From owner-pro64-support@oss.sgi.com Mon Apr 2 11:56:31 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f32IuV925223 for pro64-support-outgoing; Mon, 2 Apr 2001 11:56:31 -0700 Received: from osc.edu (osc.edu [192.148.249.4]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f32IuUM25204 for ; Mon, 2 Apr 2001 11:56:30 -0700 Received: from neptune.osc.edu (IDENT:djohnson@neptune.osc.edu [192.148.249.73]) by osc.edu (8.9.3/8.9.3/OSC 2.0) with ESMTP id OAA02124; Mon, 2 Apr 2001 14:55:49 -0400 (EDT) Date: Mon, 2 Apr 2001 14:55:44 -0400 (EDT) From: Douglas Johnson X-Sender: djohnson@localhost.localdomain To: "Chan, Sun C" cc: "'Kratzer, Willi'" , "'pro64-support@oss.sgi.com'" Subject: RE: auto-pic or non-auto-pic In-Reply-To: <9287DC1579B0D411AA2F009027F44C3F042DF9CA@FMSMSX41> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-pro64-support@oss.sgi.com Precedence: bulk If we use the command line option to pass the right flags to 'as' so that the object files are created with NOFUNCDESC_CONS_GP will this solve the problem? Or are objects created this way only compatable with other objects created this way? So, the bottom line question is, is there some flag for the linker to make these objects link??? I think the answer is no, unless the linker is generating stubs. I'd try unpacking the libmkl64_itp.a archive and replace the smpsrv64.o with a replacement routine that does exit(-1) if called. Just make sure that MKL_NUM_THREADS is not set (or whatever the thread number varibale is). Doug On Fri, 30 Mar 2001, Chan, Sun C wrote: > Looks like I've spoken too hastily. It is not simply > pic vs non-pic. I apologize. > Apparently, it has to do with the function descriptor portion > of the ABI. > Sun > From owner-pro64-support@oss.sgi.com Mon Apr 2 20:03:42 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3333gp17457 for pro64-support-outgoing; Mon, 2 Apr 2001 20:03:42 -0700 Received: from hotmail.com (f179.law14.hotmail.com [64.4.21.179]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3333fM17454 for ; Mon, 2 Apr 2001 20:03:41 -0700 Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Mon, 2 Apr 2001 20:03:36 -0700 Received: from 159.226.40.178 by lw14fd.law14.hotmail.msn.com with HTTP; Tue, 03 Apr 2001 03:03:36 GMT X-Originating-IP: [159.226.40.178] From: "Yanjun HU" To: pro64-support@oss.sgi.com Cc: huyanjun79@hotmail.com Subject: Install Fortran frontend? Date: Tue, 03 Apr 2001 11:03:36 +0800 Mime-Version: 1.0 Content-Type: text/plain; format=flowed Message-ID: X-OriginalArrivalTime: 03 Apr 2001 03:03:36.0403 (UTC) FILETIME=[ACE87230:01C0BBEA] Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Dear pro64 support team, I want to only install the frontend and whirl2c and whirl2f component of the pro64 from the osprey source tree. Using the instruction of the README.src, I have succeed in building the driver, C frontend, C++ frontend, but next when I try to build fortran frontend, I falled. I can see the error: sh: /usr/ia32-sgi-linux/bin/f90: No such file or directory gmake[1]: ***[fold.o] Error 126 When I watch the Make.cross file, I see that the fortran frontend was the last one to be built. I want to know whether the building of fortran frontend needs other components. If yes, what are they? Thanks!! Hu Yanjun _________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com. From owner-pro64-support@oss.sgi.com Mon Apr 2 22:09:06 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f33596Y20627 for pro64-support-outgoing; Mon, 2 Apr 2001 22:09:06 -0700 Received: from sgi.com (sgi.SGI.COM [192.48.153.1]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f33595M20624 for ; Mon, 2 Apr 2001 22:09:05 -0700 Received: from cchkms.engr.sgi.com ([130.62.180.48]) by sgi.com (980327.SGI.8.8.8-aspam/980304.SGI-aspam: SGI does not authorize the use of its proprietary systems or networks for unsolicited or bulk email from the Internet.) via ESMTP id WAA03257 for ; Mon, 2 Apr 2001 22:09:04 -0700 (PDT) mail_from (rat@cchkms.engr.sgi.com) Received: (from rat@localhost) by cchkms.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id WAA55515; Mon, 2 Apr 2001 22:07:22 -0700 (PDT) Date: Mon, 2 Apr 2001 22:07:22 -0700 (PDT) From: rat@cchkms.engr.sgi.com (Ross A. Towle) Message-Id: <200104030507.WAA55515@cchkms.engr.sgi.com> To: pro64-support@oss.sgi.com, "Yanjun HU" Subject: Re: Install Fortran frontend? Cc: huyanjun79@hotmail.com Sender: owner-pro64-support@oss.sgi.com Precedence: bulk You need a Fortran90 compiler to build part of the fortran90 front end. That Fortran90 compiler needs to generate code for whatever host you are intending to host your development on. So if you want your tool to run on IA32 you will need a IA32 Fortran90 front end. -Ross A. Towle From owner-pro64-support@oss.sgi.com Wed Apr 4 12:56:15 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f34JuFI30099 for pro64-support-outgoing; Wed, 4 Apr 2001 12:56:15 -0700 Received: from postal1.lbl.gov (postal1.lbl.gov [128.3.7.82]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f34JuFM30096 for ; Wed, 4 Apr 2001 12:56:15 -0700 Received: from SpamWall.lbl.gov (localhost [127.0.0.1]) by postal1.lbl.gov (8.11.2/8.9.3) with ESMTP id f34JuEo25153 for ; Wed, 4 Apr 2001 12:56:14 -0700 (PDT) Received: from lbl.gov (sensomatic.lbl.gov [131.243.240.92]) by SpamWall.lbl.gov (8.11.2/8.11.2) with ESMTP id f34JuD125149 for ; Wed, 4 Apr 2001 12:56:13 -0700 (PDT) Message-ID: <3ACB7C5D.C7CAE308@lbl.gov> Date: Wed, 04 Apr 2001 12:56:13 -0700 From: Ricardo da Silva X-Mailer: Mozilla 4.5 [en] (X11; I; SunOS 5.6 sun4u) X-Accept-Language: en MIME-Version: 1.0 CC: pro64-support@oss.sgi.com Subject: Exponation Error References: <200104030507.WAA55515@cchkms.engr.sgi.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Hi All, I' trying port my application, a atmospheric model, to IA64. I got some "unexplained" errors, the compiler (sgif90) just pointed to Front end Phase Error (anything like that). I tracked the error and found that the sgif90 compiler, do not accept the statment: 10.**2 it just accept: 10**2 I don't now if it's a bug or non-standard f77 construction (my code are in fixed form/f77), but when I change my code to the last form it work! Could anybody explain that error to me? Thanks in advance, Regards, Ricardo M. From owner-pro64-support@oss.sgi.com Wed Apr 4 17:44:32 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f350iWW06460 for pro64-support-outgoing; Wed, 4 Apr 2001 17:44:32 -0700 Received: from deliverator.sgi.com (deliverator.sgi.com [204.94.214.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f350iVM06455 for ; Wed, 4 Apr 2001 17:44:31 -0700 Received: from cchkms.engr.sgi.com (cchkms.engr.sgi.com [130.62.180.48]) by deliverator.sgi.com (980309.SGI.8.8.8-aspam-6.2/980310.SGI-aspam) via ESMTP id RAA02349 for ; Wed, 4 Apr 2001 17:43:16 -0700 (PDT) mail_from (rat@cchkms.engr.sgi.com) Received: (from rat@localhost) by cchkms.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id RAA59573; Wed, 4 Apr 2001 17:43:01 -0700 (PDT) From: "Ross A. Towle" Message-Id: <10104041743.ZM59528@cchkms.engr.sgi.com> Date: Wed, 4 Apr 2001 17:42:58 -0700 In-Reply-To: Ricardo da Silva "Exponation Error" (Apr 4, 12:56pm) References: <200104030507.WAA55515@cchkms.engr.sgi.com> <3ACB7C5D.C7CAE308@lbl.gov> X-Mailer: Z-Mail (3.2.3 08feb96 MediaMail) To: Ricardo da Silva Subject: Re: Exponation Error Cc: pro64-support@oss.sgi.com Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: owner-pro64-support@oss.sgi.com Precedence: bulk I just tried the following subroutine x(a,b,c) real*8 a,b,c a = 10. ** 2 end and it compiles fine. -Ross From owner-pro64-support@oss.sgi.com Wed Apr 4 23:53:49 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f356rnd08309 for pro64-support-outgoing; Wed, 4 Apr 2001 23:53:49 -0700 Received: from rly-ip02.mx.aol.com (rly-ip02.mx.aol.com [152.163.225.160]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f356rmM08306 for ; Wed, 4 Apr 2001 23:53:49 -0700 Received: from tot-wf1-we.proxy.aol.com (tot-wf1-we.proxy.aol.com [205.188.195.3]) by rly-ip02.mx.aol.com (8.8.8/8.8.8/AOL-5.0.0) with ESMTP id CAA21779 for ; Thu, 5 Apr 2001 02:53:09 -0400 (EDT) Received: from lbl.gov (AC8BF50D.ipt.aol.com [172.139.245.13]) by tot-wf1-we.proxy.aol.com (8.10.0/8.10.0) with ESMTP id f356r8023335 for ; Thu, 5 Apr 2001 02:53:08 -0400 (EDT) Message-ID: <3ACC1655.5DFB14D4@lbl.gov> Date: Wed, 04 Apr 2001 23:53:09 -0700 From: Ricardo Marcelo da Silva X-Mailer: Mozilla 4.73 [en]C-{C-UDP; EBM-SONY1} (Win95; U) X-Accept-Language: en MIME-Version: 1.0 To: pro64-support@oss.sgi.com Subject: Re: Exponation Error References: <200104030507.WAA55515@cchkms.engr.sgi.com> <3ACB7C5D.C7CAE308@lbl.gov> <10104041743.ZM59528@cchkms.engr.sgi.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Apparently-From: Bengamaster@aol.com Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Oh yes, it works. But if you try make little changes in the code you'll get a different result. Try it: subroutine x(a,b,c) real*8 a,b,c b=2 a = 10.**b end le'me know what happens. Thanks, Ricardo. "Ross A. Towle" wrote: > I just tried the following > > subroutine x(a,b,c) > real*8 a,b,c > > a = 10. ** 2 > end > > and it compiles fine. > > -Ross From owner-pro64-support@oss.sgi.com Thu Apr 5 06:10:17 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f35DAHa26041 for pro64-support-outgoing; Thu, 5 Apr 2001 06:10:17 -0700 Received: from narkis.wisdom.weizmann.ac.il (narkis.wisdom.weizmann.ac.il [132.76.80.32]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f35DAFM26037 for ; Thu, 5 Apr 2001 06:10:15 -0700 Received: from wisdom.weizmann.ac.il (amir8-pc.wisdom.weizmann.ac.il [132.76.81.32]) by narkis.wisdom.weizmann.ac.il (8.9.3/8.9.3) with ESMTP id PAA28849 for ; Thu, 5 Apr 2001 15:09:11 +0200 (IST) Message-ID: <3ACC6EB5.1CDFFE6A@wisdom.weizmann.ac.il> Date: Thu, 05 Apr 2001 15:10:13 +0200 From: raya Organization: Weizmann Institute of Science X-Mailer: Mozilla 4.75 [en] (WinNT; U) X-Accept-Language: il,en MIME-Version: 1.0 To: pro64 Subject: Liveness and Definition Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Hello, What is the exact definition of the following fields, which are printed to the trace file : defreach_in live_in defreach_out live_out live_def live_use Is there a specific paper that describes PRO64 data dependece graph issues ? Thanks, Raya -- Raya Leviathan Tel. 972-8-9344208 (office) Tel. 972-3-6358481 (home) Email: raya@wisdom.weizmann.ac.il From owner-pro64-support@oss.sgi.com Thu Apr 5 08:55:29 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f35FtT331168 for pro64-support-outgoing; Thu, 5 Apr 2001 08:55:29 -0700 Received: from rumor.cps.intel.com (rumor.cps.intel.com [192.102.198.242]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f35FtTM31165 for ; Thu, 5 Apr 2001 08:55:29 -0700 Received: from SMTP (fmsmsxvs05-1.fm.intel.com [132.233.42.205]) by rumor.cps.intel.com (8.9.1a+p1/8.9.1/d: relay.m4,v 1.35 2001/02/12 09:03:45 smothers Exp $) with SMTP id PAA22987; Thu, 5 Apr 2001 15:54:35 GMT Received: from fmsmsx28.fm.intel.com ([132.233.48.28]) by 132.233.48.205 (Norton AntiVirus for Internet Email Gateways 1.0) ; Thu, 05 Apr 2001 15:54:37 0000 (GMT) Received: by fmsmsx28.fm.intel.com with Internet Mail Service (5.5.2653.19) id ; Thu, 5 Apr 2001 08:54:36 -0700 Message-ID: <9287DC1579B0D411AA2F009027F44C3F042DFA09@FMSMSX41> From: "Chan, Sun C" To: "'raya'" , pro64 Subject: RE: Liveness and Definition Date: Thu, 5 Apr 2001 08:54:34 -0700 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Just look up any decent compiler text book. Sun > -----Original Message----- > From: raya [mailto:raya@wisdom.weizmann.ac.il] > Sent: Thursday, April 05, 2001 6:10 AM > To: pro64 > Subject: Liveness and Definition > > > Hello, > > What is the exact definition of the following fields, which > are printed > to the trace file : > > > defreach_in > live_in > defreach_out > live_out > live_def > live_use > > Is there a specific paper that describes PRO64 data dependece graph > issues ? > > Thanks, > Raya > -- > Raya Leviathan > Tel. 972-8-9344208 (office) > Tel. 972-3-6358481 (home) > Email: raya@wisdom.weizmann.ac.il > > > From owner-pro64-support@oss.sgi.com Thu Apr 5 10:47:47 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f35HllS02080 for pro64-support-outgoing; Thu, 5 Apr 2001 10:47:47 -0700 Received: from palrel1.hp.com (palrel1.hp.com [156.153.255.242]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f35HllM02077 for ; Thu, 5 Apr 2001 10:47:47 -0700 Received: from hpbs5001.boi.hp.com (hpbs5001.boi.hp.com [15.2.209.237]) by palrel1.hp.com (Postfix) with ESMTP id 83F89ABC; Thu, 5 Apr 2001 10:47:46 -0700 (PDT) Received: from xatlbh1.atl.hp.com (xatlbh1.atl.hp.com [15.45.89.186]) by hpbs5001.boi.hp.com with ESMTP (8.7.1/8.7.3 SMKit7.02) id LAA29732; Thu, 5 Apr 2001 11:47:45 -0600 (MDT) Received: by xatlbh1.atl.hp.com with Internet Mail Service (5.5.2653.19) id <2LGS7MW4>; Thu, 5 Apr 2001 13:46:46 -0400 Message-ID: From: "JAIN,SUNEEL (HP-Cupertino,ex1)" To: "'raya'" , "'pro64'" Subject: RE: Liveness and Definition Date: Thu, 5 Apr 2001 13:44:24 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain Sender: owner-pro64-support@oss.sgi.com Precedence: bulk You will find brief descriptions in the following files: be/cg/gra_live.h be/cg/bbregs.h - Suneel Jain > -----Original Message----- > From: owner-pro64-support@oss.sgi.com > [mailto:owner-pro64-support@oss.sgi.com]On Behalf Of raya > Sent: Thursday, April 05, 2001 6:10 AM > To: pro64 > Subject: Liveness and Definition > > > Hello, > > What is the exact definition of the following fields, which > are printed > to the trace file : > > > defreach_in > live_in > defreach_out > live_out > live_def > live_use > > Is there a specific paper that describes PRO64 data dependece graph > issues ? > > Thanks, > Raya > -- > Raya Leviathan > Tel. 972-8-9344208 (office) > Tel. 972-3-6358481 (home) > Email: raya@wisdom.weizmann.ac.il > > > From owner-pro64-support@oss.sgi.com Fri Apr 6 20:40:28 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f373eS314335 for pro64-support-outgoing; Fri, 6 Apr 2001 20:40:28 -0700 Received: from scapa.cs.ualberta.ca (root@scapa.cs.ualberta.ca [129.128.4.44]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f373eRM14332 for ; Fri, 6 Apr 2001 20:40:27 -0700 Received: (from localhost user: 'pengzhao' uid#483 fake: STDIN (pengzhao@peers)) by scapa.cs.ualberta.ca id ; Fri, 6 Apr 2001 21:40:11 -0600 Date: Fri, 6 Apr 2001 21:40:10 -0600 (MDT) From: Peng Zhao To: sgi Subject: Question on function Recompute_Addr_Taken() Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Hi, I have a question about the function ipa/local/ipl_summarize_template.h -->Recompute_Addr_Taken(), and Set_Addr_Taken_Attrib. There is some comments like "// The preopt always set the ST_addr_saved bit too conservatively. So we need to recompute them." // search for run-time address taken I don't know what do they mean and the contex. What is the meaning of "run-time address taken"? Thanks -- Regards Peng Peng Zhao pengzhao@cs.ualberta.ca http://www.cs.ualberta.ca/~pengzhao TEL (Lab): (780)492-3725 Lab: CSC251 From owner-pro64-support@oss.sgi.com Mon Apr 9 09:00:03 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f39G03v10053 for pro64-support-outgoing; Mon, 9 Apr 2001 09:00:03 -0700 Received: from hypnos.cps.intel.com (hypnos.cps.intel.com [192.198.165.17]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f39G02M10049 for ; Mon, 9 Apr 2001 09:00:02 -0700 Received: from SMTP (fmsmsxvs04-1.fm.intel.com [132.233.42.204]) by hypnos.cps.intel.com (8.9.1a+p1/8.9.1/d: relay.m4,v 1.35 2001/02/12 09:03:45 smothers Exp $) with SMTP id PAA20473; Mon, 9 Apr 2001 15:59:34 GMT Received: from fmsmsx27.FM.INTEL.COM ([132.233.48.27]) by 132.233.48.204 (Norton AntiVirus for Internet Email Gateways 1.0) ; Mon, 09 Apr 2001 15:59:25 0000 (GMT) Received: by fmsmsx27.fm.intel.com with Internet Mail Service (5.5.2653.19) id <2GMQTZ75>; Mon, 9 Apr 2001 08:59:22 -0700 Message-ID: <9287DC1579B0D411AA2F009027F44C3F042DFA26@FMSMSX41> From: "Chan, Sun C" To: "'Peng Zhao'" , sgi , "'ho@routefree.com'" , "'lo@routefree.com'" Subject: RE: Question on function Recompute_Addr_Taken() Date: Mon, 9 Apr 2001 08:59:18 -0700 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-pro64-support@oss.sgi.com Precedence: bulk > -----Original Message----- > From: Peng Zhao [mailto:pengzhao@cs.ualberta.ca] > Sent: Friday, April 06, 2001 8:40 PM > To: sgi > Subject: Question on function Recompute_Addr_Taken() > > > Hi, > > I have a question about the function > ipa/local/ipl_summarize_template.h -->Recompute_Addr_Taken(), > and Set_Addr_Taken_Attrib. > > There is some comments like > "// The preopt always set the > ST_addr_saved bit too conservatively. So we need to recompute them." Preopt operates on a per function level. Hence the "conservativeness". Address saved, I think means has address taken on the RHS of an expression and the LHS of the expression is a global, or local that will escape outside of the current function. Wilson, is that right? > > // search for run-time address taken > > I don't know what do they mean and the contex. What is the meaning of > "run-time address taken"? > From owner-pro64-support@oss.sgi.com Mon Apr 9 09:45:38 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f39GjcJ11386 for pro64-support-outgoing; Mon, 9 Apr 2001 09:45:38 -0700 Received: from therouter.routefree.com ([209.21.47.226]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f39GjcM11383 for ; Mon, 9 Apr 2001 09:45:38 -0700 Received: from routefree.com (IDENT:lo@osprey.routefree.com [10.0.0.96]) by therouter.routefree.com (8.9.3/8.9.3) with ESMTP id JAA00912; Mon, 9 Apr 2001 09:40:37 -0700 Message-ID: <3AD1E605.5E0F5237@routefree.com> Date: Mon, 09 Apr 2001 09:40:37 -0700 From: Raymond Lo Organization: RouteFree Inc. X-Mailer: Mozilla 4.73 [en] (X11; U; Linux 2.2.14-5.0 i686) X-Accept-Language: en MIME-Version: 1.0 To: "Chan, Sun C" CC: "'Peng Zhao'" , sgi , "'ho@routefree.com'" Subject: Re: Question on function Recompute_Addr_Taken() References: <9287DC1579B0D411AA2F009027F44C3F042DFA26@FMSMSX41> Content-Type: text/plain; charset=big5 Content-Transfer-Encoding: 7bit Sender: owner-pro64-support@oss.sgi.com Precedence: bulk "Chan, Sun C" wrote: > > -----Original Message----- > > From: Peng Zhao [mailto:pengzhao@cs.ualberta.ca] > > Sent: Friday, April 06, 2001 8:40 PM > > To: sgi > > Subject: Question on function Recompute_Addr_Taken() > > > > > > Hi, > > > > I have a question about the function > > ipa/local/ipl_summarize_template.h -->Recompute_Addr_Taken(), > > and Set_Addr_Taken_Attrib. > > > > There is some comments like > > "// The preopt always set the > > ST_addr_saved bit too conservatively. So we need to recompute them." > > Preopt operates on a per function level. Hence the "conservativeness". > Address saved, I think means has address taken on the RHS of an expression > and the LHS of the expression is a global, or local that will escape outside > of the current function. Wilson, is that right? You're right. IPA has the whole program context, and therefore always compute more accurate information that preopt (which only sees one function at a time). Furthermore, after IPA inlininig, IPA might introduce or remove new address taking operations. That's why IPA recomputes it. > > > > > > // search for run-time address taken > > > > I don't know what do they mean and the contex. What is the meaning of > > "run-time address taken"? > > Not sure! But there are some difference in handling taking address inside an executable statement vs in the static initializer. -Raymond From owner-pro64-support@oss.sgi.com Mon Apr 9 20:02:50 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3A32oQ24759 for pro64-support-outgoing; Mon, 9 Apr 2001 20:02:50 -0700 Received: from hotmail.com (f236.law14.hotmail.com [64.4.21.236]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3A32nM24756 for ; Mon, 9 Apr 2001 20:02:49 -0700 Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Mon, 9 Apr 2001 20:02:44 -0700 Received: from 211.94.129.37 by lw14fd.law14.hotmail.msn.com with HTTP; Tue, 10 Apr 2001 03:02:44 GMT X-Originating-IP: [211.94.129.37] From: "Yanjun HU" To: pro64-support@oss.sgi.com Cc: huyanjun@ccermail.net Subject: Can I build mfef90 through source code? Date: Tue, 10 Apr 2001 11:02:44 +0800 Mime-Version: 1.0 Content-Type: text/plain; format=flowed Message-ID: X-OriginalArrivalTime: 10 Apr 2001 03:02:44.0631 (UTC) FILETIME=[B6F0EA70:01C0C16A] Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Dear Pro64 support team, Sorry to bother you all. I want to install pro64 from source code following the instructions of README.src. However, I CANNOT build Fortran 90 Front-End with my own f90 compiler. I noticed a comment in phase of building mfef90: ˇ°NOTE: mfef90 does not build at this time because it needs an IA-32 version of f90 to compile parts of itself. While SGI has a prototype of such a compiler, it is good enough only to compile what is needed and thus we are not making it available. Mfef90 contains only code developed by SGI.