Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 18 Apr 2000 12:42:46 -0400
From:      "Matthew Zahorik" <matt@thebiz.net>
To:        <alpha@freebsd.org>
Subject:   HELP!
Message-ID:  <00c501bfa955$203546e0$1401eed8@mahatma>

next in thread | raw e-mail | index | archive | help
I have a reliable kernel panic when copying a file.

These are diskless servers.  /etc is a MFS filesystem.  /sandbox/etc is NFS
mounted.

[shell1]:DING~:~# cd /sandbox/etc
[shell1]:12:00pm:/sandobx/etc# cp /etc/sshd_config .

fatal kernel trap:

    trap entry = 0x2 (memory management fault)
    a0         = 0x0
    a1         = 0x1
    a2         = 0xffffffffffffffff
    pc         = 0x0
    ra         = 0x1
    curproc    = 0xfffffe0011456e00
        pid = 338, comm = cp

ddbprinttrap from 0x0
ddbprinttrap(0x0, 0x1, 0xffffffffffffffff, 0x2)
panic: trap
panic
Stopped at      Debugger+0x2c:  ldq     ra,0(sp) <0xfffffe0012afd3f0>\
<ra=0xfffffc00004abcc0,sp=0xfffffe0012afd3f0>
db> trace
Debugger() at Debugger+0x2c
panic() at panic+0x100
trap() at trap+0x610
XentMM() at XentMM+0x20
(null) at 0x1
db>

You can cat/more, ls -l, and otherwise manipulate the file.  You can't copy
it for the life of me.  (I haven't tried mv'ing it)

I tried different destinations, like cp /etc/sshd_config /tmp and that
still crashes.

This was a 4.0-RELEASE kernel.  I CVS'd out the current kernel (cd /usr/src,
cvs co sys) just a couple of moments ago and tried that - no change, so it's
not a bug that has a fix recently committed.

These machines were really close to hitting production with FreeBSD instead
of Linux on them, but I can't have people rebooting the machines at will (:
Any pointers would be appreciated so I can fix this quickly.

There's no PC, or any other address that points to the bad code, so I really
can't track it down.  Something is getting a null pointer, and crashing.
Since it's bringing down the machine, I assume it is in the kernel space. (I
assume a null pointer access in user space can't bring the machine down)  My
gut tells me it's MFS related.

I can't reproduce this on my Intel boxes - either the bug doesn't exist or
the setup isn't identical.

- Matt

--
Matthew Zahorik  Director of Systems and Networking - BiznessOnline.com
matt@thebiz.net  President of AlbanyNet Inc. - a BiznessOnline subsidiary
maz@albany.net        Voice: (518) 292-1001  Fax: (518) 626-0793



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-alpha" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?00c501bfa955$203546e0$1401eed8>