[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Symlink, chrdev, blkdev translators

From: Ludovic Courtès
Subject: Re: Symlink, chrdev, blkdev translators
Date: Tue, 20 Aug 2002 21:28:06 +0200
User-agent: Mutt/1.4i

On Mon, Aug 19, 2002 at 11:08:05PM +0200, Marcus Brinkmann wrote:
> > So you are saying that showtrans should not give us anything on special
> > nodes?  What advantage does it give you to be inconsistent on purpose?

> In fact, once you access that node, you are talking to the fifo translator
> which will lie and report the fifo flag in its response to io_stat.  So it
> is completely transparent to the user.  Except for symlink, for which you
> don't get a process of its own.

I see. This is because netfs and diskfs (in their dir_lookup stub) test
whether the node being accessed is a symlink or a device or a fifo and try to
create the appropriate translator behaviour.

However, if these special cases were not there, filesystems could do "funny"
things. Suppose for instance a node "foo" which is a symlink to "bar", where
"bar" doesn't exist: We could have the filesystem server provide data to read
requests on "foo", even though the normal Unix behaviour would be to return
an error! ;)


reply via email to

[Prev in Thread] Current Thread [Next in Thread]