dr bob

Tesla may be a shining star on stuff like this. They've offered significant feature upgrades to cars they've already sold. That while the big guys have only offered bug fixes and patches when a large enough owner share complains about something.

----

Software copyright protection, such as it is, keeps gas in my car and a roof over my head. And there's a good chance it helps with your electricity costs. There's a lot more to the industry than movies and such. A lot of the 'corporate bad-guys' are you and me, as share-holders and owners. Plenty goes into development and implementation. Without legal protection, a creator/developer/manufacturer needs to recover all those costs with the first sale if they know that there won't be any more. Licensing is already a part of a car purchase. My recent Honda purchase includes a massive manual for the digital entertainment system and the navigation system, incuding many pages of software license information. This strays from the original discussion, but gives an idea that we are not buying a '64 Nova with an AM radio any more.

928NOOBIE

Sue everybody...

dr bob

There ya go!


I invited some friends over to watch a movie I thought I had "purchased". I had put it on my media server. Others came and watched it with my wife. All very clear violations of DMCA.

I figured out a few ways to add life to some power plant equipment, while making the whole thing more economical and safe to operate. According to one control system manufacturer, only they can build and install code in their computer systems. Once installed, the rights to that code belong to them. Good News is that my IP attorneys blew a hole in their successive-ownership claims.

There's a lot more to digital rights than meets the eye.


Meanwhile... If you decide you want to open up someone else's source code, modify part of it, and then sell it, you are technically violating the original owner's copyright. The exception is reverse-engineering, which allows you to independently develop your code based on ultimate function, without ever looking at or knowing what was previously installed. If you have an idea to make someone's existing software better and decide to hack it and resell the modified version, contact a good attorney and invest in some of their review time. It can save you a big amount of money in the long run. Just sayin'...

The Fixer

More persecuting individualism and independent thinking.

Who'd have seen this coming?

upstate bob

I'm happy to know there is no software in my '79, or my 62 bug, or my 29 ford. Unfortunately my '14 transit connect is abounding with such crap. I'll never know what all the buttons are for, including the ones on the steering wheel. If you want to take control of a new car I would buy a standard shift model. The engine is hopelessly intertwined with the auto shifters.

Speedtoys

Anyone like me..who will bet that nobody in the thread wrote their senator(s) a letter about this yet, or would without a little shaming while being angry about it.


..just a reminder, form letters to "fill in grievance" and Email em out..

James Bailey

the 996 and 997 store the history of overrevs.... prospective buyers are now requesting the downloads as part of the inspection. Porsche on their CPO cars will not accept any cars with the higher level. The 996 logs on two but the 997 splits them into six !! Interesting read if you look into it and Porsche has used this information to deny warranty work. Clear evidence of ABUSE ...
Yes all kinds of information is being stored in those black boxes !!

The Fixer

It makes me really cherish the older cars,

(not old cars with carbs.. recently had one of those again, choke to start, then 1/2 choke, geez can't believe i used to do this every day!)

But my '83S and my '86 Carrera (RIP). They are real fun to own, old enough to be different and i don't care that they aren't the fastest car on the road aymore.

But I just don't see how a Porsche Panamera could be "restored" 30 years from now by an enthusiastic 20 years old. (the way i was at 20)

Seems they would be too complex and expensive at some point to repair.

Probably best to recycle.

But maybe the tech will help them be repaired and stay on the road.
They won't rust like the old cars did that's for sure.

Opinions?

worf928

Folks, best sit down for this...

First: Software has paid for and continues to pay for my family's Porsche habit. I've written lots and lots of software and managed teams that wrote software and ran a company that did software. I've done damn near everything from embedded control systems to user interfaces to machine learning to autonomous systems to software that controls the machines that make the chips that surround us.

Almost everything on modern cars is controlled by software. Oil Pressure? Yup. Variable based on conditions and commanded by the software. Timing? Spark? Injection? Throttle plate opening? Brake pressure? E-brake? Window switches? Yup. All of it. Controlled by software. And there's no documentation anymore. Software says a sensor is busted? Change it out. No documentation on it's specs. Can't test it on a bench.

The fundamental difference between our 928s (and other cars made in that era) and cars that started hitting the road in the mid-2000s is that it is no longer possible to know how some bit of the car is supposed to work or the principles that govern its operation without some kind of access to the design documentation (should any exist) or source code for the software.

If you understand how something is supposed to work and why it is supposed to work then it becomes possible to figure out what's busted when it doesn't work. If you don't know what it is really supposed to do - under all normal and exceptional conditions - and you don't have some understanding of the theory of its normal operation you have no hope figuring out what's wrong with it when it seems like it doesn't work.

That last sentence is the problem with today's cars. Not only do we not have knowledge of the software, neither do the Porsche techs at dealerships. Furthermore, the diagnostic tools that are made available to them provide neither understanding nor context.

