Message-ID: <Pine.LNX.4.44L0.0608311433550.5685-100000@iolanthe.rowland.org>
Date:	Thu, 31 Aug 2006 15:01:09 -0400 (EDT)
From:	Alan Stern <stern@...land.harvard.edu>
To:	Andrew Morton <akpm@...l.org>
cc:	Kernel development list <linux-kernel@...r.kernel.org>,
	<a.zummo@...ertech.it>, <adaplas@....net>, <dtor@...l.ru>,
	Greg KH <greg@...ah.com>, <jeff@...zik.org>,
	<khali@...ux-fr.org>, Stefan Richter <stefanr@...6.in-berlin.de>,
	Jonathan Corbet <corbet@....net>
Subject: [PATCH] Change return value from schedule_work()

This patch (as775b) renames:

	schedule_work()			to add_work(),
	schedule_delayed_work()		to add_delayed_work(),
	schedule_delayed_work_on()	to add_delayed_work_on().

The return value is altered, so that now 0 = success and -EBUSY = failure.

New routines with the original names are added, so that the majority
of callers don't need any changes.  The new routines call the original
functions and forget the return value.  There is no checking for
failure, because apparently several callers _expect_ the routines to
fail.  That is, they repeatedly queue the same work_struct whenever
they need it to run again, regardless of whether it is already queued.

The new routines aren't exported GPL-only because the original ones
weren't -- unlike the queue_work() routines, for some reason.

Finally, a section is added to Documentation/feature-removal-schedule.txt
to let people know that the old function names will go away next year.

Signed-off-by: Alan Stern <stern@...land.harvard.edu>

---

On Tue, 29 Aug 2006 akpm@...l.org wrote:

> The patch titled
> 
>      Change return value from schedule_work()
> 
> has been removed from the -mm tree.  Its filename is
> 
>      change-return-value-from-schedule_work.patch
> 
> This patch was dropped because it causes a warning storm - more thinking needed, I think

Fair enough.  I didn't see a problem but then I don't load all that many 
drivers.  I got to admit, Stefan was right after all.

Here's a resubmit.  This time the warnings have been left out, so the new
code behaves _exactly_ the same as the old (no more storms).  It's even
exported the same way (not GPL-only).  And I added a section to
feature-removal-schedule.txt, as Jonathan suggested.

Alan Stern


Index: mm/drivers/usb/net/usbnet.c
===================================================================
--- mm.orig/drivers/usb/net/usbnet.c
+++ mm/drivers/usb/net/usbnet.c
@@ -286,7 +286,7 @@ static void defer_bh(struct usbnet *dev,
 void usbnet_defer_kevent (struct usbnet *dev, int work)
 {
 	set_bit (work, &dev->flags);
-	if (!schedule_work (&dev->kevent))
+	if (add_work (&dev->kevent) < 0)
 		deverr (dev, "kevent %d may have been dropped", work);
 	else
 		devdbg (dev, "kevent %d scheduled", work);
Index: mm/Documentation/feature-removal-schedule.txt
===================================================================
--- mm.orig/Documentation/feature-removal-schedule.txt
+++ mm/Documentation/feature-removal-schedule.txt
@@ -285,3 +285,13 @@ Why:	The same information is available t
 
 Who:	Eric Biederman <ebiederm@...ssion.com>
 
+---------------------------
+
+What:	queue_work(), queue_delayed_work(), queue_delayed_work_on(),
+	schedule_work(), schedule_delayed_work(), schedule_delayed_work_on()
+When:	July 2007
+Why:	The return codes have been changed and the routines renamed to
+	add_work_to_q(), add_delayed_work_to_q(), add_delayed_work_to_q_on(),
+	add_work(), add_delayed_work(), and add_delayed_work_on().
+	The new function names can be used instead of the old ones.
+Who:	Alan Stern <stern@...land.harvard.edu>
Index: mm/drivers/input/serio/libps2.c
===================================================================
--- mm.orig/drivers/input/serio/libps2.c
+++ mm/drivers/input/serio/libps2.c
@@ -272,7 +272,7 @@ int ps2_schedule_command(struct ps2dev *
 	memcpy(ps2work->param, param, send);
 	INIT_WORK(&ps2work->work, ps2_execute_scheduled_command, ps2work);
 
-	if (!schedule_work(&ps2work->work)) {
+	if (add_work(&ps2work->work) < 0) {
 		kfree(ps2work);
 		return -1;
 	}
Index: mm/fs/file.c
===================================================================
--- mm.orig/fs/file.c
+++ mm/fs/file.c
@@ -85,7 +85,7 @@ static void fdtable_timer(unsigned long 
 	 */
 	if (!fddef->next)
 		goto out;
-	if (!schedule_work(&fddef->wq))
+	if (add_work(&fddef->wq) < 0)
 		mod_timer(&fddef->timer, 5);
 out:
 	spin_unlock(&fddef->lock);
@@ -147,7 +147,7 @@ static void free_fdtable_rcu(struct rcu_
 		 * If the per-cpu workqueue is running, then we
 		 * defer work scheduling through a timer.
 		 */
-		if (!schedule_work(&fddef->wq))
+		if (add_work(&fddef->wq) < 0)
 			mod_timer(&fddef->timer, 5);
 		spin_unlock(&fddef->lock);
 		put_cpu_var(fdtable_defer_list);
Index: mm/include/linux/workqueue.h
===================================================================
--- mm.orig/include/linux/workqueue.h
+++ mm/include/linux/workqueue.h
@@ -69,10 +69,12 @@ extern int add_delayed_work_to_q_on(int 
 		struct work_struct *work, unsigned long delay);
 extern void FASTCALL(flush_workqueue(struct workqueue_struct *wq));
 
-extern int FASTCALL(schedule_work(struct work_struct *work));
-extern int FASTCALL(schedule_delayed_work(struct work_struct *work, unsigned long delay));
+extern int FASTCALL(add_work(struct work_struct *work));
+extern int FASTCALL(add_delayed_work(struct work_struct *work,
+		unsigned long delay));
+extern int add_delayed_work_on(int cpu, struct work_struct *work,
+		unsigned long delay);
 
-extern int schedule_delayed_work_on(int cpu, struct work_struct *work, unsigned long delay);
 extern int schedule_on_each_cpu(void (*func)(void *info), void *info);
 extern void flush_scheduled_work(void);
 extern int current_is_keventd(void);
@@ -111,4 +113,10 @@ extern void FASTCALL(queue_delayed_work(
 extern void queue_delayed_work_on(int cpu, struct workqueue_struct *wq,
 		struct work_struct *work, unsigned long delay);
 
+extern void FASTCALL(schedule_work(struct work_struct *work));
+extern void FASTCALL(schedule_delayed_work(struct work_struct *work,
+		unsigned long delay));
+extern void schedule_delayed_work_on(int cpu, struct work_struct *work,
+		unsigned long delay);
+
 #endif
Index: mm/kernel/workqueue.c
===================================================================
--- mm.orig/kernel/workqueue.c
+++ mm/kernel/workqueue.c
@@ -449,33 +449,33 @@ EXPORT_SYMBOL_GPL(destroy_workqueue);
 static struct workqueue_struct *keventd_wq;
 
 /**
- * schedule_work - put work task in global workqueue
+ * add_work - put work task in global workqueue
  * @work: job to be done
  *
  * This puts a job in the kernel-global workqueue.
  */
-int fastcall schedule_work(struct work_struct *work)
+int fastcall add_work(struct work_struct *work)
 {
-	return !add_work_to_q(keventd_wq, work);
+	return add_work_to_q(keventd_wq, work);
 }
-EXPORT_SYMBOL(schedule_work);
+EXPORT_SYMBOL(add_work);
 
 /**
- * schedule_delayed_work - put work task in global workqueue after delay
+ * add_delayed_work - put work task in global workqueue after delay
  * @work: job to be done
  * @delay: number of jiffies to wait
  *
  * After waiting for a given time this puts a job in the kernel-global
  * workqueue.
  */
-int fastcall schedule_delayed_work(struct work_struct *work, unsigned long delay)
+int fastcall add_delayed_work(struct work_struct *work, unsigned long delay)
 {
-	return !add_delayed_work_to_q(keventd_wq, work, delay);
+	return add_delayed_work_to_q(keventd_wq, work, delay);
 }
-EXPORT_SYMBOL(schedule_delayed_work);
+EXPORT_SYMBOL(add_delayed_work);
 
