Auto merge of #83833 - jyn514:no-resolver, r=GuillaumeGomez
rustdoc: Store intra-doc links in Cache instead of on items directly
Items are first built after rustdoc creates the TyCtxt. To allow
resolving the links before the TyCtxt is built, the links can't be
stored on `clean::Item` directly.
Helps with https://github.com/rust-lang/rust/issues/83761. Opening this early because I think it might decrease memory usage.
Joshua Nelson [Sun, 4 Apr 2021 02:15:24 +0000 (22:15 -0400)]
Store links in Cache instead of on items directly
Items are first built after rustdoc creates the TyCtxt. To allow
resolving the links before the TyCtxt is built, the links can't be
stored on `clean::Item` directly.
Auto merge of #83934 - Dylan-DPC:rollup-nw5dadn, r=Dylan-DPC
Rollup of 7 pull requests
Successful merges:
- #82963 (Move `SharedContext` to `context.rs`)
- #83829 (rustc_target: Rely on defaults more in target specs)
- #83895 (Add listing of lints (eg via `-W help`) to rustdoc)
- #83902 (Update LLVM to support more wasm simd ops)
- #83903 (Fix typo in TokenStream documentation)
- #83910 (Update cargo)
- #83920 (Fix HashMap/HashSet LLDB pretty-printer after hashbrown 0.11.0)
Rollup merge of #83920 - ortem:fix-hashmap-lldb-pretty-printer-1.52, r=pnkfelix
Fix HashMap/HashSet LLDB pretty-printer after hashbrown 0.11.0
The pretty-printer was broken in https://github.com/rust-lang/rust/pull/77566 after updating hashbrown to 0.11.0.
Note that the corresponding GDB pretty-printer was updated properly.
Auto merge of #83821 - camelid:improve-thinvec, r=petrochenkov
Add `FromIterator` and `IntoIterator` impls for `ThinVec`
These should make using `ThinVec` feel much more like using `Vec`.
They will allow users of `Vec` to switch to `ThinVec` while continuing
to use `collect()`, `for` loops, and other parts of the iterator API.
I don't know if there were use cases before for using the iterator API
with `ThinVec`, but I would like to start using `ThinVec` in rustdoc,
and having it conform to the iterator API would make the transition
*a lot* easier.
I added a `FromIterator` impl, an `IntoIterator` impl that yields owned
elements, and `IntoIterator` impls that yield immutable or mutable
references to elements. I also added some unit tests for `ThinVec`.
Fix HashMap/HashSet LLDB pretty-printer after hashbrown 0.11.0
The pretty-printer was broken in https://github.com/rust-lang/rust/pull/77566
after updating hashbrown to 0.11.0.
Note that the corresponding GDB pretty-printer was updated properly.
Auto merge of #81641 - bjorn3:find_codegen_backend, r=davidtwco
Find codegen backends in more locations
* Search in the sysroot passed using `--sysroot` in addition to the default sysroot.
* Search for `librustc_codegen_$name.so` in addition to `librustc_codegen_$name-$release.so`.
This combined would allow putting `librustc_codegen_cranelift.so` in the right location of a sysroot passed using `--sysroot`.
Auto merge of #81234 - repnop:fn-alignment, r=lcnr
Allow specifying alignment for functions
Fixes #75072
This allows the user to specify alignment for functions, which can be useful for low level work where functions need to necessarily be aligned to a specific value.
I believe the error cases not covered in the match are caught earlier based on my testing so I had them just return `None`.
Add `FromIterator` and `IntoIterator` impls for `ThinVec`
These should make using `ThinVec` feel much more like using `Vec`.
They will allow users of `Vec` to switch to `ThinVec` while continuing
to use `collect()`, `for` loops, and other parts of the iterator API.
I don't know if there were use cases before for using the iterator API
with `ThinVec`, but I would like to start using `ThinVec` in rustdoc,
and having it conform to the iterator API would make the transition
*a lot* easier.
