FS#987 - libnss-mysql+pacman woes

Attached to Project: Pacman
Opened by Dale Blount (dale) - Thursday, 10 June 2004, 21:21 GMT
Task Type Bug Report
Category
Status Closed
Assigned To Judd Vinet (judd)
Architecture not specified
Severity Medium
Priority Normal
Reported Version 0.7 Wombat
Due in Version Undecided
Due Date Undecided
Percent Complete 0%
Votes 0
Private No

Details

Core was generated by `pacman -Sy'.
Program terminated with signal 11, Segmentation fault.
#0 0x00000000 in ?? ()

pacman -Sy causes segfault in pacman (libtar maybe), but works *sometimes* with nscd started.

we got more debug last time we tried to trace it back, something in geteuid() i think...

a dynamic build of pacman works as far as I can remember...
This task depends upon

Closed by  Judd Vinet (judd)
Monday, 23 August 2004, 17:46 GMT
Reason for closing:  Fixed
Comment by Jan de Groot (JGC) - Friday, 11 June 2004, 18:56 GMT
pointed that out before, but it seems the developers want a static pacman instead of a dynamic one.

What happens is that your static linked pacman does some requests to nsswitch, which links in a dynamic mysql module. Since pacman only has reserved space for the nsswitch part without additional dynamic modules, it segfaults (it also segfaults with a newer version of glibc, since the newer glibc has a larger nsswitch part I guess).

Two solutions:
- Use nscd
- Use a dynamic pacman
Comment by Dale Blount (dale) - Friday, 11 June 2004, 20:02 GMT
Thanks Jan,

Any idea why nscd fails some of the time?
Or how we'd upgrade glibc with a dynamic pacman?

Comment by Jan de Groot (JGC) - Friday, 11 June 2004, 22:01 GMT
I have the same setup with nss_ldap, but I never have any problems when NSCD is running.
About upgrading glibc with dynamic pacman: it should do, but I didn't try it. Does it give a "text file in use" error message?
Comment by Dale Blount (dale) - Saturday, 12 June 2004, 01:39 GMT
I dunno, never tried it... I just have heard it doesn't work.

About 1/4 times I can reproduce a crash WITH nscd running... maybe I'll try to trace that back and see if it differs.
Comment by Dale Blount (dale) - Monday, 26 July 2004, 15:32 GMT
Shouldn't this at least be consistant?

[root@host root]# /etc/rc.d/nscd start
:: Starting nscd [DONE]
[root@host root]# pacman -Sy
:: Synchronizing package databases...
current [#################################################################################################] 100% 37K 294.9K/s 00:00:00
Segmentation fault
[root@host root]# rm /tmp/pacman.lck
[root@host root]# pacman -Sy
:: Synchronizing package databases...
current [#################################################################################################] 100% 37K 6401K/s 00:00:00
Segmentation fault
[root@host root]# rm /tmp/pacman.lck
[root@host root]# pacman -Sy
:: Synchronizing package databases...
current [#################################################################################################] 100% 37K 6173K/s 00:00:00
extra [#################################################################################################] 100% 145K 8091K/s 00:00:00
[root@host root]# /etc/rc.d/nscd stop


I can reproduce this (as far as the inconsitancy goes) on multiple systems.
Comment by Dale Blount (dale) - Monday, 23 August 2004, 15:18 GMT
pacman 2.8.4 seems to fix this (dynamic pacman binary).

no more segfaults using libnss-mysql for me

Loading...