consistent libjvm SIGSEGV crashes with older or recent versions of CLion
Had been using CLion 2018.3.4 without issues on RedHat 7.6 previously on an Intel server and workstation without issue.
Now using CentOS 7.9, on a workstation w/ AMD Ryzen Threadripper 3960X processor, 64 GB, 2x RTX 3090s. When I load CLion with the bundled JRE, I get crashes like these:
# SIGSEGV (0xb) at pc=0x00007f4710c457fa, pid=55470, tid=0x00007f470d976700
#
# JRE version: OpenJDK Runtime Environment (8.0_152-b26) (build 1.8.0_152-release-1343-b26)
# Java VM: OpenJDK 64-Bit Server VM (25.152-b26 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# V [libjvm.so+0x8fa7fa] ParCompactionManager::follow_marking_stacks()+0x1fa
# SIGSEGV (0xb) at pc=0x00007f61499e5b20, pid=55998, tid=0x00007f6133da7700
#
# JRE version: OpenJDK Runtime Environment (8.0_152-b26) (build 1.8.0_152-release-1343-b26)
# Java VM: OpenJDK 64-Bit Server VM (25.152-b26 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# V [libjvm.so+0x904b20] ParallelCompactData::calc_new_pointer(HeapWord*)+0x30
# SIGSEGV (0xb) at pc=0x00007fee7bd1ea1a, pid=56348, tid=0x00007fee64efe700
#
# JRE version: OpenJDK Runtime Environment (8.0_152-b26) (build 1.8.0_152-release-1343-b26)
# Java VM: OpenJDK 64-Bit Server VM (25.152-b26 mixed mode linux-amd64 compressed oops)
# Problematic frame:
# V [libjvm.so+0x90ba1a] oopDesc* PSPromotionManager::copy_to_survivor_space<false>(oopDesc*)+0x11a
I have also tried the latest 2020.3.1 version, and I also get similar crashes. I have tried the following:
1) Changing JDK to 11, 13, even the latest 15 version -- all of them crash in different ways, the newer ones seem to crash in routines related to java's garbage collection
2) Trying the latest version and also changing JDKs, and also see the crashes.
The crashes seem to occur most often at the point where loading symbols is done, but not always, sometimes it even crashes before that is done. Usually doesn't crash if it doesn't pull in a CMake project file to run, which obviously makes the IDE useless.
I've spent the better part of the day trying to figure out what the issue is and have not been able to solve this.
I've attached the crash dump logs (these are on 2018.3.4 with the bundled JRE/JDK). Any ideas?
Upload id: 2021_01_13_AEhhxSosZNL3Jso7 (files: java_error_in_CLION_55470.log and 2 more)
请先登录再写评论。
Hello!
I kindly ask you to create an issue in the JetBrains Runtime tracker - https://youtrack.jetbrains.com/issues/JBR. My colleagues will assist you.
TL;DR For anyone that has repeatable crashes in different functions, check to make sure that your memory modules are not triggering memory errors with memtester / memtest86 / etc.
Anna Falevskaya Thanks for the suggestion.
It turns out that in this case, this workstation had a bad DDR4 DIMM, which explains the random crashes in different functions in the crashes noted above. I happened to verify this first with a userland memory checker (memtester) and later with a memtest86 run outside of the O/S. After the bad memory was replaced, the problem went away as expected.
Another issue that was being experienced which also was related to the bad memory was that building a large C++ project (gcc) outside of CLion, especially if using multiple jobs with make, would sometimes cause the compiler to crash, but it would later resume compilation with no issues, which now makes sense with the bad memory causing bitflips randomly.