ˇ± Also I have seen the following comment in the Prerequisites: ˇ°We also used g77 for compiling parts of the Fortran 90 front-end. The version that we used was: GNU F77 version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release) (i386-redhat)ˇ± I have some doubts on these two comments. So I want to know the following question: 1. Can I build mfef90 through source code? If yes, on which phase will it be built? 2. Can I use other IA32 version f90 compiler to build Fortran 90 Front-End except the unavailable prototype compiler of SGI? If yes, can you recommend one for me? 3. What is the relationship between GNU F77 version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release) (i386-redhat) and SGI prototype f90 compiler? Thanks in advance! Sincerely yours, Hu Yanjun _________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com. From owner-pro64-support@oss.sgi.com Tue Apr 10 15:14:18 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3AMEIJ18268 for pro64-support-outgoing; Tue, 10 Apr 2001 15:14:18 -0700 Received: from cs.rice.edu (cs.rice.edu [128.42.1.30]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3AMEDM18265 for ; Tue, 10 Apr 2001 15:14:13 -0700 Received: from una.cs.rice.edu (una.cs.rice.edu [128.42.1.160]) by cs.rice.edu (8.9.0/8.9.0) with ESMTP id RAA03775 for ; Tue, 10 Apr 2001 17:14:13 -0500 (CDT) From: Fengmei Zhao Received: (from fzhao@localhost) by una.cs.rice.edu (8.9.0/8.9.0) id RAA06112 for pro64-support@oss.sgi.com; Tue, 10 Apr 2001 17:14:12 -0500 (CDT) Message-Id: <200104102214.RAA06112@una.cs.rice.edu> Subject: whirl2f90 To: pro64-support@oss.sgi.com Date: Tue, 10 Apr 2001 17:14:12 -0500 (CDT) X-Mailer: ELM [version 2.5 PL2] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Could anybody tell me where to get whirl2f90 (it is mentioned in the whirl specification)? Thanks, Fengmei From owner-pro64-support@oss.sgi.com Tue Apr 10 16:31:47 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3ANVln20097 for pro64-support-outgoing; Tue, 10 Apr 2001 16:31:47 -0700 Received: from pneumatic-tube.sgi.com (pneumatic-tube.sgi.com [204.94.214.22]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3ANVjM20094 for ; Tue, 10 Apr 2001 16:31:45 -0700 Received: from rohi.engr.sgi.com (rohi.engr.sgi.com [130.62.180.74]) by pneumatic-tube.sgi.com (980327.SGI.8.8.8-aspam/980310.SGI-aspam) via ESMTP id PAA00692 for ; Tue, 10 Apr 2001 15:33:56 -0700 (PDT) mail_from (mpm@rohi.engr.sgi.com) Received: (from mpm@localhost) by rohi.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id PAA73018; Tue, 10 Apr 2001 15:21:44 -0700 (PDT) Date: Tue, 10 Apr 2001 15:21:44 -0700 (PDT) From: mpm@rohi.engr.sgi.com (Michael Murphy) Message-Id: <200104102221.PAA73018@rohi.engr.sgi.com> To: pro64-support@oss.sgi.com, Fengmei Zhao Subject: Re: whirl2f90 Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Sorry, there is no whirl2f90. That was an idea that never happened. I'll remove it from the whirl document. -- Mike Murphy -- mpm@sgi.com -- quote of the day: -- "Often, the most striking and innovative solutions come from realizing -- that your concept of the problem was wrong." -- (Eric S. Raymond, "The Cathedral and the Bazaar") From owner-pro64-support@oss.sgi.com Sat Apr 14 00:30:24 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3E7UO210018 for pro64-support-outgoing; Sat, 14 Apr 2001 00:30:24 -0700 Received: from hebe.or.intel.com (jffdns02.or.intel.com [134.134.248.4]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3E7UNM10015 for ; Sat, 14 Apr 2001 00:30:23 -0700 Received: from SMTP (orsmsxvs01-1.jf.intel.com [192.168.65.200]) by hebe.or.intel.com (8.9.1a+p1/8.9.1/d: relay.m4,v 1.35 2001/02/12 09:03:45 smothers Exp $) with SMTP id HAA13779 for ; Sat, 14 Apr 2001 07:30:06 GMT Received: from orsmsx28.jf.intel.com ([192.168.70.28]) by 192.168.70.200 (Norton AntiVirus for Internet Email Gateways 1.0) ; Sat, 14 Apr 2001 07:30:06 0000 (GMT) Received: by orsmsx28.jf.intel.com with Internet Mail Service (5.5.2653.19) id <27SQG2F4>; Sat, 14 Apr 2001 00:30:04 -0700 Message-ID: <9D3B898D35A5D411AC7000A0C96B7AEF0145B6D0@orsmsx33.jf.intel.com> From: "Robboy, David G" To: "'pro64-support@oss.sgi.com'" Subject: Bug in sgicc version 0.01.0-13 Date: Fri, 13 Apr 2001 22:51:56 -0700 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Compiling the attachted program with -O3 using sgicc version 0.01.0-13, the function "bar" is optimized down to no code at all. The code should access the volatile field in memory on each iteration of the loop. Is there a later release of sgicc in which this bug is fixed? David Robboy /***** start of program *****/ struct taskQ { long zilch; volatile long foo; } *tasks; void bar(void) { while (!tasks[34].foo) ; } main() { bar(); } From owner-pro64-support@oss.sgi.com Mon Apr 16 14:50:56 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3GLouE06355 for pro64-support-outgoing; Mon, 16 Apr 2001 14:50:56 -0700 Received: from pneumatic-tube.sgi.com (pneumatic-tube.sgi.com [204.94.214.22]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3GLotM06352 for ; Mon, 16 Apr 2001 14:50:55 -0700 Received: from rohi.engr.sgi.com (rohi.engr.sgi.com [130.62.180.74]) by pneumatic-tube.sgi.com (980327.SGI.8.8.8-aspam/980310.SGI-aspam) via ESMTP id PAA06274 for ; Mon, 16 Apr 2001 15:01:19 -0700 (PDT) mail_from (mpm@rohi.engr.sgi.com) Received: (from mpm@localhost) by rohi.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id OAA66376; Mon, 16 Apr 2001 14:49:33 -0700 (PDT) Date: Mon, 16 Apr 2001 14:49:33 -0700 (PDT) From: mpm@rohi.engr.sgi.com (Michael Murphy) Message-Id: <200104162149.OAA66376@rohi.engr.sgi.com> To: "'pro64-support@oss.sgi.com'" , "Robboy, David G" Subject: Re: Bug in sgicc version 0.01.0-13 Sender: owner-pro64-support@oss.sgi.com Precedence: bulk From: "Robboy, David G" To: "'pro64-support@oss.sgi.com'" Subject: Bug in sgicc version 0.01.0-13 Compiling the attachted program with -O3 using sgicc version 0.01.0-13, the function "bar" is optimized down to no code at all. The code should access the volatile field in memory on each iteration of the loop. Is there a later release of sgicc in which this bug is fixed? David Robboy /***** start of program *****/ struct taskQ { long zilch; volatile long foo; } *tasks; void bar(void) { while (!tasks[34].foo) ; } main() { bar(); } This is indeed a bug. It is LNO that removes the loop. I'll file a bug about this, but I'm not sure when it will be fixed. Any LNO experts on the list want to take a look at this? :-) -- Mike Murphy -- mpm@sgi.com -- quote of the day: -- "A man's pride will bring him low, but the humble in spirit -- will retain honor." (Proverbs 29:23) From owner-pro64-support@oss.sgi.com Mon Apr 16 15:02:09 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3GM29p06772 for pro64-support-outgoing; Mon, 16 Apr 2001 15:02:09 -0700 Received: from mail-in.hq.tensilica.com ([38.170.141.29]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3GM28M06769 for ; Mon, 16 Apr 2001 15:02:08 -0700 Received: from tensilica.com (IDENT:maydan@maydan.hq.tensilica.com [192.168.11.50]) by mail-in.hq.tensilica.com (8.9.3/8.9.3) with ESMTP id PAA08778; Mon, 16 Apr 2001 15:02:04 -0700 Message-ID: <3ADB6B31.A6FC03C1@tensilica.com> Date: Mon, 16 Apr 2001 14:59:13 -0700 From: Dror Maydan X-Mailer: Mozilla 4.74 [en] (X11; U; Linux 2.2.14-5.0 i686) X-Accept-Language: en MIME-Version: 1.0 To: Michael Murphy CC: "'pro64-support@oss.sgi.com'" , "Robboy, David G" Subject: Re: Bug in sgicc version 0.01.0-13 References: <200104162149.OAA66376@rohi.engr.sgi.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Michael Murphy wrote: > This is indeed a bug. It is LNO that removes the loop. > I'll file a bug about this, but I'm not sure when it will be fixed. > Any LNO experts on the list want to take a look at this? :-) > > -- Mike Murphy Actually, preopt removes the loop, not LNO. Dror From owner-pro64-support@oss.sgi.com Mon Apr 16 15:30:08 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3GMU8K07645 for pro64-support-outgoing; Mon, 16 Apr 2001 15:30:08 -0700 Received: from mail-in.hq.tensilica.com ([38.170.141.29]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3GMU7M07642 for ; Mon, 16 Apr 2001 15:30:07 -0700 Received: from tensilica.com (IDENT:nenad@eye.hq.tensilica.com [192.168.10.163] (may be forged)) by mail-in.hq.tensilica.com (8.9.3/8.9.3) with ESMTP id PAA29118; Mon, 16 Apr 2001 15:29:57 -0700 Message-ID: <3ADB7261.93D182B0@tensilica.com> Date: Mon, 16 Apr 2001 15:29:53 -0700 From: Nenad Nedeljkovic X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.2.18 i686) X-Accept-Language: en MIME-Version: 1.0 To: Dror Maydan CC: Michael Murphy , "'pro64-support@oss.sgi.com'" , "Robboy, David G" Subject: Re: Bug in sgicc version 0.01.0-13 References: <200104162149.OAA66376@rohi.engr.sgi.com> <3ADB6B31.A6FC03C1@tensilica.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Dror Maydan wrote: > > Michael Murphy wrote: > > > This is indeed a bug. It is LNO that removes the loop. > > I'll file a bug about this, but I'm not sure when it will be fixed. > > Any LNO experts on the list want to take a look at this? :-) > > > > -- Mike Murphy > > Actually, preopt removes the loop, not LNO. > > Dror Actually, it seems that the front end is at fault -- it doesn't mark the type for 'foo' as volatile. If 'foo' is a standalone variable (not a struct member), its type is marked as volatile, and the loop is not removed. Nenad From owner-pro64-support@oss.sgi.com Mon Apr 16 17:16:05 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3H0G5311389 for pro64-support-outgoing; Mon, 16 Apr 2001 17:16:05 -0700 Received: from mail-in.hq.tensilica.com ([38.170.141.29]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3H0G4M11386 for ; Mon, 16 Apr 2001 17:16:04 -0700 Received: from tensilica.com (sys5.hq.tensilica.com [192.168.11.63]) by mail-in.hq.tensilica.com (8.9.3/8.9.3) with ESMTP id RAA22574 for ; Mon, 16 Apr 2001 17:16:04 -0700 Message-ID: <3ADB8B47.28BCA4EC@tensilica.com> Date: Mon, 16 Apr 2001 17:16:07 -0700 From: Sterling Augustine X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.16-3 i686) X-Accept-Language: en MIME-Version: 1.0 To: pro64-support@oss.sgi.com Subject: sgicc and -fvolatile Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Although sgicc accepts -fvolatile, it does not pass it to the front or back ends (diffing the verbose output with and without the option reveals no differences.) Is there a reason for that? Do the front ends actually support -fvolatile, but there is a bug in the driver, or should the driver be warning about an option that has no effect? (Or does anyone actually know what should be happening?) Sterling sterling@tensilica.com Member of Technical Staff Tensilica, Inc. From owner-pro64-support@oss.sgi.com Mon Apr 16 17:41:37 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3H0fbg12009 for pro64-support-outgoing; Mon, 16 Apr 2001 17:41:37 -0700 Received: from sgi.com (sgi.SGI.COM [192.48.153.1]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3H0faM12006 for ; Mon, 16 Apr 2001 17:41:36 -0700 Received: from rohi.engr.sgi.com ([130.62.180.74]) by sgi.com (980327.SGI.8.8.8-aspam/980304.SGI-aspam: SGI does not authorize the use of its proprietary systems or networks for unsolicited or bulk email from the Internet.) via ESMTP id RAA06558 for ; Mon, 16 Apr 2001 17:41:21 -0700 (PDT) mail_from (mpm@rohi.engr.sgi.com) Received: (from mpm@localhost) by rohi.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id RAA68336; Mon, 16 Apr 2001 17:40:04 -0700 (PDT) Date: Mon, 16 Apr 2001 17:40:04 -0700 (PDT) From: mpm@rohi.engr.sgi.com (Michael Murphy) Message-Id: <200104170040.RAA68336@rohi.engr.sgi.com> To: pro64-support@oss.sgi.com, Sterling Augustine Subject: Re: sgicc and -fvolatile Sender: owner-pro64-support@oss.sgi.com Precedence: bulk From: Sterling Augustine Although sgicc accepts -fvolatile, it does not pass it to the front or back ends (diffing the verbose output with and without the option reveals no differences.) Is there a reason for that? Do the front ends actually support -fvolatile, but there is a bug in the driver, or should the driver be warning about an option that has no effect? (Or does anyone actually know what should be happening?) The driver does not pass it to the phases. It could pass it to the front end, but that doesn't seem to have any effect. Actually, even gcc seems to ignore -fvolatile and -fvolatile-global when I try them, but maybe I'm using it wrong? -- Mike Murphy -- mpm@sgi.com -- quote of the day: -- "A man's pride will bring him low, but the humble in spirit -- will retain honor." (Proverbs 29:23) From owner-pro64-support@oss.sgi.com Mon Apr 16 22:00:16 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3H50Gs20735 for pro64-support-outgoing; Mon, 16 Apr 2001 22:00:16 -0700 Received: from scapa.cs.ualberta.ca (root@scapa.cs.ualberta.ca [129.128.4.44]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3H50FM20732 for ; Mon, 16 Apr 2001 22:00:15 -0700 Received: (from localhost user: 'cam' uid#302 fake: STDIN (cam@sakwatamau.cs.ualberta.ca)) by scapa.cs.ualberta.ca id ; Mon, 16 Apr 2001 23:00:01 -0600 Date: Mon, 16 Apr 2001 23:00:00 -0600 (MDT) From: "A. Cameron Macdonell" To: "Chan, Sun C" cc: pro64-support@oss.sgi.com Subject: error in compiling with profiling In-Reply-To: <9287DC1579B0D411AA2F009027F44C3F042DF9F4@FMSMSX41> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-pro64-support@oss.sgi.com Precedence: bulk hi Sun, I implemented some instrumentation by modifying the wn_instrument.cxx file and commenting out the bodies of the instrumentation methods I didn't want. It worked for compiling small C programs consisting of a few files that I created, but when I tried compiling some SPEC2000 benchmarks, specifically mcf and twolf and even equake(1 C file), under NUE I get this error, I can't figure out what it means, it seems that it thinks the structure is invalid. Even when I take out my instrumentation insertions methods, leaving basically empty instrumentation methods I still get the error. ### Assertion failure at line 219 of ../../be/cg/localize.cxx: ### Compiler Error in file pbeampp.c during Localize phase: ### didn't find call BB before bb 1 for tn 9 sgicc INTERNAL ERROR: ../../../pro64/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0//be returned non-zero status 1 make: *** [mcf] Error 1 Thanks, Cam From owner-pro64-support@oss.sgi.com Tue Apr 17 08:18:31 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3HFIVD10945 for pro64-support-outgoing; Tue, 17 Apr 2001 08:18:31 -0700 Received: from mail-in.hq.tensilica.com ([38.170.141.29]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3HFIUM10942 for ; Tue, 17 Apr 2001 08:18:30 -0700 Received: from tensilica.com (sys5.hq.tensilica.com [192.168.11.63]) by mail-in.hq.tensilica.com (8.9.3/8.9.3) with ESMTP id IAA09853; Tue, 17 Apr 2001 08:18:20 -0700 Message-ID: <3ADC5EBF.D37CCC63@tensilica.com> Date: Tue, 17 Apr 2001 08:18:24 -0700 From: Sterling Augustine X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.16-3 i686) X-Accept-Language: en MIME-Version: 1.0 To: Michael Murphy CC: pro64-support@oss.sgi.com Subject: Re: sgicc and -fvolatile References: <200104170040.RAA68336@rohi.engr.sgi.