The above is not something new to me. 15 years ago I knew it was coming. I watched and hoped that the car companies would 'figure it out.' They haven't.

I got to witness the consequences with my wife's 981.

Long story. I'll try to make it short... (and will fail.)

Just after the break-in period she began to really exercise the car. After a couple of days the car began to behave bizarrely under full throttle. It would get to about 4000 rpm and then one of three things would happen:

- acceleration would stop and the car would just hold 4000 rpm (with the pedal floored) as if timing had been pulled.
- acceleration would drop-off to almost nothing for one or two seconds then pick back up.
- the engine would stutter as if it was hitting a fuel cut off limiter.

The symptoms would sometimes result in a 'code' on the dash: reducing operating power, PDK in safe mode, engine control fault, etc.

The only pattern was that these symptoms would only occur after the engine had been running for 30 or so minutes: it had to be heat soaked. Not hot. At operating temperature for a while. Drive it around for 10 minutes and it was fine at WOT. Turn it off and let it sit for 15 minutes then restart and drive and the symptoms would occur.

It was under warranty. So off to the dealer. After a couple of days of them not being able to repeat the issue (is anyone surprised?) I went to the dealer and taught the master tech how to repeat the issue.

So, we hooked up the PWIS laptop, put it in data logging mode and logged data. The log was some of the most random bizarre crap I've seen. Everything would be fine and then all of a sudden the log for throttle position would go to zero.

But, was throttle position the sensed pedal position or the commanded throttle opening or the measured throttle opening? I didn't know. The log didn't say. The master tech didn't know and Porsche wasn't telling. There were other wacky log lines. I don't recall all of the specifics. There was ton's of data in the log but there was no context with which to interpret it.

A dealer tech can't do anything under warranty and expect to get paid for it if PCNA doesn't authorize it. So, with no clear indication from the log the MT's hands were tied.

We grabbed a couple more logs and the MT did the only thing he was allowed to do: e-mail the logs to PCNA tech support.

There are three, count'em - three, tech support guys for Porsche North America.

A couple of days later the order comes in to swap coils. Nada. Then a couple of days later swap something else. Then replace the right-bank variocam actuator. Then pull an ECU from another 981 and try that. Then... wait... let me tell you about the !@#%$ing ECUs...

In modern Porsches there are a bunch of control units. Motor, front suspension, rear suspension, chassis, etc. They are all VIN coded. So are the keys. If you swap an ECU you have to recode the VIN for all control units and the all the keys. If you don't do that the car will do NOTHING. You stick an uncoded key in the ignition and you don't even get warning lights. Of COURSE you can only recode with the PWIS laptop. It's a $1700 per month lease and it only operates with an Internet connection to the Mothership.

... ok. Back to the 981. They swapped a couple of thousand dollars worth of parts in and out AT THE DIRECTION OF PCNA TECH SUPPORT over the course of 3 weeks.

In the end it was gummed-up fuel injectors.

(CarChick only, ever, uses tier 1 93 octane fuel. The 981 has never sat for more than a week without stabil in the tank and it is never driven for less than 45 minutes. So, it wasn't user error. She bought the car off the lot. It had been on the lot for 6 months. My theory is that during that period the car was started and stopped many times to move it around the lot and all on the same tank of crab pi$$ ethanol-laden gas. And that's what gummed the injectors. But it wasn't until she stated WOTing around that it became a problem.)

So... why would a couple of gummed up injectors cause the engine to hit a wall? A 928 will continue to run pretty-ok even if the injectors are thoroughly gummed-up. And how does my wife's 981 tie in with the beginning of this too long post?

I spent non-trivial time with the MT. This is what we put together: One of the things the ECU software does is protect the catalytic converters. It has to because of the US 100k-mile warranty requirement. So, ANY time the ECU sees over-fueling it acts to protect the converters. This gets more interesting because the piezoelectric injectors used in DFI engines implement a feedback mechanism to the ECU that allows the ECU to compare it's commanded pulse with the actual pulse on an injector-by-injector basis. So the ECU can look at the wide-band oxygen sensors (four of 'em) as well as the feedback from the injectors to figure out how much fuel is being delivered.

Nifty. Except that the folks that wrote the software didn't take into account that injectors can get gummed-up. So, they never wrote code to deal with a set of nonsensical, exceptional, but quite possible inputs. I guaranty you that somewhere in that code is something along the lines of an if-if-else-else(*) statement with nothing in the 'else' part because the programmer said to him/herself: "nah. that can't ever happen so I don't need to worry about it."

(*) For those of you familiar with modern languages: an empty catch block that swallows the exception silently.

I spend, literally, hours and hours per week pointing at people's code and telling them to deal with these 'conditions that can't ever happen' and also to actually put meaningful context in trace log statements. When you have an autonomous system that drives off a cliff the only thing you might have left is the trace log. If the log doesn't tell you exactly which 'impossible condition' actually happened you are screwed.

