Commit Graph

2644 Commits

Author SHA1 Message Date
Matthias Krüger
3ec21070e2
Rollup merge of #133041 - madsmtm:print-deployment-target-env-var, r=davidtwco
Print name of env var in `--print=deployment-target`

The deployment target environment variable is OS-specific, and if you're in a place where you're asking `rustc` for the deployment target, you're likely to also wanna know the name of the environment variable. I myself wanted this for some code I'm working on in bootstrap, for example.

Behaviour before this PR:
```console
$ rustc --print=deployment-target --target=aarch64-apple-darwin
deployment_target=11.0
$ rustc --print=deployment-target --target=aarch64-apple-visionos
deployment_target=1.0
```

Behaviour after this PR:
```console
$ rustc --print=deployment-target --target=aarch64-apple-darwin
MACOSX_DEPLOYMENT_TARGET=11.0
$ rustc --print=deployment-target --target=aarch64-apple-visionos
XROS_DEPLOYMENT_TARGET=1.0
```

My _belief_ is that this option is extremely rarely used in general, and a GitHub search for "rustc print deployment-target" seems to confirm this, it revealed only the following actual pieces of code using this:
- b292ef6934/src/build_context.rs (L1199-L1220)
- daab9244b0/src/lib.rs (L3422-L3426)

`maturin` does `.split('=').last()`, so it will continue to work after this change, but `cc v1.0.84` did `.strip_prefix("deployment_target=")` since [this PR](https://github.com/rust-lang/cc-rs/pull/848), so it would break. That's _probably_ fine though, it was broken in a lot of scenarios anyway, and [got](https://github.com/rust-lang/cc-rs/pull/901) [reverted](https://github.com/rust-lang/cc-rs/pull/943) in `v1.0.85`.

So while this is _technically_ a breaking change, I really doubt that anyone is going to observe it, so it's probably fine.

``@BlackHoleFox`` wdyt?

``@rustbot`` label O-apple
r? compiler
2024-12-03 07:48:31 +01:00
Matthias Krüger
ae6a7dba2a
Rollup merge of #132974 - madsmtm:linker-arguments-with-commas, r=petrochenkov
Properly pass linker arguments that contain commas

When linking with the system C compiler, we sometimes want to forward certain arguments unchanged to the linker. This can be done with `-Wl,arg1,arg2` or `-Xlinker arg1 -Xlinker arg2`. `-Wl` is used when possible, since it is more compact, but it does not support commas in the argument itself - in those cases, we need to use `-Xlinker`, and that is what this PR implements.

This also fixes using sanitizers on macOS with `-Clinker-flavor=ld`, as those were previously manually using `-Wl`/`-Xlinker` (probably since the support wasn't present in the `link_args` function).

Note that there has been [a previous PR for this](https://github.com/rust-lang/rust/pull/38798), but it only implemented this in certain cases when passing `-rpath`.

r? compiler
2024-12-01 14:30:07 +01:00
Ralf Jung
3029e09e2f
Rollup merge of #131698 - the8472:remove-set-discriminant-hack, r=RalfJung
use stores of the correct size to set discriminants

Resolves an old HACK /FIXME.

Note that I haven't worked much with codegen so I'm not sure if I'm using the functions correctly and I was surprised seeing out-of-range values being fed into `const_uint_big` but apparently they're wrapped implicitly? By making it explicit we can pass in-range values instead.
2024-11-30 19:24:40 +01:00
The 8472
26d7b5da99 use stores of the correct size to set discriminants 2024-11-30 18:33:08 +01:00
bors
d53f0b1d8e Auto merge of #123244 - Mark-Simulacrum:share-inline-never-generics, r=saethlin
Enable -Zshare-generics for inline(never) functions

This avoids inlining cross-crate generic items when possible that are
already marked inline(never), implying that the author is not intending
for the function to be inlined by callers. As such, having a local copy
may make it easier for LLVM to optimize but mostly just adds to binary
bloat and codegen time. In practice our benchmarks indicate this is
indeed a win for larger compilations, where the extra cost in dynamic
linking to these symbols is diminished compared to the advantages in
fewer copies that need optimizing in each binary.

It might also make sense it expand this with other heuristics (e.g.,
`#[cold]`) in the future, but this seems like a good starting point.

