Let's make sure these follow the rest of kernel-install and always
operate on the given root directory, even if the verb itself can't
support --root= just yet.
when KERNEL_INSTALL_UKIFY is not supplied we set ukify to $PWD/ukify
that will fail (perhaps only for manual installations):
FileNotFoundError: [Errno 2] No such file or directory: '/usr/src/linux-6.7-rc1/ukify'
this will make sure we have a sane default for UKIFY
Signed-off-by: Paymon MARANDI <paymon@utubeipod.xyz>
When we resolve symlinks, paths (especially filenames) may be changed,
but plugins may expect to see the kernel added under the name specified,
not under the final name that the symlink chain resolves to.
This makes symlinks in specified paths that passed to plugins are not
resolved when neither --root nor --image specified.
Fixes#29317.
This fixes a regression introduced by
42551ea7e9.
In the shell script version, plugin failures are propagated to the
caller. But after the commit, failures in plugins are logged, but never
propagated as the exit code of the execution.
Fixes#30087.
Let's make kernel-install a bit easier to use:
If the kernel version is not specified, let's imply "uname -r", so that
we regnerate the entries for the current kernel.
If the kernel image is not specified let's imply using
/usr/lib/modules/$version/vmlinuz, i.e. the location distros like Fedora
drop the kernel into, which we generally recommend people to use.
If the kernel is not found there, don't try to automatically pick the
kernel path, and fail, as before.
In mkosi, we can't use kernel-install directly but we do want to
mimick its behavior as much as possible. Let's make that easier by
making it easy to fetch data from kernel-install as JSON.
To get all the necessary data, we extend the inspect verb to also
allow passing in a kernel version and initrds, to mimick the "add"
verb. The kernel version is used to determine the "Entry Directory",
and in absence of auto-detection of kernel version in kernel-install
we have to allow users to pass it.
We also add --no-pager while we're at it.
Requested in https://github.com/systemd/systemd/pull/28582#issuecomment-1673300596.
The is the last requested changed, so fixes#28771.
90-loaderentry.install is modified to also check $KERNEL_INSTALL_CONF_ROOT
when looking for the devicetree file. For normal use this is probably not
needed, but it's nice to be consistent and it also makes it much easier to
write the tests.
In tests, also do 'ukify inspect' now that we have it.
/boot is not trusted, so we shouldn't use load files from there. Also, space in
/boot is limited, so it doesn't make sense to install the files under one
location there and then copy them to a different location. We should only copy
the files from /usr somewhere and then install it in the appropriate place under
/boot.
Also use "/usr/lib" instead of the "/lib" prefix. We don't support unmerged-user
anymore.
Addresses some of the feedback in
https://github.com/systemd/systemd/pull/28582#discussion_r1285820556.
For us, this is a compatibility mode, but most likely it is there to stay: the
kernel Makefile's install target expects to be able to call /bin/installkernel.
We want people who build their own kernels to use this, so that they use
kernel-install and get support for all the functionality provided by it,
including building of UKIs and other new features. So let's actually advertise
that this exists and works.
The notice in the man page is removed and the tool is moved into the $PATH.
A compat symlink is provided.
It is fairly widely used now, and realistically we need to keep backwards
compat or people will be very unhappy.
Like the cmdline file we look for a devicetree file in
$KERNEL_INSTALL_CONF_ROOT, /etc/kernel and /usr/lib/kernel. If it is
present we look for the specified device tree that comes with the kernel
we're adding and install it into $ENTRY_DIR_ABS and add a devicetree
stanza to the loader entry.
Unfortunately it seems there is no common consensus on where to install
device tree blobs, so we have to look in a few different places for it.
60-ukify.install would only work with initrd provided by command line
arguements. Fixed to look for both microcode and initrd is found in
$KERNEL_INSTALL_STAGING_AREA which is placed by initrd generator like
mkinitcpio
The shell script version of kernel-install silently ignored unexpected
arguments, but C version refused that. Unfortunately, Fedora's kernel
script specifies kernel file even for 'remove' command. Let's accept
extra arguments and silently ignore them to keep backward compatibility.
Fixes#28448.
Fixes https://bugzilla.redhat.com/show_bug.cgi?id=2223794.
kernel-install used to work without /proc mounted before the rewrite
in C. Let's restore that property by making sure we don't reopen
file descriptors via /proc. In this case, parse_env_file_fdv() calls
fdopen_independent() to get a FILE * for the given file descriptor
(which itself calls fd_reopen()). Let's avoid the call to
fdopen_independent() by using chase_and_fopenat_unlocked() which
gives us a FILE * immediately without having to reopen any file
descriptors.
We do not provide any way to override /etc/machine-info.
As the file is deprecated in kernel-install, let's skip to read it when
we test kernel-install.
Fixes#28011.