I added a `FromIterator` impl, an `IntoIterator` impl that yields owned
elements, and `IntoIterator` impls that yield immutable or mutable
references to elements. I also added some unit tests for `ThinVec`.
Auto merge of #83592 - nagisa:nagisa/dso_local, r=davidtwco
Set dso_local for hidden, private and local items
This should probably have no real effect in most cases, as e.g. `hidden`
visibility already implies `dso_local` (or at least LLVM IR does not
preserve the `dso_local` setting if the item is already `hidden`), but
it should fix `-Crelocation-model=static` and improve codegen in
executables.
Note that this PR does not exhaustively port the logic in [clang], only the
portion that is necessary to fix a regression from LLVM 12 that relates to
`-Crelocation_model=static`.
Auto merge of #83905 - JohnTitor:rollup-pa1la80, r=JohnTitor
Rollup of 8 pull requests
Successful merges:
- #83370 (Add `x.py setup tools` which enables `download-rustc` by default)
- #83489 (Properly suggest deref in else block)
- #83734 (Catch a bad placeholder type error for statics in `extern`s)
- #83814 (expand: Do not ICE when a legacy AST-based macro attribute produces and empty expression)
- #83835 (rustdoc: sort search index items for compression)
- #83849 (rustdoc: Cleanup handling of associated items for intra-doc links)
- #83881 (:arrow_up: rust-analyzer)
- #83885 (Document compiler/ with -Aprivate-intra-doc-links)
Rollup merge of #83849 - jyn514:intra-doc-cleanup, r=bugadani
rustdoc: Cleanup handling of associated items for intra-doc links
Helps with https://github.com/rust-lang/rust/issues/83761 (right now the uses of the resolver are all intermingled with uses of the tyctxt). Best reviewed one commit at a time.
r? ```@bugadani``` maybe? Feel free to reassign :)
Rollup merge of #83835 - notriddle:sort-index, r=ollie27
rustdoc: sort search index items for compression
This should not affect the appearance of the docs pages themselves.
This makes the pre-compressed search index smaller, thanks to the
empty-string path duplication format, and also the gzipped version,
by giving the algorithm more structure to work with.
Rollup merge of #83370 - jyn514:setup-tools, r=Mark-Simulacrum
Add `x.py setup tools` which enables `download-rustc` by default
Helps with https://github.com/rust-lang/rust/issues/81930. I know I said in that issue that I should fix that rebasing rebuilds bootstrap, but the compile time improvement is so good I think it's ok to leave that fix for later (I still plan to work on it). I think all the outright bugs have been fixed :)
This builds on https://github.com/rust-lang/rust/pull/83368 so I can set the option to `if-unchanged`.
Auto merge of #83152 - guswynn:jemallocator_part2, r=Mark-Simulacrum
Use tikv-jemallocator in rustc/rustdoc in addition to jemalloc-sys when enabled.
In https://github.com/rust-lang/rust/pull/81782 it was mentioned that one reason rustc may benefit from minimalloc is it doesn't use the `sdallocx` api from jemalloc.
Currently, on unix, rust uses jemalloc by importing its symbols to use them with the default, System (libc) global allocator.
This PR switches its global alloc to `tikv-jemallocator`, which correctly uses sized deallocation (https://docs.rs/tikv-jemallocator/0.4.1/src/tikv_jemallocator/lib.rs.html#121-126). `tikv-jemallocator`, as far as I can tell, is a more up-to-date set of bindings to jemalloc than `jemallocator`
The perf results of this pr are in large part due to the version upgrade of jemalloc, but sized deallocation has a non-trivial improvement, particularly to rustdoc.
This pr also includes changes to bootstrap to correctly pass the jemalloc feature through to the rustdoc build
On recursive types such as in the example given in https://github.com/rust-lang/rust/issues/83150, the current implementation of `LateBoundRegionNameCollector` has very long compilation runtimes. To prevent those we store the types visited in the `middle::ty::Visitor` implementation of `LateBoundRegionNameCollector` in a `SsoHashSet`.
