Noticed this the other day; if I run 'du -h<random dir>' and compare
the files sizes to those reported by 'ls -sh1' they don't match. For
example:
# ls -sh1 PROJECTS/LISP/code/*.lsp
4.2K PROJECTS/LISP/code/fsvm.lsp
205B PROJECTS/LISP/code/hello.lsp
1.7K PROJECTS/LISP/code/recursion.lsp
1.1K PROJECTS/LISP/code/test.lsp
4.5K PROJECTS/LISP/code/tictactoe
# du -h PROJECTS/LISP/code/*.lsp
6.0K PROJECTS/LISP/code/fsvm.lsp
2.0K PROJECTS/LISP/code/hello.lsp
2.0K PROJECTS/LISP/code/recursion.lsp
2.0K PROJECTS/LISP/code/test.lsp
6.0K PROJECTS/LISP/code/tictactoe.lsp
Not only do they not match, they don't seem to be off by any set amount
or ratio. I don't have BLOCKSIZE set and my filesystem is the default
FFS. Some system specifics below:
# sysctl kern. | head -n 4
kern.ostype = NetBSD
kern.osrelease = 5.0_STABLE
kern.osrevision = 500000003
kern.version = NetBSD 5.0_STABLE (GENERIC) #0: Fri Apr 2 03:45:48 UTC 2010
# disklabel wd0 | grep -B1 "^ a"
# size offset fstype [fsize bsize cpg/sgs]
a: 3984561 63 4.2BSD 2048 16384 0
# dumpfs -s / | head -n 9
file system: /dev/rwd0a
endian little-endian
magic 11954 (UFS1) time Fri Sep 30 17:03:32 2011
superblock location 8192 id [ 4bb67bda 87b5b2c ]
cylgrp dynamic inodes 4.4BSD sblock FFSv2 fslevel 4
nbfree 90278 ndir 1798 nifree 223202 nffree 164
ncg 11 size 996140 blocks 980555
bsize 16384 shift 14 mask 0xffffc000
fsize 2048 shift 11 mask 0xfffff800
--
Is this a known issue/bug or is there something I've overlooked?