WEBVTT

00:00.000 --> 00:10.000
Before we start, I have two announcements.

00:10.000 --> 00:16.000
First is for the people who are going to the open-ended workshop tomorrow.

00:16.000 --> 00:22.000
Check your email for the new location. It has changed.

00:22.000 --> 00:27.000
The second announcement is related to this talk.

00:27.000 --> 00:33.000
Those are the chocolates. Who has been to two of my previous talks at first

00:33.000 --> 00:39.000
them? What was yesterday? In the other words, this morning.

00:39.000 --> 00:45.000
Not too many. It didn't work. But there are still people.

00:45.000 --> 00:49.000
So I'm going to give those chocolates.

00:49.000 --> 00:53.000
The first person I've seen who attended three talks,

00:53.000 --> 00:59.000
then you organize yourself to the show chocolates.

00:59.000 --> 01:03.000
People who attended two talks qualify also. Who was that?

01:03.000 --> 01:05.000
It was you.

01:05.000 --> 01:09.000
And then you start passing the chocolates to other people.

01:09.000 --> 01:13.000
Thank you.

01:13.000 --> 01:17.000
So, formally you haven't attended the first talk.

01:17.000 --> 01:19.000
But we are going to start, right?

01:19.000 --> 01:21.000
And there are people coming in.

01:21.000 --> 01:31.000
So, this is going to be a personal selection of events that happened during the last year

01:31.000 --> 01:33.000
in the embedded space.

01:33.000 --> 01:37.000
Even with a few categories like regulation, cryptography, and so on.

01:37.000 --> 01:43.000
Maybe I'm not giving exact dates because that's not that important.

01:43.000 --> 01:47.000
But you will have in the slides later on

01:47.000 --> 01:55.000
that you will be able to download links to various talks, events, and the like.

01:55.000 --> 02:01.000
So, I was thinking a long time what I'm going to start with.

02:01.000 --> 02:05.000
If I'm going to start with regulations or I'm going to finish with regulations,

02:05.000 --> 02:15.000
in during my talks, finishing with regulations usually means that then I have an hour talking about regulations in the question and answer section.

02:15.000 --> 02:21.000
So, I'm going to start with regulations that hopefully are going to forget all your questions.

02:21.000 --> 02:23.000
So, but who am I?

02:23.000 --> 02:30.000
I'm Marka, I do embedded security and I don't have a lot during all of those conferences.

02:30.000 --> 02:32.000
So, where do we start?

02:32.000 --> 02:35.000
We are going to start with regulations.

02:35.000 --> 02:37.000
Yeah, regulations.

02:37.000 --> 02:48.000
So, the regulations, the big regulations that we have going on for the embedded space in this is the cyber resources act.

02:48.000 --> 02:51.000
Most of people here are in Europe.

02:51.000 --> 02:55.000
So, I'm going to test it again.

02:55.000 --> 03:05.000
So, who knows the cyber resources act well right your hand?

03:05.000 --> 03:08.000
Okay?

03:08.000 --> 03:12.000
Who has heard about it?

03:12.000 --> 03:14.000
Way better.

03:14.000 --> 03:17.000
Okay.

03:17.000 --> 03:24.000
So, cyber resources act mandatory cyber security requirements for all of them.

03:24.000 --> 03:35.000
But the product, it's already in place, but all of the conditions apply in December 27.

03:35.000 --> 03:40.000
And what happened in 25, about all of that?

03:40.000 --> 03:53.000
There was a work on standards and first set of standards have seen, have been seen by people other than people working.

03:53.000 --> 04:11.000
So, the big reminder is that standards are not mandatory, but in practice, especially in some categories, you will actually need them.

04:11.000 --> 04:22.000
So, there was the first review round for quite many of them and still continuing for the vulnerability management horizontal.

04:22.000 --> 04:42.000
And there is a common problem going on with the serious standards affecting especially embedded in small companies that it is still not clear if they are going to be available for a fee or not.

04:42.000 --> 04:48.000
Especially taking account that you have three horizontals that apply to everything.

04:48.000 --> 04:54.000
And then you have 40 something verticals.

04:54.000 --> 05:03.000
If you want to buy them it's 150 Swiss francs.

05:03.000 --> 05:12.000
So, this budget, so the discussion is going on and some people do not want open source people to talk about this anymore.

