Hacker News .hnnew | past | comments | ask | show | jobs | submitlogin
Jetbrains products all crash on the 2020 MacBook Air, sometimes crashing the OS (jetbrains.com)
68 points by api on May 12, 2020 | hide | past | favorite | 21 comments


(Bug report mentions PHPstorm but it manifests on CLion, GoLand, and others.)

This is clearly an Apple bug as well as possibly a bug in Jetbrains' UI or other native libraries or maybe upstream in the JVM. It's an Apple bug because it can bring down the whole system, sometimes with an absolute hardest possible crash that generates no debug logs! Sometimes it also just crashes the JVM. Sometimes it's a softer crash with logs and sometimes a hard crash with no logs. Logs show crashes happening at random different places in the code.

Absolutely nothing else does this, including heavy stuff like running two Parallels VMs at once (Windows 10 and Linux) and running heavy multithreaded compiles. Also ran a RAM and CPU burn-in test. It's not a hardware problem. Many others are reporting it too.

Given that the new 13" Pro is a slightly larger version of what seems to be the same machine (10th gen Intel core, etc.) I would speculate that it will probably show the same weird problem. (The 2020 Air is a "2020 13" Pro, No Touch Bar Edition.")

People using Jetbrains products on earlier or different MacBook models have no issues. This includes the 2019 Air.

Posting here to attempt to get Apple's attention. Drawing attention on social media is the new way to get tech support focused on an issue -- "blue checkmark service" as it's called on Twitter.

Edit: I bet there is a security vulnerability here given that it brings down the whole OS. Any security researchers in the audience? The 2020 Air has a new 10th generation Intel chip and a new version of Intel integrated GPU graphics, so my money is on drivers or kernel support for one of those. (Runs off to disable WebGL and WebGPU...)


Or possibly an Intel/hardware problem.


It'd be an extremely esoteric one if these products crash it and literally nothing else does including:

- Hours of CPU and RAM burn

- Multiple concurrent VMs running different OSes (Windows 10 and Linux).

- "make -j8" on a large C++ code base

- Visual Studio Code and other large Electron apps that also run under a JIT.

- I ran a 3D graphics demo to stress test the GPU, no issues.

Zero crashes of any kind, then I launch a Jetbrains app and boom at a frequency of about once per hour of app use. Sometimes the app just crashes, sometimes the whole machine crashes and does so very "hard."

The simplest explanation is an Apple bug being triggered by a Jetbrains runtime or JVM bug, or if not an outright bug something really odd and out of the ordinary that this software is doing via some OS API. Given that it only happens on this machine my bet would be on something video or screen related as the GPU differs slightly from other Mac laptops and that's a common place for bugs to hide.


Recent Mac GPUs have been... weird. I've had small artifacts on three different 16-inch MBPs (one that I returned, a replacement that I decided to keep, and one that I was issued at work).


GPU seems like a reasonable path to investigate. Would testing with an external GPU be able to validate this theory?


> Absolutely nothing else does this

That's not true. It's pretty easy to crash Finder, and when Finder crashes, the whole system goes with it. For example, Time Machine has a tendency to crash my devices. It takes down finder, and from there, it's impossible to do anything.

Heavy resource usage isn't usually what causes these kinds of crashes on macOS, from what I've seen (unless there's an underlying hardware issue).

But yes, I'd agree this one is probably on Apple.

> Drawing attention on social media is the new way to get tech support focused on an issue -- "blue checkmark service" as it's called on Twitter.

You're right, sadly.

> Edit: I bet there is a security vulnerability here given that it brings down the whole OS. Any security researchers in the audience?

Possibly, but crashes are a dime a dozen. Actually getting something useful out of that information can be difficult. Hopefully this will be fixed long before anyone is able to get that far.


By crash I don't mean can't use the machine. I mean hard total lock requiring holding down the power button and no dump is even generated.


Interestingly, one commenter posted that switching to OpenJDK fixes it. So it's some interaction between the hardware, the JRE, and JetBrains. I look forward to the post-mortem for this one.


Read more. It doesn't, though it may be less frequent under OpenJDK (hard to tell as its random). Disabling newer GCs, SSE, and AVX has no effect either.

My favorite is that nothing else triggers it. Just this app. So, so bizarre. I am wondering about some mega obscure Intel errata, some specific pattern of highly improbable opcodes combined with some pattern of multithreaded memory operations...

My money is still on video though.

I too await the postmortem.


My old Mac started crashing fairly regularly a couple months ago. Which are “the logs” one would look at to investigate this? I haven’t been able to get anyone to tell me anything specific.


Did you try some good SSD burn in test? Last time I observed all kinds of weird crashes and behavior on my wife’s old Air, it was due to slowly dying SSD...


I've seen that too. I highly doubt it in this case. It's 100% specific to this app and I have done a lot of other heavy things.


agree


Not real impressed with Apple Mac OS stability lately. I am currently experiencing the WindowServer crash coming out of sleep on my MacBook Pro 16.

Look like it has something to do with the AMD dGPU and Catalina so again probably a GPU driver: https://mrmacintosh.com/10-15-4-update-wake-from-sleep-kerne...

Then you have another fun one probably related like it spinning up the fans and the GPU consuming double power when you run the built in display and an external monitor, but not with just external monitors: https://forums.macrumors.com/threads/16-is-hot-noisy-with-an...

Of course the problem with Apple is zero communication even acknowledging an issue leads to all sorts of wild goose chases usually involving reset pram or whatever.


Fascinating. This could form the basis of a serious zero-day.


I wonder if this affects Android Studio.


the thing that sucks about jetbrains is that you need to keep updating license keys on your own end every year, even though you're paying a yearly fee. they really need to make this much easier. the end user shouldn't have to do anything.


No? Just login to your JetBrains Account and it should just work in my experience? No need to fiddle around with license keys.


nope, it just says "no suitable licenses found" even though i got the email that i updated.


Something is wrong with your account. You should have support look into it.

I just login into my account on the IDE and all licenses are automatically imported.


Can't you login with your Jetbrains account (see Help => Register => Get licence from: JB Account)?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: