The build has been failing occasionally, due to some kind of race condition.
The last lines of log output look like this:
Updating policy/booleans.conf and policy/modules.conf
python3 -t -t -E -W error support/sedoctool.py -b policy/booleans.conf -m policy/modules.conf -x doc/policy.xml
support/sedoctool.py exiting for: Error while parsing xml
make: *** [Makefile:415: conf.intermediate] Error 1
* ERROR: sec-policy/selinux-base-2.20200818-r2::coreos failed (configure phase):
* emake failed
Try to fix this by forcing a sequential build.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
PR https://github.com/flatcar-linux/coreos-overlay/pull/432 started
to replace `dev-lang/rust` in accept_keywords with its new version.
However, its corresponding `virtual/rust` has never been updated.
That issue had been hidden until
4463efcfd4
started adding `virtual/rust` to accept_keywords.
Unlike `dev-lang/rust`, keywords for `virtual/rust` stayed with old
versions. As a result, subsequent Github Actions PRs for rust become
all invalid, so build failures.
Fix the issue by replacing versions of `virtual/rust` with new versions.
Also try to match with version specifiers, not only `=` but also `>=`,
'<=', '~'.
trousers supports TPM 1.2, and fails for TPM 2. This commits
skips the tcsd service if TPM 2 is detected.
Signed-off-by: Sayan Chowdhury <schowdhury@microsoft.com>
with the OpenSSLv3 upgrade, `update_engine` is not fully compatible yet.
See the associated issue for more details.
Let's keep the deprecated SHA functions in the meantime to run the
build.
Signed-off-by: Mathieu Tortuyaux <mtortuyaux@microsoft.com>
trousers supports TPM 1.2, and fails for TPM 2. This commits
skips the tcsd service if TPM 2 is detected.
Uses ConditionSecurity introduced in systemd v248
Fixesflatcar-linux/Flatcar#208
Signed-off-by: Sayan Chowdhury <schowdhury@microsoft.com>
For HTTPS to work and also for it to deliver the security promises we
need to ship the latest certificate database.
Update the package version symlink in oder to fetch the database from
the newest NSS release under
https://ftp.mozilla.org/pub/security/nss/releases/
and do a "ebuild ca-certificates-3.70.ebuild manifest" run.
Signed-off-by: Guillaume Perrin <guillaume28.perrin@gmail.com>
commit 5c4d184e22fd93ab926878a131150047b54d0b6c
Author: Michael Marineau <michael.marineau@coreos.com>
Date: Fri Aug 1 14:48:59 2014 -0700
polkit: fix config install paths, use systemd-tmpfiles
All configs should be installed to /usr and tmpfiles should be used to
create and fix directory permissions instead of the ebuild's postinst.
For HTTPS to work and also for it to deliver the security promises we
need to ship the latest certificate database.
Update the package version symlink in oder to fetch the database from
the newest NSS release under
https://ftp.mozilla.org/pub/security/nss/releases/
and do a "ebuild ca-certificates-3.69.1.ebuild manifest" run.
Hgfs-mounter has been dropped from the repository and it let's make the
patch name independent of the package version so that the patch doesn't
have to be touched on every upgrade.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
systemd v249 changes the usual failed units "●" to show "×".
This commit adapts accordingly to display the correct failed units
For compatibility with the longer-cadence channels, we continue to
support "●"
Signed-off-by: Sayan Chowdhury <schowdhury@microsoft.com>
Automatically update coreos/open-vm-tools as well as
coreos-base/oem-vmware.
Get the latest open-vm-tools release number, and get its build number
from the Github repo, and replace the old build number with the new one.
Also sync coreos-base/oem-vmware in line with open-vm-tools.
We need to split the beginning of setting up the top-level git repo into
a new function prepare_git_repo, and call it in the beginning of each
script. That is to prevent some corner cases, where applying multiple
patches does not work because the latter overwrites the former patch.