05:12.000 --> 05:14.000
But we do.

05:14.000 --> 05:19.000
Then the second subject is the European vulnerability database.

05:19.000 --> 05:38.000
So, this is even more funny because there is an obligation to report exploited vulnerabilities and security issues in all products starting September 26.

05:38.000 --> 05:42.000
That's this year.

05:42.000 --> 05:58.000
And the really funny thing is that the site where you will be doing that is not that part is not even embedded.

05:58.000 --> 06:00.000
So, that's interesting.

06:00.000 --> 06:03.000
The database itself.

06:03.000 --> 06:09.000
That's basically assumes you are going to use to check your products for vulnerabilities.

06:09.000 --> 06:15.000
It is in preview.

06:15.000 --> 06:19.000
Yeah, but that's still the work on it.

06:19.000 --> 06:36.000
And then in December last year, the commission has released a fact answering some of the questions that the community was asking them to clarify for a very, very long time.

06:36.000 --> 06:40.000
And it's long.

06:40.000 --> 06:45.000
Important clarification for embedded people.

06:45.000 --> 06:49.000
It is on the support period.

06:49.000 --> 06:59.000
The support period is the number of years during which you need to release security updates for free for all of your products.

06:59.000 --> 07:10.000
And the clarification is that the start of that support period comes from the moment the manufacturer sells the product.

07:10.000 --> 07:12.000
Not from the end of design.

07:12.000 --> 07:18.000
Not from the moment you fork the object or build root to support that vote.

07:18.000 --> 07:26.000
But it starts from the invoice date of the manufacturer.

07:26.000 --> 07:33.000
So if you have one batch, there's a part of the batch on the shelf.

07:33.000 --> 07:49.000
So later, the support period for that part of the batch runs from the moment they are actually sold and not put in the warehouse.

07:49.000 --> 07:58.000
So that's a pretty important and they confirm that the risk assessment is the only line of defense that you have.

07:58.000 --> 08:05.000
The specificity of CRA is that it's a risk based approach.

08:05.000 --> 08:10.000
So there's no checklist of the features you expected to have.

08:10.000 --> 08:18.000
We expected to have a risk assessment and it's the risk assessment that will be giving you the checklist.

08:19.000 --> 08:22.000
And a lot of responsibility that comes from.

08:22.000 --> 08:27.000
So I did the presentation about risk assessments.

08:27.000 --> 08:30.000
But I do not think it, okay.

08:30.000 --> 08:43.000
But I think I think to the fact so that you can have a look, there will be at least one more on open source, but I don't know why.

08:43.000 --> 08:49.000
So what also happened in 2025 on the embedded markets now.

08:49.000 --> 08:55.000
There's more alone awareness about the series that's really really clear.

08:55.000 --> 09:01.000
And if you work on embedded products.

09:01.000 --> 09:12.000
As I recommend you to do a try and ask your vendors to provide their CRA plans.

09:12.000 --> 09:21.000
Because if in 2027 they are unable to fix the C mark on the product.

09:21.000 --> 09:24.000
The new C mark with the cybersecurity requirements.

09:24.000 --> 09:33.000
You have a pretty supply chain problem because you cannot integrate that product that component anymore into your product.

09:33.000 --> 09:36.000
Better to ask them before if they are preparing.

09:37.000 --> 09:42.000
And so vendors are even starting to provide S-bombs on their websites.

09:42.000 --> 09:45.000
That's the real revolution.

09:45.000 --> 09:57.000
This advantage of the fact that companies are starting to be aware about the CRA is them misinformation.

09:57.000 --> 10:03.000
So I've seen basically all false informations about the CRA online.

10:04.000 --> 10:14.000
Including presentations of certain features as mandatory under the CRA where they are not.

10:14.000 --> 10:19.000
So be really really careful about what your vendors tell you.

10:19.000 --> 10:27.000
And there are some vendors that are claiming CRA compliance already.

10:27.000 --> 10:39.000
I do not claim series compliance for my product yet because at least I do not know how to report those them exploited when I beat this to any site.

10:39.000 --> 10:42.000
I'm missing at least that.

10:42.000 --> 10:51.000
So if someone is claiming that that's orange orange getting into red sign.

10:51.000 --> 10:55.000
I check it for you to check if you are ready.

10:55.000 --> 10:59.000
We are not going to play with hands.

