rollup of 17 pull requests
Successful merges:
- #55576 (Clarify error message for -C opt-level)
- #55633 (Support memcpy/memmove with differing src/dst alignment)
- #55638 (Fix ICE in msg_span_from_free_region on ReEmpty)
- #55659 (rustc: Delete grouping logic from the musl target)
- #55719 (Sidestep link error from rustfix'ed code by using a *defined* static.)
- #55736 (Elide anon lifetimes in conflicting impl note)
- #55739 (Consume optimization fuel from the MIR inliner)
- #55742 (Avoid panic when matching function call)
- #55753 (borrow_set: remove a helper function and a clone it uses)
- #55755 (Improve creation of 3 IndexVecs)
- #55758 ([regression - rust2018]: unused_mut lint false positives on nightly)
- #55760 (Remove intermediate font specs)
- #55761 (mir: remove a hacky recursive helper function)
- #55774 (wasm32-unknown-emscripten expects the rust_eh_personality symbol)
- #55777 (Use `Lit` rather than `P<Lit>` in `ast::ExprKind`.)
- #55783 (Deprecate mpsc channel selection)
- #55788 (rustc: Request ansi colors if stderr isn't a tty)
Mark Rousskov [Fri, 9 Nov 2018 01:15:24 +0000 (18:15 -0700)]
Rollup merge of #55788 - alexcrichton:wincolors, r=petrochenkov
rustc: Request ansi colors if stderr isn't a tty
Currently Cargo will always capture the output of rustc meaning that
rustc is never hooked up to a tty. To retain colors Cargo uses the
`fwdansi` crate to ensure that ansi color codes are translated to
windows terminal methods (and ansi codes otherwise just go their natural
route on Unix).
Cargo passes `--color always` to rustc to ensure that using a pipe
doesn't trick it into not emitting colors at all. It turns out, however,
that `--color always` ends up still accidentally using the native shell
api on native windows shells.
The fix here is to instead pass `AlwaysAnsi` to `termcolor` instead of
`Always`, ensuring that when `--color always` is passed to rustc and its
output isn't a terminal, we're always generating ansi colors regardless
of the platform.
Mark Rousskov [Fri, 9 Nov 2018 01:15:17 +0000 (18:15 -0700)]
Rollup merge of #55774 - CryZe:patch-5, r=alexcrichton
wasm32-unknown-emscripten expects the rust_eh_personality symbol
The `wasm32-unknown-emscripten` expects the `rust_eh_personality` symbol to be there, but a cfg checking for `target_arch = "wasm32"` which was meant to remove the symbol from the `wasm32-unknown-unknown` target, didn't check for whether `emscripten` is targeted or not, so the symbol accidentally got filtered out there as well.
Mark Rousskov [Fri, 9 Nov 2018 01:15:12 +0000 (18:15 -0700)]
Rollup merge of #55760 - jonhoo:no-intermediate-fonts, r=GuillaumeGomez
Remove intermediate font specs
This is a (much) more constrained version of #54772 that also aims at improving the situation in #34681. It removes any font specifications that are not the "official" rustdoc font, and instead relies on the browser to provide the fallback font if the official on is not available. On Linux systems, this is particularly important, as fonts like Helvetica, Arial, and Times often look pretty bad since they're pulled from extracted MS fonts. A specification like `serif` or `sans-serif` lets the browser instead choose a good font.
Mark Rousskov [Fri, 9 Nov 2018 01:15:10 +0000 (18:15 -0700)]
Rollup merge of #55758 - davidtwco:issue-55344, r=pnkfelix
[regression - rust2018]: unused_mut lint false positives on nightly
Fixes #55344.
This commit filters out locals that have never been initialized for
consideration in the `unused_mut` lint.
This is intended to detect when the statement that would have
initialized the local was removed as unreachable code. In these cases,
we would not want to lint. This is the same behaviour as the AST borrow
checker.
This is achieved by taking advantage of an existing pass over the MIR
for the `unused_mut` lint and creating a set of those locals that were
never initialized.
Mark Rousskov [Fri, 9 Nov 2018 01:15:03 +0000 (18:15 -0700)]
Rollup merge of #55742 - F001:fix-55718, r=petrochenkov
Avoid panic when matching function call
Fix #55718
This bug is introduced by #53751. The original code checked `Def::AssociatedConst(..) | Def::Method(..)` before `pat_ty.no_bound_vars().expect("expected fn type")`. But somehow I exchanged the sequence carelessly. Sorry about that.
Mark Rousskov [Fri, 9 Nov 2018 01:14:56 +0000 (18:14 -0700)]
Rollup merge of #55719 - pnkfelix:issue-54388-sidestep-link-error-from-rustfixed-code, r=alexcrichton
Sidestep link error from rustfix'ed code by using a *defined* static.
As a drive-by, added `-g` to the compile-flags so that the test more
reliably fails to compile when the extern static in question is *not*
provided. (I.e. this is making the test more robust in the face of
potential future revisions.)
Mark Rousskov [Fri, 9 Nov 2018 01:14:53 +0000 (18:14 -0700)]
Rollup merge of #55659 - alexcrichton:musl-no-group, r=michaelwoerister
rustc: Delete grouping logic from the musl target
This commit deletes the injection of `-(` and `-)` options to the linker
for the musl targets. This actually causes problems today on nightly if
you execute:
you get a linker error about "cannot nest groups". This comes about
because rustc injects its own `--start-group` and `--end-group`
variables which clash with the outer `-(` and `-)` variables. It's not
entirely clear to me why this doesn't affect the musl target by default
(in `-C panic=unwind` mode).
The compiler's own injection of `--start-group` and `--end-group` should
solve the issues mentioned in the comment for injecting `-(` and `-)` as
well.
we now, instead of ICE'ing, get a diagnostic like:
```
error[E0700]: hidden type for `impl Trait` captures lifetime that does not appear in bounds
--> issue-55608.rs:3:16
|
3 | fn server() -> impl FilterBase2 {
| ^^^^^^^^^^^^^^^^
|
= note: hidden type `Map2<[closure@issue-55608.rs:4:36: 4:41]>` captures an empty lifetime
```
Mark Rousskov [Fri, 9 Nov 2018 01:14:49 +0000 (18:14 -0700)]
Rollup merge of #55633 - nikic:memcpy-align, r=nagisa
Support memcpy/memmove with differing src/dst alignment
If LLVM 7 is used, generate memcpy/memmove with differing src/dst alignment. I've added new FFI functions to construct these through the builder API, which is more convenient than dealing with differing intrinsic signatures depending on the LLVM version.
Alex Crichton [Thu, 8 Nov 2018 15:49:28 +0000 (07:49 -0800)]
rustc: Request ansi colors if stderr isn't a tty
Currently Cargo will always capture the output of rustc meaning that
rustc is never hooked up to a tty. To retain colors Cargo uses the
`fwdansi` crate to ensure that ansi color codes are translated to
windows terminal methods (and ansi codes otherwise just go their natural
route on Unix).
Cargo passes `--color always` to rustc to ensure that using a pipe
doesn't trick it into not emitting colors at all. It turns out, however,
that `--color always` ends up still accidentally using the native shell
api on native windows shells.
The fix here is to instead pass `AlwaysAnsi` to `termcolor` instead of
`Always`, ensuring that when `--color always` is passed to rustc and its
output isn't a terminal, we're always generating ansi colors regardless
of the platform.
wasm32-unknown-emscripten expects the rust_eh_personality symbol
The `wasm32-unknown-emscripten` expects the `rust_eh_personality` symbol to be there, but a cfg checking for `target_arch = "wasm32"` which was meant to remove the symbol from the `wasm32-unknown-unknown` target, didn't check for whether `emscripten` is targeted or not, so the symbol accidentally got filtered out there as well.
bors [Wed, 7 Nov 2018 17:14:34 +0000 (17:14 +0000)]
Auto merge of #55746 - kennytm:rollup, r=kennytm
Rollup of 14 pull requests
Successful merges:
- #55377 (Slight copy-editing for `std::cell::Cell` docs)
- #55441 (Remove unused re import in gdb_rust_pretty_printing)
- #55453 (Choose predicates without inference variables over those with them)
- #55495 (Don't print opt fuel messages to stdout because it breaks Rustbuild)
- #55501 (Make `process_obligations`' computation of `completed` optional.)
- #55510 (Fix feature gate only being checked on first repr attr.)
- #55609 (Run name-anon-globals after LTO passes as well)
- #55645 (do not print wrapping ranges like normal ranges in validity diagnostics)
- #55688 (Standardised names and location of ui issue tests)
- #55692 (-C remark: fix incorrect warning about requiring "--debuginfo" instead of "-C debuginfo=n")
- #55702 (Add `aarch64-pc-windows-msvc` to deployed targets)
- #55728 (Update lldb)
- #55730 (Use trait impl method span when type param mismatch is due to impl Trait)
- #55734 (refactor: use shorthand fields)
Jon Gjengset [Wed, 7 Nov 2018 16:02:08 +0000 (11:02 -0500)]
Remove intermediate font specs
This is a (much) more constrained version of #54772 that also aims at
improving the situation in #34681. It removes any font specifications
that are not the "official" rustdoc font, and instead relies on the
browser to provide the fallback font if the official on is not
available. On Linux systems, this is particularly important, as fonts
like Helvetica, Arial, and Times often look pretty bad since they're
pulled from extracted MS fonts. A specification like `serif` or
`sans-serif` lets the browser instead choose a good font.
