GCJ Build Problem (CVS)

Sat Nov 10 16:25:00 GMT 2001

On Saturday, November 10, 2001, at 06:14 PM, Bryce McKinlay wrote:
> So, the problem is that you have an old version of libgcj.so sitting 
> around somewhere and the linker is seeing that before the new libgcj.so. 
> But normally that should never happen since GCJ finds libgcj.so relative 
> to the path of the compiler. You did remember to "make install", right?

Yes I did make install.
Here's the problem. Presumably /usr/lib/libgcj* is from the RedHat gcc 
installation; I sure didn't put them there. /usr/local/lib/libgcj* is 
probably left over from my first try at upgrading gcc, but that can't be 
it because its not in /etc/ld.so.config. /usr/local/gcc31 isn't listed 
since I created it after the nightly locate -u.
[bcox@linux bcox]$ cat /etc/ld.so.conf
/usr/X11R6/lib
/usr/lib
/usr/i486-linux-libc5/lib
/usr/kerberos/lib
/usr/lib/gconv
/usr/lib/qt-2.1.0/lib
/usr/lib/qt-1.45/lib
/vame/ive/bin
/usr/lib/qt-2.2.0/lib
/usr/lib/sane
/usr/lib/mysql
/usr/local/xmms/Input
/lib
[bcox@linux bcox]$ locate libgcj.so
/usr/lib/libgcj.so.1
/usr/lib/libgcj.so.1.0.0
/usr/lib/libgcj.so
/usr/local/lib/libgcj.so.2
/usr/local/lib/libgcj.so.2.0.0
/usr/local/lib/libgcj.so
[bcox@linux bcox]$ d /usr/lib/libgcj*
-rw-r--r-- 1 root root 11240656 Aug 21 2000 /usr/lib/libgcj.a
lrwxrwxrwx 1 root root 15 Feb 26 2001 /usr/lib/libgcj.so 
-> libgcj.so.1.0.0
lrwxrwxrwx 1 root root 15 Feb 26 2001 /usr/lib/libgcj.so.
1 -> libgcj.so.1.0.0
-rwxr-xr-x 1 root root 6834973 Aug 21 2000 /usr/lib/libgcj.so.
1.0.0
-rw-r--r-- 1 root root 121064 Aug 21 2000 /usr/lib/libgcjgc.
a
lrwxrwxrwx 1 root root 17 Feb 26 2001 /usr/lib/libgcjgc.
so -> libgcjgc.so.1.0.1
lrwxrwxrwx 1 root root 17 Feb 26 2001 /usr/lib/libgcjgc.
so.1 -> libgcjgc.so.1.0.1
-rwxr-xr-x 1 root root 103373 Aug 21 2000 /usr/lib/libgcjgc.
so.1.0.1
[bcox@linux bcox]$ d /usr/local/lib/libgcj*
-rw-r--r-- 1 root root 18812824 Aug 28 11:57 
/usr/local/lib/libgcj.a
-rwxr-xr-x 1 root root 754 Aug 28 11:56 
/usr/local/lib/libgcj.la
lrwxrwxrwx 1 root root 15 Aug 28 11:56 
/usr/local/lib/libgcj.so -> libgcj.so.2.0.0
lrwxrwxrwx 1 root root 15 Aug 28 11:56 
/usr/local/lib/libgcj.so.2 -> libgcj.so.2.0.0
-rwxr-xr-x 1 root root 10760265 Aug 28 11:56 
/usr/local/lib/libgcj.so.2.0.0
-rw-r--r-- 1 root root 314 Aug 28 11:57 
/usr/local/lib/libgcj.spec
-rw-r--r-- 1 root root 670088 Aug 28 11:58 
/usr/local/lib/libgcjgc.a
-rwxr-xr-x 1 root root 712 Aug 28 11:58 
/usr/local/lib/libgcjgc.la
lrwxrwxrwx 1 root root 17 Aug 28 11:58 
/usr/local/lib/libgcjgc.so -> libgcjgc.so.1.0.1
lrwxrwxrwx 1 root root 17 Aug 28 11:58 
/usr/local/lib/libgcjgc.so.1 -> libgcjgc.so.1.0.1
-rwxr-xr-x 1 root root 321190 Aug 28 11:58 
/usr/local/lib/libgcjgc.so.1.0.1
> Usually, I think, people keep the install prefix separate from the 
> src/build directory. Maybe this could be causing problems...

The change was to fix one of the things I found confusing; source, build 
and install directories at entirely different and unrelated places in /usr/
local. The only difference is that they all share /usr/local/gcj31 in my 
scheme versys /usr/local in the original, so I doubt that's it.
But the from the gcc -v output (prior mail), gcj IS behaving as it should.
 The problem is that a.out is apparently using a different path. So I tried
test: checkbook.a Test.o
 $(GCJ) -v --main=Test Test.o
 export LD_LIBRARY_PATH="$(GCC)/lib:$(LD_LIBRARY_PATH)" && a.out
IT WORKED! At least its not dying in SecureRandom as before but much later.
 I'll dig into that next.
Thanks for the help!!