10:59.000 --> 11:02.000
At this stage next year.

11:02.000 --> 11:12.000
The first important thing is if you have an update system for your products because that's one of the basis of the components.

11:12.000 --> 11:23.000
And then if you can receive vulnerabilities of your products easily by users and then you know how to handle them.

11:23.000 --> 11:27.000
That includes updating your products.

11:27.000 --> 11:38.000
And if you have started thinking about risk assessment, I guess that people will start writing those things this year.

11:38.000 --> 11:48.000
I, I, I, I, he is the link to my to my talk last year about the basic of of risk assessment.

11:48.000 --> 11:50.000
Find that.

11:50.000 --> 11:55.000
Okay, now a little bit lighter.

11:55.000 --> 11:58.000
Subject cryptography.

11:58.000 --> 12:06.000
That apparently is a really controversial, even even more controversial than CRA.

12:06.000 --> 12:12.000
It seems when I posted on social media. So the quantum computing.

12:12.000 --> 12:18.000
As a background.

12:18.000 --> 12:22.000
Algoriths for post quantum cryptography has been standard.

12:22.000 --> 12:27.000
What it means that they are normalized.

12:27.000 --> 12:31.000
They could be used and what is what is post quantum.

12:31.000 --> 12:35.000
You've probably heard about quantum cryptography.

12:35.000 --> 12:50.000
If and when cryptography shows up, it will be all the public key algorithms that you have including your SSH keys.

12:50.000 --> 12:53.000
They will be really easy to break.

12:53.000 --> 13:00.000
But with the current knowledge, it doesn't break symmetric algorithms like I, yes.

13:00.000 --> 13:08.000
This is that subtlety that it does break certain things, but not everything that we have.

13:08.000 --> 13:16.000
The current estimations about when it can happen, they vary.

13:16.000 --> 13:20.000
The optimistic people are saying in a few years.

13:20.000 --> 13:26.000
So less optimistic people say in a 10 to 15 years.

13:26.000 --> 13:31.000
And so people are saying that it's never going to happen.

13:31.000 --> 13:46.000
That is a nice risk assessment scenario because when it happens, you are going to have devices in the field.

13:46.000 --> 13:57.000
And many of you are already working on devices that will be in the field in 15 or 20 years.

13:57.000 --> 14:01.000
The question is what do you do?

14:01.000 --> 14:13.000
You may already reserve RAM and flash space for those extra libraries that will have those algorithms to eventually update.

14:13.000 --> 14:23.000
They should further or you already link them cryptographic libraries in assuming that for product that you ship today,

14:23.000 --> 14:27.000
there's no hardware acceleration for those things.

14:27.000 --> 14:31.000
It's going to show up in a few years problem.

14:31.000 --> 14:42.000
From the good site also on the post quantum, open SSH support from the current LTS of basically everything.

14:42.000 --> 14:50.000
And they missed the window for one of the parts for the signature validation that is available only in 36.

14:50.000 --> 14:56.000
So if you want to add post quantum to your products, you can.

14:56.000 --> 15:07.000
They are all the cryptolibrary working on the support of post quantum at different stages.

15:07.000 --> 15:12.000
So post quantum is a risk assessment exercise.

15:12.000 --> 15:20.000
What do you do either you do nothing and you may wake up with devices in the field that have completely broken.

15:20.000 --> 15:30.000
Outerizations for example, or you start adding those libraries in your products.

15:30.000 --> 15:38.000
And here I will play with with hands who has already looked in the post quantum.

15:38.000 --> 15:46.000
Oh really nice nice nice nice.

15:46.000 --> 15:53.000
So on the two side.

15:53.000 --> 15:58.000
First on the typical tools on the compiler side and my security side.

15:58.000 --> 16:03.000
There is now F-houdent option in GCC.

16:03.000 --> 16:08.000
That is combining some of the nice hardening options that you have.

16:08.000 --> 16:13.000
So why not to use it?

16:13.000 --> 16:18.000
There are also hardening improvements on different platforms in LLVM.

16:18.000 --> 16:23.000
LLVM doesn't have F-houdent yet, but they are discussing.

16:23.000 --> 16:31.000
The second big subject of course in the field is the move to memory safe languages.

16:31.000 --> 16:34.000
Also coming to embedded.

16:34.000 --> 16:41.000
So with the notable even being the end of the Linux rest experiment.

16:42.000 --> 16:50.000
And I know some vendors of embedded products that they are rewriting their stacks.

16:50.000 --> 16:54.000
It memory safe languages.

16:54.000 --> 17:01.000
And then we also have AI.

17:01.000 --> 17:04.000
That can be used to accelerate development.

17:04.000 --> 17:10.000
It can be used to find vulnerabilities in the products.

17:10.000 --> 17:17.000
Or submit completely bugger reports.

17:17.000 --> 17:19.000
Depends.

17:19.000 --> 17:24.000
And there are talks about the subject all around the first them.

17:24.000 --> 17:32.000
So I'm not going to get into the details more on that.

17:32.000 --> 17:35.000
All the dependencies and as well.

17:35.000 --> 17:38.000
What do we have here?

17:38.000 --> 17:41.000
Of course, I'm going to ask a question.

17:41.000 --> 17:44.000
Who is already generating an as well?

17:44.000 --> 17:46.000
Okay.

17:46.000 --> 17:48.000
Less than a half.

17:48.000 --> 17:49.000
Okay.

17:49.000 --> 17:52.000
So it is currently working in the Octoproject.

17:52.000 --> 17:53.000
It's working in Zephyr.

17:53.000 --> 17:56.000
It's working for some air tosses.

17:56.000 --> 17:59.000
Even some SDK vendors do provide.

17:59.000 --> 18:02.000
As well, I haven't looked yet.

18:02.000 --> 18:06.000
How accurate they look.

18:06.000 --> 18:09.000
But the end result problem in embedded.

18:09.000 --> 18:12.000
I did a whole talk about the subject this morning.

18:12.000 --> 18:16.000
So if you try to merge as bombs.

18:16.000 --> 18:19.000
For example, for multiple processors.

18:19.000 --> 18:21.000
Into one.

18:21.000 --> 18:23.000
To analyze this on that.

18:23.000 --> 18:25.000
That's becoming those.

18:25.000 --> 18:29.000
They're really complicated still.

18:29.000 --> 18:33.000
And we have a result of problems on in the S bomb land.

18:33.000 --> 18:37.000
One of them is how do we handle patched software.

18:37.000 --> 18:40.000
It's pretty common in a bit.

18:40.000 --> 18:42.000
And how we handle naming or false.

18:42.000 --> 18:45.000
There's no good solution.

18:45.000 --> 18:47.000
In there.

18:47.000 --> 18:50.000
Another tendency that we have a lot in embedded is that.

18:50.000 --> 18:54.000
People copy code like libraries.

18:54.000 --> 18:57.000
Or just files somewhere.

18:57.000 --> 19:02.000
And then other developer needs to deal with that ten years later.

19:02.000 --> 19:07.000
Because of course those files haven't been updated for the last ten years.

19:07.000 --> 19:11.000
And nobody was aware that it was there.

19:11.000 --> 19:14.000
And then.

19:14.000 --> 19:19.000
We have our favorite Wi-Fi bluetooth and all the stacks that are running.

19:19.000 --> 19:24.000
Processos internally with big software stacks on them.

19:24.000 --> 19:29.000
That have a lot of other software libraries on them.

19:29.000 --> 19:33.000
And we would like sometimes to.

19:33.000 --> 19:37.000
Have a nest bomb to have an idea what.

19:37.000 --> 19:39.000
How much of a number they are.

19:39.000 --> 19:42.000
Especially if they are providing connectivity.

19:42.000 --> 19:48.000
And that's that's also a very interesting and unsolved problem.

19:49.000 --> 19:54.000
Um, this year I have also seen a lot of.

19:54.000 --> 19:58.000
Attention on bootloaders.

19:58.000 --> 20:02.000
A bootloader is something that is very close to your hardware.

20:02.000 --> 20:06.000
And it can write to the memory.

20:06.000 --> 20:09.000
Wherever it was right.

20:09.000 --> 20:12.000
And they can.

20:12.000 --> 20:16.000
If you can confirm if an attacker can compromise a bootloader.

20:16.000 --> 20:21.000
Then they can compromise the whole of your embedded system.

20:21.000 --> 20:24.000
So there are two interesting talks.

20:24.000 --> 20:28.000
I recommend you to have a look at this at the slide deck.

20:28.000 --> 20:34.000
One above hardening variables and other about finding vulnerabilities in your boot.

