This means that we'll fail hard if something goes wrong, e.g. reading
of a config file. I think this is appropriate. If errors should be ignored,
the caller should do that on their end.
In practice this makes little difference, because kernel-install will
only call the plugins for 'add' or 'remove', and if we were to add a
new verb to kernel-install, we'd just change the plugins at the same
time. But our plugins serve as documentation for external plugins too,
and there it's better to silently ignore unknown verbs so that we can
add new verbs in the future.
(50-depomod.install was already like that.)
It's pretty hard to write tests without this. I started out by adding separate
variables for each of the files we read, but there's a bunch, and in practice
it's good enough to just override the directory.
'test -r' is changed to 'test -f' everywhere. If the file exists but we
cannot read it, it would be better if we fail with a permission error. E.g. if
/etc/kernel/cmdline is unreadable, and we're running something as non-root, we
shouldn't fall back to /usr/lib/kernel/cmdline. This commit doesn't resolve
this fully, because we're not running with 'set -e', but this is a preparator
step.
Now that we can distinguish the naming of the boot loader spec
dirs/files and the machine ID let's tweak the logic for suffixing the
kernel cmdline with systemd.boot_id=: let's only do that when we
actually need the boot ID for naming these dirs/files. If we don't,
let's not bother.
This should be beneficial for "golden" images that shall not carry any
machine IDs at all, i.e acquire their identity only once the final
userspace is actually reached.
This cleans up naming of boot loader spec boot entries a bit (i.e. the
naming of the .conf snippet files, and the directory in $BOOT where the
kernel images and initrds are placed), and isolates it from the actual machine
ID concept.
Previously there was a sinlge concept for both things, because typically
the entries are just named after the machine ID. However one could also
use a different identifier, i.e. not a 128bit ID in which cases issues
pop up everywhere. For example, the "machine-id" field in the generated
snippets would not be a machine ID anymore, and the newly added
systemd.machine_id= kernel parameter would possibly get passed invalid
data.
Hence clean this up:
$MACHINE_ID → always a valid 128bit ID.
$ENTRY_TOKEN → usually the $MACHINE_ID but can be any other string too.
This is used to name the directory to put kernels/initrds in. It's also
used for naming the *.conf snippets that implement the Boot Loader Type
1 spec.
This reworks the how machine ID used by the boot loader spec snippet
generation logic. Instead of persisting it automatically to /etc/ we'll
append it via systemd.machined_id= to the kernel command line, and thus
persist it in the generated boot loader spec snippets instead. This has
nice benefits:
1. We do not collide with read-only root
2. The machine ID remains stable across factory reset, so that we can
safely recognize the path in $BOOT we drop our kernel images in
again, i.e. kernel updates will work correctly and safely across
kernel factory resets.
3. Previously regular systems had different machine IDs while in
initrd and after booting into the host system. With this change
they will now have the same.
This then drops implicit persisting of KERNEL_INSTALL_MACHINE_ID, as its
unnecessary then. The field is still honoured though, for compat
reasons.
This also drops the "Default" fallback previously used, as it actually
is without effect, the randomized ID generation already took precedence
in all cases. This means $MACHNE_ID/KERNEL_INSTALL_MACHINE_ID are now
guaranteed to look like a proper machine ID, which is useful for us,
given you need it that way to be able to pass it to the
systemd.machine_id= kernel command line option.
The general approach of kernel-install was that each plugin would drop in some
files into the entry directory. But this doesn't scale well, because if we have
multiple initrd generators, or multiple initrds, each generator would need to
recreate the logic to put the generated files in the right place.
Also, effective cleanup is impossible if anything goes wrong on the way, so we
could end up with unused files in $BOOT.
So let's invert the process: plugins drop files into $KERNEL_INSTALL_STAGING_AREA,
and at the end 90-loaderentry.install DTRT with those files.
This allow new plugins like 50-mkosi-initrd.install to be significantly simpler.
kernel-install would continue after errors… We don't want this, as it
makes the results totally unpredicatable. If we didn't install the kernel
or didn't do some important part of the setup, let's just return an error
and let the user deal with it.
When looking at output, the error was often hard to distinguish, esp.
with -v. Add "Error:" everywhere to make the output easier to parse.
341890de86 made "bootctl install" create
ESP\MID, in preparation of cf73f65089 that
followed it and created 00-entry-directory.install to make ESP\MID\KVER
if ESP\MID existed ‒ this meant that "bootctl install" followed by
"kernel-install $(uname -r) /boot/vml*$(uname -r) /boot/ini*$(uname -r)"
actually installed the kernel correctly.
