WEBVTT

00:00.000 --> 00:10.000
I think I figured out today, this is the first time I've been in the world for a long time.

00:10.000 --> 00:14.400
All right, for the next talk, we have Richard telling us everything about how we finally

00:14.400 --> 00:16.400
fixed police encryption.

00:16.400 --> 00:18.400
And welcome Richard.

00:18.400 --> 00:22.400
Thank you very much.

00:22.400 --> 00:23.400
So, teeny bit about me.

00:23.400 --> 00:28.760
I'm a distribution architect as soon as I've been doing this open sort of stuff for way too long.

00:28.760 --> 00:32.760
I think I figured out today, this is my 13th first time, the other 12th went fine.

00:32.760 --> 00:35.760
So, I guess this is all my everything's going to go wrong.

00:35.760 --> 00:40.760
And yeah, I've been working at Suse of the 10 years and I created this distribution

00:40.760 --> 00:44.760
called M, which I will talk a little bit about.

00:44.760 --> 00:51.760
And then show how we've done for this encryption with all of the system detour chain on top of that.

00:51.760 --> 00:55.760
And then talk about some of the rough edges we have left.

00:55.760 --> 01:00.760
And then the whole plan was to have a nice Q&A open-for discussion with the handheld microphone,

01:00.760 --> 01:02.760
which of course we haven't got in this room.

01:02.760 --> 01:09.760
So, I'm just going to try and figure out a way of making that work with me copying everyone's questions for the recording.

01:09.760 --> 01:13.760
So, Aeon, I have a talk about this in the digital room and stuff to mean.

01:13.760 --> 01:18.760
But basically, it's not sure it's another one of these immutable desktop concepts.

01:18.760 --> 01:23.760
Being reliable, predictable, immutable bit like microS from the server side of things.

01:23.760 --> 01:28.760
But also being really opinionated, because this is my desktop and my daily driver.

01:28.760 --> 01:34.760
So, not only very Chromebook-like experience, image-based installation,

01:34.760 --> 01:40.760
keeping it pretty minimal for everybody else, but functional, you know, all the printing and gaming development stuff

01:40.760 --> 01:42.760
should just work out of the box.

01:42.760 --> 01:47.760
This was freshly installed last week and the presentation's working without me doing anything on it.

01:47.760 --> 01:49.760
So, that's all good.

01:49.760 --> 01:54.760
And really, we don't want people having to mess around with lots of funky configuration

01:54.760 --> 01:56.760
to get the whole thing to work.

01:56.760 --> 01:59.760
And that includes also for description.

01:59.760 --> 02:07.760
So, last year July, we did release candidate 3, which was the one where we put FTE in there,

02:07.760 --> 02:09.760
and it's enabled by default.

02:09.760 --> 02:10.760
In fact, you can't turn it off.

02:10.760 --> 02:14.760
If you're installing Aeon, you are getting full description if you like it or not.

02:14.760 --> 02:17.760
By default, if your hardware supports it,

02:17.760 --> 02:20.760
it's TPM backed and automatically unlocking.

02:20.760 --> 02:26.760
And if your hardware doesn't, we have this concept of a fallback mode where you've got to put a passphrase in.

02:26.760 --> 02:32.760
And, yeah, you know, you're unlocking it sort of old fashioned lots of way every time it boots up.

02:32.760 --> 02:35.760
It's deployed calling it bare with a tool called tick.

02:35.760 --> 02:40.760
Basically, a whole new installer that I wrote, because I felt like playing around with that kind of thing.

02:40.760 --> 02:45.760
Really nothing more than it's a fancy graphical wrapper around system D-report.

02:45.760 --> 02:47.760
If system D-report is actually doing all the hard work,

02:47.760 --> 02:50.760
deploying the image, setting up the partitions,

02:50.760 --> 02:53.760
configuring the encryption, doing all the luck stuff.

02:53.760 --> 02:56.760
Everything that's remotely important gets done by system D-report.

02:56.760 --> 02:59.760
And then all of the TPM measurements.

02:59.760 --> 03:05.760
And, because this is a user style distribution, all the stitching with BTRFS subvolumes,

03:05.760 --> 03:08.760
is done by a tool we have called SDPU-U2.

03:08.760 --> 03:18.760
So, when you are deploying A on with tick, first or, in fact, any distribution of the tick,

03:18.760 --> 03:21.760
but A on is the one you're using it right now.

03:21.760 --> 03:26.760
We probably system to figure out, does it have policy authorised NV,

03:26.760 --> 03:30.760
because that's what you need to do to get it working with PCI lock.

03:30.760 --> 03:34.760
Then also checks for secure bit.

03:34.760 --> 03:40.760
And then figure out what the heck is it going to do with those two different full description modes here.

03:40.760 --> 03:45.760
Are you going to get a nice decent validated supplied boot chain,

03:45.760 --> 03:48.760
where it's actually only going to unlock if your boot chain is good,

03:48.760 --> 03:51.760
or are you just going to be shopping in a password?

03:51.760 --> 03:55.760
So, if you have a good enough beta hardware with policy author and VNGTPM,

03:55.760 --> 04:00.760
you get default mode, pretty much any other case, you get full back mode,

04:00.760 --> 04:03.760
and if you're on a machine that doesn't have good enough hardware,

04:03.760 --> 04:07.760
and you've turned off secure boot, you get full back mode with a grumpy warning,

04:07.760 --> 04:11.760
because, you know, okay, secure boot isn't great, but it's better than nothing.

04:11.760 --> 04:17.760
So, on at least people have people some concept of having some check over their boot chain

04:17.760 --> 04:19.760
when they're starting up their system.

04:19.760 --> 04:26.760
But typically speaking on a default mode, you know, I don't care with A on if secure boot is on or off.

04:26.760 --> 04:31.760
I just care whatever you've picked stays the same because I'm measuring PCR7.

04:34.760 --> 04:39.760
Like I said, tick avoid doing as much complicated work as it can itself.

04:39.760 --> 04:42.760
So, because of these two different modes, you know,

04:42.760 --> 04:47.760
because I'm doing TPM backed and traditional full description of the path raises,

04:47.760 --> 04:52.760
it means I can't use system DV parts, instant TPM pairing.

04:52.760 --> 04:57.760
So, system DV parts using a key file, so I have tick generate the key file,

04:57.760 --> 05:04.760
and then basically I just take an image, read the system DV part config from that image,

05:04.760 --> 05:07.760
and slap it on the disk, if you chose, it's part of the install.

05:07.760 --> 05:10.760
Simple, easy works.

05:10.760 --> 05:18.760
The typical system DV part config we're using is also incredibly simple, easy straightforward.

05:18.760 --> 05:22.760
Like, I'm not the large EFI partition on purpose,

05:22.760 --> 05:26.760
because I'm kind of expecting to move to UKI's one day.

05:26.760 --> 05:31.760
So, I'd rather make every of my coven, one of the reasons I'm talking about this later,

05:31.760 --> 05:35.760
like we're on RC3 as I kept on changing how big everybody's positions were.

05:35.760 --> 05:37.760
Like, this way, I'm not going to have to deal that again.

05:37.760 --> 05:40.760
I think I'm probably fine with four gig for a while,

05:40.760 --> 05:43.760
and then everything else, I just take over the entire disk,

05:43.760 --> 05:51.760
copying the blocks from the image, and encrypting that with the key file.

05:51.760 --> 05:57.760
And then, after the image is deployed, there's a little bit of stitching to be done.

05:57.760 --> 06:00.760
Of course, the image is being built in OBS.

06:00.760 --> 06:03.760
I have had some experimental ones built with MKOSI.

06:03.760 --> 06:07.760
The cover one is actually using kiwi, just because I know kiwi better.

06:07.760 --> 06:10.760
I'm going to go back to the MKOSI one soon.

06:10.760 --> 06:16.760
But either way, it's a generic image and, of course, your system isn't a generic system.

06:16.760 --> 06:18.760
So, there's a bit of stitching required, you know,

06:18.760 --> 06:23.760
correcting the FS tab to actually include the UIDs of the partition that's just deployed,

06:23.760 --> 06:26.760
correcting the command line,

06:26.760 --> 06:32.760
yeah, populated the crypto tab, that kind of stuff,

06:32.760 --> 06:38.760
and then using FD but U2 to actually put system D but in the ESP,

06:38.760 --> 06:41.760
and the Cardinals in the ID, because we're not using UKI.

06:41.760 --> 06:47.760
So, it's just putting all of that stuff in there, separated.

06:47.760 --> 06:50.760
Then, in default mode, there's a couple of extra steps,

06:50.760 --> 06:53.760
because we're doing all this TPM pairing.

06:53.760 --> 06:57.760
So, we can figure out PCR policy, which I'll talk about a bit more later,

06:57.760 --> 07:02.760
and then update those predictions for the system, and enroll it into the TPM.

07:02.760 --> 07:08.760
So, system D, SDP, U2, just reads that config,

07:08.760 --> 07:12.760
and figures out the entries for every PCR I'm interested in,

07:12.760 --> 07:18.760
and then system D cooked and roll, nicely simply pairs it all up,

07:18.760 --> 07:21.760
and away we go.

07:21.760 --> 07:25.760
And in full back mode, of course, none of that TPM stuff needed,

07:25.760 --> 07:31.760
you just have to use it for a passphrase.

