Date: Mon, 13 Jul 2009 13:58:05 +0300 From: Kostik Belousov <kostikbel@gmail.com> To: freebsd-current@freebsd.org Subject: rtld segfaults [Was Re: Process stuck in vmmaps on 8.0-BETA1] Message-ID: <20090713105805.GA55190@deviant.kiev.zoral.com.ua> In-Reply-To: <20090713104448.GE982@rwpc12.mby.riverwillow.net.au> References: <20090709073054.GB27373@rwpc12.mby.riverwillow.net.au> <20090709085242.GC27373@rwpc12.mby.riverwillow.net.au> <20090709142121.GS55190@deviant.kiev.zoral.com.ua> <20090710035849.GB31950@rwpc12.mby.riverwillow.net.au> <1247214510.2437.1693.camel@strangepork.london.mintel.ad> <20090710114234.GF32316@rwpc12.mby.riverwillow.net.au> <20090710132429.GA55190@deviant.kiev.zoral.com.ua> <20090711091050.GN32316@rwpc12.mby.riverwillow.net.au> <20090711124206.GF55190@deviant.kiev.zoral.com.ua> <20090713104448.GE982@rwpc12.mby.riverwillow.net.au>
next in thread | previous in thread | raw e-mail | index | archive | help
--WzyiqVXNYYkrrY2o Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Mon, Jul 13, 2009 at 08:44:48PM +1000, John Marshall wrote: > On Sat, 11 Jul 2009, 15:42 +0300, Kostik Belousov wrote: > > On Sat, Jul 11, 2009 at 07:10:50PM +1000, John Marshall wrote: > > > On Fri, 10 Jul 2009, 16:24 +0300, Kostik Belousov wrote: > > > > > > > > Thank you, I see what is going on. Please, try the following patch. > > >=20 > > > Thank you Kostik, > > >=20 > > > I applied your patch to vm_map.c and rebuilt the kernel. Now the > > > process crashes instead - but I think it is getting beyond the point > > > where it was hanging. > > >=20 > > > I rebuilt ntpd on the patched kernel and it still produces this crash. > >=20 > > This is obviously different issue. Besiddes, backtrace does not make > > much sense. > >=20 > > You need to rebuild and install lib/libc, lib/libthr and > > libexec/rtld-elf with debugging symbols. Easiest way to do this is to > > enter into each listed src/ subdirectory and do > > make obj > > make clean > > make depend > > make all install DEBUG_FLAGS=3D-g > > and then start ntpd again and get the backtrace, might by "bt full" will > > be more informative. >=20 > Thank you for the help. This is what I saw after following your > instructions... >=20 > (gdb) bt full > #0 memset () at /usr/src/lib/libc/i386/string/memset.S:78 > No locals. > #1 0x00006000 in ?? () > No symbol table info available. > #2 0x28085c9b in map_object (fd=3D6,=20 > path=3D0x280c0080 "/usr/local/lib/nss_winbind.so.1", sb=3D0xbfbfdbac) > at /usr/src/libexec/rtld-elf/map_object.c:196 > obj =3D Variable "obj" is not available. > Current language: auto; currently asm >=20 > > BTW, do you use NIS ? >=20 > Yes, and I also have winbind. If I remove winbind from > /etc/nsswitch.conf then ntpd runs happily. I suspect that this might be > related to the GSSAPI authentication not working for sshd. I shall now > turn my attention to that problem. >=20 > Thank you for your help in resolving the initial vmmaps problem - and > for your patch to vm_map.c. I think we've progressed far enough beyond > that problem to close this thread. This still seems to be a system bug, but now in the rtld. I need two things: 1. ktrace/kdump output for ntpd start that causes the process to crash; 2. readelf -a /usr/local/lib/nss_winbind.so.1 (both may be large enough, so that it is better to put it somewhere on web server). --WzyiqVXNYYkrrY2o Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (FreeBSD) iEYEARECAAYFAkpbEzwACgkQC3+MBN1Mb4h33ACeKgJ9zTV8rCaMtxcEjrQcB+j4 cksAoIonU2aAxIR2kwkWOg4bTL1ysMAu =v7sp -----END PGP SIGNATURE----- --WzyiqVXNYYkrrY2o--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20090713105805.GA55190>