So we should not set up the git repo again in each apply_patch, but only
in the beggining, prepare_git_repo.
`ebuild audit-2.8.5-r1.ebuild manifest` fails like that:
```
>>> Downloading
'017e6c6ab9.patch'
--2021-09-29 04:05:09--
017e6c6ab9.patch
Resolving github.com... 140.82.121.3
Connecting to github.com|140.82.121.3|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 854 [text/plain]
Saving to: /mnt/host/source/.cache/distfiles/audit-017e6c6ab95df55f34e339d2139def83e5dada1f.patch.__download__
2021-09-29 04:05:09 (57.3 MB/s) -
/mnt/host/source/.cache/distfiles/audit-017e6c6ab95df55f34e339d2139def83e5dada1f.patch.__download__ saved [854/854]
!!! Fetched file:
audit-017e6c6ab95df55f34e339d2139def83e5dada1f.patch VERIFY FAILED!
!!! Reason: Filesize does not match recorded size
!!! Got: 854
!!! Expected: 852
Refetching... File renamed to
'/mnt/host/source/.cache/distfiles/audit-017e6c6ab95df55f34e339d2139def83e5dada1f.patch._checksum_failure_.o2889wwd'
!!! Couldn't download 'audit-017e6c6ab95df55f34e339d2139def83e5dada1f.patch'. Aborting.
```
That happens because the upstream audit patch
017e6c6ab9.patch
silently changed, so it could have a git commit of 8-bytes instead 7.
Fix the hash in Manifest for now, until we could update
sys-process/audit to 3.0. Upstream Gentoo already has 3.0, dropped 2.8.
However, updating to 3.0 might not so trivial due to Flatcar changes in
audit.
The bug fix https://github.com/flatcar-linux/coreos-overlay/pull/1129
caused a regression that Github Actions cannot determine a correct
$VERSION_OLD if the old ebuild file has a suffix like `-r1`.
We need to create a function to get a correct ebuild file name, by
falling back to the most similar name, in case the expected ebuild
file does not exist.
When the GnuPG keyserver is set to `keys.openpgp.org`, `gpg --recv-keys`
occasionally fails with the following error:
```
gpg: key E52F0DB391453C45: no user ID
```
We need to make GnuPG accept keys even without UIDs.
Original patches come from
f292beac11/debian/patches/import-merge-without-userid .
See also https://dev.gnupg.org/T4393 .
Based on commit 3d9a9c9c3654c6b8c073e306636bf8dc64cfb657 .
Update app-crypt/gnupg to 2.2.29.
One of the key purposes for the update is to be able to use the new
default keyserver `keyserver.ubuntu.com`, which is provided by default
since 2.2.29. It is due to the shutdown of the SKS keyserver pools.
See also https://bugs.gentoo.org/811828 .
I think we still prefer to keep packages in portage-stable and
sometimes add an entry to the accept_keywords file instead of moving
the package to overlay just to edit a keyword. Or a PYTHON_COMPAT
field.
This changes comes together with the change made in portage-stable to
one of the python eclasses where we add support for python3 version
from 3.8 to 3.10. To make this change complete, we need to mask those
new versions, so building packages will not try to depend on python
version we haven't yet packaged.
with the recent update of `dev-lang/perl`, we added the `minimal`
useflag.
This one is not taken in account from `package.use` into the stage 2 of the boostraping,
because we do an `export USE=...`.
Following the precedence of the USE flag with Gentoo, the `export` will
be used in first, so the `package.use` with our `dev-lang/perl minimal`
won't be used.
Signed-off-by: Mathieu Tortuyaux <mtortuyaux@microsoft.com>
This seems to be still unpackaged by gentoo, so refresh it on our own
then and do some maintainance work on it:
- Update to the recent EAPI, replacing autotools-utils with autotools,
and systemd_get_unitdir with systemd_get_systemunitdir.
