2021.24.28 Rolling out to MCU1/FSD MS/X

2017 MS with MCU1/FSD

First significant firmware update in over a year!

Old firmware 2020.48.37.8 (0.6% of TeslaFi Fleet AWT upgrade)
New firmware 2021.24.28 (0.4% of TeslaFi Fleet running this version)

Guessing no FSD Beta access but I'm 6 months closer than 2020.48. Lol.

I h

 

[Workbirdy]

Yeah I'm not a fan of that update went back to an older 1 now my car work right

 

[KalJoMoS]

One thing I forgot to add to my last post was that apparently after the update the rearview camera image is now blurred, like the image is doubled. It was clear and crispy before.
And for the vampire drain section I can say (before this version) it was on my car about 5-6km a day, even if ambient temperature was freezing. No app’s connected, expect the mothership.

 

[brico84]

Yeah I'm not a fan of that update went back to an older 1 now my car work right

How did you downgrade?

My understand is this is not possible.

 

[Workbirdy]

How did you downgrade?

My understand is this is not possible.

I have a friend that dumped the software and reinstall an older version its definitely possible the car is a pc and my friend is a software engineer for another company

 

[t3sl4drvr]

Got 2021.24.28 on Feb 16 on my 2017 Model S (MCU1, FSD HW3).

Today the car asked to install 2021.24.28 again. That was quite confusing.

Are there multiple "versions" of 2021.24.28 around? Have anyone else had to install it twice?

 

[Workbirdy]

Yeah I believe it's a 2021.24.28.2 version out that fixed 2021.24.28

 

[Wallacechan]

Got an issue with this update 2021.24.28 , the cluster screen occasionally reboot while driving, brought it to Tesla service center this morning. It seems to be software issue, and need to wait for Tesla to release a new update to fix it, nothing can be done now.

 

[dgatwood]

View attachment 806059
Got an issue with this update 2021.24.28 , the cluster screen occasionally reboot while driving, brought it to Tesla service center this morning. It seems to be software issue, and need to wait for Tesla to release a new update to fix it, nothing can be done now.

It looks like Tesla changed some of the configuration silently so that the IC shuts down more often when the car is locked, i.e. it reboots almost every time I get into the car now. I'm assuming this is a workaround to try to mitigate the severity of this safety flaw.

That said, even with that mitigation it remains a very serious safety flaw. Every time the IC reboots, if autopilot is enabled at the time, autopilot disengages (fortunately with several seconds' warning, but it is still potentially dangerous). And losing your speedometer means the vehicle ceases to be street-legal, i.e. this is truly a top-tier safety violation.

Yet in spite of the fact that this bug makes MCU1 Teslas technically illegal to drive almost anywhere in the world, Tesla has chosen to leave this bug unfixed for three full months, focusing their efforts instead on AutoPilot functionality that is not safety-critical.

I'm not sure how much more patience we can afford. NHTSA needs to compel Tesla to call a "code red", making this the top priority for their entire engineering team, until this critical safety regression is fully corrected. The current state is simply unacceptable. This should not have gone a week without correction, much less three months, and the fact that Tesla shipped this fundamentally flawed update out to every MCU1 car tells me that their telemetry is woefully inadequate. This level of crashing should have been caught when there were just a single-digit number of cars on this release, and Tesla should have held the release until this flaw was fixed.

There is absolutely no excuse for Tesla sending such a fundamentally defective software update out to broad release, period. And if I were in charge at NHTSA, I'd make sure Tesla got fined a million dollars per day until they fix this grossly incompetent mistake.

 

[Tesomega]

It looks like Tesla changed some of the configuration silently so that the IC shuts down more often when the car is locked, i.e. it reboots almost every time I get into the car now. I'm assuming this is a workaround to try to mitigate the severity of this safety flaw.

That said, even with that mitigation it remains a very serious safety flaw. Every time the IC reboots, if autopilot is enabled at the time, autopilot disengages (fortunately with several seconds' warning, but it is still potentially dangerous). And losing your speedometer means the vehicle ceases to be street-legal, i.e. this is truly a top-tier safety violation.

