Java C++ 中的 JNI 启动器：在没有 JavaFX 的情况下工作，使用 JavaFX 的核心转储

Question

我写了一个 C++ JNI Java 启动器。如果我启动一个不使用 JavaFX 的 java 程序，它会工作，但如果我尝试用它启动一个 JavaFX 程序，它会创建一个核心转储。这是代码：

Without JavaFX - launch.cpp

With JavaFX - launch.ccp

仅当我使用本机启动器时才会发生核心转储。如果我运行具有相同参数的 JavaFX 程序在命令行中使用 java，则没有问题。它运行符合预期。

Here's the content of the hs_err file, in a pastebin.

我没有从少数文件中粘贴代码，而是创建了一个包含两个分支的存储库，其中包含完整的运行可用示例。

Full Example - No JavaFX

Full Example - With JavaFX

您可以运行通过编辑 build.sh 和 run-native.sh 并更改行 jdk="/usr/lib/jvm/java-11-oracle" 以使其对您的系统准确。那么：

./build.sh          #compiles the java program and cpp program
./run-native.sh     #Sets LD_LIBRARY_PATH and runs the compiled cpp program
   or
./run-with-java.sh  #only on "withjfx" branch, runs via java at cli.

如您所见，No JavaFX 版本运行可以，但 JavaFX 版本仅运行通过调用 java 可执行文件，如果通过我的本机启动器执行，它会进行核心转储。

最后，这是我尝试使用本机启动器运行 JavaFX 版本时得到的转储：

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGSEGV (0xb) at pc=0x00007f77513575d8, pid=15281, tid=15301
#
# JRE version: Java(TM) SE Runtime Environment (11.0.1+13) (build 11.0.1+13-LTS)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (11.0.1+13-LTS, mixed mode, tiered, compressed oops, g1 gc, linux-amd64)
# Problematic frame:
# j  java.util.Arrays$ArrayList.<init>([Ljava/lang/Object;)V+6 java.base@11.0.1
#
# Core dump will be written. Default location: Core dumps may be processed with "/usr/share/apport/apport %p %s %c %d %P" (or dumping to /home/joshua/work/javalaunch/core.15281)
#
# An error report file with more information is saved as:
# /home/joshua/work/javalaunch/hs_err_pid15281.log
Compiled method (c2)     326  208       4       java.util.Objects::requireNonNull (14 bytes)
 total in heap  [0x00007f7758e10d90,0x00007f7758e10fc8] = 568
 relocation     [0x00007f7758e10f08,0x00007f7758e10f18] = 16
 main code      [0x00007f7758e10f20,0x00007f7758e10f60] = 64
 stub code      [0x00007f7758e10f60,0x00007f7758e10f78] = 24
 metadata       [0x00007f7758e10f78,0x00007f7758e10f80] = 8
 scopes data    [0x00007f7758e10f80,0x00007f7758e10f90] = 16
 scopes pcs     [0x00007f7758e10f90,0x00007f7758e10fc0] = 48
 dependencies   [0x00007f7758e10fc0,0x00007f7758e10fc8] = 8
Could not load hsdis-amd64.so; library not loadable; PrintAssembly is disabled
#
# If you would like to submit a bug report, please visit:
#   http://bugreport.java.com/bugreport/crash.jsp
#
./run-native.sh: line 6: 15281 Aborted                 (core dumped) ./bin/launch

根据此处用户的推荐，基于之前的崩溃，我在我的系统上安装了 hsdis-amd64.so 并指向 LD_LIBRARY_PATH。我仍然得到一个核心转储，看起来它所做的只是让核心转储加载反汇编程序：

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGSEGV (0xb) at pc=0x00007f9e6e2015d8, pid=17530, tid=17550
#
# JRE version: Java(TM) SE Runtime Environment (11.0.1+13) (build 11.0.1+13-LTS)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (11.0.1+13-LTS, mixed mode, tiered, compressed oops, g1 gc, linux-amd64)
# Problematic frame:
# j  java.util.Arrays$ArrayList.<init>([Ljava/lang/Object;)V+6 java.base@11.0.1
#
# Core dump will be written. Default location: Core dumps may be processed with "/usr/share/apport/apport %p %s %c %d %P" (or dumping to /home/joshua/work/so-question/core.17530)
#
# An error report file with more information is saved as:
# /home/joshua/work/so-question/hs_err_pid17530.log
Compiled method (c2)     372  187       4       java.util.Objects::requireNonNull (14 bytes)
 total in heap  [0x00007f9e75cbb890,0x00007f9e75cbbac8] = 568
 relocation     [0x00007f9e75cbba08,0x00007f9e75cbba18] = 16
 main code      [0x00007f9e75cbba20,0x00007f9e75cbba60] = 64
 stub code      [0x00007f9e75cbba60,0x00007f9e75cbba78] = 24
 metadata       [0x00007f9e75cbba78,0x00007f9e75cbba80] = 8
 scopes data    [0x00007f9e75cbba80,0x00007f9e75cbba90] = 16
 scopes pcs     [0x00007f9e75cbba90,0x00007f9e75cbbac0] = 48
 dependencies   [0x00007f9e75cbbac0,0x00007f9e75cbbac8] = 8
Loaded disassembler from hsdis-amd64.so
#
# If you would like to submit a bug report, please visit:
#   http://bugreport.java.com/bugreport/crash.jsp
#
./run-native.sh: line 6: 17530 Aborted                 (core dumped) ./bin/launch

更新：我在 Ubuntu 18.04 和 Windows 7 上重复了这个问题。

Answer 1

我明白了。问题出在第 35 行，参数传递给 JVM 的主要方法：

env->CallStaticVoidMethod( cls, main, " ");

我们正在将 c 字符串传递给 Java，其中 java 期望 java.lang.String[]。一旦有人试图读取该参数，JVM 就会崩溃。 JavaFX 必须在 main 和 start() 之间的某个时刻读取参数，并因此崩溃。

正确的行是

env->CallStaticVoidMethod( cls, main, "(I)V");（我没有测试过这个，但这是 java 人员提供的示例代码中的公共行）

或：

jclass classString = env->FindClass("java/lang/String");
jobjectArray argsToJava = env->NewObjectArray(0, classString, NULL);
env->CallStaticVoidMethod( cls, main, argsToJava );

Java C++ 中的 JNI 启动器：在没有 JavaFX 的情况下工作，使用 JavaFX 的核心转储

Java JNI Launcher in C++: Works without JavaFX, Core Dump with JavaFX

java

java-native-interface

javafx

java-11

javafx-11