Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Tesla should open-source the MCU code

This site may earn commission on affiliate links.
Seriously!

  1. The company needs to cut costs. Obviously there's money to be saved by scaling down their own team.
  2. A lot of skilled people will contribute much needed bugfixes and features as merge requests for free.
  3. Maybe finally someone will fix the USB bugs.
  4. There is nothing secret in the MCU code that will significantly help other manufacturers. Any company can build something similar.
  5. Tesla would be in charge of direction and accept the merge requests of sufficient quality.
 
Strange coincidence I was just thinking the same thing. Open sourcing would be a win-win for the community and for Tesla. There is very little proprietary tech on the MCU (after all it is the "Media Control Unit"), so this should not be a barrier.

So let another company or group build something similar and open code it. Nothing is stopping them..

Actually, there is quite a bit. All software components are tied to a single package. If MCU software is updated by a third party (which would require rooting), there are version/hash mismatches. Those with the knowledge can get around it, but apparently it's getting harder and harder. Using third party MCU code would essentially leave the rest of the car's software frozen and non-updatable. Not a solution.
 
  • Like
Reactions: jerry33
Elon originally said they'd open up the system to 3rd party apps, but it never happened.

Personally I don't think the MCU software should be opened up for security/safety reasons. My solution would be to make most of the UI be a set of browser windows and allow sandboxed apps to run in them with a limited set of APIs into things like the Media Center.

If performance was that much of an issue (which I doubt), make sure it supported ASM.JS.
 
Elon originally said they'd open up the system to 3rd party apps, but it never happened.

Personally I don't think the MCU software should be opened up for security/safety reasons. My solution would be to make most of the UI be a set of browser windows and allow sandboxed apps to run in them with a limited set of APIs into things like the Media Center.

If performance was that much of an issue (which I doubt), make sure it supported ASM.JS.
What does the MCU have to do with safety?
 
I know from personal experience you can drive the car with a dead MCU... after you figure out how to open the doors to get in. The only thing you loose is the displays and audio effects. Not sure if the blinkers or wipers worked.

BTW, remember when v8 had that "Apps" screen with the lonely Calendar icon, teasing us with hopes of a Tesla App store?
 
That means they can steal/copy the most customer facing of all the software, ones of the parts that separates them most from competitors in the eyes of many people.
A handful of of junior devs could duplicate the functionality in a relatively short period. There's really not much to steal.

Or better yet, they could build it out of already-exiting open source components and come up with something much, much better.
 
  • Disagree
Reactions: APotatoGod
I'm not sure what the problem would be. Seems like Tesla uses Linux and the Linux kernel is worked on by thousands of people around the world. It is the largest software project in the world, run on more devices than anything else. It has proven to be very secure.
 
  • Like
Reactions: APotatoGod
As Tesla does NOT sell code - AND all the code is for Tesla products (internal use) - I don't think Tesla is in violation, right?
Tesla could share code they THINK might be helpful. At the kernel level - perhaps they don't get that deep?

Home - Automotive Grade Linux
Tesla not yet a member
Members - Automotive Grade Linux

I think a fine strategy for Legacy OEMs would be to join the above and take over (or work toward) managing this project.
Other more experienced in Open Source management might comment? thanks.

Germany or Japan or Koreans or even US automakers (and don't forget China or India) - any of them could take on the challenge.
This would seem to me the quickest way Legacy OEMs could start to compete with Tesla in the software area.
 
I'm not sure what the problem would be. Seems like Tesla uses Linux and the Linux kernel is worked on by thousands of people around the world. It is the largest software project in the world, run on more devices than anything else. It has proven to be very secure.

the kernel is pretty secure IF you keep up with the upstream patches (no idea if tesla does; my guess is that they, like most of the vendors that use linux, do NOT keep up with all the updates and have a guy (or even better, more than 1) who decides what version of this lib or that app they take. they have to test it and lock it down (the changes) and only permit changes they agree with and have tested (and feel safe about). that's a lot of work and most companies do NOT do this (very sad to say that but its true, and its my experience in the industry; 25+ years in bay area, etc)

userland is another story, altogether ([airplane]userland is another story[/airplane]). there are bugs and they sometimes go a long time before being found. I use linux, have been for quite a long time, I love it, but I also realize that its not the right tool for safety-critical things unless you take significant pains (MISRA, anyone?). I have my doubts as to how well tesla does all this (I have not seen any of their code, and frankly, don't want to). but I have a hunch that there are some major security holes that a good pen-testing team would find. thru CAN, thru LIN, thru places you may not even expect. I've seen it, indirectly, and it amazed me.

I'm a VERY open source person, but for cars (a field I work in, currently) I'm not sure I agree. there are even lower level OSs than linux that have more safety certification-based designs and limitations (limits are good, for embedded).

besides all that, to be really effective at working on car systems that have so many dependancies, you have to be 'in' the group, have good access to info and even then its a challenge ;) this is not for outsiders and its hard enough to manage insiders.
 
  • Helpful
Reactions: Brando

I've looked at AGL and thought it was a neat idea, but I don't see any vendor buying into it, in any serious way.

so far, there is no common vendor or source for 'smart cars', but I do think that in 10 years or less, it will be just commodity that they buy, like clicking boxes on a webpage and pressing 'customize my automotive OS and apps'. not kidding,
 
  • Helpful
Reactions: Brando