[jacquesm]

Oh dear, I wonder how I'll ever be able to use the code I wrote over the years that controls uncounted lathes, mills, plasmacutters, lasers and a whole raft of other industrial tools.

Obviously the only people that can be trusted with our safety are the manufacturers, because the people whose lives are on the line are irresponsible madmen.

> Designing a vehicle to be hackable will very likely lead to an unsafe vehicle.

Vehicles are hackable, but they're not documented which makes them more dangerous, not less dangerous. Witness comma.ai and others.

[ncts]

> I wonder how I'll ever be able to use the code I wrote over the years that controls uncounted lathes, mills, plasmacutters, lasers and a whole raft of other industrial tools.

You are knowledgeable enough to make them work. Many aren't. Some can't be. Hacking requires knowledge and skill, and most importantly, being contained. Cutting yourself with your self-programmed hackable laser in your garage is unfortunate, but cutting other people is a disaster you can't afford.

> Vehicles are hackable, but they're not documented which makes them more dangerous, not less dangerous. Witness comma.ai and others.

I see two points here.

1. Security through obscurity is bad. That's true, but we have "business" in the play, so that's how it goes. Maybe push for better regulation.

2. comma.ai, an "autopilot", based on reverse engineering, or as you put it, the base product "not documented", thus makes it "more dangerous". No, it's dangerous not because the base product is not documented, but because there's no real autopilot at the moment, and comma.ai is irresponsibly advertising as being able to "drive for hours without driver action". There are many "black box" products with a ToS that forbid reverse engineering. Does that make the product inherently more dangerous too?

Besides, you seem to suggest that, with open products, people can not make things unsafer. That's not true. Some don't know what they are doing when they "hack" things.

[adhesive_wombat]

I'm all for open things, but that's a false equivalence. You don't use those tools on a public road around unsuspecting others.

In the same way you can't just merrily hack about with a plane. The FAA don't really care that much if you die in your experiment. They do care if the burning wreckage falls on someone minding their own business.

[LeifCarrotson]

No, Jacques and I hand those things off to random unskilled laborers who come work at a robot cell or CNC for $15/hr without any real knowledge of the code and safety standards. "Push the green button, trust us, it's safe."

FAA regulations for experimental or kit aircraft are wide open. As are automotive standards, having just helped my buddy get his sand rail DOT certified for use on the road. A tube chassis, '65 Volkswagen rear end and VIN, safety glass and manually-squeezed washer bottle and manually-swung windshield wipers (for the DOT guy, then never used again), un-boosted non-ABS brakes with a cutting brake in the loop, and it's cleared for use on the Interstate if we were crazy enough to do that. It's only the manufacturers who claim they're protecting the public by locking down their designs.

[svorakang]

Road approval looks very differently for mass-produced vehicles. It is not an easy thing to pass.

[jacquesm]

And what makes you think that the current crop of automotive software written in either asm or unsafe C is going to be any better than what you or I would produce? I've had a very recent model Mercedes C-class nearly kill me twice on account of buggy software. So much for that 'stellar' (pun intended) reputation. My current car is as dumb as it possibly could be.

I'd expect that if any ECU software was to be released that we'd finally realize how bad things really are and that there would be a massive amount of work done on making sure these pieces of critical software would be as safe as they could possibly be.

Note that the norm is 'a subset of C deemed to be safe' but that what I've seen of such development would not pass my personal threshold for quality work. In fact, rather the opposite. On the plus side, the hardware people usually know their stuff and realize what is dangerous to pass to the software people so with some luck your vehicle will use an FPGA for any kind of really safety critical stuff (or processors embedded with the relevant hardware, such as ABS and so on).

[formerly_proven]

> I'd expect that if any ECU software was to be released that we'd finally realize how bad things really are and that there would be a massive amount of work done on making sure these pieces of critical software would be as safe as they could possibly be.

toyota/denso michael barr testimony cough

edit: oh, there's even slides now, you don't even have to read the court transcript https://www.safetyresearch.net/Library/BarrSlides_FINAL_SCRU...

[jacquesm]

Yes, that's one of the better known cases. But I've looked at similar stuff under NDA and I'm more than a little bit uncomfortable with some of the stuff that I've seen.

The scariest kind of developer to me are the ones that believe that they and only they are able to create safe software without outside scrutiny of what they produce.

Of course that scrutiny would have exposed emissions cheating software right on day #1.

[intern4tional]

This comment chain appears to have a fundamental misconception of what constitutes safe and what does not.

