From mboxrd@z Thu Jan 1 00:00:00 1970 From: Cyrill Gorcunov Subject: Re: [rfc 2/7] procfs: Convert /proc/pid/fdinfo/ handling routines to seq-file Date: Thu, 5 Jul 2012 21:37:29 +0400 Message-ID: <20120705173729.GC16586@moon> References: <20120627110116.201735815@openvz.org> <20120627110512.485189390@openvz.org> <4FF3F2B6.5040909@parallels.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: "linux-kernel@vger.kernel.org" , "linux-fsdevel@vger.kernel.org" , Al Viro , Alexey Dobriyan , Andrew Morton , James Bottomley To: Pavel Emelyanov Return-path: Content-Disposition: inline In-Reply-To: <4FF3F2B6.5040909@parallels.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Wed, Jul 04, 2012 at 11:37:26AM +0400, Pavel Emelyanov wrote: ... > > I believe we can still have the proc_fdinfo_read and proc_fd_link code non-splitted. > Just push a callback pointer ino the proc_fd_info (as usual -- we an opaque void *argument). Hi Pavel, sorry for delay (was busy a bit). It seems this wont work. Look, previously we have had static int proc_fd_info(struct inode *inode, struct path *path, char *info) { struct task_struct *task = get_proc_task(inode); ... if (info) { ... } this makes info argument optional } static int proc_fd_link(struct dentry *dentry, struct path *path) { return proc_fd_info(dentry->d_inode, path, NULL); } it's possible because we didn't use seq-files engine. One we switch to seq-files the old proc_fd_info become a part of static const struct file_operations proc_fdinfo_file_operations = { .open = seq_fdinfo_open, ie the declaration of seq_fdinfo_open is restricted to file_operations::open method and I can't add opaque void *argument here. Sure I can add one more wrapper function but I guess this increase code complexity which I tried to escape. Cyrill