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-pro64-support@oss.sgi.com Precedence: bulk It is supposed to force all memory accesses through pointers to be treated as volatile. Here is a simple testcase: int foo; int main() { int * foo_p = &foo; * foo_p = 1; * foo_p = 2; } if you compile the above with egcs 2.95.2 at -O3, the first indirection off foo_p is optimized away. But if you compile the above at -03 and -fvolatile, then it remains. sgicc -O3 -v -Wf,-fvolatile -S will force -fvolatile to be passed to the front end, but (as unfamiliar as I am with IA64 assembly), it still looks to me like the first line is optimized away. So -fvolatile doesn't appear to work, and should probably result in an unsupported warning. Sterling Michael Murphy wrote: > > From: Sterling Augustine > > Although sgicc accepts -fvolatile, it does not pass it to the front or > back ends (diffing the verbose output with and without the option > reveals no differences.) Is there a reason for that? Do the front ends > actually support -fvolatile, but there is a bug in the driver, or should > the driver be warning about an option that has no effect? (Or does > anyone actually know what should be happening?) > > The driver does not pass it to the phases. > It could pass it to the front end, but that doesn't seem > to have any effect. Actually, even gcc seems to ignore -fvolatile > and -fvolatile-global when I try them, but maybe I'm using it wrong? > -- Mike Murphy > -- mpm@sgi.com > -- quote of the day: > -- "A man's pride will bring him low, but the humble in spirit > -- will retain honor." (Proverbs 29:23) From owner-pro64-support@oss.sgi.com Tue Apr 17 11:24:32 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3HIOWW21213 for pro64-support-outgoing; Tue, 17 Apr 2001 11:24:32 -0700 Received: from yog-sothoth.sgi.com (eugate.sgi.com [192.48.160.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3HIOKM21193 for ; Tue, 17 Apr 2001 11:24:20 -0700 Received: from rohi.engr.sgi.com (rohi.engr.sgi.com [130.62.180.74]) by yog-sothoth.sgi.com (980305.SGI.8.8.8-aspam-6.2/980304.SGI-aspam-europe) via ESMTP id UAA7871226 for ; Tue, 17 Apr 2001 20:24:18 +0200 (CEST) mail_from (mpm@rohi.engr.sgi.com) Received: (from mpm@localhost) by rohi.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id LAA76082; Tue, 17 Apr 2001 11:22:29 -0700 (PDT) Date: Tue, 17 Apr 2001 11:22:29 -0700 (PDT) From: mpm@rohi.engr.sgi.com (Michael Murphy) Message-Id: <200104171822.LAA76082@rohi.engr.sgi.com> To: "Chan, Sun C" , "A. Cameron Macdonell" Subject: Re: error in compiling with profiling Cc: pro64-support@oss.sgi.com Sender: owner-pro64-support@oss.sgi.com Precedence: bulk From: "A. Cameron Macdonell" To: "Chan, Sun C" cc: pro64-support@oss.sgi.com Subject: error in compiling with profiling I implemented some instrumentation by modifying the wn_instrument.cxx file and commenting out the bodies of the instrumentation methods I didn't want. It worked for compiling small C programs consisting of a few files that I created, but when I tried compiling some SPEC2000 benchmarks, specifically mcf and twolf and even equake(1 C file), under NUE I get this error, I can't figure out what it means, it seems that it thinks the structure is invalid. Even when I take out my instrumentation insertions methods, leaving basically empty instrumentation methods I still get the error. ### Assertion failure at line 219 of ../../be/cg/localize.cxx: ### Compiler Error in file pbeampp.c during Localize phase: ### didn't find call BB before bb 1 for tn 9 sgicc INTERNAL ERROR: ../../../pro64/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0//be returned non-zero status 1 make: *** [mcf] Error 1 Thanks, Cam This probably means that you have an incorrect use of a return register. The localize phase is run when GRA is not done, and what it does is insert spills so that there are no global TNs in the PU. As part of this, it checks for uses of physical registers like the parameter and return registers, and makes sure they are used properly. -- Mike Murphy -- mpm@sgi.com -- quote of the day: -- "To laugh often and much; to win the respect of intelligent people and -- the affection of children; to earn the appreciation of honest critics -- and endure the betrayal of false friends; to appreciate beauty; -- to find the best in others; to leave the world a bit better, -- whether by a healthy child, a garden patch or a redeemed social condition; -- to know even one life has breathed easier because you have lived. -- This is to have succeeded." (Ralph Waldo Emerson) From owner-pro64-support@oss.sgi.com Tue Apr 17 12:02:42 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3HJ2gY23220 for pro64-support-outgoing; Tue, 17 Apr 2001 12:02:42 -0700 Received: from deliverator.sgi.com (deliverator.sgi.com [204.94.214.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3HJ2fM23217 for ; Tue, 17 Apr 2001 12:02:41 -0700 Received: from rohi.engr.sgi.com (rohi.engr.sgi.com [130.62.180.74]) by deliverator.sgi.com (980309.SGI.8.8.8-aspam-6.2/980310.SGI-aspam) via ESMTP id MAA29712 for ; Tue, 17 Apr 2001 12:01:09 -0700 (PDT) mail_from (mpm@rohi.engr.sgi.com) Received: (from mpm@localhost) by rohi.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id MAA77267; Tue, 17 Apr 2001 12:00:19 -0700 (PDT) Date: Tue, 17 Apr 2001 12:00:19 -0700 (PDT) From: mpm@rohi.engr.sgi.com (Michael Murphy) Message-Id: <200104171900.MAA77267@rohi.engr.sgi.com> To: Sterling Augustine Subject: Re: sgicc and -fvolatile Cc: pro64-support@oss.sgi.com References: <200104170040.RAA68336@rohi.engr.sgi.com> Sender: owner-pro64-support@oss.sgi.com Precedence: bulk From Sterling@tensilica.com Tue Apr 17 08:18:39 2001 It is supposed to force all memory accesses through pointers to be treated as volatile. Here is a simple testcase: int foo; int main() { int * foo_p = &foo; * foo_p = 1; * foo_p = 2; } if you compile the above with egcs 2.95.2 at -O3, the first indirection off foo_p is optimized away. But if you compile the above at -03 and -fvolatile, then it remains. If you use gcc version 2.96-ia64-000717 snap 001117 which is the latest approved ia64 compiler, -fvolatile does nothing. I'm not sure if 3.0 fixes this, but if not, then this should be reported to gcc. Once gcc supports this, passing -fvolatile to the sgicc front end may still not do anything, depending on how gcc implements this. It might require back-end work. -- Mike Murphy -- mpm@sgi.com -- quote of the day: -- "To laugh often and much; to win the respect of intelligent people and -- the affection of children; to earn the appreciation of honest critics -- and endure the betrayal of false friends; to appreciate beauty; -- to find the best in others; to leave the world a bit better, -- whether by a healthy child, a garden patch or a redeemed social condition; -- to know even one life has breathed easier because you have lived. -- This is to have succeeded." (Ralph Waldo Emerson) From owner-pro64-support@oss.sgi.com Tue Apr 17 14:11:26 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3HLBQ228824 for pro64-support-outgoing; Tue, 17 Apr 2001 14:11:26 -0700 Received: from deliverator.sgi.com (deliverator.sgi.com [204.94.214.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3HLBPM28821 for ; Tue, 17 Apr 2001 14:11:25 -0700 Received: from harpoon.engr.sgi.com (harpoon.engr.sgi.com [130.62.41.49]) by deliverator.sgi.com (980309.SGI.8.8.8-aspam-6.2/980310.SGI-aspam) via ESMTP id OAA19456 for ; Tue, 17 Apr 2001 14:10:08 -0700 (PDT) mail_from (jkingdon@cthulhu.engr.sgi.com) Received: from engr.sgi.com (kingdon.engr.sgi.com [130.62.180.80]) by harpoon.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) via ESMTP id OAA74632; Tue, 17 Apr 2001 14:10:06 -0700 (PDT) Message-ID: <3ADCB132.8F940518@engr.sgi.com> Date: Tue, 17 Apr 2001 14:10:10 -0700 From: Jim Kingdon X-Mailer: Mozilla 4.7C-SGI [en] (X11; I; IRIX 6.5 IP32) X-Accept-Language: en MIME-Version: 1.0 To: Michael Murphy CC: Sterling Augustine , pro64-support@oss.sgi.com Subject: Re: sgicc and -fvolatile References: <200104170040.RAA68336@rohi.engr.sgi.com> <200104171900.MAA77267@rohi.engr.sgi.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Michael Murphy wrote: > int foo; > > int main() > { > int * foo_p = &foo; > > * foo_p = 1; > * foo_p = 2; > } > > if you compile the above with egcs 2.95.2 at -O3, the first indirection > off foo_p is optimized away. But if you compile the above at -03 and > -fvolatile, then it remains. I tried this test case with a couple ia64 GCC's (2.96-ish and 3.1-ish) and -fvolatile had no effect. I was using -O2 in all my tests. Then I tried it on MIPS. GCC 2.95 (from freeware.sgi.com) worked right (-fvolatile made both stores appear). But GCC 2.97 which I built myself had the bug. So this appears to have bitrotted; see http://gcc.gnu.org/ for reporting bugs. > Once gcc supports this, passing -fvolatile to the > sgicc front end may still not do anything, depending on how gcc > implements this. It might require back-end work. Hmm, not sure. I suspect the front-end might require work too (haven't looked too closely). From owner-pro64-support@oss.sgi.com Tue Apr 17 15:43:33 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3HMhXs32265 for pro64-support-outgoing; Tue, 17 Apr 2001 15:43:33 -0700 Received: from hypnos.cps.intel.com (hypnos.cps.intel.com [192.198.165.17]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3HMhWM32262 for ; Tue, 17 Apr 2001 15:43:32 -0700 Received: from SMTP (fmsmsxvs03-1.fm.intel.com [132.233.42.203]) by hypnos.cps.intel.com (8.9.1a+p1/8.9.1/d: relay.m4,v 1.35 2001/02/12 09:03:45 smothers Exp $) with SMTP id WAA01041; Tue, 17 Apr 2001 22:43:01 GMT Received: from fmsmsx19.fm.intel.com ([132.233.48.19]) by 132.233.48.203 (Norton AntiVirus for Internet Email Gateways 1.0) ; Tue, 17 Apr 2001 22:43:01 0000 (GMT) Received: by fmsmsx19.fm.intel.com with Internet Mail Service (5.5.2653.19) id ; Tue, 17 Apr 2001 15:42:59 -0700 Message-ID: <9287DC1579B0D411AA2F009027F44C3F07E5CCB3@FMSMSX41> From: "Chan, Sun C" To: Sterling Augustine , Michael Murphy Cc: pro64-support@oss.sgi.com Subject: RE: sgicc and -fvolatile Date: Tue, 17 Apr 2001 15:42:52 -0700 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-pro64-support@oss.sgi.com Precedence: bulk This feature seemed like a hack to me. Looks like to do it correctly, gcc needs to make all foo_p and all that its possible aliased pointers volatile (I bet there's a bug lurking in the version you claim this feature works.) In the presence of separate compilation, and someone must have the same flag for all .o's compiled, and no way to check against the lack of it in some .o's. For a highly optimized compiler, this approach is just too conservative. Why not use the volatile key word? Compilers, written by non-optimizing people usually ends up implementing something that is just plain crude, this is indeed sad. Sun > -----Original Message----- > From: Sterling Augustine [mailto:Sterling@tensilica.com] > Sent: Tuesday, April 17, 2001 8:18 AM > To: Michael Murphy > Cc: pro64-support@oss.sgi.com > Subject: Re: sgicc and -fvolatile > > > It is supposed to force all memory accesses through pointers to be > treated as volatile. > > Here is a simple testcase: > > int foo; > > int main() > { > int * foo_p = &foo; > > * foo_p = 1; > * foo_p = 2; > } > > if you compile the above with egcs 2.95.2 at -O3, the first > indirection > off foo_p is optimized away. But if you compile the above at -03 and > -fvolatile, then it remains. > > sgicc -O3 -v -Wf,-fvolatile -S > > will force -fvolatile to be passed to the front end, but (as > unfamiliar > as I am with IA64 assembly), it still looks to me like the > first line is > optimized away. So -fvolatile doesn't appear to work, and should > probably result in an unsupported warning. > > Sterling > > > > > > > Michael Murphy wrote: > > > > From: Sterling Augustine > > > > Although sgicc accepts -fvolatile, it does not pass > it to the front or > > back ends (diffing the verbose output with and > without the option > > reveals no differences.) Is there a reason for > that? Do the front ends > > actually support -fvolatile, but there is a bug in > the driver, or should > > the driver be warning about an option that has no > effect? (Or does > > anyone actually know what should be happening?) > > > > The driver does not pass it to the phases. > > It could pass it to the front end, but that doesn't seem > > to have any effect. Actually, even gcc seems to ignore -fvolatile > > and -fvolatile-global when I try them, but maybe I'm using it wrong? > > -- Mike Murphy > > -- mpm@sgi.com > > -- quote of the day: > > -- "A man's pride will bring him low, but the humble in spirit > > -- will retain honor." (Proverbs 29:23) > From owner-pro64-support@oss.sgi.com Wed Apr 18 12:24:14 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3IJOE315763 for pro64-support-outgoing; Wed, 18 Apr 2001 12:24:14 -0700 Received: from scapa.cs.ualberta.ca (root@scapa.cs.ualberta.ca [129.128.4.44]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3IJOEM15759 for ; Wed, 18 Apr 2001 12:24:14 -0700 Received: (from localhost user: 'pengzhao' uid#483 fake: STDIN (pengzhao@peers)) by scapa.cs.ualberta.ca id ; Wed, 18 Apr 2001 13:24:02 -0600 Date: Wed, 18 Apr 2001 13:24:03 -0600 (MDT) From: Peng Zhao To: sgi Subject: question on IPO_SYMTAB::fix_base Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Hi, Can somebody tell me what " IPO_SYMTAB::fix_base" do? I am not clear about the following comments about it. What is the meaning of "different base" and "fix it"? Thanks // ====================================================================== // Walk the ST list and for those that are PU-level static that have // a different base, need to fix it //====================================================================== Regards Peng Peng Zhao pengzhao@cs.ualberta.ca http://www.cs.ualberta.ca/~pengzhao TEL (Lab): (780)492-3725 Lab: CSC251 From owner-pro64-support@oss.sgi.com Wed Apr 18 14:22:52 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3ILMq622806 for pro64-support-outgoing; Wed, 18 Apr 2001 14:22:52 -0700 Received: from scapa.cs.ualberta.ca (root@scapa.cs.ualberta.ca [129.128.4.44]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3ILMqM22803 for ; Wed, 18 Apr 2001 14:22:52 -0700 Received: (from localhost user: 'pengzhao' uid#483 fake: STDIN (pengzhao@peers)) by scapa.cs.ualberta.ca id ; Wed, 18 Apr 2001 15:22:47 -0600 Date: Wed, 18 Apr 2001 15:22:48 -0600 (MDT) From: Peng Zhao To: sgi Subject: question on inliner-->Generate_Barriers Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Hi, Another question is that what Generate_Barriers in IPO_INLINE::Post_Process_Caller does. Is there any papers deals with the implementation issues in function inlining so that I can get asscustomed to the context? Thank you. -- Regards Peng Peng Zhao pengzhao@cs.ualberta.ca http://www.cs.ualberta.ca/~pengzhao TEL (Lab): (780)492-3725 Lab: CSC251 From owner-pro64-support@oss.sgi.com Wed Apr 18 21:57:45 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3J4vje07757 for pro64-support-outgoing; Wed, 18 Apr 2001 21:57:45 -0700 Received: from scapa.cs.ualberta.ca (root@scapa.cs.ualberta.ca [129.128.4.44]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3J4viM07754 for ; Wed, 18 Apr 2001 21:57:44 -0700 Received: (from localhost user: 'mcnaught' uid#11891 fake: STDIN (mcnaught@tees)) by scapa.cs.ualberta.ca id ; Wed, 18 Apr 2001 22:57:25 -0600 Date: Wed, 18 Apr 2001 22:57:26 -0600 (MDT) From: Matthew McNaughton To: pro64-support@oss.sgi.com Subject: Hyperblock bug? Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-pro64-support@oss.sgi.com Precedence: bulk At be/cg/hb_id_candidates.cxx, ~line 627 (version 0.12): if (!BB_SET_MemberP(BB_dom_set(pdom), dom)) { for (dom = Find_Immediate_Dominator(dom); dom && !BB_SET_MemberP(BB_dom_set(pdom), dom); dom = Find_Immediate_Dominator(pdom)); } I think the last call to Find_Immediate_Dominator(pdom) should be Find_Immediate_Dominator(dom). -- Matthew McNaughton From owner-pro64-support@oss.sgi.com Wed Apr 18 22:03:17 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3J53Hr08212 for pro64-support-outgoing; Wed, 18 Apr 2001 22:03:17 -0700 Received: from scapa.cs.ualberta.ca (root@scapa.cs.ualberta.ca [129.128.4.44]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3J53GM08209 for ; Wed, 18 Apr 2001 22:03:16 -0700 Received: (from localhost user: 'mcnaught' uid#11891 fake: STDIN (mcnaught@tees)) by scapa.cs.ualberta.ca id ; Wed, 18 Apr 2001 23:03:03 -0600 Date: Wed, 18 Apr 2001 23:03:05 -0600 (MDT) From: Matthew McNaughton To: pro64-support@oss.sgi.com Subject: archives unavailable. Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-pro64-support@oss.sgi.com Precedence: bulk http://oss.sgi.com/projects/Pro64/support.html refers at bottom of page to archives at http://oss.sgi.com/projects/Pro64/mail/pro64-support/thread.html which is 404. -- Matthew McNaughton From owner-pro64-support@oss.sgi.com Wed Apr 18 23:53:54 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3J6rsO12159 for pro64-support-outgoing; Wed, 18 Apr 2001 23:53:54 -0700 Received: from web13305.mail.yahoo.com (web13305.mail.yahoo.com [216.136.175.41]) by oss.sgi.com (8.11.3/8.11.3) with SMTP id f3J6rrM12156 for ; Wed, 18 Apr 2001 23:53:53 -0700 Message-ID: <20010419065353.16559.qmail@web13305.mail.yahoo.com> Received: from [206.47.244.90] by web13305.mail.yahoo.com; Wed, 18 Apr 2001 23:53:53 PDT Date: Wed, 18 Apr 2001 23:53:53 -0700 (PDT) From: nuf si Subject: Re: archives unavailable To: pro64-support@oss.sgi.com MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: owner-pro64-support@oss.sgi.com Precedence: bulk > http://oss.sgi.com/projects/Pro64/support.html > refers at bottom of page to archives at > http://oss.sgi.com/projects/Pro64/mail/pro64-support/thread.html > which is 404. If you root around in http://oss.sgi.com/projects/Pro64/mail/pro64-support/ you'll see that the actual URL is http://oss.sgi.com/projects/Pro64/mail/pro64-support/threads.html __________________________________________________ Do You Yahoo!? Yahoo! Auctions - buy the things you want at great prices http://auctions.yahoo.com/ From owner-pro64-support@oss.sgi.com Thu Apr 19 07:13:36 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3JEDa711631 for pro64-support-outgoing; Thu, 19 Apr 2001 07:13:36 -0700 Received: from artemis.rus.uni-stuttgart.de (artemis.rus.uni-stuttgart.de [129.69.1.28]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3JEDZM11628 for ; Thu, 19 Apr 2001 07:13:35 -0700 Received: from baracke.rus.uni-stuttgart.de (baracke.rus.uni-stuttgart.de [129.69.58.100]) by artemis.rus.uni-stuttgart.de with ESMTP id QAA01862 for ; Thu, 19 Apr 2001 16:13:33 +0200 (MET DST) env-from (ruschelf@baracke.rus.uni-stuttgart.de) Received: (from ruschelf@localhost) by baracke.rus.uni-stuttgart.de (AIX4.3/UCB 8.8.8/8.8.8) id QAA42016 for pro64-support@oss.sgi.com; Thu, 19 Apr 2001 16:13:33 +0200 Date: Thu, 19 Apr 2001 16:13:33 +0200 From: Clemens Helf To: pro64-support@oss.sgi.com Subject: man pages Message-ID: <20010419161333.T65400@baracke.rus.uni-stuttgart.