Tyler Mandry [Sun, 16 Aug 2020 21:59:23 +0000 (14:59 -0700)]
Rollup merge of #74346 - eddyb:reachable-defs, r=nikomatsakis
Use LocalDefId instead of HirId for reachable_set elements.
The only `HirId`s being tracked there that don't have matching `DefId`s are local variables, and that's an accident from #44316 (where I preserved the old behavior, even if nothing relied on reachability tracking local variables).
Tyler Mandry [Sun, 16 Aug 2020 21:59:21 +0000 (14:59 -0700)]
Rollup merge of #74314 - eddyb:closure-substs-direct, r=nikomatsakis
rustc_typeck: construct {Closure,Generator}Substs more directly.
We've previously not had a way to create `{Closure,Generator}Substs` other than instantiating all generics as inference variables and unifying the inference types (extracted using the regular `{Closure,Generator}Substs` accessors), with the actual types.
With this PR, those hacks, and assumptions about the order of closure/generator-specific components, are replaced with a simple API where the base `Substs` are combined with the additional information into a `{Closure,Generator}Substs`.
This might also be faster than relying inference, although probably not by much.
Tyler Mandry [Sun, 16 Aug 2020 21:59:20 +0000 (14:59 -0700)]
Rollup merge of #74204 - ayazhafiz:i/74120, r=eddyb
Don't visit foreign function bodies when lowering ast to hir
Previously the existence of bodies inside a foreign function block would
cause a panic in the hir `NodeCollector` during its collection of crate
bodies to compute a crate hash:
The collector walks the hir tree and creates a map of hir nodes, then
attaching bodies in the crate to their owner in the map. For a code like
```rust
extern "C" {
fn f() {
fn g() {}
}
}
```
The crate bodies include the body of the function `g`. But foreign
functions cannot have bodies, and while the parser AST permits a foreign
function to have a body, the hir doesn't. This means that the body of
`f` is not present in the hir, and so neither is `g`. So when the
`NodeCollector` finishes the walking the hir, it has no record of `g`,
cannot find an owner for the body of `g` it sees in the crate bodies,
and blows up.
Why do the crate bodies include the body of `g`? The AST walker has a
need a for walking function bodies, and FFIs share the same AST node as
functions in other contexts.
There are at least two options to fix this:
- Don't unwrap the map entry for an hir node in the `NodeCollector`
- Modifier the ast->hir lowering visitor to ignore foreign function
blocks
I don't think the first is preferrable, since we want to know when we
can't find a body for an hir node that we thought had one (dropping this
information may lead to an invalid hash). So this commit implements the
second option.
bors [Sun, 16 Aug 2020 18:48:26 +0000 (18:48 +0000)]
Auto merge of #73176 - LeSeulArtichaut:tyctxtat-err, r=eddyb
Add `TyCtxtAt::{ty_error, ty_error_with_message}`
~~Only e2d957d was added, the rest comes from #70551.~~
I was unsure where to put the implementation for those methods, please tell me if there is a better place for it.
Closes #72619, ~~blocked on #70551~~.
r? @eddyb cc @mark-i-m, maybe this should be part of #70551? If so feel free to cherry-pick or ask me to file a PR against your fork.
bors [Sun, 16 Aug 2020 16:58:13 +0000 (16:58 +0000)]
Auto merge of #75472 - Mark-Simulacrum:mangling-config, r=eddyb
Add option to use the new symbol mangling in rustc/std
I don't know if this causes problems in some cases -- maybe it should be on by default for at least rustc. I've never encountered problems with it other than tools not supporting it, though.
bors [Sun, 16 Aug 2020 06:26:38 +0000 (06:26 +0000)]
Auto merge of #75561 - kornelski:stringcol, r=Dylan-DPC
Doc: String isn't a collection
On forums one user was confused by this text, interpreting it as saying that `String` is a `Vec<char>` literally, rather than figuratively for the purpose of collect. I've reworded that paragraph.
bors [Sat, 15 Aug 2020 18:01:06 +0000 (18:01 +0000)]
Auto merge of #75560 - Mark-Simulacrum:rustc-docs, r=matthiaskrgr
Add rustc-docs as a component
Previously it was listed as a package but wasn't available in the component
lists in rustup, so wasn't actually installable.
rustc-docs is also only present for x86_64-unknown-linux-gnu. Eventually it'll
also be shipped for aarch64-gnu with current CI configuration, but that builder
isn't quite up and running yet.
