Alfred M. Szmidt [Tue, Dec 31, 2002 at 08:17:16PM +0100]:
> I am using 2.5.53,ext3 on /dev/root,isofs on /mnt/dvd and df is
>
> 2.5.53 of what exactly?
the kernel. I will try 2.5.54 soon,too.
> flapp:/home/user/nico # df --version
> df (GNU fileutils) 4.0.35
>
> Could you try and see if this still presists in GNU Fileutils 4.1 [1]?
nico@flapp:~/fileutils-4.1/src $ ./df -h
Filesystem Size Used Avail Use% Mounted on
/dev/root 18G 19G 136M 100% /
/dev/cdroms/cdrom0 71M 71M 0 100% /mnt/dvd
nico@flapp:~/fileutils-4.1/src $ ./df
Filesystem 1k-blocks Used Available Use% Mounted on
/dev/root 19228276 19088984 139292 100% /
/dev/cdroms/cdrom0 72434 72434 0 100% /mnt/dvd
nico@flapp:~/fileutils-4.1/src $ ./du /mnt/dvd/ -sh
505M /mnt/dvd
yes, it does.
> Or if you feel brave GNU Coreutils (an merge of Fileutils, Shutils and
> Textutils) [2]?
nico@flapp:~/coreutils-4.5.4/src $ ./df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/root 19228276 19080532 147744 100% /
/dev/cdroms/cdrom0 72434 72434 0 100% /mnt/dvd
nico@flapp:~/coreutils-4.5.4/src $ ./df -h
Filesystem Size Used Avail Use% Mounted on
/dev/root 19G 19G 145M 100% /
/dev/cdroms/cdrom0 71M 71M 0 100% /mnt/dvd
nico@flapp:~/coreutils-4.5.4/src $ ./du /mnt/dvd/ -sh
505M /mnt/dvd/
yes, it does. but at least is the size now correct in -h.
====> now trying with 2.5.54 <====
nico@flapp:~ $ df
Filesystem 1k-blocks Used Available Use% Mounted on
/dev/root 19228276 19086992 141284 100% /
/dev/cdroms/cdrom0 72434 72434 0 100% /mnt/dvd
nico@flapp:~ $ df -h
Filesystem Size Used Avail Use% Mounted on
/dev/root 18G 19G 137M 100% /
/dev/cdroms/cdrom0 71M 71M 0 100% /mnt/dvd
nico@flapp:~ $ df --version
df (fileutils) 4.1
and the cdrom is again 505 MB big...
nico@flapp:~/coreutils-4.5.4/src $ ./df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/root 19228276 19087008 141268 100% /
/dev/cdroms/cdrom0 72434 72434 0 100% /mnt/dvd
nico@flapp:~/coreutils-4.5.4/src $ ./df -h
Filesystem Size Used Avail Use% Mounted on
/dev/root 19G 19G 138M 100% /
/dev/cdroms/cdrom0 71M 71M 0 100% /mnt/dvd
nico@flapp:~/coreutils-4.5.4/src $ ./du /mnt/dvd/ -sh
505M /mnt/dvd/
==> so coreutils at least prints correct human values...
I am still thinking it could be a kernel bug..so I am rebooting to
2.4.21-pre2.
But this does not make a difference...
> If it still exists in those, could you try debugging
> it?
nice,1002 lines of code in df.c.. okay, let's have a look into it...
(currently I think it _could_ be a kernel bug, so I will pass this message
alon to lkml)
I had a quick view at the code, but didn't have the time to look into
it closely. Perhaps the next days!
Am I the only one where df reports wrong disk space?
Nico
--
Please send your messages pgp-signed and/or pgp-encrypted (don't encrypt mails
to mailing list!). If you don't know what pgp is visit http://www.gnupg.org.
(public pgp key: ftp.schottelius.org/pub/familiy/nico/pgp-key)
nice,1002 lines of code in df.c.. okay, let's have a look into it...
(currently I think it _could_ be a kernel bug, so I will pass this message
alon to lkml)
There is no reason why you should read all of df.c, you can use the
GNU Debugger and see what is going on quickly. Possible places to set
break points might be in [coreutils]/src/df.c:show_dev and/or
[coreutils]/lib/fsusage.c:get_fs_usage.