https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84213
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |pmderodat at gcc dot gnu.org
--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
.uleb128 0x7 # (DIE (0x56) DW_TAG_variable)
.ascii "em1\0" # DW_AT_name
.byte 0x1 # DW_AT_decl_file (t.f90)
.byte 0x2 # DW_AT_decl_line
.long 0x1a # DW_AT_type
# DW_AT_external
.uleb128 0x9 # DW_AT_location
.byte 0x3 # DW_OP_addr
.quad module_ra_gfdleta.eq.0_
and we get here via
#2 0x0000000000af7740 in analyze_functions (first_time=true)
at /space/rguenther/src/svn/early-lto-debug/gcc/cgraphunit.c:1256
1256 (*debug_hooks->late_global_decl) (node->decl);
(gdb) l
1251 /* We are reclaiming totally unreachable code and
variables
1252 so they effectively appear as readonly. Show that to
1253 the debug machinery. */
1254 TREE_READONLY (node->decl) = 1;
1255 node->definition = false;
1256 (*debug_hooks->late_global_decl) (node->decl);
1257 }
but the intent doesn't work because we have a DECL_VALUE_EXPR here...
#1 0x0000000000b99a34 in dwarf2out_late_global_decl (
decl=<var_decl 0x7ffff7fefbd0 em1v>)
at /space/rguenther/src/svn/early-lto-debug/gcc/dwarf2out.c:25750
(gdb) l
25745 they are relevant for debuggers. */
25746 varpool_node *node = varpool_node::get (decl);
25747 if ((! node || ! node->definition) && ! DECL_HAS_VALUE_EXPR_P
(decl))
25748 tree_add_const_value_attribute_for_decl (die, decl);
25749 else
25750 add_location_or_const_value_attribute (die, decl, false);
25751 }
25752 }
I didn't add this extra check but this breaks things here :/ Added by
r248792.
