Subject: Re: LFS partition limitations
To: Manuel Bouyer <>
From: Tracy J. Di Marco White <>
List: current-users
Date: 10/01/2000 16:45:42
}On Sun, Oct 01, 2000 at 09:23:59AM -0500, Tracy J. Di Marco White wrote:
}> I updated my source for 1.5 to be current as of yesterday (September 30,
}> 2000) since it looked like there were some updates since the snapshot of
}> 1.5ALPHA2.  I newfs_lfs'd the disk and mounted it.  It again couldn't
}> handle the newlfs until I doubled the logical segment size from the
}> default.  The machine panic'd again, this time it wasn't the scsi
}> controller's fault, probably. :)  I suspect it was lfs_cleanerd, at a
}> guess, as I mounted the filesystem and then did nothing else to touch
}> the filesystem, and after a little while it crashed.
}> lfs_writeinode: negative bytes (segment 0 short by 128)
}> panic: lfs_writeinode: negative bytes
}> Only this time I didn't get a core dump.
}I've seen this but I though it was fixed by the last commits :(

I'd hope it might be.  Just to be clear, I'm using 1.5 branch, not
1.5[ABCDE].  It looked like most of the changes had been pulled up to
the branch, though.

I recreated the lfs filesystem again, just in case, and mounted it again.
It didn't crash in the first minute or so of being mounted, so I rebooted
lyra to see what would happen.  I get dropped to single user mode because
fsck_lfs can't deal with the filesystem:
:~# fsck_lfs -d /dev/raid0d
** /dev/rraid0d
sb0 970434535, sb1 970434535
dev_bsize = 512
lfs_bsize = 8192
lfs_fsize = 1024
lfs_frag  = 8
INOPB(fs) = 64
Segmentation fault (core dumped)

This is very repeatable.

I recompiled fsck_lfs so I could use the unstripped version.
I'm not very familiar with gdb, but I did get this:
# gdb ./fsck_lfs fsck_lfs.core
GNU gdb 4.17
Copyright 1998 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "i386--netbsd"...(no debugging symbols found)...
Core was generated by `fsck_lfs'.
Program terminated with signal 11, Segmentation fault.
#0  0x804f9e0 in bzero ()
(gdb) bt
#0  0x804f9e0 in bzero ()
#1  0x2aba9800 in ?? ()
#2  0x804aeb3 in checkfilesys ()
#3  0x804adf5 in main ()
#4  0x80481c5 in ___start ()

I assumed that since the machine was shut down cleanly, the filesystem
was more or less ok (since nothing had been written to it yet), so I
went ahead and mounted it.  It hasn't crashed in the last five minutes
or so, and now I'm copying NetBSD's pkgsrc xsrc & src to see what happens.
I'm running lfs_cleanerd in verbose debugging mode, in case there are
things to see.

Tracy J. Di Marco White
Project Vincent Systems Manager