signals: Allow rt tasks to cache one sigqueue struct

To avoid allocation allow rt tasks to cache one sigqueue struct in
task struct.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

5 files changed, 76 insertions, 10 deletions

diff --git a/include/linux/sched.h b/include/linux/sched.h
index bb28896ffd6a..b71161f2c2bc 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1701,6 +1701,7 @@ struct task_struct {
 /* signal handlers */
 	struct signal_struct *signal;
 	struct sighand_struct *sighand;
+	struct sigqueue *sigqueue_cache;
 
 	sigset_t blocked, real_blocked;
 	sigset_t saved_sigmask;	/* restored if set_restore_sigmask() was used */
diff --git a/include/linux/signal.h b/include/linux/signal.h
index ffa58ff53e22..0d7faf67a464 100644
--- a/include/linux/signal.h
+++ b/include/linux/signal.h
@@ -250,6 +250,7 @@ static inline void init_sigpending(struct sigpending *sig)
 }
 
 extern void flush_sigqueue(struct sigpending *queue);
+extern void flush_task_sigqueue(struct task_struct *tsk);
 
 /* Test if 'sig' is valid signal. Use this instead of testing _NSIG directly */
 static inline int valid_signal(unsigned long sig)
diff --git a/kernel/exit.c b/kernel/exit.c
index 8716f0780fe3..89e38e5929a9 100644
--- a/kernel/exit.c
+++ b/kernel/exit.c
@@ -143,7 +143,7 @@ static void __exit_signal(struct task_struct *tsk)
 	 * Do this under ->siglock, we can race with another thread
 	 * doing sigqueue_free() if we have SIGQUEUE_PREALLOC signals.
 	 */
-	flush_sigqueue(&tsk->pending);
+	flush_task_sigqueue(tsk);
 	tsk->sighand = NULL;
 	spin_unlock(&sighand->siglock);
 
diff --git a/kernel/fork.c b/kernel/fork.c
index 02ac2a1e2101..a434cbefcf7f 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1582,6 +1582,7 @@ static __latent_entropy struct task_struct *copy_process(
 	spin_lock_init(&p->alloc_lock);
 
 	init_sigpending(&p->pending);
+	p->sigqueue_cache = NULL;
 
 	p->utime = p->stime = p->gtime = 0;
 	p->utimescaled = p->stimescaled = 0;
diff --git a/kernel/signal.c b/kernel/signal.c
index d2edbc3f2bc5..09c0f647ce17 100644
--- a/kernel/signal.c
+++ b/kernel/signal.c
@@ -14,6 +14,7 @@
 #include <linux/export.h>
 #include <linux/init.h>
 #include <linux/sched.h>
+#include <linux/sched/rt.h>
 #include <linux/fs.h>
 #include <linux/tty.h>
 #include <linux/binfmts.h>
@@ -363,13 +364,30 @@ static bool task_participate_group_stop(struct task_struct *task)
 	return false;
 }
 
+static inline struct sigqueue *get_task_cache(struct task_struct *t)
+{
+	struct sigqueue *q = t->sigqueue_cache;
+
+	if (cmpxchg(&t->sigqueue_cache, q, NULL) != q)
+		return NULL;
+	return q;
+}
+
+static inline int put_task_cache(struct task_struct *t, struct sigqueue *q)
+{
+	if (cmpxchg(&t->sigqueue_cache, NULL, q) == NULL)
+		return 0;
+	return 1;
+}
+
 /*
  * allocate a new signal queue record
  * - this may be called without locks if and only if t == current, otherwise an
  *   appropriate lock must be held to stop the target task from exiting
  */
 static struct sigqueue *
-__sigqueue_alloc(int sig, struct task_struct *t, gfp_t flags, int override_rlimit)
+__sigqueue_do_alloc(int sig, struct task_struct *t, gfp_t flags,
+		    int override_rlimit, int fromslab)
 {
 	struct sigqueue *q = NULL;
 	struct user_struct *user;
@@ -390,8 +408,13 @@ __sigqueue_alloc(int sig, struct task_struct *t, gfp_t flags, int override_rlimi
 		get_uid(user);
 	rcu_read_unlock();
 
-	if (override_rlimit || likely(sigpending <= task_rlimit(t, RLIMIT_SIGPENDING))) {
-		q = kmem_cache_alloc(sigqueue_cachep, flags);
+	if (override_rlimit ||
+	    atomic_read(&user->sigpending) <=
+			task_rlimit(t, RLIMIT_SIGPENDING)) {
+		if (!fromslab)
+			q = get_task_cache(t);
+		if (!q)
+			q = kmem_cache_alloc(sigqueue_cachep, flags);
 	} else {
 		print_dropped_signal(sig);
 	}
@@ -408,6 +431,13 @@ __sigqueue_alloc(int sig, struct task_struct *t, gfp_t flags, int override_rlimi
 	return q;
 }
 
+static struct sigqueue *
+__sigqueue_alloc(int sig, struct task_struct *t, gfp_t flags,
+		 int override_rlimit)
+{
+	return __sigqueue_do_alloc(sig, t, flags, override_rlimit, 0);
+}
+
 static void __sigqueue_free(struct sigqueue *q)
 {
 	if (q->flags & SIGQUEUE_PREALLOC)
@@ -417,6 +447,21 @@ static void __sigqueue_free(struct sigqueue *q)
 	kmem_cache_free(sigqueue_cachep, q);
 }
 
+static void sigqueue_free_current(struct sigqueue *q)
+{
+	struct user_struct *up;
+
+	if (q->flags & SIGQUEUE_PREALLOC)
+		return;
+
+	up = q->user;
+	if (rt_prio(current->normal_prio) && !put_task_cache(current, q)) {
+		atomic_dec(&up->sigpending);
+		free_uid(up);
+	} else
+		  __sigqueue_free(q);
+}
+
 void flush_sigqueue(struct sigpending *queue)
 {
 	struct sigqueue *q;
@@ -430,6 +475,21 @@ void flush_sigqueue(struct sigpending *queue)
 }
 
 /*
+ * Called from __exit_signal. Flush tsk->pending and
+ * tsk->sigqueue_cache
+ */
+void flush_task_sigqueue(struct task_struct *tsk)
+{
+	struct sigqueue *q;
+
+	flush_sigqueue(&tsk->pending);
+
+	q = get_task_cache(tsk);
+	if (q)
+		kmem_cache_free(sigqueue_cachep, q);
+}
+
+/*
  * Flush all pending signals for this kthread.
  */
 void flush_signals(struct task_struct *t)
@@ -548,7 +608,7 @@ still_pending:
 			(info->si_code == SI_TIMER) &&
 			(info->si_sys_private);
 
-		__sigqueue_free(first);
+		sigqueue_free_current(first);
 	} else {
 		/*
 		 * Ok, it wasn't in the queue.  This must be
@@ -584,6 +644,8 @@ int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
 	bool resched_timer = false;
 	int signr;
 
+	WARN_ON_ONCE(tsk != current);
+
 	/* We only dequeue private signals from ourselves, we don't let
 	 * signalfd steal them
 	 */
@@ -1555,7 +1617,8 @@ EXPORT_SYMBOL(kill_pid);
  */
 struct sigqueue *sigqueue_alloc(void)
 {
-	struct sigqueue *q = __sigqueue_alloc(-1, current, GFP_KERNEL, 0);
+	/* Preallocated sigqueue objects always from the slabcache ! */
+	struct sigqueue *q = __sigqueue_do_alloc(-1, current, GFP_KERNEL, 0, 1);
 
 	if (q)
 		q->flags |= SIGQUEUE_PREALLOC;
@@ -2395,7 +2458,7 @@ relock:
 }
 
 /**
- * signal_delivered - 
+ * signal_delivered -
  * @ksig:		kernel signal struct
  * @stepping:		nonzero if debugger single-step or block-step in use
  *
@@ -3339,7 +3402,7 @@ int __compat_save_altstack(compat_stack_t __user *uss, unsigned long sp)
  */
 SYSCALL_DEFINE1(sigpending, old_sigset_t __user *, set)
 {
-	return sys_rt_sigpending((sigset_t __user *)set, sizeof(old_sigset_t)); 
+	return sys_rt_sigpending((sigset_t __user *)set, sizeof(old_sigset_t));
 }
 
 #endif
@@ -3464,7 +3527,7 @@ COMPAT_SYSCALL_DEFINE4(rt_sigaction, int, sig,
 	ret = do_sigaction(sig, act ? &new_ka : NULL, oact ? &old_ka : NULL);
 	if (!ret && oact) {
 		sigset_to_compat(&mask, &old_ka.sa.sa_mask);
-		ret = put_user(ptr_to_compat(old_ka.sa.sa_handler), 
+		ret = put_user(ptr_to_compat(old_ka.sa.sa_handler),
 			       &oact->sa_handler);
 		ret |= copy_to_user(&oact->sa_mask, &mask, sizeof(mask));
 		ret |= put_user(old_ka.sa.sa_flags, &oact->sa_flags);
@@ -3642,7 +3705,7 @@ SYSCALL_DEFINE2(rt_sigsuspend, sigset_t __user *, unewset, size_t, sigsetsize)
 		return -EFAULT;
 	return sigsuspend(&newset);
 }
- 
+
 #ifdef CONFIG_COMPAT
 COMPAT_SYSCALL_DEFINE2(rt_sigsuspend, compat_sigset_t __user *, unewset, compat_size_t, sigsetsize)
 {