On Tuesday, October 14, 2003, at 08:37  AM, Charles M. Hannum wrote:

>> Darrin pointed out to me that it should be allowed to mount an unclean
>> FFS that was softdep-enabled, since the only inconsistencies in that
>> case should be lost file system blocks.
>
> That theory dates back to something Kirk said years ago, which he has 
> now
> recanted because I pointed out several cases where he was wrong.

Well, that theory is restated in his "background fsck" paper, so if 
he's recanted it, he un-recanted it.

Personally, I would rather disallow mounts of unclean softdep file 
systems, too... My opinion is that mounting an unclean softdep file 
system is only useful for background fsck, and I happen to disagree 
with the background fsck approach in general :-)

But Darrin's argument is that it might be useful to bring up a softdep 
file system immediately, and then take it offline later to run the fsck.

         -- Jason R. Thorpe <thorpej@wasabisystems.com>