Comment 53 for bug 1899308

Hi,

Chris Guiver (guiverc) wrote:
> (1) wiped thumb-drive
> (2) `mkusb-plug groovy-desktop-amd64.iso` (option N or non-persistent
> was selected)

Is (2) equivalent to a plain copy by dd or cp ?

> I booted again (still same thumb-drive) without issue on
> - hp dc7700
> and booted successfully on hp dc7900

Did they previously fail with plain copies of the Ubuntu ISOs ?

If so, then it could be that step (1) erased an old GPT backup at the
end of the USB stick, which made the whole stick suspicious when there
was a valid but not matching GPT at its start.
With MBR partition table in the ISOs, there would have been no need to
look for a backup GPT at the end.

This theory would be wrong if the plain copy methods included a wipe
of the last block of the USB stick, as i let xorriso-dd-target do.

@sudodus: Can you confirm or refute this theory from the reports in
this thread ?

If the theory stays valid:
Have those machines, which demanded a boot flag, been tested with no GPT
backup at the end of the USB stick ?

It could be that the boot flag keeps EFI from looking for a backup GPT
and thus from taking offense if it does not match the main GPT.
(I confess that this is far fetched. But we already went through most of
the not so wayward possibilities.)

---

Always valid is the theory about hardware allergies. It is not explainable
to me why my "128 GB" Intenso Ultra Line rev 1.00 does not work in the
USB 3 sockets of my new machine, although it was sold and is advertised as
USB 3.0 device.

Have a nice day :)

Thomas