[Bug java/35979] New: JNI method NewStringUTF crashes when passed a NULL pointer

[Bug java/35979] New: JNI method NewStringUTF crashes when passed a NULL pointer

[thomas dot g dot girard at free dot fr]

Hello,

when running java-gnome 4 unit tests, I've found out that gij NewStringUTF
implementation crashes when its argument is NULL. Indeed in natString.cc the
very first line of _Jv_NewStringUTF(const char *bytes) calls strlen(bytes),
without checking whether bytes is NULL or not. Hence the crash.

Sun's implementation does not crash: it returns NULL instead.

However it seems gij implementation is correct according to
http://java.sun.com/docs/books/jni/html/functions.html#65644:
  "Return Values. Returns a local reference to a string object, or NULL if the
string cannot be constructed. Returns NULL if and only if an invocation of this
function has thrown an exception."

Should gij implementation be changed to handle NULL? I've checked that the
crash should still occur on the trunk, r134466


-- 
           Summary: JNI method NewStringUTF crashes when passed a NULL
                    pointer
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: java
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: thomas dot g dot girard at free dot fr
 GCC build triplet: x86_64-unknown-linux-gnu
  GCC host triplet: x86_64-unknown-linux-gnu
GCC target triplet: x86_64-unknown-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35979

[Bug libgcj/35979] JNI method NewStringUTF crashes when passed a NULL pointer

[tromey at gcc dot gnu dot org]

------- Comment #1 from tromey at gcc dot gnu dot org  2008-04-19 19:55 -------
>From your comment I assume that java-gnome is using the JNI NewStringUTF call.
This does not directly call _Jv_NewStringUTF.  Instead, it calls
_Jv_JNI_NewStringUTF
(in jni.cc).  That code wraps the call to _Jv_NewStringUTF in a
try...catch.

So, if this is crashing for you, something else is going wrong.
Ordinarily libgcj will turn the SEGV into a NullPointerException, which
should be handled by that try...catch.

If you have a minimal test case, I could take a look.


-- 

tromey at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |tromey at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35979

[Bug libgcj/35979] JNI method NewStringUTF crashes when passed a NULL pointer

[thomas dot g dot girard at free dot fr]

------- Comment #2 from thomas dot g dot girard at free dot fr  2008-04-19 20:51 -------
Created an attachment (id=15497)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=15497&action=view)
testcase for NewStringUTF


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35979

[Bug libgcj/35979] JNI method NewStringUTF crashes when passed a NULL pointer

[thomas dot g dot girard at free dot fr]

------- Comment #3 from thomas dot g dot girard at free dot fr  2008-04-19 20:54 -------
You are absolutely right, thanks for pointing this out. The SIGSEGV gets
converted to an NPE.

But Sun's implementation does not throw an NPE. Please have a look at the
attached testcase. You can launch it using 'make JAVA=/path/to/sun/vm' to use a
different VM.

Outputs: 'Got: null' with Sun's VM, and 'Got: java.lang.NullPointerException'
with gij.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35979

[Bug libgcj/35979] JNI method NewStringUTF throws NPE when passed a NULL pointer

[tromey at gcc dot gnu dot org]

------- Comment #4 from tromey at gcc dot gnu dot org  2008-04-19 21:16 -------
Testing a patch.


-- 

tromey at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |tromey at gcc dot gnu dot
                   |dot org                     |org
             Status|UNCONFIRMED                 |ASSIGNED
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2008-04-19 21:16:34
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35979

[Bug libgcj/35979] JNI method NewStringUTF throws NPE when passed a NULL pointer

[tromey at gcc dot gnu dot org]

------- Comment #5 from tromey at gcc dot gnu dot org  2008-04-19 21:35 -------
Fix checked in.


-- 

tromey at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|---                         |4.4.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35979

[Bug libgcj/35979] JNI method NewStringUTF throws NPE when passed a NULL pointer

[tromey at gcc dot gnu dot org]

------- Comment #6 from tromey at gcc dot gnu dot org  2008-04-19 21:35 -------
Subject: Bug 35979

Author: tromey
Date: Sat Apr 19 21:35:02 2008
New Revision: 134471

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=134471
Log:
        PR libgcj/35979:
        * jni.cc (_Jv_JNI_NewStringUTF): Return NULL if bytes==NULL.

Modified:
    trunk/libjava/ChangeLog
    trunk/libjava/jni.cc


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35979