How works a loop device (especially a mounted OS Image)

Question

I know, it's a often asked question on many boards and I have read the wikipedia article. But those articles aren't satisfactory. I want to know it because I have Ubuntu 12.04 for Android on my chrooted HTC One X. For normal file systems like a disk or a hard drive image it seems to be clear but how works a operating system loop device? My thought is, that a simualted disk is a static thing which has no own life but an operating system has it's own life with processes and threads which are running.

• Notices the kernel (here Android) that the loop device is an os image which needs a special mounting/handling? How is the native kernel in contact with the virtual os so that I can use an os within an os?
• Obviously it's mounted like a file system but communicates it with the kernel via the loop back adapter or has it nothing to do with that? (just the normal network communication)

I hope my problem isn't asked to complicated. Thanks for your help.

Answer 1

Loop devices generally provides an interface to access a file as a file-system. In order to loop mount an image you have to place the image somewhere in your real file-systems, i.e. disk-partitions.

So while loop devices are virtual, the files they provides access to are not and those files occupy space.

When you add some data to the loop-mounted file-system the data is actually added to the image file, as if you are appending/inserting some bytes of data somewhere in a file. Loop device decides where to put those bytes. So the image file grows in size and deducts space from the hosting file-system.

To answer the question about how OS within an OS runs, it can be said that they either share kernels or use a hypervisor to emulate an encapsulated environment which translate another OS'es instructions to hosting OS'es instructions.