[Bug 213049] New: Emacs CVS build fails in 10.2 alpha 5
https://bugzilla.novell.com/show_bug.cgi?id=213049 Summary: Emacs CVS build fails in 10.2 alpha 5 Product: openSUSE 10.2 Version: Alpha 5 Platform: i686 OS/Version: Other Status: NEW Severity: Major Priority: P5 - None Component: Development AssignedTo: pth@novell.com ReportedBy: dobey@ximian.com QAContact: qa@suse.de When attempting to build emacs CVS on 10.2a5, the make bootstrap phase fails, with bootstrap-emacs aborting. The same source tree builds fine on SLED 10, and Fedora Core 5. I have tried using both gcc 4.1.3 and the gcc41 4.1.2 package. I believe this to be an issue in either gcc or glibc. The backtrace that I can get is pasted below. (gdb) r Starting program: /home/dobey/Projects/emacs/src/bootstrap-emacs -geometry 80x40+0+0 Failed to read a valid object file image from memory. [Thread debugging using libthread_db enabled] [New Thread -1218861376 (LWP 23434)] Program received signal SIGABRT, Aborted. [Switching to Thread -1218861376 (LWP 23434)] 0xb7fbe410 in ?? () (gdb) thread apply all bt Thread 1 (Thread -1218861376 (LWP 23434)): #0 0xb7fbe410 in ?? () #1 0xbfcde74c in ?? () #2 0x00000006 in ?? () #3 0x00005b8a in ?? () #4 0xb764e060 in raise () from /lib/libc.so.6 #5 0xb764f801 in abort () from /lib/libc.so.6 #6 0xb7689719 in malloc_printerr () from /lib/libc.so.6 #7 0xb768ad45 in free () from /lib/libc.so.6 #8 0x080ed8ba in malloc_initialize_hook () at emacs.c:774 #9 0xb7689ce0 in ptmalloc_init () from /lib/libc.so.6 #10 0xb768d913 in malloc_hook_ini () from /lib/libc.so.6 #11 0xb768d095 in malloc () from /lib/libc.so.6 #12 0x0813e0ae in xmalloc (size=12) at alloc.c:756 #13 0x080ee39d in sort_args (argc=0, argv=0x6) at emacs.c:1908 #14 0x080ee7cb in main (argc=0, argv=0xbfcdee24) at emacs.c:883 Lisp Backtrace: Cannot access memory at address 0xbfda8b94 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=213049 pth@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|pth@novell.com |werner@novell.com -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=213049 werner@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- Severity|Major |Enhancement Status|NEW |NEEDINFO Info Provider| |dobey@ximian.com ------- Comment #1 from werner@novell.com 2006-11-02 03:40 MST ------- Hmmmm ... why you're open a bug if you are not able to build emacs CVS? You may ask gcc or glibc people if there are some changes you may also debug the problem and give me some feedback about how to get emacs CVS running but open a bug isn't very nice. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=213049 dobey@ximian.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |ASSIGNED Info Provider|dobey@ximian.com | ------- Comment #2 from dobey@ximian.com 2006-11-02 06:40 MST ------- I filed a bug because it is evident to me that there is a bug in openSUSE in glibc or gcc. I am not sure how to debug the problem. The stack trace above is all I get when emacs aborts. This same source tree builds and runs perfectly fine on both SLED 10 and Fedora Core 5. If there is some specific information available to me that isn't in the stack trace already posted, I don't know what it is, and it isn't obvious. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=213049 werner@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |werner@novell.com, pbaudis@novell.com AssignedTo|werner@novell.com |matz@novell.com Status|ASSIGNED |NEW ------- Comment #3 from werner@novell.com 2006-11-02 07:02 MST ------- Then not the emacs maintainer but the gcc maintainer should be assigned. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=213049 ------- Comment #4 from pbaudis@novell.com 2006-11-02 09:50 MST ------- It aborted - did it print any error? What happens if you run it in valgrind? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=213049 matz@novell.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Comment #5 from matz@novell.com 2006-11-27 09:07 MST ------- I've tried to compile emacs CVS with the Beta2plus of 10.2, with these package versions: gcc-4.1.3-26 gcc41-4.1.2_20061109-2 glibc-2.5-19 It worked just fine. From the backtrace and looking at the emacs.c sources it seems that it plays some heavy magic with the glibc allocator. The free() call that is shown in the backtrace is only done when XMALLOC_OVERRUN_CHECK is defined, which never is defined. It can be defined to change a 0 into a 1 in lisp.h, but CVS doesn't do that. That means you didn't test emacs CVS but a locally modified version. Nevertheless I also turned it on for testing, and it still worked for me. So, let's assume it's fixed for the final 10.2. If that turns out to be false for you, please reopen with exact steps to reproduce. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=213049 ------- Comment #6 from matz@novell.com 2006-11-27 09:10 MST ------- Oops, sorry. emacs.c actually contains: #ifndef XMALLOC_OVERRUN_CHECK free (malloc_state_ptr); #endif so, it does the free() when the symbols is _not_ defined, i.e. like it comes from CVS. That doesn't change the outcome that it works both ways for me, though. Sorry again. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
https://bugzilla.novell.com/show_bug.cgi?id=213049 ------- Comment #7 from dobey@ximian.com 2006-11-27 09:34 MST ------- It does seem to work in 10.2 beta 2 for me. I imagine something might have changed in glibc that now causes XMALLOC_OVERRUN_CHECK to be defined by configure. Anyway, it does build now. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.
participants (1)
-
bugzilla_noreply@novell.com