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Dear all, I am trying to compile a self-written C++ code with your compiler on an ia64-system. After reading the compiler man-page I wanted to look at the opt(5)-pages. but I could not find them. Are these pages part of the installation ? Thanks for your help Clemens Helf helf@hlrs.de From owner-pro64-support@oss.sgi.com Thu Apr 19 08:56:07 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3JFu7k15135 for pro64-support-outgoing; Thu, 19 Apr 2001 08:56:07 -0700 Received: from gate.init.com ([38.138.181.132]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3JFu3M15117 for ; Thu, 19 Apr 2001 08:56:04 -0700 Received: from denali.abinitio.com (localmail.init.com [10.50.30.91]) by gate.init.com (8.9.3/8.9.3/ab-gate-1.1) with ESMTP id LAA25279; Thu, 19 Apr 2001 11:51:04 -0400 Received: from localhost.localdomain (neo.init.com [10.50.20.32]) by denali.abinitio.com (8.9.3/8.9.3/ab-hub-1.2) with ESMTP id LAA25490; Thu, 19 Apr 2001 11:50:53 -0400 (EDT) Received: (from rshapiro@localhost) by localhost.localdomain (8.9.3/8.8.8/ab-cli-1.1) id LAA10319; Thu, 19 Apr 2001 11:50:55 -0400 Date: Thu, 19 Apr 2001 11:50:55 -0400 Message-Id: <200104191550.LAA10319@localhost.localdomain> X-Authentication-Warning: localhost.localdomain: rshapiro set sender to rshapiro@localhost.localdomain using -f From: Richard Shapiro To: mcnaught@cs.ualberta.ca CC: pro64-support@oss.sgi.com In-reply-to: (message from Matthew McNaughton on Wed, 18 Apr 2001 22:57:26 -0600 (MDT)) Subject: Re: Hyperblock bug? References: Sender: owner-pro64-support@oss.sgi.com Precedence: bulk At be/cg/hb_id_candidates.cxx, ~line 627 (version 0.12): if (!BB_SET_MemberP(BB_dom_set(pdom), dom)) { for (dom = Find_Immediate_Dominator(dom); dom && !BB_SET_MemberP(BB_dom_set(pdom), dom); dom = Find_Immediate_Dominator(pdom)); } I think the last call to Find_Immediate_Dominator(pdom) should be Find_Immediate_Dominator(dom). Refresh my memory. Why do you think this? Could you send out a larger chunk of the code (since I don;t have an active development set right now). -- Richard Shapiro Ab Initio Software Corporation rshapiro@abinitio.com (781) 301-2311 From owner-pro64-support@oss.sgi.com Thu Apr 19 09:43:03 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3JGh3x16625 for pro64-support-outgoing; Thu, 19 Apr 2001 09:43:03 -0700 Received: from yog-sothoth.sgi.com (eugate.sgi.com [192.48.160.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3JGh0M16620 for ; Thu, 19 Apr 2001 09:43:01 -0700 Received: from cchkms.engr.sgi.com (cchkms.engr.sgi.com [130.62.180.48]) by yog-sothoth.sgi.com (980305.SGI.8.8.8-aspam-6.2/980304.SGI-aspam-europe) via ESMTP id SAA8065623 for ; Thu, 19 Apr 2001 18:42:58 +0200 (CEST) mail_from (rat@cchkms.engr.sgi.com) Received: (from rat@localhost) by cchkms.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id JAA19658; Thu, 19 Apr 2001 09:41:36 -0700 (PDT) From: "Ross A. Towle" Message-Id: <10104190941.ZM118778@cchkms.engr.sgi.com> Date: Thu, 19 Apr 2001 09:41:35 -0700 In-Reply-To: Clemens Helf "man pages" (Apr 19, 4:13pm) References: <20010419161333.T65400@baracke.rus.uni-stuttgart.de> X-Mailer: Z-Mail (3.2.3 08feb96 MediaMail) To: Clemens Helf , pro64-support@oss.sgi.com Subject: Re: man pages Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Sorry but we have not released the remaining man pages. -Ross From owner-pro64-support@oss.sgi.com Thu Apr 19 09:43:33 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3JGhXe16697 for pro64-support-outgoing; Thu, 19 Apr 2001 09:43:33 -0700 Received: from sgi.com (sgi.SGI.COM [192.48.153.1]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3JGhXM16694 for ; Thu, 19 Apr 2001 09:43:33 -0700 Received: from rohi.engr.sgi.com ([130.62.180.74]) by sgi.com (980327.SGI.8.8.8-aspam/980304.SGI-aspam: SGI does not authorize the use of its proprietary systems or networks for unsolicited or bulk email from the Internet.) via ESMTP id JAA06577 for ; Thu, 19 Apr 2001 09:43:31 -0700 (PDT) mail_from (mpm@rohi.engr.sgi.com) Received: (from mpm@localhost) by rohi.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id JAA87835; Thu, 19 Apr 2001 09:42:13 -0700 (PDT) Date: Thu, 19 Apr 2001 09:42:13 -0700 (PDT) From: mpm@rohi.engr.sgi.com (Michael Murphy) Message-Id: <200104191642.JAA87835@rohi.engr.sgi.com> To: pro64-support@oss.sgi.com, Clemens Helf Subject: Re: man pages Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Unfortunately, the group man pages like opt(5) are not released yet. We're getting them ready now, so they should be in the next release. -- Mike Murphy -- mpm@sgi.com -- quote of the day: -- "Sometimes we prefer to stay busy -- so that we don't have to think about where we are going." From owner-pro64-support@oss.sgi.com Thu Apr 19 09:56:48 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3JGumC17150 for pro64-support-outgoing; Thu, 19 Apr 2001 09:56:48 -0700 Received: from mail-in.hq.tensilica.com (hq.tensilica.com [38.170.141.29]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3JGuiM17147 for ; Thu, 19 Apr 2001 09:56:44 -0700 Received: from gobi ([192.168.11.226]) by mail-in.hq.tensilica.com (8.9.3/8.9.3) with SMTP id JAA18376; Thu, 19 Apr 2001 09:51:39 -0700 From: "Peng Tu" To: "Richard Shapiro" , Cc: Subject: RE: Hyperblock bug? Date: Thu, 19 Apr 2001 09:50:17 -0700 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) In-Reply-To: <200104191550.LAA10319@localhost.localdomain> Importance: Normal X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400 Sender: owner-pro64-support@oss.sgi.com Precedence: bulk The pdom never changes in the for statement. It does look suspicious. -----Original Message----- From: owner-pro64-support@oss.sgi.com [mailto:owner-pro64-support@oss.sgi.com]On Behalf Of Richard Shapiro Sent: Thursday, April 19, 2001 8:51 AM To: mcnaught@cs.ualberta.ca Cc: pro64-support@oss.sgi.com Subject: Re: Hyperblock bug? At be/cg/hb_id_candidates.cxx, ~line 627 (version 0.12): if (!BB_SET_MemberP(BB_dom_set(pdom), dom)) { for (dom = Find_Immediate_Dominator(dom); dom && !BB_SET_MemberP(BB_dom_set(pdom), dom); dom = Find_Immediate_Dominator(pdom)); } I think the last call to Find_Immediate_Dominator(pdom) should be Find_Immediate_Dominator(dom). Refresh my memory. Why do you think this? Could you send out a larger chunk of the code (since I don;t have an active development set right now). -- Richard Shapiro Ab Initio Software Corporation rshapiro@abinitio.com (781) 301-2311 From owner-pro64-support@oss.sgi.com Thu Apr 19 17:09:50 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3K09oh31403 for pro64-support-outgoing; Thu, 19 Apr 2001 17:09:50 -0700 Received: from scapa.cs.ualberta.ca (root@scapa.cs.ualberta.ca [129.128.4.44]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3K09jM31399 for ; Thu, 19 Apr 2001 17:09:45 -0700 Received: (from localhost user: 'mcnaught' uid#11891 fake: STDIN (mcnaught@tees)) by scapa.cs.ualberta.ca id ; Thu, 19 Apr 2001 18:09:32 -0600 Date: Thu, 19 Apr 2001 18:09:34 -0600 (MDT) From: Matthew McNaughton To: Richard Shapiro cc: pro64-support@oss.sgi.com Subject: Re: Hyperblock bug? In-Reply-To: <200104191550.LAA10319@localhost.localdomain> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-pro64-support@oss.sgi.com Precedence: bulk On Thu, 19 Apr 2001, Richard Shapiro wrote: > At be/cg/hb_id_candidates.cxx, ~line 627 (version 0.12): > > if (!BB_SET_MemberP(BB_dom_set(pdom), dom)) { > for (dom = Find_Immediate_Dominator(dom); > dom && !BB_SET_MemberP(BB_dom_set(pdom), dom); > dom = Find_Immediate_Dominator(pdom)); > } > > I think the last call to Find_Immediate_Dominator(pdom) should be > Find_Immediate_Dominator(dom). > > Refresh my memory. Why do you think this? Could you send out a larger chunk > of the code (since I don;t have an active development set right now). I put a copy at http://www.cs.ualberta.ca/~mcnaught/c680/hb_id_candidates.cxx This is in the function Check_Region() The point of this snip of code is that we start with a tentative entry ("dom") to a hammock region, and a tentative exit ("pdom") for the same hammock region, and we are looking upwards in the dominance hierarchy from dom until we find a block that dominates pdom. But in this loop, we may jump downwards in the hierarchy to just above pdom, which undoes a whole bunch of work. It doesn't end up making an illegal candidate region, but it may fail to find the desired one. If you look just below this point you'll see a loop that parallels this one, looking for a post-dominator of our tentative entry point. if (dom && !BB_SET_MemberP(BB_pdom_set(dom), pdom)) { for (pdom = Find_Immediate_Postdominator(pdom); pdom && !BB_SET_MemberP(BB_pdom_set(dom), pdom); pdom = Find_Immediate_Postdominator(pdom)); } I hope this makes it clear. I also found some other minor bugs I'll write up later. -- Matthew McNaughton From owner-pro64-support@oss.sgi.com Thu Apr 19 19:39:10 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3K2dAU04244 for pro64-support-outgoing; Thu, 19 Apr 2001 19:39:10 -0700 Received: from mail.ict.ac.cn ([159.226.39.4]) by oss.sgi.com (8.11.3/8.11.3) with SMTP id f3K2d9M04239 for ; Thu, 19 Apr 2001 19:39:09 -0700 Received: (qmail 20429 invoked from network); 20 Apr 2001 02:34:08 -0000 Received: from unknown (HELO DXQ) (@159.226.40.246) by 159.226.39.4 with SMTP; 20 Apr 2001 02:34:08 -0000 Message-ID: <001601c0c943$b910c650$2a0379c8@DXQ> From: "dxq" To: "A. Cameron Macdonell" , References: Subject: Re: A question on feedback in the Pro64 Date: Fri, 20 Apr 2001 10:43:44 +0800 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.00.2919.6700 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700 Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from base64 to 8bit by oss.sgi.com id f3K2d9M04240 Sender: owner-pro64-support@oss.sgi.com Precedence: bulk If you want to enable dynamic profile which has feedback,you should compile as next steps: 1) sgicc -fb_create feed_back_file matrix.c (do instrumentation work) 2)a.out ( I don't know whether your program read the input file itself, if it can't, you should manually send the input to a.out) 3) sgicc -fb feed_back_file matrix.c (read feedback info and annotation) 4) a.out but the older version Pro64 did't have a lib "libinstr.so", it is said that the newest version Pro64 publish the lib. In driver.cxx , the function WN_Annotate is used to read the feedback and annotation. There are transforms and optimizations, so in hyperblock formation phase, the edge freq maybe is not the true frequency. BTW: In now Pro64 edge does not have not freq info,but has probability info. basic block has freq info and does not have prob info. a beginner. ----- Original Message ----- From: "A. Cameron Macdonell" To: Sent: Wednesday, March 21, 2001 5:44 AM Subject: A question on feedback in the Pro64 > > My question is: do you know if SGI introduced automatic > profiling in the PRo64. Or else any facilities to enable > profile feedback optimization. > > For instance if I have an aplication, say matrix.c, and > a typical input file, say matrix.inp. What > would be the compilation steps that I would have to do > to make sure that the Pro64 hyperblock formation uses > the edge frequency in the CFG from my input file. > > I can find lots of places in the code where it says it will > feedback data if exists but I can't determine how or where > its being outputted or how to trace it. > > A pointer to the right place is as good as a direct answer. > > Thanks, > Cam > > > > > > From owner-pro64-support@oss.sgi.com Thu Apr 19 21:29:39 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3K4TdE08294 for pro64-support-outgoing; Thu, 19 Apr 2001 21:29:39 -0700 Received: from pneumatic-tube.sgi.com (pneumatic-tube.sgi.com [204.94.214.22]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3K4TXM08287 for ; Thu, 19 Apr 2001 21:29:33 -0700 Received: from dsl-proof.corp.sgi.com (dsl-proof.corp.sgi.com [192.102.142.250]) by pneumatic-tube.sgi.com (980327.SGI.8.8.8-aspam/980310.SGI-aspam) via ESMTP id VAA03113 for ; Thu, 19 Apr 2001 21:40:02 -0700 (PDT) mail_from (dlstephe@sgi.com) Received: from sgi.com (localhost [127.0.0.1]) by dsl-proof.corp.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) via ESMTP id VAA77988; Thu, 19 Apr 2001 21:31:57 -0700 (PDT) Message-ID: <3ADFBBBC.11236D1E@sgi.com> Date: Thu, 19 Apr 2001 21:31:56 -0700 From: David Stephenson Organization: SGI -- Compilers X-Mailer: Mozilla 4.7C-SGI [en] (X11; I; IRIX 6.5 IP32) X-Accept-Language: en MIME-Version: 1.0 To: dxq , pro64-support@oss.sgi.com CC: "A. Cameron Macdonell" Subject: Re: A question on feedback in the Pro64 References: <001601c0c943$b910c650$2a0379c8@DXQ> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-pro64-support@oss.sgi.com Precedence: bulk dxq wrote: > If you want to enable dynamic profile which has feedback,you > should compile as next steps: > 1) sgicc -fb_create feed_back_file matrix.c (do > instrumentation work) > 2)a.out ( I don't know whether your program read the input > file itself, if it can't, you should manually send the > input to a.out) > 3) sgicc -fb feed_back_file matrix.c (read feedback info > and annotation) > 4) a.out Use -fb_opt instead of -fb in stage 3. > but the older version Pro64 did't have a lib "libinstr.so", > it is said that the newest version Pro64 publish the lib. Correct. The source for libinstr.so was released with version 0.13 but not 0.12. > In driver.cxx , the function WN_Annotate is used to read > the feedback and annotation. There are transforms and > optimizations, so in hyperblock formation phase, the edge freq > maybe is not the true frequency. To clarify: Both instrumentation and annotation are performed in be just before lowering from VH WHIRL to H WHIRL. At this point, the annotated frequencies are exact counts. After this point, the compiler will continue running the other backend phases, and perform some code transformations. Transformations that affect control flow require that the frequency data be updated to reflect the change. In a few cases (primarily transformation that involve code cloning, such as inlining) the frequencies cannot be exactly updated; educated guesses are made instead. Consequently, when the compiler reaches code generation phase (including hyperblock formation), some of the frequencies may not represent exact known counts from the training run. Limited experimental evidence suggests that this doesn't have much impact on performance. > BTW: In now Pro64 edge does not have not freq info, but has > probability info. basic block has freq info and does not have > prob info. Within the code generator phase, all blocks are assigned frequencies counts (normalized to 1 at the procedure entry) and branches are assigned probabilities. These frequencies are assigned using heuristics if feedback data is not available. Other phases of the compiler only have frequency data available if compiled with -fb_opt. Except in the code generator, frequencies are stored in the FEEDBACK class, using frequency values of type FB_FREQ. - David From owner-pro64-support@oss.sgi.com Thu Apr 19 22:13:43 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3K5DhX09917 for pro64-support-outgoing; Thu, 19 Apr 2001 22:13:43 -0700 Received: from deliverator.sgi.com (deliverator.sgi.com [204.94.214.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3K5DhM09914 for ; Thu, 19 Apr 2001 22:13:43 -0700 Received: from rohi.engr.sgi.com (rohi.engr.sgi.com [130.62.180.74]) by deliverator.sgi.com (980309.SGI.8.8.8-aspam-6.2/980310.SGI-aspam) via ESMTP id WAA16272 for ; Thu, 19 Apr 2001 22:12:25 -0700 (PDT) mail_from (mpm@rohi.engr.sgi.com) Received: (from mpm@localhost) by rohi.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id WAA03568; Thu, 19 Apr 2001 22:10:50 -0700 (PDT) Date: Thu, 19 Apr 2001 22:10:50 -0700 (PDT) From: mpm@rohi.engr.sgi.com (Michael Murphy) Message-Id: <200104200510.WAA03568@rohi.engr.sgi.com> To: Richard Shapiro , Matthew McNaughton Subject: Re: Hyperblock bug? Cc: pro64-support@oss.sgi.com Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Thanks for notifying us of this bug. I am currently testing the fix, and assuming all goes well, I will check in the fix to the baseline. -- Mike Murphy -- mpm@sgi.com -- quote of the day: -- "Sometimes we prefer to stay busy -- so that we don't have to think about where we are going." From owner-pro64-support@oss.sgi.com Fri Apr 20 01:59:04 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3K8x4g15994 for pro64-support-outgoing; Fri, 20 Apr 2001 01:59:04 -0700 Received: from artemis.rus.uni-stuttgart.de (artemis.rus.uni-stuttgart.de [129.69.1.28]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3K8x2M15990 for ; Fri, 20 Apr 2001 01:59:03 -0700 Received: from baracke.rus.uni-stuttgart.de (baracke.rus.uni-stuttgart.de [129.69.58.100]) by artemis.rus.uni-stuttgart.de with ESMTP id KAA14774 for ; Fri, 20 Apr 2001 10:59:01 +0200 (MET DST) env-from (ruschelf@baracke.rus.uni-stuttgart.de) Received: (from ruschelf@localhost) by baracke.rus.uni-stuttgart.de (AIX4.3/UCB 8.8.8/8.8.8) id KAA80006 for pro64-support@oss.sgi.com; Fri, 20 Apr 2001 10:59:01 +0200 Date: Fri, 20 Apr 2001 10:59:01 +0200 From: Clemens Helf To: pro64-support@oss.sgi.com Subject: Linker Error Message-ID: <20010420105901.U65400@baracke.rus.uni-stuttgart.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Dear all, thanks for your reply on missing man-pages. I am still trying to compile my C++ code. When I link my object files into an exceutable, an unresolved symbol error appears. The missing symbol is 'Formal_Arg_StkSeg' and is missing in nearly all object files. Where does this come from ? The problem does not show up with a simple hello-world program. Thanks for your help Clemens Helf helf@hlrs.de From owner-pro64-support@oss.sgi.com Fri Apr 20 04:43:25 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3KBhPB20959 for pro64-support-outgoing; Fri, 20 Apr 2001 04:43:25 -0700 Received: from artemis.rus.uni-stuttgart.de (artemis.rus.uni-stuttgart.de [129.69.1.28]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3KBhOM20955 for ; Fri, 20 Apr 2001 04:43:24 -0700 Received: from baracke.rus.uni-stuttgart.de (baracke.rus.uni-stuttgart.de [129.69.58.100]) by artemis.rus.uni-stuttgart.de with ESMTP id NAA19977 for ; Fri, 20 Apr 2001 13:43:21 +0200 (MET DST) env-from (ruschelf@baracke.rus.uni-stuttgart.de) Received: (from ruschelf@localhost) by baracke.rus.uni-stuttgart.de (AIX4.3/UCB 8.8.8/8.8.8) id NAA36602 for pro64-support@oss.sgi.com; Fri, 20 Apr 2001 13:43:21 +0200 Date: Fri, 20 Apr 2001 13:43:21 +0200 From: Clemens Helf To: pro64-support@oss.sgi.com Subject: comparison gcc - sgiCC Message-ID: <20010420134320.V65400@baracke.rus.uni-stuttgart.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Dear all, I am trying to compile my self-written C++ code on an ia64 system. My source files are located on an nfs-mounted filesystem, because I frequently compile my code on different machines. I am using 'make' to control compilation. A noteable difference between gcc and sgiCC is, that there are frequent nfs failures when using sgiCC, but not with gcc. About every fifth to tenth file fails. Is there any known problem with I/O ? Using similar options, compile time for the whole package differs substantially. With gcc it takes about 6:30 minutes (obtained by time command), with sgiCC it takes several hours. Due to the nfs failures, I did not make a timing. With a local copy of the sources, compile time was 1:48:00 hours (obtained by time command). Is there specific reason for this ? Compiler versions: ------------------ sgiCC -version SGIcc Compilers: Version 0.01.0-13 gcc --version 2.96-ia64-000717 Compile time options: --------------------- sgiCC -c -I. -I -fullwarn -O2 -woff 1155,1174,1375,1468 -g3 g++ -c -I. -I -O -g3 Thanks for your help Clemens Helf helf@hlrs.de From owner-pro64-support@oss.sgi.com Fri Apr 20 16:23:03 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3KNN3e15095 for pro64-support-outgoing; Fri, 20 Apr 2001 16:23:03 -0700 Received: from yog-sothoth.sgi.com (eugate.sgi.com [192.48.160.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3KNN1M15089 for ; Fri, 20 Apr 2001 16:23:02 -0700 Received: from rohi.engr.sgi.com (rohi.engr.sgi.com [130.62.180.74]) by yog-sothoth.sgi.com (980305.SGI.8.8.8-aspam-6.2/980304.SGI-aspam-europe) via ESMTP id BAA5623113 for ; Sat, 21 Apr 2001 01:22:59 +0200 (CEST) mail_from (mpm@rohi.engr.sgi.com) Received: (from mpm@localhost) by rohi.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id QAA06925; Fri, 20 Apr 2001 16:20:10 -0700 (PDT) Date: Fri, 20 Apr 2001 16:20:10 -0700 (PDT) From: mpm@rohi.engr.sgi.com (Michael Murphy) Message-Id: <200104202320.QAA06925@rohi.engr.sgi.com> To: pro64-support@oss.sgi.com, Clemens Helf Subject: Re: Linker Error Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Formal_Arg_StkSeg is an intermediate base symbol for formal parameters. If you see it used in a .o, that implies a data layout error. I'm not aware of any such error currently (are you using the stock sgicc binaries? If so, send me an example preprocessed source and command line, I'll look into it). -- Mike Murphy -- mpm@sgi.com -- quote of the day: -- "When you encourage a person to see himself as a victim of anything - -- crime, poverty, bigotry, bad luck - you are piling bricks on his chest." -- (David Gelernter, injured by mail bomb from Unabomber) From owner-pro64-support@oss.sgi.com Sat Apr 21 08:36:18 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3LFaIJ08298 for pro64-support-outgoing; Sat, 21 Apr 2001 08:36:18 -0700 Received: from pneumatic-tube.sgi.com (pneumatic-tube.sgi.com [204.94.214.22]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3LFaEM08291 for ; Sat, 21 Apr 2001 08:36:15 -0700 Received: from cchkms.engr.sgi.com (cchkms.engr.sgi.com [130.62.180.48]) by pneumatic-tube.sgi.com (980327.SGI.8.8.8-aspam/980310.SGI-aspam) via ESMTP id IAA01061 for ; Sat, 21 Apr 2001 08:46:41 -0700 (PDT) mail_from (rat@cchkms.engr.sgi.com) Received: (from rat@localhost) by cchkms.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id IAA22881; Sat, 21 Apr 2001 08:32:12 -0700 (PDT) Date: Sat, 21 Apr 2001 08:32:12 -0700 (PDT) From: rat@cchkms.engr.sgi.com (Ross A. Towle) Message-Id: <200104211532.IAA22881@cchkms.engr.sgi.com> To: pro64-support@oss.sgi.com, Clemens Helf Subject: Re: comparison gcc - sgiCC Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Remember that the pre-MR versions of the SGI Pro64 compilers run in IA-32 compatability mode. 1. nfs problems. We don't see that problem running TurboLinux. This is not a compiler problem but either a Linux problem or a problem with your system. 2. Compile speed. The SGI Pro64 compilers do suffer by running in IA-32 compatability mode. The truely native versions (undergoing testing) run faster. Also we found a bug that causes some programs to take lots of extra time to compile. If you want, I could tr compiling your sample program with the compilers undergoing testing. If you are interested, contact by email directly (rat@sgi.com). -Ross From owner-pro64-support@oss.sgi.com Tue Apr 24 04:40:09 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3OBe9x12728 for pro64-support-outgoing; Tue, 24 Apr 2001 04:40:09 -0700 Received: from artemis.rus.uni-stuttgart.de (artemis.rus.uni-stuttgart.de [129.69.1.28]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3OBe6M12724 for ; Tue, 24 Apr 2001 04:40:07 -0700 Received: from baracke.rus.uni-stuttgart.de (baracke.rus.uni-stuttgart.de [129.69.58.100]) by artemis.rus.uni-stuttgart.de with ESMTP id NAA05787 for ; Tue, 24 Apr 2001 13:40:02 +0200 (MET DST) env-from (ruschelf@baracke.rus.uni-stuttgart.de) Received: (from ruschelf@localhost) by baracke.rus.uni-stuttgart.de (AIX4.3/UCB 8.8.8/8.8.8) id NAA72944 for pro64-support@oss.sgi.com; Tue, 24 Apr 2001 13:40:02 +0200 Date: Tue, 24 Apr 2001 13:40:02 +0200 From: Clemens Helf To: pro64-support@oss.sgi.com Subject: libstdc++ linkage problem Message-ID: <20010424134002.B65400@baracke.rus.uni-stuttgart.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Dear all, sgiCC does not properly locate its 'libstdc++.a' under all circumstances. Giving the linker option '-L/usr/lib', sgiCC will prefer to link 'libstdc++' from '/usr/lib' instead of from somewhere within the compiler installation, in my case probably 'usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0.new' The superflous mention of '/usr/lib' came from an environment variable, specifying the location of libtcl.so, which was by chance installed in '/usr/lib'. '/usr/lib/ libstdc++.a' is the library for gcc, installed as well. It will be preferable, if the compiler finds the correct library even in this case. I only realized, that I got the wrong library, because an unresolved symbol error occured. But how will you find the problem, if it occurs at runtime ? Clemens Helf helf@hlrs.de ----main.cpp---- #include void main (void) { cout << "hello world !\n"; } -------- compile/link: sgiCC main.cpp -L/usr/lib From owner-pro64-support@oss.sgi.com Tue Apr 24 06:17:26 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3ODHQO16394 for pro64-support-outgoing; Tue, 24 Apr 2001 06:17:26 -0700 Received: from sunshine.math.utah.edu (IDENT:4okVESivfre0KuUFhxhwt9j7PEd2e7PD@sunshine.math.utah.edu [128.110.198.2]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3ODHPM16391 for ; Tue, 24 Apr 2001 06:17:25 -0700 Received: from suncore.math.utah.edu (IDENT:79GXqAPCMhfc/4XzhRlXGs71l3NI9A1q@suncore0.math.utah.edu [128.110.198.5]) by sunshine.math.utah.edu (8.9.3/8.9.3) with ESMTP id HAA01060; Tue, 24 Apr 2001 07:17:23 -0600 (MDT) Received: (from beebe@localhost) by suncore.math.utah.edu (8.9.3/8.9.3) id HAA22617; Tue, 24 Apr 2001 07:17:22 -0600 (MDT) Date: Tue, 24 Apr 2001 07:17:22 -0600 (MDT) From: "Nelson H. F. Beebe" To: pro64-support@oss.sgi.com Cc: beebe@math.utah.edu X-US-Mail: "Center for Scientific Computing, Department of Mathematics, 322 INSCC, University of Utah, 155 S 1400 E RM 233, Salt Lake City, UT 84112-0090, USA" X-Telephone: +1 801 581 5254 X-FAX: +1 801 585 1640, +1 801 581 4148 X-URL: http://www.math.utah.edu/~beebe Subject: pro64-0.01.0-13.ia64 sgicc compiler assertion failure Message-ID: Sender: owner-pro64-support@oss.sgi.com Precedence: bulk I've been testing the sgicc compiler from the pro64-0.01.0-13.ia64.rpm file obtained from http://oss.sgi.com/projects/Pro64/ on the HP IA-64 emulator running on a Red Hat Linux 6.2 installation on an dual-CPU Intel Pentium III 600MHz system. With the struct.c test file from the lcc compiler suite (available in a new autoconfigurized release at ftp://ftp.math.utah.edu/pub/lcc/lcc-4.1.beta.8.tar.gz http://www.math.utah.edu/pub/lcc/lcc-4.1.beta.8.tar.gz ), I get this compiler assertion failure: % sgicc -O2 struct.c ### Assertion failure at line 328 of ../../be/cg/cgexp.cxx: ### Compiler Error in file struct.c during Code_Expansion phase: ### Expand_OP: unexpected opcode OPC_MLOAD sgicc INTERNAL ERROR: /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/be returned non-zero status 1 The code is short enough that I'm including it here: % cat struct.c #line 1 "struct.c" typedef struct point { int x,y; } point; typedef struct rect { point pt1, pt2; } rect; point addpoint(point p1, point p2) { p1.x += p2.x; p1.y += p2.y; return p1; } rect canonrect(rect r) { rect temp; temp.pt1.x =((r.pt1.x) < (r.pt2.x) ? (r.pt1.x) : (r.pt2.x)); temp.pt1.y =((r.pt1.y) < (r.pt2.y) ? (r.pt1.y) : (r.pt2.y)); temp.pt2.x =((r.pt1.x) > (r.pt2.x) ? (r.pt1.x) : (r.pt2.x)); temp.pt2.y =((r.pt1.y) > (r.pt2.y) ? (r.pt1.y) : (r.pt2.y)); return temp; } point makepoint(int x, int y) { point p; p.x = x; p.y = y; return p; } rect makerect(point p1, point p2) { rect r; r.pt1 = p1; r.pt2 = p2; return canonrect(r); } int ptinrect(point p, rect r) { return p.x >= r.pt1.x && p.x < r.pt2.x && p.y >= r.pt1.y && p.y < r.pt2.y; } struct odd {char a[3]; } y = {'a', 'b', 0}; odd(struct odd y) { struct odd x = y; printf("%s\n", x.a); } main() { int i; point x, origin = { 0, 0 }, maxpt = { 320, 320 }; point pts[] = { -1, -1, 1, 1, 20, 300, 500, 400 }; rect screen = makerect(addpoint(maxpt, makepoint(-10, -10)), addpoint(origin, makepoint(10, 10))); for (i = 0; i < sizeof pts/sizeof pts[0]; i++) { printf("(%d,%d) is ", pts[i].x, (x = makepoint(pts[i].x, pts[i].y)).y); if (ptinrect(x, screen) == 0) printf("not "); printf("within [%d,%d; %d,%d]\n", screen.pt1.x, screen.pt1.y, screen.pt2.x, screen.pt2.y); } odd(y); exit(0); } ------------------------------------------------------------------------------- - Nelson H. F. Beebe Tel: +1 801 581 5254 - - Center for Scientific Computing FAX: +1 801 585 1640, +1 801 581 4148 - - University of Utah Internet e-mail: beebe@math.utah.edu - - Department of Mathematics, 322 INSCC beebe@acm.org beebe@computer.org - - 155 S 1400 E RM 233 beebe@ieee.org - - Salt Lake City, UT 84112-0090, USA URL: http://www.math.utah.edu/~beebe - ------------------------------------------------------------------------------- From owner-pro64-support@oss.sgi.com Tue Apr 24 06:47:03 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3ODl3e17764 for pro64-support-outgoing; Tue, 24 Apr 2001 06:47:03 -0700 Received: from sunshine.math.utah.edu (IDENT:yOhOnArhx8rWeLcbhbcSY2AxYCyudh5a@sunshine.math.utah.edu [128.110.198.2]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3ODl2M17760 for ; Tue, 24 Apr 2001 06:47:02 -0700 Received: from suncore.math.utah.edu (IDENT:7Ddl2tvS+xZvCgMEFWuBOQbXrWPGfgbQ@suncore0.math.utah.edu [128.110.198.5]) by sunshine.math.utah.edu (8.9.3/8.9.3) with ESMTP id HAA01225; Tue, 24 Apr 2001 07:46:57 -0600 (MDT) Received: (from beebe@localhost) by suncore.math.utah.edu (8.9.3/8.9.3) id HAA22735; Tue, 24 Apr 2001 07:46:57 -0600 (MDT) Date: Tue, 24 Apr 2001 07:46:57 -0600 (MDT) From: "Nelson H. F. Beebe" To: pro64-support@oss.sgi.com Cc: beebe@math.utah.edu, meyering@math.utah.edu X-US-Mail: "Center for Scientific Computing, Department of Mathematics, 322 INSCC, University of Utah, 155 S 1400 E RM 233, Salt Lake City, UT 84112-0090, USA" X-Telephone: +1 801 581 5254 X-FAX: +1 801 585 1640, +1 801 581 4148 X-URL: http://www.math.utah.edu/~beebe Subject: pro64-0.01.0-13.ia64 sgicc compiler with -O2 produces unresolved function references Message-ID: Sender: owner-pro64-support@oss.sgi.com Precedence: bulk I've been testing the sgicc compiler from the pro64-0.01.0-13.ia64.rpm file obtained from http://oss.sgi.com/projects/Pro64/ on the HP IA-64 emulator running on a Red Hat Linux 6.2 installation on an dual-CPU Intel Pentium III 600MHz system. With several programs compiled with "sgicc -O2", I've gotten link-time errors like this, always to the same symbol, Formal_Arg_StkSeg: array.o(.debug_info+0x1f46): undefined reference to `Formal_Arg_StkSeg' If optimization is reduced to -O1, the link succeeds. I have not been able to reproduce this with a short test program, and indeed, all of the lcc compiler test programs built fine with -O2, except for the assertion failure for struct.c reported in a previous message. Nor did it occur in a build of Brian Kernighan's latest awk release (http://cm.bell-labs.com/who/bwk/awk.tar.gz). However, the bug can be reproduced with the latest GNU fileutils and textutils beta test releases, available at either ftp://alpha.gnu.org/gnu/fetish/fileutils-4.0.45.tar.gz ftp://alpha.gnu.org/gnu/fetish/textutils-2.0.14.tar.gz or ftp://freefriends.org/gnu/fetish/fileutils-4.0.45.tar.gz ftp://freefriends.org/gnu/fetish/textutils-2.0.14.tar.gz Just unpack them, and do env CC=sgicc ./configure && make The first failure I get in fileutils is % sgicc -g -O2 -o chgrp chgrp.o chown-core.o ../lib/libfetish.a chgrp.o(.debug_info+0xe85): undefined reference to `Formal_Arg_StkSeg' chgrp.o(.debug_info+0xe98): undefined reference to `Formal_Arg_StkSeg' chgrp.o(.debug_info+0x10d5): undefined reference to `Formal_Arg_StkSeg' chgrp.o(.debug_info+0x1250): undefined reference to `Formal_Arg_StkSeg' chgrp.o(.debug_info+0x1266): undefined reference to `Formal_Arg_StkSeg' and in textutils, % sgicc -DLOCALEDIR=\"/usr/local/share/locale\" -DHAVE_CONFIG_H -I.. -I. -I../lib -I../intl -g -O2 -c -o cat.o `test -f cat.c || echo './'`cat.c sgicc -g -O2 -o cat cat.o ../lib/libfetish.a cat.o(.debug_info+0xc03): undefined reference to `Formal_Arg_StkSeg' cat.o(.debug_info+0xdb5): undefined reference to `Formal_Arg_StkSeg' cat.o(.debug_info+0xdce): undefined reference to `Formal_Arg_StkSeg' cat.o(.debug_info+0x1055): undefined reference to `Formal_Arg_StkSeg' cat.o(.debug_info+0x106c): undefined reference to `Formal_Arg_StkSeg' ------------------------------------------------------------------------------- - Nelson H. F. Beebe Tel: +1 801 581 5254 - - Center for Scientific Computing FAX: +1 801 585 1640, +1 801 581 4148 - - University of Utah Internet e-mail: beebe@math.utah.edu - - Department of Mathematics, 322 INSCC beebe@acm.org beebe@computer.org - - 155 S 1400 E RM 233 beebe@ieee.org - - Salt Lake City, UT 84112-0090, USA URL: http://www.math.utah.edu/~beebe - ------------------------------------------------------------------------------- From owner-pro64-support@oss.sgi.com Tue Apr 24 07:20:22 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3OEKMZ19701 for pro64-support-outgoing; Tue, 24 Apr 2001 07:20:22 -0700 Received: from mail.eecis.udel.edu (louie.udel.edu [128.175.2.33]) by oss.sgi.com (8.11.3/8.11.3) with SMTP id f3OEKMM19697 for ; Tue, 24 Apr 2001 07:20:22 -0700 Received: from louie.udel.edu by mail.eecis.udel.edu id aa29206; 24 Apr 2001 10:19 EDT Date: Tue, 24 Apr 2001 10:19:39 -0400 (EDT) From: Ziang Hu To: pro64-support@oss.sgi.com MMDF-Warning: Parse error in original version of preceding line at mail.eecis.udel.edu Subject: Where to get gcc-19991114 snapshot ? Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-pro64-support@oss.sgi.com Precedence: bulk In case we don't have CVS account from GNU, whereelse we can get this gcc version? Thanks Ziang From owner-pro64-support@oss.sgi.com Tue Apr 24 07:22:38 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3OEMc419884 for pro64-support-outgoing; Tue, 24 Apr 2001 07:22:38 -0700 Received: from deliverator.sgi.com (deliverator.sgi.com [204.94.214.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3OEMbM19881 for ; Tue, 24 Apr 2001 07:22:37 -0700 Received: from sgihud.hudson.sgi.com (sgihud.hudson.sgi.com [169.238.41.4]) by deliverator.sgi.com (980309.SGI.8.8.8-aspam-6.2/980310.SGI-aspam) via ESMTP id HAA01747 for ; Tue, 24 Apr 2001 07:21:17 -0700 (PDT) mail_from (lesniak@sgihud.hudson.sgi.com) Received: (from lesniak@localhost) by sgihud.hudson.sgi.com (980427.SGI.8.8.8/980728.SGI.AUTOCF) id KAA26366; Tue, 24 Apr 2001 10:19:04 -0400 (EDT) Date: Tue, 24 Apr 2001 10:19:04 -0400 (EDT) From: lesniak@sgihud.hudson.sgi.com (Ken Lesniak) Message-Id: <200104241419.KAA26366@sgihud.hudson.sgi.com> To: pro64-support@oss.sgi.com, "Nelson H. F. Beebe" Subject: Re: pro64-0.01.0-13.ia64 sgicc compiler assertion failure Cc: beebe@math.utah.edu Reply-To: lesniak@sgihud.hudson.sgi.com Sender: owner-pro64-support@oss.sgi.com Precedence: bulk >Date: Tue, 24 Apr 2001 07:17:22 -0600 (MDT) >From: "Nelson H. F. Beebe" >To: pro64-support@oss.sgi.com > >I've been testing the sgicc compiler from the pro64-0.01.0-13.ia64.rpm >file obtained from http://oss.sgi.com/projects/Pro64/ on the >HP IA-64 emulator running on a Red Hat Linux 6.2 installation >on an dual-CPU Intel Pentium III 600MHz system. > >With the struct.c test file from the lcc compiler suite (available >in a new autoconfigurized release at > ftp://ftp.math.utah.edu/pub/lcc/lcc-4.1.beta.8.tar.gz > http://www.math.utah.edu/pub/lcc/lcc-4.1.beta.8.tar.gz >), I get this compiler assertion failure: > >% sgicc -O2 struct.c >### Assertion failure at line 328 of ../../be/cg/cgexp.cxx: >### Compiler Error in file struct.c during Code_Expansion phase: >### Expand_OP: unexpected opcode OPC_MLOAD >sgicc INTERNAL ERROR: /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/be returned >non-zero status 1 >... Thank you for reporting this and including the test case. It is still broken in the current compiler so I have opened an internal bug report on it. Ken From owner-pro64-support@oss.sgi.com Tue Apr 24 10:22:51 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3OHMpI28083 for pro64-support-outgoing; Tue, 24 Apr 2001 10:22:51 -0700 Received: from pneumatic-tube.sgi.com (pneumatic-tube.sgi.com [204.94.214.22]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3OHMoM28080 for ; Tue, 24 Apr 2001 10:22:50 -0700 Received: from rohi.engr.sgi.com (rohi.engr.sgi.com [130.62.180.74]) by pneumatic-tube.sgi.com (980327.SGI.8.8.8-aspam/980310.SGI-aspam) via ESMTP id KAA07916 for ; Tue, 24 Apr 2001 10:33:25 -0700 (PDT) mail_from (mpm@rohi.engr.sgi.com) Received: (from mpm@localhost) by rohi.