Using all-Apple Thunderbolt hardware to kill my MacBook | Hacker News Reader

Using all-Apple Thunderbolt hardware to kill my MacBook

rachelbythebay.com

145 pointsLaSombra46 comments

I have a 2018 Mac Mini and a Dell 4K monitor (U2720Q).

Unbelievably, if the monitor is connected over USB-C, the FileVault unlock screen will not render at 4K60, which means that if you enable FileVault, and then reboot, you just stare at a "no signal" screen, forever.

I had to switch to HDMI just because of this bug which has apparently existed since the 2018 Mac Mini was released.

I have the MacBook Pro 16" 2019. I can confirm you don't have to try very hard for this to break. Disconnect the thunderbolt connected display when asleep and boom. I've learnt that if you close the lid then this seems to help avoid the issue.

The AMD GPUs power consumption shoots up from 5 to 20+ watts when you connect an external monitor. Closing the laptop's lid is the only reliable workaround, but even then you only have to fire up Photoshop to make the device's fans go wild.

Zero confirmation of this widespread issue from Apple or AMD.

€6000+ machine

No more Apple after this for me

Thunderbolt/USB-C definitely has some non-intuitive behavior. I accidentally fried a dock and one of the USB-C ports on my macbook plugging in the power supply while it was docked (the OS said it wasn't charging so I thought it just disconnected PD from the dock, apparently I was wrong.)

I had a similar problem with a 13 inch MacBook pro in 2018, after they had swapped out the motherboard due to battery swelling under warranty. The computer kept restarting itself every single day for a year after hibernating. The solution was to change the auto sleep settings in terminal. Yes I was using a Thunderbolt display as well, but didn’t think it was related.

I got that worked-around (not resolved but also not still happening) by switching off graphics card energy saving (switching between external to onboard graphics).

Only rebooted after sleep if I was using a display through a thunderbolt dock and had graphics switching enabled.

I use a Thunderbolt Display through a TB2 -> 3 adaptor too and it’s an unreliable piece of crap.

I haven’t had any panics from that (although plenty of unrelated panics), but I have to unplug it before I close the laptop, else it’s very unlikely to sleep. Despite not having any external peripherals connected, it always goes into clamshell mode rather than sleep. It then appears to sleep (i.e. the display goes off), but then I come back to my desk the next morning to find the laptop uncomfortably warm with the fans running, while closed and doing literally nothing.

The more I spend on MacBooks the more issues they seem to have.

At least a few years ago the UltraFine 5K monitor on display in the Apple Store connected to the 2018 Mac Mini exhibited some odd behaviour. Not sure if it was Thunderbolt but it was definitely USB-C, but I believe it was changing the display scaling that would result in no display for quite a while.

I had this problem, downgraded to Mojave and it went away. Saw recently someone say they talked to CalDigit support and they said to clear the NVRAM with sudo nvram -c

I upgraded to Catalina again, ran this to be certain, and it worked! No more crashing in a whole month. Previous reinstalls of Catalina did not fix it.

My theory is the NVRAM seems to hold some values that look like hints of hardware configuration (I forget the label this long string is under) and if this string gets anything incorrect it panics on wake or even on boot because it doesn't see all the hardware supposedly plugged in and can't speed up the boot/wake using these hints.

This makes sense why it would panic if you unplug while it's sleeping and then it crashes on wake.

I think the NVRAM string that gets corrupted is called HW_BOOT_DATA

I have my own USB-C crappy story. I have a 2020 MacBook Pro 16” with bootcamp installed. Recently I bought a USB-C hub, if I try to boot into windows (hold alt at Apple logo) the machine literally cannot boot. At all. I have to disconnect the hub in order to make any progress.

I've got dual monitors on a Mac Mini. The primary monitor uses the HDMI port, and it works fine. The secondary monitor runs off Thunderbolt and a couple of weeks ago it started going black, randomly. This isn't a hardware issue, because I have used multiple cables and monitors, and I have used both Thunderbolt-to-HDMI and Thunderbolt-to-VGA adapters and I get the same issue. Oddly, the black screen still shows the mouse pointer, and if I take a screen shot it shows what the monitor should be displaying.

The fact that it just started happening leads me to believe that it came with an operating system update, and it's probably a driver issue.

Apple really needs to get its act together on USB.

I too experience kernel panics from connecting my 2017 MBP to a Thunderbolt display through an Apple Thunderbolt adapter. Apparently this issue is more common than I thought.

This sounds like possibly a variant of this ridiculous issue.

https://mrmacintosh.com/10-15-4-update-wake-from-sleep-kerne...

I've set my Mac to never sleep when plugged in. Disappointing.

My Apple first-party USBC hub has bizarre ordering issues with my Macbook (have to plug it in, in the order HDMI, pass-through USBc charging, then the USB-A peripheral), and the hub doens't work with my Chromebook. A third-party hub does work, but my Macbook seems to recognize each different hub configuration as a new device, so every time I disconnect a USB-A peripheral, the screen flickers as it reestablishes a connection. Inevitably, my internal sound card is no longer recognized. If I disconnect the pass-through USB-C power cable, the screen goes black and takes 1-2 minutes to reload the login screen. Performing an NVRAM reset seems to fix most of these issues, but I now have to do it every few weeks.

The Apple-branded USB-C ecosystem is a giant mess right now.

https://developer.apple.com/bug-reporting/

If you have steps to repro, this is likely the best option for having it fixed

Disagree. I've filed bugs with trivial repro steps and have never gotten any feedback of any kind.

Even just closing them as "wontfix" would be kind.

Apple's penchant for silence or not responding shows in bug reports too. Most of the time there are no responses or changes in the status to any issues filed. Apple uses an internal tracking system for itself and an external system for anyone outside Apple, and hardly any information flows back from the internal system.

Here is the specific issue that was breaking our CI system[0]. I don't have XCode set up anywhere to see if this is still broken.

[0]: https://gist.github.com/ryjones/3e9a851e92794d67b5e37197608a...

I am fairly sure this heuristic was implemented on purpose.

Agreed; however, it ignores return value of the build step.

This logs as a warning, right? “Step produced errors even though tool exited successfully?”

no, the build step was info, not warn. It is detected as an error, and stops CI

I often did get replies, but usually they were "We've fixed it now. Simply upgrade to the new version of macOS." Of course, that wasn't necessarily true. Once, I got a whole correspondence over non-working WiFi (which they didn't resolve, but which I eventually traced to, indeed, a Trust USB-C splitter).

However, if you file bugs for Logic Pro, they get addressed rather quickly. They even take suggestions. I guess it depends on which team gets the bug.

My recent favorite mac kernel panic starts to happen when I connect my samsung t5 ssd to create a time machine backup. After some time macos hangs and then panics. I needed like 10 kernel panics to make it through the whole backup. Even better, I took my months of frustration to find out that my ssd was causing those panics. I mean who would suspect your backup device to be the root cause of kernel panics? It was connected to my screen so I completely forgot about it‘s existence.

Samsung does issue firmware upgrades for these drives, perhaps they fix this.

I was just listening to some downloaded podcast backlog, and one of the episodes was Accidental Tech Podcast where they spent a while discussing the merits of a desktop versus a docked laptop.

And the main thing they complained about wasn’t performance, it was how much MacBooks like to crash in this arrangement. Get an iMac and an old laptop instead for when you need to be mobile, at least those will work.

It’s been what, four years now since thunderbolt only MBP, and still not fixed.

I don’t have any thunderbolt accessories (just a DisplayPort screen) and I used to get “sleep wake failures” at least once a week when it was new. That one at least has improved.

This is a situation that I expect to get much better under the Apple silicon Macs. Under the current situation, Apple is writing drivers for hardware they did not design, and switching between discrete and integrated GPUs has been a flaky issue going all the way back to G4 PowerBooks. A lot of this complexity is going to just evaporate.

Incidentally, I was also using a Thunderbolt Display with a 2016 usb-c MacBook for about three years, it was similarly flakey. Switched to an LG 5k (found an inexpensive refurb) a while back and 95% of the problems went away.

> Get an iMac

The problem is you end up paying for a crappy monitor. So you are stuck either looking at a Mac Mini, or going all the way up to a Mac Pro. There is no middle ground.

There’s the whole X86 arena of middle ground.

Most people don’t need a Mac.

Well yeah, but for context the ATP podcast is hosted by Mac/iOS developers, so he was looking at his Mac options specifically

I love the iMac 27 monitor. Curious what is not to like? I prefer it over the LG Ultrafine.

>The problem is you end up paying for a crappy monitor

Huh? It's one of the best monitors out there...

If anything the main problem with the imac is that you're forced to pay for a very good monitor even if that's not what you'd like to spend your money on.

In Windows/Linux world we have been craving for a 5k/27" monitor for years so that we don't have to deal with non-integer scaling.

In Windows a restart makes everything work properly. But it is a pain, specially in work setups where you return to your desk from a meeting and you want to plug-in your laptop to the Thunderbolt dock and then immediately start working.

Now I have come to terms. After a meeting I save everything and shutdown my machine. I connect to the dock and then boot.

Something I don't understand:

Surfaces have a magnetic charging port that doubles as a dock and it... just works?

USB-C hubs work better on Apple’s devices than Thunderbolt if you want an apples to apples comparison, but they can’t do as much. Surface just doesn’t offer Thunderbolt.

First post about this was from Sep 14th. This has been going on for a while.

https://rachelbythebay.com/w/2020/09/14/crash/

How well does Linux work with Thunderbolt 3 docks?

Chris from Linux Unplugged says OWC docks work well in Linux.

I have 2 Lenovo X1's of different generations running Linux, the Lenovo thunderbolt dock and another USB C dock, and there are problems for every configuration of devices. I get dropped keystrokes, flakey network connections, flakey audio, and monitor flickering. I don't know if the problem is Linux, Lenovo, or both docks. I'm not even sure how to start debugging this problem.

I used to have one of those Apple thunderbolt displays and without getting into all the same technical stuff I can say it is likely that it did the same thing. I tried it on at least five computers and always had the same problem - machine would go to sleep and kernel panic on wake. Took display to shop - “nothing wrong with it”.

The title is misleading: "kill my MacBook" vs "crash and reboot"

Try running the ‘killall’ command, you’ll either be disappointed or relieved.

That's not a good comparison. 'kill' has that meaning in the context of OS processes, while the title talks about MacBook as being a hardware product.

Killing hardware has a very different meaning generally.

I had an issue with a 2016 Mac Book Pro, where a USB-C to HDMI, USB-A, SD card reader, etc would cause random kernel panics when it was plugged in. I don't recall the brand but I think it was on Kickstarter.

Sounds like it was just a crash/kernel panic rather than actual damage to the hardware?

Reply on news.ycombinator.com