Skip to contentSkip to conversation

Convert message queue timeouts to hrtimer

PATCH In review Carsten Emde 1 participant 5 messages Apr 2, 2010 – Apr 6, 2010 High-resolution timers, timer wheel, clockevents

lore
Share & tools
lore

Carsten Emde

author
#0 lore raw

cover Convert message queue timeouts to hrtimer

The message queue system calls that may timeout, mq_timedsend() and mq_timedreceive(), still use the old jiffie-based timer interface. Convert them to hrtimer.

The first patch provides the hrtimer infrastructure, the second patch does the actual conversion to hrtimer.

Carsten.

author
#1 lore raw

PATCH 1/2 provide-schedule_hrtimeout-for-wallclock.patch

The current version of schedule_hrtimeout() always uses the monotonic clock. Some system calls such as mq_timedsend() and mq_timedreceive(), however, require the use of the wall clock due to the definition of the system call.

This patch provides the infrastructure to use schedule_hrtimeout() with a wallclock timer.

CC Andrew Morton <[email protected]>
CC Arjan van de Veen <[email protected]>
CC Thomas Gleixner <[email protected]>
Tested-by Pradyumna Sampath <[email protected]>
Signed-off-by Carsten Emde <[email protected]>
---
 include/linux/hrtimer.h |    2 +
 kernel/hrtimer.c        |   66 +++++++++++++++++++++++++++++-------------------
 2 files changed, 42 insertions(+), 26 deletions(-)
include/linux/hrtimer.h +2 -0
Index: head/include/linux/hrtimer.h
===================================================================
@@ -422,6 +422,8 @@ extern void hrtimer_init_sleeper(struct 
 
 extern int schedule_hrtimeout_range(ktime_t *expires, unsigned long delta,
 						const enum hrtimer_mode mode);
+extern int schedule_hrtimeout_range_clock(ktime_t *expires,
+		unsigned long delta, const enum hrtimer_mode mode, int clock);
 extern int schedule_hrtimeout(ktime_t *expires, const enum hrtimer_mode mode);
 
 /* Soft interrupt function to run the hrtimer queues: */
kernel/hrtimer.c +40 -26

88 lines not on this page · raw message

author
#2 lore raw

[tip:timers/core] hrtimers: Provide schedule_hrtimeout for CLOCK_REALTIME

Commit-ID:  351b3f7a21e413a9b14d0393171497d2373bd702
Gitweb:     http://git.kernel.org/tip/351b3f7a21e413a9b14d0393171497d2373bd702
Author:     Carsten Emde <[email protected]>
AuthorDate: Fri, 2 Apr 2010 22:40:19 +0200
Committer:  Thomas Gleixner <[email protected]>
CommitDate: Tue, 6 Apr 2010 21:50:03 +0200

hrtimers: Provide schedule_hrtimeout for CLOCK_REALTIME

The current version of schedule_hrtimeout() always uses the monotonic clock. Some system calls such as mq_timedsend() and mq_timedreceive(), however, require the use of the wall clock due to the definition of the system call.

This patch provides the infrastructure to use schedule_hrtimeout() with a CLOCK_REALTIME timer.

Signed-off-by Carsten Emde <[email protected]>
Tested-by Pradyumna Sampath <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Arjan van de Veen <[email protected]>

LKML-Reference: <[email protected]>

Signed-off-by Thomas Gleixner <[email protected]>
---
 include/linux/hrtimer.h |    2 +
 kernel/hrtimer.c        |   67 ++++++++++++++++++++++++++++------------------
 2 files changed, 43 insertions(+), 26 deletions(-)
include/linux/hrtimer.h +2 -0
index 5d86fb2..fd0c1b8 100644
@@ -422,6 +422,8 @@ extern void hrtimer_init_sleeper(struct hrtimer_sleeper *sl,
 
 extern int schedule_hrtimeout_range(ktime_t *expires, unsigned long delta,
 						const enum hrtimer_mode mode);
+extern int schedule_hrtimeout_range_clock(ktime_t *expires,
+		unsigned long delta, const enum hrtimer_mode mode, int clock);
 extern int schedule_hrtimeout(ktime_t *expires, const enum hrtimer_mode mode);
 
 /* Soft interrupt function to run the hrtimer queues: */
kernel/hrtimer.c +41 -26

89 lines not on this page · raw message

author
#3 lore raw

PATCH 2/2 let-message-queue-timeout-use-hrtimer.patch

In reply to Carsten Emde

The message queue functions mq_timedsend() and mq_timedreceive() have not yet been converted to use the hrtimer interface.

This patch replaces the call to schedule_timeout() by a call to schedule_hrtimeout() and transforms the expiration time from timespec to ktime as required.

