From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 14717 invoked by alias); 11 Apr 2004 23:18:05 -0000 Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-bugs-owner@gcc.gnu.org Received: (qmail 14698 invoked by uid 48); 11 Apr 2004 23:18:00 -0000 Date: Mon, 12 Apr 2004 00:38:00 -0000 Message-ID: <20040411231800.14697.qmail@sources.redhat.com> From: "pinskia at gcc dot gnu dot org" To: gcc-bugs@gcc.gnu.org In-Reply-To: <20040411221509.14919.hjl@lucon.org> References: <20040411221509.14919.hjl@lucon.org> Reply-To: gcc-bugzilla@gcc.gnu.org Subject: [Bug debug/14919] [3.3 Regression]: DW_AT_comp_dir is missing X-Bugzilla-Reason: CC X-SW-Source: 2004-04/txt/msg00952.txt.bz2 List-Id: ------- Additional Comments From pinskia at gcc dot gnu dot org 2004-04-11 23:17 ------- This has not changed since: 1.126 (jason 23-Nov-99): if (wd != NULL && filename[0] != DIR_SEPARATOR) 1.126 (jason 23-Nov-99): add_AT_string (die, DW_AT_comp_dir, wd); So I do not know how 3.2 worked at all. 3.4 changed it to: if (filename[0] != DIR_SEPARATOR) add_comp_dir_attribute (comp_unit_die); else if (get_AT (comp_unit_die, DW_AT_comp_dir) == NULL) { size_t i; for (i = 1; i < VARRAY_ACTIVE_SIZE (file_table); i++) if (VARRAY_CHAR_PTR (file_table, i)[0] != DIR_SEPARATOR /* Don't add cwd for . */ && VARRAY_CHAR_PTR (file_table, i)[0] != '<') { add_comp_dir_attribute (comp_unit_die); break; } } This was changed with: 2003-01-16 Richard Henderson * dwarf2out.c (struct file_table): Remove. (FILE_TABLE_INCREMENT): Remove. (file_table): Make a varray; mark for GC. Update all users. (file_table_last_lookup_index): Extract from struct file_table. (output_file_names): Fix unsigned compare warnings. (add_name_attribute): Remove inline marker. (add_comp_dir_attribute): Split out from gen_compile_unit_die. (lookup_filename): Don't manage size of file_table. (init_file_table): Allocate file_table with GC. (dwarf2out_init): Don't record main_input_filename here. (dwarf2out_finish): Do it here instead. See for the patch but this was needed for PCH only so something else has changed in the compiler, I am starting to think the preprocessor changed. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14919