- Add a patch from upstream fixing the pkg-config detection, so our
hack during configure phase is not necessary any more.
- Patch the configure script to put the D-Bus policy files in
/usr/share instead of /etc. This removes a need for a hack in the
install step.
This is to get rid of EAPI 5 in the package.
We are going to update it to a newer version, which fixes the build
system issue that was a reason for putting the package in overlay.
This is to get rid of EAPI 5 in the package.
Incase the ebtables tables are not set, the ebuilds links the
ebtables binaries to the legacy version instead of the nft version
Moving to coreos-overlay to link it to xtables-nft-multi.
The next step could be upstream the patches, incase of the usage of
nftables USE flag.
Signed-off-by: Sayan Chowdhury <schowdhury@microsoft.com>
We had an accept_keywords for versioned rust in base profile already,
but it was outdated due to a bug in github action. So update it and
move the lines from sdk target to base profile. The accept_keywords
for virtual/cargo package are dropped, because there is no such
package.
Replace any dev-lang/rust version with the current one, and make sure
that the modified files in the profiles directory is actually included
in the patch.
This is a backport of https://github.com/SSSD/sssd/pull/5748 adapted to 2.3.1.
A change was necessary: src/tools/sssctl/sssctl_logs.c wasn't passing
'--no-create' to truncate in 2.3.1 yet.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
- Drop the init.d files.
- Remove the socket unit's rate limiting.
Instead of dropping bindist, enable it with the profiles now so it
doesn't need to be modified on future updates.
Imported commit 20d298fb282ec9d5a060f12aef64c47aede0904d .
Update net-misc/openssh to 8.7_p1-r1, mainly to address CVE-2020-15778.
Goal of the package update is to add the support of a new option `-s`
of scp, i.e. "sftp mode of scp". Openssh 8.7 started to support the
flag, but it is disabled by default. So at the moment users need to
explicitly run `scp -s` to test the feature.
Gentoo ref: 11d6f23704e7ab84191e28e034816bfdb151d406
Now that we started encoding strings to unicode by default,
we should also take care of corner cases, where LC_CYPTE is set to a
different value from the systemd default value in `/etc/locale.gen`.
For example, under a build environment with `LC_CTYPE=C`, when the UTF-8
file name is `AC_Ra�z_Certic�mara_S.A..pem`,
build fails like that.
```
Traceback (most recent call last):
File "/var/tmp/portage/app-misc/ca-certificates-3.27.1-r2/files/certdata2pem.py",
line 127, in <module>
f = open(fname, 'w')
UnicodeEncodeError: 'ascii' codec can't encode character '\xed' in position 5: ordinal not in range(128)
* ERROR: app-misc/ca-certificates-3.27.1-r2::coreos failed (compile phase):
```
To fix that, encode filename with system encoding when opening the file.
This package contained no Flatcar modifications, so in theory it could
be moved to portage-stable. But we also will want to update it to some
recent version that does not depend on python2. But the recent
versions in gentoo use python3.{7,9}, so we will need to change it for
now, since we still use python3.6.
WALinuxAgent falls back to using the `distro` module to figure out the
distribution details in case the `get_linux_distribution` function
from the builtin `platform` module is not able to do it. With the
update of python-oem to python3, the distribution detection broke,
because we stopped carrying a patch that implemented fetching the
distribution information from `/etc/os-release`. It does not make
sense to backport that patch though, because
`platform.get_linux_distribution` is deprecated and removed in python
3.7 or 3.8. So when we update python3 to the newer version, we would
need to add the `distro` module anyway.
Maybe we can drop `distro-oem` module in future, when python-oem will
use version 3.10 and WALinuxAgent starts using the newly added
functionality in 3.10 to figure out the distribution information.
- unmask amd64 and arm64
- remove tmpfiles from ebuild inherit so we don't run into a circular
dep with systemd, use systemd_tmpfilesd instead
- take care of nscd.conf via systemd_tmpfilesd,
add files/nscd-conf.tmpfiles.