FWIW, I expect that doing cleanup in where we make the decision
what should/shouldn't be shared is also a good idea. Way too
much code needed to be tweaked to check this. But I'm hoping
to leave that for a follow-up PR rather than blocking this on it.
2024-11-28 21:44:34 +00:00
Mark Rousskov
4a216a25d1 Share inline(never) generics across crates
This reduces code sizes and better respects programmer intent when
marking inline(never). Previously such a marking was essentially ignored
for generic functions, as we'd still inline them in remote crates.
2024-11-28 13:43:05 -05:00
Monadic Cat
ca55eeeaf3
use intra-doc links for CodegenBackend::link 2024-11-27 18:42:14 -06:00
Monadic Cat
52684a4c52
update comment (codegen_backend -> codegen_crate)
use intra-doc links so there'll be a doc gen fail next time this becomes
wrong
2024-11-27 18:26:08 -06:00
许杰友 Jieyou Xu (Joe)
c50e19b69c
Rollup merge of #133384 - RalfJung:vector-abi-check-tests, r=jieyouxu
add a test for target-feature-ABI warnings in closures and when calling extern fn

Also update the comment regarding the inheritance of target features into closures, to make it more clear that we really shouldn't do this right now.
2024-11-25 00:39:06 +08:00
许杰友 Jieyou Xu (Joe)
8d20d71256
Rollup merge of #133297 - DianQK:embed-bitcode-ios, r=nikic
Remove legacy bitcode for iOS

Follow #117364.
2024-11-25 00:39:05 +08:00
Ralf Jung
b9a0e57b0c add a test for target-feature-ABI warnings in closures 2024-11-24 09:20:10 +01:00
DianQK
7cc5feea4d
Remove forces_embed_bitcode 2024-11-24 15:51:47 +08:00
Mads Marquart
6bbf832cf2 Remove unnecessary 0 link args optimization 2024-11-24 01:23:19 +01:00
Mads Marquart
cb6f8fa422 Support rpath with -Clinker-flavor=ld
Using `cc_args` panics when using `-Clinker-flavor=ld`, because the
arguments are in a form tailored for `-Clinker-flavor=gcc`.

So instead, we use `link_args` and let that wrap the arguments with the
appropriate `-Wl` or `-Xlinker` when needed.
2024-11-24 01:20:29 +01:00
Mads Marquart
45791dde8e Support linker arguments that contain commas 2024-11-24 01:20:28 +01:00
许杰友 Jieyou Xu (Joe)
92f5144c9c
Rollup merge of #133217 - xingxue-ibm:fix-strip, r=compiler-errors
[AIX] Add option -X32_64 to the "strip" command

The AIX `strip` utility requires option `-X` to specify the object mode. This patch adds the `-X32_64` option to the `strip` command so that it can handle both 32-bit and 64-bit objects. The parameter `option` of function `strip_symbols_with_external_utility`, previously a single string, has been changed to `options`, an array of string slices, to accommodate multiple `strip` options.
2024-11-23 20:50:14 +08:00
Mads Marquart
431c500c60 Print env var in --print=deployment-target
The deployment target environment variable is OS-specific, and if you're
in a place where you're asking `rustc` for the deployment target, you're
likely to also wanna know the environment variable.
2024-11-22 20:46:05 +01:00
bors
5d3c6ee9b3 Auto merge of #132362 - mustartt:aix-dylib-detection, r=jieyouxu
[AIX] change system dynamic library format

Historically on AIX, almost all dynamic libraries are distributed in `.a` Big Archive Format which can consists of both static and shared objects in the same archive (e.g. `libc++abi.a(libc++abi.so.1)`). During the initial porting process, the dynamic libraries are kept as `.a` to simplify the migration, but semantically having an XCOFF object under the archive extension is wrong. For crate type `cdylib` we want to be able to distribute the libraries as archives as well.

We are migrating to archives with the following format:
```
$ ar -t lib<name>.a
lib<name>.so
```
where each archive contains a single member that is a shared XCOFF object that can be loaded.
2024-11-21 21:36:47 +00:00
Xing Xue
02f51ec2bd Change to pass "strip" options in an array of string slices and add option "-X32_64" for AIX. 2024-11-21 12:16:06 -05:00
Henry Jiang
0db9059726 aix: fix archive format
fmt

fix cfg for windows

remove unused imports

address comments

update libc to 0.2.164

fmt

remove unused imports
2024-11-21 10:33:07 -05:00
bors
2d0ea7956c Auto merge of #133261 - matthiaskrgr:rollup-ekui4we, r=matthiaskrgr
Rollup of 6 pull requests

Successful merges:

