Laboratory assignment  3 -  Filesystem

Purpose

To investigate a filesystem.

Local  filsystem

This exercise must be done on a computer with a local file system.

Assignment:

1. Investigate with the mount command what the file system on your computer looks like. How is the tree of  different mounted filesystems put together? Which parts are local and which parts are fetched over the network. Which special purpose filesystems exist?
   
2. Write a  C program that duplicates some of the functions of the Linux stat(1) command. The program should take the name of a file as parameter and do stat(2) or lstat(2) on this file. The results should be written to the terminal in a suitable formate (hexadecimal is probabely best) for analyzing the file status. Use this program on simple files, directories, special files and symbolic links to find out the differences between them.
   
3. Write a C program that creates a file and writes 100 bytes with known content (not binary zero), seeks to position 40000 and writes another 100 bytes of nonzero data. After this the file is closed and the program terminated. This program creates a file with a "hole" in it. Create such a file in the local filesystem. Use the command od to list the content of the file. Run ls -l and take note of the length of the file. Use the program from step 2 to check the  lenght of the file and how many blocks the file use. Do the results agree with the theory? Write a simple copying program that copies the file by reading and writing one character in a loop until end-of-file. Copy the file with the "hole" to new file with your copying program. Do the same checks as previously on the copied file. What have happened? Copy the file with the "hole" to a new file using the cp command. Do the same checks again. What have happened?
   
4. Write a C program that searches the local filesystem tree and gathers information about all the files. The information is used to calculate the wasted space as a function of block size and to recreate table 8.12 (page 367) from the course book. To calculate data waste for different block sizes, the calculation must be done for every file as the filesystem is searched and summed up for each block size. The filesystem on the lab machines (Reiserfs) is different from the BSD file system. This do not matter if only the size of each file is used to calculate the waste for every block size. Assume that every file and directory requires one inode and that the size of the inode is 128 bytes. You must avoid searching NFS filesystems because there are too many files there and it would generate to high load at the file server. There are also some other special filesystems that should be avoided. Only "normal" file systems should be searched. One way to avoid the unwanted filesystems is to make a list of the beginning path name component of these filesystems that the program use to avoid them. Also, you should not follow symbolic links! Because it is difficult to get everything correct from the beginning, the program should use two arguments to limit the search. The first argument gives the total number of files to search and the second argument indicates how often information about the current file name should be written to the terminal. For example   <program_name 100 1>  means that 100 files are processed and information is written for every file.   <program_name 1000 50>  means that 1000 files are processed and information is written for every 50:th file. When everything works both values are set high for the final run. Check that your results are reasonable and try to explain any differences between your values and the values from the book.
   

Reporting: