At 2004-08-09T13:49:17+1200, Volker Kuhlmann wrote:
> However that is what it (wrongly) implies to be.
Yes, thus my comments (which you snipped) about the documentation
requiring an update.
> Considering drives have a minimum of 2MB, often 8MB, cache I'd
> certainly try and stay above that (the man page explicitly says so) to
That recommendation has been removed--it's not in the man page for
badblocks from e2fsprogs 1.35. It could be that the comment has been
invalidated by recent versions ensuring that they read/write through the
drive's cache, but I haven't confirmed that other than noting the recent
versions are attempting to use O_DIRECT.
> So what... If I only want to check parts of a partition (which may not
> even have a filesystem!), anyfsck -c is no good and I have to get into
> numbers.
Which is exactly why badblocks will let you do such things in those
cases. Since you removed the context of the quote, it's pretty easy to
say "oh, but it's wrong in <this other case>".
> Reiser rightfully doesn't put badblock handling on a high priority
> because usable hard disks don't have user-visible bad blocks (ok,
> small amount of generalisation).
Well, reiserfs provides all of the support for bad blocks that ext[23]
do (and, in fact, you can even add bad blocks to the reiserfs bad blocks
list while the filesystem is mounted), except that the userland tools
are lacking the 'fsck -c' stuff.
Cheers,
-mjg
--
Matthew Gregan |/
/| [EMAIL PROTECTED]
