dwarf2out questi

Aldy Hernandez aldyh@redhat.com
Sun Mar 3 23:42:00 GMT 2002


On Monday, March 4, 2002, at 02:35  PM, David Edelsohn wrote:

> 	I think it is a bug that dwarf2out.c mem_loc_descriptor does not
> handle LO_SUM.  But I also think that your patch is wrong.  I would have
> expected LO_SUM to be handled like PLUS.
>
> David
>
>

well, what the heck, let's include everyone on the discussion.

here is a patch to handle LO_SUM in mem_loc_descriptor.  i have
a testcase, which unfortunately does not trigger the error in
FSF gcc, but on a branch cut a few weeks ago, which aborts
because mem_loc_descriptor does not handle LO_SUM.

is this ok or is this something to be done in

	ASM_SIMPLIFY_DWARF_ADDR

2002-03-04  Aldy Hernandez  <aldyh@redhat.com>

         * dwarf2out.c (mem_loc_descriptor): Handle LO_SUM.

Index: dwarf2out.c
===================================================================
RCS file: /cvs/uberbaum/gcc/dwarf2out.c,v
retrieving revision 1.357
diff -c -p -r1.357 dwarf2out.c
*** dwarf2out.c 2002/02/26 06:57:59     1.357
--- dwarf2out.c 2002/03/04 07:39:56
*************** mem_loc_descriptor (rtl, mode)
*** 7849,7854 ****
--- 7849,7855 ----
         /* ... fall through ...  */

       case PLUS:
+     case LO_SUM:
       plus:
         if (is_based_loc (rtl))
         mem_loc_result = based_loc_descr (reg_number (XEXP (rtl, 0)),



More information about the Gcc-patches mailing list