The NHTSA wouldn't know about this yet if they kept file sizes down. Logging uses up memory but there was enough before the bloat. increased firmware made failures happen in 3-5 years instead of 12-20 because they ran out of unused space for the logging to burn. They never stopped logging even now, they reduced a lot of the unnecessary linux file system logging but it still records everything you do. Logging isn't how they could fix this - the only way is to have more memory to burn whether increasing chip size or decreasing used portion.First, a decade is not enough. NHTSA said 15 years is the minimum for automotive products.
Second, software bloat is not the issue. It's the amount of datalogging Tesla did. Flash wears out due to writes. A few 1GB updates doesn't wear it out- constant background writing which adds up to terrabytes does.
They stopped bloat updating MCU1 when they realized it was going to get very expensive. Bloat is why 2017s wear out in 3 years but 2012s took 6+. The used firmware portion quadrupled in size and that means logging had 25% of it's original planned scratch area to burn. 1.2GB updates aren't causing the chip to fail but running out of memory sooner is. They designed the chip with 250mb firmwares and a lot more scratch. BLOAT reduced scratch and causes failure when the new firmware is overlapping burned scratch areas.
A lot of us have had temporary "fixes" applied by the service center using their "Double update" method where they update the same version twice, bouncing your active partition from the "A" side to "B" side when the scratch area is below 0% left and encroaches on one of the usable firmware areas. As those usable areas grew the time for this to happen was dropped by years. Downgrading to 1.0 would fix an MCU1 for a few more years but they won't.
MCU2 could see the same problem in a few years if they bloat it with enough games. It will take a lot more used space but it has the same underlying issues with memory, logging, and planned scratch areas to burn.
Last edited: