Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755602AbZGBCL4 (ORCPT ); Wed, 1 Jul 2009 22:11:56 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754743AbZGBCLV (ORCPT ); Wed, 1 Jul 2009 22:11:21 -0400 Received: from smtp-out.google.com ([216.239.33.17]:22106 "EHLO smtp-out.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754388AbZGBCLU (ORCPT ); Wed, 1 Jul 2009 22:11:20 -0400 DomainKey-Signature: a=rsa-sha1; s=beta; d=google.com; c=nofws; q=dns; h=from:subject:to:cc:date:message-id:in-reply-to:references: user-agent:mime-version:content-type: content-transfer-encoding:x-system-of-record; b=SKNcoFvXSBLfws4gKlGbGGXex0EqDZM+I+65XUX2vt28W1YyrBtMvogW9gpHEEnKR P/rs+k55GNJ851bsXtytQ== From: Paul Menage Subject: [PATCH 1/9] [RFC] Support named cgroups hierarchies To: lizf@cn.fujitsu.com, balbir@linux.vnet.ibm.com Cc: linux-kernel@vger.kernel.org, akpm@linux-foundation.org, containers@lists.linux-foundation.org, kamezawa.hiroyu@jp.fujitsu.com Date: Wed, 01 Jul 2009 19:10:58 -0700 Message-ID: <20090702021057.14469.37548.stgit@menage.mtv.corp.google.com> In-Reply-To: <20090702020624.14469.47066.stgit@menage.mtv.corp.google.com> References: <20090702020624.14469.47066.stgit@menage.mtv.corp.google.com> User-Agent: StGIT/0.14.3 MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit X-System-Of-Record: true Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 9272 Lines: 321 [RFC] Support named cgroups hierarchies To simplify referring to cgroup hierarchies in mount statements, and to allow disambiguation in the presence of empty hierarchies and multiply-bindable subsystems (see later patches in series) this patch adds support for naming a new cgroup hierarchy via the "name=" mount option A pre-existing hierarchy may be specified by either name or by subsystems; a hierarchy's name cannot be changed by a remount operation. Example usage: # To create a hierarchy called "foo" containing the "cpu" subsystem mount -t cgroup -oname=foo,cpu cgroup /mnt/cgroup1 # To mount the "foo" hierarchy on a second location mount -t cgroup -oname=foo cgroup /mnt/cgroup2 Open issues: - should the specification be via a name= option as in this patch, or should we simply use the "device name" as passed to the mount() system call? Using the device name is more conceptually clean and consistent with the filesystem API; however, given that the device name is currently ignored by cgroups, this would lead to a user-visible behaviour change. Signed-off-by: Paul Menage --- kernel/cgroup.c | 136 ++++++++++++++++++++++++++++++++++++------------------- 1 files changed, 88 insertions(+), 48 deletions(-) diff --git a/kernel/cgroup.c b/kernel/cgroup.c index ea255fe..940f28d 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -59,6 +59,8 @@ static struct cgroup_subsys *subsys[] = { #include }; +#define MAX_CGROUP_ROOT_NAMELEN 64 + /* * A cgroupfs_root represents the root of a cgroup hierarchy, * and may be associated with a superblock to form an active @@ -93,6 +95,9 @@ struct cgroupfs_root { /* The path to use for release notifications. */ char release_agent_path[PATH_MAX]; + + /* The name for this hierarchy - may be empty */ + char name[MAX_CGROUP_ROOT_NAMELEN]; }; /* @@ -828,6 +833,8 @@ static int cgroup_show_options(struct seq_file *seq, struct vfsmount *vfs) seq_puts(seq, ",noprefix"); if (strlen(root->release_agent_path)) seq_printf(seq, ",release_agent=%s", root->release_agent_path); + if (strlen(root->name)) + seq_printf(seq, ",name=%s", root->name); mutex_unlock(&cgroup_mutex); return 0; } @@ -836,12 +843,15 @@ struct cgroup_sb_opts { unsigned long subsys_bits; unsigned long flags; char *release_agent; + char *name; + /* A flag indicating that a root was created from this options block */ + bool created_root; }; /* Convert a hierarchy specifier into a bitmask of subsystems and * flags. */ static int parse_cgroupfs_options(char *data, - struct cgroup_sb_opts *opts) + struct cgroup_sb_opts *opts) { char *token, *o = data ?: "all"; unsigned long mask = (unsigned long)-1; @@ -850,9 +860,7 @@ static int parse_cgroupfs_options(char *data, mask = ~(1UL << cpuset_subsys_id); #endif - opts->subsys_bits = 0; - opts->flags = 0; - opts->release_agent = NULL; + memset(opts, 0, sizeof(*opts)); while ((token = strsep(&o, ",")) != NULL) { if (!*token) @@ -872,11 +880,19 @@ static int parse_cgroupfs_options(char *data, /* Specifying two release agents is forbidden */ if (opts->release_agent) return -EINVAL; - opts->release_agent = kzalloc(PATH_MAX, GFP_KERNEL); + opts->release_agent = + kstrndup(token + 14, PATH_MAX, GFP_KERNEL); if (!opts->release_agent) return -ENOMEM; - strncpy(opts->release_agent, token + 14, PATH_MAX - 1); - opts->release_agent[PATH_MAX - 1] = 0; + } else if (!strncmp(token, "name=", 5)) { + /* Specifying two names is forbidden */ + if (opts->name) + return -EINVAL; + opts->name = kstrndup(token + 5, + MAX_CGROUP_ROOT_NAMELEN, + GFP_KERNEL); + if (!opts->name) + return -ENOMEM; } else { struct cgroup_subsys *ss; int i; @@ -903,7 +919,7 @@ static int parse_cgroupfs_options(char *data, return -EINVAL; /* We can't have an empty hierarchy */ - if (!opts->subsys_bits) + if (!opts->subsys_bits && !opts->name) return -EINVAL; return 0; @@ -931,6 +947,12 @@ static int cgroup_remount(struct super_block *sb, int *flags, char *data) goto out_unlock; } + /* Don't allow name to change at remount */ + if (opts.name && strcmp(opts.name, root->name)) { + ret = -EINVAL; + goto out_unlock; + } + ret = rebind_subsystems(root, opts.subsys_bits); if (ret) goto out_unlock; @@ -942,6 +964,7 @@ static int cgroup_remount(struct super_block *sb, int *flags, char *data) strcpy(root->release_agent_path, opts.release_agent); out_unlock: kfree(opts.release_agent); + kfree(opts.name); mutex_unlock(&cgroup_mutex); mutex_unlock(&cgrp->dentry->d_inode->i_mutex); unlock_kernel(); @@ -963,6 +986,7 @@ static void init_cgroup_housekeeping(struct cgroup *cgrp) INIT_LIST_HEAD(&cgrp->release_list); init_rwsem(&cgrp->pids_mutex); } + static void init_cgroup_root(struct cgroupfs_root *root) { struct cgroup *cgrp = &root->top_cgroup; @@ -976,28 +1000,56 @@ static void init_cgroup_root(struct cgroupfs_root *root) static int cgroup_test_super(struct super_block *sb, void *data) { - struct cgroupfs_root *new = data; + struct cgroup_sb_opts *new = data; struct cgroupfs_root *root = sb->s_fs_info; - /* First check subsystems */ - if (new->subsys_bits != root->subsys_bits) - return 0; + /* If we asked for a name then it must match */ + if (new->name && strcmp(new->name, root->name)) + return 0; - /* Next check flags */ - if (new->flags != root->flags) + /* If we asked for subsystems then they must match */ + if (new->subsys_bits && new->subsys_bits != root->subsys_bits) return 0; return 1; } +static struct cgroupfs_root *cgroup_root_from_opts(struct cgroup_sb_opts *opts) +{ + struct cgroupfs_root *root; + + if (!opts->subsys_bits) + return ERR_PTR(-EINVAL); + + root = kzalloc(sizeof(*root), GFP_KERNEL); + if (!root) + return ERR_PTR(-ENOMEM); + + init_cgroup_root(root); + root->subsys_bits = opts->subsys_bits; + root->flags = opts->flags; + if (opts->release_agent) + strcpy(root->release_agent_path, opts->release_agent); + if (opts->name) + strcpy(root->name, opts->name); + opts->created_root = true; + return root; +} + static int cgroup_set_super(struct super_block *sb, void *data) { int ret; - struct cgroupfs_root *root = data; + struct cgroup_sb_opts *opts = data; + struct cgroupfs_root *root; + root = cgroup_root_from_opts(opts); + if (IS_ERR(root)) + return PTR_ERR(root); ret = set_anon_super(sb, NULL); - if (ret) + if (ret) { + kfree(root); return ret; + } sb->s_fs_info = root; root->sb = sb; @@ -1039,44 +1091,23 @@ static int cgroup_get_sb(struct file_system_type *fs_type, struct cgroup_sb_opts opts; int ret = 0; struct super_block *sb; - struct cgroupfs_root *root; - struct list_head tmp_cg_links; /* First find the desired set of subsystems */ ret = parse_cgroupfs_options(data, &opts); - if (ret) { - kfree(opts.release_agent); - return ret; - } - - root = kzalloc(sizeof(*root), GFP_KERNEL); - if (!root) { - kfree(opts.release_agent); - return -ENOMEM; - } - - init_cgroup_root(root); - root->subsys_bits = opts.subsys_bits; - root->flags = opts.flags; - if (opts.release_agent) { - strcpy(root->release_agent_path, opts.release_agent); - kfree(opts.release_agent); - } + if (ret) + goto out_err; - sb = sget(fs_type, cgroup_test_super, cgroup_set_super, root); + sb = sget(fs_type, cgroup_test_super, cgroup_set_super, &opts); if (IS_ERR(sb)) { - kfree(root); - return PTR_ERR(sb); + ret = PTR_ERR(sb); + goto out_err; } - if (sb->s_fs_info != root) { - /* Reusing an existing superblock */ - BUG_ON(sb->s_root == NULL); - kfree(root); - root = NULL; - } else { + if (opts.created_root) { /* New superblock */ + struct cgroupfs_root *root = sb->s_fs_info; + struct list_head tmp_cg_links; struct cgroup *root_cgrp = &root->top_cgroup; struct inode *inode; int i; @@ -1109,7 +1140,8 @@ static int cgroup_get_sb(struct file_system_type *fs_type, if (ret == -EBUSY) { mutex_unlock(&cgroup_mutex); mutex_unlock(&inode->i_mutex); - goto free_cg_links; + free_cg_links(&tmp_cg_links); + goto drop_new_super; } /* EBUSY should be the only error here */ @@ -1146,12 +1178,17 @@ static int cgroup_get_sb(struct file_system_type *fs_type, } simple_set_mnt(mnt, sb); + kfree(opts.release_agent); + kfree(opts.name); return 0; - free_cg_links: - free_cg_links(&tmp_cg_links); drop_new_super: deactivate_locked_super(sb); + + out_err: + kfree(opts.release_agent); + kfree(opts.name); + return ret; } @@ -2923,6 +2960,9 @@ static int proc_cgroup_show(struct seq_file *m, void *v) seq_printf(m, "%lu:", root->subsys_bits); for_each_subsys(root, ss) seq_printf(m, "%s%s", count++ ? "," : "", ss->name); + if (strlen(root->name)) + seq_printf(m, "%sname=%s", + count ? "," : "", root->name); seq_putc(m, ':'); get_first_subsys(&root->top_cgroup, NULL, &subsys_id); cgrp = task_cgroup(tsk, subsys_id); -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/