 - #129838 (uefi: process: Add args support)
 - #130800 (Mark `get_mut` and `set_position` in `std::io::Cursor` as const.)
 - #132708 (Point at `const` definition when used instead of a binding in a `let` statement)
 - #133226 (Make `PointerLike` opt-in instead of built-in)
 - #133244 (Account for `wasm32v1-none` when exporting TLS symbols)
 - #133257 (Add `UnordMap::clear` method)

r? `@ghost`
`@rustbot` modify labels: rollup
2024-11-20 21:58:38 +00:00
daxpedda
f37d021d6c
Account for wasm32v1-none when exporting TLS symbols 2024-11-20 14:02:25 +01:00
Ding Xiang Fei
297b618944
reduce false positives of tail-expr-drop-order from consumed values
take 2

open up coroutines

tweak the wordings

the lint works up until 2021

We were missing one case, for ADTs, which was
causing `Result` to yield incorrect results.

only include field spans with significant types

deduplicate and eliminate field spans

switch to emit spans to impl Drops

Co-authored-by: Niko Matsakis <nikomat@amazon.com>

collect drops instead of taking liveness diff

apply some suggestions and add explantory notes

small fix on the cache

let the query recurse through coroutine

new suggestion format with extracted variable name

fine-tune the drop span and messages

bugfix on runtime borrows

tweak message wording

filter out ecosystem types earlier

apply suggestions

clippy

check lint level at session level

further restrict applicability of the lint

translate bid into nop for stable mir

detect cycle in type structure
2024-11-20 20:53:11 +08:00
bors
70e814bd9e Auto merge of #133212 - lcnr:questionable-uwu, r=compiler-errors
continue `ParamEnv` to `TypingEnv` transition

cc #132279

r? `@compiler-errors`
2024-11-20 06:22:01 +00:00
lcnr
002efeb72a additional TypingEnv cleanups 2024-11-19 21:36:23 +01:00
lcnr
7a90e84f4d InterpCx store TypingEnv instead of a ParamEnv 2024-11-19 21:36:23 +01:00
lcnr
948cec0fad move fn is_item_raw to TypingEnv 2024-11-19 18:06:20 +01:00
Kyle Huey
f5b023bd9c When the required discriminator value exceeds LLVM's limits, drop the debug info for the function instead of panicking.
The maximum discriminator value LLVM can currently encode is 2^12. If macro use
results in more than 2^12 calls to the same function attributed to the same
callsite, and those calls are MIR-inlined, we will require more than the maximum
discriminator value to completely represent the debug information. Once we reach
that point drop the debug info instead.
2024-11-19 05:19:09 -08:00
lcnr
9cba14b95b use TypingEnv when no infcx is available
the behavior of the type system not only depends on the current
assumptions, but also the currentnphase of the compiler. This is
mostly necessary as we need to decide whether and how to reveal
opaque types. We track this via the `TypingMode`.
2024-11-18 10:38:56 +01:00
Jiri Bobek
777003ae9f Likely unlikely fix 2024-11-17 21:49:10 +01:00
Piotr Osiewicz
42e71bb8ea rustc_metadata: Preprocess search paths for better performance
Over in Zed we've noticed that loading crates for a large-ish workspace can take almost 200ms. We've pinned it down to how rustc searches for paths, as it performs a linear search over the list of candidate paths. In our case the candidate list had about 20k entries which we had to iterate over for each dependency being loaded.

This commit introduces a simple FilesIndex that's just a sorted Vec under the hood. Since crates are looked up by both prefix and suffix, we perform a range search on said Vec (which constraints the search space based on prefix) and follow up with a linear scan of entries with matching suffixes.
FilesIndex is also pre-filtered before any queries are performed using available target information; query prefixes/sufixes are based on the target we are compiling for, so we can remove entries that can never match up front.

Overall, this commit brings down build time for us in dev scenarios by about 6%.
100ms might not seem like much, but this is a constant cost that each of our workspace crates has to pay, even when said crate is miniscule.
2024-11-15 10:35:33 +01:00
Matthias Krüger
bd79fe7a94
Rollup merge of #132702 - 1c3t3a:issue-132615, r=rcvalle
CFI: Append debug location to CFI blocks

Currently we're not appending debug locations to the inserted CFI blocks. This shows up in #132615 and #100783. This change fixes that by passing down the debug location to the CFI type-test generation and appending it to the blocks.

