[Bug debug/48853] [4.7 Regression] Wrong DWARF codegen when Pmode != ptr_mode

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48853

--- Comment #17 from H.J. Lu <hjl.tools at gmail dot com> 2011-05-06 22:38:34 UTC ---
loc_descriptor change in

http://gcc.gnu.org/ml/gcc-cvs/2011-04/msg01406.html

@@ -14469,7 +15018,10 @@
      up an entire register.  For now, just assume that it is
      legitimate to make the Dwarf info refer to the whole register which
      contains the given subreg.  */
-      loc_result = loc_descriptor (SUBREG_REG (rtl), mode, initialized);
+      if (REG_P (SUBREG_REG (rtl)) && subreg_lowpart_p (rtl))
+    loc_result = loc_descriptor (SUBREG_REG (rtl), mode, initialized);
+      else
+    goto do_default;
       break;

     case REG:

doesn't handle:

Breakpoint 5, loc_descriptor (rtl=0x7ffff0b24c30, mode=SImode, 
    initialized=VAR_INIT_STATUS_INITIALIZED)
    at /export/gnu/import/git/gcc-x32/gcc/dwarf2out.c:15035
15035          if ((REG_P (SUBREG_REG (rtl))
(gdb) call debug_rtx (rtl)
(subreg:SI (symbol_ref:DI ("a") [flags 0x2] <var_decl 0x7ffff0a9c500 a>) 0)
(gdb) 

It happens when Pmode != ptr_mode in gcc.dg/guality/inline-params.c.