- Don't run sanity checks in pkg_pretend to prevent gcc checks when
only the binary package is installed.
- comment out 'dostrip -x' to force the OS image binaries to be stripped
- remove everything glibc wants to put under /etc since we use
baselayout to provide that
Add flatcar specific changes to the build recipe.
Move PYTHON_DEPS to DEPEND so things can build.
Don't run sanity checks in pkg_pretend
(similar change as in glibc-2.29) to prevent
gcc checks when only the binary package is installed.
Based on commit f7a8cd5f1fcc.
Signed-off-by: Thilo Fromm <thilo@kinvolk.io>
Signed-off-by: Dongsu Park <dongsupark@microsoft.com>
Now that sys-apps/policycoreutils is pulled in explicitly for both
architectures, we should be able to pull in its dependencies, e.g.
sys-apps/semodule-utils, sys-libs/libselinux, sys-libs/libsemanage,
sys-libs/libsepol. In case of arm64, however, all the ebuilds have
only `~arm64`. So we need to enable the keywords for the ebuilds.
Without the changes, build fails like:
```
!!! All ebuilds that could satisfy
">=sys-libs/libselinux-3.1:=[python?,python_targets_python3_6(-)?,-python_single_target_python3_6(-)]"
for /build/arm64-usr/ have been masked.
!!! One of the following masked packages is required to complete your
request:
- sys-libs/libselinux-9999::coreos (masked by: missing keyword)
- sys-libs/libselinux-3.2::coreos (masked by: ~arm64 keyword)
- sys-libs/libselinux-3.1-r1::coreos (masked by: ~arm64 keyword)
```
Now that Kernel config `CONFIG_ICE` is enabled, its corresponding
firmware file needs to be also in place. However, upstream
linux-firmware tarball does not contain a correct symlink to
`intel/ice/ddp/ice-1.3.26.0.pkg`, but `modinfo ice.ko` shows it
requires `ice.pkg`. So we need to create the symlink to avoid failures
at the firmware scanning stage like below:
```
Missing firmware: intel/ice/ddp/ice.pkg (ice.ko.xz)
```
The image contents are defined by the list in this package and the
dependencies pulled in. Once we would lose some dependency due to
a package change, that would also meant that this dependency's
binaries are not available to the user anymore. To prevent user
binaries from being lost we have to explicitly list them in this
package.
Add the packages that have binaries relevant to the user and are
currently installed (seen in flatcar_production_image_packages.txt
and checked manually). Also add sys-apps/acl which got lost when
removing rkt.
This pulls in
https://github.com/kinvolk/init/pull/47
to randomize OEM filesystem UUID if mounting fails, and to avoid trying
to install the QEMU qcow2 images.
Current cross builds of perl segfault on simple operations such as `perl -V`.
This appears to be due to the cross-build not getting `-fwrapv -fno-strict-aliasing`
passed from the configure script. While we try to get this fixed upstream, we
can monkeypatch our old version of perl to fix this.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
This change results in building the pam_tty_audit additionally, nothing else.
Related to https://github.com/kinvolk/Flatcar/issues/485.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
It produces files with the same contents as the python2 version of the
script, but the filename handling is a bit different wrt. filenames
with weird, non-unicode characters. But overall, it does not affect
anything.
This change adds the "slirp" use flag to qemu (SDK only), enabling
qemu's user networking. This fixes a bug where qemu is unable to start
the Flatcar qemu image:
$ ./flatcar_production_qemu.sh
qemu-system-x86_64: Parameter 'type' expects a netdev backend type
The issue has been discussed on the qemu mailing list:
https://www.mail-archive.com/qemu-devel@nongnu.org/msg786275.html
Signed-off-by: Thilo Fromm <thilo@kinvolk.io>
It contained some chromium version of flatcar scripts, from which we
were using the common.sh script in the cros-workon script (from the
now-removed coreos-base/cros-devutils package). It's not used any more
- we updated flatcar scripts to call into its internal copy of
cros-workon.
The package contained scripts that are not used in our workflow, are
unmaintained by us for a number of years now and it presents an
obstacle in porting the packages to python3.
Our scripts are using cert-to-efi-sig-list and flash-var from
efitools, and sbsign from sbsigntools. Currently the cros-devutils
package is pulling in the efitools package, which in turn pull in the
sbsigntools package.
We plan to drop the cros-devutils package, so better be explicit about
the dependencies.
We dropped the installation of the gmerge script a while ago, because
it was not used anywhere and that made one script less to port to
python3 at the time of updating portage to a recent version. Now we
only install the emerge-gitclone script, so rename the package to
reflect this fact.
- Drop binddist from RESTRICT variable
- Drop pkg_postinst
- Create /etc/ssl with tmpfiles (and package it for the SDK).
Signed-off-by: Sayan Chowdhury <sayan@kinvolk.io>
These are no longer used by anything in the tree, after removing old versions
of docker-runc/docker-proxy/containerd.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
Upstream builds with go1.16. Additionally fixup the VERSION variable specify the
current Flatcar Docker version 20.10.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
Upstream builds go through github.com/docker/docker repo and that builds
with go1.16 with module support disabled.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
I'm not sure if we ever built it - it's not a dependency of anything
at all. Also one of its dependencies, dev-python/pyxenstore, was
dropped in 2014, so the package is broken for about seven years
now. Looks like that the rackspace oem package is rather pulling
nova-agent through the coreos-base/nova-agent-container package.
The containerd config works in mysterious ways - sometimes it acts hierarchical
with respect to the section headers, other times not. In this case, setting
runc.options resets all the fields of the runc section, including
'runtime_type'. Having an unset runtime_type causes containerd to fail to spawn
containers (but the daemon itself starts succesfully) returning the error:
kubelet[13148]: E0823 11:57:17.030551 13148 remote_runtime.go:116] "RunPodSandbox from runtime service failed" err="rpc error: code = InvalidArgument desc = failed to create containerd container: create container failed validation: container.Runtime.Name must be set: invalid argument"
Explicitly set the runtime_type in all containerd configs, and bump the config
version to 2.
Reported as https://github.com/kinvolk/Flatcar/issues/484
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
Now that we have dev-util/pkgconfig 0.29.2, there is no need to
keep third-party patch for avoiding cross-build issues in
dev-util/strace. Let's simply drop the patch, and move strace to
portage-stable.
Apparently the `coreos-devel/sdk-extras` was originally meant to work
as a meta package to pull in all the optional packages in the SDK at once.
It has been unmaintained since 2~3 years, so an attempt of `emerge
coreos-devel/sdk-extras` will give you a huge list of conflicts to
resolve. It is difficult to resurrect sdk-extras at the moment.
Delete `coreos-devel/sdk-extras` completely. Doing that, we can delete
more than 20 other packages from the source tree.
Now that coreos-devel/sdk-extras are gone, delete unnecessary configs
in profiles, for app-portage/repoman, dev-go/glide, dev-go/godep,
dev-python/awscli, dev-python/botocore, dev-python/s3transfer.
This version has an officially documented support for python3, so it
plays along our plans of removing python2 in favor of python3. When
the switch actually happens, we will need to update the ebuild to
mention the correct path to python modules. The path contains python
version, which is a hindrance. Would be nice to have it hidden behind
some variable.
There is also a version 2.4.0.2, but it's marked as a prerelease on
github, so decided to package 2.3.1.1 instead.
Upstream has switched to go 1.16, but still doesn't use go modules. The ebuilds
needed fixing up after the automated PR was created.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
Set PYTHON_COMPAT to python 3.6 and 3.7 to be suitable for the current
code base.