07:31.760 --> 07:35.760
I decided, when I was doing this, for a TPM environment,

07:35.760 --> 07:38.760
if you're TPM break, so you'll be a measurements don't match up,

07:38.760 --> 07:42.760
you need to have a recovery key to get into the system unlock it manually,

07:42.760 --> 07:46.760
and for feedback mode as well, I also kind of figured,

07:46.760 --> 07:49.760
you know, users can be silly sometimes and forget their own passwords,

07:49.760 --> 07:53.760
so generating a recovery key for everybody in every case all the time,

07:53.760 --> 07:57.760
and then presenting it to them with a nice long string,

07:57.760 --> 08:00.760
and a nice QR code that you can store on your phone or whatever,

08:00.760 --> 08:04.760
to remember what's going on.

08:04.760 --> 08:07.760
And then last, but definitely not least,

08:07.760 --> 08:10.760
wiping that key further used for the, yeah,

08:10.760 --> 08:14.760
to get the whole thing running, so the only keys that are enrolled in the end,

08:14.760 --> 08:21.760
are your passphrase, all the recovery key, or the TPM.

08:21.760 --> 08:24.760
And that's been how every single Ion user,

08:24.760 --> 08:28.760
including myself, has been running their laptops since July,

08:28.760 --> 08:31.760
and to be honest, it's been mostly painless.

08:31.760 --> 08:35.760
It's just working, SD but you two are doing a wonderful job of every time

08:35.760 --> 08:39.760
there's a system update updating all of the predictions.

08:39.760 --> 08:43.760
Of course, we also have automatic rollbacks and failure detection,

08:43.760 --> 08:45.760
and of course, system DB, you choose a way of that too,

08:45.760 --> 08:48.760
so it'll even update the predictions when you roll back from user space,

08:48.760 --> 08:51.760
which is really nice.

08:51.760 --> 08:55.760
As I talk about in a bit, we did mess around with the PCR policy a bit,

08:55.760 --> 08:58.760
so I did have to make it a little tool to go and check everybody's machine

08:58.760 --> 09:00.760
to fix which PCR's are being checked,

09:00.760 --> 09:05.760
but besides that, it's mostly just working.

09:05.760 --> 09:09.760
But not 100% perfect.

09:09.760 --> 09:14.760
So the first thing we had issues with is this lovely recovery key.

09:14.760 --> 09:18.760
I don't know if this is just an SD but usual thing,

09:18.760 --> 09:21.760
or also if it's system decrypt enroll,

09:21.760 --> 09:26.760
but there's a whole bunch of effort taken to only use characters

09:26.760 --> 09:29.760
which are all in the same location on everybody's keyboard,

09:29.760 --> 09:30.760
which is awesome.

09:30.760 --> 09:32.760
So, you know, German keyboard, I've got a German keyboard here,

09:32.760 --> 09:34.760
but I type in English.

09:34.760 --> 09:36.760
It's great, these French whatever,

09:36.760 --> 09:39.760
the character set is ridiculously constrained,

09:39.760 --> 09:41.760
so it's just the keys that don't move around

09:41.760 --> 09:43.760
or all these different keyboard layouts.

09:43.760 --> 09:44.760
Brilliant work.

09:44.760 --> 09:46.760
Those make the string a little bit longer,

09:46.760 --> 09:48.760
because you still want to have some entropy.

09:48.760 --> 09:49.760
That's fine.

09:50.760 --> 09:52.760
But we're using dashes to separate it,

09:52.760 --> 09:56.760
and that's the one key that moves on every single keyboard layout.

09:58.760 --> 09:59.760
Again?

09:59.760 --> 10:01.760
They have no value, they do nothing.

10:07.760 --> 10:09.760
When I drop the dashes,

10:09.760 --> 10:10.760
it fails for me.

10:10.760 --> 10:11.760
I'm doing something wrong.

10:11.760 --> 10:15.760
Well, you have to use this as a new tool to do this

10:15.760 --> 10:18.760
because they will drop the dashes for you.

10:18.760 --> 10:19.760
Okay.

10:19.760 --> 10:20.760
I've been with dashes a loop.

10:20.760 --> 10:21.760
Consider that a password.

10:21.760 --> 10:22.760
Yeah.

10:22.760 --> 10:23.760
Okay.

10:23.760 --> 10:24.760
Yeah.

10:24.760 --> 10:26.760
You can type it in upper case and lower case and drop it that.

10:26.760 --> 10:27.760
So, that's awesome.

10:27.760 --> 10:28.760
Yeah.

10:28.760 --> 10:30.760
My own implementation is the problem,

10:30.760 --> 10:31.760
which is perfect.

10:31.760 --> 10:32.760
Thank you very much.

10:32.760 --> 10:33.760
Yes.

10:38.760 --> 10:41.760
Well, I was going to just use spaces or something.

10:41.760 --> 10:42.760
Yeah.

10:42.760 --> 10:45.760
The question is why don't I just use point or something.

10:45.760 --> 10:47.760
I could just use spaces.

10:47.760 --> 10:50.760
I could also just do what system these doing with the tool.

