Bug#365409: [Pkg-openldap-devel] Bug#365409: slapd: segfaults on entry modify
On Sun, Jun 04, 2006 at 07:57:20PM -0700, Quanah Gibson-Mount wrote: If it isn't an OpenLDAP thing, then I guess no ITS. ;) I thought it wasn't completely specific to the ARM platform. I haven't seen anybody being able to reproduce it except on arm, and certainly such brokenness should be detected nearly at once. /* Steinar */ -- Homepage: http://www.sesse.net/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#365409: [Pkg-openldap-devel] Bug#365409: slapd: segfaults on entry modify
On Sun, Jun 04, 2006 at 07:36:11PM -0700, Steve Langasek wrote: Since this is a bug with OpenLDAP Is that certain? It looks just as likely to be a compiler bug to me. After testing some more, it seems fairly likely to be a compiler bug. Compiling the file in question (mods.c) with -O0 makes the problem go away. Likewise, compiling it with gcc-4.1 (with -O2) makes the problem go away. I'm unable to point at anything directly wrong in the code, but I'm not familiar with arm assembler. I'll try building the entire package with gcc-4.1, but that will take another eight hours. (Perhaps I should try bisecting the list of available -f flags, but I'm not sure how useful that would be...) Cc'ing Martin Michlmayr, since this looks like another case for switching to gcc-4.1 in etch. /* Steinar */ -- Homepage: http://www.sesse.net/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#365409: [Pkg-openldap-devel] Bug#365409: slapd: segfaults on entry modify
On Mon, Jun 05, 2006 at 02:01:02PM +0200, Steinar H. Gunderson wrote: Compiling the file in question (mods.c) with -O0 makes the problem go away. Likewise, compiling it with gcc-4.1 (with -O2) makes the problem go away. I'm unable to point at anything directly wrong in the code, but I'm not familiar with arm assembler. I'll try building the entire package with gcc-4.1, but that will take another eight hours. (Perhaps I should try bisecting the list of available -f flags, but I'm not sure how useful that would be...) I also tried compiling the entire package, unmodified, with /usr/bin/gcc pointing to gcc-4.1. It compiled without a hitch, and ldapmodify works without segfaulting slapd. At this point, I throw up my hands and call it a compiler bug (very comforting, no? :-) ); as GCC 4.1 will be the default in Debian from tomorrow, this problem will probably be solved by issuing a binNMU for arm. /* Steinar */ -- Homepage: http://www.sesse.net/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#365409: [Pkg-openldap-devel] Bug#365409: slapd: segfaults on entry modify
On Mon, Jun 05, 2006 at 07:06:11PM +0200, Steinar H. Gunderson wrote: On Mon, Jun 05, 2006 at 02:01:02PM +0200, Steinar H. Gunderson wrote: Compiling the file in question (mods.c) with -O0 makes the problem go away. Likewise, compiling it with gcc-4.1 (with -O2) makes the problem go away. I'm unable to point at anything directly wrong in the code, but I'm not familiar with arm assembler. I'll try building the entire package with gcc-4.1, but that will take another eight hours. (Perhaps I should try bisecting the list of available -f flags, but I'm not sure how useful that would be...) I also tried compiling the entire package, unmodified, with /usr/bin/gcc pointing to gcc-4.1. It compiled without a hitch, and ldapmodify works without segfaulting slapd. At this point, I throw up my hands and call it a compiler bug (very comforting, no? :-) ); as GCC 4.1 will be the default in Debian from tomorrow, this problem will probably be solved by issuing a binNMU for arm. Thanks very much for your work on this! I'll plan to schedule a binNMU for openldap2.3 on arm as soon as I can confirm the new gcc-defaults are available and installed on all buildds. Cheers, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. [EMAIL PROTECTED] http://www.debian.org/ signature.asc Description: Digital signature
Bug#365409: [Pkg-openldap-devel] Bug#365409: slapd: segfaults on entry modify
