[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <E1JznjI-0003fD-9i@pomaz-ex.szeredi.hu>
Date: Sat, 24 May 2008 08:59:48 +0200
From: Miklos Szeredi <miklos@...redi.hu>
To: zippel@...ux-m68k.org
CC: miklos@...redi.hu, hch@...radead.org,
linux-fsdevel@...r.kernel.org, viro@...IV.linux.org.uk,
linux-kernel@...r.kernel.org
Subject: Re: [patch 06/14] hfsplus: remove hfsplus_permission()
> > Semi-concrete: link(2) locks the target's parent and the source.
> > Cross-directory rename(2) locks both parents. If link's target is a
> > file which has children, this can result in an ABBA deadlock. That's
> > _before_ the filesystem's ->link() or ->rename() function is called.
>
> I'm afraid you have to be more concrete than this.
> hfsplus_file_inode_operations doesn't have a link or rename operation, so
> if one your source parents above involves a file, it shouldn't get this
> far or you have to describe your scenario in more detail.
Heh, you just want me to do all the hard work :)
OK, I'll go along:
/mnt/a/x is a regular file
ln /mnt/a/x /mnt/b/x
rename /mnt/a/x /mnt/b/x/p & ln /mnt/b/x /mnt/a/q
Neither ->rename nor ->link is required to be defined on x. The
rename will first lock x, then a. The link will first lock a then x,
AFAICS.
This is probably just one of several ways to make it deadlock.
Hmm?
Miklos
--
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/
Powered by blists - more mailing lists