Re: Support for plain dm-crypt and detached LUKS header

[Xen]

Maxim Fomin schreef op 11-02-2017 19:09:

> Well, this explains me your line of thinking. Sure, prober cannot
> detect required
> parameters for plain dm-crypt, but this is theoretically impossible,
> so why bother?
>
> Either user supplies command with dm-crypt parameters into config file
> or executes
> manually during the boot. What is the problem here?

My apologies, I suspected there was a misunderstanding, but I thought 
Andrei was mainly talking about the generation phase, and not the 
execution.

To me it seems unlikely that it would be theoretically impossible for a 
grub module to use dm-crypt, but I suspect Andrei was mainly talking 
about (a) the existing cryptomount module and (b) the generation phase.

That leaves (c) the manual configuration though a module (another 
module, perhaps) that Maxim requires (or desires) as an option, right?

Meaning, if the plain (existing) cryptomount module does not and could 
not support it, then the suggestion of another module would of course 
cover the use case, right?

I mean the plain (or existing) module.

However I believe Maxim suggests that it is hard to find technical 
grounds for this limitation in the existing cryptomount?

Regardless, either Maxim's usecase would be possible with an extra 
module (the generation phase is not required) or potentially, if at all 
possible, with the existing module.

I do not wish to meddle with your dialogue, I only sought to see if 
there was maybe some misunderstanding still.

Regards, Bart.