16.1 The Reduced Binary Seed Bootstrap

Guix—like other GNU/Linux distributions—is traditionally bootstrapped from a set of bootstrap binaries: Bourne shell, command-line tools provided by GNU Coreutils, Awk, Findutils, ‘sed’, and ‘grep’ and Guile, GCC, Binutils, and the GNU C Library (see Bootstrapping). Usually, these bootstrap binaries are “taken for granted.”

Taking the bootstrap binaries for granted means that we consider them to be a correct and trustworthy “seed” for building the complete system. Therein lies a problem: the combined size of these bootstrap binaries is about 250MB (see Bootstrappable Builds in GNU Mes). Auditing or even inspecting these is next to impossible.

For i686-linux and x86_64-linux, Guix now features a “Reduced Binary Seed” bootstrap ³³.

The Reduced Binary Seed bootstrap removes the most critical tools—from a trust perspective—from the bootstrap binaries: GCC, Binutils and the GNU C Library are replaced by: bootstrap-mescc-tools (a tiny assembler and linker) and bootstrap-mes (a small Scheme Interpreter and a C compiler written in Scheme and the Mes C Library, built for TinyCC and for GCC).

Using these new binary seeds the “missing” Binutils, GCC, and the GNU C Library are built from source. From here on the more traditional bootstrap process resumes. This approach has reduced the bootstrap binaries in size to about 145MB in Guix v1.1.

The next step that Guix has taken is to replace the shell and all its utilities with implementations in Guile Scheme, the Scheme-only bootstrap. Gash (see Gash in The Gash manual) is a POSIX-compatible shell that replaces Bash, and it comes with Gash Utils which has minimalist replacements for Awk, the GNU Core Utilities, Grep, Gzip, Sed, and Tar. The rest of the bootstrap binary seeds that were removed are now built from source.

Building the GNU System from source is currently only possible by adding some historical GNU packages as intermediate steps³⁴. As Gash and Gash Utils mature, and GNU packages become more bootstrappable again (e.g., new releases of GNU Sed will also ship as gzipped tarballs again, as alternative to the hard to bootstrap xz-compression), this set of added packages can hopefully be reduced again.

The graph below shows the resulting dependency graph for gcc-core-mesboot0, the bootstrap compiler used for the traditional bootstrap of the rest of the Guix System.

The only significant binary bootstrap seeds that remain³⁵ are a Scheme interpreter and a Scheme compiler: GNU Mes and GNU Guile³⁶.

This further reduction has brought down the size of the binary seed to about 60MB for i686-linux and x86_64-linux.

Work is ongoing to remove all binary blobs from our free software bootstrap stack, working towards a Full Source Bootstrap. Also ongoing is work to bring these bootstraps to the arm-linux and aarch64-linux architectures and to the Hurd.

If you are interested, join us on ‘#bootstrappable’ on the Freenode IRC network or discuss on bug-mes@gnu.org or gash-devel@nongnu.org.