Fix current EFI boot dir discovery script.
Also, adjust scripts order:
50-dracut generates initramfs in /boot/(efi/)?/$MACHINE_ID/.../initrd
80-grub2 copies it to /boot/initramfs-....img
90-xen-efi copies it to /boot/efi/EFI/qubes/initramfs-....img
Make the above order explicit, rather than relying on xen sorted later
than grub2.
QubesOS/qubes-issues#3234
Remove also EFI version of BootLoader Specification dirs. This will:
- really force to re-generate initramfs during installation, after all
relevant configs are updated; previously, dracut (called by anaconda
through kernel-install) refuse to update already existing
/boot/efi/.../initrd file.
- save some precious space in /boot/efi
FixesQubesOS/qubes-issues#3234
During installation, /usr/lib/kernel/install.d/50-dracut.install
generate initramfs in $BOOT_DIR_ABS. It is important to use that one,
even if there is one in /boot/initramfs-*.img already, because it was
generated later and contains all required config files (including
keyboard layout for entering LUKS passphrase).
This fixes d1f3be0eed "kernel-install:
avoid creating initramfs multiple times".
FixesQubesOS/qubes-issues#3234
On LVM thin it is easy to fill the pool if fstrim (or 'discard' mount
option) isn't used from time to time. Enable fstrim.timer by default,
which will do fstrim once a week.
There are multiple places where initramfs can be created:
- /boot/iniramfs-*.img
- /boot/$MACHINE_ID/.../initrd (unused on Qubes, but created by Fedora
scripts)
- /boot/efi/EFI/.../initramfs-*.img
Do not generate all of those from scratch, but try to reuse existing
image (if exists). Since one dracut call may last even 5 minutes, this
change should greatly reduce installation time.
FixesQubesOS/qubes-issues#3637
clock synchronization mechanism rewritten to use systemd-timesync instead of NtpDate; at the moment, requires:
- modifying /etc/qubes-rpc/policy/qubes.GetDate to redirect GetDate to designated clockvm
- enabling clocksync service in clockvm ( qvm-features clockvm-name service/clocksync true )
Works as specified in issue listed below, except for:
- each VM synces with clockvm after boot and every 6h
- clockvm synces time with the Internet using systemd-timesync
- dom0 synces itself with clockvm every 1h (using cron)
fixesQubesOS/qubes-issues#1230
Use UDEV_DISABLE_PERSISTENT_STORAGE_RULES_FLAG instead, which is
available since systemd 231.
- Do not merge to branches where dom0 is older than Fedora 25 -
I need to set some flags in order to boot as described here:
https://www.qubes-os.org/doc/uefi-troubleshooting/
My settings look like this:
$ efibootmgr -v
BootCurrent: 0000
Boot0000* Qubes HD(...)/File(\EFI\qubes\xen.efi)p.l.a.c.e.h.o...
which causes awk to get confused and think my $EFI_DIR should be:
/EFI/qubesp.l.a.c.e.h.o.l.d.e.r. ./.m.a.p.b.s. ./.n.o.e.x.i.t.b.o.o.t.
This causes the script to later bail:
if [ ! -d "$EFI_DIR" ]; then
# non-EFI system
exit 0;
fi
So my xen.cfg did not get new entries when installing dom0 kernel packages.
The entry may be already present for example when reinstalling package,
or calling the script multiple times (which apparently is the case
during system installation).
In non-EFI installation /boot/efi/EFI/qubes may not exists. In this case
do not try to touch (non-existing) files there.
FixesQubesOS/qubes-issues#1829
The specification doesn't cover how to boot Xen (or any other multiboot
binary), but the sole presence of such directory changes dracut default
path. So get rid of that directory.
Since we now allow using Fedora kernel, add a script to generate proper
bootloader configuration then. Standard Fedora mechanism relies on
Boot Loader Specification support in grub2, which sadly does not support
Xen, so it is useless in Qubes.