Automotive standards and automotive coding standards approach safety in a different way than most people think (and given your comments I would say this includes you). If you're curious, you can have a look at some rules to evaluate automotive code that are published here: https://github.com/github/codeql-coding-standards

In short, the rules do not aim to eliminate failure or crashes, but rather make the crash predictable and uniform when a crash occurs so that it can be dealt with. This is further complicated by where and how the automotive manufacture chooses to implement safety controls. It is entirely possible to have a bunch of unsafe code running somewhere on a car, and simply have a small safety shim around said code that prevents the unsafe code from impacting the safe operation of the vehicle.

With that in mind, let's take the example that you use here of emissions cheating software. Emissions is likely not considered safety relevant (it might not even be QM, it just might be some code) and so no safety requirement applies to it. So, no real scrutiny would happen there regardless, at least from a safety perspective. See, validating that software passes a particular safety certification is time and money intensive and manufacturers therefore keep the amount of code that they qualify as safe to a minimum. This means as an example that the infotainment systems of many manufacturers are not safety relevant and no safety function should exist on or interact with them.

A few other things to consider from other threads:

- Telsa doesn't necessarily follow or adhere to safety standards. They (Telsa) are explicitly non-compliant in some cases, and this is partially why there are investigations into their practices.

- Industrial robotics code is just as bad if not worse than most automotive software from what I've seen. As you note, its that these robots are not under manual control

- None of this prevents the software from being open source. There are plenty of safety qualified open source projects. This simply limits who can contribute and how contributions are managed. The main reason why many things in automotive are not open source is that the ECU manufacturer isn't interested in doing so, and the Tier 1/2/3 that does the implementation is even less so.

[jacquesm]

There is a difference between what I with my layman-user cap on would consider unsafe (say, releasing purposefully harmful software into the environment or software with bugs like phantom braking) and what I with my embedded software programmer cap on (machine control, aerospace related) would consider to be unsafe. I spend a lot of my time reading root cause analysis on software failures and my only conclusion is that this is a very immature industry with a lot of players that should probably not be trusted as much as we do.

As for safety shims: for instance, watchdog timers are often used for this purpose, to bring a system that is exhibiting buggy behavior to behave more predictable. Personally I would consider any watchdog error that was not directly related to a hardware fault (say a bitflip) as a failure and I'd like to have my car report such failures.

Tesla being non-compliant is precisely my point: they get away with that because they don't have to open up their code to scrutiny. But I'll bet that if they would that their marketing department would not be able to spin stuff the way they do at present.

All of those would take into account the relevant context and I think that's where things go off the rails here: embedded software developers do not get to claim the high ground around what they consider safe or unsafe given the code from that world that I've had my eyes on. If anything it is amazing that it works as well as it does, the only two industries that get a pass based on my experience so far is medical embedded and avionics. Everybody else has the exact same problems as any other software project and would benefit from opening themselves up to scrutiny.

[mikrotikker]

I got a friend who was almost killed by traction control. Pulling out of his driveway onto a 3 lane road, and having 2 cars 1 overtaking come around the corner very fast. He tried to punch it to get out of the way but: ”computer says no”. Longest 2 seconds of his life he reckons, waiting for the gas pedal to work again. Both cars had to brake very heavily.

"Safety"

[jacquesm]

Yes. That sort of thing. I had a 2015 model C-class Mercedes make two fairly serious attempts at killing me before I got rid of it. The first time it could have been a glitch, the second was so obviously borked that I still wonder why they released that stuff on the road. And no way to disable that 'feature'. I'd have loved to analyze what exactly went wrong there and why a bridge and an advertisement by the side of the road were classified as imminent frontal collision.

My present car just does what I tell it to, no gizmos. It may be statistically less safe (I'm willing to believe that) but at least it doesn't actively try to kill me in the name of keeping me safe.

[sircastor]

You don’t drive your lathe down the freeway at 80mph. And your lathe probably isn’t surrounded by hundreds of other lathes also going 80mph. And probably not being run by people only half paying attention.

Tell me about the liability laws in place related to you operating your lathe, or the state-required licensing and insurance that each lathe operator holds.

A machining tool is worlds away from a motor vehicle.

[jacquesm]

Control software for a lathe (or, in fact any industrial robot) is far more dangerous than a car. Go visit any metal workshop and take an inventory of the number of people and then count the number of digits. The fact that they don't cause nearly as many accidents is because they are not normally under manual control. But when they are better find some (preferably solid) cover.

And meanwhile, Tesla gets to beta-test their self driving stuff in public and I can't look at the source code to my ECU to figure out whether or not a certain behavior is by design or an indication that something is broken.

