On Mo, 20.10.25 16:28, Demi Marie Obenour ([email protected]) wrote: > > LoaderDevicePartUUID reports the partition UUID of the ESP > > systemd-boot or systemd-stub first were invoked from. > > > > StubDevicePartUUID reports the partition UUID of the partition the UKI > > was invoked from (which is typically also the ESP, but could also be > > XBOOTLDR or in fact any oher partition). > > > > But note that the ESP/XBOOTLDR partition UUIDs should generally be > > understood to be as useful for identifying specific drives as the > > GPT partition table UUID, as the ESPs/XBOOTLDR have to be generated > > individually for each drive, as they are writable and recognzable by > > firmware, as they are referenced by firmware via their partition UUID > > too. > > Can some sort of ACPI path be used? The firmware uses fairly long > paths internally.
This is not realistic. How the EFI device paths are assembled varies widely from firmware to firmware. As a particularly bad example: my Intel NUC I test things on doesn't bother putting anything useful into device path, it just puts the partition UUID in there and is done with it. My guess is that it's far from the only firmware implementation doing that. (Also PCI/USB enumeration indexing might differ between Linux and firmware anyway, so the references are not as translatable as one might hope even if they are fully specified by the firmware). Lennart -- Lennart Poettering, Berlin
