Before the patch, the default behavior of the compiler was to omit frame
pointers. However, even though it looks fine for the code produced
by the compiler, the compiler itself was broken. When there was a
compilation error, rustc panics to report that error. While handling
that panic, it tries to unwind the stack but fails to do so, which
results in an abort(3).
Here is an example:
$ rustc hello.rs
error: invalid reference to argument `0` (no arguments given)
--> hello.rs:2:5
|
2 | println!("Hello {}")
| ^^^^^^^^^^^^^^^^^^^^
|
= note: this error originates in a macro outside of the current crate
error: aborting due to previous error
fatal runtime error: failed to initiate panic, error 5
Abort trap (core dumped)
The problem also exists on other systems such as OpenBSD and
i686-unknown-linux-musl. "omit frame pointers" was already disabled for
i686-unknown-linux-musl:
88a28ff602/src/librustc_back/target/i686_unknown_linux_musl.rs (L22)
The patch does the same for *-unknown-freebsd.
See the upstream bug for more informations:
https://github.com/rust-lang/rust/issues/43575
This may help ports like textproc/ripgrep to run on old hardware.
Rust itself still requires SSE2 until bootstrap is regenerated.
PR: 223415
Approved by: rust (dumbbell)
FTR, this option allowed to used LLVM from ports instead of building the
bundled copy.
The problem is that this combination isn't really tested upstream. This
led to regressions which are difficult to diagnose. For instance, in
Rust 1.21.0, the bundled LLVM provided a new API to query the features
supported by the target arch. The equivalent code inside Rust was
removed to use that new API. Unfortunately, building Rust 1.21.0+ with a
copy of LLVM not providing this API didn't failed but instead made that
list of CPU features empty. This resulted in the following obscure build
failure in Firefox:
error[E0432]: unresolved import `x86::sse2`
To avoid future pain for both end users and maintainers, we decided to
remove that option. Yes, it will increase the (already long) time to
build Rust, but it should save time wasted on debugging what is not
really supported anyway.
PR: 223342, 223300
Reported by: Many users
Support for LibreSSL up-to 2.6.3 is now included, thus we can remove the
patches to the `openssl` and `openssl-sys` crates.
The renewed bootstrap for aarch64 was created on 11.0-RELEASE, instead
of 11.1-STABLE. It should allow it to run on all 11.x and thus fix the
build on 11.0-RELEASE. [1]
PR: 223009 [1]
As this fixes Stylo on i386, remove the workaround in www/firefox.
PR: 223047
Suggested by: Sebastien Marie <semarie@openbsd.org>
Approved by: rust (dumbbell)
MFH: 2017Q4 (required by Firefox 57 scheduled on 2017-11-14)
This port now provides Cargo. This is the recommended now because Cargo
won't be provided separately in the future.
To build Cargo, we set `extended = true` in `config.toml`. As a side
effect, this flag also installs Rust source code. The port has a new
`SOURCES` option (disabled by default) to keep those sources.
As a consequence of this, `devel/cargo` is removed. Several ports
and Makefiles in Mk were updated to depend on `lang/rust` instead of
`devel/cargo`.
The other big change in this patch is the use of the bundled crates,
instead of relying on Cargo's registry (which was part of the distfiles,
in order to allow offline builds). So now, we don't need to prepare the
registry when updating this port.
This has several other benefits:
* It fixes the build with sudo(8).
* It fixes the use of the ino-64 patch (it was not applied to the
registry, thus not used).
Compilation errors were fixed in the ino-64 patch.
Various `.cargo-checksum.json` files are updated after the sources are
patched (FBSD10_FIX, ino-64, and so on). This fixes builds which were
failing with errors such as:
error: the listed checksum of `.../rustc-1.19.0-src/src/vendor/lzma-sys/xz-5.2.3/build-aux/config.rpath` has changed:
expected: c8b4c017079da9dfb3086a0583e60ffe736184d89005dc5973f0bb0fd17c04bb
actual: 561b00eb30ecaef2c9da17bc195e7d2a7ea63facea38ea9849fbb0ed340bebba
PR: 221088
Reported by: joneum@, nwhitehorn@, romain@,
Ekaterina Vaartis <vaartis@cock.li>,
david@catwhisker.org,
fullermd@over-yonder.net,
rum1cro@yandex.ru,
w.schwarzenfeld@utanet.at
Differential Revision: https://reviews.freebsd.org/D11783
The main change in the port is the switch from configure+make to the
`x.py` build script. This is the recommended way of building Rust. It
takes a configuration file, `config.toml`, which, in the port case,
indicates:
* where to install things;
* which release channel to use;
* where to find Python and llvm-config;
* if documentation should be also built.