And let me tell ya: in a 24x7 mission critical system that's doing real work the 'impossible condition' will happen every couple of days.

So, in summary:

The software is a black box to us because:
- the car companies want to keep independents out of the business. Maybe.
- the car companies are worried about liability. Definitely. But why?

... because they absolutely do not want independent reviewers to see the crap that they ship.

- the car companies(*) don't have a clue about how to make it transparent in a way that doesn't release their IP to the world or give their lawyers fits.

(*) Apparently Mercedes may be the joker in the deck here. The MT told me that, according to his Merc Master-tech friend, Mercedes is actually doing extensive training course for dealer techs on the software. Who knows.

Last, the techs who weren't born with a smart phone in their hand are really not happy with this situation either. They've become parts-swappers that follow trails of (trace log) bread crumbs off a cliff. They are screaming for knowledge of the software so that they can actually use their brains and make their customers less-pi$$ed off.

kevinr

As does my Cayman. Four levels as I recall on mine, the lowest (most sensitive) level being a simple rev limiter event.

RKD in OKC

If you are having trouble with an issue on a new Porsche, hope and pray the factory tech responsible for your issue goes on vacation. Instead of going to same level techs it goes up a level for review and THAT guy will double check everything and typically find the problem.

Oh, and they do not listen to the owner no matter how good of a problem solver they may be or how correct their evaluation of the problem is. They will dismiss it because you are NOT a tech, what do you know.

I am happy that the data logging is there. When the issue and attempted solution created a condition where the car spun a rod/crank bearing, the logs determined it was not driver error and Porsche put in a new motor. The dealer tech didn't even get to inspect anything. He checked the logs and sent them to Porsche, they said the problem was not driver error. Drained the oil, and it had metal shavings. Porsche said swap engines. Tech didn't even get to crack it open. I was very grateful of the log results because it spun the bearing at an autocross event.

This was on a new 2001 Boxster S. It was a great car for two years and ran like it had more than the stated HP. Then one morning it started taking more than one turn of the key to start. I took it in cause that should not be. They determined the MAF had gotten dirty and out of calibration and replaced it. That fixed the starting issue, but introduced a new issue. On deceleration it would backfire out the exhaust. On three occasions the backfire was violent enough that it blew apart the welded seams on the cats.

They tried all kinds of things over a year's time including replacing the MAF again. They even did a valve cleaning with some kind of solvent, then did not change the oil after. Two weeks later it spun a bearing and Porsche replaced the motor. Which did not fix the issue. When I took it in for yet another cat replacement the factory tech was on vacation. The techs supervisor reviewed the case and discovered there were two different MAFs for that year of Boxster and after checking the part numbers my dealer had put the wrong one on my car, then replaced it with an identical wrong one. New correct MAF and the problems were all sorted. The real downer was the the replacement motor was at least 15 hp down from the original. It was noticeable.

Each time I took the boxster in, I told the service advisor, tech, and even the factory rep that the backfiring problem started with the MAF replacement. There response was, it's and new one, that couldn't be it. Bet there is a tech bulletin about multiple MAFs now.

There was also an issue where the clear coat started chunking off the rear brake calipers. The service advisor and factory rep said it was because I raced it. I asked why racing would cause the clear coat to come off. They said it was the extra heat. I said why then was the clear coat coming off the rear calipers and not the fronts when the front brakes get a lot hotter than the rears when racing. The factory rep approved the $8000 replacement of the rear calipers.

Adk46

Excellent post, good story. I know enough about coding to understand the empty catch block error. What will people do when they find one of these in a barn someday?

One detail that I don't think has been mentioned: the ECU code is actually encrypted. You cannot read it, modify it, and burn a ROM chip. I've read a thread about tuning a 981 that described how you can open up the intake for more airflow, but the ECU will just close the throttle to compensate. It will even do this on an unmodified car in certain operating regimes. [WOT-driver] != [WOT-engine].

GlenL

Software pays for my Porsche habit, too. One thing I did recently was ban "try/catch" in a new, life-sustaining medical device.

Your point on software quality in embedded system is well taken. Eventually that unpredictable thing, or combination of things, happens and the code needs to be ready.

I'd like to understand the nature of the copyright enforcement that the car maker's want. Not allowing changes to their code provides revenue from upgrades and avoids taking responsibility for others' mistakes.

SeanR

Great stuff and I'm going to share a small story that my father (RIP) and I had one day.

He operated his own Euro shop for a good few decades and really hated the 928 when it came out. It wasn't a 911, was not easy to work on, was new and "had too many Goddamn computers"

So forward to a few years ago after he retired. We were discussing some of the cars I was working on that week and made the comment that I did not want to work on anything newer than 2006, he asked why and I said "too many Goddamn computers". It was silent for a few moment and then we both busted out laughing.

Drewster67

Time to return to old school engines