Flat File System

I am sorry, I have already did a thread about this but for some reason ie (it's a public library) doesnt let me make a reply so I did another one, I apologise.

So I have an assignment to do a very simple flat file system on unix which supports some basic functions i.e ls, cp, mv, shownig file id and etc...

I asked for some guide and I was given some pages and I think I realised the basic concept but I honestly dont know how to start. I know that I will create header files (library) and link them with c files which will have the functions that I want supported. I think that I have to allocate the space from my disk in the header files while specifying the block size and number of blocks and I probably have to think of it as a big file which contains other files so I guess I will nead a head pointer of sorts which will show the start of my file system and will move while I create files and save the functions (or a copy of it I guess so I wont lose the beginning). I dont know if my thought trail is correct but this is all I have. I dont know how to start writing code. In the lesson they just gave as some functions of c and the library the are in (open, close, creat, lseek, read and write) but I alredy knew them and I suppose that I will use them in the implementation of the functions of the file system...

So basicly I am lost... I know that I cant ask for code, so any help is gladly appreciated...

Code:

---

typedef struct {
    char name[10];  // name[0] = '\0' is a free entry
    int size;
    int blocks[100];  // max file size is 100 * 256 bytes
} dirEnt;

---

Each file has
- a name
- a size
- a number of blocks of 'storage' associated with it.

You then have

Code:

---

dirEnt files[MAXFILES];

---

So very crudely, you fread() and fwrite() the files array to the start of some large file.

Say the first file had blocks 2 and 7 associated with it.
To get the first block, you would fseek( sizeof(files) + 256 * 2 ); and read 256 bytes
To get the second block, you would fseek( sizeof(files) + 256 * 7 ); and read 256 bytes

You might want to keep a separate structure which is a list of all the free blocks in the file system, so whenever you create files (or append to files), you have a ready source of blocks to use.

Ok, I think I understand it, but wouldnt
fseek( sizeof(files) + 256 * 2 );
read the third block since it sets the seek pointer two blocks after the files array so if read after that I would read the third block, or not?

Quote:

---

Originally Posted by Phoenix_Rebirth

Ok, I think I understand it, but wouldnt
fseek( sizeof(files) + 256 * 2 );
read the third block since it sets the seek pointer two blocks after the files array so if read after that I would read the third block, or not?

---

Yes, but if you count blocks as 0, 1, 2 then the third block is what you want.

--
Mats

Ah, yes. If you look at it like this yes (classic c always starting from 0... :P ).
Oh another dumb question, If I remember correct for allocating memory I can use malloc. Can I also use it for space on a hard drive or must I go another way (probably more complicated)

Depends on how much seeking you want to do.

At a minimum, 1 directory entry and one block of file data.
At a maximum (since this is just a simulation), the whole virtual file system.

Malloc can only allocate memory and not hard drive space, if that's what you're asking.

Quote:

---

Originally Posted by Phoenix_Rebirth

Ah, yes. If you look at it like this yes (classic c always starting from 0... :P ).

---

Yes, there's a difference between the second sector [index 1] and sector number 2 [index 2].

Quote:

---

Oh another dumb question, If I remember correct for allocating memory I can use malloc. Can I also use it for space on a hard drive or must I go another way (probably more complicated)

---

Not really - if we assume that you have "a large file that is your disk", then you have to keep track of which block you use for what in some sense. There are many methods which all works more or less well under various circumstances. Two of the more popular methods are:
1. Bitmap - one bit for each block, 0 -> free, 1 -> used, then search the bitmap until you find a free block. Drawback is that the list gets very large if you have large disks [e.g. 4GB of 512byte sectors -> 1MB of bitmap to search within].
2. List of free blocks - Use some sort of "linked list" where you store a (set of) block(s) that are free. If you use a structure like this:

Code:

---

struct freelist
{
    int nextfreelistblock;  /* disk block that contains the next list of free blocks. 0 means no more list */
    struct freeblocks {   
      int nblocks;              /* Number of contiguous blocks that are free */
      int firstblock;          /* First block in this particular free section */
    } freeblocks[1];
};

---

You can put multiple structs in one block, so you don't take up that much space.

--
Mats