[PATCH] make ITIMER_REAL per-process

POSIX requires that setitimer, getitimer, and alarm work on a per-process
basis.  Currently, Linux implements these for individual threads.  This patch
fixes these semantics for the ITIMER_REAL timer (which generates SIGALRM),
making it shared by all threads in a process (thread group).

Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

3 files changed, 80 insertions, 49 deletions

diff --git a/kernel/exit.c b/kernel/exit.c
index fbe293b10a47..db204cd02d8b 100644
--- a/kernel/exit.c
+++ b/kernel/exit.c
@@ -795,7 +795,6 @@ fastcall NORET_TYPE void do_exit(long code)
 	}
 
 	tsk->flags |= PF_EXITING;
-	del_timer_sync(&tsk->real_timer);
 
 	/*
 	 * Make sure we don't try to process any timer firings
@@ -813,8 +812,10 @@ fastcall NORET_TYPE void do_exit(long code)
 	acct_update_integrals(tsk);
 	update_mem_hiwater(tsk);
 	group_dead = atomic_dec_and_test(&tsk->signal->live);
-	if (group_dead)
+	if (group_dead) {
+ 		del_timer_sync(&tsk->signal->real_timer);
 		acct_process(code);
+	}
 	exit_mm(tsk);
 
 	exit_sem(tsk);
diff --git a/kernel/fork.c b/kernel/fork.c
index 718eaf0bb1cd..a1d1939f596c 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -740,6 +740,11 @@ static inline int copy_signal(unsigned long clone_flags, struct task_struct * ts
 	init_sigpending(&sig->shared_pending);
 	INIT_LIST_HEAD(&sig->posix_timers);
 
+	sig->it_real_value = sig->it_real_incr = 0;
+	sig->real_timer.function = it_real_fn;
+	sig->real_timer.data = (unsigned long) tsk;
+	init_timer(&sig->real_timer);
+
 	sig->tty = current->signal->tty;
 	sig->pgrp = process_group(current);
 	sig->session = current->signal->session;
@@ -870,14 +875,10 @@ static task_t *copy_process(unsigned long clone_flags,
 	clear_tsk_thread_flag(p, TIF_SIGPENDING);
 	init_sigpending(&p->pending);
 
-	p->it_real_value = 0;
-	p->it_real_incr = 0;
 	p->it_virt_value = cputime_zero;
 	p->it_virt_incr = cputime_zero;
 	p->it_prof_value = cputime_zero;
 	p->it_prof_incr = cputime_zero;
-	init_timer(&p->real_timer);
-	p->real_timer.data = (unsigned long) p;
 
 	p->utime = cputime_zero;
 	p->stime = cputime_zero;
diff --git a/kernel/itimer.c b/kernel/itimer.c
index e1743c563206..c9cc2a4cb40a 100644
--- a/kernel/itimer.c
+++ b/kernel/itimer.c
@@ -14,25 +14,31 @@
 
 #include <asm/uaccess.h>
 
+static unsigned long it_real_value(struct signal_struct *sig)
+{
+	unsigned long val = 0;
+	if (timer_pending(&sig->real_timer)) {
+		val = sig->real_timer.expires - jiffies;
+
+		/* look out for negative/zero itimer.. */
+		if ((long) val <= 0)
+			val = 1;
+	}
+	return val;
+}
+
 int do_getitimer(int which, struct itimerval *value)
 {
-	register unsigned long val;
+	unsigned long interval, val;
 
 	switch (which) {
 	case ITIMER_REAL:
-		val = 0;
-		/* 
-		 * FIXME! This needs to be atomic, in case the kernel timer happens!
-		 */
-		if (timer_pending(&current->real_timer)) {
-			val = current->real_timer.expires - jiffies;
-
-			/* look out for negative/zero itimer.. */
-			if ((long) val <= 0)
-				val = 1;
-		}
+		spin_lock_irq(&current->sighand->siglock);
+		interval = current->signal->it_real_incr;
+		val = it_real_value(current->signal);
+		spin_unlock_irq(&current->sighand->siglock);
 		jiffies_to_timeval(val, &value->it_value);
-		jiffies_to_timeval(current->it_real_incr, &value->it_interval);
+		jiffies_to_timeval(interval, &value->it_interval);
 		break;
 	case ITIMER_VIRTUAL:
 		cputime_to_timeval(current->it_virt_value, &value->it_value);
@@ -48,7 +54,6 @@ int do_getitimer(int which, struct itimerval *value)
 	return 0;
 }
 
-/* SMP: Only we modify our itimer values. */
 asmlinkage long sys_getitimer(int which, struct itimerval __user *value)
 {
 	int error = -EFAULT;
@@ -63,60 +68,87 @@ asmlinkage long sys_getitimer(int which, struct itimerval __user *value)
 	return error;
 }
 
+/*
+ * Called with P->sighand->siglock held and P->signal->real_timer inactive.
+ * If interval is nonzero, arm the timer for interval ticks from now.
+ */
+static inline void it_real_arm(struct task_struct *p, unsigned long interval)
+{
+	p->signal->it_real_value = interval; /* XXX unnecessary field?? */
+	if (interval == 0)
+		return;
+	if (interval > (unsigned long) LONG_MAX)
+		interval = LONG_MAX;
+	p->signal->real_timer.expires = jiffies + interval;
+	add_timer(&p->signal->real_timer);
+}
+
 void it_real_fn(unsigned long __data)
 {
 	struct task_struct * p = (struct task_struct *) __data;
-	unsigned long interval;
 
 	send_group_sig_info(SIGALRM, SEND_SIG_PRIV, p);
-	interval = p->it_real_incr;
-	if (interval) {
-		if (interval > (unsigned long) LONG_MAX)
-			interval = LONG_MAX;
-		p->real_timer.expires = jiffies + interval;
-		add_timer(&p->real_timer);
-	}
+
+	/*
+	 * Now restart the timer if necessary.  We don't need any locking
+	 * here because do_setitimer makes sure we have finished running
+	 * before it touches anything.
+	 */
+	it_real_arm(p, p->signal->it_real_incr);
 }
 
 int do_setitimer(int which, struct itimerval *value, struct itimerval *ovalue)
 {
-	unsigned long expire;
+	struct task_struct *tsk = current;
+ 	unsigned long val, interval;
 	cputime_t cputime;
-	int k;
 
-	if (ovalue && (k = do_getitimer(which, ovalue)) < 0)
-		return k;
 	switch (which) {
 		case ITIMER_REAL:
-			del_timer_sync(&current->real_timer);
-			expire = timeval_to_jiffies(&value->it_value);
-			current->it_real_value = expire;
-			current->it_real_incr =
+ 			spin_lock_irq(&tsk->sighand->siglock);
+ 			interval = tsk->signal->it_real_incr;
+ 			val = it_real_value(tsk->signal);
+ 			if (val)
+ 				del_timer_sync(&tsk->signal->real_timer);
+ 			tsk->signal->it_real_incr =
 				timeval_to_jiffies(&value->it_interval);
-			if (!expire)
-				break;
-			if (expire > (unsigned long) LONG_MAX)
-				expire = LONG_MAX;
-			current->real_timer.expires = jiffies + expire;
-			add_timer(&current->real_timer);
+ 			it_real_arm(tsk, timeval_to_jiffies(&value->it_value));
+ 			spin_unlock_irq(&tsk->sighand->siglock);
+			if (ovalue) {
+				jiffies_to_timeval(val, &ovalue->it_value);
+				jiffies_to_timeval(interval,
+						   &ovalue->it_interval);
+			}
 			break;
 		case ITIMER_VIRTUAL:
+			if (ovalue) {
+				cputime_to_timeval(tsk->it_virt_value,
+						   &ovalue->it_value);
+				cputime_to_timeval(tsk->it_virt_incr,
+						   &ovalue->it_interval);
+			}
 			cputime = timeval_to_cputime(&value->it_value);
 			if (cputime_gt(cputime, cputime_zero))
 				cputime = cputime_add(cputime,
 						      jiffies_to_cputime(1));
-			current->it_virt_value = cputime;
+			tsk->it_virt_value = cputime;
 			cputime = timeval_to_cputime(&value->it_interval);
-			current->it_virt_incr = cputime;
+			tsk->it_virt_incr = cputime;
 			break;
 		case ITIMER_PROF:
+			if (ovalue) {
+				cputime_to_timeval(tsk->it_prof_value,
+						   &ovalue->it_value);
+				cputime_to_timeval(tsk->it_prof_incr,
+						   &ovalue->it_interval);
+			}
 			cputime = timeval_to_cputime(&value->it_value);
 			if (cputime_gt(cputime, cputime_zero))
 				cputime = cputime_add(cputime,
 						      jiffies_to_cputime(1));
-			current->it_prof_value = cputime;
+			tsk->it_prof_value = cputime;
 			cputime = timeval_to_cputime(&value->it_interval);
-			current->it_prof_incr = cputime;
+			tsk->it_prof_incr = cputime;
 			break;
 		default:
 			return -EINVAL;
@@ -124,9 +156,6 @@ int do_setitimer(int which, struct itimerval *value, struct itimerval *ovalue)
 	return 0;
 }
 
-/* SMP: Again, only we play with our itimers, and signals are SMP safe
- *      now so that is not an issue at all anymore.
- */
 asmlinkage long sys_setitimer(int which,
 			      struct itimerval __user *value,
 			      struct itimerval __user *ovalue)