We probably want to ship compiler docs for other platforms as well, though, but
this commit doesn't enable that quite yet. A future PR may do so by adding
--enable-compiler-docs to the relevant builders (but it would also need to
decide the set of builders which we'd ship on).
bors [Sat, 15 Aug 2020 15:51:55 +0000 (15:51 +0000)]
Auto merge of #75483 - mati865:mingw-lld-flags, r=petrochenkov
Add LLD flags for MinGW
Tested locally and this now works:
- `RUSTFLAGS="-Zlink-self-contained=yes -Clinker=rust-lld" cargo b`
- `RUSTFLAGS="-Zlink-self-contained=no -Clinker=rust-lld -Zpre-link-arg=-Ld:/msys64/mingw64/x86_64-w64-mingw32/lib -Zpre-link-arg=-Ld:/msys64/mingw64/lib/gcc/x86_64-w64-mingw32/10.2.0 -Zpre-link-arg=crt2.o" cargo b`
This is "harmless" part of the changes to make possible linking with bare LLD with windows-gnu target. More debatable changes should follow in next PRs soon.
The match branch simplification is applied when target blocks contain
statements that are either equal or perform a const bool assignment with
different values to the same place.
Previously, when constructing new statements, only statements from a
single block had been examined. This lead to a misoptimization when
statements are equal because the assign the *same* const bool value to
the same place.
Fix the issue by examining statements from both blocks when deciding on
replacement.
Additionally:
* Copy discriminant instead of moving it since it might be necessary to use its
value more than once.
* Optimize when switching on copy operand
bors [Sat, 15 Aug 2020 11:38:24 +0000 (11:38 +0000)]
Auto merge of #74576 - myfreeweb:freebsd-sanitizers, r=oli-obk
Add sanitizer support on FreeBSD
Restarting #47337. Everything is better now, no more weird llvm problems, well not everything:
Unfortunately, the sanitizers don't have proper support for versioned symbols (https://github.com/google/sanitizers/issues/628), so `libc`'s usage of `stat@FBSD_1.0` and so on explodes, e.g. in calling `std::fs::metadata`.
Building std (now easy thanks to cargo `-Zbuild-std`) and libc with `freebsd12/13` config via the `LIBC_CI=1` env variable is a good workaround…
Mark Rousskov [Sat, 15 Aug 2020 11:06:18 +0000 (07:06 -0400)]
Add rustc-docs as a component
Previously it was listed as a package but wasn't available in the component
lists in rustup, so wasn't actually installable.
rustc-docs is also only present for x86_64-unknown-linux-gnu. Eventually it'll
also be shipped for aarch64-gnu with current CI configuration, but that builder
isn't quite up and running yet.
We probably want to ship compiler docs for other platforms as well, though, but
this commit doesn't enable that quite yet. A future PR may do so by adding
--enable-compiler-docs to the relevant builders (but it would also need to
decide the set of builders which we'd ship on).
bors [Sat, 15 Aug 2020 09:46:22 +0000 (09:46 +0000)]
Auto merge of #75518 - davidtwco:issue-75326-polymorphization-symbol-mangling-v0-predicates, r=lcnr
polymorphize: `I` used if `T` used in `I: Foo<T>`
Fixes #75326.
This PR adjusts polymorphization's handling of predicates so that after ensuring that `T` is used in `I: Foo<T>` if `I` is used, it now ensures that `I` is used if `T` is used in `I: Foo<T>`. This is necessary to mark generic parameters that only exist in impl parameters as used - thereby avoiding symbol clashes when using the new mangling scheme.
