search

eclipse-openj9 / openj9

Eclipse OpenJ9: A Java Virtual Machine for OpenJDK that's optimized for small footprint, fast start-up, and high throughput. Builds on Eclipse OMR (https://github.com/eclipse/omr) and combines with the Extensions for OpenJDK for OpenJ9 repo.
Other
3.28k stars 721 forks source link

SIGFPE not consumed by the JVM when trap is enabled #2887

Open enasser opened 6 years ago

enasser commented 6 years ago

As per the JVM specification https://docs.oracle.com/javase/specs/jvms/se10/html/jvms-2.html#jvms-2.8.1, SIGFPE should be consumed by the JVM even if trap is enabled. Hotspot handles SIGFPE and provides Infinity for a divide-by-zero where as OpenJ9 crash.

The key differences between the floating-point arithmetic supported by the Java Virtual Machine and the IEEE 754 standard are:

The floating-point operations of the Java Virtual Machine do not throw exceptions, trap, or otherwise signal the IEEE 754 exceptional conditions of invalid operation, division by zero, overflow, underflow, or inexact. The Java Virtual Machine has no signaling NaN value.

Please find below the test case to recreate the issue: test_case.zip

Recreate step:

With Hotspot:

$ export JAVA_HOME=/home/enasser/openjdk/builds/hs/jdk-10.0.2+13/
$ export LD_LIBRARY_PATH=$JAVA_HOME/lib/server
$ c++ -Wl,--no-as-needed -ldl -I$JAVA_HOME/include -I$JAVA_HOME/include/linux -L$LD_LIBRARY_PATH -o invoke invoke.cpp -ljvm -lpthread
invoke.cpp: In function ‘int main()’:
invoke.cpp:27:30: warning: deprecated conversion from string constant to ‘char*’ [-Wwrite-strings]
     options[0].optionString = "-Djava.class.path=.";
                             ^
$ $JAVA_HOME/bin/javac Prog.java
$ ./invoke
Inverse of 0.0 is Infinity

With J9:

