Project 4: Implementing a Simple File System on Top of a Virtual Disk Project Goals
The goals of this project are:
to implement a simple file system on top of a virtual disk
to understand implementation details of file systems Implementing a simple file system
The goal of this project is to implement a simple file system on top of a virtual disk. To this end, you will implement a library of functions that offers a set of basic file system calls such as open, read, write, … to applications. The file data and file system metainformation will be stored on a virtual disk. This virtual disk is actually a single file that is stored on the real file system provided by the Linux operating system. That is, you are basically implementing your file system on top of the Linux file system.
To create and access the virtual disk, we have provided a few definitions and helper functions that you can find in this header file and this source file. Note that, in your library, you are not allowed to create any real files on the Linux file system itself. Instead, you have to use the provided helper functions and store all the data that you need on the virtual disk. As you can see by looking at the provided header and source files, the virtual disk has 16,384 blocks, and each block holds 4KB. You can create an empty disk, open and close a disk, and read and write entire blocks by providing a block number in the range between 0 and 16,383 inclusive.
Your file system has to support a directory hierarchy, to at least one level below the root directory. Files are stored in a directory and on the virtual disk. In addition, your file system does not have to store more than 256 files of course, you can create and delete files, and deleted files do not count against this 256 file limit. Finally, out of the 16384 blocks available on disk, only 8,192 must be reserved as data blocks. That is, you have ample of space to store your metainformation. However, you have to free data blocks make them available again when the corresponding file is deleted. The maximum file size is 32 megabytes all 8,192 data blocks, each with 4KB.
To manage your file system, you have to provide the following three functions: int makefschar diskname;
This function creates a fresh and empty file system on the virtual disk with name diskname. As part of this function, you should first invoke makediskdiskname to create a new disk.
Then, open this disk and writeinitialize the necessary metainformation for your file system so that it can be later used mounted. The function returns 0 on success, and 1 when the disk diskname could not be created, opened, or properly initialized.
int mountfschar diskname;
This function mounts a file system that is stored on a virtual disk with name diskname. With the mount operation, a file system becomes ready for use. You need to open the disk and then load the metainformation that is necessary to handle the file system operations that are discussed below. The function returns 0 on success, and 1 when the disk diskname could not be opened or when the disk does not contain a valid file system that you previously created with makefs.
int umountfschar diskname;
This function unmounts your file system from a virtual disk with name diskname. As part of this operation, you need to write back all metainformation so that the disk persistently reflects all changes that were made to the file system such as new files that are created, data that is written, …. You should also close the disk. The function returns 0 on success, and 1 when the disk diskname could not be closed or when data could not be written to the disk this should not happen. If there are any open file descriptors that point to files in the file system that is about to be unmounted, umountfs will close them.
It is important to observe that your file system must provide persistent storage. That is, assume that you have created a file system on a virtual disk and mounted it. Then, you create a few files and write some data to them. Finally, you unmount the file system. At this point, all data must be written onto the virtual disk. Another program that mounts the file system at a later point in time must see the previously created files and the data that was written. This means that whenever umountfs is called, all metainformation and file data that you could temporarily have only in memory; depending on your implementation must be written out to disk.
In addition to the management routines listed above, you are to implement the following file system functions which are very similar to the corresponding Linux file system operations. These file system functions require that a file system was previously mounted.
int fsopenchar name;
The file specified by name in this case, name is a path to the file that is to be opened, including the actual filename part of the path is opened for reading and writing, and the file descriptor corresponding to this file is returned to the calling function. If successful, fsopen returns a non negative integer, which is a file descriptor that can be used to subsequently access this file. Note that the same file file with the same name can be opened multiple times. When this
happens, your file system is supposed to provide multiple, independent file descriptors. Your library must support a maximum of 64 file descriptors that can be open simultaneously. fsopen returns 1 on failure. It is a failure when the file with name cannot be found i.e., it has not been created previously or is already deleted. It is also a failure when there are already 64 file descriptors active. When a file is opened, the file offset seek pointer is set to 0 the beginning of the file.
int fscloseint fildes;
The file descriptor fildes is closed. A closed file descriptor can no longer be used to access the corresponding file. Upon successful completion, a value of 0 is returned. In case the file descriptor fildes does not exist or is not open, the function returns 1.
int fscreatechar name;
This function creates a new file with name name in your file system name is the path to the file including the name of the file itself. The file is initially empty. The maximum length for a file name is 15 characters this is also the maximum length of a directory name. Also, there can be at most 256 files in the directory. Upon successful completion, a value of 0 is returned. fscreate returns 1 on failure. It is a failure when the file with name already exists using the full path specified in name, when the file name is too long it exceeds 15 characters for the directory name and 15 characters for the name of the file, or when there are already 256 files present in the specified directory. Note that to access a file that is created, it has to be subsequently opened.
int fsdeletechar name;
This function deletes the file with the path and name name from the directory of your file system and frees all data blocks and metainformation that correspond to that file. The file that is being deleted must not be open. That is, there cannot be any open file descriptor that refers to the file name. When the file is open at the time that fsdelete is called, the call fails and the file is not deleted. Upon successful completion, a value of 0 is returned. fsdelete returns 1 on failure. It is a failure when the file with name does not exist. It is also a failure when the file is currently open i.e., there exists at least one open file descriptor that is associated with this file.
int fsmkdirchar name;
This function attempts to create a directory with the name name. fsmkdir returns zero on success, or 1 if an error occurred. It is a failure when the directory with name already exists, when the file name is too long it exceeds 15 characters for the directory name, or when there are already 256 files present in the directory.