[adhesive_wombat]

Cars will do a lot worse than reduce your finger count. In the UK, 68 people died in work-related accidents. Nearly 1700 died on the roads. (The same figures in the US are 5k/40k, go figure).

And I agree that Tesla shouldn't be beta testing on the roads. But neither should anyone else.

[jacquesm]

The main reason for that is that there are a lot more drivers than there are lathe operators, and that lathe operators usually injure themselves.

There is a relatively low chance that you'll find a lathe on a busy intersection or on a freeway during rush hour. I thought that was obvious.

[ahmedfromtunis]

Not necessarily irresponsible madmen; just curious.

Because I bet you if I buy a new car and discover that I can access its internal components via an API, I will be toying with it.

On any other platform that would never be a problem: found a bug? Just restart the container!

But with a car, this might mean a bug in my code manifesting itself while I'm driving 120 kph. And maybe there's a pedestrian crossing the road and I can't stop in time because the bug makes the brake 60% weaker.

This time however, there's not a restart docker button.

I'm sure if this happens people would be attacking Ferreri viciously the way they pile up on Tesla whenever a douche sleeps at the wheel going 100 kph, even though the company said before that that's not safe.

[jacquesm]

Then we should all go back to security-by-obscurity and trust in the man behind the curtain for computer security as well. And we all know that that doesn't work, so why is there this conviction that the embedded programmers at car companies are made from magic?

It's precisely because cars are so dangerous that the code should be open to scrutiny. And of course - at least in the past - the argument has been made that more eyes do not make the bugs more shallow, but in practice if there is an incentive (such as personal safety) people will expend a lot of effort to figure out why stuff goes wrong.

What it would do is to take away any kind of excuse that manufacturers have in those cases where their gear is suspect to claim that their wares are perfect and that it must have been user error. Because I can pretty much guarantee you that if you were to inspect your average automotive code-base that you'd find errors, and not just minor ones. From accidental erroneous emergency braking, untended acceleration to outright malicious ones such as planned obsolescence drivers, emission controls defeat code and so on.

[adhesive_wombat]

Open to scrutiny, absolutely. Anything safety-critical should be freely available to those it can harm. Cars, trains, planes, nuclear reactors, lathes, the lot. I hope your code and schematics is fully provided to worker relying on it being correct. I indeed don't have faith in regulators auditing it properly.

That said I still don't want someone to plonk some GitHub code into the brake controllers, take it for a spin and turn me and mine into meat salsa.

On private land, surrounded by informed and consenting people, sure, go nuts.

[jacquesm]

> That said I still don't want someone to plonk some GitHub code into the brake controllers, take it for a spin and turn me and mine into meat salsa.

The chances of that happening, versus brake fluid contamination, bad lines, seized rotors, rusted rotors, rotors and pads with grease on them and a thousand other mechanical failures are nil. Because brake controllers are always backed up by a mechanical system and the worst thing about a brake is that it could fail.

The bigger problem is that manufacturers that could barely create functional entertainment systems are now actually creating software and hardware combos that can override driver input to the steering wheel and the brakes and in my own experience they are absolutely not qualified to do this. Car software is crap, you can take my word on that. Very, very few manufacturers have software as a core competency.

[elzbardico]

Please define car software.

You have user facing functions, and you have engine control functions, ABS, transmission, and so on.

The first one, I agree, is generally crap.

For the second one, in a lot of models, your manufacturer haven't even written the code, because they buy it from some OEM manufacturer like Bosch.

And I am pretty sure that Bosch is pretty good at writing this kind of software.

[jacquesm]

> Please define car software.

The totality of all code running on a particular vehicle that was part of that vehicle when it was sold to the end user.

> You have user facing functions, and you have engine control functions, ABS, transmission, and so on.

Yes.

> The first one, I agree, is generally crap.

Ok.

> For the second one, in a lot of models, your manufacturer haven't even written the code, because they buy it from some OEM manufacturer like Bosch.

That depends. If they buy a whole unit there is a chance it is 'stock', there is a chance that the firmware was modified by the manufacturer or there is a chance that development of the software is insourced. All of that depends on volume, cost, licensing, purpose.

> And I am pretty sure that Bosch is pretty good at writing this kind of software.

Based on what evidence?

[mikrotikker]

Why do we even need computers in cars. The woz said never trust a computer you can't throw out a window. The only computer in my car is the radio.

It's just excessive consumerism and marketing crap. It's not needed.

[jacquesm]

The initial reason was emissions, but now that we have them all kinds of other stuff gets tacked on.