OS development: Implementing FAT12 Write Function

Question

So im pretty new at OS development, right now im writing my FAT12 filesystem code. I got all the writing to floppy disk using the FDC code working but I cant seem to wrap my head around how I should proceed when writing a file to disk.

I know where my root directory is and all its related info, but how should I look for the correct sector where to write my file? And after that, how should I add that entry to the FAT?

This is the code I use to traverse the Root Directory and find a file:

FILE fsysFatDirectory (const char* DirectoryName) {

        FILE file;
    unsigned char* buf;
    PDIRECTORY directory;

    //! get 8.3 directory name
    char DosFileName[11];
    ToDosFileName (DirectoryName, DosFileName, 11);
    DosFileName[11]=0;

    //! 14 sectors per directory
    int sector;
    for (sector=0; sector<14; sector++) {

        //! read in sector of root directory
        buf = (unsigned char*) flpydsk_read_sector (_MountInfo.rootOffset + sector );

        //! get directory info
        directory = (PDIRECTORY) buf;

        //! 16 entries per sector
        int i;
        for (i=0; i<16; i++) {

            //! get current filename
            char name[11];
            kmemcpy (name, directory->Filename, 11);
            name[11]=0;

            //! find a match?
            if (kstrcmp (DosFileName, name) == 0) {

                //! found it, set up file info
                kstrcpy (file.name, DirectoryName);
                file.id             = 0;
                file.currentCluster = directory->FirstCluster;
                file.fileLength     = directory->FileSize;
                file.eof            = 0;
                file.fileLength     = directory->FileSize;

                //! set file type
                if (directory->Attrib == 0x10)
                    file.flags = FS_DIRECTORY;
                else
                    file.flags = FS_FILE;

                //! return file
                return file;
            }

            //! go to next directory
            directory++;
        }
    }

    //! unable to find file
    file.flags = FS_INVALID;
    return file;
    }

I was planning to write something similar and walk through the root directory, entry by entry until I find a spot. As for adding/walking through FAT, I know each entry represents a cluster(entry 1 = cluster 1). But I dont know if i should be walking through the FAT instead of the root directory or both.

Most of my code is based off this tutorial: http://www.brokenthorn.com/Resources/OSDev22.html But he never added the Creating/Writing a File bit, so Im doing it on my own.

Any help is appreciated.

UPDATE: I just realized I never actually wrote the FAT tables to disk... how can I manually write the tables to my disk image? — Andres Bejarano, Jun 22 '14 at 21:22

score 1 · Accepted Answer · answered Jul 25 '14 at 08:29

One implementation I can think of is to use a tree-like structure wherein the root directory is the parent directory.

struct fat_component{
    fat_component *parent;
    /*other fat components here*/
};

Once we have that we can then invoke something like an inline assembly call to int 13h/AH=03. I'm not too familiar with it but here's a link: http://stanislavs.org/helppc/int_13-3.html.

inline_asm("mov ah, 03h");
inline_asm("mov ch, %s", fc->cylindernum);
inline_asm("mov cl, %s", fc->sectornum);
/*other int13h initializations here*/
inline_asm ("int 13h");

OS development: Implementing FAT12 Write Function

1 Answers1