10:50.760 --> 10:51.760
So.

10:51.760 --> 10:52.760
Yeah.

10:54.760 --> 10:55.760
I know.

10:55.760 --> 10:56.760
It's a little thing.

10:56.760 --> 11:00.760
But it's the one that my users hit like so many times.

11:00.760 --> 11:02.760
It's like, if this is the biggest problem,

11:02.760 --> 11:04.760
we have really FTE is really doing well.

11:04.760 --> 11:06.760
But this is the biggest problem we have.

11:08.760 --> 11:09.760
Yeah.

11:17.760 --> 11:18.760
Okay.

11:18.760 --> 11:21.760
What the question is, why can't you just have your native keyboard layout running?

11:21.760 --> 11:23.760
I'm, you know, we're very early in the group process here.

11:23.760 --> 11:25.760
You know, like, you haven't unlocked, you haven't unlocked your disk yet.

11:25.760 --> 11:36.760
So, you've got only got what you've got in your air, your ESP.

11:36.760 --> 11:37.760
Yeah.

11:37.760 --> 11:40.760
But when you type your key, you've generated your key.

11:40.760 --> 11:43.760
So, you've stored that key in whatever keyboard layout you've done.

11:43.760 --> 11:45.760
So, that's all fine.

11:45.760 --> 11:48.760
In this case, this has been randomly generated by us.

11:48.760 --> 11:53.760
So, if there's a mismatch, then, yeah.

11:53.760 --> 11:56.760
One thing I would really, really like is multi-factor authentication.

11:56.760 --> 12:00.760
But, you know, the only option we have for that right now is TPM and PIN.

12:00.760 --> 12:02.760
And TPM and PIN is scary.

12:02.760 --> 12:06.760
Not for anything that we control and software because we don't control that.

12:06.760 --> 12:09.760
That's really what the manufacturers do.

12:09.760 --> 12:13.760
And the other problem we have had with this has been how different manufacturers

12:13.760 --> 12:16.760
with different firmware do things somewhat subtly differently.

12:16.760 --> 12:21.760
And, you know, certainly I have someone's machine that the TPM just doesn't want to respond anymore.

12:21.760 --> 12:23.760
And they do things never unlocks.

12:23.760 --> 12:25.760
So, yeah.

12:25.760 --> 12:29.760
The kind of open question is, are we really stuck with just, you know.

12:29.760 --> 12:34.760
I know exactly what we need to do for the PIN file or stuff, which is what the Chromebooks do.

12:34.760 --> 12:36.760
It's just somebody's simple like that.

12:36.760 --> 12:39.760
So, what do the Chromebooks do?

12:39.760 --> 12:45.760
They do it really nicely because the TPM basically asks the challenge to the file.

12:45.760 --> 12:49.760
The thing, the file does a regular file or stuff like what it always does.

12:49.760 --> 12:50.760
Yeah.

12:50.760 --> 12:53.760
And then the TPM gets the signature back, right?

12:53.760 --> 12:54.760
Yeah.

12:54.760 --> 12:56.760
And then it will release the encryption key to the...

12:56.760 --> 13:00.760
And that would all be in addition to all the other PCR checks that have done.

13:00.760 --> 13:04.760
So, the comment was basically doing what the Chromebooks do.

13:04.760 --> 13:08.760
So, having the TPM challenge and the photo key separately,

13:08.760 --> 13:11.760
and then unlocking the only one both part.

13:11.760 --> 13:12.760
Yeah.

13:12.760 --> 13:13.760
Yeah.

13:13.760 --> 13:14.760
Yeah.

13:14.760 --> 13:15.760
That's fine.

13:15.760 --> 13:16.760
That's good.

13:16.760 --> 13:17.760
I'd like that.

13:17.760 --> 13:18.760
And maybe I'll find time to do it.

13:18.760 --> 13:20.760
But maybe not.

13:20.760 --> 13:22.760
And so, yes.

13:22.760 --> 13:24.760
There are PCR policy.

13:24.760 --> 13:28.760
We started initially following what opens using microRFs was doing,

13:28.760 --> 13:32.760
which is PCR 04579.

13:32.760 --> 13:34.760
Well, I think I might have added.

13:34.760 --> 13:37.760
It's about 44790 for sure.

13:37.760 --> 13:42.760
Which, of course, was great apart from every time somebody did a firmware update.

13:42.760 --> 13:47.760
Then PCR 0 and no longer matched because you've just changed your firmware.

13:47.760 --> 13:49.760
And of course, with Alan, it's a desktop OS.

13:49.760 --> 13:54.760
I have FFW update there with no software, and it was running all the time.

13:54.760 --> 13:58.760
You know, reading the system decrypts at a box of a good starting point,

13:58.760 --> 14:02.760
you know, it's recommended there to not measure PCR 0 for this kind of reason.

