I'm suddenly booting into BIOS/UEFI instead of to 22.04 login screen

[oldfred]

I have tried renaming entries in UEFI by changing /etc/default/grub. But Ubuntu's grubx64.efi seems to be hard coded to use the /EFI/ubuntu folder.
For example my grub entry for external drive.
GRUB_DISTRIBUTOR=ssk
And I keep grub entries in 40_custom simple. Either just boot stanza or configfile.

menuentry "Ubuntu 23.04 lunar  (on /dev/sda9)" { 
    #set root=(hd1,gpt9)     
    search --set=root --label lunar  --hint hd1,gpt9 
    configfile /boot/grub/grub.cfg 
}

Using 40_custom & Custom menu
https://help.ubuntu.com/community/Grub2/CustomMenus

[watchpocket]

I would also check the voltage of the CMOS battery. A long-shot, but suspect.
There is a way to do that "programatically"... On shutdown, do

dd if=/dev/nvram of=cmos_ram_01.bin

On startup do it again to a comparison file

dd if=/dev/nvram of=cmos_ram_02.bin 
TEST=$(cmp -b cmos_ram_01.bin cmos_ram_02.bin) # This will printout any bytes that differ
if [! -z "$TEST" ]
then 
  logger "NVRAM corrupted." # This will enter a message into the syslog
fi

I am also at this point suspecting the CMOS battery, but could I maybe get some clarification about running the the above commands?

So just before shutting down, I created a cmos_ram_01.bin file with your first line above.

Then I re-started and created the cmos_ram_02.bin file immediately after starting up.


The rest is what I have trouble with. I can enter this line at my prompt:
TEST=$(cmp -b cmos_ram_01.bin cmos_ram_02.bin)

and nothing happens, except that I am returned to my shell prompt. (Maybe nothing is supposed to happen yet.)

But when I enter:
if [! -z "$TEST" ]

I am put into a secondary "if -->" prompt (which I may have long ago set up as a python prompt, something like that) and nothing else happens. I can then enter logger "NVRAM corrupted." but none of these entries return any info on the commandline.

Maybe what I need to see from this is in the syslog, but I don't recall exactly where in the file system syslog is located.

Wait -- just now opening the lnav app, it puts me into syslog_log, and I see this:

Dec 1 16:07:43 rjbox rj: NVRAM corrupted.

But that's only because I wrote "logger "NVRAM corrupted" at the prompt, right? Shouldn't I see somewhere a voltage difference?

I'm afraid I'm not clear on how to run the commands you put above.

(For what it's worth, I did this (thoug it may not tell the whole story):

--> cat /proc/driver/rtc | grep batt 
batt_status    : okay

[watchpocket]

SOLVED. It's the CMOS battery.

I always turn off the power supply as well as the motherboard. Yesterday I left the power supply on all night and booted today directly into the preferred OS install, instead of to the UEFI setup.

That means the battery got enough power overnight to keep all its memory intact.

Now I know that I just need to replace the battery soon.