Firmware Update Storm?

[JohnnyG]

I got 16.32.6 last Wednesday when I took my MX2.0 in for service. By Saturday my mirrors stopped folding in when I locked the car and I could not fold them in with the button on drivers side door either. Took the car into the service center today and they said it appears to be a new firmware issue. They had to reset the "module" to fix the issue. Tech said another MX was coming in tomorrow with the same issue. Of course I left with a new firmware update (still downloading). He also stated owners could not reset the module on there own.

Here is my concern, I used to think updates were a plus however I asked the service advisor what this would have cost out of warranty and he said it would be $175 because they had to diagnose the issue. My problem is why should we have to pay Tesla to submit bugs in the firmware? Isn't being unconvinced enough?

Also I called tech support first to verify it wasn't a known firmware issue and he stated that since the button didn't work it had to be hardware.

To be clear I am still under warranty so I didn't have to pay the $175 but I would have been very upset had I been out of warranty.

FWIW, I've noticed that the SC will blame firmware anytime something malfunctions that they can't figure out the reason for; with the exception of something physically braking. They need to give some explanation, because any explanation is better than a shrug of the shoulders. I'm not saying that there aren't firmware bugs and regression, just that it becomes an easy excuse for things they may not understand. That being said, if a firmware update did somehow cause an independent control module to enter a defunct state, they would be hard pressed to get any money out of me to reset it.

 

[mrkisskiss]

It's definitely a new strategy by Tesla - probably put in place by Jim Keller now that he runs the Autopilot program.

Maybe Jim favours a "RERO" cadence - (Release Early Release Often), as it gives them a chance to test out new features, minor behavioural tweaks in AP, maybe "shadow" test new features whilst receiving useful comparator telemetry from a large fleet (one of Tesla's big advantages, so I'd rather they use that ability to full effect)

I have a P100D built in Jan 2017 - I don't get the "little" updates - only the major fleet-wide rollouts. Presumably I'm just... not that useful!

Anyway, I would imagine they flight various release candidates to certain cars for testing, maybe some builds configured to gather general telemetry, some to address specific issues (P90D issue etc), some to "shadow" trial new features and report back... and then they make their decisions whether or not to roll it out fleet wide. It does make sense - imagine if they pushed an update to every Tesla in the world that bricked it, or noticeably regressed for everyone; it'd be a PR nightmare. It's probably also linked into their FSD dev efforts; you can see that it'd be useful to be able to run simulations on the real cars, to compare what their FSD code would have done vs what you actually did (whilst driving in non-AP2 situations!)

So, I think we're in for a gamut of firmwares for the foreseeable future; potentially more than one a week, which is probably why they changed to a more granular naming scheme than Year/Month/Week.

 

[yoyodyne]

My car has spent about 7 days this summer in the shop trying to diagnose, first, a play-audio-from-USB drive issue which cropped up after a firmware update in late spring. They told me engineering had determined it's a firmware bug and they are working on it. Then again later after the shop erased my HomeLink settings trying to figure out the above and I found I could no longer initialize HomeLink to receive input from my garage remote. The shop was leaning towards replacing the entire MCU (no longer in warranty, ~$3000 for part) when, surprise-surprise, engineering said there have been some others with the same complaint so they are working on a fix for that too.

The upside is I haven't been charged for the many hours of diagnosis which is more than appropriate since their firmware update took a car that was working fine and screwed it up.

A little QA would be nice before updates get pushed out.

 

[outie]

My car has spent about 7 days this summer in the shop trying to diagnose, first, a play-audio-from-USB drive issue which cropped up after a firmware update in late spring. They told me engineering had determined it's a firmware bug and they are working on it. Then again later after the shop erased my HomeLink settings trying to figure out the above and I found I could no longer initialize HomeLink to receive input from my garage remote. The shop was leaning towards replacing the entire MCU (no longer in warranty, ~$3000 for part) when, surprise-surprise, engineering said there have been some others with the same complaint so they are working on a fix for that too.

The upside is I haven't been charged for the many hours of diagnosis which is more than appropriate since their firmware update took a car that was working fine and screwed it up.

A little QA would be nice before updates get pushed out.

They use the end users as QA. That’s why they aren’t charging you for it.

 

[jdjeff88]

I spent many, many months after a firmware update in January of 2107 trying to get my car to connect to wifi. Pre-update, no problem. Post-update, big problem. After 7 months of rangers and service centers and 7 different routers and endless settings changes I was told...it's a firmware issue and it is known and the mythical engineers are working on it. When I lost my car in the floods, my first thought (ok, not literally my first thought, but my first Tesla-related thought) was that my new Tesla damn well better connect to my wifi! Picking it up today so we will see. But I agree that the firmware is the ultimate refuge for the bug that just won't die when it comes to service.

 

[boaterva]

Are you sure nothing changed at home originally? New DVR? New TV service? Many times we’ve seen it’s the presence of DLNA servers (Plex, TiVo) on the network and not a router issue. Especially if the car can connect to wifi elsewhere.

