Re. the radio: Now its a big useless screen that shows me useless data while still hiding all the useful data that I can get over OBD-II. And whats worse, that screen is tied to your fucking cars computer and configures your car so you cant remove it, no matter how much the software sucks. I hate my 2022 CR-V's garbage infotainment screen. Its a shit UI, shit audio quality, and the Bluetooth is bugged to all hell. I already have a computer with me in my car called a phone that does everything but better. And that's not saying much.
I also own a 05 55 AMG, also all mechanical, but oh so impracticable :D
I also like sensors and crash avoidance tech.
So instead of cleanly pulling off my merge into a lane going 10mph faster than me I look like a goddamn moron for zipping over and then hard braking away 20mph of speed. All because some programmers buried in Toyota HQ somewhere spent too much time on the HN or Reddit or whatever circle jerking it in the comments with the "you can never go wrong by braking" crowd. Could have been way worse had it been a spicer situation, like merging into traffic with a disabled vehicle at the end of the merge ramp or just about any other case with equal or great speed differential and equal or lesser margin.
A car should do what I say. I can understand doing something when I have provided no input or perhaps ignore a 0-100% press to prevent wrong pedal accidents but this is just horrible systems design. If I'm traveling at speed and mash the gas it stands to reason I did that on purpose.
A lot of the other stuff, though, I agree with you.
But, it was a beauty.
On the other hand, coupes with rear hatches can be particularly bad: https://www.motortrend.com/uploads/sites/10/2015/11/2004-toy...
Looking at the extremes like the Honda CRZ, it does seem to be a low priority area!
There's plenty of kids on my street, and I'm much more comfortable knowing everyone has one when backing out of a driveway, and not just the people who bothered to go get one installed aftermarket.
But the realistic option that worked immediately was mandating backup cameras.
(I would note that even in a world where everyone backed up into their driveways and parking spots, mandated backup cameras would still be a good thing.)
I'd feel more comfortable if we'd raise the standards for the driving test so that only responsible people can drive.
To give a programming analogy, this is like saying "we can prevent memory safety issues by only allowing good programmers to use C". Everyone makes mistakes.
Everyone else can enjoy their reflection/replay attacks or whatever.
Honorable mention to Toyota who has still not completely abandoned this simple, functional technology for a clunky fob that can be easily hacked.
FWIW, fobs are not for your convenience. It's for theirs.
Same with touch buttons. Not for you convenience, it's for theirs.
I do think the writing is on the wall for old fashioned keys, though? For one, they don't really give you that much protection. As laughable as poorly done key fobs are, a physical key is a pretty low bar as far as deterrence goes.
It can be annoying to consider, but cultural norms protect cars far more than anything else. Is why many in suburban areas can get away with having their keys in the cars at all times.
Fobs just created another attack vector catering to people too lazy to take it out of their pocket or purse.
Keyless start has another legitimate function besides laziness: it allows you to leave your car locked with the engine (and AC) running while a baby or dog is inside.
Of course, you can accomplish the same by having two keys with you; you decide whether that's another example of laziness. :-)
Some aftermarket remote starts have this feature.
However, in many states it's illegal to leave a car running unattended.
Though one could argue in court the baby or dog could serve as the attendant. Having said that, leaving a baby or dog unattended, AC or not, is just stupid.
Here's the repo: https://github.com/joelsernamoreno/EvilCrowRF-V2
Chevy's pre-2008 were in a good spot, maybe 2007 for the avalanche body change? Quite a few Hondas and Toyotas were good through the early teens, especially the 4 cylinders.
I have a late 80s GMC pickup, 2005 Buick, and a Chevy Volt. The only one I have any real issues with is the Volt, though that's only been the last couple years as the battery is getting old; the most frustrating thing is needing to run questionable software on an airgapped laptop just to turn the Volt back on when a high voltage safety flag is flipped tripped in the computer.
Sometimes the platforms that the OEMs don't care about are great because the idiot dick swinging engineers who want to hit their KPIs neglect them. Sometimes they're terrible because they get phoned in. The flagship platforms are usually safe but sometimes they put too much bleeding edge tech in them.
This situation is not a recipe for good code. Now that BLE has audio (the last thing from classic that it lacked), we can begin phasing out BT classic and this mess. However, it will be a decade before anyone can safely drop bt classic interop.
Basically: anywhere you have a Bluetooth stack that supports bt classic, feel free to ASSUME there are RCEs and DOSs lurking. You will not be wrong.
Source: a full blown case of PTSD from having written and debugged a few BT stacks
Of course I don't expect it's implemented anywhere near securely, but in theory it's very possible. Game console companies have this stuff pretty solved.
The infosec community loves their weasel words don't they?
The only other career path other than "meteorologist" where they get it wrong half the time with the burden of proof on the recipient, and everyone looks the other way.
Show your work, or it's not possible.
There are cases where vulnerable code is found, but it may take weeks of tinkering to actually build an exploit that gets arbitrary RCE.
An example could be a buffer overflow that only allows a few bytes to be written. At first, you're likely just causing segmentation faults. DEP and ASLR will make writing an exploit that gives RCE difficult. This is when an attacker "may" be able to do something, if there's an attacker determined enough to figure out a full exploit.
The original researcher might not be interested in spending that time and just wants the vendor to fix it.
https://arstechnica.com/information-technology/2023/04/crook...
minusLik•7h ago
noman-land•7h ago
Ccecil•6h ago
It is an interesting topic for sure.
minusLik•6h ago