http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54837
--- Comment #2 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-10-09
14:14:36 UTC ---
DECL_ABSTRACT_ORIGIN (parm) == aexp
(gdb) call debug_tree (parm)
<parm_decl 0x7ffff5f34000 unit_num
type <integer_type 0x7ffff60c85e8 int asm_written public SI
size <integer_cst 0x7ffff60cc0a0 constant 32>
unit size <integer_cst 0x7ffff60cc0c0 constant 4>
align 32 symtab -165706496 alias set 2 canonical type 0x7ffff60c85e8
precision 32 min <integer_cst 0x7ffff60cc040 -2147483648> max <integer_cst
0x7ffff60cc060 2147483647>
pointer_to_this <pointer_type 0x7ffff60d02a0>>
used SI file /space/rguenther/src/svn/trunk/gcc/genautomata.c line 3573 col
61 size <integer_cst 0x7ffff60cc0a0 32> unit size <integer_cst 0x7ffff60cc0c0
4>
align 32 context <function_decl 0x7ffff61fff00
_ZL16test_unit_reservPmii.3701.constprop.227> abstract_origin <parm_decl
0x7ffff61fce80 unit_num>
(reg/v:SI 71 [ unit_num ]) arg-type <integer_type 0x7ffff60c85e8 int>
incoming-rtl (reg:SI 4 si [ unit_num ])>
(gdb) call debug_tree (aexp)
<parm_decl 0x7ffff61fce80 unit_num
type <integer_type 0x7ffff60c85e8 int asm_written public SI
size <integer_cst 0x7ffff60cc0a0 constant 32>
unit size <integer_cst 0x7ffff60cc0c0 constant 4>
align 32 symtab -165706496 alias set 2 canonical type 0x7ffff60c85e8
precision 32 min <integer_cst 0x7ffff60cc040 -2147483648> max <integer_cst
0x7ffff60cc060 2147483647>
pointer_to_this <pointer_type 0x7ffff60d02a0>>
used SI file /space/rguenther/src/svn/trunk/gcc/genautomata.c line 3573 col
61 size <integer_cst 0x7ffff60cc0a0 32> unit size <integer_cst 0x7ffff60cc0c0
4>
align 32 context <function_decl 0x7ffff61ffe00 test_unit_reserv> arg-type
<integer_type 0x7ffff60c85e8 int>>
I'm testing a quick hack. Broken LTO bootstrap is annoying.
