bug-fileutils
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Error in du running on Linux for a mounted HP-UX filesystem


From: Jens Liebehenschel
Subject: Re: Error in du running on Linux for a mounted HP-UX filesystem
Date: Thu, 21 Jun 2001 14:38:12 +0200 (CEST)

Jim,

> Thanks for the detailed bug report!
> If someone can debug it, that'd be great.

unfortunately I cannot give you access to our system, but I could try
to help you debugging the du program.

I installed the sources of the fileutils and had a look at the program
flow. I created a file of length 10010 Bytes and copied the file to my
local disk (Linux) and to a mountd disk (HP-UX 10.20). Then I compared
where the the du program has different behaviour for the local file and
the file on the mounted disk.

In the function function count_entry in du.c we have to compute the size
of the file:
...
size = ST_NBLOCKS (stat_buf);
tot_size += size;
...

I debugged the program and found the following results:

elements of     |       file on                 file on
stat_buf        |       HP-UX 10.20 FS          Linux FS
----------------+-------------------------------------------------------------
st_dev          |       3                       2049
...             |       ...                     ...
__st_ino        |       1708                    36287
...             |       ...                     ...
st_size         |       10010                   10010
st_blksize      |       8192                    4096
st_blocks       |       10                      24
...             |       ...                     ...

In this table you find all variables in the struct stat with different
values and the variable st_size.
Unfortunately, I did not find any information on the struct stat. I think
- if you have access to the documentation on the elements in stat - it is
an easy task to fix the bug. Perhaps you have to take the elements
st_blksize and st_dev (is this a number describing the filesystem ?) into
account.

> Otherwise, if someone gives me temporary access
> to such a system, I'll do it.

I hope we can solve the problem this way. If you can provide some
information an struct stat, I can think about the problem.
But I should give you a warning: neither I am familar with system
programming nor I have access to many differnt systems to test the
changes.
On the other hand I am sure that we are able to solve the problem.

In any case, please let me know what you think about solving the problem
together.

Regards
   Jens

------------------------------------------------------------------------
  Dr. Jens Liebehenschel                        Phone: 069 / 798-28253
  Fachbereich Biologie und Informatik           Fax:   069 / 798-28870
  Institut für Informatik
  Universität Frankfurt
  Postfach 11 19 32
  D-60054 Frankfurt am Main

  Email: address@hidden
  WWW: http://boa.sads.informatik.uni-frankfurt.de:8000/jens/jens.html
------------------------------------------------------------------------




reply via email to

[Prev in Thread] Current Thread [Next in Thread]