Patch: RFA: fix PR java/17329

[Andrew Haley <aph at redhat dot com>]

Tom Tromey writes:
 > In PR 17329, we are compiling code like this, where Options is a
 > different class:
 > 
 >                 Options.verbose += 2;
 > 
 > This results in a save_expr the _Jv_InitClass call; there is special
 > code in gimplify_save_expr() to handle this (searching for "java" in
 > this file was enlightening, I had no idea about all these special
 > cases).
 > 
 > Anyway, gimplify_save_expr will set the save_expr argument to NULL the
 > first time it is seen, but java_gimplify_expr doesn't account for this
 > case and crashes.
 > 
 > 
 > Index: gcc/java/ChangeLog
 > from  Tom Tromey  <tromey@redhat.com>
 > 	PR java/17329:
 > 	* java-gimplify.c (java_gimplify_expr) <SAVE_EXPR>: Ignore case
 > 	where operand is null.

This is fine. 

I guess this happens when we generate a SAVE_EXPR with void type.  I
wanted to be able to say "is this a variable?" for which 
  (TREE_CODE (TREE_OPERAND (*expr_p, 0)) == VAR_DECL)
looked fine.  But I'd either forgotten, or never knew, that we
{ab}used SAVE_EXPRs with void type.

Andrew.