14:02.760 --> 14:10.760
But I do always ask myself, you know, is this enough?

14:10.760 --> 14:14.760
And of course, I wrote this like three weeks ago, and now there's a nice blog post.

14:14.760 --> 14:17.760
It's telling me definitely probably not enough.

14:17.760 --> 14:21.760
You know, there is cases like potentially.

14:21.760 --> 14:26.760
You know, with a really strong way, basically the kind of potential bypass here is, you know,

14:26.760 --> 14:30.760
some very fancy person could theoretically inject a different operating system,

14:30.760 --> 14:33.760
kind of in front of your one.

14:33.760 --> 14:37.760
So, you know, potentially you could end up booting into a different environment,

14:37.760 --> 14:43.760
and you were expecting, but that environment still gets the keys unlocked, so we can still read your disk.

14:43.760 --> 14:48.760
Where is the recommendation there is to, yeah, measure PCR 15.

14:48.760 --> 14:53.760
System D has all this wonderful tooling for measuring lots of file system information,

14:53.760 --> 14:57.760
and UUIDs and recovery keys and all that stuff.

14:57.760 --> 15:00.760
And none of all that stuff only works if I'm using System D,

15:00.760 --> 15:02.760
to stop with UKIs.

15:02.760 --> 15:06.760
Is that really the one true way for, sort of, am I?

15:06.760 --> 15:09.760
Why not go on with UKIs?

15:09.760 --> 15:11.760
Yes, fine.

15:11.760 --> 15:15.760
Well, the reason we haven't used UKIs up to now with A-on is, of course,

15:15.760 --> 15:21.760
we're using BTRFS with a Susan-style fancy snapshotting and rollback.

15:21.760 --> 15:25.760
So, every update to the system creates a new BTR snapshot,

15:25.760 --> 15:31.760
the main magic with SD boot UTIL is basically creating boot entries in System D boot.

15:31.760 --> 15:35.760
So, when your system no longer boots, you can pick a different boot entry to go to...

15:35.760 --> 15:39.760
The reason to say no to UKIs, that's the reason why you want Type 1 and just, right?

15:39.760 --> 15:45.760
So, my suggestion would always, like, you can use a Type 1 SD boot, like a BLS entry,

15:45.760 --> 15:48.760
point it through UKIs, so you're basically...

15:48.760 --> 15:53.760
Yep, so the answer was, you know, you could just have a Type 1 entry put it through UKIs,

15:53.760 --> 15:58.760
but that UKIs has a command line that, you know, is signed to check, et cetera,

15:58.760 --> 16:03.760
and with our snapshots, you know, we are typically changing the snapshot by changing that command line

16:03.760 --> 16:05.760
to point to the different snapshot number, which...

16:05.760 --> 16:06.760
Yeah.

16:06.760 --> 16:07.760
Yeah.

16:07.760 --> 16:09.760
What about UTIL?

16:09.760 --> 16:10.760
What's it?

16:10.760 --> 16:11.760
I can't do this.

16:11.760 --> 16:12.760
Yeah.

16:12.760 --> 16:13.760
We've got UTKI profiles.

16:13.760 --> 16:14.760
Yeah.

16:14.760 --> 16:15.760
Yeah.

16:15.760 --> 16:19.760
I actually, I actually have an idea that I came up with while writing these slides,

16:19.760 --> 16:23.760
which would basically just be creating a directory structure, you know,

16:23.760 --> 16:28.760
on my system for each UKI version that I know the system's deployed.

16:28.760 --> 16:34.760
So, you know, UKI version 1 points to this snapshot, and that's contained in the OS,

16:34.760 --> 16:39.760
therefore when it boots find, and that means in user space I can do all the fancy rollbacks to whatever

16:39.760 --> 16:45.760
snapshot I want, and a boot time I just have a list of UKIs, which would be simple and easy.

16:45.760 --> 16:51.760
And actually, probably more sensible for our users, because most of the time when you're using that boot menu,

16:51.760 --> 16:54.760
you're only trying to recover because your kernel's messed up.

16:54.760 --> 16:59.760
So, booting to a slightly older snapshot with the same kernel is actually the worst thing.

16:59.760 --> 17:00.760
Yeah.

17:00.760 --> 17:03.760
Are you considering the potential to use snapshots?

17:03.760 --> 17:08.760
Because you can still then, against the rollback of the VM, you're creating them locally.

17:08.760 --> 17:09.760
I had...

17:09.760 --> 17:11.760
I had the connection in the USB, and we used that.

17:11.760 --> 17:13.760
So, I had no idea we could do that.

17:13.760 --> 17:14.760
That's a cool idea.

17:14.760 --> 17:17.760
The suggestion is like, why don't we use good entrances?

17:17.760 --> 17:19.760
Yeah, that's an awesome idea.

17:19.760 --> 17:20.760
I'll look into that.

