Thread: Some (Much) help for implementing a flat filesystem

The file system will support certain number of files and the file name will also have a limited length. Free space will be managed by a file allocation table and will have to hand over an API, a library, utilities used by a basic shell. The file system is comprised by a Directory Service, File Service and Block Service. At the highest level the user will issue commands in the shell which will call functions from the directory and file service which in turn will call functions from the block service where they are responcibe for moving to/from the "virtual" disk.
The shell will be very basic and will show a prompt where the user can input commands. Commands supported will be: my_mkfs, my_ls, my_cp, my_mv, my_rm,
my_df, my_cat, my_echo, my_mount, my_umount, my_help, my_quit.

Directory Service will correspond ufid with filenames. It will be a flat file system without access control. Directory service will offer the following methods
• int my_files (char **filenames);
• int my_create (const char *filename); create a new file with 0 bytes as size
• int my_delete (const char *filename);
• int my_open (const char *filename);

File Service:
• int my_close (int ufid);
• int my_file_size (int ufid);
• int my_read (int ufid, char *buf, int num, int pos);
• int my_write (int ufid, char *buf, int num, int pos);

The file system will have to conserve some basic structures for servicing requests using block service functions. These will be DIrectory table, file allocation table, free block list and file descriptor table. The first three will be stored on the disk and will be loaded on main memory when mounting the system and will be written when unmounting with their changes. The last will be only on the main memory
Directory Table: filename, size, fat index(first block of file on disk)
File Allocation Table: global index; int fat[nu,ber of blocks] Each cell/block contains the index for the next block of the file or eof if last one
Free Block List: char [number of blocks] ; 1:used, 0:unused
File Descriptor Table: ufid, directory table index

The disk which is a huge file defined by the user when calling my_mkfs(blocks, blocksize)
Block Service will supply functions for allocation, deallocation, read and write of a
block. They will be used by the Directory and File Services
For bringing the blocks and writing them lseek(), read() and write() from UNIX will be used. We will lseek() in block sizes.

This is what I have done up to now "http://www.4shared.com/file/83812550/b371d6a5/filesystem.html" which are only the shell recognizing the commands used and I begun trying to make the services but after creating the file/disk I am stucked. I try to make the functions for creating and writing the files but I cant see to grasp how the blocks are written. I mean A file is created with 0 bytes so it's just a record for the time being. Then I open the file. When I write in it do I store the writing in a buffer of sorts in main memory and when closed I calculate the size and accordingly find blocks and write on them. Then I will have to have a large buffer in main memory. And even so say my block size is 512 bytes and I have opened a file where I wrote data worth 1000 bytes how dow I write them in the blocks. I mean how do I write first the first 512 and then only the remaining 498. Also the write works with a poitner to char right.
And also about the basic structures I want to save which are not character but are int or structs how dow I write them as files...