Another test is to use your phone as a hotspot just to prove it’s something in the house network or not (using the LTE from the phone as the internet).

 

[JohnnyG]

Are you sure nothing changed at home originally? New DVR? New TV service? Many times we’ve seen it’s the presence of DLNA servers (Plex, TiVo) on the network and not a router issue. Especially if the car can connect to wifi elsewhere.

Another test is to use your phone as a hotspot just to prove it’s something in the house network or not (using the LTE from the phone as the internet).

I have a mobile hotspot (Franklin R850) that my car refuses to connect to. I literally have not found one other device that won't connect to this hotspot. Funny thing is that, I bought this hotspot to leave in the car, both for passenger's devices, along with as a back-up for the built-in AT&T LTE, in the case of no signal from AT&T.

 

[Kant.Ing]

They are doing the verification by simulating the fleet's response with different firmware versions. For example, if the silky smooth algorithm or the computer vision for AP in version x gets more driver intervention than version y, the networks or whatever in version x is outright inferior than those in version y. The team or the engineer proposing version x also gets a pay cut.

 

[Kant.Ing]

I spent many, many months after a firmware update in January of 2107 trying to get my car to connect to wifi. Pre-update, no problem. Post-update, big problem. After 7 months of rangers and service centers and 7 different routers and endless settings changes I was told...it's a firmware issue and it is known and the mythical engineers are working on it. When I lost my car in the floods, my first thought (ok, not literally my first thought, but my first Tesla-related thought) was that my new Tesla damn well better connect to my wifi! Picking it up today so we will see. But I agree that the firmware is the ultimate refuge for the bug that just won't die when it comes to service.

Try avoiding assigning 192.168.x.y as the IP for the car's Wifi, and use the other private addresses 172.16.x.y instead. The car's internal network is on 192.168.x.y so it might get confused and abort such Wifi connection.

 

[boaterva]

Try avoiding assigning 192.168.x.y as the IP for the car's Wifi, and use the other private addresses 172.16.x.y instead. The car's internal network is on 192.168.x.y so it might get confused and abort such Wifi connection.

Sorry, not so... there may be a network that Tesla uses that you shouldn't use, but it's not all of 192.168.0.0 I'm using 192.168.100.0 with no problems and have a car on it right now.

The TeslaTap article has some good ideas. Others I recall (and I can't find it now) are not to use ones that Tesla uses. May have been 192.168.0.0/24 or 192.168.1.0/24) (That's 192.168.0.0 mask 255.255.255.0 which means don't use addresses in 192.168.0.x or 192.168.1.x for the second one.)

There was also a note about changing the password for the network, not using ones with single or double quotes. A LOT of programmers don't escape special characters correctly and you need to use simpler passwords, which aren't really simpler if long enough.

WiFi Guide and Troubleshooter for Tesla Vehicles | TeslaTap

-------------

From a tesla.com forum post:

TeslaTap.com | February 16, 2017
Thanks for all the great input. I've made a number of changes to update the article.

On the passwords - after about an hour of testing, I found the single quote symbol anywhere in a password will make Tesla fail to accept the password, even though it's a valid password symbol. I've let Tesla know about the bug and a few other quirks I discovered.

----------------

Mine is a VERY long password that is all upper and lower case and numbers. Easy to enter, and nothing to upset any program, several cars have worked perfectly and so do all my Apple and other devices.

Since it doesn't work on the hotspot, I'm assuming it's not the DLNA issue; it's probably the network itself or the password. I'd try the password as easier to change (you can always change it back).

 

[SomeJoe7777]

On the passwords - after about an hour of testing, I found the single quote symbol anywhere in a password will make Tesla fail to accept the password, even though it's a valid password symbol. I've let Tesla know about the bug and a few other quirks I discovered.

Tesla's WiFi implementation in the car actually doesn't meet the acceptance requirements of the WiFi Alliance, so they're not supposed to be using the "WiFi" branding for it. WiFi Alliance requires that the system be able to connect to WAP-Enterprise access points that authenticate with PEAP, EAP-TLS, etc. Tesla's wireless doesn't support that.

It means I can't connect to the wireless network at my workplace and at home I use WAP-Enterprise as well, so I had to create a separate wireless network that uses WPA-PSK authentication.

 

[boaterva]

Tesla's WiFi implementation in the car actually doesn't meet the acceptance requirements of the WiFi Alliance, so they're not supposed to be using the "WiFi" branding for it. WiFi Alliance requires that the system be able to connect to WAP-Enterprise access points that authenticate with PEAP, EAP-TLS, etc. Tesla's wireless doesn't support that.

It means I can't connect to the wireless network at my workplace and at home I use WAP-Enterprise as well, so I had to create a separate wireless network that uses WPA-PSK authentication.

I assume also not being able to use any (all) of the 'private' networks would also exclude them from 'acceptance'. All this reverse engineering people have had to do (and the bug reports about telling Tesla to turn off the services in the kernal that didn't need to be running that mess up things when you have DLNA servers.... ) are literally nuts.