RUSTC_TARGET_ARCH comes from the GNU triple, so every 32-bit PowerPC
subtarget gets powerpc-unknown-linux-musl. That triple is hard-float and
mpc85xx is not: its e500v2 has no floating-point unit, and the toolchain
reports _SOFT_FLOAT and calls __divdf3 for a division.
So rustc emits lfd, stfd and fdiv for a core with no floating-point
registers, and passes doubles in a way nothing else in the image does. It
does not fail outright — a speed test reported negative throughput, and
formatting one of those values tripped an assertion inside core.
Pass -Ctarget-feature=-hard-float where CONFIG_SOFT_FLOAT says so. The same
flag has to reach the std that rust/host builds, or std and the packages
linked against it disagree about how a double is passed.
Signed-off-by: Josef Schlehofer <pepe@bloodkings.eu>
rust/host fails to compile on macOS running on Intel x64
because the host target triple is autogenerated to be
'arm64-unknown-linux-'. Rust doesn't have such a target triple, thus the
build fails because there are no pre-built artefacts for bootstrapping.
Fix this by setting RUSTC_HOST_ARCH to 'x86_64-apple-darwin' in case
our host is HOST_ARCH=x86_64 and HOST_OS=Darwin.
This fix is based on the existing fix for Apple silicon [1].
Fixes:
File "/Volumes/pepe/py/openwrt/build_dir/target-x86_64_musl/host/rustc-1.96.0-src/x.py", line 53, in <module>
bootstrap.main()
~~~~~~~~~~~~~~^^
File "/Volumes/pepe/py/openwrt/build_dir/target-x86_64_musl/host/rustc-1.96.0-src/src/bootstrap/bootstrap.py", line 1418, in main
bootstrap(args)
~~~~~~~~~^^^^^^
File "/Volumes/pepe/py/openwrt/build_dir/target-x86_64_musl/host/rustc-1.96.0-src/src/bootstrap/bootstrap.py", line 1366, in bootstrap
build.download_toolchain()
~~~~~~~~~~~~~~~~~~~~~~~~^^
File "/Volumes/pepe/py/openwrt/build_dir/target-x86_64_musl/host/rustc-1.96.0-src/src/bootstrap/bootstrap.py", line 697, in download_toolchain
download_component(download_info)
~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^
File "/Volumes/pepe/py/openwrt/build_dir/target-x86_64_musl/host/rustc-1.96.0-src/src/bootstrap/bootstrap.py", line 529, in download_component
get(
~~~^
download_info.base_download_url,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
...<3 lines>...
verbose=download_info.verbose,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/Volumes/pepe/py/openwrt/build_dir/target-x86_64_musl/host/rustc-1.96.0-src/src/bootstrap/bootstrap.py", line 58, in get
raise RuntimeError(
...<6 lines>...
)
RuntimeError: src/stage0 doesn't contain a checksum for dist/2026-04-16/rust-std-1.95.0-x86_64-unknown-linux-darwin24.6.0.tar.xz. Pre-built artifacts might not be available for this target at this time, see https://doc.rust-lang.org/nightly/rustc/platform-support.html for more information.
[1] https://github.com/openwrt/packages/commit/105fa3920e12f557bdf1fcbc566fc286fb53e319
Signed-off-by: Georgi Valkov <gvalkov@gmail.com>
Currently, armv5 and armv6 targets are both using armv6 rustc.
Without this patch, rust programs in armv5 targets throw illegal instruction
error.
Signed-off-by: Lu jicong <jiconglu58@gmail.com>
Users might configure their own env variables on the host, and sometimes
it can lead build failure or unexpected behavior.
Fixes: #22889
Signed-off-by: Tianling Shen <cnsztl@immortalwrt.org>
Patch the target triple for Rust with glibc to include hard floating
point support.
The GNU target triple used elsewhere does not include hard float support,
instead `-mfloat-abi=hard` is passed separately. For Rust it must be
included in the target triple. This was already being done for musl,
this commit adds the same patching for glibc.
Without this patch Rust compilation fails with an error like this
(abbreviated to fit the line length):
ld: error: libstd.so uses VFP register arguments, ... does not
ld: failed to merge target specific data of file ...
Signed-off-by: Drew Young <dyoung@viridiparente.com>
Using sccache makes recompilation of rustc and Rust packages faster.
This also makes the rust package visible in menuconfig, in order for the
sccache options to be accessible.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
This consolidates all environment variables for cargo into:
* CARGO_HOST_CONFIG_VARS / CARGO_PKG_CONFIG_VARS
These contain all cargo-specific environment variables, i.e. without
"common" variables like CC.
* CARGO_HOST_VARS / CARGO_PKG_VARS (renamed from CARGO_VARS)
These contain all environment variables to be passed to cargo.
This also:
* Set the CARGO_BUILD_TARGET environment variable instead of using the
--target command-line option
* Update Python include files to use CARGO_HOST_CONFIG_VARS /
CARGO_PKG_CONFIG_VARS
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
As CARGO_HOME mainly functions as a download and source cache[1], moving
it into $(DL_DIR) allows it to persist and be reused between different
buildroots/sdks (when DL_DIR is set to a custom/external location).
