bootstrap: x.py dist rustc-src should keep LLVM's siphash

[lambdageek]

Fixes #145117

[rustbot]

r? @jieyouxu

rustbot has assigned @jieyouxu.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[Kobzol]

Thanks for taking a look! I think that something like this makes sense, yeah. I wonder why distcheck didn't catch this though.. 🤔 Ah, because it only checks plain source tarball and src, and not rustc-src.. we have too many source tarballs.

r? @Kobzol

[lambdageek]

@Kobzol rustc-src is the plain source tarball, isn't it?

rust/src/bootstrap/src/core/build_steps/dist.rs

Line 1061 in 2886b36

run.alias("rustc-src").default_condition(builder.config.rust_dist_src)

[lambdageek]

does distcheck build LLVM? Could it be using the cached one?

[Kobzol]

Right, the naming confused me yet again, sorry. Hmm, then that's weird, as distcheck should do some simple build after extracing the source tarball.. I'll have to investigate that.

[Kobzol]

rust/src/ci/docker/host-x86_64/x86_64-gnu-distcheck/Dockerfile

Line 37 in d43c2e7

ENV NO_DOWNLOAD_CI_LLVM 1

it shouldn't be using cached LLVM.

[lambdageek]

So here's a suspicious thing that happens when I do x.py test distcheck locally on an x86_64-unknown-linux-gnu with master:

running: cd "/home/aleksey/hacking/upstream-rust/build/tmp/distcheck/build/x86_64-unknown-linux-gnu/llvm/build" && CMAKE_PREFIX_PATH="" DESTDIR="" LC_ALL="C" "cmake" "/home/aleksey/hacking/upstream-rust/src/llvm-project/llvm" "-G" "Ninja" "-DLLVM_ENABLE_ASSERTIONS=OFF" "-DLLVM_UNREACHABLE_OPTIMIZE=OFF" "-DLLVM_ENABLE_PLUGINS=OFF" "-DLLVM_TARGETS_TO_BUILD=AArch64;AMDGPU;ARM;BPF;Hexagon;LoongArch;MSP430;Mips;NVPTX;PowerPC;RISCV;Sparc;SystemZ;WebAssembly;X86" "-DLLVM_EXPERIMENTAL_TARGETS_TO_BUILD=AVR;M68k;CSKY;Xtensa" "-DLLVM_INCLUDE_EXAMPLES=OFF" "-DLLVM_INCLUDE_DOCS=OFF" "-DLLVM_INCLUDE_BENCHMARKS=OFF" "-DLLVM_INCLUDE_TESTS=OFF" "-DLLVM_ENABLE_LIBEDIT=OFF" "-DLLVM_ENABLE_BINDINGS=OFF" "-DLLVM_ENABLE_Z3_SOLVER=OFF" "-DLLVM_PARALLEL_COMPILE_JOBS=16" "-DLLVM_TARGET_ARCH=x86_64" "-DLLVM_DEFAULT_TARGET_TRIPLE=x86_64-unknown-linux-gnu" "-DLLVM_ENABLE_WARNINGS=OFF" "-DLLVM_INSTALL_UTILS=ON" "-DLLVM_ENABLE_ZLIB=ON" "-DLLVM_ENABLE_LIBXML2=OFF" "-DLLVM_VERSION_SUFFIX=-rust-1.91.0-nightly" "-DCMAKE_INSTALL_MESSAGE=LAZY" "-DCMAKE_C_COMPILER=cc" "-DCMAKE_CXX_COMPILER=c++" "-DCMAKE_ASM_COMPILER=cc" "-DCMAKE_C_FLAGS=-ffunction-sections -fdata-sections -fPIC -m64" "-DCMAKE_CXX_FLAGS=-ffunction-sections -fdata-sections -fPIC -m64" "-DCMAKE_SHARED_LINKER_FLAGS=" "-DCMAKE_MODULE_LINKER_FLAGS=" "-DCMAKE_EXE_LINKER_FLAGS=" "-DLLVM_ENABLE_ZSTD=OFF" "-DCMAKE_INSTALL_PREFIX=/home/aleksey/hacking/upstream-rust/build/tmp/distcheck/build/x86_64-unknown-linux-gnu/llvm" "-DCMAKE_ASM_FLAGS= -ffunction-sections -fdata-sections -fPIC -m64" "-DCMAKE_BUILD_TYPE=Release"

we're in the <git_root>/build/tmp/distcheck/build/... directory, so it's the build that distcheck invokes,
but we're calling cmake on <git_root>/src/llvm-project/llvm not on the llvm from the tarball.

