Created attachment 293439 [details]
If genkernel doesnt find *the expected* key device at the address specified, it needs manual intervention. However it would be perfect to autoprobe for the key device at other locations. It might have changed the address /dev/sdX for /dev/sdY due to a new device plugged in or other reasons disturbing the order.
genkernel already autoprobes for the key device if the specified device doesnt exist.
The proposed patch makes sys-kernel/genkernel fallback to autoprobe in the following cases:
1. The specified key device exists but fails to mount for any reason.
2. The specified key device exists, is mounted successfully, but without the expected key on it.
If autoprobing doesnt work neither, it still asks for user intervention...
I have written and tested thoroughly the proposed patch.
Sebastian, what do you think of this patch proposition?
I have tested the whole combination of cases thoroughly:
- key dev available (plugged in) / absent (plugged off)
- non available block dev / non mountable (i.e. swap) key dev / mountable but other dev (i.e. root) / key dev
In all the four cases with key dev available the autodetection works, decrypts device and starts the computer.
In the other four cases where the key dev is plugged off, you will get a prompt asking for the device path. -> even if you enter a wrong path if only the key dev is available somewhere, partition is decrypted and the computer starts.
(In reply to comment #2)
Zero time until January, sorry.
ping again... I would be very happy to see the proposed patch or any code for device autodetection going in. Please give it a chance. It works well for me, probably for others too.
I know this has been ignored for a while now, but if you are willing to update the patch against genkernel git I'd really like to add this feature.