Re: [BUG] garbage instead of zeroes in UFS

["Tomasz Kvarsin" <kvarsin@gmail.com>]

On 12/20/06, Andrew Morton <akpm@osdl.org> wrote:
> On Wed, 20 Dec 2006 14:04:06 +0300
> "Tomasz Kvarsin" <kvarsin@gmail.com> wrote:
>
> > Forgot to say I use linux-2.6.20-rc1-mm1
> >
> > On 12/20/06, Tomasz Kvarsin <kvarsin@gmail.com> wrote:
> > > I have some problems with write support of UFS.
> > > Here is script which demonstrate problem:
> > >
> > > #create image
> > > mkdir /tmp/ufs-expirements && cd /tmp/ufs-expirements/
> > > for ((i=0; i<1024*1024*2; ++i)); do printf "z"; done > image
> > >
> > > #build ufs tools
> > > wget 'http://heanet.dl.sourceforge.net/sourceforge/ufs-linux/ufs-tools-0.1.tar.bz2'
> > > && tar xjf ufs-tools-0.1.tar.bz2 && cd ufs-tools-0.1
> > > wget http://lkml.org/lkml/diff/2006/5/20/48/1 -O build.patch
> > > patch -p1 < build.patch && make
> > >
> > > #create UFS file system on image
> > > ./mkufs -O 1 -b 16384 -f 2048 ../image
> > > cd .. && mkdir root
> > > mount -t ufs image root -o loop,ufstype=44bsd
> > > cd root/
> > > touch a.txt
> > > echo "END" > end.txt
> > > dd if=./end.txt of=./a.txt bs=16384 seek=1
> > >
> > > and at the end content of "a.txt" not only  "END" and zeroes,
> > > "a.txt" also contains "z".
> > >
> > > The real situation happened when I deleted big file,
> > > and create new one with holes. This script just easy way to reproduce bug.
> > >
>
> Does 2.6.20-rc1 have the same problem?
>

Yes.
Actually, if it is important,
I start my searching of stable UFS write support from 2.6.10 + RH patches.
It just hang up time to time,
I try 2.6.19, it is pretty stable, but has this bug,
2.6.20-rc1 has it.