Re: QAPI schema for desired state of LUKS keyslots

[Markus Armbruster]

Max Reitz <address@hidden> writes:

> On 25.02.20 17:48, Markus Armbruster wrote:
>> Max Reitz <address@hidden> writes:
>> 
>>> On 15.02.20 15:51, Markus Armbruster wrote:
>>>> Review of this patch led to a lengthy QAPI schema design discussion.
>>>> Let me try to condense it into a concrete proposal.
>>>>
>>>> This is about the QAPI schema, and therefore about QMP.  The
>>>> human-friendly interface is out of scope.  Not because it's not
>>>> important (it clearly is!), only because we need to *focus* to have a
>>>> chance at success.
>>>>
>>>> I'm going to include a few design options.  I'll mark them "Option:".
>>>>
>>>> The proposed "amend" interface takes a specification of desired state,
>>>> and figures out how to get from here to there by itself.  LUKS keyslots
>>>> are one part of desired state.
>>>>
>>>> We commonly have eight LUKS keyslots.  Each keyslot is either active or
>>>> inactive.  An active keyslot holds a secret.
>>>>
>>>> Goal: a QAPI type for specifying desired state of LUKS keyslots.
>>>>
>>>> Proposal:
>>>>
>>>>     { 'enum': 'LUKSKeyslotState',
>>>>       'data': [ 'active', 'inactive' ] }
>>>>
>>>>     { 'struct': 'LUKSKeyslotActive',
>>>>       'data': { 'secret': 'str',
>>>>                 '*iter-time': 'int } }
>>>>
>>>>     { 'struct': 'LUKSKeyslotInactive',
>>>>       'data': { '*old-secret': 'str' } }
>>>>
>>>>     { 'union': 'LUKSKeyslotAmend',
>>>>       'base': { '*keyslot': 'int',
>>>>                 'state': 'LUKSKeyslotState' }
>>>>       'discriminator': 'state',
>>>>       'data': { 'active': 'LUKSKeyslotActive',
>>>>                 'inactive': 'LUKSKeyslotInactive' } }
>>>
>>> Looks OK to me.  The only thing is that @old-secret kind of works as an
>>> address, just like @keyslot,
>> 
>> It does.
>> 
>>>                              so it might also make sense to me to put
>>> @keyslot/@old-secret into a union in the base structure.
>> 
>> I'm fine with state-specific extra adressing modes (I better be, I
>> proposed them).
>> 
>> I'd also be fine with a single state-independent addressing mode, as
>> long as we can come up with sane semantics.  Less flexible when adding
>> states, but we almost certainly won't.
>> 
>> Let's see how we could merge my two addressing modes into one.
>> 
>> The two are
>> 
>> * active
>> 
>>   keyslot     old-secret      slot(s) selected
>>   absent      N/A             one inactive slot if exist, else error
>>   present     N/A             the slot given by @keyslot
>
> Oh, I thought that maybe we could use old-secret here, too, for
> modifying the iter-time.

Update in place is unsafe.

>                           But if old-secret makes no sense for
> to-be-active slots, then there’s little point in putting old-secret in
> the base.
>
> (OTOH, specifying old-secret for to-be-active slots does have a sensible
> meaning; it’s just that we won’t support changing anything about
> already-active slots, except making them inactive.  So that might be an
> argument for not making it a syntactic error, but just a semantic error.)

Matter of taste.  I like to keep simple things syntactic, and thus
visible in introspection.

> [...]
>
>> Note we we don't really care what "inactive, both absent" does.  My
>> proposed semantics are just the most regular I could find.  We can
>> therefore resolve the conflict by picking "active, both absent":
>> 
>>   keyslot     old-secret      slot(s) selected
>>   absent      absent          one inactive slot if exist, else error
>>   present     absent          the slot given by @keyslot
>>   absent      present         all active slots holding @old-secret
>>   present     present         the slot given by @keyslot, error unless
>>                               it's active holding @old-secret
>> 
>> Changes:
>> 
>> * inactive, both absent: changed; we select "one inactive slot" instead of
>>   "all slots".
>> 
>>   "All slots" is a no-op when the current state has no active keyslots,
>>   else error.
>> 
>>   "One inactive slot" is a no-op when the current state has one, else
>>   error.  Thus, we no-op rather than error in some states.
>> 
>> * active, keyslot absent or present, old-secret present: new; selects
>>   active slot(s) holding @old-secret, no-op when old-secret == secret,
>>   else error (no in place update)
>> 
>> Can do.  It's differently irregular, and has a few more combinations
>> that are basically useless, which I find unappealing.  Matter of taste,
>> I guess.
>> 
>> Anyone got strong feelings here?
>
> The only strong feeling I have is that I absolutely don’t have a strong
> feeling about this. :)
>
> As such, I think we should just treat my rambling as such and stick to
> your proposal, since we’ve already gathered support for it.

Thanks!