The HyperNews Linux KHG Discussion Pages

Question: inode semantics?

Forum: A tour of the Linux VFS
Keywords: VFS filesystems
Date: Wed, 13 Aug 1997 08:09:54 GMT
From: Rick van Rein <vanrein@cs.utwente.nl>

Hello,

I have read your KHG and am pleased with it. I started looking at the filesystem, but found the docs to reveal only minimal things. I hope you can answer some of the questions that follow, or maybe you can given pointers to places of more information - my sources have dried out on this.

1. I know what an inode is. But who creates it, who destroys it, when do these things happen?

2. I have found the important actions of the VFS on anyFS named readsuper, writesuper, putsuper, readinode, writeinode, putinode. I have a suspicion on their order of calling, but I couldn't find any information on what protocol they are, in general, supposed to follow. Should it be that you first read, then do an arbitrary number of writes, then put it? What changes to this scheme for creation and deletion of inodes? Is it true that inode manipulations can only done between readsuper (i.e, mount) and putsuper (i.e, umount)?

Is there any other reason than mount/umount for which to do a readsuper or putsuper?

Please note, the KHG *only* explains mounting, not umounting.

3. What is the meaning of putinode/getinode? My suspicion is that putinode makes an inode available for use by the VFS (inserting it in the space of known inodes, so to say) and that getinode takes it out of that space again (requesting and/or awaiting it if it isn't there, I presume).

Groeten,

Rick van Rein,
        Spiritus flexibilis in corpore flexibile
--
Postbus 217
7500 AE Enschede
053-489 4291