When threadgroup_change_begin/end() are called from fork path, pass in @child and @clone_flags so that fork path can be distinguished and fork related information is available.
While at it, un-inline cgroup_threadgroup_change_begin/end() and fold cgroup_fork() into cgroup_threadgroup_change_begin(). These changes will be used to imlement in-process resource control. Signed-off-by: Tejun Heo <t...@kernel.org> Cc: Ingo Molnar <mi...@redhat.com> Cc: Peter Zijlstra <pet...@infradead.org> Cc: Oleg Nesterov <o...@redhat.com> --- fs/exec.c | 6 +++--- include/linux/cgroup-defs.h | 39 ++++++++++++------------------------- include/linux/cgroup.h | 2 -- include/linux/sched.h | 16 +++++++++++---- kernel/cgroup.c | 47 ++++++++++++++++++++++++++++++++++++--------- kernel/fork.c | 7 +++---- kernel/signal.c | 6 +++--- 7 files changed, 71 insertions(+), 52 deletions(-) diff --git a/fs/exec.c b/fs/exec.c index 828ec5f..5b81bbb 100644 --- a/fs/exec.c +++ b/fs/exec.c @@ -936,7 +936,7 @@ static int de_thread(struct task_struct *tsk) struct task_struct *leader = tsk->group_leader; for (;;) { - threadgroup_change_begin(tsk); + threadgroup_change_begin(tsk, NULL, 0); write_lock_irq(&tasklist_lock); /* * Do this under tasklist_lock to ensure that @@ -947,7 +947,7 @@ static int de_thread(struct task_struct *tsk) break; __set_current_state(TASK_KILLABLE); write_unlock_irq(&tasklist_lock); - threadgroup_change_end(tsk); + threadgroup_change_end(tsk, NULL, 0); schedule(); if (unlikely(__fatal_signal_pending(tsk))) goto killed; @@ -1005,7 +1005,7 @@ static int de_thread(struct task_struct *tsk) if (unlikely(leader->ptrace)) __wake_up_parent(leader, leader->parent); write_unlock_irq(&tasklist_lock); - threadgroup_change_end(tsk); + threadgroup_change_end(tsk, NULL, 0); release_task(leader); } diff --git a/include/linux/cgroup-defs.h b/include/linux/cgroup-defs.h index d3d1f92..3c4a75b 100644 --- a/include/linux/cgroup-defs.h +++ b/include/linux/cgroup-defs.h @@ -514,31 +514,12 @@ struct cgroup_subsys { unsigned int depends_on; }; -extern struct percpu_rw_semaphore cgroup_threadgroup_rwsem; - -/** - * cgroup_threadgroup_change_begin - threadgroup exclusion for cgroups - * @tsk: target task - * - * Called from threadgroup_change_begin() and allows cgroup operations to - * synchronize against threadgroup changes using a percpu_rw_semaphore. - */ -static inline void cgroup_threadgroup_change_begin(struct task_struct *tsk) -{ - percpu_down_read(&cgroup_threadgroup_rwsem); -} - -/** - * cgroup_threadgroup_change_end - threadgroup exclusion for cgroups - * @tsk: target task - * - * Called from threadgroup_change_end(). Counterpart of - * cgroup_threadcgroup_change_begin(). - */ -static inline void cgroup_threadgroup_change_end(struct task_struct *tsk) -{ - percpu_up_read(&cgroup_threadgroup_rwsem); -} +void cgroup_threadgroup_change_begin(struct task_struct *tsk, + struct task_struct *child, + unsigned long clone_flags); +void cgroup_threadgroup_change_end(struct task_struct *tsk, + struct task_struct *child, + unsigned long clone_flags); #else /* CONFIG_CGROUPS */ @@ -546,8 +527,12 @@ struct css_set; #define CGROUP_SUBSYS_COUNT 0 -static inline void cgroup_threadgroup_change_begin(struct task_struct *tsk) {} -static inline void cgroup_threadgroup_change_end(struct task_struct *tsk) {} +static inline void cgroup_threadgroup_change_begin(struct task_struct *tsk, + struct task_struct *child, + unsigned long clone_flags) {} +static inline void cgroup_threadgroup_change_end(struct task_struct *tsk, + struct task_struct *child, + unsigned long clone_flags) {} #endif /* CONFIG_CGROUPS */ diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h index ebcf21f..1e00fc0 100644 --- a/include/linux/cgroup.h +++ b/include/linux/cgroup.h @@ -101,7 +101,6 @@ int cgroupstats_build(struct cgroupstats *stats, struct dentry *dentry); int proc_cgroup_show(struct seq_file *m, struct pid_namespace *ns, struct pid *pid, struct task_struct *tsk); -void cgroup_fork(struct task_struct *p); extern int cgroup_can_fork(struct task_struct *p, unsigned long clone_flags, struct css_set **new_rgrp_csetp); extern void cgroup_cancel_fork(struct task_struct *p, unsigned long clone_flags, @@ -540,7 +539,6 @@ static inline int cgroup_attach_task_all(struct task_struct *from, static inline int cgroupstats_build(struct cgroupstats *stats, struct dentry *dentry) { return -EINVAL; } -static inline void cgroup_fork(struct task_struct *p) {} static inline int cgroup_can_fork(struct task_struct *p, unsigned long clone_flags, struct css_set **new_rgrp_csetp) { return 0; } diff --git a/include/linux/sched.h b/include/linux/sched.h index 80d6ed1..d4ae795 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -2783,6 +2783,8 @@ static inline void unlock_task_sighand(struct task_struct *tsk, /** * threadgroup_change_begin - mark the beginning of changes to a threadgroup * @tsk: task causing the changes + * @child: child task if forking, NULL otherwise + * @clone_flags: clone flags if forking * * All operations which modify a threadgroup - a new thread joining the * group, death of a member thread (the assertion of PF_EXITING) and @@ -2791,21 +2793,27 @@ static inline void unlock_task_sighand(struct task_struct *tsk, * subsystems needing threadgroup stability can hook into for * synchronization. */ -static inline void threadgroup_change_begin(struct task_struct *tsk) +static inline void threadgroup_change_begin(struct task_struct *tsk, + struct task_struct *child, + unsigned long clone_flags) { might_sleep(); - cgroup_threadgroup_change_begin(tsk); + cgroup_threadgroup_change_begin(tsk, child, clone_flags); } /** * threadgroup_change_end - mark the end of changes to a threadgroup * @tsk: task causing the changes + * @child: child task if forking, NULL otherwise + * @clone_flags: clone flags if forking * * See threadgroup_change_begin(). */ -static inline void threadgroup_change_end(struct task_struct *tsk) +static inline void threadgroup_change_end(struct task_struct *tsk, + struct task_struct *child, + unsigned long clone_flags) { - cgroup_threadgroup_change_end(tsk); + cgroup_threadgroup_change_end(tsk, child, clone_flags); } #ifndef __HAVE_THREAD_FUNCTIONS diff --git a/kernel/cgroup.c b/kernel/cgroup.c index ac207ae..70f9985 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -110,7 +110,7 @@ static DEFINE_SPINLOCK(cgroup_file_kn_lock); */ static DEFINE_SPINLOCK(release_agent_path_lock); -struct percpu_rw_semaphore cgroup_threadgroup_rwsem; +static struct percpu_rw_semaphore cgroup_threadgroup_rwsem; #define cgroup_assert_mutex_or_rcu_locked() \ RCU_LOCKDEP_WARN(!rcu_read_lock_held() && \ @@ -5688,17 +5688,46 @@ static const struct file_operations proc_cgroupstats_operations = { }; /** - * cgroup_fork - initialize cgroup related fields during copy_process() - * @child: pointer to task_struct of forking parent process. + * cgroup_threadgroup_change_begin - threadgroup exclusion for cgroups + * @tsk: target task + * @child: child task if forking, NULL otherwise + * @clone_flags: clone flags if forking * - * A task is associated with the init_css_set until cgroup_post_fork() - * attaches it to the parent's css_set. Empty cg_list indicates that - * @child isn't holding reference to its css_set. + * Called from threadgroup_change_begin() and allows cgroup operations to + * synchronize against threadgroup changes using a percpu_rw_semaphore. */ -void cgroup_fork(struct task_struct *child) +void cgroup_threadgroup_change_begin(struct task_struct *tsk, + struct task_struct *child, + unsigned long clone_flags) { - RCU_INIT_POINTER(child->cgroups, &init_css_set); - INIT_LIST_HEAD(&child->cg_list); + if (child) { + /* + * A task is associated with the init_css_set until + * cgroup_post_fork() attaches it to the parent's css_set. + * Empty cg_list indicates that @child isn't holding + * reference to its css_set. + */ + RCU_INIT_POINTER(child->cgroups, &init_css_set); + INIT_LIST_HEAD(&child->cg_list); + } + + percpu_down_read(&cgroup_threadgroup_rwsem); +} + +/** + * cgroup_threadgroup_change_end - threadgroup exclusion for cgroups + * @tsk: target task + * @child: child task if forking, NULL otherwise + * @clone_flags: clone flags if forking + * + * Called from threadgroup_change_end(). Counterpart of + * cgroup_threadcgroup_change_begin(). + */ +void cgroup_threadgroup_change_end(struct task_struct *tsk, + struct task_struct *child, + unsigned long clone_flags) +{ + percpu_up_read(&cgroup_threadgroup_rwsem); } /** diff --git a/kernel/fork.c b/kernel/fork.c index 812d477..840b662 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -1368,8 +1368,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, p->real_start_time = ktime_get_boot_ns(); p->io_context = NULL; p->audit_context = NULL; - threadgroup_change_begin(current); - cgroup_fork(p); + threadgroup_change_begin(current, p, clone_flags); #ifdef CONFIG_NUMA p->mempolicy = mpol_dup(p->mempolicy); if (IS_ERR(p->mempolicy)) { @@ -1609,7 +1608,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, proc_fork_connector(p); cgroup_post_fork(p, clone_flags, new_rgrp_cset); - threadgroup_change_end(current); + threadgroup_change_end(current, p, clone_flags); perf_event_fork(p); trace_task_newtask(p, clone_flags); @@ -1650,7 +1649,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, mpol_put(p->mempolicy); bad_fork_cleanup_threadgroup_lock: #endif - threadgroup_change_end(current); + threadgroup_change_end(current, p, clone_flags); delayacct_tsk_free(p); bad_fork_cleanup_count: atomic_dec(&p->cred->user->processes); diff --git a/kernel/signal.c b/kernel/signal.c index f3f1f7a..1679c02 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -2389,11 +2389,11 @@ void exit_signals(struct task_struct *tsk) * @tsk is about to have PF_EXITING set - lock out users which * expect stable threadgroup. */ - threadgroup_change_begin(tsk); + threadgroup_change_begin(tsk, NULL, 0); if (thread_group_empty(tsk) || signal_group_exit(tsk->signal)) { tsk->flags |= PF_EXITING; - threadgroup_change_end(tsk); + threadgroup_change_end(tsk, NULL, 0); return; } @@ -2404,7 +2404,7 @@ void exit_signals(struct task_struct *tsk) */ tsk->flags |= PF_EXITING; - threadgroup_change_end(tsk); + threadgroup_change_end(tsk, NULL, 0); if (!signal_pending(tsk)) goto out; -- 2.5.0