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id KAA15118; Tue, 24 Apr 2001 10:19:39 -0700 (PDT) Date: Tue, 24 Apr 2001 10:19:39 -0700 (PDT) From: mpm@rohi.engr.sgi.com (Michael Murphy) Message-Id: <200104241719.KAA15118@rohi.engr.sgi.com> To: pro64-support@oss.sgi.com, "Nelson H. F. Beebe" Subject: Re: pro64-0.01.0-13.ia64 sgicc compiler with -O2 produces unresolved function references Cc: meyering@math.utah.edu, beebe@math.utah.edu Sender: owner-pro64-support@oss.sgi.com Precedence: bulk From: "Nelson H. F. Beebe" I've been testing the sgicc compiler from the pro64-0.01.0-13.ia64.rpm file obtained from http://oss.sgi.com/projects/Pro64/ on the HP IA-64 emulator running on a Red Hat Linux 6.2 installation on an dual-CPU Intel Pentium III 600MHz system. With several programs compiled with "sgicc -O2", I've gotten link-time errors like this, always to the same symbol, Formal_Arg_StkSeg: array.o(.debug_info+0x1f46): undefined reference to `Formal_Arg_StkSeg' If optimization is reduced to -O1, the link succeeds. This is the same error that Helf reported a few days ago. It happens when you do -O -g. If you just do -O it won't happen. I have a fix checked in; if you don't want to wait for new binaries, you can apply this patch to the source and build: be/cg/cgdwarf.cxx: *** 708,713 **** --- 708,714 ---- if (ST_is_not_used(st)) return; Base_Symbol_And_Offset (st, &base_st, &base_ofst); + if (ST_is_not_used(base_st)) return; deref = FALSE; if (DST_IS_deref(flag)) /* f90 formals, dope, etc */ -- Mike Murphy -- mpm@sgi.com -- quote of the day: -- "The only thing necessary for the triumph of evil -- is for good men to do nothing." (Edmund Burke, 1795) From owner-pro64-support@oss.sgi.com Wed Apr 25 07:10:54 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3PEAsl02531 for pro64-support-outgoing; Wed, 25 Apr 2001 07:10:54 -0700 Received: from web510.mail.yahoo.com (web510.mail.yahoo.com [216.115.104.225]) by oss.sgi.com (8.11.3/8.11.3) with SMTP id f3PEArM02528 for ; Wed, 25 Apr 2001 07:10:53 -0700 Message-ID: <20010425141053.13338.qmail@web510.mail.yahoo.com> Received: from [63.167.1.26] by web510.mail.yahoo.com; Wed, 25 Apr 2001 07:10:53 PDT Date: Wed, 25 Apr 2001 07:10:53 -0700 (PDT) From: Brian Makin Subject: internal error: To: pro64-support@oss.sgi.com MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: owner-pro64-support@oss.sgi.com Precedence: bulk I have included as much information as I thought would be usefull, I don't have a small testcase which I can share but I could gather some debugging data if anyone has an idea. Linux itanium1 2.4.0test10-001115-58smp #1 SMP Wed Dec 20 14:11:57 PST 2000 ia64 unknown SGIcc Compilers: Version 0.01.0-13 ### Assertion failure at line 3170 of ../../be/cg/ia64/expand.cxx: ### Compiler Error in file file.c during Code_Expansion phase: ### NYI: Exp_COPY inter-class copy for predicate to integer sgicc INTERNAL ERROR: /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/be returned non-zero status 1 Any assistance would be greately appreciated. merimus@yahoo.com __________________________________________________ Do You Yahoo!? Yahoo! Auctions - buy the things you want at great prices http://auctions.yahoo.com/ From owner-pro64-support@oss.sgi.com Thu Apr 26 02:02:23 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3Q92N804665 for pro64-support-outgoing; Thu, 26 Apr 2001 02:02:23 -0700 Received: from artemis.rus.uni-stuttgart.de (artemis.rus.uni-stuttgart.de [129.69.1.28]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3Q92MM04662 for ; Thu, 26 Apr 2001 02:02:22 -0700 Received: from baracke.rus.uni-stuttgart.de (baracke.rus.uni-stuttgart.de [129.69.58.100]) by artemis.rus.uni-stuttgart.de with ESMTP id LAA25586 for ; Thu, 26 Apr 2001 11:02:20 +0200 (MET DST) env-from (ruschelf@baracke.rus.uni-stuttgart.de) Received: (from ruschelf@localhost) by baracke.rus.uni-stuttgart.de (AIX4.3/UCB 8.8.8/8.8.8) id LAA79616 for pro64-support@oss.sgi.com; Thu, 26 Apr 2001 11:02:20 +0200 Date: Thu, 26 Apr 2001 11:02:20 +0200 From: Clemens Helf To: pro64-support@oss.sgi.com Subject: perfomance analysis Message-ID: <20010426110220.E49280@baracke.rus.uni-stuttgart.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Dear all, Is there a profiling tool available, which cooperates with the compiler. I was told, that the current versions of prof are not working on ia64 systems. Clemens Helf helf@hlrs.de From owner-pro64-support@oss.sgi.com Thu Apr 26 02:40:32 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3Q9eWn05792 for pro64-support-outgoing; Thu, 26 Apr 2001 02:40:32 -0700 Received: from artemis.rus.uni-stuttgart.de (artemis.rus.uni-stuttgart.de [129.69.1.28]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3Q9eUM05789 for ; Thu, 26 Apr 2001 02:40:30 -0700 Received: from baracke.rus.uni-stuttgart.de (baracke.rus.uni-stuttgart.de [129.69.58.100]) by artemis.rus.uni-stuttgart.de with ESMTP id LAA04026 for ; Thu, 26 Apr 2001 11:40:29 +0200 (MET DST) env-from (ruschelf@baracke.rus.uni-stuttgart.de) Received: (from ruschelf@localhost) by baracke.rus.uni-stuttgart.de (AIX4.3/UCB 8.8.8/8.8.8) id LAA31362 for pro64-support@oss.sgi.com; Thu, 26 Apr 2001 11:40:29 +0200 Date: Thu, 26 Apr 2001 11:40:29 +0200 From: Clemens Helf To: pro64-support@oss.sgi.com Subject: linker error due to -Ofast Message-ID: <20010426114029.F49280@baracke.rus.uni-stuttgart.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Dear all, After having done a first test case with my code, I wanted see the effects of -Ofast. Especially I do not understand, why the last message appears as I can see the symbols with nm. Clemens Helf helf@hlrs.de At link time, the following errors are issued: sgiCC -I. -I/home/rus/rus/ruschelf/CeqHome/sources/ceq-3d -fullwarn -woff 1155,1174,1375,1468 -Ofast -o ceq1 IniArray2Num.o Num2IniArray.o Ini2Num.o Res2Num.o InFiles.o OutStream.o InOut.o GridManager.o Grid.o GridObj.o NumObjMap.o CopyGrid.o Conn.o ConnOf.o GeomObj.o Reconst.o MLevData.o Objects.o MultiLevelGrid.o check.o check_geom.o check_conn.o init_state.o moments.o neighbour.o DivElem.o DivOf.o HyperPlane.o IspNode.o IspEdge.o InterSectPar.o adapt.o adapt_elem.o adapt_side.o colour.o divide.o divide_bound.o merge.o move.o smooth.o Statistic.o HSDIdraw.o HSDIstate.o toHSDI.o toHSDIprivate.o MethodInfo.o GrpInfo.o SimInfo.o SimMonitor.o Command.o SimParam.o simcontrol.o faenger.o shell.o tclinterp.o ceq.o svd.o wall.o outflow_super.o in_lpc.o inside.o inside_sconv.o inside_euler.o inside_burger.o boundary.o gradient.o grad1_df.o grad1_dx.o limiter.o extrema.o fluxes.o calculate.o multilevel_scheme.o -L../ceq-lib -lceqbase -ltcl8.0 -lm /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/ipa_link: bfd assertion fail ../../cygnus_20000828/bfd/elflink.h:1789 /usr/lib/gcc-lib/ia64-cygnus-linux/2.96-ia64-000717//libstdc++.a(stdstreams.o)(.data+0x100): multiple definition of `clog' /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/ipa_link: Warning: type of symbol `clog' changed from 2 to 1 in stdstreams.o /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/ipa_link: bfd assertion fail ../../cygnus_20000828/bfd/elflink.h:1789 /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/ipa_link: bfd assertion fail ../../cygnus_20000828/bfd/elflink.h:1789 /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/ipa_link: bfd assertion fail ../../cygnus_20000828/bfd/elflink.h:1789 /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/ipa_link: bfd assertion fail ../../cygnus_20000828/bfd/elflink.h:1789 ../ceq-lib/libceqbase.a: could not read symbols: Bad value Output of nm: nm ../ceq-lib/libceqbase.a | more Compress.o: U Cerr U Clog U Cout U Cr U GReconstMethDesc 0000000000000000 W _._10IniArrayId U _._10istrstream U _._10ostrstream U _._12strstreambuf 0000000000000000 W _._13strstreambase U _._22_IO_istream_withassign U _._22_IO_ostream_withassign U _._3ios 0000000000000000 W _._7istream 0000000000000000 W _._7ostream 0000000000000000 A _._8Compress 0000000000000000 W _._8iostream U _._9Tabulator 0000000000000000 W _._9strstream 0000000000000000 W _._t10DynArrayOf1Zi 0000000000000000 W _._t8Array1Of1Zi U _12GeomDataSide.default_comp U _7AscText.flag_ashlar0 U _7AscText.flag_ashlar1 ...contd... From owner-pro64-support@oss.sgi.com Thu Apr 26 06:02:32 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3QD2Wi10920 for pro64-support-outgoing; Thu, 26 Apr 2001 06:02:32 -0700 Received: from sunshine.math.utah.edu (IDENT:wNleKtpYSPldb7hcZlqCD1xXa/1Tw6yJ@sunshine.math.utah.edu [128.110.198.2]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3QD2VM10917 for ; Thu, 26 Apr 2001 06:02:31 -0700 Received: from suncore.math.utah.edu (IDENT:sFo20dY0dE+YdsEzqSCcUJYGn8Sk+JI9@suncore0.math.utah.edu [128.110.198.5]) by sunshine.math.utah.edu (8.9.3/8.9.3) with ESMTP id HAA02334; Thu, 26 Apr 2001 07:02:26 -0600 (MDT) Received: (from beebe@localhost) by suncore.math.utah.edu (8.9.3/8.9.3) id HAA01035; Thu, 26 Apr 2001 07:02:26 -0600 (MDT) Date: Thu, 26 Apr 2001 07:02:26 -0600 (MDT) From: "Nelson H. F. Beebe" To: pro64-support@oss.sgi.com Cc: beebe@math.utah.edu, arnold@math.utah.edu X-US-Mail: "Center for Scientific Computing, Department of Mathematics, 322 INSCC, University of Utah, 155 S 1400 E RM 233, Salt Lake City, UT 84112-0090, USA" X-Telephone: +1 801 581 5254 X-FAX: +1 801 585 1640, +1 801 581 4148 X-URL: http://www.math.utah.edu/~beebe Subject: sgicc, __GNUC__, and -rdynamic Message-ID: Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Arnold Robbins (author and maintainer of GNU gawk) and I have been collaborating on testing of gawk development releases under the NUE IA-64 environment, using both gcc 2.9-ia64-000216, and sgicc 0.01.0-13. While the builds with gcc went as expected, the build with sgicc failed. Some detective work found the cause, and its effect: (1) The configure script tests whether the selected C compiler is GNU gcc, not by name, but rather, by whether the symbol __GNUC__ is known. (2) Based on the outcome of that test, configure adds a gcc-specific flag to LDFLAGS: if test "$GCC" = yes then # Add others here as appropriate, # one day use GNU libtool. if uname | egrep -i linux > /dev/null then LDFLAGS="$LDFLAGS -rdynamic" fi fi The -rdynamic flag is rarely seen, and indeed, even in this week's latest gcc 3.x development snapshot, remains undocumented. gawk uses it to support dynamic loading of modules for language extensions. (3) sgicc uses "gcc -E" as its preprocessor, and this results in __GNUC__ being defined. (4) While gcc recognizes -rdynamic and converts it to flags -export-dynamic -dynamic-linker passed to the linker, collect2, sgicc does not recognize -rdynamic, and linking fails. Because of the rarity of the -rdynamic switch, very few programs would ever uncover this problem. Two fixes are possible: (1) When sgicc invokes "gcc -E" for the preprocessing step, do it as "gcc -E -U__GNUC__". This fixes the gawk build problem, since % env CC='sgicc -U__GNUC__' ./configure produces a Makefile that does not use -rdynamic. (2) Extend sgicc to recognize -rdynamic. Fix (1) is certainly the easiest, and will ensure that other dependencies on the symbol __GNU__ in configure scripts and inside source code will not be foiled by suggesting that a gcc extension is available, when in fact, the host compiler is completely different, and the extension is not there. Fix (2) is less desirable, since leaving __GNUC__ defined can have other side effects, as noted in the previous paragraph. However, it still may be useful to add support for -rdynamic, possibly under a different name, provided __GNUC__ remains undefined. ------------------------------------------------------------------------------- - Nelson H. F. Beebe Tel: +1 801 581 5254 - - Center for Scientific Computing FAX: +1 801 585 1640, +1 801 581 4148 - - University of Utah Internet e-mail: beebe@math.utah.edu - - Department of Mathematics, 322 INSCC beebe@acm.org beebe@computer.org - - 155 S 1400 E RM 233 beebe@ieee.org - - Salt Lake City, UT 84112-0090, USA URL: http://www.math.utah.edu/~beebe - ------------------------------------------------------------------------------- From owner-pro64-support@oss.sgi.com Thu Apr 26 07:30:08 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3QEU8B13530 for pro64-support-outgoing; Thu, 26 Apr 2001 07:30:08 -0700 Received: from skeeve.com (IDENT:root@ads5.ads.active.net.il [192.117.127.213]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3QEU6M13527 for ; Thu, 26 Apr 2001 07:30:07 -0700 Received: (from arnold@localhost) by skeeve.com (8.11.0/8.11.0) id f3QEOEK01223; Thu, 26 Apr 2001 17:24:14 +0300 Date: Thu, 26 Apr 2001 17:24:14 +0300 From: Aharon Robbins Message-Id: <200104261424.f3QEOEK01223@skeeve.com> To: beebe@math.utah.edu, pro64-support@oss.sgi.com Subject: Re: sgicc, __GNUC__, and -rdynamic Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Greetings. > (1) When sgicc invokes "gcc -E" for the preprocessing step, do it as > "gcc -E -U__GNUC__". This fixes the gawk build problem, since > > % env CC='sgicc -U__GNUC__' ./configure > > produces a Makefile that does not use -rdynamic. There may be other GCC-specific symbols that should be undefined as well; it's definitely worth checking the documentation. > (2) Extend sgicc to recognize -rdynamic. > > Fix (1) is certainly the easiest, and will ensure that other > dependencies on the symbol __GNU__ in configure scripts and inside > source code will not be foiled by suggesting that a gcc extension is > available, when in fact, the host compiler is completely different, > and the extension is not there. Nelson understates the case. The -rdynamic flag can't be the only part of GCC that sgicc doesn't support. There are a slew of language extensions that GNU code often makes use of --- properly ifdef-ed, of course --- and defining __GNUC__ is just shouting a request for trouble. Hmmm, I guess I just paraphrased what Nelson said, didn't I? Oh well. :-) > Fix (2) is less desirable, since leaving __GNUC__ defined can have > other side effects, as noted in the previous paragraph. However, it > still may be useful to add support for -rdynamic, possibly under a > different name, provided __GNUC__ remains undefined. The two issues are and should be treated orthogonally. Undefining __GNUC__ for sgicc should be done in any case, whether or not support for -rdynamic is added. Thanks! Arnold Robbins -- Arnold Robbins --- Pioneer Consulting Ltd. arnold@skeeve.com P.O. Box 354 Home Phone: +972 8 979-0381 Fax: +1 603 761-6761 Nof Ayalon Cell Phone: +972 51 297-545 (See www.efax.com) D.N. Shimshon 99785 Laundry increases exponentially in the ISRAEL number of children. -- Miriam Robbins From owner-pro64-support@oss.sgi.com Thu Apr 26 08:57:33 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3QFvXU16105 for pro64-support-outgoing; Thu, 26 Apr 2001 08:57:33 -0700 Received: from palrel2.hp.com (palrel2.hp.com [156.153.255.234]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3QFvWM16102 for ; Thu, 26 Apr 2001 08:57:32 -0700 Received: from hpbs5002.boi.hp.com (hpbs5002.boi.hp.com [15.2.216.28]) by palrel2.hp.com (Postfix) with ESMTP id AED411CE1; Thu, 26 Apr 2001 08:57:31 -0700 (PDT) Received: from xpabh4.corp.hp.com (xpabh4.corp.hp.com [15.58.136.1]) by hpbs5002.boi.hp.com with ESMTP (8.7.1/8.7.3 SMKit7.02) id JAA08277; Thu, 26 Apr 2001 09:57:30 -0600 (MDT) Received: by xpabh4.corp.hp.com with Internet Mail Service (5.5.2653.19) id ; Thu, 26 Apr 2001 08:51:19 -0700 Message-ID: From: "JAIN,SUNEEL (HP-Cupertino,ex1)" To: "'Aharon Robbins'" , "'pro64-support@oss.sgi.com'" Subject: RE: sgicc, __GNUC__, and -rdynamic Date: Thu, 26 Apr 2001 08:51:19 -0700 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-pro64-support@oss.sgi.com Precedence: bulk > > Nelson understates the case. The -rdynamic flag can't be the only part > of GCC that sgicc doesn't support. There are a slew of > language extensions > that GNU code often makes use of --- properly ifdef-ed, of > course --- and > defining __GNUC__ is just shouting a request for trouble. > > Hmmm, I guess I just paraphrased what Nelson said, didn't I? Oh well. > :-) > > > Fix (2) is less desirable, since leaving __GNUC__ defined can have > > other side effects, as noted in the previous paragraph. However, it > > still may be useful to add support for -rdynamic, possibly under a > > different name, provided __GNUC__ remains undefined. > > The two issues are and should be treated orthogonally. > Undefining __GNUC__ > for sgicc should be done in any case, whether or not support > for -rdynamic > is added. > The sgicc and sgiCC compilers share frontends, preprocessor, assembler, linker, debugger with GNU gcc/g++ compilers. They support the same language extensions that the GNU compilers do. One way to view the SGI compilers is that they provide an alternate backend for the IA64 architecture. Given this, I think it is perfectly reasonable for sgicc to define __GNUC__. It should be plug compatible to gcc and adding support for -rdynamic seems like the right answer. - Suneel Jain From owner-pro64-support@oss.sgi.com Thu Apr 26 09:16:31 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3QGGVL16699 for pro64-support-outgoing; Thu, 26 Apr 2001 09:16:31 -0700 Received: from skeeve.com (IDENT:root@ads5.ads.active.net.il [192.117.127.213]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3QGGSM16696 for ; Thu, 26 Apr 2001 09:16:29 -0700 Received: (from arnold@localhost) by skeeve.com (8.11.0/8.11.0) id f3QGASt01348; Thu, 26 Apr 2001 19:10:28 +0300 Date: Thu, 26 Apr 2001 19:10:28 +0300 From: Aharon Robbins Message-Id: <200104261610.f3QGASt01348@skeeve.com> To: arnold@skeeve.com, pro64-support@oss.sgi.com, suneel_jain@hp.com Subject: RE: sgicc, __GNUC__, and -rdynamic Cc: beebe@math.utah.edu Sender: owner-pro64-support@oss.sgi.com Precedence: bulk I stand corrected. That the two compilers share the same front-end wasn't obvious. Since that is the case, I'll agree that adding -rdynamic support is the right way to go. This information then leads one to ask if the IA-64 back end will eventually be merged into the main GCC code base? And in the meantime, for IA-64, why not just make sgicc be gcc? Thanks! Arnold > From: "JAIN,SUNEEL (HP-Cupertino,ex1)" > To: "'Aharon Robbins'" , > "'pro64-support@oss.sgi.com'" > Subject: RE: sgicc, __GNUC__, and -rdynamic > Date: Thu, 26 Apr 2001 08:51:19 -0700 > > > Nelson understates the case. The -rdynamic flag can't be the only part > > of GCC that sgicc doesn't support. There are a slew of > > language extensions > > that GNU code often makes use of --- properly ifdef-ed, of > > course --- and > > defining __GNUC__ is just shouting a request for trouble. > > > > Hmmm, I guess I just paraphrased what Nelson said, didn't I? Oh well. > >:-) > > > > > Fix (2) is less desirable, since leaving __GNUC__ defined can have > > > other side effects, as noted in the previous paragraph. However, it > > > still may be useful to add support for -rdynamic, possibly under a > > > different name, provided __GNUC__ remains undefined. > > > > The two issues are and should be treated orthogonally. > > Undefining __GNUC__ > > for sgicc should be done in any case, whether or not support > > for -rdynamic > > is added. > > The sgicc and sgiCC compilers share frontends, preprocessor, > assembler, linker, debugger with GNU gcc/g++ compilers. They support > the same language extensions that the GNU compilers do. One way to > view the SGI compilers is that they provide an alternate > backend for the IA64 architecture. > > Given this, I think it is perfectly reasonable for sgicc to > define __GNUC__. It should be plug compatible to gcc and > adding support for -rdynamic seems like the right answer. > > - Suneel Jain From owner-pro64-support@oss.sgi.com Thu Apr 26 10:34:10 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3QHYAu19569 for pro64-support-outgoing; Thu, 26 Apr 2001 10:34:10 -0700 Received: from deliverator.sgi.com (deliverator.sgi.com [204.94.214.