Auto merge of #83890 - Dylan-DPC:rollup-9fqy3fe, r=Dylan-DPC
Rollup of 5 pull requests
Successful merges:
- #83368 (Add `download-rustc = "if-unchanged"`)
- #83705 (Give a better error when --theme is not a CSS file)
- #83853 (Disallow the use of high byte registes as operands on x86_64)
- #83877 (Remove unnecessary exceptions to the platform-specific code check)
- #83878 (Fix racing file access in tidy)
Rollup merge of #83878 - the8472:fix-concurrent-tidy-access, r=Mark-Simulacrum
Fix racing file access in tidy
That should fix the failure in https://github.com/rust-lang/rust/pull/83776#issuecomment-813311289
The file is only created for a brief moment during the bins checks in the source directories while other checks may also be visiting the same directory. By skipping it we avoid file not found errors.
Rollup merge of #83877 - jyn514:exceptions, r=Mark-Simulacrum
Remove unnecessary exceptions to the platform-specific code check
Some of these were just wrong, like src/librustc. Some looked outdated,
like std::f64. Not sure what was going on with the others - maybe this
check isn't as smart as it needs to be? But in the meantime it seems
silly to ignore the check if it will pass anyway.
Joshua Nelson [Sun, 4 Apr 2021 20:23:08 +0000 (16:23 -0400)]
Use more appropriate return type for `resolve_associated_item`
Previously, the types looked like this:
- None means this is not an associated item (but may be a variant field)
- Some(Err) means this is known to be an error. I think the only way that can happen is if it resolved and but you had your own anchor.
- Some(Ok(_, None)) was impossible.
Now, this returns a nested Option and does the error handling and
fiddling with the side channel in the caller. As a side-effect, it also
removes duplicate error handling.
This has one small change in behavior, which is that
`resolve_primitive_associated_item` now goes through `variant_field` if
it fails to resolve something. This is not ideal, but since it will be
quickly rejected anyway, I think the performance hit is worth the
cleanup.
This also fixes a bug where struct fields would forget to set the side
channel, adds a test for the bug, and ignores `private_intra_doc_links`
in rustc_resolve (since it's always documented with
--document-private-items).
Auto merge of #83880 - Dylan-DPC:rollup-hz9ph0e, r=Dylan-DPC
Rollup of 8 pull requests
Successful merges:
- #81922 (Let `#[allow(unstable_name_collisions)]` work for things other than function)
- #82483 (Use FromStr trait for number option parsing)
- #82739 (Use the beta compiler for building bootstrap tools when `download-rustc` is set)
- #83650 (Update Source Serif to release 4.004)
- #83826 (List trait impls before deref methods in doc's sidebar)
- #83831 (Add `#[inline]` to IpAddr methods)
- #83863 (Render destructured struct function param names as underscore)
- #83865 (Don't report disambiguator error if link would have been ignored)
Rollup merge of #83865 - camelid:disamb-err-fix, r=jyn514
Don't report disambiguator error if link would have been ignored
Fixes #83859.
This prevents us from warning on links such as `<hello@example.com>`.
Note that we still warn on links such as `<hello@localhost>` because
they have no dots in them. However, the links will still work, even
though a warning is reported.
Rollup merge of #83826 - slightlyoutofphase:rustdoc-sidebar-order-shuffle, r=jyn514
List trait impls before deref methods in doc's sidebar
This PR is acting directly on a suggestion made by ```````@jyn514``````` in #83133. I've tested the changes locally, and can confirm that it does in fact properly achieve what he thought it would. This PR also in turn closes #83133.
Rollup merge of #82739 - jyn514:separate-stage0-stage1, r=Mark-Simulacrum
Use the beta compiler for building bootstrap tools when `download-rustc` is set
## Motivation
This avoids having to rebuild bootstrap and tidy each time you rebase
over master. In particular, it makes rebasing and running `x.py fmt` on
each commit in a branch significantly faster. It also avoids having to
rebuild bootstrap after setting `download-rustc = true`.
## Implementation
Instead of extracting the CI artifacts directly to `stage0/`, extract
them to `ci-rustc/` instead. Continue to copy them to the proper
sysroots as necessary for all stages except stage 0.
This also requires `bootstrap.py` to download both stage0 and CI
artifacts and distinguish between the two when checking stamp files.
Note that since tools have to be built by the same compiler that built
`rustc-dev` and the standard library, the downloaded artifacts can't be
reused when building with the beta compiler. To make sure this is still
a good user experience, warn when building with the beta compiler, and
default to building with stage 2.
I tested this by rebasing this PR from edeee915b1c52f97411e57ef6b1a8bd46548a37a over 1c77a1fa3ca574f2a40056f64d498db8efe0d8a8 and confirming that only the bootstrap library itself had to be rebuilt, not any dependencies and not `tidy`. I also tested that a clean build with `x.py build` builds rustdoc exactly once and does no other work, and that `touch src/librustdoc/lib.rs && x.py build` works. `x.py check` still behaves as before (checks using the beta compiler, even if there are changes to `compiler/`).
Helps with https://github.com/rust-lang/rust/issues/81930.
Rollup merge of #81922 - magurotuna:issue81522, r=matthewjasper
Let `#[allow(unstable_name_collisions)]` work for things other than function
Fixes #81522
In addition to the report in #81522, currently `#[allow(unstable_name_collisions)]` doesn't suppress the corresponding diagnostics even if this attribute is appended to an expression statement or a let statement. It seems like this is because the wrong `HirId` is passed to `struct_span_lint_hir`.
It's fixed in this PR, and a regression test for it is also added.
this file is only created for a brief moment during the bins checks
in the source directories while other checks may also be visiting
that directory. skip processing it to avoid missing file errors
Joshua Nelson [Mon, 5 Apr 2021 10:02:29 +0000 (06:02 -0400)]
Remove unnecessary exceptions to the platform-specific code check
Some of these were just wrong, like src/librustc. Some looked outdated,
like std::f64. Not sure what was going on with the others - maybe this
check isn't as smart as it needs to be? But it the meantime it seems
silly to ignore the check if it will pass anyway.
Auto merge of #83858 - joshtriplett:unsafe-cell-always-inline, r=Mark-Simulacrum
Use `#[inline(always)]` on trivial UnsafeCell methods
UnsafeCell is the standard building block for shared mutable data
structures. UnsafeCell should add zero overhead compared to using raw
pointers directly.
Some reports suggest that debug builds, or even builds at opt-level 1,
may not always be inlining its methods. Mark the methods as
`#[inline(always)]`, since once inlined the methods should result in no
actual code other than field accesses.
Don't report disambiguator error if link would have been ignored
This prevents us from warning on links such as `<hello@example.com>`.
Note that we still warn on links such as `<hello@localhost>` because
they have no dots in them. However, the links will still work, even
though a warning is reported.
The attribute could always be replaced with `-C link-arg`, but cargo didn't provide a reasonable way to pass such flags to rustc.
Now cargo supports `cargo:rustc-link-arg*` directives in build scripts (https://doc.rust-lang.org/cargo/reference/unstable.html#extra-link-arg), so this attribute can be removed.
In the codegen and codegen-units test suites, the `//` comment markers
were kept in order not to affect any source locations. This is because
these tests cannot be automatically `--bless`ed.
Rollup merge of #83019 - eddyb:spirv-no-block-swap, r=nagisa
core: disable `ptr::swap_nonoverlapping_one`'s block optimization on SPIR-V.
SPIR-V primarily supports what it calls the "Logical addressing model" (and AFAIK for graphical shaders it's the only option), and what that implies is that there is no "memory" to uniformly address at some byte/word level, and that you can't really talk about values having a "raw representation" in terms of sequences of bytes. Therefore, the "block"-wise swapping optimization employed by `ptr::swap_nonoverlapping_one` (where a "block" is 32 bytes, currently), is fundamentally incompatible with SPIR-V "memory".
As such, [Rust-GPU](https://github.com/EmbarkStudios/rust-gpu/)'s `rustc_codegen_spirv` backend cannot currently allow the use of `ptr::swap_nonoverlapping_one` - but that comes at a great price, since it's the building block of `mem::{swap,replace}`, and those in turn are used by e.g. `Option::take` and `Range`'s `Iterator` implementation (the latter blocking the use of `for i in 0..n` loops).
There's 4 options I can see in terms of supporting `ptr::swap_nonoverlapping_one` in `rustc_codegen_spirv`:
* legalize the block-wise swap loop back into swapping whole values, for SPIR-V
* this is made borderline impossible by the fact that the size of the state "on the stack" is a block, and has to be expanded back to the appropriate size of the value being swapped, so in practice this would have to effectively pattern-match on the exact shape of the block-wise swapping algorithm, as a roundabout way of "patching `core::ptr` on the fly"
* (**this PR**) disable the block-wise swap optimization altogether when `#[cfg(target_arch = "spirv")`
* I've tested it and it does in fact allow compiling `for i in 0..n` loops, which was my primary motivation
* main downside IMO is the fact that `core` now acknowledges an out-of-tree backend
* as a counterpoint, any attempt to compile Rust to SPIR-V would run into this problem, one way or another
* only enable the block-wise swap optimization on targets where it's been empirically proven to be an improvement
* would avoid any surprises in terms of potentially-broken/inefficient codegen, in general
* however, it may be universally applicable (thanks to caches), even if the optimal block size could differ
* move low-level swapping into an intrinsic, where the backend can choose any optimization approach it wants
* this also has an impact on MIR optimizations (cc ``@rust-lang/wg-mir-opt)`` - which currently cannot hope to make sense of e.g. `Option::take` despite it being effectively `_0 = *_1;` `*_1 = None;` `return;`
* long-term this is my preferred approach, and I can start working on it if that's desired, but I wanted to confirm that this swapping optimization is the final blocker for [Rust-GPU](https://github.com/EmbarkStudios/rust-gpu/) supporting e.g. range `for` loops
Auto merge of #82907 - petrochenkov:dercache, r=Aaron1011
resolve/expand: Cache intermediate results of `#[derive]` expansion
Expansion function for `#[derive]` (`rustc_builtin_macros::derive::Expander::expand`) may return an indeterminate result, and therefore can be called multiple times.
Previously we parsed the `#[derive(Foo, Bar)]`'s input and tried to resolve `Foo` and `Bar` on every such call.
Now we maintain a cache `Resolver::derive_data` and take all the necessary data from it if it was computed previously.
So `Foo, Bar` is now parsed at most once, and `Foo` and `Bar` are successfully resolved at most once.
Joshua Nelson [Wed, 10 Feb 2021 05:33:17 +0000 (00:33 -0500)]
Use the beta compiler for building bootstrap tools when `download-rustc` is set
## Motivation
This avoids having to rebuild bootstrap and tidy each time you rebase
over master. In particular, it makes rebasing and running `x.py fmt` on
each commit in a branch significantly faster. It also avoids having to
rebuild bootstrap after setting `download-rustc = true`.
## Implementation
Instead of extracting the CI artifacts directly to `stage0/`, extract
them to `ci-rustc/` instead. Continue to copy them to the proper
sysroots as necessary for all stages except stage 0.
This also requires `bootstrap.py` to download both stage0 and CI
artifacts and distinguish between the two when checking stamp files.
Note that since tools have to be built by the same compiler that built
`rustc-dev` and the standard library, the downloaded artifacts can't be
reused when building with the beta compiler. To make sure this is still
a good user experience, warn when building with the beta compiler, and
default to building with stage 2.
Mark Rousskov [Sun, 4 Apr 2021 18:51:22 +0000 (14:51 -0400)]
Workaround increased cache clearing in Cargo
1.52 Cargo adds rust-lang/cargo#8640 which means that cargo will try to purge
the doc directory caches for us. In theory this may mean that we can jettison
the clear_if_dirty for rustdoc versioning entirely, but for now just workaround
the effects of this change in a less principled but more local way.