|> Currently fs_may_umount_ro has a problem which allows it to permit a
|> readonly remount even if there is write activity pending on the fs. The
|> problem is the special case recently pointed out on l-k where we have an
|> open file descriptor on a deleted file (i_nlink==0). Closing that fd
|> will cause the file to be deleted on disk, even though the filesystem
|> has been remounted readonly.
|>
|> In the absence of explicit filesystem support for this special case
|> (such as the readonly remount recording a list of files waiting for
|> delete before freezing the disk), we need to prevent such dismounts from
|> happening. Trivial patch against 2.1.117 follows.
|>
|> This should not break anything new, but it may convert problems with
|> existing, broken shutdown scripts from silent failures to noisy ones.
|> It's best to know about it if it happens, though.
This will break unmounting the root fs on shutdown for many systems,
because /sbin/init will often have a reference to a deleted file (eg
/etc/ld.so.cache), and init never exits.
-- Andreas Schwab "And now for something schwab@issan.informatik.uni-dortmund.de completely different" schwab@gnu.org- 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.altern.org/andrebalsa/doc/lkml-faq.html