17:20.760 --> 17:23.760
I mean, this is the use case for...

17:23.760 --> 17:25.760
Is this a credentials, right?

17:25.760 --> 17:28.760
Like that you've put them locked to the TVM next to the kernel.

17:28.760 --> 17:32.760
Now, in your cases, it's not going to work one to one, because you only want one kernel,

17:32.760 --> 17:35.760
but it's 500 boot menu entries after all.

17:35.760 --> 17:36.760
Yep.

17:36.760 --> 17:37.760
So, we would probably...

17:37.760 --> 17:38.760
There's a...

17:38.760 --> 17:42.760
There's an issue open already about something like this, where you...

17:43.760 --> 17:45.760
Well, no, you actually can use the global grid.

17:45.760 --> 17:48.760
And to, like, we have a directory for global credentials, you can just stuff your...

17:48.760 --> 17:50.760
No, you need the other way around.

17:50.760 --> 17:51.760
Anyway, but...

17:51.760 --> 17:52.760
Yeah, especially...

17:52.760 --> 17:53.760
Yes, the discussion is going on.

17:53.760 --> 17:55.760
We're precisely about your problem.

17:55.760 --> 17:56.760
You get a issue.

17:56.760 --> 17:57.760
Yeah.

17:57.760 --> 18:02.760
It's really easy that you take a BLS1 thing that points to the UKI.

18:02.760 --> 18:08.760
But then, also, we teach BLS101 to point to a credentials file, right?

18:08.760 --> 18:11.760
So, and then, for every boot part of our press,

18:11.760 --> 18:14.760
not sure that you have, you generate one credentials.

18:14.760 --> 18:15.760
Yeah.

18:15.760 --> 18:16.760
You have a dedicated information.

18:16.760 --> 18:20.760
And then, one BLS1 entry that goes together with the UKI.

18:20.760 --> 18:21.760
Yep.

18:21.760 --> 18:24.760
That itself is not often indicated, but that's not a problem.

18:24.760 --> 18:25.760
That's... Yeah.

18:25.760 --> 18:27.760
So, yeah, that's what that answer is.

18:27.760 --> 18:28.760
Sort it then.

18:28.760 --> 18:29.760
So...

18:29.760 --> 18:31.760
I'm expecting a patch about the new.

18:31.760 --> 18:33.760
Yes, you can definitely get a patch from it.

18:33.760 --> 18:35.760
I would definitely be looking at that.

18:35.760 --> 18:36.760
That seems to...

18:36.760 --> 18:39.760
So, that's actually my topics.

18:39.760 --> 18:42.760
So, does anybody have any questions or any more feedback?

18:42.760 --> 18:44.760
Although, I think we covered the burning topics already?

18:44.760 --> 18:45.760
Yes.

18:45.760 --> 18:47.760
So, I want to mention...

18:47.760 --> 18:50.760
Go away from always mentioning, like the non-TPM thing,

18:50.760 --> 18:52.760
as the fallback for that, for any sort of desktop systems,

18:52.760 --> 18:53.760
isn't it?

18:53.760 --> 18:56.760
Doesn't it make more sense for the PIDOS stuff to be the primary thing?

18:56.760 --> 18:58.760
And then the TPM be the add-on to that?

18:58.760 --> 19:02.760
Instead of making the focusing so much on the TPM,

19:02.760 --> 19:04.760
then, like, this fall face or whatever, we're in the fall.

19:04.760 --> 19:08.760
So, yeah, the suggestion is, like, why do the TPM is the default,

19:08.760 --> 19:12.760
and not the passphrases is the fallback?

19:12.760 --> 19:16.760
My logic with that is really, in the case of A on,

19:16.760 --> 19:19.760
you know, being partly because I'm lazy.

19:19.760 --> 19:21.760
You know, I just want my laptop to boot.

19:21.760 --> 19:23.760
I don't want to actually enter a passphrase.

19:23.760 --> 19:27.760
You know, ideally, I want to know that I'm booting, you know,

19:27.760 --> 19:29.760
the boot chain that I should be.

19:29.760 --> 19:33.760
So, I'm using FTE with TPM to really give me that verification

19:33.760 --> 19:34.760
that my system is good.

19:34.760 --> 19:39.760
And, yeah, I'm not interested in having a passphrase

19:39.760 --> 19:40.760
to unlock my machine.

19:40.760 --> 19:43.760
So, that's, you know, that's the mindset there.

19:43.760 --> 19:48.760
Yeah, I think it's, of course, a nice thing with all of this,

19:48.760 --> 19:52.760
because of stuff like the command line being validated,

19:52.760 --> 19:54.760
you know, then I know no one's going in and changing,

19:54.760 --> 19:57.760
you know, boot to bin batch and doing silly stuff like that.

19:57.760 --> 20:01.760
So, my Linux actual login passphrase is suddenly

20:01.760 --> 20:03.760
actually meaningful.

