On Fri, 2007-07-06 at 10:52 -0400, David Patrick Quigley wrote: > Hello, > I have been reading through the NFSv4 specification and I came > across this piece of information. > > "Each of the Mandatory and Recommended attributes can be classified in > one of three categories: per server, per filesystem, or per > filesystem object." Section 5.4 RFC3530 pg 37 > > What I am wondering is if these classifications are enforced by > convention or if there is something in the data structures that specify > this. I attempted to look through the Linux implementation to find the > code pertaining to this but I couldn't find it. I'm not sure I understand you. The protocol lists each attribute and classifies it into a particular category. It is then up to the client and server to ensure that their implementations reflect this categorisation. As you can see, though, most of these attributes reflect standard POSIX filesystem and file attributes, and so we automatically categorise them properly into 'cache per inode', 'cache per superblock', ... Note however that there are a couple of mistakes that should probably be fixed in the upcoming errata version of the RFC: Firstly, most of the attributes that are classified as "The per filesystem object attributes" are in fact "The per file object attributes". Secondly, the "fsid" should probably be listed as a "per filesystem attribute" (since, in fact, the fsid _defines_ what constitutes a filesystem). I think the reason that it was listed among the per-file objects is that you need to read it on a per-file basis in order to figure out when you cross the boundaries of the filesystem. Trond - To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