 /**
- * schedule_delayed_work_on - queue work in global workqueue on CPU after delay
+ * add_delayed_work_on - queue work in global workqueue on CPU after delay
  * @cpu: cpu to use
  * @work: job to be done
  * @delay: number of jiffies to wait
@@ -483,12 +483,12 @@ EXPORT_SYMBOL(schedule_delayed_work);
  * After waiting for a given time this puts a job in the kernel-global
  * workqueue on the specified CPU.
  */
-int schedule_delayed_work_on(int cpu,
+int add_delayed_work_on(int cpu,
 			struct work_struct *work, unsigned long delay)
 {
-	return !add_delayed_work_to_q_on(cpu, keventd_wq, work, delay);
+	return add_delayed_work_to_q_on(cpu, keventd_wq, work, delay);
 }
-EXPORT_SYMBOL(schedule_delayed_work_on);
+EXPORT_SYMBOL(add_delayed_work_on);
 
 /*
  * Legacy API for use when the return codes aren't needed.
@@ -523,6 +523,26 @@ void queue_delayed_work_on(int cpu, stru
 }
 EXPORT_SYMBOL_GPL(queue_delayed_work_on);
 
+void fastcall schedule_work(struct work_struct *work)
+{
+	(void) add_work_to_q(keventd_wq, work);
+}
+EXPORT_SYMBOL(schedule_work);
+
+void fastcall schedule_delayed_work(struct work_struct *work,
+		unsigned long delay)
+{
+	(void) add_delayed_work_to_q(keventd_wq, work, delay);
+}
+EXPORT_SYMBOL(schedule_delayed_work);
+
+void schedule_delayed_work_on(int cpu, struct work_struct *work,
+		unsigned long delay)
+{
+	(void) add_delayed_work_to_q_on(cpu, keventd_wq, work, delay);
+}
+EXPORT_SYMBOL(schedule_delayed_work_on);
+
 /**
  * schedule_on_each_cpu - call a function on each online CPU from keventd
  * @func: the function to call
Index: mm/drivers/net/smc91x.c
===================================================================
--- mm.orig/drivers/net/smc91x.c
+++ mm/drivers/net/smc91x.c
@@ -1436,9 +1436,8 @@ static void smc_timeout(struct net_devic
 	 * which calls schedule().  Hence we use a work queue.
 	 */
 	if (lp->phy_type != 0) {
-		if (schedule_work(&lp->phy_configure)) {
+		if (add_work(&lp->phy_configure) == 0)
 			lp->work_pending = 1;
-		}
 	}
 
 	/* We can accept TX packets again */
Index: mm/drivers/char/genrtc.c
===================================================================
--- mm.orig/drivers/char/genrtc.c
+++ mm/drivers/char/genrtc.c
@@ -122,7 +122,7 @@ static void genrtc_troutine(void *data)
 		add_timer(&timer_task);
 
 		gen_rtc_interrupt(0);
-	} else if (schedule_work(&genrtc_task) == 0)
+	} else if (add_work(&genrtc_task) < 0)
 		stask_active = 0;
 }
 
@@ -136,7 +136,7 @@ static void gen_rtc_timer(unsigned long 
 		       jiffies-tt_exp);
 	ttask_active=0;
 	stask_active=1;
-	if ((schedule_work(&genrtc_task) == 0))
+	if (add_work(&genrtc_task) < 0)
 		stask_active = 0;
 }
 
