From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 3914 invoked by alias); 8 May 2011 13:44:18 -0000 Received: (qmail 3906 invoked by uid 22791); 8 May 2011 13:44:17 -0000 X-SWARE-Spam-Status: No, hits=-2.3 required=5.0 tests=AWL,BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,FREEMAIL_FROM,RCVD_IN_DNSWL_LOW,RFC_ABUSE_POST X-Spam-Check-By: sourceware.org Received: from mail-qw0-f47.google.com (HELO mail-qw0-f47.google.com) (209.85.216.47) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Sun, 08 May 2011 13:44:03 +0000 Received: by qwh5 with SMTP id 5so3049834qwh.20 for ; Sun, 08 May 2011 06:44:02 -0700 (PDT) MIME-Version: 1.0 Received: by 10.229.50.193 with SMTP id a1mr4166697qcg.177.1304862242756; Sun, 08 May 2011 06:44:02 -0700 (PDT) Received: by 10.229.229.196 with HTTP; Sun, 8 May 2011 06:44:02 -0700 (PDT) In-Reply-To: References: <20110325113237.GY18914@tyan-ft48-01.lab.bos.redhat.com> <20110416081117.GV17079@tyan-ft48-01.lab.bos.redhat.com> Date: Sun, 08 May 2011 18:57:00 -0000 Message-ID: Subject: Re: [PATCH] Typed DWARF stack From: "H.J. Lu" To: Jakub Jelinek Cc: Richard Henderson , Jason Merrill , Cary Coutant , gcc-patches@gcc.gnu.org, Roland McGrath , Jan Kratochvil , Tom Tromey , Mark Wielaard Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable X-IsSubscribed: yes Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org X-SW-Source: 2011-05/txt/msg00617.txt.bz2 On Tue, May 3, 2011 at 6:33 AM, H.J. Lu wrote: > On Sat, Apr 16, 2011 at 1:11 AM, Jakub Jelinek wrote: >> On Fri, Mar 25, 2011 at 12:32:37PM +0100, Jakub Jelinek wrote: >>> As I said in my GCC Summit talk, currently we just give up on >>> any floating point/_Decimal*/__int128 and for 32-bit targets even >>> long long expressions, as those can't be represented in DWARF4, >>> while var-tracking has all that often available. >> >> ... >> >> This is an updated patch for the typed DWARF stack support, >> against current trunk. =A0I've changed DW_OP_GNU_reinterpret >> from 0xf8 to 0xf9, because 0xf8 is taken by PGI and e.g. binutils prefer= s no >> clashes between extension opcodes, otherwise readers would need to switch >> based on producer. =A0When we are close to running out of extension opco= des, >> we may reconsider (especially as the HP block is quite large). >> >> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk? >> >> 2011-04-16 =A0Jakub Jelinek =A0 >> >> =A0 =A0 =A0 =A0* dwarf2.h (DW_OP_GNU_const_type, DW_OP_GNU_regval_type, >> =A0 =A0 =A0 =A0DW_OP_GNU_deref_type, DW_OP_GNU_convert, DW_OP_GNU_reinte= rpret): New. >> >> =A0 =A0 =A0 =A0* dwarf2out.c (get_address_mode): New inline. >> =A0 =A0 =A0 =A0(mem_loc_descriptor): Add MEM_MODE parameter, adjust recu= rsive calls, >> =A0 =A0 =A0 =A0if not dwarf_strict emit >> =A0 =A0 =A0 =A0DW_OP_GNU_{{const,regval,deref}_type,convert,reinterpret}= when >> =A0 =A0 =A0 =A0desirable. =A0Handle FLOAT_EXTEND, FLOAT_TRUNCATE, FLOAT, >> =A0 =A0 =A0 =A0UNSIGNED_FLOAT, FIX and UNSIGNED_FIX. =A0Just return NULL= for >> =A0 =A0 =A0 =A0FMA, STRICT_LOW_PART, CONST_VECTOR and CONST_FIXED. >> =A0 =A0 =A0 =A0(dwarf2out_frame_debug_cfa_expression, reg_loc_descriptor, >> =A0 =A0 =A0 =A0dw_loc_list_1, cst_pool_loc_descr, loc_list_from_tree): A= djust >> =A0 =A0 =A0 =A0mem_loc_descriptor callers. >> =A0 =A0 =A0 =A0(dwarf_stack_op_name, size_of_loc_descr, output_loc_opera= nds, >> =A0 =A0 =A0 =A0output_loc_operands_raw, hash_loc_operands, compare_loc_o= perands): >> =A0 =A0 =A0 =A0Handle DW_OP_GNU_const_type, DW_OP_GNU_regval_type, >> =A0 =A0 =A0 =A0DW_OP_GNU_deref_type, DW_OP_GNU_convert and DW_OP_GNU_rei= nterpret. >> =A0 =A0 =A0 =A0(base_types): New variable. >> =A0 =A0 =A0 =A0(get_base_type_offset, calc_base_type_die_sizes, >> =A0 =A0 =A0 =A0base_type_for_mode, mark_base_types, base_type_cmp, >> =A0 =A0 =A0 =A0move_marked_base_types): New functions. >> =A0 =A0 =A0 =A0(calc_die_sizes): Assert that die_offset is 0 or equal to >> =A0 =A0 =A0 =A0next_die_offset. >> =A0 =A0 =A0 =A0(loc_descriptor): Only handle here lowpart SUBREGs of REG= , for >> =A0 =A0 =A0 =A0others defer to mem_loc_descriptor. =A0Adjust mem_loc_des= criptor >> =A0 =A0 =A0 =A0callers. =A0If not dwarf_strict, call mem_loc_descriptor = even for >> =A0 =A0 =A0 =A0non-MODE_INT modes or MODE_INT modes larger than DWARF2_A= DDR_SIZE. >> =A0 =A0 =A0 =A0(gen_subprogram_die): Don't give up on call site paramete= rs >> =A0 =A0 =A0 =A0with non-integral or large integral modes. =A0Adjust >> =A0 =A0 =A0 =A0mem_loc_descriptor callers. >> =A0 =A0 =A0 =A0(prune_unused_types): Call prune_unused_types_mark on bas= e_types >> =A0 =A0 =A0 =A0vector entries. >> =A0 =A0 =A0 =A0(resolve_addr): Call mark_base_types. >> =A0 =A0 =A0 =A0(dwarf2out_finish): Call move_marked_base_types. > > This caused: > > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D48853 > This also caused: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D48928 --=20 H.J.