* fixed optimised form of a_cmp_const_reg_label(): when inverting the branch, we still

    have to emit the label we originally allocated for the fallthrough path (otherwise
    the branch is not inverted at all, and moreover the caller will also emit the
    original target label somewhere so it will be doubly defined)

git-svn-id: branches/hlcgllvm@28367 -

compiler/llvm/hlcgllvm.pas

@@ -769,7 +769,7 @@ implementation
     var
       tmpreg : tregister;
       invert: boolean;
-      falselab, tmplab: tasmlabel;
+      fallthroughlab, falselab, tmplab: tasmlabel;
     begin
       { since all comparisons return their results in a register, we'll often
         get comparisons against true/false -> optimise }
@@ -783,6 +783,7 @@ implementation
               invert:=a=1;
             end;
           current_asmdata.getjumplabel(falselab);
+          fallthroughlab:=falselab;
           if invert then
             begin
               tmplab:=l;
@@ -790,7 +791,7 @@ implementation
               falselab:=tmplab;
             end;
           list.concat(taillvm.op_size_reg_lab_lab(la_br,pasbool8type,reg,l,falselab));
-          a_label(list,falselab);
+          a_label(list,fallthroughlab);
           exit;
         end;
       tmpreg:=getregisterfordef(list,size);