int fsreadint fildes, void buf, sizet nbyte;
This function attempts to read nbyte bytes of data from the file referenced by the descriptor fildes into the buffer pointed to by buf. The function assumes that the buffer buf is large enough to hold at least nbyte bytes. When the function attempts to read past the end of the file, it reads all bytes until the end of the file. Upon successful completion, the number of bytes that were actually read is returned. This number could be smaller than nbyte when attempting to read past the end of the file when trying to read while the file pointer is at the end of the file, the function returns zero. In case of failure, the function returns 1. It is a failure when the file descriptor fildes is not valid. The read function implicitly increments the file pointer by the number of bytes that were actually read.
int fswriteint fildes, void buf, sizet nbyte;
This function attempts to write nbyte bytes of data to the file referenced by the descriptor fildes from the buffer pointed to by buf. The function assumes that the buffer buf holds at least nbyte bytes. When the function attempts to write past the end of the file, the file is automatically extended to hold the additional bytes. It is possible that the disk runs out of space while performing a write operation. In this case, the function attempts to write as many bytes as possible i.e., to fill up the entire space that is left. The maximum file size is 32 Mbytes which is, 8192 blocks, each 4K. Upon successful completion, the number of bytes that were actually written is returned. This number could be smaller than nbyte when the disk runs out of space when writing to a full disk, the function returns zero. In case of failure, the function returns 1. It is a failure when the file descriptor fildes is not valid. The write function implicitly increments the file pointer by the number of bytes that were actually written.
int fsgetfilesizeint fildes;
This function returns the current size of the file pointed to by the file descriptor fildes. In case
fildes is invalid, the function returns 1. int fslseekint fildes, offt offset;
This function sets the file pointer the offset used for read and write operations associated with the file descriptor fildes to the argument offset. It is an error to set the file pointer beyond the end of the file. To append to a file, one can set the file pointer to the end of a file, for example, by calling fslseekfd, fsgetfilesizefd;. Upon successful completion, a value of 0 is returned. fslseek returns 1 on failure. It is a failure when the file descriptor fildes is invalid, when the requested offset is larger than the file size, or when offset is less than zero.
int fstruncateint fildes, offt length;
This function causes the file referenced by fildes to be truncated to length bytes in size. If the file was previously larger than this new size, the extra data are lost and the corresponding data blocks on disk if any must be freed. It is not possible to extend a file using fstruncate. When the file pointer is larger than the new length, then it is also set to length the end of the file. Upon successful completion, a value of 0 is returned. fslseek returns 1 on failure. It is a failure when the file descriptor fildes is invalid or the requested length is larger than the file size.
Implementation Filesystem
In principle, you can implement the file system in any way that you want as long as 8,192 blocks of the disk remain available to store file data. However, it might be easier when you borrow ideas from existing file system designs. We recommend you model your file system after the FAT file allocation table design, although it is also possible though likely more complex to use a Unix inodebased design.
In general, you will likely need a number of data structures on disk, including a super block, a root directory, information about free and empty blocks on disk, file metainformation such as file size, and a mapping from files to data blocks.
The super block is typically the first block of the disk, and it stores information about the location of the other data structures. For example, you can store in the super block the whereabouts of the file allocation table, the directory, and the start of the data blocks.
The directory holds the names of the files. When using a FATbased design, the directory also stores, for each file, its file size and the head of the list of corresponding data blocks. When you use inodes, the directory only stores the mapping from file names to inodes.
The file allocation table FAT is convenient because it can be used to keep track of empty blocks and the mapping between files and their data blocks. When you use an inodebased design, you will need a bitmap to mark disk blocks as used and an inode array to hold file information including the file size and pointers to data blocks.
In addition to the filesystemrelated data structures on disk, you also need support for file descriptors. A file descriptor is an integer in the range between 0 and 63 inclusive that is returned when a file is opened, and it is used for subsequent file operations such as reading and writing. A file descriptor is associated with a file, and it also contains a file offset seek pointer. This offset indicates the point in the file where read and write operations start. It is implicitly updated incremented whenever you perform a fsread or fswrite operation, and it can be explicitly moved within the file by calling fslseek. Note that file descriptors are not
stored on disk. They are only meaningful while an application is running and the file system is mounted. Once the file system is unmounted, file descriptors are no longer meaningful and, hence, should be all closed before a call to umountfs.
Test Program
You should write a small program that demonstrates the creation of the virtual disk and the file system. To demonstrate your file system implementation, you are required to develop a program that uses all of the functions in the filesystem. This program should mount the virtual disk and then demonstrate copying a file from the real OS file system to your file system and copying a file from your file system to the OS file system.
You should be able to test the persistence of your file system by rerunning the test application and then showing the directory and virtual disk contents.
Deliverables
1. You will submit all files that are part of your project via the assignment in Canvas.
2. You should submit all source files required to build your file system implementation and
the test programs.
3. Your code must be implemented in CC. You are to include disk.h as well as disk.c to
build a complete executable, but do not make any changes to these two files. The
source code should be well commented and readable.
4. As in previous projects, you are to use Github as part of your development process,
making regular commits as you design and develop your solution.
5. Include a README with this project. A separate component of the project is to describe
the detailed design and development of this project Project 4A. The README should be an overview of the features of the library components and the test application. Include a discussion of your component testing during the development process. If you had problems, tell us why and what.
Reviews
There are no reviews yet.