Here's my bootstrap.toml

# Use different pre-set defaults than the global defaults.
#
# See `src/bootstrap/defaults` for more information.
# Note that this has no default value (x.py uses the defaults in `bootstrap.example.toml`).
profile = 'dist'

[llvm]
download-ci-llvm = false
targets = "AArch64;X86"

[gcc]

[build]

# Arguments passed to the `./configure` script, used during distcheck. You
# probably won't fill this in but rather it's filled in by the `./configure`
# script. Useful for debugging.
configure-args = []

[install]

[rust]
download-rustc = false

[dist]
vendor = false
compression-formats = ["gz"]

[target.x86_64-unknown-linux-gnu]

(and I also modify distcheck not to pass --enable-vendor; it's just to speed things up a bit; but i'll try a pristine tree, soon, too)

[lambdageek]

Yeah you can see the same thing on CI

for example this build the raw log has this:

2025-08-08T21:21:12.0222503Z running: cd "/checkout/obj/build/tmp/distcheck/build/x86_64-unknown-linux-gnu/llvm/build" && CMAKE_PREFIX_PATH="" DESTDIR="" LC_ALL="C" "cmake" "/checkout/src/llvm-project/llvm" "-G" "Ninja" "-DLLVM_ENABLE_ASSERTIONS=ON" "-DLLVM_UNREACHABLE_OPTIMIZE=OFF" "-DLLVM_ENABLE_PLUGINS=OFF" "-DLLVM_TARGETS_TO_BUILD=AArch64;AMDGPU;ARM;BPF;Hexagon;LoongArch;MSP430;Mips;NVPTX;PowerPC;RISCV;Sparc;SystemZ;WebAssembly;X86" "-DLLVM_EXPERIMENTAL_TARGETS_TO_BUILD=AVR;M68k;CSKY;Xtensa" "-DLLVM_INCLUDE_EXAMPLES=OFF" "-DLLVM_INCLUDE_DOCS=OFF" "-DLLVM_INCLUDE_BENCHMARKS=OFF" "-DLLVM_INCLUDE_TESTS=OFF" "-DLLVM_ENABLE_LIBEDIT=OFF" "-DLLVM_ENABLE_BINDINGS=OFF" "-DLLVM_ENABLE_Z3_SOLVER=OFF" "-DLLVM_PARALLEL_COMPILE_JOBS=4" "-DLLVM_TARGET_ARCH=x86_64" "-DLLVM_DEFAULT_TARGET_TRIPLE=x86_64-unknown-linux-gnu" "-DLLVM_ENABLE_WARNINGS=OFF" "-DLLVM_INSTALL_UTILS=ON" "-DLLVM_ENABLE_ZLIB=ON" "-DLLVM_ENABLE_LIBXML2=OFF" "-DLLVM_VERSION_SUFFIX=-rust-1.91.0-nightly" "-DCMAKE_INSTALL_MESSAGE=LAZY" "-DCMAKE_C_COMPILER_LAUNCHER=sccache" "-DCMAKE_CXX_COMPILER_LAUNCHER=sccache" "-DCMAKE_C_COMPILER=cc" "-DCMAKE_CXX_COMPILER=c++" "-DCMAKE_ASM_COMPILER=cc" "-DCMAKE_C_FLAGS=-ffunction-sections -fdata-sections -fPIC -m64" "-DCMAKE_CXX_FLAGS=-ffunction-sections -fdata-sections -fPIC -m64" "-DCMAKE_SHARED_LINKER_FLAGS= -Wl,-Bsymbolic -static-libstdc++" "-DCMAKE_MODULE_LINKER_FLAGS= -Wl,-Bsymbolic -static-libstdc++" "-DCMAKE_EXE_LINKER_FLAGS= -Wl,-Bsymbolic -static-libstdc++" "-DLLVM_ENABLE_ZSTD=OFF" "-DCMAKE_INSTALL_PREFIX=/checkout/obj/build/tmp/distcheck/build/x86_64-unknown-linux-gnu/llvm" "-DCMAKE_ASM_FLAGS= -ffunction-sections -fdata-sections -fPIC -m64" "-DCMAKE_BUILD_TYPE=Release"

we're in the tmp/distcheck build but the sources we're compiling are /checkout/src/llvm-project/llvm not build/tmp/distcheck/src/llvm-project

[lambdageek]

Ok, it's happening because distcheck just invokes make check which runs .../build/tmp/distcheck/src/bootstrap/bootstrap.py test --stage 2

when the inner bootstrap runs, it calls compute_src_directory where flags_src is None

rust/src/bootstrap/src/core/config/config.rs

Lines 473 to 475 in ffb9d94

	if let Some(src) = compute_src_directory(flags_src, &config.exec_ctx) {
	config.src = src;
	}

And compute_src_directory ends up executing git rev-parse --show-cdup with cwd in .../build/tmp/distcheck so it ends up setting src to the outer git checkout, not the current extracted tarball. From there, all the git_info for the submodules ends up finding them in the outer git checkout.

I tried a naive change to add --src $(CFG_SRC_DIR) here:

rust/src/bootstrap/mk/Makefile.in

Line 16 in ffb9d94

BOOTSTRAP := $(CFG_PYTHON) $(CFG_SRC_DIR)src/bootstrap/bootstrap.py

but then .../bootstrap.py --src ... test --stage 2 fails when it runs the tidy check:

Commit SHA of the src/gcc submodule (``) does not match the required GCC version of the GCC codegen backend (`04ce66d8c918de9273bd7101638ad8724edf5e21`).
Make sure to set the src/gcc submodule to commit 04ce66d8c918de9273bd7101638ad8724edf5e21.
The GCC codegen backend commit is configured at /home/aleksey/hacking/upstream-rust/build/tmp/distcheck/compiler/rustc_codegen_gcc/libgccjit.version.

Which makes sense because we specifically exclude src/gcc from the plain tarball.

And at this point I'm stuck. I can't tell if there's some way to configure tidy not to do the gcc_submodule check or maybe the plain tarball should include some breadcrumb that the tidy check could look for

---

Update yeah if i comment out the tidy gcc_submodule check, and pass --src in Makefile.in, I can get distcheck to do the right thing:

running: cd "/home/aleksey/hacking/upstream-rust/build/tmp/distcheck/build/x86_64-unknown-linux-gnu/llvm/build" && CMAKE_PREFIX_PATH="" DESTDIR="" LC_ALL="C" "cmake" "/home/aleksey/hacking/upstream-rust/build/tmp/distcheck/src/llvm-project/llvm ..."

---

Update2:

@Kobzol I went ahead and added the distcheck fix here too:

1. pass --src in Makefile.in
2. teach tidy gcc_submodule to look for the src/gcc/notice.txt that we create in the tarball

Does that sound right?

I am not sure if passing --src in the Makefile is ok. It probably affects a lot more than just distcheck. Maybe some less invasive change would be better? Is there some way to directly invoke bootstrap.py without going through the Makefile?

[Kobzol]

Thanks a lot for investigating this! ❤️ It seems like distcheck is due for a big overhaul.. I would probably modify it so that it extracts itself to some temporary directory, and then invoke the build from there, to minimize the chances that it will pick up something unrelated from the source checkout. And also we should pass --set llvm.download-ci-llvm = false to it explicitly, and not read configure_args from the external environment.

Let's avoid any changes to distcheck and tidy in this PR, and just fix the LLVM build.

[lambdageek]

@Kobzol Ok, sounds good. I reverted the distcheck and tidy changes in this PR, and created a new issue about distcheck: #145183

[lambdageek]

@Kobzol I'm somewhat unhappy with all the "foo/bar", "foo\\bar" duplication all over the filter. I assume this is to avoid doing allocations. Is this filter on the hot path? What if instead of &[&str] we had LazyLock<Vec<&Path>> for all the interesting constants?

[Kobzol]

It's called for every path, so I imagine it might be hot, yeah. But probably there's a much better way of implementing this.. Anyway, we can resolve that in a separate PR. Thank you for investigating this! I can't reproduce the original issue on Linux, but the directory is now there in the archive, so hopefully it should help on macOS.

@bors r+ rollup

[bors]

📌 Commit cba5918 has been approved by Kobzol

It is now in the queue for this repository.