$ export JAVA_HOME=/home/enasser/openjdk/builds/j9/jdk-10.0.2+13/
$ export LD_LIBRARY_PATH=$JAVA_HOME/lib/compressedrefs
$ c++ -Wl,--no-as-needed -ldl -I$JAVA_HOME/include -I$JAVA_HOME/include/linux -L$LD_LIBRARY_PATH -o invoke invoke.cpp -ljvm
invoke.cpp: In function ‘int main()’:
invoke.cpp:27:30: warning: deprecated conversion from string constant to ‘char*’ [-Wwrite-strings]
     options[0].optionString = "-Djava.class.path=.";
                             ^
$ ./invoke
Unhandled exception
Type=Floating point error vmState=0x00000000
J9Generic_Signal_Number=00040020 Signal_Number=00000008 Error_Value=00000000 Signal_Code=00000003
Handler1=00007F292C408470 Handler2=00007F292BD0DC40
RDI=000000000230C7B0 RSI=000000000230C7A0 RAX=000000000230C7A0 RBX=000000000230C7B0
RCX=0000000000000000 RDX=000000000230C7B0 R8=0000000000000000 R9=0000000002261FE8
R10=0000000002262100 R11=0000000000000000 R12=0000000002261FE8 R13=000000000230C7D0
R14=000000000221EA00 R15=00000000023AAE58
RIP=00007F292C446140 GS=0000 FS=0000 RSP=00007FFD6F5491F8
EFlags=0000000000010202 CS=0033 RBP=00007F28FE232688 ERR=0000000000000000
TRAPNO=0000000000000013 OLDMASK=0000000000000000 CR2=0000000000000000
xmm0 3ff0000000000000 (f: 0.000000, d: 1.000000e+00)
xmm1 0000000000000000 (f: 0.000000, d: 0.000000e+00)
xmm2 0000000000000000 (f: 0.000000, d: 0.000000e+00)
xmm3 0000000000000000 (f: 0.000000, d: 0.000000e+00)
xmm4 0000000000000000 (f: 0.000000, d: 0.000000e+00)
xmm5 402f2491c1246069 (f: 3240386560.000000, d: 1.557142e+01)
xmm6 0000000000000000 (f: 0.000000, d: 0.000000e+00)
xmm7 0000000000000000 (f: 0.000000, d: 0.000000e+00)
xmm8 3bbcc86800000000 (f: 0.000000, d: 6.095003e-21)
xmm9 3fd49ec4cd62d92f (f: 3445807360.000000, d: 3.221905e-01)
xmm10 3d754b1c5ea09e82 (f: 1587584640.000000, d: 1.210390e-12)
xmm11 402f2492450323c9 (f: 1157833728.000000, d: 1.557143e+01)
xmm12 3cce000000000000 (f: 0.000000, d: 8.326673e-16)
xmm13 bfd49ec4cd62d920 (f: 3445807360.000000, d: -3.221905e-01)
xmm14 402e7f9c1e980d00 (f: 513281280.000000, d: 1.524924e+01)
xmm15 bc69edf0f0de144d (f: 4041085952.000000, d: -1.124511e-17)
Module=/home/enasser/openjdk/builds/j9/jdk-10.0.2+13//lib/compressedrefs/libj9vm29.so
Module_base_address=00007F292C36F000
Target=2_90_20180813_102 (Linux 4.15.0-33-generic)
CPU=amd64 (8 logical CPUs) (0x3e5c88000 RAM)
----------- Stack Backtrace -----------
(0x00007F292C446140 [libj9vm29.so+0xd7140])
(0x00007F292C446161 [libj9vm29.so+0xd7161])
(0x00007F292C3982C1 [libj9vm29.so+0x292c1])
(0x00007F292C387DCB [libj9vm29.so+0x18dcb])
(0x00007F292C4434C2 [libj9vm29.so+0xd44c2])
JVMDUMP039I Processing dump event "gpf", detail "" at 2018/09/16 14:28:59 - please wait.
JVMDUMP032I JVM requested System dump using '/home/enasser/testpgm/jni/core.20180916.142859.23107.0001.dmp' in response to an event
JVMPORT030W /proc/sys/kernel/core_pattern setting "|/usr/share/apport/apport %p %s %c %d %P" specifies that the core dump is to be piped to an external program.  Attempting to rename either core or core.23131.
JVMDUMP010I System dump written to /home/enasser/testpgm/jni/core.20180916.142859.23107.0001.dmp
JVMDUMP032I JVM requested Java dump using '/home/enasser/testpgm/jni/javacore.20180916.142859.23107.0002.txt' in response to an event
JVMDUMP010I Java dump written to /home/enasser/testpgm/jni/javacore.20180916.142859.23107.0002.txt
JVMDUMP032I JVM requested Snap dump using '/home/enasser/testpgm/jni/Snap.20180916.142859.23107.0003.trc' in response to an event
JVMDUMP010I Snap dump written to /home/enasser/testpgm/jni/Snap.20180916.142859.23107.0003.trc
JVMDUMP007I JVM Requesting JIT dump using '/home/enasser/testpgm/jni/jitdump.20180916.142859.23107.0004.dmp'
JVMDUMP013I Processed dump event "gpf", detail "".
JasonFengJ9 commented 6 years ago

Additional experiment results as per @DanHeidinga suggestion:

A JNI native (not customer launcher) invoking feenableexcept() before Java performing divided-by-zero produces following result against RI JDK8 (and JDK11):

Exception in thread "main" java.lang.ArithmeticException: / by zero
    at ProgV2.main(ProgV2.java:11)

It looks like RI signal handlers catch the error and map it to an exception.

Java code snippets:

   public static native void setfp();
   static {
        System.loadLibrary("ProgV2");
   }
   public static void main(String[] args) {
      setfp();
      double input = 00.0000;
      System.out.println("Inverse of "+ input + " is " + 1/input);
   }

Native code snippets:

JNIEXPORT void JNICALL Java_ProgV2_setfp(JNIEnv *env, jclass thisClz) {
    feenableexcept(FE_INVALID | FE_DIVBYZERO | FE_OVERFLOW  | FE_UNDERFLOW);
    return;
}
enasser commented 6 years ago

@JasonFengJ9 That is an interesting observation. So with RI JDK, there is difference in behavior when the SIGFPE is enabled before and after create JVM. If the trap is enabled before create JVM, it just catch the exception and continue where as if the trap is enabled after create JVM, then it result in the ArithmeticException.

JasonFengJ9 commented 6 years ago

@enasser that's correct.