Later, 31e57550b5 reverted the first
commit, meaning, that now running those two commands first installs
sd-boot, but then does nothing. Everything appears to work right,
nothing errors out, but no changes are actually done. To the untrained
eye (all of them), even running with -v appears to work:
all the hooks are run, as is depmod, but, again, nothing happens.
This is horrible. Nothing in either manpage suggests what to do
(nor should it, really), but the user is left with a bootloader that
appears fully funxional, since nothing suggests a failure in the output,
but with an unbootable machine, /no way to boot it/, even if they drop
to an EFI shell, since the boot bundle isn't present on the ESP,
and no real recourse even if they boot into a recovery system,
apart from installing like GRUB or whatever.
00- is purely instrumentation for 90-,
and separating one from the other has led to downstream dissatisfaxion
(indeed, the last mentioned commit cited cited exactly that as the
reversion reason), while creating $ENTRY_DIR_ABS is only required
for bootloaders using the BLS, and shouldn't itself toggle anything.
To that end, introduce an /{e,l}/k/install.conf file that allows
overriding the detected layout, and detect it as "bls" if
$BOOT_ROOT/$MACHINE_ID ($ENTRY_DIR_ABS/..) exists, otherwise "other" ‒
if a user wishes to select a different bootloader,
like GRUB, they (or, indeed, the postinst script) can specify
layout=grub. This disables 90- and $ENTRY_DIR_ABS manipulation.
The previous approach, to strip "$MACHINE_ID/$KERNEL_VERSION" from the
end, is pretty bad and encourages this for users, which makes them
inflexible to this being modified locally
Follow-up for 1cdbff1c84.
After the commit 1cdbff1c84, each entry .conf contains
redundant slash like the following:
```
$ cat xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx-5.9.8-200.fc33.x86_64.conf
title Fedora 33 (Thirty Three)
version 5.9.8-200.fc33.x86_64
machine-id xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
options root=/dev/nvme0n1p2 ro rootflags=subvol=system/fedora selinux=0 audit=0
linux //xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/5.9.8-200.fc33.x86_64/linux
initrd //xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/5.9.8-200.fc33.x86_64/initrd
```
Similarly to "setup" vs. "set up", "fallback" is a noun, and "fall back"
is the verb. (This is pretty clear when we construct a sentence in the
present continous: "we are falling back" not "we are fallbacking").
I happen to have a machine where /boot is not a separate mountpoint,
but rather just a directory under /. After upgrade to recent Fedora,
I found out that grub2 can't find any new kernels.
This happens because loadentry script generates kernel and initrd file
paths relative to /boot, while grub2 expects path to be relative to the
root of filesystem on which they are residing.
This commit fixes this issue by using stat's %m to find the mount point
of a partition holding the images, and using it as a prefix to be
removed from ENTRY_DIR_ABS.
Note that %m for stat requires coreutils 8.6, released in Oct 2010.
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
As in 2a5fcfae02
and in 3e67e5c992
using /usr/bin/env allows bash to be looked up in PATH
rather than being hard-coded.
As with the previous changes the same arguments apply
- distributions have scripts to rewrite shebangs on installation and
they know what locations to rely on.
- For tests/compilation we should rather rely on the user to have setup
there PATH correctly.
In particular this makes testing from git easier on NixOS where do not provide
/bin/bash to improve compose-ability.
When booting with Fedora-Server-dvd-x86_64-30-20190411.n.0.iso,
/proc/cmdline is empty (libvirt, qemu host with bios, not sure if that
matters), after installation to disk, anaconda would "crash" in kernel-core
%posttrans, after calling kernel-install, because dracut would fail
with
> Could not determine the kernel command line parameters.
> Please specify the kernel command line in /etc/kernel/cmdline!
I guess it's legitimate, even if unusual, to have no cmdline parameters.
Two changes are done in this patch:
1. do not fail if the cmdline is empty.
2. if /usr/lib/kernel/cmdline or /etc/kernel/cmdline are present, but
empty, ignore /proc/cmdline. If there's explicit configuration to
have empty cmdline, don't ignore it.
The same change was done in dracut:
https://github.com/dracutdevs/dracut/pull/561.
"BOOT" is misleading, because it sounds like this refers to /boot or $BOOT,
when in fact it refers to some subdirectory. Those variable names are purely
interal, so we can change them. $BOOT_DIR_ABS was used in NEWS, but it should
not be (because it is an internal detail), so the old NEWS entry is reworded to
use "entry directory".
This makes it easier to see what is going on. Documentation for
--verbose and --help is added to the man page. Our plugins are updated
to also log a bit.