With this PR, rustc will now fully bootstrap with polymorphization and the new symbol mangling scheme enabled - not all tests pass, but I'm not sure how much of that is the interaction of the two features, I'll be looking into that soon. All tests pass with only polymorphization enabled.
r? @lcnr (this isn't sufficiently complex that I need to add to eddy's review queue)
cc @eddyb
bors [Sat, 15 Aug 2020 07:55:58 +0000 (07:55 +0000)]
Auto merge of #75539 - ehuss:fix-crate-version-rustdoc-bootstrap, r=Mark-Simulacrum
Fix crate-version with rustdoc in bootstrap.
Cargo will now automatically use the `--crate-version` flag (see https://github.com/rust-lang/cargo/pull/8509). Cargo has special handling to avoid passing the flag if it is passed in via RUSTDOCFLAGS, but the `rustdoc` wrapper circumvents that check. This causes a problem because rustdoc will fail if the flag is passed in twice. Fix this by using RUSTDOCFLAGS.
This will be necessary when 1.47 is promoted to beta, but should be safe to do now.
bors [Sat, 15 Aug 2020 03:08:03 +0000 (03:08 +0000)]
Auto merge of #75549 - tmandry:rollup-sxjwa0w, r=tmandry
Rollup of 4 pull requests
Successful merges:
- #75376 (Set CMAKE_SYSTEM_NAME when cross-compiling)
- #75448 (merge `as_local_hir_id` with `local_def_id_to_hir_id`)
- #75513 (Recover gracefully from `struct` parse errors)
- #75545 (std/sys/unix/time: make it easier for LLVM to optimize `Instant` subtraction.)
Tyler Mandry [Sat, 15 Aug 2020 03:07:16 +0000 (20:07 -0700)]
Rollup merge of #75545 - eddyb:instant-sub-branchless, r=sfackler
std/sys/unix/time: make it easier for LLVM to optimize `Instant` subtraction.
This PR is the minimal change necessary to get LLVM to optimize `if self.t.tv_nsec >= other.t.tv_nsec` to branchless instructions (at least on x86_64), inspired by @m-ou-se's own attempts at optimizing `Instant` subtraction.
I stumbled over this by looking at the total number of instructions executed by `rustc -Z self-profile`, and found that after disabling ASLR, the largest source of non-determinism remaining was from this `if` taking one branch or the other, depending on the values involved.
The reason this code is even called so many times to make a difference, is that `measureme` (the `-Z self-profile` implementation) currently uses `Instant::elapsed` for its event timestamps (of which there can be millions).
I doubt it's critical to land this, although perhaps it could slightly improve some forms of benchmarking.
Tyler Mandry [Sat, 15 Aug 2020 03:07:13 +0000 (20:07 -0700)]
Rollup merge of #75513 - estebank:confused-parser, r=davidtwco
Recover gracefully from `struct` parse errors
Currently the parser tries to recover from finding a keyword where a field name was expected, but this causes extra knock down parse errors that are completely irrelevant. Instead, bail out early in the parsing of the field and consume the remaining tokens in the block. This can reduce output significantly.
_Improvements based on the narrative in https://fasterthanli.me/articles/i-am-a-java-csharp-c-or-cplusplus-dev-time-to-do-some-rust_
Tyler Mandry [Sat, 15 Aug 2020 03:07:11 +0000 (20:07 -0700)]
Rollup merge of #75448 - lcnr:rn-as_local_hir_id, r=davidtwco
merge `as_local_hir_id` with `local_def_id_to_hir_id`
`as_local_hir_id` was defined as just calling `local_def_id_to_hir_id` and I think that having two different ways to call the same method is somewhat confusing.
Don't really care about which of these 2 methods we want to keep.
Does this require an MCP, considering that these methods are fairly frequently used?
Tyler Mandry [Sat, 15 Aug 2020 03:07:10 +0000 (20:07 -0700)]
Rollup merge of #75376 - tmiasko:cmake-system-name, r=Mark-Simulacrum
Set CMAKE_SYSTEM_NAME when cross-compiling
Configure CMAKE_SYSTEM_NAME when cross-compiling in `configure_cmake`,
to tell CMake about target system. Previously this was done only for
LLVM step and now applies more generally to steps using cmake.
bors [Sat, 15 Aug 2020 00:45:13 +0000 (00:45 +0000)]
Auto merge of #73851 - matthewjasper:serialize-not-special, r=oli-obk
Remove most specialization use in serialization
Switching from specialization to min_specialization in the compiler made the unsoundness of how we used these traits pretty clear. This changes how the `Encodable` and `Decodable` traits work to be more friendly for types need a `TyCtxt` to deserialize.
The alternative design of having both `Encodable` and `TyEncodable` traits was considered, but doesn't really work because the following impls would conflict:
```
impl<E: Ecodable> TyEncodable for Encodable
impl<E: TyEcodable> TyEncodable for [E]
```
## How-to guide
- `Rustc(De|En)codable` is now spelled `Ty(De|En)coable` in `rustc_middle`, `Metadata(En|De)codable` in `rustc_metadata` where needed, and `(De|En)codable` everywhere else.
- Manual implementations of `(De|En)codable` shouldn't be much different.
- If you're adding a new interned type that needs to be en/decodable then the simplest thing way to handle this is:
- Have the type be a wrapper around a reference to the interned data (i.e. do what `ty::Predicate` does, and not what all of the other interned types do)
- Derive `Ty(En|De)codable` on the inner type
- Implement `Encodable<impl TyEncoder>` by forwarding to the inner type.
- Implement `Decodable<impl TyDecoder>` by decoding the inner type and then creating the wrapper around that (using the `tcx` from the decoder as needed).
cc @rust-lang/compiler for opinions on this change
r? @oli-obk
The match branch simplification is applied when target blocks contain
statements that are either equal or perform a const bool assignment with
different values to the same place.
Previously, when constructing new statements, only statements from a
single block had been examined. This lead to a misoptimization when
statements are equal because the assign the *same* const bool value to
the same place.
Fix the issue by examining statements from both blocks when deciding on
replacement.
bors [Fri, 14 Aug 2020 22:52:57 +0000 (22:52 +0000)]
Auto merge of #75538 - tmandry:rollup-i8miv8q, r=tmandry
Rollup of 17 pull requests
Successful merges:
- #73943 (Document the unsafe keyword)
- #74062 (deny(unsafe_op_in_unsafe_fn) in libstd/ffi/c_str.rs)
- #74185 (Remove liballoc unneeded explicit link)
- #74192 (Improve documentation on process::Child.std* fields)
- #74409 (Change Debug impl of SocketAddr and IpAddr to match their Display output)
- #75195 (BTreeMap: purge innocent use of into_kv_mut)
- #75214 (Use intra-doc links in `mem::manually_drop` & `mem::maybe_uninit`)
- #75432 (Switch to intra-doc links in `std::process`)
- #75482 (Clean up E0752 explanation)
- #75501 (Move to intra doc links in std::ffi)
- #75509 (Tweak suggestion for `this` -> `self`)
- #75511 (Do not emit E0228 when it is implied by E0106)
- #75515 (Bump std's libc version to 0.2.74)
- #75517 (Promotion and const interning comments)
- #75519 (BTreeMap: refactor splitpoint and move testing over to unit test)
- #75530 (Switch to intra-doc links in os/raw/*.md)
- #75531 (Migrate unit tests of btree collections to their native breeding ground)
David Wood [Fri, 14 Aug 2020 11:33:20 +0000 (12:33 +0100)]
polymorphize: `I` used if `T` used in `I: Foo<T>`
This commit adjusts polymorphization's handling of predicates so that
after ensuring that `T` is used in `I: Foo<T>` if `I` is used, it now
ensures that `I` is used if `T` is used in `I: Foo<T>`. This is
necessary to mark generic parameters that only exist in impl parameters
as used - thereby avoiding symbol clashes when using the new mangling
scheme.
Tyler Mandry [Fri, 14 Aug 2020 21:47:04 +0000 (14:47 -0700)]
Rollup merge of #75531 - ssomers:btree_tests_migration, r=Mark-Simulacrum
Migrate unit tests of btree collections to their native breeding ground
There's one BTreeSet test case that I couldn't easily convince to come along, maybe because it truly is an integration test. But leaving it in place would mean git wouldn't see the move so I also moved it to a new file.