20:34.000 --> 20:42.000
And I think that this is going to count as a tendency over the next year too.

20:42.000 --> 20:45.000
Because there's a lot of work to do.

20:45.000 --> 20:50.000
And also on those lower level.

20:50.000 --> 20:54.000
In parts of the stack.

20:54.000 --> 20:57.000
And then on the vulnerabilities side.

20:57.000 --> 21:02.000
I wanted to show you a tendency.

21:02.000 --> 21:06.000
A choice of three vulnerabilities that affect.

21:06.000 --> 21:10.000
That was discovered last year and effect embedded.

21:10.000 --> 21:15.000
If you have a look at the short descriptions.

21:15.000 --> 21:20.000
All of them are related to the code.

21:20.000 --> 21:28.000
Parts in data that comes either from remote service.

21:28.000 --> 21:31.000
Or is given by the user.

21:31.000 --> 21:35.000
So those are not memory safety issues.

21:35.000 --> 21:43.000
Those are issues of parsing actual messages.

21:43.000 --> 21:46.000
That could be controlled by attacker.

21:46.000 --> 21:54.000
We've known many embedded devices that we're trusting everything that is being sent to it.

21:54.000 --> 21:58.000
Not, not, not a good choice.

21:58.000 --> 22:01.000
So to wrap up.

22:01.000 --> 22:03.000
There's a lot of going on.

22:03.000 --> 22:04.000
We have regulations.

22:04.000 --> 22:06.000
We have changes in cryptography.

22:06.000 --> 22:10.000
And I see more and more interest in security subjects.

22:10.000 --> 22:16.000
For example, at the ERT version, we have a separate track on security.

22:16.000 --> 22:19.000
Something I've never seen before.

22:19.000 --> 22:21.000
So that's pretty interesting.

22:21.000 --> 22:26.000
And a lot of security docs are different embedded conferences.

22:27.000 --> 22:31.000
That's, I find that's good.

22:31.000 --> 22:37.000
I will be doing series of setups to resemble information from different subjects.

22:37.000 --> 22:44.000
If you have something that should show up in a future edition of such a talk,

22:44.000 --> 22:46.000
I'm easy to find online.

22:46.000 --> 22:48.000
Let me know.

22:48.000 --> 22:51.000
So do we have time for questions?

22:51.000 --> 23:01.000
Maybe.

23:01.000 --> 23:03.000
Thanks as ever, Marta.

23:03.000 --> 23:08.000
Just going back to manufacturing and risks and obligations.

23:08.000 --> 23:12.000
If we produce OEM hardware with the software stack,

23:12.000 --> 23:14.000
it might be October's effort on top of that.

23:14.000 --> 23:19.000
We sell that into vertical markets as our product to them.

23:19.000 --> 23:23.000
And then they sell on box it up and sell it as part of their service.

23:23.000 --> 23:25.000
Does that mean we're both manufacturers.

23:25.000 --> 23:28.000
And we both have manufacturing obligations.

23:28.000 --> 23:30.000
Or how does that split out?

23:30.000 --> 23:35.000
And what do we need to be doing as we sell on before something actually hits,

23:35.000 --> 23:39.000
as an end product into the market, if that makes sense?

23:39.000 --> 23:41.000
So first, I'm not a lawyer.

23:41.000 --> 23:46.000
A lawyer is somewhere else in this room.

23:46.000 --> 23:53.000
So the situation correctly, there is new manufacturing devices,

23:53.000 --> 23:58.000
and selling them to your customers,

23:58.000 --> 24:02.000
then they are repackaging, adding stuff and selling that again.

24:02.000 --> 24:04.000
In my opinion, they are both manufacturers,

24:04.000 --> 24:07.000
because you are putting devices on the market,

24:07.000 --> 24:10.000
and they are also putting devices on the market.

24:10.000 --> 24:14.000
So in my opinion, they are both manufacturers.

24:15.000 --> 24:20.000
I didn't manage to stop thinking about Syria.

24:20.000 --> 24:23.000
Maybe another question.

24:26.000 --> 24:28.000
We are out of time.

24:28.000 --> 24:29.000
We are on time.

24:29.000 --> 24:30.000
Thank you.

24:30.000 --> 24:32.000
Thank you all for coming.

24:44.000 --> 24:45.000
Thank you.

