diff options
Diffstat (limited to 'gcc/cp/class.c')
| -rw-r--r-- | gcc/cp/class.c | 42 |
1 files changed, 21 insertions, 21 deletions
diff --git a/gcc/cp/class.c b/gcc/cp/class.c index c7b0c687db6..597ec92e78f 100644 --- a/gcc/cp/class.c +++ b/gcc/cp/class.c @@ -659,28 +659,27 @@ build_vtable (tree class_type, tree name, tree vtable_type) DECL_EXTERNAL (decl) = 1; DECL_NOT_REALLY_EXTERN (decl) = 1; - if (write_symbols == DWARF2_DEBUG) - /* Mark the VAR_DECL node representing the vtable itself as a - "gratuitous" one, thereby forcing dwarfout.c to ignore it. It - is rather important that such things be ignored because any - effort to actually generate DWARF for them will run into - trouble when/if we encounter code like: + /* Mark the VAR_DECL node representing the vtable itself as a + "gratuitous" one, thereby forcing dwarfout.c to ignore it. It + is rather important that such things be ignored because any + effort to actually generate DWARF for them will run into + trouble when/if we encounter code like: - #pragma interface - struct S { virtual void member (); }; + #pragma interface + struct S { virtual void member (); }; - because the artificial declaration of the vtable itself (as - manufactured by the g++ front end) will say that the vtable is - a static member of `S' but only *after* the debug output for - the definition of `S' has already been output. This causes - grief because the DWARF entry for the definition of the vtable - will try to refer back to an earlier *declaration* of the - vtable as a static member of `S' and there won't be one. We - might be able to arrange to have the "vtable static member" - attached to the member list for `S' before the debug info for - `S' get written (which would solve the problem) but that would - require more intrusive changes to the g++ front end. */ - DECL_IGNORED_P (decl) = 1; + because the artificial declaration of the vtable itself (as + manufactured by the g++ front end) will say that the vtable is + a static member of `S' but only *after* the debug output for + the definition of `S' has already been output. This causes + grief because the DWARF entry for the definition of the vtable + will try to refer back to an earlier *declaration* of the + vtable as a static member of `S' and there won't be one. We + might be able to arrange to have the "vtable static member" + attached to the member list for `S' before the debug info for + `S' get written (which would solve the problem) but that would + require more intrusive changes to the g++ front end. */ + DECL_IGNORED_P (decl) = 1; return decl; } @@ -3529,13 +3528,13 @@ build_base_field (record_layout_info rli, tree binfo, /* Create the FIELD_DECL. */ decl = build_decl (FIELD_DECL, NULL_TREE, CLASSTYPE_AS_BASE (basetype)); DECL_ARTIFICIAL (decl) = 1; + DECL_IGNORED_P (decl) = 1; DECL_FIELD_CONTEXT (decl) = t; DECL_SIZE (decl) = CLASSTYPE_SIZE (basetype); DECL_SIZE_UNIT (decl) = CLASSTYPE_SIZE_UNIT (basetype); DECL_ALIGN (decl) = CLASSTYPE_ALIGN (basetype); DECL_USER_ALIGN (decl) = CLASSTYPE_USER_ALIGN (basetype); DECL_MODE (decl) = TYPE_MODE (basetype); - DECL_IGNORED_P (decl) = 1; DECL_FIELD_IS_BASE (decl) = 1; /* Try to place the field. It may take more than one try if we @@ -4697,6 +4696,7 @@ layout_class_type (tree t, tree *virtuals_p) DECL_SIZE (padding_field) = padding; DECL_CONTEXT (padding_field) = t; DECL_ARTIFICIAL (padding_field) = 1; + DECL_IGNORED_P (padding_field) = 1; layout_nonempty_base_or_field (rli, padding_field, NULL_TREE, empty_base_offsets); |
