Linux Device Drivers

How Mounting Works

Block devices differ from char devices and normal files in that they can be mounted on the computer’s filesystem. This is different from the normal access through a struct file, where the structure is bound to a specific process and exists only from open to close. When a filesystem is mounted, there’s no process holding a filp.

When the kernel mounts a device in the filesystem, it invokes the normal open method to access the driver. However, in this case the filp argument to open is a dummy variable, almost a placeholder, whose only meaningful field is f_mode. The remaining fields hold random values and should not be used. The value of f_mode tells the driver whether the device is to be mounted read-only (f_mode == FMODE_READ) or read/write (f_mode == (FMODE_READ|FMODE_WRITE)). A dummy variable is used instead of a file structure because a real struct file would be released at process termination, while a mounted filesystem survives after the mount command has done its job.

At mount time, the only thing that is invoked in the driver is the open method. While the disk is mounted, the kernel invokes the read and write methods in the device (which map to request_fn) to manage files in the filesystem. The driver can’t tell if request_fn is servicing a process (like fsck) or the filesystem layer of the kernel.

As far as umount is concerned, it just flushes the buffer cache and calls the release (close) driver method. Since there is no meaningful filp to pass to ...