20:03.760 --> 20:05.760
So, I rely on that to do my data.

20:05.760 --> 20:08.760
So, that's, that's the mindset switch that I've done with that

20:08.760 --> 20:10.760
and that's kind of way to reflect it with that.

20:10.760 --> 20:12.760
I get, like, through the traditional way,

20:12.760 --> 20:14.760
is thinking the other way around.

20:14.760 --> 20:20.760
But, like, then if I do, then I lose all of my checking of my entire boot chain.

20:20.760 --> 20:23.760
Like, that's, that's what I want this for.

20:23.760 --> 20:25.760
Like, yeah.

20:25.760 --> 20:27.760
Yeah, so, that's the logic.

20:27.760 --> 20:28.760
Yes.

20:29.760 --> 20:34.760
If someone still, if someone stills your laptop,

20:34.760 --> 20:37.760
I'm relying on your, your Linux authentication

20:37.760 --> 20:40.760
to actually work and not be easily bypass,

20:40.760 --> 20:43.760
because you can't easily bypass it, because you can't go

20:43.760 --> 20:45.760
and do stuff like changing the same thing.

20:45.760 --> 20:47.760
So, yeah, totally different approach.

20:47.760 --> 20:48.760
I get it.

20:48.760 --> 20:51.760
I know that freaks some people out, but I'm happy with that.

20:51.760 --> 20:53.760
But with that caveat of, like, you, yeah,

20:53.760 --> 20:56.760
the command line checking really has to be good.

20:57.760 --> 20:59.760
Yeah.

20:59.760 --> 21:02.760
Any more questions?

21:02.760 --> 21:04.760
Five more minutes.

21:04.760 --> 21:07.760
Yeah, go, five more minutes.

21:07.760 --> 21:09.760
Yes.

21:09.760 --> 21:22.760
So, the suggestion is, yeah, using the camera on boot to read the QR code

21:22.760 --> 21:25.760
from your phone for the recovery key.

21:25.760 --> 21:27.760
I mean, I like that idea.

21:27.760 --> 21:30.760
I don't think I'm going to get away with putting a camera driver

21:30.760 --> 21:33.760
into a system D-boot, maybe.

21:33.760 --> 21:36.760
Yeah.

21:36.760 --> 21:38.760
Yeah.

21:38.760 --> 21:39.760
Yeah.

21:39.760 --> 21:43.760
Like, that's, that's a lot of, like, how do you show that to the user

21:43.760 --> 21:45.760
so they know where to hold the phone?

21:45.760 --> 21:47.760
Like, so then you need, then you need to, like,

21:47.760 --> 21:50.760
wail in the system D-boot.

21:50.760 --> 21:52.760
Yeah.

21:52.760 --> 21:53.760
Cool, cool.

21:53.760 --> 21:54.760
Doable.

21:54.760 --> 21:55.760
Yeah.

21:55.760 --> 21:56.760
Yeah.

21:56.760 --> 21:58.760
You think with the duty of the icon packages, then we just have to pull in

21:58.760 --> 22:00.760
the right packages from there, then we'll see.

22:00.760 --> 22:01.760
And then we'll be blown.

22:01.760 --> 22:02.760
Yeah.

22:02.760 --> 22:05.760
Because I like this, like, the other, this, the person who got the cursor in the

22:05.760 --> 22:10.760
there, means, for example, it is taking, having the user to

22:10.760 --> 22:13.760
need, how do you talk to your user, to be, you know,

22:13.760 --> 22:17.760
show me, so we, because I was talking to the,

22:17.760 --> 22:19.760
the idea is that we need to put the rules of something.

22:19.760 --> 22:20.760
That's what we need to be.

22:21.760 --> 22:23.760
And we're heading to camera.

22:23.760 --> 22:24.760
It's just more efficient.

22:24.760 --> 22:26.760
I think it's just coming.

22:35.760 --> 22:36.760
Let's go.

22:36.760 --> 22:37.760
Let's go.

22:37.760 --> 22:38.760
Yeah.

22:38.760 --> 22:40.760
It's a little exalogy.

22:40.760 --> 22:41.760
Have the camera check it your face.

22:41.760 --> 22:42.760
Yeah.

22:42.760 --> 22:43.760
Sure.

22:43.760 --> 22:46.760
One thing I wanted to mention is that, you know,

22:46.760 --> 22:48.760
we do the pilot thing, was it did the end stuff?

22:49.760 --> 22:50.760
Mm-hmm.

22:50.760 --> 22:52.760
Some of the else we can do is, like,

22:52.760 --> 22:54.760
like a TOTC thing against the TBM.

22:54.760 --> 22:55.760
Yeah.

22:55.760 --> 22:58.760
So, and that, that wouldn't make the case so much shorter.

22:58.760 --> 22:59.760
Yeah.

