From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: [PATCH 08/10] locks: move i_lock acquisition into generic_*_lease handlers Date: Sun, 24 Aug 2014 09:06:34 -0700 Message-ID: <20140824160634.GG15908@infradead.org> References: <1408804878-1331-1-git-send-email-jlayton@primarydata.com> <1408804878-1331-9-git-send-email-jlayton@primarydata.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-fsdevel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, bfields-uC3wQj2KruNg9hUCZPvPmw@public.gmane.org, hch-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org, cluster-devel-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org, linux-cifs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-nfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Jeff Layton Return-path: Content-Disposition: inline In-Reply-To: <1408804878-1331-9-git-send-email-jlayton-7I+n7zu2hftEKMMhf/gKZA@public.gmane.org> Sender: linux-cifs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: Looks good, Reviewed-by: Christoph Hellwig Some comments on further work I'd like to see in this area, though: > + spin_lock(&inode->i_lock); > + time_out_leases(inode); > for (before = &inode->i_flock; > ((fl = *before) != NULL) && IS_LEASE(fl); > before = &fl->fl_next) { > if (fl->fl_file != filp) > continue; > - return fl->fl_lmops->lm_change(before, F_UNLCK); > + error = fl->fl_lmops->lm_change(before, F_UNLCK); > } We really should split a lm_release from lm_change, the way it is used is highly confusing. In addition I think a lot of code currently in lease_modify should move here instead, e.g. something like: if (fl->fl_file != filp) continue; fl = *before; fl->fl_type = F_UNLCK; lease_clear_pending(fl, F_UNLCK); locks_wake_up_blocks(fl); if (fl->fl_lmops->lm_delete) fl->fl_lmops->lm_delete(fl); locks_delete_lock(before, NULL); with lm_delete for user space leases as: static void lease_delete(struct file_lock *fl) { struct file *filp = fl->fl_file; f_delown(filp); filp->f_owner.signum = 0; fasync_helper(0, fl->fl_file, 0, &fl->fl_fasync); if (fl->fl_fasync != NULL) { printk(KERN_ERR "locks_delete_lock: fasync == %p\n", fl->fl_fasync); fl->fl_fasync = NULL; } } and a NULL implementation for delegations. From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: linux-nfs-owner@vger.kernel.org Received: from bombadil.infradead.org ([198.137.202.9]:58676 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752961AbaHXQGf (ORCPT ); Sun, 24 Aug 2014 12:06:35 -0400 Date: Sun, 24 Aug 2014 09:06:34 -0700 From: Christoph Hellwig To: Jeff Layton Cc: linux-fsdevel@vger.kernel.org, bfields@fieldses.org, hch@infradead.org, cluster-devel@redhat.com, linux-cifs@vger.kernel.org, linux-nfs@vger.kernel.org Subject: Re: [PATCH 08/10] locks: move i_lock acquisition into generic_*_lease handlers Message-ID: <20140824160634.GG15908@infradead.org> References: <1408804878-1331-1-git-send-email-jlayton@primarydata.com> <1408804878-1331-9-git-send-email-jlayton@primarydata.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <1408804878-1331-9-git-send-email-jlayton@primarydata.com> Sender: linux-nfs-owner@vger.kernel.org List-ID: Looks good, Reviewed-by: Christoph Hellwig Some comments on further work I'd like to see in this area, though: > + spin_lock(&inode->i_lock); > + time_out_leases(inode); > for (before = &inode->i_flock; > ((fl = *before) != NULL) && IS_LEASE(fl); > before = &fl->fl_next) { > if (fl->fl_file != filp) > continue; > - return fl->fl_lmops->lm_change(before, F_UNLCK); > + error = fl->fl_lmops->lm_change(before, F_UNLCK); > } We really should split a lm_release from lm_change, the way it is used is highly confusing. In addition I think a lot of code currently in lease_modify should move here instead, e.g. something like: if (fl->fl_file != filp) continue; fl = *before; fl->fl_type = F_UNLCK; lease_clear_pending(fl, F_UNLCK); locks_wake_up_blocks(fl); if (fl->fl_lmops->lm_delete) fl->fl_lmops->lm_delete(fl); locks_delete_lock(before, NULL); with lm_delete for user space leases as: static void lease_delete(struct file_lock *fl) { struct file *filp = fl->fl_file; f_delown(filp); filp->f_owner.signum = 0; fasync_helper(0, fl->fl_file, 0, &fl->fl_fasync); if (fl->fl_fasync != NULL) { printk(KERN_ERR "locks_delete_lock: fasync == %p\n", fl->fl_fasync); fl->fl_fasync = NULL; } } and a NULL implementation for delegations. From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Date: Sun, 24 Aug 2014 09:06:34 -0700 Subject: [Cluster-devel] [PATCH 08/10] locks: move i_lock acquisition into generic_*_lease handlers In-Reply-To: <1408804878-1331-9-git-send-email-jlayton@primarydata.com> References: <1408804878-1331-1-git-send-email-jlayton@primarydata.com> <1408804878-1331-9-git-send-email-jlayton@primarydata.com> Message-ID: <20140824160634.GG15908@infradead.org> List-Id: To: cluster-devel.redhat.com MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Looks good, Reviewed-by: Christoph Hellwig Some comments on further work I'd like to see in this area, though: > + spin_lock(&inode->i_lock); > + time_out_leases(inode); > for (before = &inode->i_flock; > ((fl = *before) != NULL) && IS_LEASE(fl); > before = &fl->fl_next) { > if (fl->fl_file != filp) > continue; > - return fl->fl_lmops->lm_change(before, F_UNLCK); > + error = fl->fl_lmops->lm_change(before, F_UNLCK); > } We really should split a lm_release from lm_change, the way it is used is highly confusing. In addition I think a lot of code currently in lease_modify should move here instead, e.g. something like: if (fl->fl_file != filp) continue; fl = *before; fl->fl_type = F_UNLCK; lease_clear_pending(fl, F_UNLCK); locks_wake_up_blocks(fl); if (fl->fl_lmops->lm_delete) fl->fl_lmops->lm_delete(fl); locks_delete_lock(before, NULL); with lm_delete for user space leases as: static void lease_delete(struct file_lock *fl) { struct file *filp = fl->fl_file; f_delown(filp); filp->f_owner.signum = 0; fasync_helper(0, fl->fl_file, 0, &fl->fl_fasync); if (fl->fl_fasync != NULL) { printk(KERN_ERR "locks_delete_lock: fasync == %p\n", fl->fl_fasync); fl->fl_fasync = NULL; } } and a NULL implementation for delegations.