Add a custom patch to replace error with warning when running autoconf
for cross builds, because libkrb5 is not able to detect
cross-compilation.
Based on 64e33c9f826d8fd951fd58ba1ed70debaf65be8d .
The SystemdCgroup=true setting is incompatible with kubelet
cgroupDriver: cgroupfs. So to prevent kube clusters from failing, we
will be freezing a nodes config.toml during an update. For that purpose,
we install a second configuration file that can then be selected using a
systemd drop-in unit.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
Now that Docker has been updated to 20.10, we can use cgroupv2 so have
systemd mount the unified cgroup hierarchy by default. Other ways of
achieving the same would have been to pass 'systemd.unified_cgroup_hierarchy=1'
on the kernel cmdline, but this way the change propagates nicely to all
OEM consumers.
Signed-off-by: Jeremi Piotrowski <jeremi.piotrowski@gmail.com>
The upstream docker repository location has changed to docker/docker.
Additionally, the cli component has been split out which which requires
fetching two hashes and updating two ebuilds. We also took the chance to
align the ebuild with gentoo's, which means there are is no more live ebuild
and no symlink.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
We are switching flatcar to cgroupv2 which is support by docker 20.10 and
kubernetes 1.19. This requires setting the systemd cgroup driver in the kubelet
config.
Due to the unified cgroup hierarchy, kubernetes <1.19 will not work so
remove all older versions.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
Compared to previous torcx images the docker-cli package is a separate
package, following upstream Docker repo layout changes.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
The patches do the following:
* install flatcar specific wrappers and systemd config
* force some USE flags to default on
* allow injecting CFLAGS/LDFLAGS so that torcx can work
* force building with go1.13 (like upstream does) - this won't be
necessary next time because docker master already uses go1.16
This is the version needed by docker 20.10.7. ROADMAP.md doesn't exist so it
has been removed from src_install.
Signed-off-by: Jeremi Piotrowski <jeremi.piotrowski@gmail.com>
This is the version used by docker-19.03. We will be updating the live
ebuild to build docker 20.10 dependencies.
Signed-off-by: Jeremi Piotrowski <jeremi.piotrowski@gmail.com>
We use coreos-go* eclass so that we can override several environment
variables and build with the same go version as docker upstream. These
changes are modeled after what was previously done in app-emulation/docker,
the cli ebuild has only been split out since v20.10.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
Docker upstream split the cli component into a separate repo, so there is
a separate ebuild that builds the docker utility. This is a prerequisite
of the update of docker to 20.10.
This is an import from portage commit 69d01a4273a556b1205a7a575cb3811ab7e2443d.
Signed-off-by: Jeremi Piotrowski <jeremi.piotrowski@gmail.com>
We use a custom build system to remove the cmake dependency and hardcode
relevant configuration.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
Now that runc version follows simple semver semantics, we do not have to
care about number of patches up to an rc version. Remove the obsolete
comments.
flannel will write into /run/flannel/... so we need to provide
correct labelling for dir created by docker daemon
Signed-off-by: Mathieu Tortuyaux <mathieu@kinvolk.io>
flannel uses an init container to pull CNI from container to the host
system in `/etc/cni`.
With SELinux, the permission is denied because `/etc/cni` is labelled
with `etc_t` so it can't be access by Docker since it expects `svirt_lxc_file_t`.
Using `filetrans_pattern` we can define a mechanism to create `/etc/cni`
with the correct labels even if it's not yet created - which avoid to
run `restorecon` on `/etc/cni`.
Signed-off-by: Mathieu Tortuyaux <mathieu@kinvolk.io>
The patches applies does not make sense to be removed, hence it would
be better to move `expat` back to portage-stable
Signed-off-by: Sayan Chowdhury <sayan.chowdhury2012@gmail.com>
- unmask amd64 and arm64
- remove tmpfiles from ebuild inherit so we don't run into a circular
dep with systemd, use systemd_tmpfilesd instead
- take care of nscd.conf via systemd_tmpfilesd,
add files/nscd-conf.tmpfiles.