Credits also belong to `@jakos-sec` who worked with me on this.
2024-11-12 23:26:41 +01:00
Matthias Krüger
35225d61f4
Rollup merge of #132820 - bjorn3:default_backend_link_impl, r=jieyouxu
Add a default implementation for CodegenBackend::link

As a side effect this should add raw-dylib support to cg_gcc as the default ArchiveBuilderBuilder that is used implements create_dll_import_lib. I haven't tested if the raw-dylib support actually works however.
2024-11-11 21:58:32 +01:00
Bastian Kersting
c2102259a0 CFI: Append debug location to CFI blocks 2024-11-11 09:17:43 +00:00
bjorn3
0a619dbc5d Pass owned CodegenResults to link_binary
After link_binary the temporary files referenced by CodegenResults are
deleted, so calling link_binary again with the same CodegenResults
should not be allowed.
2024-11-09 21:22:00 +00:00
bjorn3
cb44c0c8b6 Add a default implementation for CodegenBackend::link
As a side effect this should add raw-dylib support to cg_gcc as the
default ArchiveBuilderBuilder that is used implements
create_dll_import_lib. I haven't tested if the raw-dylib support
actually works however.
2024-11-09 20:42:56 +00:00
Matthias Krüger
9b47807b13
Rollup merge of #132774 - bjorn3:cranelift_lld, r=lqd
Use lld with non-LLVM backends

On arm64, Cranelift used to produce object files that don't work with lld. This has since been fixed. The GCC backend should always produce object files that work with lld unless lld for whatever reason drops GCC support. Most of the other more niche backends don't use cg_ssa's linker code at all. If they do and don't work with lld, they can always disable lld usage using a cli argument.

 Without this commit using cg_clif is by default in a non-trivial amount of cases a perf regression on Linux due to ld.bfd being a fair bit slower than lld. It is possible to explicitly enable it without this commit, but most users are unlikely to do this.
2024-11-09 19:16:45 +01:00
Matthias Krüger
b9d4ef16c9
Rollup merge of #132552 - taiki-e:sparc-target-feature, r=workingjubilee
Add v9, v8plus, and leoncasa target feature to sparc and use v8plus in create_object_file

This adds the following three unstable target features:

- `v9`: SPARC-V9 instructions ([LLVM definition][sparc-v9])
  - Relevant to https://github.com/rust-lang/rust/pull/131222#issuecomment-2453310963
  - Relevant to https://github.com/rust-lang/rust/pull/132472#discussion_r1832606081
  - This is also needed to implement https://github.com/taiki-e/atomic-maybe-uninit/pull/31 (depends on inline assembly support) more robustly.
- `v8plus`: SPARC-V8+ ABI ([LLVM definition][sparc-v8plus])
  - This is added in LLVM 20. In LLVM 19 and older, it is emulated to work the same way as LLVM in each LLVM version.
  - See https://github.com/rust-lang/rust/issues/132585#issuecomment-2453926257 for more.
- `leoncasa`: CASA instruction[^1] of LEON3 and LEON4 processors ([LLVM definition][sparc-leoncasa], LLVM feature name: `hasleoncasa`)
  - This is needed to implement https://github.com/taiki-e/atomic-maybe-uninit/pull/31 (depends on inline assembly support) more robustly.

[^1]: Atomic CAS instruction

[sparc-v9]: f5e4ffaa49/llvm/lib/Target/Sparc/Sparc.td (L37-L39)
[sparc-v8plus]: f5e4ffaa49/llvm/lib/Target/Sparc/Sparc.td (L37-L39)
[sparc-leoncasa]: https://github.com/llvm/llvm-project/blob/llvmorg-19.1.0/llvm/lib/Target/Sparc/LeonFeatures.td#L32-L37
2024-11-09 10:52:03 +01:00
Taiki Endo
c059eb7750 Add v8plus target feature to sparc and use it in create_object_file 2024-11-09 03:22:09 +09:00
bjorn3
6ffab47e55 Use lld with non-LLVM backends
On arm64, Cranelift used to produce object files that don't work with
lld. This has since been fixed. The GCC backend should always produce
object files that work with lld unless lld for whatever reason drops GCC
support. Most of the other more niche backends don't use cg_ssa's linker
code at all. If they do and don't work with lld, they can always disable
lld usage using a cli argument.

Without this commit using cg_clif is by default in a non-trivial amount
of cases a perf regression on Linux due to ld.bfd being a fair bit
slower than lld. It is possible to explicitly enable it without this
commit, but most users are unlikely to do this.
2024-11-08 15:20:20 +01:00
Jubilee
60e8ab6ba8
Rollup merge of #130586 - dpaoliello:fixrawdylib, r=wesleywiser
Set "symbol name" in raw-dylib import libraries to the decorated name

`windows-rs` received a bug report that mixing raw-dylib generated and the Windows SDK import libraries was causing linker failures: <https://github.com/microsoft/windows-rs/issues/3285>

The root cause turned out to be #124958, that is we are not including the decorated name in the import library and so the import name type is also not being correctly set.