* kernel-install: fix initrd when called as installkernel
Running make install from the kernel runs e.g.:
installkernel 4.20.5 arch/x86/boot/bzImage System.map "/boot"
Since 0912c0b80e this would
cal 90-loaderentry.install with those arguments:
add 4.20.5 /boot/... arch/x86/boot/bzImage System.map "/boot"
The two last arguments would then be handled as the initrd files.
As System.map exists in current directory but not in /boot/...
it would get copied there, and used as initrd intead of the initrd
which has been generated by dracut.
With this change, nothing changes when kernel-install is called
directly, but when it's called as installkernel, we now pass
thos arguments to 90-loaderentry.install:
add 4.20.5 /boot/... arch/x86/boot/bzImage initrd
initrd is thus detected as the file to use for the initrd, and as it
exists, nothing is copied over and the initrd line generated is
consistent with what one would expect
* kernel-install: fix dracut initrd detection when called directly
This brings back the systemd 240 behaviour when called directly too
* kernel-install: unify initrd fallback
* kernel-install: move initrd fallback handling to 90-loaderentry.install
* kernel-install: move initrd fallback just before creating loader entry
Instead of having just a single INITRD field, add support for all
additional parameters being INITRD fields in order.
Signed-off-by: Mike Auty <mike.auty@gmail.com>
The current support in kernel-install for initrd images doesn't copy
over the initrd file or allow a means for it to be specified (it
requires a specific filename in a particular directory).
This patchset adds support for (optionally) providing the name of
initial ramdisk file to copied over and used by kernel-install.
This makes two changes:
1. When called for "remove" any drop-ins with "+" suffix are removed
too, so that the logic works for entries with boot counting enabled
too and we don't lose track of configuration snippets created that
way.
2. When called for "add" we optionally generate a "+" suffix, based on
the data in /etc/kernel/tries if it exists.
This basically means after "echo 5 > /etc/kernel/tries" any installed
kernels will automatically set up for 5 boot tries before older
kernels will be tried.
The script shouldn't rely on a previous script exiting with a status code
that prevents it to be executed. Instead, should check if the destination
directory for the BLS kernel image exists and exit otherwise.
Some .install plugins does not require that machine ID is set such as
20-grubby.install for Fedora and 50-depmod.install.
To support such plugins to run without valid machine-id, this commit
makes the following change:
* if /etc/machine-id is missing or empty, create temporary directory
and set its path to BOOT_DIR_ABS,
* run the .install helpers with KERNEL_INSTALL_MACHINE_ID environment
variable that'd be empty if /etc/machine-id is missing or empty.
This may be useful for installing kernel for e.g. stateless systems
which initialize machine-id while booting the systems.
A recent commit (2f3a215) changed the parsing of /proc/cmdline to use a
shell array. Unfortunately, this introduced a bug: "read -ar line"
populates the shell variable $r, not $line. This breaks installation of
new loader entries:
# kernel-install add 3.17.1-304.fc21.x86_64 \
/boot/vmlinuz-3.17.1-304.fc21.x86_64
Could not determine the kernel command line parameters.
Please specify the kernel command line in /etc/kernel/cmdline!
This commit alters the read command to correctly populate the $line
array instead.
If /etc/kernel/cmdline is missing or empty, we read /proc/cmdline and
want to filter out the initrd line. Due to a bug, the whole contents was
filtered out.
The file should have been in /usr/lib/ in the first place, since it
describes the OS container in /usr (and not the configuration in /etc),
hence, let's support os-release files in /usr/lib as fallback if no
version in /etc exists, following the usual override logic.
A prior commit already enabled tmpfiles to create /etc/os-release as a
symlink to /usr/lib/os-release should it be missing, thus providing nice
compatibility with applications only checking in /etc.
While it's probably a good idea if all apps check both locations via a
fallback logic, it is only necessary in the early boot process, as long
as the /etc/os-release symlink has not been restored, in case we boot
with an empty /etc.
Do the depmod in the kernel-install hooks, so hooks can produce/install
kernel modules and be part of the depmod.
Also move the basic boot loader entry creation and removal to a
plugin script.
If PRETTY_NAME is not defined in /etc/os-release, fallback to
PRETTY_NAME="Linux $KERNEL_VERSION".
Add documentation for everything in the man page.
наб
Zbigniew Jędrzejewski-Szmek
Michael Biebl
Lennart Poettering
Yu Watanabe
Kir Kolyshkin
Jörg Thalheim
Marc-Antoine Perennou
Mike Auty
Javier Martinez Canillas
Harald Hoyer
Michael Chapman
Tom Gundersen