- Don't run sanity checks in pkg_pretend to prevent gcc checks when
only the binary package is installed.
- comment out 'dostrip -x' to force the OS image binaries to be stripped
- remove everything glibc wants to put under /etc since we use
baselayout to provide that
Add flatcar specific changes to the build recipe.
Move PYTHON_DEPS to DEPEND so things can build.
Don't run sanity checks in pkg_pretend
(similar change as in glibc-2.29) to prevent
gcc checks when only the binary package is installed.
Based on commit 8d040f93c289.
Signed-off-by: Thilo Fromm <thilo@kinvolk.io>
Signed-off-by: Dongsu Park <dongsupark@microsoft.com>
Now that the OEM partition is a btrfs partition with compression, we have
enough space to install ssm agent.
This reverts commit b6abb59c544be13e923a3e7240b5c9395c281fca.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
The ebuild was missing a call to go_export() which exports GOARCH, and so was
always built for host architecture. While COREOS_GO_VERSION was specified as
go1.12, src_compile() has to use '${EGO}' to make use of it, so we were
building with go1.16 (latest). Upstream builds with 1.12 for this version, so
we will do the same.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
Sysroot-wrappers contains binaries installed to /usr/lib64/sysroot-wrappers,
but the profile referenced them through the 'lib -> lib64' symlink. Stop
relying on that symlink, which is not present in arm64 profiles, and is
not part of 17.1 amd64 profiles.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
With the switch from rkt to systemd-nspawn the ability for the service
to set the routing entries for the TCP load balancer got lost,
resulting in an unreachable LB as reported in
https://github.com/kinvolk/Flatcar/issues/459
The fix also reported there is to retain CAP_NET_ADMIN when starting
the service.
Btrfs filesystems do not support a non-standard 64k page size on arm64
when the filesystem was created by a 4k page size system.
Use the default page size for arm64 to ensure compatibility with
btrfs filesystems created by amd64 systems.
With the default gzip compression the 60 MB limit for the vmlinuz
bundle of kernel+initramfs was reached. The limit comes from the size
of the /boot partition which is 128 MB large and the kernel needs to
fit twice, in addition to GRUB.
Use zstd for the initramfs as it provides a similar speed but better
compression. For the kernel we can't switch yet to zstd for arm64
but for amd64 it works.
Due to unnecessary wildcard listings, ebuild files including all rc or
beta are being listed. Since `VERSION_OLD` is already generated as a
unique version, we do not need to list multiple files to filter by
running `head -n1` etc. We just need to use only the specific ebuild.
Simply list only the unique ebuild file.
Before passing runc versions to `sed '/-/!{s/$/_/}'`, we need to replace
`_` with `-`, because runc tarball files already have names like
`1.0.0_rc2`. Without the fix, version sort would `1.0.0` come before
`1.0.0_rc2`, which is not expected in the later steps.
This does not work because the host and cross rust targets share the
same name. This needs to be reworked to (potentially) enable x86 cross
targets for aarch64 targets.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
glib-utils are used during the build so they need to be part of host
dependencies for update_engine. This only really pops up during a repeat
bootstrap, when update_engine is being built from source but glib has
been installed from a binary. BDEPEND would be the correct variable but
that requires EAPI=7, so additionally added it to DEPEND for now.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
kola-data and google-cloud-sdk install pre-built amd64 binaries, so
there's no point installing them right now. Both could be made to work
at a later time. iucode and syslinux and are x86 specific and won't
build. selinux related packages *currently* don't work/build on arm64
but could be made to work.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
Where the packages are part of coreos-overlay, I keyworded the ebuilds
directly to the same level of stability as amd64. Other packages have
been keyworded through the profile, as close to the amd64 level as I
could manage.
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
