From: "Michal Koutný" <mkoutny@suse.com> To: Christian Brauner <christian.brauner@ubuntu.com> Cc: linux-api@vger.kernel.org, linux-kernel@vger.kernel.org, Tejun Heo <tj@kernel.org>, Oleg Nesterov <oleg@redhat.com>, Ingo Molnar <mingo@redhat.com>, Johannes Weiner <hannes@cmpxchg.org>, Li Zefan <lizefan@huawei.com>, Peter Zijlstra <peterz@infradead.org>, cgroups@vger.kernel.org Subject: Re: [PATCH v5 5/6] clone3: allow spawning processes into cgroups Date: Wed, 29 Jan 2020 14:27:19 +0100 [thread overview] Message-ID: <20200129132719.GD11384@blackbody.suse.cz> (raw) In-Reply-To: <20200121154844.411-6-christian.brauner@ubuntu.com> Hello. On Tue, Jan 21, 2020 at 04:48:43PM +0100, Christian Brauner <christian.brauner@ubuntu.com> wrote: > +static int cgroup_css_set_fork(struct kernel_clone_args *kargs) > + __acquires(&cgroup_mutex) __acquires(&cgroup_threadgroup_rwsem) > +{ > + int ret; > + struct cgroup *dst_cgrp = NULL; > + struct css_set *cset; > + struct super_block *sb; > + struct file *f; > + > + if (kargs->flags & CLONE_INTO_CGROUP) > + mutex_lock(&cgroup_mutex); > + > + cgroup_threadgroup_change_begin(current); > + > + spin_lock_irq(&css_set_lock); > + cset = task_css_set(current); > + get_css_set(cset); > + spin_unlock_irq(&css_set_lock); > + > + if (!(kargs->flags & CLONE_INTO_CGROUP)) { > + kargs->cset = cset; Where is this css_set put when CLONE_INTO_CGROUP isn't used? (Aha, it's passed to child's tsk->cgroups but see my other note below.) > + dst_cgrp = cgroup_get_from_file(f); > + if (IS_ERR(dst_cgrp)) { > + ret = PTR_ERR(dst_cgrp); > + dst_cgrp = NULL; > + goto err; > + } > + > + /* > + * Verify that we the target cgroup is writable for us. This is > + * usually done by the vfs layer but since we're not going through > + * the vfs layer here we need to do it "manually". > + */ > + ret = cgroup_may_write(dst_cgrp, sb); > + if (ret) > + goto err; > + > + ret = cgroup_attach_permissions(cset->dfl_cgrp, dst_cgrp, sb, > + !!(kargs->flags & CLONE_THREAD)); > + if (ret) > + goto err; > + > + kargs->cset = find_css_set(cset, dst_cgrp); > + if (!kargs->cset) { > + ret = -ENOMEM; > + goto err; > + } > + > + if (cgroup_is_dead(dst_cgrp)) { > + ret = -ENODEV; > + goto err; > + } I'd move this check right after cgroup_get_from_file. The fork-migration path is synchrinized via cgroup_mutex with cgroup_destroy_locked and there's no need checking permissions on cgroup that's going away anyway. > +static void cgroup_css_set_put_fork(struct kernel_clone_args *kargs) > + __releases(&cgroup_threadgroup_rwsem) __releases(&cgroup_mutex) > +{ > + cgroup_threadgroup_change_end(current); > + > + if (kargs->flags & CLONE_INTO_CGROUP) { > + struct cgroup *cgrp = kargs->cgrp; > + struct css_set *cset = kargs->cset; > + > + mutex_unlock(&cgroup_mutex); > + > + if (cset) { > + put_css_set(cset); > + kargs->cset = NULL; > + } > + > + if (cgrp) { > + cgroup_put(cgrp); > + kargs->cgrp = NULL; > + } > + } I don't see any function problem with this ordering, however, I'd prefer symmetry with the "allocation" path (in cgroup_css_set_fork), i.e. cgroup_put, put_css_set and lastly mutex_unlock. > +void cgroup_post_fork(struct task_struct *child, > + struct kernel_clone_args *kargs) > + __releases(&cgroup_threadgroup_rwsem) __releases(&cgroup_mutex) > { > struct cgroup_subsys *ss; > - struct css_set *cset; > + struct css_set *cset = kargs->cset; > int i; > > spin_lock_irq(&css_set_lock); > > WARN_ON_ONCE(!list_empty(&child->cg_list)); > - cset = task_css_set(current); /* current is @child's parent */ > - get_css_set(cset); > cset->nr_tasks++; > css_set_move_task(child, NULL, cset, false); So, the reference is passed over from kargs->cset to task->cgroups. I think it's necessary to zero kargs->cset in order to prevent droping the reference