Tyler Mandry [Fri, 14 Aug 2020 21:46:56 +0000 (14:46 -0700)]
Rollup merge of #75511 - estebank:elide-trait-object-lt-error, r=lcnr
Do not emit E0228 when it is implied by E0106
Emit E0288 (lifetime bound for trait object cannot be deduced) only on bare trait objects. When the trait object is in the form of `&dyn Trait`, E0106 (missing lifetime specifier) will have been emitted, making the former redundant.
Tyler Mandry [Fri, 14 Aug 2020 21:46:54 +0000 (14:46 -0700)]
Rollup merge of #75509 - estebank:coming-merrily-from-java-land, r=lcnr
Tweak suggestion for `this` -> `self`
* When referring to `this` in associated `fn`s always suggest `self`.
* Point at ident for `fn` lacking `self`
* Suggest adding `self` to assoc `fn`s when appropriate
_Improvements based on the narrative in https://fasterthanli.me/articles/i-am-a-java-csharp-c-or-cplusplus-dev-time-to-do-some-rust_
Tyler Mandry [Fri, 14 Aug 2020 21:46:45 +0000 (14:46 -0700)]
Rollup merge of #75195 - ssomers:btree_split_up_into_kv_mut, r=Mark-Simulacrum
BTreeMap: purge innocent use of into_kv_mut
Replace the use of `into_kv_mut` into more precise calls. This makes more sense if you know that the single remaining use of `into_kv_mut` is in fact evil and can be trialled in court (#75200) and sent to a correction facility (#73971).
No real performance difference reported (but functions that might benefit a tiny constant bit like `BTreeMap::get_mut` aren't benchmarked):
```
benchcmp old new --threshold 5
name old ns/iter new ns/iter diff ns/iter diff % speedup
btree::map::clone_fat_100 63,073 59,256 -3,817 -6.05% x 1.06
btree::map::iter_100 3,514 3,235 -279 -7.94% x 1.09
```
Tyler Mandry [Fri, 14 Aug 2020 21:46:44 +0000 (14:46 -0700)]
Rollup merge of #74409 - LukasKalbertodt:improve-debug-impl-of-socketaddr-ipaddr, r=Amanieu
Change Debug impl of SocketAddr and IpAddr to match their Display output
This has already been done for `SocketAddrV4`, `SocketAddrV6`, `IpAddrV4` and `IpAddrV6`. I don't see a point to keep the rather bad to read derived impl, especially so when pretty printing:
V4(
127.0.0.1
)
From the `Display`, one can easily and unambiguously see if it's V4 or V6. Two examples:
Luckily the docs explicitly state that `Debug` output is not stable and that it may be changed at any time.
Using `Display` as `Debug` is very convenient for configuration structs (e.g. for webservers) that often just have a `derive(Debug)` and are printed that way to the one starting the server.
Tyler Mandry [Fri, 14 Aug 2020 21:46:41 +0000 (14:46 -0700)]
Rollup merge of #74192 - xkr47:patch-1, r=Mark-Simulacrum
Improve documentation on process::Child.std* fields
As a relative beginner, it took a while for me to figure out I could just steal the references to avoid partially moving the child and thus retain ability to call functions on it (and store it in structs etc).
Matthew Jasper [Thu, 11 Jun 2020 14:49:57 +0000 (15:49 +0100)]
Rework `rustc_serialize`
- Move the type parameter from `encode` and `decode` methods to
the trait.
- Remove `UseSpecialized(En|De)codable` traits.
- Remove blanket impls for references.
- Add `RefDecodable` trait to allow deserializing to arena-allocated
references safely.
- Remove ability to (de)serialize HIR.
- Create proc-macros `(Ty)?(En|De)codable` to help implement these new
traits.
bors [Fri, 14 Aug 2020 16:27:00 +0000 (16:27 +0000)]
Auto merge of #75416 - richkadel:llvm-coverage-map-gen-5.3, r=richkadel
LLVM IR coverage encoding aligns closer to Clang's
I found some areas for improvement while attempting to debug the
SegFault issue when running rust programs compiled using MSVC, with
coverage instrumentation.
I discovered that LLVM's coverage writer was generating incomplete
function name variable names (that's not a typo: the name of the
variable that holds a function name).
