Apple's T2 chip may be behind small number of crashes in iMac Pro, new MacBook Pro
A number of iMac Pro and MacBook Pro owners have reported issues with macOS enduring kernel panics, with initial investigations into the problems suggesting something connected to Apple's T2 security chip is to blame -- but actual service numbers don't point to a hardware problem.
Threads in Apple's support forums discuss issues with the iMac Pro, and more recently the 2018 MacBook Pro. All describe how the Mac crashes, entering a kernel panic and requiring it to be restarted, or in some cases rebooting itself automatically.
DigitalTrends noted the thread on Wednesday, attributing it to the T2 chip in the MacBook Pro. The publication believes that the problem manifested in the iMac Pro as well, and has been carried forward to the new hardware.
A common trait found in the error messages is the mention of Bridge OS. The Bridge OS is an embedded operating system that is used by the T2 chip, which Apple uses to perform a variety of functions instead of making the processor handle the tasks.
The T2 sub-processor manages the image quality of the FaceTime camera, as well as regulating boot safety and boot security, encrypting and decrypting data sent to flash storage, fan speeds, speakers, and secures built-in microphones. It is also used to power "Hey Siri" vocal prompts for Apple's digital assistant.
While the cause of the kernel panics are unclear, with symptoms and hardware setups varying between users, Apple is seemingly aware of the problem and has been working on it since the start of 2018. One user notes the Kernel Version increase from 17.3.0 to 17.4.0 resulted in fewer panics, reducing the count from a daily occurrence to one that happens once or twice a week.
In some cases, Apple has reportedly replaced the hardware, but the issue seemingly continued even on the new unit for a few users, suggesting a software conflict of some sort.
In the posts, Apple support staff suggests iMac Pro users wipe and reinstall MacOS, disable FileVault, and disable Power Nap. For those not willing to perform those, other forum posts suggested crashes could be reduced by not daisy-chaining devices, not using a Thunderbolt 3-to-Thunderbolt 2 adapter, disable Secure Boot, turn off power management options, remove third-party kernel extensions, avoid leaving it to idle overnight, and to not unlock it with an Apple Watch.
Across all the centers we use since the iMac Pro release, there have been 103 iMac Pros coming in for some form of service under warranty out of a pool of well over 100,000 incidents across all of Apple's product lines in total. Of those 103, four of them were for unexplained system crashes linked to the "bridgeOS." Of those four, a fresh macOS reinstall fixed them all.
This is obviously not a large sample of iMac Pro service incidents, from a computer that isn't selling in massive quantities. However, if there were a wide-spread hardware problem, it would be apparent that there was in even this small sample.
Threads in Apple's support forums discuss issues with the iMac Pro, and more recently the 2018 MacBook Pro. All describe how the Mac crashes, entering a kernel panic and requiring it to be restarted, or in some cases rebooting itself automatically.
DigitalTrends noted the thread on Wednesday, attributing it to the T2 chip in the MacBook Pro. The publication believes that the problem manifested in the iMac Pro as well, and has been carried forward to the new hardware.
A common trait found in the error messages is the mention of Bridge OS. The Bridge OS is an embedded operating system that is used by the T2 chip, which Apple uses to perform a variety of functions instead of making the processor handle the tasks.
The T2 sub-processor manages the image quality of the FaceTime camera, as well as regulating boot safety and boot security, encrypting and decrypting data sent to flash storage, fan speeds, speakers, and secures built-in microphones. It is also used to power "Hey Siri" vocal prompts for Apple's digital assistant.
While the cause of the kernel panics are unclear, with symptoms and hardware setups varying between users, Apple is seemingly aware of the problem and has been working on it since the start of 2018. One user notes the Kernel Version increase from 17.3.0 to 17.4.0 resulted in fewer panics, reducing the count from a daily occurrence to one that happens once or twice a week.
In some cases, Apple has reportedly replaced the hardware, but the issue seemingly continued even on the new unit for a few users, suggesting a software conflict of some sort.
In the posts, Apple support staff suggests iMac Pro users wipe and reinstall MacOS, disable FileVault, and disable Power Nap. For those not willing to perform those, other forum posts suggested crashes could be reduced by not daisy-chaining devices, not using a Thunderbolt 3-to-Thunderbolt 2 adapter, disable Secure Boot, turn off power management options, remove third-party kernel extensions, avoid leaving it to idle overnight, and to not unlock it with an Apple Watch.
By the numbers
Using the same service centers that we've used for MacBook Pro keyboard failure rates, as well as for "bendgate" data collation from about two years ago, AppleInsider has collected data on possible T2 issues.Across all the centers we use since the iMac Pro release, there have been 103 iMac Pros coming in for some form of service under warranty out of a pool of well over 100,000 incidents across all of Apple's product lines in total. Of those 103, four of them were for unexplained system crashes linked to the "bridgeOS." Of those four, a fresh macOS reinstall fixed them all.
This is obviously not a large sample of iMac Pro service incidents, from a computer that isn't selling in massive quantities. However, if there were a wide-spread hardware problem, it would be apparent that there was in even this small sample.
