Tesla's world of horror...

[Karel]

"A former Tesla employee, who worked on their IT infrastructure, is posting in a subforum of a subforum,
a little-known place for funy computer forgotten by time. His NDA has expired.

He has such sights to show us. Join me and I will be your silent guide into a world of horror."

https://mobile.twitter.com/atomicthumbs/status/1032939617404645376?s=21

[Halcyon]

 

[ebastler]

If you don't want this second-hand via Twitter, here's the actual forum thread, filtered for posts by that specific user:
https://forums.somethingawful.com/showthread.php?threadid=3862643&userid=20544

He might not be entirely impartial. 
But an interesting read anyway.

[Vtile]

Like all other fkn IoT shits.. Blob of shits glued to gether.

[Marco]

Should have included this post in the OP :

man twitter is so dumb this isn't even news. big shock - a large company has some issues with development and some of the issues were lol bad but I guarantee you the technical issues i posted are long gone now

bringing up hardware is always messy and with tight deadlines, poo poo happens, so to be real for a moment: in my opinion tesla is shipping good (model s/x) products, they are road safe and tested, there are growing pains, quality pains and such but nothing they can't fix. model 3 is another story, I hope they fix it

honestly i started out really liking working there and I feel like my 8 years there had an impact and tesla to this day is still full of good folks doing the right thing. i know for a fact all the crap i worked on is probably gone by now as it should be - they are evolving. you'll notice they haven't had further incidents like the thing I posed earlier with the reboot loop, so at the very least rca's are paid attention.

most of tesla's problems are self-imposed management style issues, they'll be forced to fix them at some point, one way or another, but that's up to the board

He also thinks Musk is pretty much par for the course in Silicon Valley.

[coppercone2]

looks like typical corporate? What do you expect working for a company?

no one ever wants to fuck with a cash flow that is working. I am surprised they don't have priests walking around blessing stuff there. The man for got his psychosis hat before work.

[Bud]

If you don't want this second-hand via Twitter, here's the actual forum thread, filtered for posts by that specific user:
https://forums.somethingawful.com/showthread.php?threadid=3862643&userid=20544

Man, this is a golden mine of awesome stuff

we once patched openssl to ignore client cert expiry because somebody forgot to create a process to update keys in the field and all the customer cars started falling offline because their certs had expired.

tesla isn't encrypting their firmware and it's really easy to glean information from the vpn with a packet cap because nothing inside the vpn (was) encrypted

they're [fw updates] done over cell and wifi, depending on how big or urgent the update is. the network does generally provide security in the transport sense - the backend systems are what worry me and tesla is a big enough target that a determined actor could gain control

china has a law in place that mandates all electric cars send real time telemetry to their government servers

also on the supercharger note - you can get blacklisted from using them if you charge on them all the time

The last one is just blows my mind. No juice for you buddy, if we decide so. So you may want to keep your bicycle.

[amyk]

I do electronics and software myself, and even I don't want a Tesla. They're locked-down surveillance machines, and apparently quite buggy too.

If you want to drive something that you actually own, get into the hobby of classic cars. It's fun.

[Hypernova]

It's funny that they are effectively DDOSing themselves every rush hour.

there was no battery backed rtc

Why...

[Ice-Tea]

there was no battery backed rtc

Why...

My guess: because batteries run out.

..

I know, kinda funny in a battery-on-wheels-car.

[JPortici]

the assumption may be that until the main battery dies or it get disconnected for servicing, that part of the circuit will always be powered

[Hypernova]

the assumption may be that until the main battery dies or it get disconnected for servicing, that part of the circuit will always be powered

RTC dying due to main battery being disconnected sounds semi-reasonable as that happens with normal cars too, but then:

when the system reboots (sleep, deep sleep, reboot, whatever) the car is at tyool 1970 until it gets ntp again.

Are they not using a dedicated I2C RTC chip on their motherboard? Or did someone in a fit of insanity decided to add a line to their boot code to reset it on every SoC reset/wake.

