On Mon, 2008-03-24 at 20:58 -0400, Dave Quigley wrote: > Hello Everyone, > I was working through Christoph's comments on the Labeled NFS work > and I ran into the wall that sent me on tangent about the setxattr > hook. > > The problem is we are using the iattr structure to send the label > information down into the appropriate NFS calls. The reasons we do this > are 1) that is how it is currently being done and 2) it would involve > adding parameters in quite a few functions all down the call chain to do > it as an additional parameter. Now the issue at hand is that by adding a > field into the iattr structure we should also add the appropriate > functionality to inotify and its ilk. This is a problem because, to set > the inode on disk we need a dentry, but the iattr interfaces only take > inodes. Clearly I have several things here that are mutually exclusive. > > Christoph has already expressed his desire for me not to overload iattr > with this and for the NFS server code I can accommodate this in a > reasonable way. However, the client code relies on the iattr structure > pretty extensively for passing this information around. Since this is > going to involve extensive modifications no matter which way it goes I > figured I would ask for suggestions from the key players before moving > forward. > > Dave I apologize for having to forward this to fs-devel but I didn't realize before I sent this mail last night that I had pasted the mailing list into the subject rather into the CC box. It was getting late and I was about to head out so I rushed it a bit. Sorry for the repost those who already got this mail. Dave -- 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
