Accepted: cryptsetup 2:1.0.6-1ubuntu1 (source)

[Reinhard Tartler]

Martin Pitt <martin.pitt at ubuntu.com> writes:

> Hi Reinhard,
>
> Reinhard Tartler [2008-05-25 21:10 -0000]:
>>    * Merge new debian version. Remaining changes:
>
> Thanks for the merge!

No problem!

Btw, I did this merge by heavily using bzr. For the interested parties,
I did (manually) maintain an upstream and debian branch to import and
merge version. It was pretty convenient for being able to instantly
compare files to the lastest debian package. If someone wants to have a
look, I'm happy to share my small shell functions that assist with that.

>
>>      - cryptsetup is linked dynamically against libgcrypt and libgpg-error.
>>        This will break systems where /usr is a separate encrypted filesystem
>>        but not have other bad consequences (in particular, systems with
>>        encrypted root are still fine).  The upsides include better
>>        security supportability and smaller packages.
>
> How so? For that very reason we ship these libraries in /lib?

Right, this is a copy and paste error, I have copied over the
description from the previous merge. Since the libraries have moved now,
this note is wrong. Shall I fixup the changelog entry?

>>      - stop usplash on user input. LP #62751
>>      - debian/cryptdisks.functions: Always output and read from the console.
>>        LP #58794.
>
> Hm, password input for the root fs under usplash works fine. What's
> the reason why we need to keep using the console for further
> passwords?

We currently have an inconsistent behavior wrt asking for password from
within initramfs (which uses usplash) and from the init scripts that is
run during boot. The reason for this is that switching VT causes usplash
to exit. When this is done during the password entry, the user has no
longer the chance to enter his password. (LP #203683).

TBH, I think the behavior should be consistent in the two cases. Since
LP #203683 is pretty severe in my opinion, we should quit usplash before
asking for passphrases in both cases. In fact, I uploaded such an
usplash package in the past. However it was decided that if that happens
for the root filesystem, the system is not able to boot and the user
needs to restart anyway, so this may not be as severe. This is why we
ended up with this inconsistent behaviour.

-- 
Gruesse/greetings,
Reinhard Tartler, KeyID 945348A4