Bugzilla – Bug 388435
nscd segfaults
Last modified: 2008-06-27 00:38:11 UTC
After clean install & update from factory /var/log/messages contains nscd errors cat /var/log/messages | grep nscd May 6 08:20:26 linux klogd: type=1505 audit(1210062017.523:9): operation="profile_load" name="/usr/sbin/nscd" name2="default" pid=1747 May 6 08:58:11 linux klogd: type=1505 audit(1210057083.963:9): operation="profile_load" name="/usr/sbin/nscd" name2="default" pid=1834 May 6 09:12:12 linux klogd: type=1505 audit(1210057925.180:9): operation="profile_load" name="/usr/sbin/nscd" name2="default" pid=1826 May 6 10:15:06 linux klogd: type=1505 audit(1210061698.295:9): operation="profile_load" name="/usr/sbin/nscd" name2="default" pid=1750 May 7 10:19:21 linux klogd: type=1505 audit(1210148353.947:7): operation="profile_load" name="/usr/sbin/nscd" name2="default" pid=1707 May 8 08:41:17 linux klogd: type=1505 audit(1210228870.004:9): operation="profile_load" name="/usr/sbin/nscd" name2="default" pid=1751 May 8 09:43:37 reached klogd: nscd[4224]: segfault at ffd1f0e8 ip b80e33ba sp afd1efd8 error 6 in nscd[b80d3000+1c000] May 8 19:22:28 workstation6l klogd: nscd[3957]: segfault at fffde10d ip b80623ba sp afc9dfe0 error 6 in nscd[b8052000+1c000] Notebook: Fujitsu Siemens Amilo Si 1520 Graphics: Fujitsu Siemens Mobile 945GM/GMS/GME, 943/940GML Express Monitor: QUANTADISPLAY LCD Monitor 1280x800@60Hz Wireless: Intel PRO/Wireless 3945ABG Network Connection Sound: 82801G (ICH7 Family) High Definition Audio Controller Desktop: gnome-desktop-2.22.1-14 YaST GUI: yast2-qt-2.16.45-2 OS: openSUSE 11.0 (i586) Beta2 VERSION = 11.0 Kernel: 2.6.25-26-pae libnscd-2.0.2-76 nscd-2.8-9
I can confirm this fact: orodruin:~ # grep nscd /var/log/messages Jun 20 23:24:57 orodruin kernel: nscd[3004]: segfault at b7ffa10c ip b7f957dc sp afdd2ff0 error 4 in nscd[b7f85000+1c000] Jun 23 09:11:52 orodruin kernel: nscd[3058]: segfault at 2120d937 ip b80e999d sp afd24180 error 4 in nscd[b80da000+1c000] Jun 24 10:05:37 orodruin kernel: nscd[2908]: segfault at b7fc82f8 ip b7f3b7bb sp afb75004 error 4 in nscd[b7f2b000+1c000] Jun 25 08:59:20 orodruin kernel: nscd[3480]: segfault at 9b0 ip b80727c1 sp afcab014 error 4 in nscd[b8062000+1c000] OS: openSUSE 11.0 GM (i586) Kernel: 2.6.25.5-1.1-default nscd's rpms: nscd-2.8-14.1, libnscd-2.0.2-81.1 Fresh installation, no update from 10.3.
Doesn't seem to happen here anymore: nscd-2.8-14.1 libnscd-2.0.2-81.1 # cat /var/log/messages | grep nscd Jun 24 18:08:19 workstation6l kernel: type=1505 audit(1214323691.502:9): operation="profile_load" name="/usr/sbin/nscd" name2="default" pid=1838 Jun 25 07:02:10 workstation6l kernel: type=1505 audit(1214370121.783:9): operation="profile_load" name="/usr/sbin/nscd" name2="default" pid=1802
Andrea, can you collect and provide some nscd cores, please? (E.g. ulimit -c unlimited; nscd.)
I can try, but the segfault appears very randomly. I'm not familiar with process segfaults, so it is better if you tell me exactly what commands I must use to collect informations you need (and where such informations are stored). Now I've launched the "ulimit -c unlimited;nscd" command, and I hope nscd generates a segfault.
Yes (please make sure nscd was not running before; maybe to be sure, run nscd -d instead of just nscd). When nscd segfaults, you should find some core* file in the currrent directory, please gzip it and attach it in this bug.
Created attachment 224540 [details] core dump of nscd daemon Finally I've obtained the segfault for nscd -d daemon. Last lines printed by nscd are (where ori and 1000 are the ids of my user): 4678: remove GETPWBYUID entry "1000" 4678: remove GETPWBYNAME entry "ori" nscd: mem.c:392: gc: Assertion `off_alloc == off_allocend' failed. Aborted (core dumped) I do not know how much time is passed between the last remove and the failed assertion.
Created attachment 224580 [details] another nscd core dump on segfault Another segfault core dump. This time, last output lines are: 7931: remove GETPWBYUID entry "0" 7931: remove GETPWBYNAME entry "root" nscd: mem.c:392: gc: Assertion `off_alloc == off_allocend' failed. Next time I want to try with --debug option
Ops... -d and --debug are the same option... Anyway, I've started nscd again and the first printed message is: 8736: invalid persistent database file "/var/run/nscd/passwd": verification failed
Thanks. This looks to be the same as bug 387202, I think I know what the culprit is; I will prepare a patch. *** This bug has been marked as a duplicate of bug 387202 ***