This change modifies the generation of import libraries to set the "symbol name" to the fully decorated name and correctly marks the import as being data vs function.

Note that this also required some changes to how the symbol is named within Rust: for MSVC we now need to use the decorated name but for MinGW we still need to use partially decorated (or undecorated) name.

Fixes #124958

Passing i686 MSVC and MinGW build: <https://github.com/rust-lang/rust/actions/runs/11000433888?pr=130586>

r? `@ChrisDenton`
2024-11-07 18:48:20 -08:00
clubby789
b480f0f224 Remove unused intercrate dependencies 2024-11-07 14:17:16 +00:00
Matthias Krüger
8dee3e978a
Rollup merge of #131405 - davidtwco:hardcoded-strip-macos, r=jieyouxu,albertlarsan68
bootstrap/codegen_ssa: ship llvm-strip and use it for -Cstrip

Fixes #131206.

- Includes `llvm-strip` (a symlink to `llvm-objcopy`) in the compiler dist artifact so that it can be used for `-Cstrip` instead of the system tooling.
- Uses `llvm-strip` instead of `/usr/bin/strip` for macOS. macOS needs a specific linker and the system one is preferred, hence #130781 but that doesn't work when cross-compiling, so use the `llvm-strip` utility instead.

cc #123151
2024-11-05 23:43:56 +01:00
bors
e8c698bb3b Auto merge of #129884 - RalfJung:forbidden-target-features, r=workingjubilee
mark some target features as 'forbidden' so they cannot be (un)set with -Ctarget-feature

The context for this is https://github.com/rust-lang/rust/issues/116344: some target features change the way floats are passed between functions. Changing those target features is unsound as code compiled for the same target may now use different ABIs.

So this introduces a new concept of "forbidden" target features (on top of the existing "stable " and "unstable" categories), and makes it a hard error to (un)set such a target feature. For now, the x86 and ARM feature `soft-float` is on that list. We'll have to make some effort to collect more relevant features, and similar features from other targets, but that can happen after the basic infrastructure for this landed. (These features are being collected in https://github.com/rust-lang/rust/issues/131799.)

I've made this a warning for now to give people some time to speak up if this would break something.

MCP: https://github.com/rust-lang/compiler-team/issues/780
2024-11-05 16:25:45 +00:00
David Wood
f745467cd9
codegen_ssa: use llvm-objcopy for macOS strip 2024-11-05 11:49:37 +00:00
bors
096277e989 Auto merge of #132580 - compiler-errors:globs, r=Noratrieb
Remove unnecessary pub enum glob-imports from `rustc_middle::ty`

We used to have an idiom in the compiler where we'd prefix or suffix all the variants of an enum, for example `BoundRegionKind`, with something like `Br`, and then *glob-import* that enum variant directly.

`@noratrieb` brought this up, and I think that it's easier to read when we just use the normal style `EnumName::Variant`.

This PR is a bit large, but it's just naming.

The only somewhat opinionated change that this PR does is rename `BorrowKind::Imm` to `BorrowKind::Immutable` and same for the other variants. I think these enums are used sparingly enough that the extra length is fine.

r? `@noratrieb` or reassign
2024-11-05 08:30:56 +00:00
Ralf Jung
ffad9aac27 mark some target features as 'forbidden' so they cannot be (un)set
For now, this is just a warning, but should become a hard error in the future
2024-11-04 22:56:47 +01:00
Michael Goulet
be4b0261c2 ty::KContainer -> ty::AssocItemContainer::K 2024-11-04 04:45:52 +00:00
Jubilee
7155c65d68
Rollup merge of #132565 - bjorn3:less_target_name_dependence, r=workingjubilee
Reduce dependence on the target name

The target name can be anything with custom target specs. Matching on fields inside the target spec is much more robust than matching on the target name.

Also remove the unused is_builtin target spec field.
2024-11-03 20:08:14 -08:00
Jubilee
f35433e250
Rollup merge of #131222 - thejpster:fix-sparc-v7-symbol-o, r=workingjubilee
Generate correct symbols.o for sparc-unknown-none-elf

This fixes #130172 by selecting the correct ELF Machine type for sparc-unknown-none-elf (which has a baseline of SPARC V7).
2024-11-03 20:08:13 -08:00