From mboxrd@z Thu Jan 1 00:00:00 1970 Path: news.gmane.org!not-for-mail From: Paul Eggert Newsgroups: gmane.emacs.devel Subject: Re: VIRT_ADDR_VARIES Date: Thu, 10 Nov 2011 00:17:06 -0800 Organization: UCLA Computer Science Department Message-ID: <4EBB8882.6080504@cs.ucla.edu> References: <4EB73983.1060000@cs.ucla.edu> <4EB86864.5080709@cs.ucla.edu> <4EB96841.7020701@cs.ucla.edu> <4EB99C7D.8000407@cs.ucla.edu> <4EBABBF5.5030602@cs.ucla.edu> NNTP-Posting-Host: lo.gmane.org Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit X-Trace: dough.gmane.org 1320913043 805 80.91.229.12 (10 Nov 2011 08:17:23 GMT) X-Complaints-To: usenet@dough.gmane.org NNTP-Posting-Date: Thu, 10 Nov 2011 08:17:23 +0000 (UTC) Cc: emacs-devel@gnu.org To: Andreas Schwab Original-X-From: emacs-devel-bounces+ged-emacs-devel=m.gmane.org@gnu.org Thu Nov 10 09:17:17 2011 Return-path: Envelope-to: ged-emacs-devel@m.gmane.org Original-Received: from lists.gnu.org ([140.186.70.17]) by lo.gmane.org with esmtp (Exim 4.69) (envelope-from ) id 1ROPoy-00076I-Ox for ged-emacs-devel@m.gmane.org; Thu, 10 Nov 2011 09:17:17 +0100 Original-Received: from localhost ([::1]:42983 helo=lists.gnu.org) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ROPoy-0006MB-8I for ged-emacs-devel@m.gmane.org; Thu, 10 Nov 2011 03:17:16 -0500 Original-Received: from eggs.gnu.org ([140.186.70.92]:37827) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ROPou-0006Lv-8z for emacs-devel@gnu.org; Thu, 10 Nov 2011 03:17:13 -0500 Original-Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ROPos-0007r0-OA for emacs-devel@gnu.org; Thu, 10 Nov 2011 03:17:12 -0500 Original-Received: from smtp.cs.ucla.edu ([131.179.128.62]:55904) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ROPos-0007qo-D3 for emacs-devel@gnu.org; Thu, 10 Nov 2011 03:17:10 -0500 Original-Received: from localhost (localhost.localdomain [127.0.0.1]) by smtp.cs.ucla.edu (Postfix) with ESMTP id 55B38A60003; Thu, 10 Nov 2011 00:17:08 -0800 (PST) X-Virus-Scanned: amavisd-new at smtp.cs.ucla.edu Original-Received: from smtp.cs.ucla.edu ([127.0.0.1]) by localhost (smtp.cs.ucla.edu [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id qbt-eBl1mrEu; Thu, 10 Nov 2011 00:17:06 -0800 (PST) Original-Received: from [192.168.1.10] (pool-71-189-109-235.lsanca.fios.verizon.net [71.189.109.235]) by smtp.cs.ucla.edu (Postfix) with ESMTPSA id 04FF8A60001; Thu, 10 Nov 2011 00:17:06 -0800 (PST) User-Agent: Mozilla/5.0 (X11; Linux i686; rv:7.0.1) Gecko/20110929 Thunderbird/7.0.1 In-Reply-To: X-detected-operating-system: by eggs.gnu.org: GNU/Linux 2.6 (newer, 3) X-Received-From: 131.179.128.62 X-BeenThere: emacs-devel@gnu.org X-Mailman-Version: 2.1.14 Precedence: list List-Id: "Emacs development discussions." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: emacs-devel-bounces+ged-emacs-devel=m.gmane.org@gnu.org Original-Sender: emacs-devel-bounces+ged-emacs-devel=m.gmane.org@gnu.org Xref: news.gmane.org gmane.emacs.devel:145964 Archived-At: How about this patch? Does it fix your problem with valgrind? === modified file 'src/ChangeLog' --- src/ChangeLog 2011-11-09 14:29:23 +0000 +++ src/ChangeLog 2011-11-10 08:14:27 +0000 @@ -1,3 +1,22 @@ +2011-11-10 Paul Eggert + + Standardize on VIRT_ADDR_VARIES behavior; otherwise, valgrind + does not work on some platforms. Problem reported by Andreas Schwab in + . + * puresize.h (pure, PURE_P): Always behave as if VIRT_ADDR_VARIES + is set, removing the need for VIRT_ADDRESS_VARIES. + (PURE_P): Use a more-efficient implementation that needs just one + comparison, not two: on x86-64 with GCC 4.6.2, this cut down the + number of instructions from 6 (xorl, cmpq, jge, xorl, cmpq, setge) + to 4 (xorl, subq, cmpq, setbe). + * alloc.c (pure): Always extern now, since that's the + VIRT_ADDR_VARIES behavior. + (PURE_POINTER_P): Use a single comparison, not two, for + consistency with the new puresize.h. + * lisp.h (PNTR_COMPARISON_TYPE): Remove; no longer needed. + * m/ibms390.h, m/intel386.h, m/template.h, s/cygwin.h, s/hpux10-20.h: + Remove VIRT_ADDR_VARIES no longer needed. + 2011-11-09 Chong Yidong * window.c (Fwindow_inside_edges, Fwindow_inside_pixel_edges) === modified file 'src/alloc.c' --- src/alloc.c 2011-11-07 05:37:49 +0000 +++ src/alloc.c 2011-11-10 08:14:27 +0000 @@ -203,9 +203,6 @@ remapping on more recent systems because this is less important nowadays than in the days of small memories and timesharing. */ -#ifndef VIRT_ADDR_VARIES -static -#endif EMACS_INT pure[(PURESIZE + sizeof (EMACS_INT) - 1) / sizeof (EMACS_INT)] = {1,}; #define PUREBEG (char *) pure @@ -222,10 +219,7 @@ /* Value is non-zero if P points into pure space. */ #define PURE_POINTER_P(P) \ - (((PNTR_COMPARISON_TYPE) (P) \ - < (PNTR_COMPARISON_TYPE) ((char *) purebeg + pure_size)) \ - && ((PNTR_COMPARISON_TYPE) (P) \ - >= (PNTR_COMPARISON_TYPE) purebeg)) + ((uintptr_t) (P) - (uintptr_t) purebeg <= pure_size) /* Index in pure at which next pure Lisp object will be allocated.. */ === modified file 'src/lisp.h' --- src/lisp.h 2011-11-07 17:04:01 +0000 +++ src/lisp.h 2011-11-10 08:14:27 +0000 @@ -1877,9 +1877,6 @@ CHECK_NATNUM (tmp); \ XSETCDR ((x), tmp); \ } while (0) - -/* Cast pointers to this type to compare them. */ -#define PNTR_COMPARISON_TYPE uintptr_t /* Define a built-in function for calling from Lisp. `lname' should be the name to give the function in Lisp, === modified file 'src/m/ibms390.h' --- src/m/ibms390.h 2011-02-16 01:35:20 +0000 +++ src/m/ibms390.h 2011-11-10 08:14:27 +0000 @@ -17,11 +17,6 @@ You should have received a copy of the GNU General Public License along with GNU Emacs. If not, see . */ - -/* Define VIRT_ADDR_VARIES if the virtual addresses of - pure and impure space as loaded can vary, and even their - relative order cannot be relied on. - - Otherwise Emacs assumes that text space precedes data space, - numerically. */ -#define VIRT_ADDR_VARIES +/* This file is a placeholder -- it does not contain any definitions. + At some point we should probably fix this by removing the file + and removing all uses of it. */ === modified file 'src/m/intel386.h' --- src/m/intel386.h 2011-01-31 23:54:50 +0000 +++ src/m/intel386.h 2011-11-10 08:14:27 +0000 @@ -19,7 +19,6 @@ #ifdef WINDOWSNT -#define VIRT_ADDR_VARIES #define DATA_START get_data_start () #endif @@ -28,4 +27,3 @@ /* we cannot get the maximum address for brk */ #define ULIMIT_BREAK_VALUE (32*1024*1024) #endif - === modified file 'src/m/template.h' --- src/m/template.h 2011-02-16 01:35:20 +0000 +++ src/m/template.h 2011-11-10 08:14:27 +0000 @@ -21,14 +21,6 @@ does not define it automatically. Ones defined so far include m68k and many others */ -/* Define VIRT_ADDR_VARIES if the virtual addresses of - pure and impure space as loaded can vary, and even their - relative order cannot be relied on. - - Otherwise Emacs assumes that text space precedes data space, - numerically. */ -#define VIRT_ADDR_VARIES - /* After adding support for a new machine, modify the large case statement in configure.in to recognize reasonable configuration names, and add a description of the system to === modified file 'src/puresize.h' --- src/puresize.h 2011-06-09 19:08:29 +0000 +++ src/puresize.h 2011-11-10 08:14:27 +0000 @@ -75,21 +75,7 @@ /* Define PURE_P. */ -#ifdef VIRT_ADDR_VARIES -/* For machines where text and data can go anywhere - in virtual memory. */ - extern EMACS_INT pure[]; #define PURE_P(obj) \ - ((PNTR_COMPARISON_TYPE) XPNTR (obj) < (PNTR_COMPARISON_TYPE) ((char *) pure + PURESIZE) \ - && (PNTR_COMPARISON_TYPE) XPNTR (obj) >= (PNTR_COMPARISON_TYPE) pure) - -#else /* not VIRT_ADDR_VARIES */ - -extern char my_edata[]; - -#define PURE_P(obj) \ - ((PNTR_COMPARISON_TYPE) XPNTR (obj) < (PNTR_COMPARISON_TYPE) my_edata) - -#endif /* VIRT_ADDRESS_VARIES */ + ((uintptr_t) XPNTR (obj) - (uintptr_t) pure <= PURESIZE) === modified file 'src/s/cygwin.h' --- src/s/cygwin.h 2011-03-17 05:15:08 +0000 +++ src/s/cygwin.h 2011-11-10 08:14:27 +0000 @@ -91,9 +91,6 @@ why it needed to be changed. */ #define GC_MARK_STACK GC_MAKE_GCPROS_NOOPS -/* Virtual addresses of pure and impure space can vary, as on Windows. */ -#define VIRT_ADDR_VARIES - /* Emacs supplies its own malloc, but glib (part of Gtk+) calls memalign and on Cygwin, that becomes the Cygwin-supplied memalign. As malloc is not the Cygwin malloc, the Cygwin memalign always === modified file 'src/s/hpux10-20.h' --- src/s/hpux10-20.h 2011-02-16 01:35:20 +0000 +++ src/s/hpux10-20.h 2011-11-10 08:14:27 +0000 @@ -100,14 +100,6 @@ header sections which lose when `static' is defined away, as it is on HP-UX. (You get duplicate symbol errors on linking). */ #undef _FILE_OFFSET_BITS - -/* Define VIRT_ADDR_VARIES if the virtual addresses of - pure and impure space as loaded can vary, and even their - relative order cannot be relied on. - - Otherwise Emacs assumes that text space precedes data space, - numerically. */ -#define VIRT_ADDR_VARIES /* The data segment on this machine always starts at address 0x40000000. */ #define DATA_SEG_BITS 0x40000000