David Wood [Wed, 7 Nov 2018 12:40:55 +0000 (13:40 +0100)]
Ignore never-initialized locals for `unused_mut`.
This commit filters out locals that have never been initialized for
consideration in the `unused_mut` lint.
This is intended to detect when the statement that would have
initialized the local was removed as unreachable code. In these cases,
we would not want to lint. This is the same behaviour as the AST borrow
checker.
This is achieved by taking advantage of an existing pass over the MIR
for the `unused_mut` lint and creating a set of those locals that were
never initialized.
Removed `#[rustc_error]` from tests that are all `// compile-pass`.
I also added `// skip-codegen` to each one, to address potential concerns
that this change would otherwise slow down our test suite spending time
generating code for files that are really just meant to be checks of
compiler diagnostics.
(However, I will say: My preference is to not use `// skip-codegen` if
one can avoid it. We can use all the testing of how we drive LLVM that
we can get...)
Replaced use of `#[rustc_error]` with `// skip-codegen`.
This test specifically notes that it does not want to invoke the
linker, due to the way it (IMO weakly) exercises the `#[link=...]`
attribute.
In any case, removing the the `#[rustc_error]` here uncovered an
"invalid windows subsystem" error that was previously not included in
the transcript of diagnostic output. So that's a step forward, (right?).
kennytm [Wed, 7 Nov 2018 10:01:54 +0000 (18:01 +0800)]
Rollup merge of #55609 - nikic:fix-51947, r=nagisa
Run name-anon-globals after LTO passes as well
If we're going to emit bitcode (through ThinLTOBuffer), then we need to ensure that anon globals are named. This was already done after optimization passes, but also has to happen after LTO passes, as we always emit the final result in a ThinLTO-compatible manner.
I added the test as `run-make`. The important bit is that we emit bitcode in some way (e.g. `--crate-type rlib` or `--emit=llvm-bc`). Please tell me if there is a better way to test for that.
kennytm [Wed, 7 Nov 2018 10:01:50 +0000 (18:01 +0800)]
Rollup merge of #55495 - wesleywiser:opt_fuel_rustbuild, r=nikomatsakis
Don't print opt fuel messages to stdout because it breaks Rustbuild
Rustbuild passes `--message-format json` to the compiler invocations
which causes JSON to be emitted on stdout. Printing optimization fuel
messages to stdout breaks the json and causes Rustbuild to fail.
Work around this by emitting optimization fuel related messages on
stderr instead.
kennytm [Wed, 7 Nov 2018 10:01:48 +0000 (18:01 +0800)]
Rollup merge of #55453 - Aaron1011:fix/rustdoc-lifetime, r=pnkfelix
Choose predicates without inference variables over those with them
Fixes #54705
When constructing synthetic auto trait impls, we may come across
multiple predicates involving the same type, trait, and substitutions.
Since we can only display one of these, we pick the one with the 'most
strict' lifetime paramters. This ensures that the impl we render the
user is actually valid (that is, a struct matching that impl will
actually implement the auto trait in question).
This commit exapnds the definition of 'more strict' to take into account
inference variables. We always choose a predicate without inference
variables over a predicate with inference variables.
bors [Wed, 7 Nov 2018 09:14:12 +0000 (09:14 +0000)]
Auto merge of #55549 - RalfJung:miri-visitor, r=oli-obk
Value visitors for miri
Generalize the traversal part of validation to a `ValueVisitor`.
~~This includes https://github.com/rust-lang/rust/pull/55316, [click here](https://github.com/RalfJung/rust/compare/retagging...RalfJung:miri-visitor) for just the new commits.~~
bors [Tue, 6 Nov 2018 18:58:31 +0000 (18:58 +0000)]
Auto merge of #53996 - sekineh:thumb-run, r=japaric
[CI] Run a `thumbv7m-none-eabi` binary using `qemu-system-arm` [IRR-2018-embedded]
## What's included?
- Run a `thumbv7m-none-eabi` binary using `qemu-system-arm`
- We are using `cortex-m-rt = "=0.5.4"` which does not use `proc_macro`.
(reason: stage2 build of rustc does not work well with `proc_macro` in `run-make` phase.)
- We are using GNU LD for now.
## Blocker
All resolved.
- ~[Waiting] `#[panic_handler]` is not available in stable.~
- [Merged] https://github.com/rust-lang/rust/pull/53619
- ~[Waiting] https://github.com/japaric/lm3s6965evb: does not compile on stable.~
- [OK] dependent crate ~`panic-abort`~ `panic-halt`: already moved to use `#[panic_handler]`.