CC Andrew Morton <[email protected]>
CC Arjan van de Veen <[email protected]>
CC Thomas Gleixner <[email protected]>
Tested-by Pradyumna Sampath <[email protected]>
Signed-off-by Carsten Emde <[email protected]>
---
 ipc/mqueue.c |   74 +++++++++++++++++++----------------------------------------
 1 file changed, 25 insertions(+), 49 deletions(-)
ipc/mqueue.c +25 -49
Index: head/ipc/mqueue.c
===================================================================
@@ -428,7 +428,7 @@ static void wq_add(struct mqueue_inode_i
  * sr: SEND or RECV
  */
 static int wq_sleep(struct mqueue_inode_info *info, int sr,
-			long timeout, struct ext_wait_queue *ewp)
+		    ktime_t *timeout, struct ext_wait_queue *ewp)
 {
 	int retval;
 	signed long time;

@@ -439,7 +439,8 @@ static int wq_sleep(struct mqueue_inode_
 		set_current_state(TASK_INTERRUPTIBLE);
 
 		spin_unlock(&info->lock);
-		time = schedule_timeout(timeout);
+		time = schedule_hrtimeout_range_clock(timeout,
+		    HRTIMER_MODE_ABS, 0, CLOCK_REALTIME);
 
 		while (ewp->state == STATE_PENDING)
 			cpu_relax();

@@ -551,31 +552,16 @@ static void __do_notify(struct mqueue_in
 	wake_up(&info->wait_q);
 }
 
-static long prepare_timeout(struct timespec *p)
+static int prepare_timeout(const struct timespec __user *u_abs_timeout,
+			   ktime_t *expires, struct timespec *ts)
 {
-	struct timespec nowts;
-	long timeout;
-
-	if (p) {
-		if (unlikely(p->tv_nsec < 0 || p->tv_sec < 0
-			|| p->tv_nsec >= NSEC_PER_SEC))
-			return -EINVAL;
-		nowts = CURRENT_TIME;
-		/* first subtract as jiffies can't be too big */
-		p->tv_sec -= nowts.tv_sec;
-		if (p->tv_nsec < nowts.tv_nsec) {
-			p->tv_nsec += NSEC_PER_SEC;
-			p->tv_sec--;
-		}
-		p->tv_nsec -= nowts.tv_nsec;
-		if (p->tv_sec < 0)
-			return 0;
-
-		timeout = timespec_to_jiffies(p) + 1;
-	} else
-		return MAX_SCHEDULE_TIMEOUT;
+	if (copy_from_user(ts, u_abs_timeout, sizeof(struct timespec)))
+		return -EFAULT;
+	if (!timespec_valid(ts))
+		return -EINVAL;
 
-	return timeout;
+	*expires = timespec_to_ktime(*ts);
+	return 0;
 }
 
 static void remove_notification(struct mqueue_inode_info *info)

@@ -861,22 +847,21 @@ SYSCALL_DEFINE5(mq_timedsend, mqd_t, mqd
 	struct ext_wait_queue *receiver;
 	struct msg_msg *msg_ptr;
 	struct mqueue_inode_info *info;
-	struct timespec ts, *p = NULL;
-	long timeout;
+ 	ktime_t expires, *timeout = NULL;
+	struct timespec ts;
 	int ret;
 
 	if (u_abs_timeout) {
-		if (copy_from_user(&ts, u_abs_timeout, 
-					sizeof(struct timespec)))
-			return -EFAULT;
-		p = &ts;
+		int res = prepare_timeout(u_abs_timeout, &expires, &ts);
+		if (res)
+			return res;
+		timeout = &expires;
 	}
 
 	if (unlikely(msg_prio >= (unsigned long) MQ_PRIO_MAX))
 		return -EINVAL;
 
-	audit_mq_sendrecv(mqdes, msg_len, msg_prio, p);
-	timeout = prepare_timeout(p);
+	audit_mq_sendrecv(mqdes, msg_len, msg_prio, timeout ? &ts : NULL);
 
 	filp = fget(mqdes);
 	if (unlikely(!filp)) {

@@ -918,9 +903,6 @@ SYSCALL_DEFINE5(mq_timedsend, mqd_t, mqd
 		if (filp->f_flags & O_NONBLOCK) {
 			spin_unlock(&info->lock);
 			ret = -EAGAIN;
-		} else if (unlikely(timeout < 0)) {
-			spin_unlock(&info->lock);
-			ret = timeout;
 		} else {
 			wait.task = current;
 			wait.msg = (void *) msg_ptr;

