Subject: i386 1.4Q hangs nonrandomly?
To: maximum entropy <entropy@zippy.bernstein.com>
From: Mason Loring Bliss <mason@acheron.middleboro.ma.us>
List: current-users
Date: 01/24/2000 12:23:59
On Fri, Jan 21, 2000 at 10:22:44PM -0500, maximum entropy wrote:

> Every time my system has hung with 1.4P it has been during a "make
> build" -- high disk IO and lots of process creation.

I'm seeing the same behaviour with 1.4Q. If I run something that matches what
you described, above, I crash. It takes a "make -j 3" or something to do it.
I can do lesser things without problem.

I have crash dumps available, if this is still a vaguely unknown problem.

On coming back up, I get:

Jan 23 15:04:28 acheron savecore: reboot after panic: trap

Now, something which MAY be unrelated:

Jan 23 23:29:01 acheron /netbsd: Data modified on freelist: word 0 of object 0xf05b0550 size 4 previous type UVM amap (0xfeadbeef != 0xdeadbeef)
Jan 23 23:29:54 acheron /netbsd: Data modified on freelist: word 0 of object 0xf05b0550 size 4 previous type UVM amap (0xfeadbeef != 0xdeadbeef)
Jan 23 23:29:55 acheron /netbsd: Data modified on freelist: word 0 of object 0xf05b0550 size 8 previous type UVM amap (0xfeadbeef != 0xdeadbeef)
Jan 23 23:32:53 acheron /netbsd: Data modified on freelist: word 0 of object 0xf05b0550 size 4 previous type UVM amap (0xfeadbeef != 0xdeadbeef)
Jan 23 23:34:40 acheron /netbsd: Data modified on freelist: word 0 of object 0xf05b0550 size 4 previous type UVM amap (0xfeadbeef != 0xdeadbeef)

Is it possible that this last is the consequence of bad RAM? The RAM in
question has lived through a system which died as a consequence of my
zapping it with static. However, the memory test when the system boots
up doesn't indicate problems. Today (or something) I'm going to whip
up something to allocate huge chunks of memory and write some pattern
of data to it, to try to rule out bad memory. (The chances of my getting
at ALL of the RAM are essentially NIL, but I can't think of anything more
useful to do to check it.)

In the past couple days I've also seen some stuff like, for instance, gcc
dying with a "signal 10." I haven't seen this very recently, though - I
believe I'm running a new kernel since the last time I saw this.

-- 
    Mason Loring Bliss  mason@acheron.middleboro.ma.us  If a cow laughed, would
awake ? sleep : dream;  http://acheron.ne.mediaone.net  milk come out her nose?