in cgroup_css_set_put_fork. Perhaps, a general comment about css_set whereabouts during fork and kargs passing would be useful. > @@ -6016,6 +6146,17 @@ void cgroup_post_fork(struct task_struct *child) > } while_each_subsys_mask(); > > cgroup_threadgroup_change_end(current); > + > + if (kargs->flags & CLONE_INTO_CGROUP) { > + mutex_unlock(&cgroup_mutex); > + > + cgroup_put(kargs->cgrp); > + kargs->cgrp = NULL; > + } > + > + /* Make the new cset the root_cset of the new cgroup namespace. */ > + if (kargs->flags & CLONE_NEWCGROUP) > + child->nsproxy->cgroup_ns->root_cset = cset; root_cset reference (from copy_cgroup_ns) seems leaked here and where is the additional reference to new cset obtained? Thanks, Michal
WARNING: multiple messages have this Message-ID (diff)
From: "Michal Koutný" <mkoutny-IBi9RG/b67k@public.gmane.org> To: Christian Brauner <christian.brauner-GeWIH/nMZzLQT0dZR+AlfA@public.gmane.org> Cc: linux-api-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>, Oleg Nesterov <oleg-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>, Ingo Molnar <mingo-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>, Johannes Weiner <hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org>, Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>, Peter Zijlstra <peterz-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>, cgroups-u79uwXL29TY76Z2rM5mHXA@public.gmane.org Subject: Re: [PATCH v5 5/6] clone3: allow spawning processes into cgroups Date: Wed, 29 Jan 2020 14:27:19 +0100 [thread overview] Message-ID: <20200129132719.GD11384@blackbody.suse.cz> (raw) In-Reply-To: <20200121154844.411-6-christian.brauner-GeWIH/nMZzLQT0dZR+AlfA@public.gmane.org> Hello. On Tue, Jan 21, 2020 at 04:48:43PM +0100, Christian Brauner <christian.brauner-GeWIH/nMZzLQT0dZR+AlfA@public.gmane.org> wrote: > +static int cgroup_css_set_fork(struct kernel_clone_args *kargs) > + __acquires(&cgroup_mutex) __acquires(&cgroup_threadgroup_rwsem) > +{ > + int ret; > + struct cgroup *dst_cgrp = NULL; > + struct css_set *cset; > + struct super_block *sb; > + struct file *f; > + > + if (kargs->flags & CLONE_INTO_CGROUP) > + mutex_lock(&cgroup_mutex); > + > + cgroup_threadgroup_change_begin(current); > + > + spin_lock_irq(&css_set_lock); > + cset = task_css_set(current); > + get_css_set(cset); > + spin_unlock_irq(&css_set_lock); > + > + if (!(kargs->flags & CLONE_INTO_CGROUP)) { > + kargs->cset = cset; Where is this css_set put when CLONE_INTO_CGROUP isn't used? (Aha, it's passed to child's tsk->cgroups but see my other note below.) > + dst_cgrp = cgroup_get_from_file(f); > + if (IS_ERR(dst_cgrp)) { > + ret = PTR_ERR(dst_cgrp); > + dst_cgrp = NULL; > + goto err; > + } > + > + /* > + * Verify that we the target cgroup is writable for us. This is > + * usually done by the vfs layer but since we're not going through > + * the vfs layer here we need to do it "manually". > + */ > + ret = cgroup_may_write(dst_cgrp, sb); > + if (ret) > + goto err; > + > + ret = cgroup_attach_permissions(cset->dfl_cgrp, dst_cgrp, sb, > + !!(kargs->flags & CLONE_THREAD)); > + if (ret) > + goto err; > + > + kargs->cset = find_css_set(cset, dst_cgrp); > + if (!kargs->cset) { > + ret = -ENOMEM; > + goto err; > + } > + > + if (cgroup_is_dead(dst_cgrp)) { > + ret = -ENODEV; > + goto err; > + } I'd move this check right after cgroup_get_from_file. The fork-migration path is synchrinized via cgroup_mutex with cgroup_destroy_locked and there's no need checking permissions on cgroup that's going away anyway. > +static void cgroup_css_set_put_fork(struct kernel_clone_args *kargs) > + __releases(&cgroup_threadgroup_rwsem) __releases(&cgroup_mutex) > +{ > + cgroup_threadgroup_change_end(current); > + > + if (kargs->flags & CLONE_INTO_CGROUP) { > + struct cgroup *cgrp = kargs->cgrp; > + struct css_set *cset = kargs->cset; > + > + mutex_unlock(&cgroup_mutex); > + > + if (cset) { > + put_css_set(cset); > + kargs->cset = NULL; > + } > + > + if (cgrp) { > + cgroup_put(cgrp); > + kargs->cgrp = NULL; > + } > + } I don't see any function problem with this ordering, however, I'd prefer symmetry with the "allocation" path (in cgroup_css_set_fork), i.e. cgroup_put, put_css_set and lastly mutex_unlock. > +void cgroup_post_fork(struct task_struct *child, > + struct kernel_clone_args *kargs) > + __releases(&cgroup_threadgroup_rwsem) __releases(&cgroup_mutex) > { > struct cgroup_subsys *ss; > - struct css_set *cset; > + struct css_set *cset = kargs->cset; > int i; > > spin_lock_irq(&css_set_lock); > > WARN_ON_ONCE(!list_empty(&child->cg_list)); > - cset = task_css_set(current); /* current is @child's parent */ > - get_css_set(cset); > cset->nr_tasks++; > css_set_move_task(child, NULL, cset, false); So, the reference is passed over from kargs->cset to task->cgroups. I think it's necessary to zero kargs->cset in order to prevent droping the reference in cgroup_css_set_put_fork. Perhaps, a general comment about css_set whereabouts during fork and kargs passing would be useful. > @@ -6016,6 +6146,17 @@ void cgroup_post_fork(struct task_struct *child) > } while_each_subsys_mask(); > > cgroup_threadgroup_change_end(current); > + > + if (kargs->flags & CLONE_INTO_CGROUP) { > + mutex_unlock(&cgroup_mutex); > + > + cgroup_put(kargs->cgrp); > + kargs->cgrp = NULL; > + } > + > + /* Make the new cset the root_cset of the new cgroup namespace. */ > + if (kargs->flags & CLONE_NEWCGROUP) > + child->nsproxy->cgroup_ns->root_cset = cset; root_cset reference (from copy_cgroup_ns) seems leaked here and where is the additional reference to new cset obtained? Thanks, Michal
next prev parent reply other threads:[~2020-01-29 13:27 UTC|newest] Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top 2020-01-21 15:48 [PATCH v5 0/6] clone3 & cgroups: allow spawning processes into cgroups Christian Brauner 2020-01-21 15:48 ` Christian Brauner 2020-01-21 15:48 ` [PATCH v5 1/6] cgroup: unify attach permission checking Christian Brauner 2020-01-29 13:25 ` Michal Koutný 2020-01-21 15:48 ` [PATCH v5 2/6] cgroup: add cgroup_get_from_file() helper Christian Brauner 2020-01-29 13:25 ` Michal Koutný 2020-01-29 13:25 ` Michal Koutný 2020-01-21 15:48 ` [PATCH v5 3/6] cgroup: refactor fork helpers Christian Brauner 2020-01-29 13:26 ` Michal Koutný 2020-01-29 13:26 ` Michal Koutný 2020-01-21 15:48 ` [PATCH v5 4/6] cgroup: add cgroup_may_write() helper Christian Brauner 2020-01-21 15:48 ` [PATCH v5 5/6] clone3: allow spawning processes into cgroups Christian Brauner 2020-01-21 15:48 ` Christian Brauner 2020-01-21 15:48 ` Christian Brauner 2020-01-29 13:27 ` Michal Koutný [this message] 2020-01-29 13:27 ` Michal Koutný 2020-02-02 9:37 ` Christian Brauner 2020-02-02 9:37 ` Christian Brauner 2020-02-02 9:37 ` Christian Brauner 2020-02-03 14:32 ` Michal Koutný 2020-02-03 14:32 ` Michal Koutný 2020-02-04 11:13 ` Christian Brauner 2020-02-04 11:13 ` Christian Brauner 2020-02-04 11:13 ` Christian Brauner 2020-02-04 11:53 ` Peter Zijlstra 2020-02-04 15:01 ` Christian Brauner 2020-02-04 15:01 ` Christian Brauner 2020-01-21 15:48 ` [PATCH v5 6/6] selftests/cgroup: add tests for cloning " Christian Brauner
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20200129132719.GD11384@blackbody.suse.cz \ --to=mkoutny@suse.com \ --cc=cgroups@vger.kernel.org \ --cc=christian.brauner@ubuntu.com \ --cc=hannes@cmpxchg.org \ --cc=linux-api@vger.kernel.org \ --cc=linux-kernel@vger.kernel.org \ --cc=lizefan@huawei.com \ --cc=mingo@redhat.com \ --cc=oleg@redhat.com \ --cc=peterz@infradead.org \ --cc=tj@kernel.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.