@@ -953,24 +935,23 @@ SYSCALL_DEFINE5(mq_timedreceive, mqd_t, 
 		size_t, msg_len, unsigned int __user *, u_msg_prio,
 		const struct timespec __user *, u_abs_timeout)
 {
-	long timeout;
 	ssize_t ret;
 	struct msg_msg *msg_ptr;
 	struct file *filp;
 	struct inode *inode;
 	struct mqueue_inode_info *info;
 	struct ext_wait_queue wait;
-	struct timespec ts, *p = NULL;
+   	ktime_t expires, *timeout = NULL;
+	struct timespec ts;
 
 	if (u_abs_timeout) {
-		if (copy_from_user(&ts, u_abs_timeout, 
-					sizeof(struct timespec)))
-			return -EFAULT;
-		p = &ts;
+		int res = prepare_timeout(u_abs_timeout, &expires, &ts);
+		if (res)
+			return res;
+		timeout = &expires;
 	}
 
-	audit_mq_sendrecv(mqdes, msg_len, 0, p);
-	timeout = prepare_timeout(p);
+	audit_mq_sendrecv(mqdes, msg_len, 0, timeout ? &ts : NULL);
 
 	filp = fget(mqdes);
 	if (unlikely(!filp)) {

@@ -1002,11 +983,6 @@ SYSCALL_DEFINE5(mq_timedreceive, mqd_t, 
 		if (filp->f_flags & O_NONBLOCK) {
 			spin_unlock(&info->lock);
 			ret = -EAGAIN;
-			msg_ptr = NULL;
-		} else if (unlikely(timeout < 0)) {
-			spin_unlock(&info->lock);
-			ret = timeout;
-			msg_ptr = NULL;
 		} else {
 			wait.task = current;
 			wait.state = STATE_NONE;
author
#4 lore raw

[tip:timers/core] mqueue: Convert message queue timeout to use hrtimers

Commit-ID:  9ca7d8e6834c40a99622bbe4a88aaf64313ae43c
Gitweb:     http://git.kernel.org/tip/9ca7d8e6834c40a99622bbe4a88aaf64313ae43c
Author:     Carsten Emde <[email protected]>
AuthorDate: Fri, 2 Apr 2010 22:40:20 +0200
Committer:  Thomas Gleixner <[email protected]>
CommitDate: Tue, 6 Apr 2010 21:50:03 +0200

mqueue: Convert message queue timeout to use hrtimers

The message queue functions mq_timedsend() and mq_timedreceive() have not yet been converted to use the hrtimer interface.

This patch replaces the call to schedule_timeout() by a call to schedule_hrtimeout() and transforms the expiration time from timespec to ktime as required.

[ tglx: Fixed whitespace wreckage ]

Signed-off-by Carsten Emde <[email protected]>
Tested-by Pradyumna Sampath <[email protected]>
Cc Arjan van de Veen <[email protected]>
Cc Andrew Morton <[email protected]>

LKML-Reference: <[email protected]>

Signed-off-by Thomas Gleixner <[email protected]>
---
 ipc/mqueue.c |   74 +++++++++++++++++++--------------------------------------
 1 files changed, 25 insertions(+), 49 deletions(-)
ipc/mqueue.c +25 -49
index e4e3f04..a9d8b0c 100644
@@ -428,7 +428,7 @@ static void wq_add(struct mqueue_inode_info *info, int sr,
  * sr: SEND or RECV
  */
 static int wq_sleep(struct mqueue_inode_info *info, int sr,
-			long timeout, struct ext_wait_queue *ewp)
+		    ktime_t *timeout, struct ext_wait_queue *ewp)
 {
 	int retval;
 	signed long time;

@@ -439,7 +439,8 @@ static int wq_sleep(struct mqueue_inode_info *info, int sr,
 		set_current_state(TASK_INTERRUPTIBLE);
 
 		spin_unlock(&info->lock);
-		time = schedule_timeout(timeout);
+		time = schedule_hrtimeout_range_clock(timeout,
+		    HRTIMER_MODE_ABS, 0, CLOCK_REALTIME);
 
 		while (ewp->state == STATE_PENDING)
 			cpu_relax();

@@ -551,31 +552,16 @@ static void __do_notify(struct mqueue_inode_info *info)
 	wake_up(&info->wait_q);
 }
 