@@ -260,9 +260,8 @@ static inline int gen_set_rtc_irq_bit(un
 		init_timer(&timer_task);
 
 		stask_active = 1;
-		if (schedule_work(&genrtc_task) == 0){
+		if (add_work(&genrtc_task) < 0)
 			stask_active = 0;
-		}
 	}
 	spin_unlock(&gen_rtc_lock);
 	gen_rtc_irq_data = 0;
Index: mm/drivers/usb/gadget/ether.c
===================================================================
--- mm.orig/drivers/usb/gadget/ether.c
+++ mm/drivers/usb/gadget/ether.c
@@ -1604,7 +1604,7 @@ static void defer_kevent (struct eth_dev
 {
 	if (test_and_set_bit (flag, &dev->todo))
 		return;
-	if (!schedule_work (&dev->work))
+	if (add_work (&dev->work) < 0)
 		ERROR (dev, "kevent %d may have been dropped\n", flag);
 	else
 		DEBUG (dev, "kevent %d scheduled\n", flag);
Index: mm/drivers/rtc/rtc-dev.c
===================================================================
--- mm.orig/drivers/rtc/rtc-dev.c
+++ mm/drivers/rtc/rtc-dev.c
@@ -71,7 +71,7 @@ static void rtc_uie_task(void *data)
 		rtc->uie_timer_active = 1;
 		rtc->uie_task_active = 0;
 		add_timer(&rtc->uie_timer);
-	} else if (schedule_work(&rtc->uie_task) == 0) {
+	} else if (add_work(&rtc->uie_task) < 0) {
 		rtc->uie_task_active = 0;
 	}
 	spin_unlock_irq(&rtc->irq_lock);
@@ -87,7 +87,7 @@ static void rtc_uie_timer(unsigned long 
 	spin_lock_irqsave(&rtc->irq_lock, flags);
 	rtc->uie_timer_active = 0;
 	rtc->uie_task_active = 1;
-	if ((schedule_work(&rtc->uie_task) == 0))
+	if (add_work(&rtc->uie_task) < 0)
 		rtc->uie_task_active = 0;
 	spin_unlock_irqrestore(&rtc->irq_lock, flags);
 }
@@ -127,7 +127,7 @@ static int set_uie(struct rtc_device *rt
 		rtc->stop_uie_polling = 0;
 		rtc->oldsecs = tm.tm_sec;
 		rtc->uie_task_active = 1;
-		if (schedule_work(&rtc->uie_task) == 0)
+		if (add_work(&rtc->uie_task) < 0)
 			rtc->uie_task_active = 0;
 	}
 	rtc->irq_data = 0;
Index: mm/drivers/usb/input/hid-core.c
===================================================================
--- mm.orig/drivers/usb/input/hid-core.c
+++ mm/drivers/usb/input/hid-core.c
@@ -988,7 +988,7 @@ static void hid_io_error(struct hid_devi
 
 		/* Retries failed, so do a port reset */
 		if (!test_and_set_bit(HID_RESET_PENDING, &hid->iofl)) {
-			if (schedule_work(&hid->reset_work))
+			if (add_work(&hid->reset_work) == 0)
 				goto done;
 			clear_bit(HID_RESET_PENDING, &hid->iofl);
 		}
Index: mm/drivers/i2c/chips/isp1301_omap.c
===================================================================
--- mm.orig/drivers/i2c/chips/isp1301_omap.c
+++ mm/drivers/i2c/chips/isp1301_omap.c
@@ -357,8 +357,8 @@ isp1301_defer_work(struct isp1301 *isp, 
 
 	if (isp && !test_and_set_bit(work, &isp->todo)) {
 		(void) get_device(&isp->client.dev);
-		status = schedule_work(&isp->work);
-		if (!status && !isp->working)
+		status = add_work(&isp->work);
+		if (status < 0 && !isp->working)
 			dev_vdbg(&isp->client.dev,
 				"work item %d may be lost\n", work);
 	}
Index: mm/drivers/net/smc911x.c
===================================================================
--- mm.orig/drivers/net/smc911x.c
+++ mm/drivers/net/smc911x.c
@@ -1362,9 +1362,8 @@ static void smc911x_timeout(struct net_d
 	 * which calls schedule().	 Hence we use a work queue.
 	 */
 	if (lp->phy_type != 0) {
-		if (schedule_work(&lp->phy_configure)) {
+		if (add_work(&lp->phy_configure) == 0)
 			lp->work_pending = 1;
-		}
 	}
 
 	/* We can accept TX packets again */

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists