UEFI Debian RT worker nodes fail to boot post-install

[Bob Church]

Brief Description
-----------------
Computes are stuck in PXE reboot loop, because they can't boot from disk.

Severity
--------
Critical

Steps to Reproduce
------------------
Reinstall UEFI H/W lab, configuring a low-latency worker

Expected Behavior
-----------------
Computes install successfully

Actual Behavior
---------------
workers are in a reboot loop.

Reproducibility
---------------
100%

System Configuration
--------------------
Standard IPv6 lowlatency.

Load info
---------
Any recent master branch build

Last Pass
---------
??

Timestamp/Logs
--------------
compute-0 reboots after kickstarts finish, but fails booting from disk, so it will PXEboot.

Booting from starlingx
Boot Failed: starlingx
Booting from CentOS
Boot Failed: CentOS
Booting from PXE Device 1: Integrated NIC 1 Port 1 Partition 1
>>Start PXE over IPv4.
  Station IP address is 192.168.202.242
  Server IP address is 192.168.202.1
  NBP filename is EFI/BOOT/bootx64-nosig.efi
  NBP filesize is 765952 Bytes
 Downloading NBP file...
  Succeed to download NBP file.
Welcome to GRUB!
error: no such device: ((tftp,192.168.202.1)EFI/BOOT)/EFI/BOOT/grub.cfg.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/starlingx/metal/+/859347

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.opendev.org/c/starlingx/metal/+/859347
Committed: https://opendev.org/starlingx/metal/commit/0b614831c84768013645d07151619a94487f343c
Submitter: "Zuul (22348)"
Branch: master

commit 0b614831c84768013645d07151619a94487f343c
Author: Robert Church <email address hidden>
Date: Mon Sep 26 01:52:13 2022 -0500

    Debian: UEFI RT pxeboot installs need efi=runtime option

    Add the efi=runtime kernel command line option for all lowlatency UEFI
    installs so that the efibootmgr has access to EFI non-volatile
    variables. LAT will run the efibootmgr as one of that last steps of the
    ISO/pxeboot install to change the device boot order of the host.

    When the PREEMPT_RT kernel is enable this is required otherwise we will
    rely on the existing boot order of the system, often resulting in a
    pxeboot loop.

    As reference here is the difference in the std and rt kernel requiring
    this change:

        $ diff linux-yocto-{std,rt}/drivers/firmware/efi/efi.c
        69c69
        < static bool disable_runtime;
        ---
        > static bool disable_runtime = IS_ENABLED(CONFIG_PREEMPT_RT);
        98a99,101
        >
        > if (parse_option_str(str, "runtime"))
        > disable_runtime = false;

    Test Plan:
    PASS - Added parameter on a UEFI RT worker install in a H/W lab and
           observe that the previously failing post install reboot now
           boots successfully.

    Signed-off-by: Robert Church <email address hidden>
    Closes-Bug: #1990895
    Change-Id: Iff41e1125bb5a6e27f7de92862da1ef4899de794