-static long prepare_timeout(struct timespec *p)
+static int prepare_timeout(const struct timespec __user *u_abs_timeout,
+			   ktime_t *expires, struct timespec *ts)
 {
-	struct timespec nowts;
-	long timeout;
-
-	if (p) {
-		if (unlikely(p->tv_nsec < 0 || p->tv_sec < 0
-			|| p->tv_nsec >= NSEC_PER_SEC))
-			return -EINVAL;
-		nowts = CURRENT_TIME;
-		/* first subtract as jiffies can't be too big */
-		p->tv_sec -= nowts.tv_sec;
-		if (p->tv_nsec < nowts.tv_nsec) {
-			p->tv_nsec += NSEC_PER_SEC;
-			p->tv_sec--;
-		}
-		p->tv_nsec -= nowts.tv_nsec;
-		if (p->tv_sec < 0)
-			return 0;
-
-		timeout = timespec_to_jiffies(p) + 1;
-	} else
-		return MAX_SCHEDULE_TIMEOUT;
+	if (copy_from_user(ts, u_abs_timeout, sizeof(struct timespec)))
+		return -EFAULT;
+	if (!timespec_valid(ts))
+		return -EINVAL;
 
-	return timeout;
+	*expires = timespec_to_ktime(*ts);
+	return 0;
 }
 
 static void remove_notification(struct mqueue_inode_info *info)

@@ -861,22 +847,21 @@ SYSCALL_DEFINE5(mq_timedsend, mqd_t, mqdes, const char __user *, u_msg_ptr,
 	struct ext_wait_queue *receiver;
 	struct msg_msg *msg_ptr;
 	struct mqueue_inode_info *info;
-	struct timespec ts, *p = NULL;
-	long timeout;
+	ktime_t expires, *timeout = NULL;
+	struct timespec ts;
 	int ret;
 
 	if (u_abs_timeout) {
-		if (copy_from_user(&ts, u_abs_timeout, 
-					sizeof(struct timespec)))
-			return -EFAULT;
-		p = &ts;
+		int res = prepare_timeout(u_abs_timeout, &expires, &ts);
+		if (res)
+			return res;
+		timeout = &expires;
 	}
 
 	if (unlikely(msg_prio >= (unsigned long) MQ_PRIO_MAX))
 		return -EINVAL;
 
-	audit_mq_sendrecv(mqdes, msg_len, msg_prio, p);
-	timeout = prepare_timeout(p);
+	audit_mq_sendrecv(mqdes, msg_len, msg_prio, timeout ? &ts : NULL);
 
 	filp = fget(mqdes);
 	if (unlikely(!filp)) {

@@ -918,9 +903,6 @@ SYSCALL_DEFINE5(mq_timedsend, mqd_t, mqdes, const char __user *, u_msg_ptr,
 		if (filp->f_flags & O_NONBLOCK) {
 			spin_unlock(&info->lock);
 			ret = -EAGAIN;
-		} else if (unlikely(timeout < 0)) {
-			spin_unlock(&info->lock);
-			ret = timeout;
 		} else {
 			wait.task = current;
 			wait.msg = (void *) msg_ptr;

@@ -953,24 +935,23 @@ SYSCALL_DEFINE5(mq_timedreceive, mqd_t, mqdes, char __user *, u_msg_ptr,
 		size_t, msg_len, unsigned int __user *, u_msg_prio,
 		const struct timespec __user *, u_abs_timeout)
 {
-	long timeout;
 	ssize_t ret;
 	struct msg_msg *msg_ptr;
 	struct file *filp;
 	struct inode *inode;
 	struct mqueue_inode_info *info;
 	struct ext_wait_queue wait;
-	struct timespec ts, *p = NULL;
+	ktime_t expires, *timeout = NULL;
+	struct timespec ts;
 
 	if (u_abs_timeout) {
-		if (copy_from_user(&ts, u_abs_timeout, 
-					sizeof(struct timespec)))
-			return -EFAULT;
-		p = &ts;
+		int res = prepare_timeout(u_abs_timeout, &expires, &ts);
+		if (res)
+			return res;
+		timeout = &expires;
 	}
 
-	audit_mq_sendrecv(mqdes, msg_len, 0, p);
-	timeout = prepare_timeout(p);
+	audit_mq_sendrecv(mqdes, msg_len, 0, timeout ? &ts : NULL);
 
 	filp = fget(mqdes);
 	if (unlikely(!filp)) {

@@ -1002,11 +983,6 @@ SYSCALL_DEFINE5(mq_timedreceive, mqd_t, mqdes, char __user *, u_msg_ptr,
 		if (filp->f_flags & O_NONBLOCK) {
 			spin_unlock(&info->lock);
 			ret = -EAGAIN;
-			msg_ptr = NULL;
-		} else if (unlikely(timeout < 0)) {
-			spin_unlock(&info->lock);
-			ret = timeout;
-			msg_ptr = NULL;
 		} else {
 			wait.task = current;
 			wait.state = STATE_NONE;