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3QHY9M19566 for ; Thu, 26 Apr 2001 10:34:09 -0700 Received: from rohi.engr.sgi.com (rohi.engr.sgi.com [130.62.180.74]) by deliverator.sgi.com (980309.SGI.8.8.8-aspam-6.2/980310.SGI-aspam) via ESMTP id KAA13177 for ; Thu, 26 Apr 2001 10:32:50 -0700 (PDT) mail_from (mpm@rohi.engr.sgi.com) Received: (from mpm@localhost) by rohi.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id KAA23247; Thu, 26 Apr 2001 10:30:50 -0700 (PDT) Date: Thu, 26 Apr 2001 10:30:50 -0700 (PDT) From: mpm@rohi.engr.sgi.com (Michael Murphy) Message-Id: <200104261730.KAA23247@rohi.engr.sgi.com> To: suneel_jain@hp.com, pro64-support@oss.sgi.com, arnold@skeeve.com, Aharon Robbins Subject: RE: sgicc, __GNUC__, and -rdynamic Cc: beebe@math.utah.edu Sender: owner-pro64-support@oss.sgi.com Precedence: bulk From: Aharon Robbins I stand corrected. That the two compilers share the same front-end wasn't obvious. Since that is the case, I'll agree that adding -rdynamic support is the right way to go. As Suneel explained, I think we should define __GNUC__. It looks like all gcc does with -rdynamic is to pass -export-dynamic to collect2. Is that correct? One question is why not just pass -export-dynamic instead of -rdynamic? sgicc currently doesn't handle either one, but I can add that easily enough. In the meantime you could use -Wl,-export-dynamic. This information then leads one to ask if the IA-64 back end will eventually be merged into the main GCC code base? And in the meantime, for IA-64, why not just make sgicc be gcc? There is a separate ia64 gcc, that uses the gcc back-end. GCC (at least the people we talked to), is not interested in using our backend. So we have two competing compilers. Thanks! Arnold > From: "JAIN,SUNEEL (HP-Cupertino,ex1)" > To: "'Aharon Robbins'" , > "'pro64-support@oss.sgi.com'" > Subject: RE: sgicc, __GNUC__, and -rdynamic > Date: Thu, 26 Apr 2001 08:51:19 -0700 > > > Nelson understates the case. The -rdynamic flag can't be the only part > > of GCC that sgicc doesn't support. There are a slew of > > language extensions > > that GNU code often makes use of --- properly ifdef-ed, of > > course --- and > > defining __GNUC__ is just shouting a request for trouble. > > > > Hmmm, I guess I just paraphrased what Nelson said, didn't I? Oh well. > >:-) > > > > > Fix (2) is less desirable, since leaving __GNUC__ defined can have > > > other side effects, as noted in the previous paragraph. However, it > > > still may be useful to add support for -rdynamic, possibly under a > > > different name, provided __GNUC__ remains undefined. > > > > The two issues are and should be treated orthogonally. > > Undefining __GNUC__ > > for sgicc should be done in any case, whether or not support > > for -rdynamic > > is added. > > The sgicc and sgiCC compilers share frontends, preprocessor, > assembler, linker, debugger with GNU gcc/g++ compilers. They support > the same language extensions that the GNU compilers do. One way to > view the SGI compilers is that they provide an alternate > backend for the IA64 architecture. > > Given this, I think it is perfectly reasonable for sgicc to > define __GNUC__. It should be plug compatible to gcc and > adding support for -rdynamic seems like the right answer. > > - Suneel Jain -- Mike Murphy -- mpm@sgi.com -- quote of the day: -- "Time is the most valuable coin in your life. You and you alone will -- determine how that coin will be spent. Be careful that you do not let -- other people spend it for you." (Carl Sandburg) From owner-pro64-support@oss.sgi.com Thu Apr 26 10:39:04 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3QHd4p19798 for pro64-support-outgoing; Thu, 26 Apr 2001 10:39:04 -0700 Received: from rumor.cps.intel.com (rumor.cps.intel.com [192.102.198.242]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3QHd2M19795 for ; Thu, 26 Apr 2001 10:39:02 -0700 Received: from SMTP (fmsmsxvs05-1.fm.intel.com [132.233.42.205]) by rumor.cps.intel.com (8.9.1a+p1/8.9.1/d: relay.m4,v 1.36 2001/04/18 16:16:02 root Exp $) with SMTP id RAA26182; Thu, 26 Apr 2001 17:38:23 GMT Received: from fmsmsx28.fm.intel.com ([132.233.48.28]) by 132.233.48.205 (Norton AntiVirus for Internet Email Gateways 1.0) ; Thu, 26 Apr 2001 17:38:26 0000 (GMT) Received: by fmsmsx28.fm.intel.com with Internet Mail Service (5.5.2653.19) id ; Thu, 26 Apr 2001 10:38:24 -0700 Message-ID: <9287DC1579B0D411AA2F009027F44C3F042DFA49@FMSMSX41> From: "Chan, Sun C" To: "'mpm@rohi.engr.sgi.com'" , suneel_jain@hp.com, pro64-support@oss.sgi.com, arnold@skeeve.com, Aharon Robbins Cc: beebe@math.utah.edu Subject: RE: sgicc, __GNUC__, and -rdynamic Date: Thu, 26 Apr 2001 10:38:21 -0700 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Can someone give a more detailed explanation of what -rdynamic does exactly? As far as I can see in the explanation so far, it's just a flag passed down to collect2 which is the linker? Does that require the compiler to behavior different in terms of code generations, such as "stubs" generation for function calls etc? How will that different from the ABI as specified by the Intel documents? Sun > -----Original Message----- > From: mpm@rohi.engr.sgi.com [mailto:mpm@rohi.engr.sgi.com] > Sent: Thursday, April 26, 2001 10:31 AM > To: suneel_jain@hp.com; pro64-support@oss.sgi.com; arnold@skeeve.com; > Aharon Robbins > Cc: beebe@math.utah.edu > Subject: RE: sgicc, __GNUC__, and -rdynamic > > > From: Aharon Robbins > > I stand corrected. That the two compilers share the > same front-end > wasn't obvious. Since that is the case, I'll agree that adding > -rdynamic support is the right way to go. > > As Suneel explained, I think we should define __GNUC__. > It looks like all gcc does with -rdynamic is to pass -export-dynamic > to collect2. Is that correct? One question is why not just pass > -export-dynamic instead of -rdynamic? sgicc currently doesn't handle > either one, but I can add that easily enough. In the meantime you > could use -Wl,-export-dynamic. > > This information then leads one to ask if the IA-64 > back end will > eventually be merged into the main GCC code base? And in the > meantime, for IA-64, why not just make sgicc be gcc? > > There is a separate ia64 gcc, that uses the gcc back-end. > GCC (at least the people we talked to), is not interested in using > our backend. So we have two competing compilers. > > Thanks! > > Arnold > > > From: "JAIN,SUNEEL (HP-Cupertino,ex1)" > > To: "'Aharon Robbins'" , > > "'pro64-support@oss.sgi.com'" > > > Subject: RE: sgicc, __GNUC__, and -rdynamic > > Date: Thu, 26 Apr 2001 08:51:19 -0700 > > > > > Nelson understates the case. The -rdynamic flag > can't be the only part > > > of GCC that sgicc doesn't support. There are a slew of > > > language extensions > > > that GNU code often makes use of --- properly ifdef-ed, of > > > course --- and > > > defining __GNUC__ is just shouting a request for trouble. > > > > > > Hmmm, I guess I just paraphrased what Nelson said, > didn't I? Oh well. > > >:-) > > > > > > > Fix (2) is less desirable, since leaving __GNUC__ > defined can have > > > > other side effects, as noted in the previous > paragraph. However, it > > > > still may be useful to add support for -rdynamic, > possibly under a > > > > different name, provided __GNUC__ remains undefined. > > > > > > The two issues are and should be treated orthogonally. > > > Undefining __GNUC__ > > > for sgicc should be done in any case, whether or > not support > > > for -rdynamic > > > is added. > > > > The sgicc and sgiCC compilers share frontends, preprocessor, > > assembler, linker, debugger with GNU gcc/g++ > compilers. They support > > the same language extensions that the GNU compilers > do. One way to > > view the SGI compilers is that they provide an alternate > > backend for the IA64 architecture. > > > > Given this, I think it is perfectly reasonable for sgicc to > > define __GNUC__. It should be plug compatible to gcc and > > adding support for -rdynamic seems like the right answer. > > > > - Suneel Jain > > -- Mike Murphy > -- mpm@sgi.com > -- quote of the day: > -- "Time is the most valuable coin in your life. You and > you alone will > -- determine how that coin will be spent. Be careful that > you do not let > -- other people spend it for you." (Carl Sandburg) > From owner-pro64-support@oss.sgi.com Thu Apr 26 11:01:18 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3QI1IN20582 for pro64-support-outgoing; Thu, 26 Apr 2001 11:01:18 -0700 Received: from sunshine.math.utah.edu (IDENT:89WF8KLAVEtkwynv/IGIdeAx/B5ZJv0p@sunshine.math.utah.edu [128.110.198.2]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3QI1HM20579 for ; Thu, 26 Apr 2001 11:01:17 -0700 Received: from suncore.math.utah.edu (IDENT:ud+kzwWp8mks1wkD8ncv83jP+3AIbVdi@suncore0.math.utah.edu [128.110.198.5]) by sunshine.math.utah.edu (8.9.3/8.9.3) with ESMTP id MAA06677; Thu, 26 Apr 2001 12:01:08 -0600 (MDT) Received: (from beebe@localhost) by suncore.math.utah.edu (8.9.3/8.9.3) id MAA18488; Thu, 26 Apr 2001 12:01:07 -0600 (MDT) Date: Thu, 26 Apr 2001 12:01:07 -0600 (MDT) From: "Nelson H. F. Beebe" To: "Chan, Sun C" Cc: beebe@math.utah.edu, "'mpm@rohi.engr.sgi.com'" , suneel_jain@hp.com, pro64-support@oss.sgi.com, arnold@skeeve.com, Aharon Robbins , beebe@math.utah.edu X-US-Mail: "Center for Scientific Computing, Department of Mathematics, 322 INSCC, University of Utah, 155 S 1400 E RM 233, Salt Lake City, UT 84112-0090, USA" X-Telephone: +1 801 581 5254 X-FAX: +1 801 585 1640, +1 801 581 4148 X-URL: http://www.math.utah.edu/~beebe Subject: RE: sgicc, __GNUC__, and -rdynamic In-Reply-To: Your message of Thu, 26 Apr 2001 10:38:21 -0700 Message-ID: Sender: owner-pro64-support@oss.sgi.com Precedence: bulk >> Can someone give a more detailed explanation of what -rdynamic >> does exactly? I'm unclear of its exact function at this point. A scan of the gcc-2.95.2, gcc-2.95.3, and Monday's gcc-20010423 snapshot of the 3.0 development info files, man pages, and --help output revealed no mention of -rdynamic, so this morning, I filed a bug report on its lack of documentation to gcc-bugs@gcc.gnu.org. A scan of the gcc-2.95.3 source tree turns up these only these references, which all look like a simple remap of the switch to -export-dynamic passed to the linker, collect2. % grep rdynamic `find . -type f` ./gcc/config/alpha/elf.h: %{rdynamic:-export-dynamic} \ ./gcc/config/arm/linux-elf.h: %{rdynamic:-export-dynamic} \ ./gcc/config/i386/freebsd-elf.h: %{rdynamic:-export-dynamic} \ ./gcc/config/i386/gnu.h: %{rdynamic:-export-dynamic} \ ./gcc/config/i386/linux.h: %{rdynamic:-export-dynamic} \ ./gcc/config/i386/linux.h: %{rdynamic:-export-dynamic} \ ./gcc/config/i386/linux.h: %{rdynamic:-export-dynamic} \ ./gcc/config/m68k/linux.h: %{rdynamic:-export-dynamic} \ ./gcc/config/m68k/linux.h: %{rdynamic:-export-dynamic} \ ./gcc/config/m68k/linux.h: %{rdynamic:-export-dynamic} \ ./gcc/config/mips/linux.h: %{rdynamic:-export-dynamic} \ ./gcc/config/rs6000/linux.h: %{rdynamic:-export-dynamic} \ ./gcc/config/sparc/linux.h: %{rdynamic:-export-dynamic} \ ./gcc/config/sparc/linux.h: %{rdynamic:-export-dynamic} \ ./gcc/config/sparc/linux.h: %{rdynamic:-export-dynamic} \ ./gcc/config/sparc/linux64.h: %{rdynamic:-export-dynamic} \ ./gcc/config/sparc/linux64.h: %{rdynamic:-export-dynamic} \ ./gcc/config/sparc/linux64.h: %{rdynamic:-export-dynamic} \ Another grep for export-dynamic showed the above, plus these: ./ltconfig: export_dynamic_flag_spec='${wl}--export-dynamic' ./ltmain.sh: -export-dynamic) ./ltmain.sh: # If -module or -export-dynamic was specified, set the dlname. ./ltmain.sh: $echo "$modename: error: \`-dlopen self' requires \`-export-dynamic'" 1>&2 ./ltmain.sh: test -n "$library_names" && $echo "$modename: warning: \`$file' was not linked with \`-export-dynamic'" ./ltmain.sh: -export-dynamic allow symbols from OUTPUT-FILE to be resolved with dlsym(3) In particular, I see no evidence of the -rdynamic flag affecting code generation. ------------------------------------------------------------------------------- - Nelson H. F. Beebe Tel: +1 801 581 5254 - - Center for Scientific Computing FAX: +1 801 585 1640, +1 801 581 4148 - - University of Utah Internet e-mail: beebe@math.utah.edu - - Department of Mathematics, 322 INSCC beebe@acm.org beebe@computer.org - - 155 S 1400 E RM 233 beebe@ieee.org - - Salt Lake City, UT 84112-0090, USA URL: http://www.math.utah.edu/~beebe - ------------------------------------------------------------------------------- From owner-pro64-support@oss.sgi.com Thu Apr 26 11:22:57 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3QIMv221160 for pro64-support-outgoing; Thu, 26 Apr 2001 11:22:57 -0700 Received: from deliverator.sgi.com (deliverator.sgi.com [204.94.214.10]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3QIMqM21157 for ; Thu, 26 Apr 2001 11:22:52 -0700 Received: from johmar.engr.sgi.com (johmar.engr.sgi.com [130.62.180.56]) by deliverator.sgi.com (980309.SGI.8.8.8-aspam-6.2/980310.SGI-aspam) via ESMTP id LAA21131 for ; Thu, 26 Apr 2001 11:21:32 -0700 (PDT) mail_from (jcarter@johmar.engr.sgi.com) Received: (from jcarter@localhost) by johmar.engr.sgi.com (SGI-8.9.3/8.9.3) id LAA90671; Thu, 26 Apr 2001 11:21:29 -0700 (PDT) Date: Thu, 26 Apr 2001 11:21:29 -0700 (PDT) From: Jack Carter Message-Id: <10104261121.ZM90589@johmar.engr.sgi.com> In-Reply-To: "Chan, Sun C" "RE: sgicc, __GNUC__, and -rdynamic" (Apr 26, 10:38am) References: <9287DC1579B0D411AA2F009027F44C3F042DFA49@FMSMSX41> X-Mailer: Z-Mail-SGI (3.2S.3 08feb96 MediaMail) To: "Chan, Sun C" , "'mpm@rohi.engr.sgi.com '" , suneel_jain@hp.com, pro64-support@oss.sgi.com, arnold@skeeve.com Subject: Re: sgicc, __GNUC__, and -rdynamic Cc: beebe@math.utah.edu Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Collect2 is not the linker and does not use linker code. It is a prelinker and resides in the gcc source tree, not the bfd or ld trees. This is a mistake based on the difficulty of using bfd in my humble opinion. Jack From owner-pro64-support@oss.sgi.com Thu Apr 26 11:29:57 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3QITv721585 for pro64-support-outgoing; Thu, 26 Apr 2001 11:29:57 -0700 Received: from Cantor.suse.de (ns.suse.de [213.95.15.193]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3QITrM21581 for ; Thu, 26 Apr 2001 11:29:54 -0700 Received: from Hermes.suse.de (Hermes.suse.de [213.95.15.136]) by Cantor.suse.de (Postfix) with ESMTP id 07DB91E27C; Thu, 26 Apr 2001 20:29:53 +0200 (MEST) Date: Thu, 26 Apr 2001 20:29:44 +0200 From: Andi Kleen To: Jack Carter Cc: "Chan, Sun C" , "'mpm@rohi.engr.sgi.com '" , suneel_jain@hp.com, pro64-support@oss.sgi.com, arnold@skeeve.com, beebe@math.utah.edu Subject: Re: sgicc, __GNUC__, and -rdynamic Message-ID: <20010426202944.A26882@gruyere.muc.suse.de> References: <9287DC1579B0D411AA2F009027F44C3F042DFA49@FMSMSX41> <10104261121.ZM90589@johmar.engr.sgi.