segmentation fault in InitializeLocks

[Lars Damerow]

Hi folks,

I sent a mail about this a few weeks ago:

    http://lists.gnu.org/archive/html/help-cfengine/2005-01/msg00136.html

I didn't get any responses, so I wanted to try again.

I've rebuilt cfengine and modified our kickstart post scripts so that I'm
positive there's only one version of the Berkeley DB available while this is
happening, and that it's the same one used to build cfengine in the first
place.

I've also made sure that none of the UNIX groups referenced in the cfengine
script are referenced by name, only by number, so NIS problems or changes to
/etc/nsswitch.conf shouldn't be a factor here.

The backtrace I get out of gdb is often slightly different when I try
rekickstarting the machine. For example, the last backtrace I got was this:

#0  0x4027ceba in _int_malloc () from /lib/tls/libc.so.6
#1  0x4027c09d in malloc () from /lib/tls/libc.so.6
#2  0x400ce328 in __os_malloc () from /lib/tls/libdb-4.2.so
#3  0x400cf835 in __os_r_attach () from /lib/tls/libdb-4.2.so
#4  0x400b20b2 in __db_r_attach () from /lib/tls/libdb-4.2.so
#5  0x400cb208 in __memp_open () from /lib/tls/libdb-4.2.so
#6  0x400ada33 in __dbenv_open () from /lib/tls/libdb-4.2.so
#7  0x40088589 in __db_dbenv_setup () from /lib/tls/libdb-4.2.so
#8  0x4009e552 in __db_open () from /lib/tls/libdb-4.2.so
#9  0x40099195 in __db_open_pp () from /lib/tls/libdb-4.2.so
#10 0x080854ed in InitializeLocks () at locks.c:152
#11 0x080860fe in DeleteLock (
    name = 0x8125420 
"lock._var_cfengine_inputs_kickstart_dev_cf_systems..editfile.editfile_2741") 
at locks.c:571
#12 0x08085a1d in ReleaseCurrentLock () at locks.c:339
#13 0x0806f27b in DoEditFile (ptr=0x81678c0,
    filename=0x8167868 "/etc/sysctl.conf") at edittools.c:1231
#14 0x0806ecf2 in WrapDoEditFile (ptr=0x81678c0,
    filename=0x8167868 "/etc/sysctl.conf") at edittools.c:319
#15 0x0805385c in EditFiles () at do.c:2255
#16 0x0804d4f3 in DoTree (passes=2, info=0x809aa2f "Main Tree")
    at cfagent.c:1259
#17 0x0804b1ac in main (argc=11, argv=0xbffffa84) at cfagent.c:180
#18 0x40230ad4 in __libc_start_main () from /lib/tls/libc.so.6

While still a segfault processing database locks, it's not the same as the
backtrace I got in the previous message I sent.

I'm about to build a statically-linked version of cfengine to see if it helps,
but if anyone can give me some clues I'd be most grateful. I know I've pestered
folks here before about similar problems, so I apologize for sounding like a
broken record.

Thanks,
lars

--
lars damerow
button pusher
pixar animation studios
lars@pixar.com

"During the brief interview I was asked why I wanted to work for UPS and I
answered that I wanted to work for UPS because I like the brown uniforms.
What did they expect me to say?"            -David Sedaris