linux-kernel - Re: [PATCH] knfsd: Fix stale file handle problem with subtree

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Sat, 29 Jul 2006 15:29:33 +1000
From:	Neil Brown <neilb@...e.de>
To:	"J. Bruce Fields" <bfields@...ldses.org>
Cc:	Andrew Morton <akpm@...l.org>, nfs@...ts.sourceforge.net,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] knfsd: Fix stale file handle problem with subtree_checking.

On Friday July 28, bfields@...ldses.org wrote:
> 
> It'd be great if we could deprecate subtree checking some day in the
> distant future....

The first step would be to stop it from being the default (as Trond
has suggested a number of times :-)

How about this.
 I release a 1.0.10 shortly which addresses some 'portlist' related
 breakage and prints a nasty warning if you have neither subtree_check
 or no_subtree_check, but still defaults to subtree_check.

 Then the next release will be 1.1.0 which prints the same warning,
 but defaults the other way - and probably removed the warning if you
 include neither sync not async.

That should at least get subtree_check to be used less.

> 
> Would it be feasible to add filesystem support for some sort of
> subvolume-like thing that acted like a mountpoint (in the sense that it
> restricted hardlinks and renames) but that didn't require setting aside
> a separate partition?  I imagine that'd probably do what most people
> exporting subtrees want without forcing us to do dubious tricks with
> filehandles.

I think it is a great idea for a 'filesystem' to support multiple
independent file-trees within the one storage set, which is roughly
what you are saying I think (though probably not quite).

However I suspect that most people don't actually want subtrees.  They
just get it as the default.  It isn't something that I would have
implemented if I hadn't inherited the requirement, and no other OS
that I know of provides that particular semantic.

Maybe we should make it non-default, and then in one year, subtly
break it (maybe reintroduce this bug) and see if anyone notices :-) 
(No, that's unethical, I wouldn't do that - really ....)

NeilBrown
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/