jkimball@SRC.Honeywell.COM (John Kimball) (11/20/90)
I'm having a problem with mthreads, after re-installing it. mthreads
always dies wtih "signal 11".
Situation: trn was installed here originally by an admin in another group.
I (and some others) got hooked, but I didn't have/make the time to install
it properly as part of our standard news setup. Time passes. In
mid-October, I upgraded us from Bnews to Cnews (and to the latest NNTP). I
did some directory re-organization, so I decided to rebuild trn
simultaneously. I grabbed the distribution from caesar.cs.montana.edu,
built it with our new directory setup, and tried to run mthreads. mthreads
dies with "signal 11, segmentation fault". (One other difference --
previously mthreads was running as root, I've been running it as news.) I
tried running it under gdb to get a feel for the problem (information
below), but nothing obvious popped out.
Is this a familiar problem? Any suggestions? (I'm hoping that this is a
known situation, otherwise it's "Read the source, Luke", and it'll be a
while before I can make time for that, sigh . . .)
Thanks!
:::::::
mt.log:
:::::::
Oct 29 15:57 mthreads daemon started for 'all'
Oct 29 15:57 ** interrupt 11 **
Oct 29 16:08 mthreads daemon started for 'comp'
Oct 29 16:09 ** interrupt 11 **
Nov 4 15:34 mthreads single pass for 'all'
Nov 4 15:34 ** interrupt 11 **
Nov 9 12:20 mthreads single pass for 'all'
Nov 9 12:29 mthreads single pass for 'all'
Nov 9 12:37 mthreads single pass for 'all'
Nov 9 12:39 mthreads single pass for 'all'
:::::::::::::::::
running under gdb
:::::::::::::::::
(gdb) run -v
Starting program: /news/trn/mthreads -v
Program received signal 11, Segmentation fault
Reading in symbols for mt-read.c...done.
0x40b0 in read_roots () (mt-read.c line 221)
(gdb) print last_root
$1 = (struct Root *) 0x25f40
(gdb) print root
$2 = (struct Root *) 0x320c8
(gdb) print *last_root
$3 = {link = 0x26c64, articles = 0x26c7c, subjects = 0x26c94, root_num = 158892, thread_cnt = 2, subject_cnt = 27844, seq = 2}
(gdb) print *root
$4 = {link = 0x0, articles = 0x0, subjects = 0x0, root_num = 131073, thread_cnt = 0, subject_cnt = 0, seq = 1}
(gdb) print last_root->link
$5 = (struct Root *) 0x26c64
(gdb) print *last_root->link
$6 = {link = 0x26c7c, articles = 0x26c74, subjects = 0x2, root_num = 8, thread_cnt = 4096, subject_cnt = 0, seq = 0}
(gdb)
::::::::::::::::::
mt-read.c line 221
::::::::::::::::::
last_subject = subject;
}
last_subject->link = Null(SUBJECT*);
--> last_root->link = root; <------
last_root = root;
}
ret = 1;
John Kimball
domain: jkimball@src.honeywell.com Honeywell Systems and Research Center
postmaster@src.honeywell.com Computer Sciences/Software Technology
uucp: <any-smart-host>!srcsip!jkimball 3660 Technology Drive, MN65-2100
voice: +1 612/782-7343 Minneapolis, MN 55418-1006
fax: +1 612/782-7438
disclaimer: The only opinion Honeywell authorizes me to have is:
"Thermostats are Good"jkimball@SRC.Honeywell.COM (John Kimball) (12/10/90)
In article <1990Nov19.175517.26757@src.honeywell.com> jkimball@SRC.Honeywell.COM (John Kimball) writes: > >I'm having a problem with mthreads, after re-installing it. mthreads >always dies wtih "signal 11". Thanks to randy@chinet.chi.il.us (Randy Suess) and Michael E. Dobson <rdc30med@nmrdc1.nmrdc.nnmc.navy.mil> for their suggestions. There were old .thread files lying about, owned by root. Changing their ownership to news was insufficient; I needed to rm them, then mthreads was happy. (I didn't suspect that permissions could cause a segment violation problem.) John Kimball domain: jkimball@src.honeywell.com Honeywell Systems and Research Center postmaster@src.honeywell.com Computer Sciences/Software Technology uucp: <any-smart-host>!srcsip!jkimball 3660 Technology Drive, MN65-2100 voice: +1 612/782-7343 Minneapolis, MN 55418-1006 fax: +1 612/782-7438 disclaimer: The only opinion Honeywell authorizes me to have is: "Thermostats are Good"
spike@world.std.com (Joe Ilacqua) (12/14/90)
In article <1990Dec10.150638.16632@src.honeywell.com> jkimball@SRC.Honeywell.COM (John Kimball) writes: <Thanks to randy@chinet.chi.il.us (Randy Suess) and Michael E. Dobson ><rdc30med@nmrdc1.nmrdc.nnmc.navy.mil> for their suggestions. There were <old .thread files lying about, owned by root. Changing their ownership to >news was insufficient; I needed to rm them, then mthreads was happy. (I <didn't suspect that permissions could cause a segment violation problem.) The same thing just happen to me and it overflowed the disk with 24 MB of: Dec 13 10:34 ** interrupt 11 ** This is a excellent answer to "Why run NN when there is TRN?". 'Cause 'nnmaster' has yet to f*ck me over where as 'mthreads' does it all the time. ->Spike (In a ranting mood) -- The World - Public Access Unix - +1 617-739-9753 24hrs {3,12,24,96,192}00bps