bootstrap: vendor crates required by opt-dist to collect profiles
These are the default package set required by opt-dist to correctly work,
hence for people wanting to build a production grade of rustc in a
sandboxed / air-gapped environment, these need to be vendored.
The size of `rustc-src-nightly.tar.xz` before and after this change:
* Before: 298M
* After: 323M (+8%)
Size change might or might not be a concern.
See the previous discussion: https://github.com/rust-lang/rust/pull/125166#issuecomment-2113626468
Previous efforts on making:
* https://github.com/rust-lang/rust/pull/125125
* https://github.com/rust-lang/rust/pull/125166
---
Note that extra works still need to be done to make it fully vendored.
* The current pinned rustc-perf uses `tempfile::Tempdir` as the working
directory when collecting profiles from some of these packages.
This "tmp" working directory usage make it impossible for Cargo to pick
up the correct vendor sources setting in `.cargo/config.toml` bundled
in the rustc-src tarball. [^1]
* opt-dist verifies the final built rustc against a subset of rustc test
suite. However it rolls out its own `config.toml` without setting
`vendor = true`, and that results in `./vendor/` directory removed.
[^2]
[^1]: 4f313add60/collector/src/compile/benchmark/mod.rs (L164-L173)
[^2]: 606afbb617/src/tools/opt-dist/src/tests.rs (L62-L77)
Add `FRAC_1_SQRT_2PI` constant to f16/f32/f64/f128
This adds the `FRAC_1_SQRT_2PI` to the `f16`, `f32`, `f64` and `f128` as [`1/√(2π)`](https://www.wolframalpha.com/input?i=1%2Fsqrt%282*pi%29). The rationale is that while `FRAC_1_SQRT_PI` already exists, [Gaussian calculations](https://en.wikipedia.org/wiki/Gaussian_function) for random normal distributions require a `1/(σ√(2π))` term, which could then be directly expressed e.g. as `f32::FRAC_1_SQRT_2PI / sigma`.
The actual value is approximately `1/√(2π) = 0.3989422804014326779399460599343818684758586311649346576659258296…`. Truncated/rounded forms were used for the individual types.
---
~~I did not any of the `#[unstable]` attributes since I am not aware of their implications.~~
**Edit:** I applied the stability attributes from the surrounding types according to what seemed most likely correct. I believe the `more_float_constants` feature marker is incorrectly applied, but I wasn't sure how to proceed.
Enable GVN for `AggregateKind::RawPtr`
Looks like I was worried for nothing; this seems like it's much easier than I was originally thinking it would be.
r? `@cjgillot`
This should be useful for `x[..4]`-like things, should those start inlining enough to expose the lengths.
Adds --remap-path-prefix as an unstable option. This is implemented to
mimic the behavior of rustc's --remap-path-prefix but with minor
adjustments.
This flag similarly takes in two paths, a prefix to replace and a
replacement string.
ci: use GCC 13 as cross compiler in `dist-aarch64-linux`
I'm proposing this GCC upgrade since it addresses bug https://github.com/rust-lang/rust/issues/125619. The
regression in question affects stable release consumers who tend to have
no exposure to Rust build tools, so if at all possible, I would like to
have it resolved in the next stable release. I have tried to fix the bug
in `compiler-builtins`, which led to submitting a PR for `compiler-rt`
in upstream LLVM, but it may take a long time before these upstreams
address this regression.
A summary of why upgrading GCC solves the regression follows.
`__multc3()` is a builtin function `compiler-builtins` exposes for
specifically aarch64, non-Windows targets [1]. The object file for it is
included in `staticlib` archives through `libstd`. The implementation
for `__multc3()` is from `multc3.c`, part of LLVM's `compiler-rt`.
Upstream `compiler-rt` normally builds the C file using the Clang
from the same LLVM version. On the other hand, `compiler-builtins`
builds the C file using GCC, outside of the usual LLVM build system.
The upstream implementation doesn't have feature detection which
works for GCC version older than 10, and ends up producing an
unlinkable object.
Upstream LLVM might be slow to respond to this issue as they might deem
`compiler-builtin` as doing something out of the ordinary from their
perspective. They might reasonably assume everyone builds `compiler-rt`
through LLVM's build system.
I have done the following to test this change:
- verified that a local build without this patch exhibits the
regression.
- verified that with this patch, the object for `__multc3()` has no
reference to undefined functions in the symbol table.
- verified that with this patch, `rustc` is usable to build Ruby with
YJIT, and that the reported regression is resolved.
Since `loongarch64-linux-gnu` already uses GCC 13.2.0, I hope we can upgrade without issues.
try-job: dist-aarch64-linux
[1]: c04eb9e1af/build.rs (L524-L539)
`ring` is included because it is an optional dependency of `hyper`,
which is a training data in rustc-pref for optimized build.