Yet in spite of the fact that this bug makes MCU1 Teslas technically illegal to drive almost anywhere in the world, Tesla has chosen to leave this bug unfixed for three full months, focusing their efforts instead on AutoPilot functionality that is not safety-critical.

I'm not sure how much more patience we can afford. NHTSA needs to compel Tesla to call a "code red", making this the top priority for their entire engineering team, until this critical safety regression is fully corrected. The current state is simply unacceptable. This should not have gone a week without correction, much less three months, and the fact that Tesla shipped this fundamentally flawed update out to every MCU1 car tells me that their telemetry is woefully inadequate. This level of crashing should have been caught when there were just a single-digit number of cars on this release, and Tesla should have held the release until this flaw was fixed.

There is absolutely no excuse for Tesla sending such a fundamentally defective software update out to broad release, period. And if I were in charge at NHTSA, I'd make sure Tesla got fined a million dollars per day until they fix this grossly incompetent mistake.

I am assuming here that you have reported this issue with NHTSA ?. I have MCU1, don't have the issue you had reported. Also, I have turned off Energy Savings features as it make it very slow for MCU1 to boot up...

 

[dgatwood]

I am assuming here that you have reported this issue with NHTSA ?. I have MCU1, don't have the issue you had reported. Also, I have turned off Energy Savings features as it make it very slow for MCU1 to boot up...

Yes, I reported it months ago.

Which Autopilot hardware are you using? I strongly suspect that the problems are specific to the MCU1-HW3 combination, and might even be specific to the HW2.5 -> HW3 upgrade path, either because of the camera differences or the presence of a second RADAR input.

The failure is likely not energy-related, as I initially saw this manifest itself as:

• Randomly being unable to use the driver's side door controls for half a minute at a time
• Messages on the IC screen saying that it is unable to communicate with the MCU for half a minute
• MCU becoming unresponsive to touches for up to half a minute

with some of these symptoms sometimes occurring simultaneously. At this point, the other communication problems have subsided (I haven't seen them since a week or so after the "upgrade"), but the IC reboots and autopilot disengagement persists.

What's interesting is that when it fails, you get a "Take over immediately", but Autopilot remains engaged, and the IC remains lit. And then as soon as you disengage Autopilot, that's when the IC reboots. It makes me wonder if there is a memory leak that causes the IC computer to hit some low water mark memory-wise, and then reboot itself as soon as it thinks it is "safe" to do so.

 

[Wallacechan]

Yes, I reported it months ago.

Which Autopilot hardware are you using? I strongly suspect that the problems are specific to the MCU1-HW3 combination, and might even be specific to the HW2.5 -> HW3 upgrade path, either because of the camera differences or the presence of a second RADAR input.

The failure is likely not energy-related, as I initially saw this manifest itself as:

• Randomly being unable to use the driver's side door controls for half a minute at a time
• Messages on the IC screen saying that it is unable to communicate with the MCU for half a minute
• MCU becoming unresponsive to touches for up to half a minute

with some of these symptoms sometimes occurring simultaneously. At this point, the other communication problems have subsided (I haven't seen them since a week or so after the "upgrade"), but the IC reboots and autopilot disengagement persists.

What's interesting is that when it fails, you get a "Take over immediately", but Autopilot remains engaged, and the IC remains lit. And then as soon as you disengage Autopilot, that's when the IC reboots. It makes me wonder if there is a memory leak that causes the IC computer to hit some low water mark memory-wise, and then reboot itself as soon as it thinks it is "safe" to do so.

