> > > swapon should keep a handle on the inode, which means the file won't
> > > be removed from the physical media even if it is unlinked (just like
> > > open files aren't.)
> >
> > Unfortunately, there are a few issues with that:
> > - it doesn't work on inodeless FSes (msdosfs)
> > - you can't swapoff() a deleted file, leaving minor
> > fs corruption on reboot (and major corruption on
> > umsdos and the like)
> > - since you can't swapoff() the deleted file, you
> > can't reclaim the disk space
> >
> > This basically means that we have a real problem at
> > hand that should be fixed.
>
> Sure it works on inodeless filesystems -- *ALL* filesystems are
> inode-based as far as the Linux kernel are concerned. However, your
> objection about swapoff() is well taken -- we need a variant that can
> take an index to a swap area rather than a filename.
...or make the swap inodes appear in /proc.
Like for deleted files still accessed by a process are *still* in
/proc/<pid>/fds/<fdnum> (though you cannot link them back for some
strange reason).
It would be more unix-like IMHO.
Phil.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/