[1]: https://doc.rust-lang.org/cargo/guide/cargo-home.html
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
This also:
* Modify the "release" profile in place of adding the "stripped" profile
Only the profile for target is modified; there are no file size
constraints for host.
* For host, build with the "release" profile
* For target, build with either the "dev" or "release" profile based on
CONFIG_DEBUG
There is no environment variable to specify the "strip" option, but
enabling this option is not necessary as the build system will already
strip binaries based on CONFIG_NO_STRIP / CONFIG_USE_STRIP /
CONFIG_USE_SSTRIP.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
In order to use $(RUSTC_TARGET_ARCH) in HOST_BUILD_DIR, the line to
include rust-values.mk in the Rust makefile was moved in
f489e019ac, causing it to be included
before package.mk is included.
This had the side effect of preventing "-lssp_nonshared" from being
added to RUSTC_LDFLAGS, because PKG_SSP is indirectly set by package.mk
(package.mk includes hardening.mk, hardening.mk sets PKG_SSP).
There is a deeper issue; it is the Rust package's PKG_SSP value that
causes RUSTC_LDFLAGS to be set and written to the Cargo config file. For
packages that use Rust to build, their PKG_SSP value does not affect the
linker flag.
This sets rustflags with the RUSTFLAGS environment variable, instead of
writing the value to the Cargo config file, allowing PKG_SSP from the
package being built to be used and for the package being built to modify
the rustflags used.
This also:
* Fix "-lssp_nonshared" being added to TARGET_CFLAGS instead of
RUSTC_LDFLAGS, when CONFIG_PKG_CC_STACKPROTECTOR_STRONG is set.
* Remove the use of $(RUSTC_TARGET_ARCH) in HOST_BUILD_DIR and move the
include line for rust-values.mk back to after package.mk.
Since the host build directory was moved under the target build
directory in efdbac38dc, it is no longer
necessary to separate build directories with RUSTC_TARGET_ARCH;
$(BUILD_DIR) already separates build directories by target.
* Add BUILDONLY:=1, as the Rust package does not build a target package.
* Install the Cargo config file as "config.toml" instead of "config", as
this is the preferred form[1].
* Rename RUST_CFLAGS to RUSTC_CFLAGS and CONFIG_HOST_SUFFIX to
RUSTC_HOST_SUFFIX, for consistency.
* Allow CARGO_VARS to be set before rust-values.mk is included.
[1]: https://doc.rust-lang.org/cargo/reference/config.html#hierarchical-structure
Fixes: f489e019ac ("rust: compile host package per target")
Fixes: 83785a7ce0 ("rust-lang: Add the rust language support")
Fixes: https://github.com/openwrt/packages/issues/22133
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
This allows other packages access to RUST_ARCH_DEPENDS by including
rust-values.mk, instead of rust-package.mk which also sets
Build/Compile.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
Fixes the commit 105fa3920e which was intended to make rust/host build
on aarch64 darwin working again. However, the fix contains a mistake
because it sets RUSTC_TARGET_ARCH instead of RUSTC_HOST_ARCH. Thus, the
fix doesn't work.
This properly sets the correct variable RUSTC_HOST_ARCH.
Fixes: 105fa3920e ("rust: fix host build on aarch64 darwin")
Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
rust/host failed to compile on macOS running on Apple Silicon M1 Pro
because the host target triple is autogenerated to be
'arm64-unknown-linux-'. Rust doesn't have such a target triple, thus the
build failes because there are no pre-built artifacts for bootstrapping.
Fix this by setting RUSTC_HOST_ARCH to 'aarch64-apple-darwin' in case
our host is HOST_ARCH=arm64 and HOST_OS=darwin.
Signed-off-by: Jonas Jelonek <jelonek.jonas@gmail.com>
While the compiled binaries are intended to run on the host system, the
rust/host package does include the target matching the configured
OpenWrt target.
If using (for example) ./scripts/env to switch between different
OpenWrt configurations, this will cause issues if the different
configuration is for a different target. In such case there will be a
mismatch between the available Rust target and OpenWrt target and the
following error will be printed:
> error[E0463]: can't find crate for `core`
> note: the `XXX` target may not be installed
This fix will add the RUSTC_TARGET_ARCH as HOST_BUILD_DIR and CARGO_HOME
suffix, such that rust/host will be compiled in case an OpenWrt
configuration change causes the RUSTC_TARGET_ARCH to change.
Fixes: #21530
Signed-off-by: Orne Brocaar <info@brocaar.com>
[Applied Jeffery To's suggestion for build and install path]
Signed-off-by: Tianling Shen <cnsztl@immortalwrt.org>
also add new variable CARGO_VARS to make it possible
to pass environment variables for cargo process.
This is necessary when for example, cross-compiling
netavark.
Signed-off-by: Oskari Rauta <oskari.rauta@gmail.com>
Added new RustBinPackage, RustBinHostBuild wrapper.
Added new RUST_PKG_FEATURES flag.
Moved CARGO_HOME to STAGING_DIR_HOSTPKG.
Overrode default Build/Compile and Host/Compile to Cargo build.
Signed-off-by: Tianling Shen <cnsztl@immortalwrt.org>
Josef Schlehofer
Georgi Valkov
George Sapkin
Tianling Shen
hingbong lo
Lu jicong
Drew Young
Jeffery To
Jonas Jelonek
Orne Brocaar
Oskari Rauta