Yes , I am also having MCU1+HW3

 

[Workbirdy]

My Tesla has no 3g service today anyone having a problem? Is fine on wifi reset the car but nothing has worked

 

[t3sl4drvr]

My Tesla has no 3g service today anyone having a problem? Is fine on wifi reset the car but nothing has worked

3G Cellular Network Retirement | Tesla Support

AT&T has recently announced that it will discontinue its 3G cellular service by February 2022. Once AT&T’s 3G service is discontinued, Model S vehicles built before June 2015 without a 4G/LTE capable modem will not be able to enjoy some features unless connected to Wi-Fi.

www.tesla.com

 

[dbldwn02]

I'm wondering how Tesla will figure out how to push firmware you don't want, if they don't have a cell service to use.

If you keep your 3g car off wifi, they really couldn't do much, huh?

 

[Workbirdy]

I'm wondering how Tesla will figure out how to push firmware you don't want, if they don't have a cell service to use.

If you keep your 3g car off wifi, they really couldn't do much, huh?

Yeah getting the LTE upgrade on the 15th I'm scared because I have older software loaded that I paid for and hope they don't try a fast one! It's a @ home service so ik be watching them

 

[Solar40]

Nope, even with emmc change, my mcu1 still crashes. Getting worse too. My voice command barely works. Fsd on mcu1 is a joke. No way we getting it. If we were going to get it. Elon would’ve been touting it like he does everything else he got to work. People have been all over Twitter asking him about this. People he answers all the time. Yet he won’t give a answer on the subject or even engage it. So it’s looking like we’re all screwed until we upgrade to mcu2.

 

[dgpcolorado]

^ I have MCU1 without the emmc upgrade (so far, still waiting). It is LTE (4g) and voice commands work fine when the car has a bar or two of cell reception, which is most places nowadays.

 

[dgatwood]

The failure is likely not energy-related, as I initially saw this manifest itself as:

• Randomly being unable to use the driver's side door controls for half a minute at a time
• Messages on the IC screen saying that it is unable to communicate with the MCU for half a minute
• MCU becoming unresponsive to touches for up to half a minute

with some of these symptoms sometimes occurring simultaneously. At this point, the other communication problems have subsided (I haven't seen them since a week or so after the "upgrade"), but the IC reboots and autopilot disengagement persists.

As do the window problems, to be clear.

Here we are in mid-June, with tens of O(6%) of Tesla cars out there still stuck on a dangerously defective firmware released to us in mid-February that also caused some of the same problems on MCU2 when they shipped the early 2021 releases to those cars.

Does Tesla not have a release manager? EngProd team? Release QA? How many different layers of engineering had to completely fail at their jobs to end up with this sort of outcome? Because this really blows my mind. It's a level of incompetence that exceeds anything I've ever seen at any tech company with the sole exception of Facebook.

 

[Workbirdy]

As do the window problems, to be clear.

Here we are in mid-June, with tens of O(6%) of Tesla cars out there still stuck on a dangerously defective firmware released to us in mid-February that also caused some of the same problems on MCU2 when they shipped the early 2021 releases to those cars.

Does Tesla not have a release manager? EngProd team? Release QA? How many different layers of engineering had to completely fail at their jobs to end up with this sort of outcome? Because this really blows my mind. It's a level of incompetence that exceeds anything I've ever seen at any tech company with the sole exception of Facebook.

As do the window problems, to be clear.

Here we are in mid-June, with tens of O(6%) of Tesla cars out there still stuck on a dangerously defective firmware released to us in mid-February that also caused some of the same problems on MCU2 when they shipped the early 2021 releases to those cars.

Does Tesla not have a release manager? EngProd team? Release QA? How many different layers of engineering had to completely fail at their jobs to end up with this sort of outcome? Because this really blows my mind. It's a level of incompetence that exceeds anything I've ever seen at any tech company with the sole exception of Facebook.

 

[Workbirdy]

Stop going to Tesla service center people Tesla told me I need a 20k battery replacement that was not true I took it to a 3rd party Tesla service they downgrade my software and now I charge fast and my battery charge to full with only 5% Degradation. The technician told me tesla adds a file to gimp the older cars range and even when we take it to them they tell us they want 20k for a new battery this is all a lie, I can no longer update my software and this only works on mcu1 computers