The license of it is generally `ISC AND MIT AND OpenSSL`,
though the `package.license` field is not set.
See https://github.com/briansmith/ring/issues/902
`git+https://github.com/rust-lang/team` git source is from
`rust_team_data`, which is used by `site` in src/tools/rustc-perf.
This doesn't really a part of the compiler,
so doesn't really affect the bootstrapping process.
See https://github.com/rust-lang/rustc-perf/pull/1914.
These are the default package set required by opt-dist to correctly work,
hence for people wanting to build a production grade of rustc in a
sandboxed / air-gapped environment, these need to be vendored.
The size of `rustc-src-nightly.tar.xz` before and after this change:
* Before: 298M
* After: 323M (+8%)
These crates are the default set of packages required by opt-dist
to correctly work, hence for people wanting to build a production grade
of rustc in an sandboxed / air-gapped environment, these need to be vendored.
The size of `rustc-src-nightly.tar.xz` before and after this change:
* Before: 298M
* After: 323M (+8%)
Size change might or might not be a concern.
See the previous discussion: https://github.com/rust-lang/rust/pull/125166#issuecomment-2113626468
Previous efforts on making:
* https://github.com/rust-lang/rust/pull/125125
* https://github.com/rust-lang/rust/pull/125166
---
Note that extra works still need to be done to make it fully vendored.
* The current pinned rustc-perf uses `tempfile::Tempdir` as the working
directory when collecting profiles from some of these packages.
This "tmp" working directory usage make it impossible for Cargo to pick
up the correct vendor sources setting in `.cargo/config.toml` bundled
in the rustc-src tarball. [^1]
* opt-dist verifies the final built rustc against a subset of rustc test
suite. However it rolls out its own `config.toml` without setting
`vendor = true`, and that results in `./vendor/` directory removed.
[^2]
[^1]: 4f313add60/collector/src/compile/benchmark/mod.rs (L164-L173)
[^2]: 606afbb617/src/tools/opt-dist/src/tests.rs (L62-L77)
- Autolabel PRs modifying `tests/run-make/` and
`src/tools/run-make-support/` with `X-run-make` label.
- Add reminder to update the tracking issue
<https://github.com/rust-lang/rust/issues/121876>
if applicable when `tests/run-make/` is modified by a PR.
Fix futex with large timeout ICE
Fixes#3647.
This PR changed the type of ``nanoseconds`` from ``u64`` to ``u128``. In ``duration_since``, nanoseconds is manually converted to second by dividing it with 1e9. But overflow is still possible.
Rollup of 5 pull requests
Successful merges:
- #126137 (tests: Add ui/higher-ranked/trait-bounds/normalize-generic-arg.rs)
- #126146 (std::unix::process adding few specific freebsd signals to be able to id.)
- #126155 (Remove empty test suite `tests/run-make-fulldeps`)
- #126168 (std::unix::os current_exe implementation simplification for haiku.)
- #126175 (Use --quiet flag when installing pip dependencies)
r? `@ghost`
`@rustbot` modify labels: rollup
std::unix::os current_exe implementation simplification for haiku.
_get_net_image_info is a bit overkill as it allows to get broader informations about the process.
Remove empty test suite `tests/run-make-fulldeps`
After #109770, there were only a handful of tests left in the run-make-fulldeps suite.
As of #126111, there are no longer *any* run-make-fulldeps tests, so now we can:
- Remove the directory
- Remove related bootstrap/compiletest code
- Remove various other references in CI scripts and documentation.
By removing this suite, we also no longer need to worry about discrepancies between it and ui-fulldeps, and we don't have to worry about porting tests from Makefile to [rmake](https://github.com/rust-lang/rust/issues/121876) (or whether rmake even works with fulldeps).
tests: Add ui/higher-ranked/trait-bounds/normalize-generic-arg.rs
This adds a regression test for an ICE "accidentally" fixed by https://github.com/rust-lang/rust/pull/101947 that does not add a test for this particular case.
Closes#107564.
I have confirmed the added test code fails with `nightly-2023-01-09` (and passes with `nightly-2023-01-10` and of course recent `nightly`).
simd packed types: remove outdated comment, extend codegen test
It seems like https://github.com/rust-lang/rust/pull/125311 made that check in codegen unnecessary?
r? `@workingjubilee` `@calebzulawski`
It looks like this job was intending to run all of the `needs-matching-clang`
tests (since they don't run without `RUSTBUILD_FORCE_CLANG_BASED_TESTS`), but
over time developed two problems:
- The tests it cares about were moved from run-make-fulldeps to run-make.
- Some of the relevant tests don't actually have "clang" in their name.
Switching to run-make solves the first problem, but we still don't run the
tests without "clang" in their name, because some of them are currently broken.
