Sure it is. And at worst you'll just have to have a dummy inode to
attach to it.
Could we split up just the important fields and create a new level of
abstraction? Sure. But since there are no major users, and since it
would make the common case less readable, there simply isn't any point.
Over-abstraction is a real problem - it makes the code less obvious, and
if there is no real gain from it then over-abstracting things is a bad
thing.
>Just a very basic example: Inode 0 on ntfs is a file named $MFT. It
>contains two data parts: one containing the actual on-disk inode metadata
>and one containing the inode allocation bitmap. At the moment I have placed
>the inode metadata in the "normal" inode address_space mapping and of
>course ->host points back to inode 0. But for the inode allocation bitmap I
>set ->host to the current ntfs_volume structure instead and I am at present
>using a special readpage function to access this address space.
Why not just allocate a separate per-volume "bitmap" inode? It makes
sense, and means that you can then use all the generic routines to let
it be automatically written back to disk etc.
Also, don't you get the "ntfs_volume" simply through "inode->i_sb"? It
looks like you're just trying to save off the same piece of information
that you already have in another place. Redundancy isn't a virtue.
>The mere fact that the VM requires to know the size of the data in an
>address space just indicates to me that struct address_space ought to
>contain a size field in it.
It doesn't "require" it - in the sense that you can certainly use
address spaces without using those routines that assume that it has an
inode. It just means that you don't get the advantage of the
higher-level concepts.
>So while something should be fixed I think it is the kernel and not ntfs. (-;
There are a few fundamental concepts in UNIX, and one of them is
"everything is a file".
And like it or not, the Linux concept of a file _requires_ an inode (it
also requires a dentry to describe the path to that inode, and it
requires the "struct file" itself).
The notion of "struct inode" as the lowest level of generic IO entity is
very basic in the whole design of Linux. If you try to avoid it, you're
most likely doing something wrong. You think your special case is
somehow independent and more important than one of the basic building
blocks of the whole system.
Linus
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/