GCJ problem

Wed Aug 15 14:36:00 GMT 2001

On 2001年8月15日, Antonio Ake wrote:
> (I had no idea when you talked about the nice "stack backtrace", beacuse
> in Alpha is not shown, why?)

Alpha backtraces aren't possible without debug info, unfortunately.
I find even gdb sometimes cannot print a backtrace. (I suspect .eh_frame
has enough information to support backtraces, but we cannot use that yet.)
> java::lang::ArrayIndexOutOfBoundsException::ArrayIndexOutOfBoundsException(java::lang::String*)
> (/home/min/g/ake/ss/build/i686-pc-linux-gnu/libjava/../../../gcc-3.0.1-20010813/libjava/java/lang/ArrayIndexOutOfBoundsException.java:36)
> at 0x0809f138: _Jv_ThrowBadArrayIndex
> (/home/min/g/ake/ss/build/i686-pc-linux-gnu/libjava/../../../gcc-3.0.1-20010813/libjava/gcj/cni.h:114)
> at 0x080cc9ac: java::util::Hashtable__U24_Enumerator::nextElement()

Just a wild guess: perhaps the hashtable was modified while the
enumerator is live? The JCL says that will evoke undefined behavior. It
could be a case of something that will work with Sun's Hashtable but not
libgcj's, even if both are correct.
Jeff