[Bug-tar] an exactly duplicated file was created by tar

[Oliver Kullmann]

Hello,

I have created a tar-ball from a directory on one machine, running
1> tar --version
tar (GNU tar) 1.26
1> uname -a
Linux linux-vp82.site 2.6.39.3-0.5-desktop #1 SMP PREEMPT Sun Jul 31 02:04:11 
BST 2011 x86_64 x86_64 x86_64 GNU/Linux
(Suse 11.4)

and unpacked it on a machine running
2> tar --version
tar (GNU tar) 1.19
2> uname -a
Linux cs-oksvr 2.6.25.20-0.1-default #1 SMP 2008-12-12 20:30:38 +0100 x86_64 
x86_64 x86_64 GNU/Linux
(Suse 11.4)

The unpacked archive contains an additional file, which is an exact(!) 
duplicate of an
existing file. The details are as follows:

We have a directory
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738
which contains
1> cd SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738/
1> ls -l
total 55756
-rw-r--r-- 1 kullmann users 23720739 Aug 17 20:41 Data
-rw-r--r-- 1 kullmann users       32 Aug 16 01:47 F
drwxr-xr-x 2 kullmann users 32620544 Aug 17 04:41 Instances
-rw-r--r-- 1 kullmann users       33 Aug 17 20:36 Md5sum
-rw-r--r-- 1 kullmann users        8 Aug 17 04:41 N
-rw-r--r-- 1 kullmann users      117 Aug 16 01:47 Parameters
-rw-r--r-- 1 kullmann users     1285 Aug 17 04:41 Result
-rw-r--r-- 1 kullmann users      651 Aug 17 20:41 Statistics
-rw-r--r-- 1 kullmann users   715832 Aug 16 01:47 
VanDerWaerden_pd_2-4-13_520.cnf
-rw-r--r-- 1 kullmann users      219 Aug 16 01:47 Version
1> ls -1 Instances/ | wc -l
 1365418

Via 
1> tar -czf 
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738.tar.gz 
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738

we obtain
1> ls -l 
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738.tar.gz 
-rw-r--r-- 1 kullmann users 103401616 Aug 18 03:53 
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738.tar.gz
1> md5sum 
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738.tar.gz 
99dfe82839d104abc8279dabe70a35c4  
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738.tar.gz

On the second machine we untar it:
2> ls -l 
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738.tar.gz 
-rw-r--r-- 1 csoliver users 103401616 2011-08-18 08:06 
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738.tar.gz 
2> md5sum 
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738.tar.gz 
99dfe82839d104abc8279dabe70a35c4  
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738.tar.gz
2> tar -xzf 
SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738.tar.gz
2> cd SplitViaOKsolver_D36VanDerWaerden_pd_2413_520cnf_2011-08-16-014738/
2> ls -l
total 54968
-rw-r--r-- 1 csoliver users 23720739 2011-08-17 20:41 Data
-rw-r--r-- 1 csoliver users       32 2011-08-16 01:47 F
drwxr-xr-x 2 csoliver users 31748096 2011-08-18 17:21 Instances
-rw-r--r-- 1 csoliver users       33 2011-08-17 20:36 Md5sum
-rw-r--r-- 1 csoliver users        8 2011-08-17 04:41 N
-rw-r--r-- 1 csoliver users      117 2011-08-16 01:47 Parameters
-rw-r--r-- 1 csoliver users     1285 2011-08-17 04:41 Result
-rw-r--r-- 1 csoliver users      651 2011-08-17 20:41 Statistics
-rw-r--r-- 1 csoliver users   715832 2011-08-16 01:47 
VanDerWaerden_pd_2-4-13_520.cnf
-rw-r--r-- 1 csoliver users      219 2011-08-16 01:47 Version
2> ls -1 Instances/ | wc -l
1365419

We see that one file has been added to the subdirectory Instances.
The original directory Instances contained text-files named "1", "2", ..., 
"1365418".
The (false) copy-directory Instances contains the file "1025185" now in two 
apparently
identical instances:
2> cd Instances
2> > ls -ali 1025185*
7551518 -rw-r--r-- 1 csoliver users 164 2011-08-16 21:23 1025185
7551518 -rw-r--r-- 1 csoliver users 164 2011-08-16 21:23 1025185


I think that's a bug. Hopefully you can do something about it.

As you can see, the files and directories are somewhat bigger. If you wish,
I can put them on some web page where you can download it.

Best wishes

Oliver

-- 
Dr. Oliver Kullmann
Department of Computer Science
College of Science, Swansea University
Faraday Building, Singleton Park
Swansea SA2 8PP, UK
http://cs.swan.ac.uk/~csoliver/