Re: [PATCH v3 3/3] Improve container_notify_cb() to support container hot-remove.

[Rafael J. Wysocki]

On Thursday, November 01, 2012 04:39:50 PM Yinghai Lu wrote:
> On Thu, Nov 1, 2012 at 4:15 PM, Rafael J. Wysocki <rjw@xxxxxxx> wrote:
> > From: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
> > Subject: ACPI: Make seemingly useless check in osl.c more understandable
> >
> > There is a seemingly useless check in drivers/acpi/osl.c added by
> > commit bc73675 (ACPI: fixes a false alarm from lockdep), which really
> > is necessary to avoid false positive lockdep complaints.  Document
> > this and rearrange the code related to it so that it makes fewer
> > checks.
> >
> > Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
> > ---
> >  drivers/acpi/osl.c |   21 ++++++++++++++-------
> >  1 file changed, 14 insertions(+), 7 deletions(-)
> >
> > Index: linux/drivers/acpi/osl.c
> > ===================================================================
> > --- linux.orig/drivers/acpi/osl.c
> > +++ linux/drivers/acpi/osl.c
> > @@ -944,17 +944,24 @@ static acpi_status __acpi_os_execute(acp
> >          * because the hotplug code may call driver .remove() functions,
> >          * which invoke flush_scheduled_work/acpi_os_wait_events_complete
> >          * to flush these workqueues.
> > +        *
> > +        * To prevent lockdep from complaining unnecessarily, make sure that
> > +        * there is a different static lockdep key for each workqueue by using
> > +        * INIT_WORK() for each of them separately.
> >          */
> > -       queue = hp ? kacpi_hotplug_wq :
> > -               (type == OSL_NOTIFY_HANDLER ? kacpi_notify_wq : kacpid_wq);
> > -       dpc->wait = hp ? 1 : 0;
> > -
> > -       if (queue == kacpi_hotplug_wq)
> > +       if (hp) {
> > +               queue = kacpi_hotplug_wq;
> > +               dpc->wait = 1;
> >                 INIT_WORK(&dpc->work, acpi_os_execute_deferred);
> > -       else if (queue == kacpi_notify_wq)
> > +       } else if (type == OSL_NOTIFY_HANDLER) {
> > +               queue = kacpi_notify_wq;
> > +               dpc->wait = 0;
> 
> yes, much clear.
> 
> at the same can you changne
> dpc allocation from kmalloc with kzalloc instead.
> 
> then we save two lines for dpc->wait = 0

Good idea. :-)

> After that
> 
> Acked-by: Yinghai Lu <yinghai@xxxxxxxxxx>

For completeness:

---
From: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
Subject: ACPI: Make seemingly useless check in osl.c more understandable

There is a seemingly useless check in drivers/acpi/osl.c added by
commit bc73675 (ACPI: fixes a false alarm from lockdep), which really
is necessary to avoid false positive lockdep complaints.  Document
this and rearrange the code related to it so that it makes fewer
checks.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
Acked-by: Yinghai Lu <yinghai@xxxxxxxxxx>
---
 drivers/acpi/osl.c |   21 +++++++++++++--------
 1 file changed, 13 insertions(+), 8 deletions(-)

Index: linux/drivers/acpi/osl.c
===================================================================
--- linux.orig/drivers/acpi/osl.c
+++ linux/drivers/acpi/osl.c
@@ -932,7 +932,7 @@ static acpi_status __acpi_os_execute(acp
 	 * having a static work_struct.
 	 */
 
-	dpc = kmalloc(sizeof(struct acpi_os_dpc), GFP_ATOMIC);
+	dpc = kzalloc(sizeof(struct acpi_os_dpc), GFP_ATOMIC);
 	if (!dpc)
 		return AE_NO_MEMORY;
 
@@ -944,17 +944,22 @@ static acpi_status __acpi_os_execute(acp
 	 * because the hotplug code may call driver .remove() functions,
 	 * which invoke flush_scheduled_work/acpi_os_wait_events_complete
 	 * to flush these workqueues.
+	 *
+	 * To prevent lockdep from complaining unnecessarily, make sure that
+	 * there is a different static lockdep key for each workqueue by using
+	 * INIT_WORK() for each of them separately.
 	 */
-	queue = hp ? kacpi_hotplug_wq :
-		(type == OSL_NOTIFY_HANDLER ? kacpi_notify_wq : kacpid_wq);
-	dpc->wait = hp ? 1 : 0;
-
-	if (queue == kacpi_hotplug_wq)
+	if (hp) {
+		queue = kacpi_hotplug_wq;
+		dpc->wait = 1;
 		INIT_WORK(&dpc->work, acpi_os_execute_deferred);
-	else if (queue == kacpi_notify_wq)
+	} else if (type == OSL_NOTIFY_HANDLER) {
+		queue = kacpi_notify_wq;
 		INIT_WORK(&dpc->work, acpi_os_execute_deferred);
-	else
+	} else {
+		queue = kacpid_wq;
 		INIT_WORK(&dpc->work, acpi_os_execute_deferred);
+	}
 
 	/*
 	 * On some machines, a software-initiated SMI causes corruption unless


-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/