The existing implementation used one-up numbers to distinguish
variables, and correcting the names did not fix the MSVC coverage bug,
but the fix in this PR makes the names and resulting LLVM IR easier to
follow and more consistent with Clang's implementation.
I also changed the way the `-Zinstrument-coverage` option is supported in
symbol_export.rs. The original implementation was incorrect, and the
corrected version matches the handling for `-Zprofile-generate`, as it
turns out.
(An argument could be made that maybe `-Zinstrument-coverage` should
automatically enable `-Cprofile-generate`. In fact, if
`-Cprofile-generate` is analagous to Clang's `-fprofile-generate`, as
some documentation implies, Clang always requires this flag for its
implementation of source-based code coverage. This would require a
little more validation, and if implemented, would probably require
updating some of the user-facing messages related to
`-Cprofile-generate` to not be so specific to the PGO use case.)
None of these changes fixed the MSVC coverage problems, but they should
still be welcome improvements.
Lastly, I added some additional FIXME comments in instrument_coverage.rs
describing issues I found with the generated LLVM IR that would be
resolved if the coverage instrumentation is injected with a `Statement`
instead of as a new `BasicBlock`. I describe seven advantages of this
change, but it requires some discussion before making a change like
this.
Jonas Berlin [Thu, 9 Jul 2020 18:18:06 +0000 (21:18 +0300)]
Improve documentation on process::Child.std* fields
As a relative beginner, it took a while for me to figure out I could just steal the references to avoid partially moving the child and thus retain ability to call functions on it (and store it in structs etc).
bors [Fri, 14 Aug 2020 12:00:59 +0000 (12:00 +0000)]
Auto merge of #74777 - ssomers:btree_cleanup_7, r=Mark-Simulacrum
Stop BTreeMap casts from reborrowing
Down in btree/node.rs, the interface and use of `cast_unchecked` look a bit shady. It's really just there for inverting `forget_type` which does not borrow. By borrowing we can't write the same `cast_unchecked` in the same way at the Handle level.
Rich Kadel [Thu, 6 Aug 2020 05:53:11 +0000 (22:53 -0700)]
LLVM IR coverage encoding aligns closer to Clang's
I found some areas for improvement while attempting to debug the
SegFault issue when running rust programs compiled using MSVC, with
coverage instrumentation.
I discovered that LLVM's coverage writer was generating incomplete
function name variable names (that's not a typo: the name of the
variable that holds a function name).
The existing implementation used one-up numbers to distinguish
variables, and correcting the names did not fix the MSVC coverage bug,
but the fix in this PR makes the names and resulting LLVM IR easier to
follow and more consistent with Clang's implementation.
I also changed the way the `-Zinstrument-coverage` option is supported
in symbol_export.rs. The original implementation was incorrect, and the
corrected version matches the handling for `-Zprofile-generate`, as it
turns out.
(An argument could be made that maybe `-Zinstrument-coverage` should
automatically enable `-Cprofile-generate`. In fact, if
`-Cprofile-generate` is analagous to Clang's `-fprofile-generate`, as
some documentation implies, Clang always requires this flag for its
implementation of source-based code coverage. This would require a
little more validation, and if implemented, would probably require
updating some of the user-facing messages related to
`-Cprofile-generate` to not be so specific to the PGO use case.)
None of these changes fixed the MSVC coverage problems, but they should
still be welcome improvements.
Lastly, I added some additional FIXME comments in instrument_coverage.rs
describing issues I found with the generated LLVM IR that would be
resolved if the coverage instrumentation is injected with a `Statement`
instead of as a new `BasicBlock`. I describe seven advantages of this
change, but it requires some discussion before making a change like
this.
bors [Fri, 14 Aug 2020 09:23:22 +0000 (09:23 +0000)]
Auto merge of #75503 - JulianKnodt:opt_opt, r=oli-obk
Clean up some mir transform passes
I noticed a few places where there were intermediates being created
in MIR optimization passes, so I removed them.
I also changed some `Some(..)` into just `..` and wrap `Some(..)` at the function end, doing early returns for `None`.
I was generally looking for some easy optimizations in theses passes, and hopefully these should improve the runtime of these passes by a tinnnyyyyy bit.