[JPortici]

ah, yes i forgot about that line 

[EEVblog]

Our own Free_Electron (head PCB designer at Tesla) is suspiciously quiet 

[jmelson]

Our own Free_Electron (head PCB designer at Tesla) is suspiciously quiet 

He might have been among the 5000 that got fired a few weeks ago.

Jon

[NiHaoMike]

If you want to drive something that you actually own, get into the hobby of classic cars. It's fun.

I have wanted a first generation Honda Insight with the electronically controlled manual transmission, but then I read about all the problems they have. Namely in that it's infamous for the batteries failing, and although it's possible to retrofit in a pack based on Nissan Leaf cells, it's difficult to make effective use of it within the limitations of the Honda hybrid system. Toyota's hybrid system is a much better match for an upgraded battery pack.

[coppercone2]

do they have to sign a NDA to work there since there is so much automotive shit going on right now?


if vincent got fired he might have to sit quiet about it for 3 years. they probably have a vicious lawyer squad

[dr.diesel]

if vincent got fired he might have to sit quiet about it for 3 years. they probably have a vicious lawyer squad

I would seriously doubt that, but in any case they can't stop him from acknowledging his employment status, but just about any other detail they could.

[coppercone2]

if vincent got fired he might have to sit quiet about it for 3 years. they probably have a vicious lawyer squad

I would seriously doubt that, but in any case they can't stop him from acknowledging his employment status, but just about any other detail they could.

I heard a NDA is like 6 months in a dead industry and 3 years in a booming competitive one right?

[dr.diesel]

I heard a NDA is like 6 months in a dead industry and 3 years in a booming competitive one right?

Most of my recent NDAs have been 3 years.

[SparkyFX]

Pretty much snafu for infotainment in connected cars, if you ask me. The whole telemetry idea is to lay the foundation for shaving money off an already sold product, but it tears down the term ownership in a legal sense by converting functions to services.

Then there are the paradigms of automation/mobile colliding with those of web technology.

[janoc]

I have read that thread and, frankly, apart from the known, done on purpose stuff (telemetry, remote access/disabling your car, the supercharger blocking), there are no real smoking guns or insanely crazy/unsafe things.

Is it crap engineering in many places? Sure it is. However this is pretty much what you get in any company where software isn't the main focus. Fixing bugs and rewriting legacy stinkers isn't done because it costs money and takes time from getting the new stuff (that actually will hopefully bring money) out to the market.

If you are horrified or laughing at the supposed incompetence then you have really not seen how most software is being developed, tested and deployed ... It is a wild west, despite boatloads of books being written on software engineering methodologies, new methodology du jour appearing every few years (it was extreme programming, then agile, then scrum, before we had SSADM and Prince ...). However, no matter which methodology and process you have, it always gets trumped by management decisions prioritizing other things because they don't see value in doing things properly (costs money, time, resources, nobody cares about that it will cost 10x as much money, time and resources to fix when it blows up 6 months down the road at the most inopportune moment).

And especially automotive (and embedded firmware in general), where the code is often written by some intern or the lowest bidder sub-sub-subcontractor instead of someone who actually has a clue how to write software is likely one of the places where the worst crap gets written.

This isn't new or somehow specific to Tesla - the guy explicitly named Bosch (Bosch was actually the supplier who delivered the cheating code in the original Dieselgate), but similar things were discovered elsewhere too. E.g. Toyota after the unintended acceleration fiasco:
http://www.safetyresearch.net/blog/articles/toyota-unintended-acceleration-and-big-bowl-%E2%80%9Cspaghetti%E2%80%9D-code

Considering that stuff is actually controlling safety critical functions and that there is a heavy pressure by car makers on the regulators to allow fully "drive-by-wire" (i.e. with no mechanical connections between the steering wheel and the wheels) and autonomous cars, it should give you a pause.

There is a running joke that the IP protections, keeping everything proprietary and NDAed is not meant to protect the IP of a company but the brand in case the horrors the company products contain leak out ...

[Cyberdragon]

It shouldn't be too hard to hotwire on or even permanantly bypass the "screw you" cutoff in a supercharger (assuming you owned it of course).

[Zucca]

china has a law in place that mandates all electric cars send real time telemetry to their government servers

Yes, I confirm if you want to sell e-cars there you need to do that.