On 08/27/2014 03:19 PM, Anton Aylward wrote:
The real problem arises when file systems get corrupt and things walk over the inode. That's why we have FSCK. Yes, this was worse in the days of the V7 FS and even the later 20th century file systems where the inode space was separate from the data space. The V7 file system was the limiting case of that! With b-tree based FS the distinction either no longer exists or is very fuzzy.
But an intermediate problem is when the file-that-is-the-directory gets corrupted. Plonk random characters in that and who knows.
Essentially the structure used by readdir has the name and inode. The name_should_ be null terminated. If the directory entry gets corrupted in just the right way, that is the name gets hit rather than the inode field or one of the other supported fields then we have the oddity that the record size and the name length don't match. Yes its a YMMV situation, but I've BTDT. Often you recognise it by very weird 'ls' listings or names you know getting truncated.
yes, that's bad. In such a situation I'd hopefully not try to remove such a file, but rather copy as much as I can into a fresh file system, and then care about those oddities there instead of touching the original, corrupted file system. Have a nice day, Berny -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org