The build process should now honor ${MAKE_JOBS_NUMBER} instead of always
using all available CPU cores.
The Cargo registry archive is significantly larger. It includes the
whole registry index (not only the crates needed by Rust). If the index
is missing, Cargo wants to download it from crates.io and I couldn't
find a way to convince it skip it.
New bootstraps for FreeBSD/aarch64 were prepared for this update. Those
are based on Rust 1.18.0 and Cargo 0.19.0.
CONFLICTS_BUILD with installed lang/rust and lang/rust-nightly is
commented out for now. I couldn't see any build failure with an installed
lang/rust so it looks promising. But I couldn't test with an installed
lang/rust-nightly because the latter is marked BROKEN for now. Once it's
restored, I'll do more testing and hopefully completely remove this
CONFLICTS_BUILD variables.
This port now uses an unofficial bootstrap of Rust 1.17.0 and Cargo
0.18.0 for FreeBSD/aarch64. Here are my notes to create this boostrap:
https://gist.github.com/dumbbell/b587da50ef014078da9e732a4331ebad
The port's Makefile was changed to:
- allow to override the versions and directories of the bootstrap
archives;
- patch the bootstrap script and manifest with those
versions/directories.
Beside changes to support FreeBSD/aarch64, the port received the
following fixes:
- Pass python path to the configure script.
- Fix and simplify the extraction of rust-std bootstrap by re-using
existing variables.
- Make sure tar(1) doesn't produce sparse file entries when it
recreates the rust-std bootstrap because Python 2's tarfile module
doesn't support them (PR 219842).
PR: 216143, 219842
Differential Revision: https://reviews.freebsd.org/D10857
Both versions of Rust seen changes to their build system. Now they
require a bootstrap of Cargo and thus, a snapshot of Cargo's registry.
Each library has its own suffix instead of a common suffix for a given
version of Rust. Thus all future updates of the ports will include
changes to their plist.
$DISTFILES are now all under the `rust` directory in `$DISTDIR`. A
similar change will be committed to devel/cargo later.
lang/rust-nightly is enabled on i386 (bugzilla 216143). The source
snapshot also contains everything to build it on aarch64, but enabling
it will be committed later.
PR: 216143
Approved by: riggs, bapt (mentor), antoine (mentor)
Differential Revision: https://reviews.freebsd.org/D9286
Language changes require very recent stage0 snapshots to build
rust 1.8.0. Upstream forgot to make these snapshots known to the
build system. While this was patched for FreeBSD in r413501, it
was missed for DragonFly. This changeset fixes this.
- Provide target for 'make test': Use bundled rust regression test suite
- Use bundled LLVM for now: Built with it, rust passes more regression tests
- Bump PORTREVISION
Dependencies are updated: Perl is not a dependency at all and Python is
a build dependency only.
Rust may fail to build if a previous version is already installed
because the bootstrapped rustc puts "-L/usr/local/lib" at the beginning
of the linker flags. Therefore, mark rust as a CONFLICTS_BUILD with
itself.
Fix the build as a normal user by setting permissions on all directories
to 0755 [1].
Be closer to the standalone installer's file hierarchy by installing
files such as "components", "manifest-rustc" and "uninstall.sh".
Add "files/patch-src_librustc__back_target_freebsd__base.rs"
to remove "-L/usr/local/lib", "-L/usr/local/lib/gcc46" and
"-L/usr/local/lib/gcc44" from the builtin linker flags.
Differential Revision: https://reviews.freebsd.org/D2466
Submitted by: Chris Hutchinson [1]
Reviewed by: jonathan, kwm
Approved by: jonathan
- Remove no-op patch file (builds just fine with -Werror as it seems)
- In BATCH or PACKAGE_BUILDING modes, produce verbose build logs
- Rework post-extract target: do not extract bootstrap compiler ourselves,
put it into expected download directory instead; this also allows us to
optimize away RUST_TARGET knob and post-patch hack
- Since the port is for FreeBSD 10+, no need to explicitly depend on
libexecinfo.so
- Use OPTIONS helpers, perform some other minor Makefile cleanups
- Augment port description text while I am here