## Update
`#[panic_handler]` will be stabilized in Rust 1.30.
Nikita Popov [Fri, 2 Nov 2018 12:22:48 +0000 (13:22 +0100)]
Run name-anon-globals after LTO passes as well
If we're going to emit bitcode (through ThinLTOBuffer), then we
need to ensure that anon globals are named. This was already done
after optimization passes, but also has to happen after LTO passes,
as we always emit the final result in a ThinLTO-compatible manner.
Sidestep link error from rustfix'ed code by using a *defined* static.
As a drive-by, added `-g` to the compile-flags so that the test more
reliably fails to compile when the extern static in question is *not*
provided. (I.e. this is making the test more robust in the face of
potential future revisions.)
kennytm [Tue, 6 Nov 2018 07:21:10 +0000 (15:21 +0800)]
Rollup merge of #55696 - davidtwco:issue-55675, r=pnkfelix
NLL Diagnostic Review 3: Missing errors for borrows of union fields
Fixes #55675.
This PR modifies a test to make it more robust (it also fixes indentation on a doc comment, but that's not the point of the PR). See the linked issue for details.
kennytm [Tue, 6 Nov 2018 07:21:01 +0000 (15:21 +0800)]
Rollup merge of #55601 - petrochenkov:featissue, r=pnkfelix
Fix tracking issue numbers for some unstable features
And also remove deprecated unstable `#[panic_implementation]` attribute that was superseded by stable `#[panic_handler]` and doesn't have an open tracking issue.
kennytm [Tue, 6 Nov 2018 07:20:59 +0000 (15:20 +0800)]
Rollup merge of #55597 - alexcrichton:thread-local-inner, r=KodrAus
std: Enable usage of `thread_local!` through imports
The `thread_local!` macro delegated to an internal macro but it didn't
do so in a macros-and-the-module-system compatible fashion, meaning if a
`#![no_std]` crate imported `std` and tried to use `thread_local!` it
would fail due to missing a lookup of an internal macro.
This commit switches the macro to instead use `$crate` to invoke other
macros, ensuring that it'll work when `thread_local!` is imported alone.
bors [Tue, 6 Nov 2018 04:04:33 +0000 (04:04 +0000)]
Auto merge of #55518 - alexcrichton:smaller-wasm, r=sfackler
std: Improve codegen size of accessing TLS
Some code in the TLS implementation in libstd stores `Some(val)` into an
`&mut Option<T>` (effectively) and then pulls out `&T`, but it currently
uses `.unwrap()` which can codegen into a panic even though it can never
panic. With sufficient optimizations enabled (like LTO) the compiler can
see through this but this commit helps it along in normal mode
(`--release` with Cargo by default) to avoid codegen'ing the panic path.
This ends up improving the optimized codegen on wasm by ensuring that a
call to panic pulling in more file size doesn't stick around.
bors [Tue, 6 Nov 2018 01:20:58 +0000 (01:20 +0000)]
Auto merge of #55106 - petrhosek:fuchsia-lld, r=alexcrichton
Use lld directly for Fuchsia target
Fuchsia already uses lld as the default linker, so there's no reason
to always invoke it through Clang, instead we can simply invoke lld
directly and pass the set of flags that matches Clang.
Petr Hosek [Tue, 16 Oct 2018 00:27:07 +0000 (17:27 -0700)]
Use lld directly for Fuchsia target
Fuchsia already uses lld as the default linker, so there's no reason
to always invoke it through Clang, instead we can simply invoke lld
directly and pass the set of flags that matches Clang.
bors [Mon, 5 Nov 2018 19:29:57 +0000 (19:29 +0000)]
Auto merge of #55410 - nagisa:atomic-align, r=pnkfelix
Correct alignment of atomic types and (re)add Atomic{I,U}128
This is a updated https://github.com/rust-lang/rust/pull/53514 to also make atomic types `repr(C)` as per comment in https://github.com/rust-lang/rust/pull/53514#issuecomment-431042767.
Use `// revisions` in the dropck-eyepatch tests instead of relying on compare-mode=nll.
NLL has increased precision in its analysis of drop order, and we want
the test annotations to deliberately reflect this by having fewer
ERROR annotations for NLL than for AST-borrowck. The best way to get
this effect is via `// revisions`.
As a drive-by, also added uses of all the borrows just to make it
clear that NLL isn't somehow sidestepping things by using shorter
borrows than you might have otherwise expected. (Of course, the added
uses do not make all that much difference since the relevant types all
declare `impl Drop` and thus those drops have implicit uses anyway.)