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: <10104261121.ZM90589@johmar.engr.sgi.com>; from jcarter@johmar.engr.sgi.com on Thu, Apr 26, 2001 at 11:21:29AM -0700 Sender: owner-pro64-support@oss.sgi.com Precedence: bulk On Thu, Apr 26, 2001 at 11:21:29AM -0700, Jack Carter wrote: > Collect2 is not the linker and does not use linker > code. It is a prelinker and resides in the gcc source > tree, not the bfd or ld trees. > > This is a mistake based on the difficulty of using bfd > in my humble opinion. collect2 is also used for non GNU linkers, which gcc uses/supports on some platforms. It also used to do the C++ template generation, which probably does not belong in the linker. -Andi From owner-pro64-support@oss.sgi.com Thu Apr 26 18:37:06 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3R1b6032623 for pro64-support-outgoing; Thu, 26 Apr 2001 18:37:06 -0700 Received: from hotmail.com (f213.law14.hotmail.com [64.4.21.213]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3R1b4M32620; Thu, 26 Apr 2001 18:37:04 -0700 Received: from mail pickup service by hotmail.com with Microsoft SMTPSVC; Thu, 26 Apr 2001 18:36:48 -0700 Received: from 211.94.129.37 by lw14fd.law14.hotmail.msn.com with HTTP; Fri, 27 Apr 2001 01:36:48 GMT X-Originating-IP: [211.94.129.37] From: "Yanjun HU" To: pro64-support@oss.sgi.com Cc: owner-pro64-support@oss.sgi.com Subject: Call_Graph? Date: Fri, 27 Apr 2001 09:36:48 +0800 Mime-Version: 1.0 Content-Type: text/plain; format=flowed Message-ID: X-OriginalArrivalTime: 27 Apr 2001 01:36:48.0852 (UTC) FILETIME=[86E13940:01C0CEBA] Sender: owner-pro64-support@oss.sgi.com Precedence: bulk Dear Pro64 support team, Can I use some options to display the Call_Graph using daVinci, just like using "-S -O3 -Wb,-tt57:0x100" to get the Control_Flow_Graph? Anymore, except /ipa/inline/inline.cxx /ipa/common/ip_graph.h, any other source code about Call_Graph will be useful for me? Thanks! Regards, Hu Yanjun _________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com. From owner-pro64-support@oss.sgi.com Thu Apr 26 19:29:11 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3R2TBq01479 for pro64-support-outgoing; Thu, 26 Apr 2001 19:29:11 -0700 Received: from mail.ict.ac.cn ([159.226.39.4]) by oss.sgi.com (8.11.3/8.11.3) with SMTP id f3R2TAM01476 for ; Thu, 26 Apr 2001 19:29:10 -0700 Received: (qmail 1696 invoked from network); 27 Apr 2001 02:23:55 -0000 Received: from unknown (HELO zsk) (@159.226.40.246) by 159.226.39.4 with SMTP; 27 Apr 2001 02:23:55 -0000 Message-ID: <006701c0cec2$2af32910$280379c8@zsk> From: "Shukang ZHOU" To: "Yanjun HU" , References: Subject: Re: Call_Graph? Date: Fri, 27 Apr 2001 10:31:29 +0800 MIME-Version: 1.0 Content-Type: text/plain; charset="gb2312" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.00.2919.6700 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700 Sender: owner-pro64-support@oss.sgi.com Precedence: bulk As far as I know, sgicc cannot display call graph by daVinci. In addition, I have a question: what's the option of display WHIRL? There is a function named dV_view_whirl(WN *wn, const char *title = NULL, BOOL show_expr = FALSE, FILE *trace_fp = NULL) in osprey1.0/be/com/com_whirlview.h. Is it usable? Shukang ZHOU Advanced Compiler Technology Group Institute of Computing Technology, Chinese Academy of Sciences E-mail: zhshk@ict.ac.cn ----- Original Message ----- From: "Yanjun HU" To: Cc: Sent: Friday, April 27, 2001 9:36 AM Subject: Call_Graph? > Dear Pro64 support team, > > Can I use some options to display the Call_Graph using daVinci, just like > using "-S -O3 -Wb,-tt57:0x100" to get the Control_Flow_Graph? > > Anymore, except /ipa/inline/inline.cxx /ipa/common/ip_graph.h, any other > source code about Call_Graph will be useful for me? > > Thanks! > > Regards, > Hu Yanjun > _________________________________________________________________________ > Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com. > > From owner-pro64-support@oss.sgi.com Thu Apr 26 21:53:50 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3R4rox05045 for pro64-support-outgoing; Thu, 26 Apr 2001 21:53:50 -0700 Received: from hypnos.cps.intel.com (hypnos.cps.intel.com [192.198.165.17]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3R4rnM05042 for ; Thu, 26 Apr 2001 21:53:49 -0700 Received: from SMTP (fmsmsxvs03-1.fm.intel.com [132.233.42.203]) by hypnos.cps.intel.com (8.9.1a+p1/8.9.1/d: relay.m4,v 1.36 2001/04/18 16:16:02 root Exp $) with SMTP id EAA07166; Fri, 27 Apr 2001 04:52:47 GMT Received: from fmsmsx19.fm.intel.com ([132.233.48.19]) by 132.233.48.203 (Norton AntiVirus for Internet Email Gateways 1.0) ; Fri, 27 Apr 2001 04:52:47 0000 (GMT) Received: by fmsmsx19.fm.intel.com with Internet Mail Service (5.5.2653.19) id ; Thu, 26 Apr 2001 21:52:45 -0700 Message-ID: <9287DC1579B0D411AA2F009027F44C3F084E825F@FMSMSX41> From: "Chan, Sun C" To: Shukang ZHOU , Yanjun HU , pro64-support@oss.sgi.com Subject: RE: Call_Graph? Date: Thu, 26 Apr 2001 21:52:42 -0700 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="gb2312" Sender: owner-pro64-support@oss.sgi.com Precedence: bulk I thought IPA does display call graph by daVinci. Wilson/Lilian? Sun > -----Original Message----- > From: Shukang ZHOU [mailto:zhshk@ict.ac.cn] > Sent: Thursday, April 26, 2001 7:31 PM > To: Yanjun HU; pro64-support@oss.sgi.com > Subject: Re: Call_Graph? > > > As far as I know, sgicc cannot display call graph by daVinci. > > In addition, I have a question: what's the option of display > WHIRL? There is > a function named dV_view_whirl(WN *wn, const char *title = NULL, BOOL > show_expr = FALSE, FILE *trace_fp = NULL) in > osprey1.0/be/com/com_whirlview.h. Is it usable? > > Shukang ZHOU > Advanced Compiler Technology Group > Institute of Computing Technology, Chinese Academy of Sciences > E-mail: zhshk@ict.ac.cn > > > ----- Original Message ----- > From: "Yanjun HU" > To: > Cc: > Sent: Friday, April 27, 2001 9:36 AM > Subject: Call_Graph? > > > > Dear Pro64 support team, > > > > Can I use some options to display the Call_Graph using > daVinci, just like > > using "-S -O3 -Wb,-tt57:0x100" to get the Control_Flow_Graph? > > > > Anymore, except /ipa/inline/inline.cxx > /ipa/common/ip_graph.h, any other > > source code about Call_Graph will be useful for me? > > > > Thanks! > > > > Regards, > > Hu Yanjun > > > ______________________________________________________________ > ___________ > > Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com. > > From owner-pro64-support@oss.sgi.com Fri Apr 27 09:39:02 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3RGd2F26109 for pro64-support-outgoing; Fri, 27 Apr 2001 09:39:02 -0700 Received: from sgi.com (sgi.SGI.COM [192.48.153.1]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3RGcuM26098 for ; Fri, 27 Apr 2001 09:38:56 -0700 Received: from rohi.engr.sgi.com ([130.62.180.74]) by sgi.com (980327.SGI.8.8.8-aspam/980304.SGI-aspam: SGI does not authorize the use of its proprietary systems or networks for unsolicited or bulk email from the Internet.) via ESMTP id JAA05865 for ; Fri, 27 Apr 2001 09:38:55 -0700 (PDT) mail_from (mpm@rohi.engr.sgi.com) Received: (from mpm@localhost) by rohi.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id JAA26148; Fri, 27 Apr 2001 09:35:55 -0700 (PDT) Date: Fri, 27 Apr 2001 09:35:55 -0700 (PDT) From: mpm@rohi.engr.sgi.com (Michael Murphy) Message-Id: <200104271635.JAA26148@rohi.engr.sgi.com> To: , "Yanjun HU" , "Shukang ZHOU" Subject: Re: Call_Graph? References: Sender: owner-pro64-support@oss.sgi.com Precedence: bulk It is possible to use DaVinci; it is one of the feedback traces. Look at the howto-debug-compiler file for info. -- Mike Murphy -- mpm@sgi.com -- quote of the day: -- "The value of life is computed not by its duration but by its donation". -- (William James) From owner-pro64-support@oss.sgi.com Sat Apr 28 15:34:43 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3SMYho08516 for pro64-support-outgoing; Sat, 28 Apr 2001 15:34:43 -0700 Received: from sunshine.math.utah.edu (IDENT:9oWurQ84WQTpKWoid3Jv9CmhuRuJrqZU@sunshine.math.utah.edu [128.110.198.2]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3SMYfM08513 for ; Sat, 28 Apr 2001 15:34:41 -0700 Received: from suncore.math.utah.edu (IDENT:SlJbz/Q2IDF5v2O3EViiSE+cuPLpGNoN@suncore0.math.utah.edu [128.110.198.5]) by sunshine.math.utah.edu (8.9.3/8.9.3) with ESMTP id QAA04767; Sat, 28 Apr 2001 16:34:34 -0600 (MDT) Received: (from beebe@localhost) by suncore.math.utah.edu (8.9.3/8.9.3) id QAA22835; Sat, 28 Apr 2001 16:34:33 -0600 (MDT) Date: Sat, 28 Apr 2001 16:34:33 -0600 (MDT) From: "Nelson H. F. Beebe" To: lia64-sim@napali.hpl.hp.com, pro64-support@oss.sgi.com, gcc-bugs@gcc.gnu.org Cc: beebe@math.utah.edu X-US-Mail: "Center for Scientific Computing, Department of Mathematics, 322 INSCC, University of Utah, 155 S 1400 E RM 233, Salt Lake City, UT 84112-0090, USA" X-Telephone: +1 801 581 5254 X-FAX: +1 801 585 1640, +1 801 581 4148 X-URL: http://www.math.utah.edu/~beebe Subject: C++ compiler assertion errors Message-ID: Sender: owner-pro64-support@oss.sgi.com Precedence: bulk I've been working today on attempting to build on NUE IA-64 the most recent version of James Clark's SGML parser, SP, available in source form at ftp://ftp.math.utah.edu/pub/sgml/sp-1.3.4.1.tar.gz http://www.math.utah.edu/pub/sgml/sp-1.3.4.1.tar.gz That directory also contains binary distributions for versions 1.3.4 and 1.3.4.1 for 12 UNIX flavors, demonstrating that it IS possible to build this complex package with several other C++ compilers, even though most C++ compilers fail to handle the code. The 1.3.4.1 builds from today were done with the gcc-2.95.3 release of 22-Mar-2001. There is no support for IA-64 in that compiler version, but the weekly gcc-3.0 development snapshots support IA-64 (see e.g., ftp://gcc.gnu.org/pub/gcc/snapshots/2001-04-23/gcc-20010423.tar.gz). Unfortunately, they still only build successfully on 4 of my 11 local UNIX flavors. During this work, I've encountered fatal compiler assertion failures on IA-64 in both g++ (version 2.9-ia64-000216) and sgiCC (version 0.01.0-13). Since the SGI Pro64 compilers use part of the gcc components, and since it is unclear which group (GNU, HP, or SGI) is managing the IA-64 support in gcc, I'm sending this report to all three. If the real maintainers will stand up, I'll direct future reports to them only! % c++ -fno-implicit-templates -O2 -ansi -I. -I./../include -I./../generic \ -DSP_HAVE_BOOL -DSP_ANSI_CLASS_INST -DSP_MULTI_BYTE -c parseMode.cxx parseMode.cxx: In method `void Parser::compileModes (const Mode *, int, const Dtd *)': parseMode.cxx:403: Internal compiler error in `scan_region', at except.c:2757 Please submit a full bug report. See for instructions. % sgiCC -ansi -I. -I./../include -I./../generic -DSP_HAVE_BOOL \ -DSP_ANSI_CLASS_INST -DSP_MULTI_BYTE -c ParserApp.cxx ### Assertion failure at line 1164 of ../../g++fe/wfe_decl.cxx: ### Compiler Error during Writing WHIRL file phase: ### unexpected tree code ptrmem_cst sgiCC INTERNAL ERROR: /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/gfecc returned non-zero status 1 Attempts to switch compilers allow the compilations to proceed, but unresolved references at link time prevent completion of the builds. When I attempt a build without -DSP_ANSI_CLASS_INST, I get another failure: % g++ -I. -I./../include -DSP_HAVE_BOOL -DSP_MULTI_BYTE -o nsgmls \ nsgmls.o SgmlsEventHandler.o RastEventHandler.o StringSet.o \ nsgmls_inst.o ../lib/libsp.a -lm nsgmls.o: In function `XRastEventHandler::XRastEventHandler(SgmlParser *, char const *, String const &, OutputCodingSystem const *, CmdLineApp *, Messenger *)': nsgmls.o(.text+0x25b2): undefined reference to `ErrnoMessageArg::ErrnoMessageArg(int)' nsgmls.o: In function `XRastEventHandler::truncateOutput(void)': nsgmls.o(.text+0x2d82): undefined reference to `ErrnoMessageArg::ErrnoMessageArg(int)' nsgmls.o(.text+0x3072): undefined reference to `ErrnoMessageArg::ErrnoMessageArg This failure is peculiar, because the preprocessed code shows that the missing function is indeed defined as a public inline member function (after prettyprinting with "astyle --style=ansi"): class ErrnoMessageArg : public OtherMessageArg { public: virtual TypeId dynamicType() const; static inline TypeId staticType() { return TypeId(RTTI_bases_); } protected: static const void *RTTI_bases_[]; private: public: ErrnoMessageArg(int errnum) : errno_(errnum) { } MessageArg *copy() const; int errnum() const; private: int errno_; }; The only other large package written in C++ that I have extensive experience with is the GNU implementation of troff, groff, which is also written by, but not maintained by, James Clark. The latest version is available at ftp://ftp.gnu.org/pub/gnu/groff/groff-1.16.1.tar.gz or at any of the GNU mirrors listed at http://www.gnu.org/order/ftp.html I therefore tried a build of it to see if further compiler assertion failures would turn up, and indeed, one does: % cd groff-1.16.1/ % env CC=sgicc CXX=sgiCC ./configure && make all check % sgiCC -I. -I/nue/local/build/groff-1.16.1/src/utils/addftinfo \ -I/nue/local/build/groff-1.16.1/src/include \ -I/nue/local/build/groff-1.16.1/src/include -DHAVE_STDLIB_H=1 \ -DHAVE_UNISTD_H=1 -DHAVE_DIRENT_H=1 -DHAVE_LIMITS_H=1 \ -DHAVE_SYS_DIR_H=1 -DHAVE_STRING_H=1 -DHAVE_STRINGS_H=1 \ -DHAVE_MATH_H=1 -DRET_TYPE_SRAND_IS_VOID=1 -DHAVE_SYS_NERR=1 \ -DHAVE_SYS_ERRLIST=1 -DHAVE_CC_LIMITS_H=1 -DRETSIGTYPE=void \ -DHAVE_STRUCT_EXCEPTION=1 -DHAVE_UNISTD_H=1 -DHAVE_GETPAGESIZE=1 \ -DHAVE_FMOD=1 -DHAVE_STRTOL=1 -DHAVE_GETCWD=1 -DHAVE_STRERROR=1 \ -DHAVE_PUTENV=1 -DHAVE_RENAME=1 -DHAVE_MKSTEMP=1 -DHAVE_STRCASECMP=1 \ -DHAVE_STRNCASECMP=1 -DHAVE_STRSEP=1 -DHAVE_STRDUP=1 \ -DSYS_SIGLIST_DECLARED=1 -g -O1 -c addftinfo.cc ### Assertion failure at line 1164 of ../../g++fe/wfe_decl.cxx: ### Compiler Error during Writing WHIRL file phase: ### unexpected tree code ptrmem_cst sgiCC INTERNAL ERROR: /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/gfecc returned non-zero status 1 Notice that it occurs at the SAME location as in the compilation of the SP file, ParserApp.cxx. ------------------------------------------------------------------------------- - Nelson H. F. Beebe Tel: +1 801 581 5254 - - Center for Scientific Computing FAX: +1 801 585 1640, +1 801 581 4148 - - University of Utah Internet e-mail: beebe@math.utah.edu - - Department of Mathematics, 322 INSCC beebe@acm.org beebe@computer.org - - 155 S 1400 E RM 233 beebe@ieee.org - - Salt Lake City, UT 84112-0090, USA URL: http://www.math.utah.edu/~beebe - ------------------------------------------------------------------------------- From owner-pro64-support@oss.sgi.com Sat Apr 28 19:02:50 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3T22os13255 for pro64-support-outgoing; Sat, 28 Apr 2001 19:02:50 -0700 Received: from hebe.or.intel.com (jffdns02.or.intel.com [134.134.248.4]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3T22nM13252 for ; Sat, 28 Apr 2001 19:02:49 -0700 Received: from SMTP (orsmsx227.jf.intel.com [192.168.65.227]) by hebe.or.intel.com (8.9.1a+p1/8.9.1/d: relay.m4,v 1.36 2001/04/18 16:16:02 root Exp $) with SMTP id CAA10409; Sun, 29 Apr 2001 02:01:51 GMT Received: from orsmsx229.JF.INTEL.COM ([192.168.70.229]) by 192.168.70.227 (Norton AntiVirus for Internet Email Gateways 1.0) ; Sun, 29 Apr 2001 02:01:51 0000 (GMT) Received: by ORSMSX229 with Internet Mail Service (5.5.2653.19) id ; Sat, 28 Apr 2001 19:01:43 -0700 Message-ID: <9287DC1579B0D411AA2F009027F44C3F084E850F@FMSMSX41> From: "Chan, Sun C" To: "Nelson H. F. Beebe" , pro64-support@oss.sgi.com Subject: RE: C++ compiler assertion errors Date: Sat, 28 Apr 2001 19:01:48 -0700 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2653.19) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-pro64-support@oss.sgi.com Precedence: bulk > -----Original Message----- > From: Nelson H. F. Beebe [mailto:beebe@math.utah.edu] > Sent: Saturday, April 28, 2001 3:35 PM > To: lia64-sim@napali.hpl.hp.com; pro64-support@oss.sgi.com; > gcc-bugs@gcc.gnu.org > Cc: beebe@math.utah.edu > Subject: C++ compiler assertion errors > % c++ -fno-implicit-templates -O2 -ansi -I. > -I./../include -I./../generic \ > -DSP_HAVE_BOOL -DSP_ANSI_CLASS_INST > -DSP_MULTI_BYTE -c parseMode.cxx > parseMode.cxx: In method `void Parser::compileModes > (const Mode *, int, const Dtd *)': > parseMode.cxx:403: Internal compiler error in > `scan_region', at except.c:2757 > Please submit a full bug report. > See > for > instructions. > This is clearly a frontend error, with full instruction on how to report bug > % sgiCC -ansi -I. -I./../include -I./../generic > -DSP_HAVE_BOOL \ > -DSP_ANSI_CLASS_INST -DSP_MULTI_BYTE -c ParserApp.cxx > ### Assertion failure at line 1164 of ../../g++fe/wfe_decl.cxx: > ### Compiler Error during Writing WHIRL file phase: > ### unexpected tree code ptrmem_cst > sgiCC INTERNAL ERROR: > /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/gfecc returned > non-zero status 1 > This is a bug in translating gccIR to Whirl. Bug is at SGI's court. I suggest you cpp the source, reduce that to, hopefully a minimum, test case. I'm sure Murthy will get to it very quickly. > Attempts to switch compilers allow the compilations to proceed, but > unresolved references at link time prevent completion of the builds. > > When I attempt a build without -DSP_ANSI_CLASS_INST, I get another > failure: > > % g++ -I. -I./../include -DSP_HAVE_BOOL > -DSP_MULTI_BYTE -o nsgmls \ > nsgmls.o SgmlsEventHandler.o RastEventHandler.o > StringSet.o \ > nsgmls_inst.o ../lib/libsp.a -lm > nsgmls.o: In function > `XRastEventHandler::XRastEventHandler(SgmlParser *, char > const *, String const &, OutputCodingSystem > const *, CmdLineApp *, Messenger *)': > nsgmls.o(.text+0x25b2): undefined reference to > `ErrnoMessageArg::ErrnoMessageArg(int)' > nsgmls.o: In function `XRastEventHandler::truncateOutput(void)': > nsgmls.o(.text+0x2d82): undefined reference to > `ErrnoMessageArg::ErrnoMessageArg(int)' > nsgmls.o(.text+0x3072): undefined reference to > `ErrnoMessageArg::ErrnoMessageArg > > This failure is peculiar, because the preprocessed code shows that the > missing function is indeed defined as a public inline member function > (after prettyprinting with "astyle --style=ansi"): > I've seen these kind of failures when there is inconsistent optimize options used that does not ended up inlining all functions as intended. > % sgiCC -I. > -I/nue/local/build/groff-1.16.1/src/utils/addftinfo \ > -I/nue/local/build/groff-1.16.1/src/include \ > -I/nue/local/build/groff-1.16.1/src/include > -DHAVE_STDLIB_H=1 \ > -DHAVE_UNISTD_H=1 -DHAVE_DIRENT_H=1 -DHAVE_LIMITS_H=1 \ > -DHAVE_SYS_DIR_H=1 -DHAVE_STRING_H=1 > -DHAVE_STRINGS_H=1 \ > -DHAVE_MATH_H=1 -DRET_TYPE_SRAND_IS_VOID=1 > -DHAVE_SYS_NERR=1 \ > -DHAVE_SYS_ERRLIST=1 -DHAVE_CC_LIMITS_H=1 > -DRETSIGTYPE=void \ > -DHAVE_STRUCT_EXCEPTION=1 -DHAVE_UNISTD_H=1 > -DHAVE_GETPAGESIZE=1 \ > -DHAVE_FMOD=1 -DHAVE_STRTOL=1 -DHAVE_GETCWD=1 > -DHAVE_STRERROR=1 \ > -DHAVE_PUTENV=1 -DHAVE_RENAME=1 > -DHAVE_MKSTEMP=1 -DHAVE_STRCASECMP=1 \ > -DHAVE_STRNCASECMP=1 -DHAVE_STRSEP=1 -DHAVE_STRDUP=1 \ > -DSYS_SIGLIST_DECLARED=1 -g -O1 -c addftinfo.cc > > ### Assertion failure at line 1164 of ../../g++fe/wfe_decl.cxx: > ### Compiler Error during Writing WHIRL file phase: > ### unexpected tree code ptrmem_cst > sgiCC INTERNAL ERROR: > /usr/lib/gcc-lib/ia64-sgi-linux/sgicc-1.0/gfecc returned > non-zero status 1 Sounds line the same gccIR to Whirl translation assertion. Sun From owner-pro64-support@oss.sgi.com Sun Apr 29 09:38:42 2001 Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f3TGcgZ06101 for pro64-support-outgoing; Sun, 29 Apr 2001 09:38:42 -0700 Received: from sgi.com (sgi.SGI.COM [192.48.153.1]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f3TGccM06098 for ; Sun, 29 Apr 2001 09:38:38 -0700 Received: from gaea.engr.sgi.com ([130.62.180.97]) by sgi.com (980327.SGI.8.8.8-aspam/980304.SGI-aspam: SGI does not authorize the use of its proprietary systems or networks for unsolicited or bulk email from the Internet.) via ESMTP id JAA06427 for ; Sun, 29 Apr 2001 09:38:35 -0700 (PDT) mail_from (murthy@gaea.engr.sgi.com) Received: (from murthy@localhost) by gaea.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id JAA50306; Sun, 29 Apr 2001 09:36:34 -0700 (PDT) Date: Sun, 29 Apr 2001 09:36:34 -0700 (PDT) From: murthy@gaea.engr.sgi.com (Chandrasekhar Murthy) Message-Id: <200104291636.JAA50306@gaea.engr.sgi.com> To: gcc-bugs@gcc.gnu.org, pro64-support@oss.sgi.com, lia64-sim@napali.hpl.hp.com, "Nelson H. F. Beebe" Subject: Re: C++ compiler assertion errors Cc: beebe@math.utah.edu Sender: owner-pro64-support@oss.sgi.com Precedence: bulk As Sun mentioned, please send me a preprocessed file for both cases along with the commandline option used. Murthy