Comments
Didn't Scotty say, "the more they overthink the plumbing, the easier it is to stop up the drain"?
Or in other words avoid using functions of the T2 chip as much as possible?
The most heinous thing I’ve ever seen a software developer do is to put in exception handling logic that silently handles the exception and continues on without crashing at the point where the exception occurred. Two minutes, hours, days, or weeks later the whole thing blows up in a seemingly innocuous part of the code and the original cause of the crash is long gone. Crashing is a good thing when it helps capture the root cause of the problem. Any form of crash prevention that leaves the system running in a non deterministic and crippled state is pure evil, unless it’s part of a life, limb, or money damage control scenario.
Then you have little understanding of what went wrong and where. This especially applies to Java where sometimes the sheer depth of the procedure calls is staggering.
I've had beginner software devs look at me in total surprise when I've described some of the EH designs I've used over the years.
As for rolling back the current operation.... Oh! That's too hard.
There is no one panacea for excellent EH design but there a is are millions of really bad ones.
My iMac Pro rebooted more than half the time during iMovie exports. The longer the export (higher res or longer clip), the greater the chance of a crash. I haven't tried it since 10.13.6 arrived.
My i9 MBP spontaneously logs me out under high compute (not memory) load.
When logging into both systems, some apps (most often Terminal and Safari) have trouble reopening. They often launch and crash. On second launch, they're stable. I can't remember a login that wasn't accompanied by at least one app crash.
OT: in my opinion, MBP throttling isn't fixed. I see over-throttling of the CPU (down to ~2 GHz, just 11 watts power consumption and under 70 degrees C, with no GPU involvement) under real conditions (multithreaded jobs using more than ~4 GB memory and accessing memory randomly, not sequentially). I don't see such over-throttling on the iMac Pro. Yes, the iMac Pro has a better cooling system, but -really- why should the MBP be throttled so low when the CPU is >50% utilized?
The difficult thing for support staff is that people come to them expecting that they have all of the answers to their problems. When really, only the people who are directly working on the hardware and software can determine the cause (and solution) for these types of problems. Apple should really tell support staff that if they don't know the solution, and can't find it in their knowledge base, they should tell people that information instead of providing speculation-based answers.
8 core iMac Pro. Vega 56 (vs 64 in most of the cases I saw). 2TB SSD. 1 external monitor. 1 external drive TB drive. 1GB Ethernet connection.
Not a single Kernel Panic in 6+ months. Machine has been like a rock.
Cancer is another example of the difficulty of getting down to the facts. It has taken decades to get to the point where we have associated specific DNA configurations with specific cancers. That only takes care of a few of the multitude of cancers, we are a very long ways from having the facts there.
In a nut shell i become suspiceous of anybody that uses the word 'facts' to prove their point. We certainly have people that have given up on a rational view of the world but that does not make a more rational view of the world all truth. Think about it scietists have been working for decades to resolve the connection between the quantum workd and the classical. We live in a world of theories largely and often struggle to explain the facts.
If you take Auxio's original sentence, he equates people buying into creationism and flat Earth theories as people not willing to accept well research facts. If we take "fact" as being:
"An objective consensus on a fundamental reality that has been agreed upon by a substantial number of people."
I think we can assuming the Earth being round and Evolution are both "An objective consensus on a fundamental reality that has been agreed upon by a substantial number of people."
In an observational world, our theories do an amazing job of explaining our observations (what you seem to be calling "facts"). For example: If I drop a brick from 4 meters high, Newton's basic theory of motion (a = dv/dt = d^2/dt^2) predicts it will take about 0.8 seconds to hit the ground and this fits observation extremely well. Theory even predicts interference patterns in double slit experiments with lights as well as quantum entanglement as demonstrated in (and in variations) of the quantum eraser experiments. Theory and observation match very well. Our theories, based on our observation of things around us, were good enough to get to the moon and send Voyager to other planets in a grand tour of the solar system.
Darwin made an amazing amount of predictions based on evolution such as predicting that precursors to the trilobite would be found in pre-Silurian rocks and he was right. He also predicted a method would be found allowing traits to be passed down from parents to offspring and he was right (as confirmed by Watson and Crick in 1953).
While it is true we don't have a single "Theory of Everything" explaining 100.000000% of everything we observe, saying we live in a world where "theories largely and often struggle to explain [observations]" is short sighted and poorly represents what we really know. There is lots left to explain and understand but we have an amazing set of theories today capable of explaining a vast array of data we observe every day.
I was glad I had just read this article because it made me think that the newly made Daisy chain connection may be the trigger. After the reboot from the kernel panic, I unplugged the new RAID and plugged it directly into the iMac. I was able to initialize the drive with no problems. I think this issue is a flaw in the latest builds of MacOS and that the new computers are seeing this more often because when people get new gear they have lots of stuff to set up including redoing their daisy chained drives or adding new ones.
Just look at the movie 2001. How much of the special effects ended up being true to reality when we finally could launch spacecrafts.
The point he and Scotty were making is not to make something over complicated if you don't need to. This is something that is happening in real life right now in the tech world.