Backport upstream patches to enable native musl toolchain support in
GCC. Only the changes required for i386 were taken, excluding
changes for libgo and libfortran.
These patches enable binaries built using gcc and g++ to automatically
use musl's dynamic linker as their interpreter when present during the
build.
Instead of using the pre-generated "bootstrap" script, execute the
relevant bootstrap operations manually. This doesn't actually change
the build output; the final package hash remains identical.
This should avoid configuration errors caused by combining an older
automake that doesn't support Go with a newer libtool that does.
No change in package hash.
- disk to be created (blank disk given to live-bootstrap) (default)
- disk to already exist but sources downloaded within live-bootstrap
- sources to be downloaded outside live-bootstrap (non-blank disk given
to live-bootstrap)
Also migrate sysb to use sys_transfer in QEMU mode also.
Note that this means copy_sysc is now irrelevant. sysc is *always*
sourced from sysa.
* Bzip2 was manually installed directly into filesystem, so bzip2 package was empty.
Fixed by installing it to destdir. bzip2 moves out its binary before installing
its own package.
* sha256sum from stage0-posix was not checking any checksums because it does not
accept piped input. Fixed by using temporary file.
* grep was broken for a short time (but with fixed bzip2 package this caused failures)
due to touch creating grep file of zero size (egrep symlink was touched).
Fixed by implementing touch -h
Resolves#156, resolves#166, resolves#167
XBPS packages don't seem to have an option to manipulate file
permissions at package creation time.
So as a best effort, set the umask as soon as possible to ensure
file permissions are the same when running the bootstrap.
This is enough to allow reproducible XBPS packages when starting the
bootstrap with a umask different than 0022.
This makes the order of objects in archives created by libtool
consistent.
It is known to affect cases where the *_LIBADD automake variable is
used to add extra objects from a separate archive.
Fixing this allows us to remove a few workarounds.
- Rename sources to distfiles for clarity.
- Per sys(a/c) distfiles to reduce rootfs.py processing and reduce RAM
usage in sysa.
- Canonicalise early kaem mes/tcc files to kaem script conventions.
- Cleanup unused setup in python.
Include:
- Not regenerated man pages.
- Automake tarball with bad time (and completely useless).
- Ordering of files within tarballs.
- Resetting timestamp various fixes.
- Older tars not properly overwriting files leading to many issues.
- Weird lack of reproducibility in libtool scripts paths to tools.
- and more
Utilising previously introduced DESTDIR support, everything is installed
to aformentioned DESTDIR.
Prior to the building of XBPS, we use gzip compressed tarballs as
packages. This requires a lot of strange hacks especially for old tar +
gzip (timestamps, etc causing reproducibility problems). Then we use
XBPS once built.
In rootless boostrap, dev nodes must be mounted before starting the
bootstrap process. However, some of these aren't necessarily mounted,
such as when relying upon BuildStream's sandbox.
Since these are not actually required for either a rootless or
chroot-based bootstrap, simply avoid them.
- The date reproducibility issue was not actually set to any date, which
could cause issues in the fiture, fix that sed.
- Fix a reproducibility issue (under QEMU), where the ordering of files
within libguile can be random. This also affects guile binary so we
need to rebuild that.
- Some change in kernel versions means that guile binary is
significantly different in newer kernel versions. Because we cannot
know the version of the host under chroot mode we must disable guile
checksum entirely.
- Update checksums accordingly.
- Add parts.rst documentation for Linux kernel.
- Completely fix problems caused by new bootstrap, update checksums for
/usr.
- Globalise populate_device_nodes.
- Enable deblobbing.
- This is much more standard and replaces /image in sysa and is the
standard in sysc (avoids many issues).
- GCC needs to have a file created for some unknown reason.
- Checksums updated.
Now that we have the Linux Kernel built, we move to a full-disk (rather
than initramfs) setup in sysc. However, we cannot assume the seed kernel
has support for mounting hard drives. So, first we need to kexec into
sysb, which is used as a jumping off point to create the hard drive for
sysc.
Additionally, since 2.6.16 does not have support for on-demand initramfs
(initramfs must be built into kernel), we will have to rebuild the linux
kernel within sysb without the initramfs.
All of this process is not performed for chroot mode. Instead, we skip
sysb and jump straight to sysc, copying over appropriate data.
The python scripts have been changed slightly. Each sys* inherits
SysGeneral, which contains various functions which are not specific to
any sys* and simplifies those files. rootfs now also handles sysb and
sysc.
bootstrap.cfg also gives an indication whether we are running in a
chroot to avoid attempting to kexec/mount within a chroot.
Andrius Štikonas
rick-masters
Dor Askayo
fosslinux