From: Spencer Shepler (shepler@eng.sun.com)
Date: 06/25/01-02:24:27 AM Z
Date: Mon, 25 Jun 2001 02:24:27 -0500 From: Spencer Shepler <shepler@eng.sun.com> Subject: Re: READ results on non-regular files Message-ID: <20010625022427.D553@shepnet150.eng.sun.com> On Mon, Martin Pool wrote: > It's not mentioned in RFC3010 14.2.23, but I assume that READ on a > filehandle identifying directory is meant to return NFS4ERR_ISDIR. Is > that the intention? It seems like it would not be very useful to > return the contents of the directory in the server's native format, as > read(2) on a directory does on Solaris and some other unices. With the current list of error returns, NFS4ERR_INVAL would need to be used in this instance. I will add an item to add NFS4ERR_ISDIR for the READ operation and add clarifying text. > Also, I would expect that NFSv4 continues to export devices as device > nodes, rather than allowing clients to directly access from the device > on the server. What is the correct error for READ to return if the > client tries to READ a character or block device? NFS4ERR_INVAL? You are correct, the model for device special files has not changed from NFSv2/v3. INVAL would be most appropriate. Spencer
This archive was generated by hypermail 2.1.2 : 03/04/05-01:48:52 AM Z CST