From mboxrd@z Thu Jan 1 00:00:00 1970 From: Daniel Phillips Subject: Re: [PATCH] Introduce BSD-style user credential [3/3] Date: Sun, 1 Sep 2002 17:23:57 +0200 Sender: linux-fsdevel-owner@vger.kernel.org Message-ID: References: <15728.7151.27079.551845@charged.uio.no> <15728.61204.381468.238609@charged.uio.no> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7BIT Cc: Trond Myklebust , Linux FSdevel , Linux Kernel , Dave McCracken Return-path: To: trond.myklebust@fys.uio.no, Linus Torvalds In-Reply-To: <15728.61204.381468.238609@charged.uio.no> List-Id: linux-fsdevel.vger.kernel.org On Saturday 31 August 2002 18:30, Trond Myklebust wrote: > >>>>> " " == Linus Torvalds writes: > > > One thing that may be interesting (I certainly think it migth > > be), would be to add a "struct user_struct *" pointer to the > > vfs_cred as well. This is because I'd just _love_ to have that > > "user_struct" fed down to the VFS layer, since I think that is > > where we may some day want to put things like user-supplied > > cryptographic keys etc. > > > The advantage of "struct user_struct" (as opposed to just a > > uid_t) is that it can have information that lives for the whole > > duration of a login, and it's really the only kind of data > > structure in the kernel that can track that kind of > > information. > > No problem at all with this. Indeed I agree it makes a lot of sense... > > The only thing is if you'd allow me to do it as an incremental patch > to the initial one? > I don't see 'struct user_struct *' as replacing the existing 'uid' > entry, so there should be no need to change the existing API. Instead, > we can just add in the necessary call to alloc_uid() to > vfscred_create() and/or setfsuid()... I really do like Kai's name suggestion 'struct session'. -- Daniel