Skip to contentSkip to conversation

signal, procfs: lock_task_sighand() do not need rcu_read_lock()

PATCH In review Lai Jiangshan 3 participants 3 messages Oct 3, 2008 – Oct 4, 2008 Proc filesystem

lore
Share & tools
lore

Lai Jiangshan

author
#0 lore raw

PATCH 1/1 signal, procfs: lock_task_sighand() do not need rcu_read_lock()

lock_task_sighand() make sure task->sighand is being protected, so we do not need rcu_read_lock().
[ exec() will get task->sighand->siglock before change task->sighand! ]

But code using rcu_read_lock() _just_ to protect lock_task_sighand() only appear in procfs. (and some code in procfs use lock_task_sighand() without such redundant protection.)

Other subsystem may put lock_task_sighand() into rcu_read_lock() critical region, but these rcu_read_lock() are used for protecting "for_each_process()", "find_task_by_vpid()" etc. , not for protecting lock_task_sighand().

Signed-off-by Lai Jiangshan <[email protected]>

---

fs/proc/array.c +0 -2
index 71c9be5..1c8d7b5 100644
@@ -261,7 +261,6 @@ static inline void task_sig(struct seq_file *m, struct task_struct *p)
 	sigemptyset(&ignored);
 	sigemptyset(&caught);
 
-	rcu_read_lock();
 	if (lock_task_sighand(p, &flags)) {
 		pending = p->pending.signal;
 		shpending = p->signal->shared_pending.signal;

@@ -272,7 +271,6 @@ static inline void task_sig(struct seq_file *m, struct task_struct *p)
 		qlim = p->signal->rlim[RLIMIT_SIGPENDING].rlim_cur;
 		unlock_task_sighand(p, &flags);
 	}
-	rcu_read_unlock();
 
 	seq_printf(m, "Threads:\t%d\n", num_threads);
 	seq_printf(m, "SigQ:\t%lu/%lu\n", qsize, qlim);
fs/proc/base.c +1 -8
index a28840b..bb63fa1 100644
@@ -164,7 +164,6 @@ static struct fs_struct *get_fs_struct(struct task_struct *task)
 
 static int get_nr_threads(struct task_struct *tsk)
 {
-	/* Must be called with the rcu_read_lock held */
 	unsigned long flags;
 	int count = 0;
 

@@ -471,14 +470,10 @@ static int proc_pid_limits(struct task_struct *task, char *buffer)
 
 	struct rlimit rlim[RLIM_NLIMITS];
 
-	rcu_read_lock();
-	if (!lock_task_sighand(task,&flags)) {
-		rcu_read_unlock();
+	if (!lock_task_sighand(task, &flags))
 		return 0;
-	}
 	memcpy(rlim, task->signal->rlim, sizeof(struct rlimit) * RLIM_NLIMITS);
 	unlock_task_sighand(task, &flags);
-	rcu_read_unlock();
 
 	/*
 	 * print the file header

@@ -3088,9 +3083,7 @@ static int proc_task_getattr(struct vfsmount *mnt, struct dentry *dentry, struct
 	generic_fillattr(inode, stat);
 
 	if (p) {
-		rcu_read_lock();
 		stat->nlink += get_nr_threads(p);
-		rcu_read_unlock();
 		put_task_struct(p);
 	}
kernel/sched_debug.c +0 -2
index bbe6b31..ad958c1 100644
@@ -333,12 +333,10 @@ void proc_sched_show_task(struct task_struct *p, struct seq_file *m)
 	unsigned long flags;
 	int num_threads = 1;
 
-	rcu_read_lock();
 	if (lock_task_sighand(p, &flags)) {
 		num_threads = atomic_read(&p->signal->count);
 		unlock_task_sighand(p, &flags);
 	}
-	rcu_read_unlock();
 
 	SEQ_printf(m, "%s (%d, #threads: %d)\n", p->comm, p->pid, num_threads);
 	SEQ_printf(m,
#1 lore raw
Quoted 8 lines · Lai Jiangshan

lock_task_sighand() make sure task->sighand is being protected, so we do not need rcu_read_lock().
[ exec() will get task->sighand->siglock before change task->sighand! ]

But code using rcu_read_lock() _just_ to protect lock_task_sighand() only appear in procfs. (and some code in procfs use lock_task_sighand() without such redundant protection.)

Yes, the patch looks correct.

The initial implementaion of lock_task_sighand() did not take RCU lock, that is why we still have the callers which call rcu_read_lock(). See 1406f2d321bae5ac5ff729dcb773336d9c05ec74.

Thanks!

Oleg.

#2 lore raw
Quoted 11 lines · Oleg Nesterov
Quoted 8 lines · Lai Jiangshan

lock_task_sighand() make sure task->sighand is being protected, so we do not need rcu_read_lock().
[ exec() will get task->sighand->siglock before change task->sighand! ]

But code using rcu_read_lock() _just_ to protect lock_task_sighand() only appear in procfs. (and some code in procfs use lock_task_sighand() without such redundant protection.)

Yes, the patch looks correct.

Yeah, applied to proc.git