--On Sunday, June 04, 2006 9:15 PM +0200 Steinar H. Gunderson [EMAIL PROTECTED] wrote: On Sun, Jun 04, 2006 at 08:43:47PM +0200, Steinar H. Gunderson wrote: I'm completely unable to reproduce this. Scrap that; I can reproduce it now. I just didn't see that slapd segfaulted... Debugging now. Is it possible to reproduce in 2.3.24? Or just the rather old 2.2.x version? --Quanah -- Quanah Gibson-Mount Principal Software Developer ITS/Shared Application Services Stanford University GnuPG Public Key: http://www.stanford.edu/~quanah/pgp.html -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#365409: [Pkg-openldap-devel] Bug#365409: slapd: segfaults on entry modify
On Sun, Jun 04, 2006 at 01:17:51PM -0700, Quanah Gibson-Mount wrote: Is it possible to reproduce in 2.3.24? Or just the rather old 2.2.x version? 2.3.24. I'm currently compiling a noopt/nostrip-version, but at 133MHz ARM it's going to take a while. /* Steinar */ -- Homepage: http://www.sesse.net/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#365409: [Pkg-openldap-devel] Bug#365409: slapd: segfaults on entry modify
On Sun, Jun 04, 2006 at 10:19:44PM +0200, Steinar H. Gunderson wrote: I'm currently compiling a noopt/nostrip-version, but at 133MHz ARM it's going to take a while. OK, the problem doesn't show up at -O0, which supports the theory that it's some kind of overflow issue. I'm starting a build with -O2 now -- perhaps it can be caught using valgrind on a slightly faster platform, though... /* Steinar */ -- Homepage: http://www.sesse.net/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#365409: [Pkg-openldap-devel] Bug#365409: slapd: segfaults on entry modify
--On Monday, June 05, 2006 1:58 AM +0200 Steinar H. Gunderson [EMAIL PROTECTED] wrote: On Mon, Jun 05, 2006 at 12:42:57AM +0200, Steinar H. Gunderson wrote: OK, the problem doesn't show up at -O0, which supports the theory that it's some kind of overflow issue. I'm starting a build with -O2 now -- perhaps it can be caught using valgrind on a slightly faster platform, though... OK, valgrind didn't catch it, but I believe in the theory of a stack smashing problem (which valgrind won't catch; bounds-checking gcc would, though). Since this is a bug with OpenLDAP, I would highly advise opening an ITS with OpenLDAP at: http://www.openldap.org/its/ I would include the URL to the bug in debian, since it has the gdb backtrace and other information contained in it. --Quanah -- Quanah Gibson-Mount QA Engineer http://www.openldap.org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#365409: [Pkg-openldap-devel] Bug#365409: slapd: segfaults on entry modify
On Sun, Jun 04, 2006 at 07:28:14PM -0700, Quanah Gibson-Mount wrote: On Mon, Jun 05, 2006 at 12:42:57AM +0200, Steinar H. Gunderson wrote: OK, the problem doesn't show up at -O0, which supports the theory that it's some kind of overflow issue. I'm starting a build with -O2 now -- perhaps it can be caught using valgrind on a slightly faster platform, though... OK, valgrind didn't catch it, but I believe in the theory of a stack smashing problem (which valgrind won't catch; bounds-checking gcc would, though). Since this is a bug with OpenLDAP Is that certain? It looks just as likely to be a compiler bug to me. -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. [EMAIL PROTECTED] http://www.debian.org/ signature.asc Description: Digital signature
Bug#365409: [Pkg-openldap-devel] Bug#365409: slapd: segfaults on entry modify
--On Sunday, June 04, 2006 7:36 PM -0700 Steve Langasek [EMAIL PROTECTED] wrote: On Sun, Jun 04, 2006 at 07:28:14PM -0700, Quanah Gibson-Mount wrote: On Mon, Jun 05, 2006 at 12:42:57AM +0200, Steinar H. Gunderson wrote: OK, the problem doesn't show up at -O0, which supports the theory that it's some kind of overflow issue. I'm starting a build with -O2 now -- perhaps it can be caught using valgrind on a slightly faster platform, though... OK, valgrind didn't catch it, but I believe in the theory of a stack smashing problem (which valgrind won't catch; bounds-checking gcc would, though). Since this is a bug with OpenLDAP Is that certain? It looks just as likely to be a compiler bug to me. Oh.. I misunderstood. ;) If it isn't an OpenLDAP thing, then I guess no ITS. ;) I thought it wasn't completely specific to the ARM platform. --Quanah -- Quanah Gibson-Mount QA Engineer http://www.openldap.org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]