From: chad@src.umd.edu (R Michael McMahon) Subject: Re: Badblocks revisited (**sigh**) Date: 14 Jul 1992 02:15:12 GMT
> At that time, there
> was an option to mkfs that would cause it to scan the disk, and put
> any badblocks into a file, which I believe was /.badblocks.
> [ edited blurb about tar backups]
> This is certainly
> a low-tech solution to bad blocks, but it seems reasonable. Ideally
> the bad blocks would be put somewhere that doesn't appear in the
> file system at all, but this would require modifications to both
> the kernel and fsck.
Sun does it this way: inode 0 is unusable, inode 1 is used for
bad-blocks (these are all linked to inode 1), and the root inode is
therefore 2. There is no real reason to use inode 1 for this purpose
(a dumpsite for bad-blocks) : any inode could be similarly reserved
in the super-block (if I'm not mistaken). This is the BSD way- I
haven't used SysV to know if its any different.
R Michael McMahon chad@src.umd.edu