22:59.760 --> 23:00.760
Yeah.

23:00.760 --> 23:03.760
That would be the regular TOTB, that everybody has.

23:03.760 --> 23:04.760
Yeah.

23:04.760 --> 23:06.760
Somewhere on my tip, do was, more so than,

23:06.760 --> 23:09.760
fight over to support with TBM, at the same time,

23:09.760 --> 23:11.760
as to just TOTOTB.

23:11.760 --> 23:13.760
TOTB with that, that would, that would be,

23:13.760 --> 23:14.760
yeah.

23:14.760 --> 23:15.760
I mean, when it comes to multi-factor,

23:15.760 --> 23:17.760
I'd like to have more options than just relying on the TPM.

23:17.760 --> 23:18.760
Yeah.

23:18.760 --> 23:21.760
This isn't using as a really nice thing that our,

23:21.760 --> 23:23.760
uh, cell phone operating system should be nice, absolutely.

23:23.760 --> 23:24.760
Yeah.

23:24.760 --> 23:25.760
Yeah.

23:25.760 --> 23:27.760
By a while, the PCR, we're, like, the QR code that we currently show,

23:27.760 --> 23:29.760
um, that's a fucking text rate.

23:29.760 --> 23:30.760
Yeah.

23:30.760 --> 23:31.760
Yeah.

23:31.760 --> 23:32.760
Yeah.

23:32.760 --> 23:33.760
Yeah.

23:33.760 --> 23:34.760
It's not great.

23:34.760 --> 23:36.760
But it's better than nothing, and, yeah.

23:36.760 --> 23:37.760
Yeah.

23:37.760 --> 23:44.760
Do you support having multiple factors in combination?

23:44.760 --> 23:47.760
So five of key plus something?

23:47.760 --> 23:48.760
No.

23:48.760 --> 23:49.760
That's kind of what we're talking about.

23:49.760 --> 23:50.760
Like, I'd, yeah.

23:50.760 --> 23:51.760
Yeah.

23:51.760 --> 23:52.760
Yeah.

23:52.760 --> 23:53.760
That's interesting.

23:53.760 --> 23:54.760
There always comes up.

23:54.760 --> 23:56.760
People want to use, like, as I said, that's, like, the,

23:56.760 --> 23:57.760
some is secret charity.

23:57.760 --> 24:00.760
Um, I think that's, uh, there would be income model, actually,

24:00.760 --> 24:02.760
then, not the five or two plus TBMs,

24:02.760 --> 24:04.760
that, that's the, that's the problem that people are doing,

24:04.760 --> 24:07.760
because, um, you know, if you use as a success,

24:07.760 --> 24:09.760
and then combine this in user space, the keys,

24:09.760 --> 24:12.760
um, that you get from the digital hardware,

24:12.760 --> 24:15.760
and, um, there's all done in, in user space, right, like, um,

24:15.760 --> 24:17.760
and there's no way to do it in the TBM.

24:17.760 --> 24:18.760
Because I don't think you can.

24:18.760 --> 24:19.760
No.

24:19.760 --> 24:20.760
And that's, yeah.

24:20.760 --> 24:21.760
Yeah.

24:21.760 --> 24:25.760
So, uh, yeah, the next best thing is, like,

24:25.760 --> 24:28.760
it's much nicer than if, for the five or two TBMs stuff,

24:28.760 --> 24:30.760
um, there is no, yeah, all the calculations that

24:30.760 --> 24:32.760
are done in the TBMs, and it's by the device,

24:32.760 --> 24:34.760
and, uh, only when everything checks out,

24:34.760 --> 24:37.760
the TBM will release the volume P to you,

24:37.760 --> 24:39.760
and that's a really nice model.

24:40.760 --> 24:44.760
One last comment, as soon as the timer just might be ours,

24:44.760 --> 24:47.760
you can clear the eyes and the campaign.

24:47.760 --> 24:48.760
Yeah.

24:48.760 --> 24:51.760
You'll be able to do the device for the end and sign them,

24:51.760 --> 24:55.760
only the way the PCR stuff sign as well.

24:55.760 --> 24:58.760
That's, that's, that's, as of last week.

24:58.760 --> 25:00.760
So that's, that's, that's, that's my, yeah,

25:00.760 --> 25:01.760
on the back there for the network.

25:01.760 --> 25:02.760
Perfect.

25:02.760 --> 25:03.760
So, yeah.

25:03.760 --> 25:06.760
So, so, even if I do, I'm stuck on doing you guys,

25:06.760 --> 25:07.760
it's going to be easy.

25:07.760 --> 25:10.760
That's, awesome news. Thank you.

25:10.760 --> 25:11.760
Great.

25:11.760 --> 25:12.760
Okay.

25:12.760 --> 25:14.760
Uh, yes, I think we're all the time, so let's thank Richard.

