Re: Could you please check this into 3.0.1, too?

[Martin Kahlert <martin dot kahlert at infineon dot com>]

Hi Jeff,
thanks for committing the two patches.

On Tue, Sep 18, 2001 at 12:40:37AM -0400, Jeff Sturm wrote:
> I was mildly surprised that nothing had been done in libjava since the
> 3.0.1 release.  Are there other simple patches that could go up now?  Some
> pre-3.0.2 application testing might be a good idea.

Here are my two favorites:

2001-05-03  Tom Tromey  <tromey@redhat.com>

	* java/lang/natRuntime.cc (init): Call add_library on the program
	itself.
	(_Jv_FindSymbolInExecutable): Return NULL if no library on the
	list has the symbol.

diff -rc gcc-20010910.orig/libjava/java/lang/natRuntime.cc gcc-20010910/libjava/java/lang/natRuntime.cc
*** gcc-20010910.orig/libjava/java/lang/natRuntime.cc	Mon Mar 12 08:33:57 2001
--- gcc-20010910/libjava/java/lang/natRuntime.cc	Tue Sep 11 09:07:28 2001
***************
*** 69,75 ****
  	return r;
      }
  
!   return lt_dlsym (NULL, symname);
  }
  
  #endif /* USE_LTDL */
--- 69,75 ----
  	return r;
      }
  
!   return NULL;
  }
  
  #endif /* USE_LTDL */
***************
*** 191,196 ****
--- 191,199 ----
    finalize_on_exit = false;
  #ifdef USE_LTDL
    lt_dlinit ();
+   lt_dlhandle self = lt_dlopen (NULL);
+   if (self != NULL)
+     add_library (self);
  #endif
  }
  

and this important one (it causes an infinite loop for my application):

2001-06-15  Tom Tromey  <tromey@redhat.com>

	* jni.cc (_Jv_JNI_NewLocalRef): Search other frames.

Index: jni.cc
===================================================================
RCS file: /cvs/gcc/gcc/libjava/jni.cc,v
retrieving revision 1.46
diff -u -r1.46 jni.cc
--- jni.cc      2001/06/15 22:09:10     1.46
+++ jni.cc      2001/06/15 23:40:45
@@ -278,16 +278,23 @@
   // Try to find an open slot somewhere in the topmost frame.
   _Jv_JNI_LocalFrame *frame = env->locals;
   bool done = false, set = false;
-  while (frame != NULL && ! done)
+  for (; frame != NULL && ! done; frame = frame->next)
     {
       for (int i = 0; i < frame->size; ++i)
-       if (frame->vec[i] == NULL)
-         {
-           set = true;
-           done = true;
-           frame->vec[i] = obj;
-           break;
-         }
+       {
+         if (frame->vec[i] == NULL)
+           {
+             set = true;
+             done = true;
+             frame->vec[i] = obj;
+             break;
+           }
+       }
+
+      // If we found a slot, or if the frame we just searched is the
+      // mark frame, then we are done.
+      if (done || frame->marker != MARK_NONE)
+       break;
     }

   if (! set)


I would be glad, if these patches could be applied, too.

Thanks, Martin.

PS: I cc'ed the java list, so other people can make suggestions, too.

-- 
The early bird catches the worm. If you want something else for       
breakfast, get up later.