diff mbox

genirq: provide means to retrigger parent

Message ID 1350425269-11489-1-git-send-email-khilman@deeprootsystems.com (mailing list archive)
State New, archived
Headers show

Commit Message

Kevin Hilman Oct. 16, 2012, 10:07 p.m. UTC
From: Thomas Gleixner <tglx@linutronix.de>

Attempts to retrigger nested threaded IRQs currently fail because they
have no primary handler.  In order to support retrigger of nested
IRQs, the parent IRQ needs to be retriggered.

To fix, when an IRQ needs to be resent, if the interrupt has a parent
IRQ and runs in the context of the parent IRQ, then resend the parent.

Also, handle_nested_irq() needs to clear the replay flag like the
other handlers, otherwise check_irq_resend() will set it and it will
never be cleared.  Without clearing, it results in the first resend
working fine, but check_irq_resend() returning early on subsequent
resends because the replay flag is still set.

Problem discovered on ARM/OMAP platforms where a nested IRQ that's
also a wakeup IRQ happens late in suspend and needed to be retriggered
during the resume process.

Reported-by: Kevin Hilman <khilman@ti.com>
Tested-by: Kevin Hilman <khilman@ti.com>
[khilman@ti.com: changelog edits, clear IRQS_REPLAY in handle_nested_irq()]
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
Applies on v3.7-rc1

 include/linux/irq.h     |    9 +++++++++
 include/linux/irqdesc.h |    3 +++
 kernel/irq/chip.c       |    1 +
 kernel/irq/manage.c     |   16 ++++++++++++++++
 kernel/irq/resend.c     |    8 ++++++++
 5 files changed, 37 insertions(+)

Comments

Russell King - ARM Linux Oct. 16, 2012, 10:15 p.m. UTC | #1
On Tue, Oct 16, 2012 at 03:07:49PM -0700, Kevin Hilman wrote:
> From: Thomas Gleixner <tglx@linutronix.de>
> 
> Attempts to retrigger nested threaded IRQs currently fail because they
> have no primary handler.  In order to support retrigger of nested
> IRQs, the parent IRQ needs to be retriggered.
> 
> To fix, when an IRQ needs to be resent, if the interrupt has a parent
> IRQ and runs in the context of the parent IRQ, then resend the parent.
> 
> Also, handle_nested_irq() needs to clear the replay flag like the
> other handlers, otherwise check_irq_resend() will set it and it will
> never be cleared.  Without clearing, it results in the first resend
> working fine, but check_irq_resend() returning early on subsequent
> resends because the replay flag is still set.
> 
> Problem discovered on ARM/OMAP platforms where a nested IRQ that's
> also a wakeup IRQ happens late in suspend and needed to be retriggered
> during the resume process.
> 
> Reported-by: Kevin Hilman <khilman@ti.com>
> Tested-by: Kevin Hilman <khilman@ti.com>
> [khilman@ti.com: changelog edits, clear IRQS_REPLAY in handle_nested_irq()]
> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

Umm, we also have the converse situation.  We have platforms where the
resend has to be done from the child IRQ, and the parent must not be
touched.  I hope that doesn't break those.
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Kevin Hilman Oct. 23, 2012, 10:23 p.m. UTC | #2
Russell King - ARM Linux <linux@arm.linux.org.uk> writes:

> On Tue, Oct 16, 2012 at 03:07:49PM -0700, Kevin Hilman wrote:
>> From: Thomas Gleixner <tglx@linutronix.de>
>> 
>> Attempts to retrigger nested threaded IRQs currently fail because they
>> have no primary handler.  In order to support retrigger of nested
>> IRQs, the parent IRQ needs to be retriggered.
>> 
>> To fix, when an IRQ needs to be resent, if the interrupt has a parent
>> IRQ and runs in the context of the parent IRQ, then resend the parent.
>> 
>> Also, handle_nested_irq() needs to clear the replay flag like the
>> other handlers, otherwise check_irq_resend() will set it and it will
>> never be cleared.  Without clearing, it results in the first resend
>> working fine, but check_irq_resend() returning early on subsequent
>> resends because the replay flag is still set.
>> 
>> Problem discovered on ARM/OMAP platforms where a nested IRQ that's
>> also a wakeup IRQ happens late in suspend and needed to be retriggered
>> during the resume process.
>> 
>> Reported-by: Kevin Hilman <khilman@ti.com>
>> Tested-by: Kevin Hilman <khilman@ti.com>
>> [khilman@ti.com: changelog edits, clear IRQS_REPLAY in handle_nested_irq()]
>> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
>
> Umm, we also have the converse situation.  We have platforms where the
> resend has to be done from the child IRQ, and the parent must not be
> touched.  I hope that doesn't break those.

I'm assuming the child IRQs you're concerned with are not threaded,
right?  This patch only addresses nested, threaded IRQs, and these don't
have a primary handler to run at all, so cannot do any triggering.

Kevin

--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Thomas Gleixner Oct. 23, 2012, 10:36 p.m. UTC | #3
On Tue, 23 Oct 2012, Kevin Hilman wrote:

> Russell King - ARM Linux <linux@arm.linux.org.uk> writes:
> 
> > On Tue, Oct 16, 2012 at 03:07:49PM -0700, Kevin Hilman wrote:
> >> From: Thomas Gleixner <tglx@linutronix.de>
> >> 
> >> Attempts to retrigger nested threaded IRQs currently fail because they
> >> have no primary handler.  In order to support retrigger of nested
> >> IRQs, the parent IRQ needs to be retriggered.
> >> 
> >> To fix, when an IRQ needs to be resent, if the interrupt has a parent
> >> IRQ and runs in the context of the parent IRQ, then resend the parent.
> >> 
> >> Also, handle_nested_irq() needs to clear the replay flag like the
> >> other handlers, otherwise check_irq_resend() will set it and it will
> >> never be cleared.  Without clearing, it results in the first resend
> >> working fine, but check_irq_resend() returning early on subsequent
> >> resends because the replay flag is still set.
> >> 
> >> Problem discovered on ARM/OMAP platforms where a nested IRQ that's
> >> also a wakeup IRQ happens late in suspend and needed to be retriggered
> >> during the resume process.
> >> 
> >> Reported-by: Kevin Hilman <khilman@ti.com>
> >> Tested-by: Kevin Hilman <khilman@ti.com>
> >> [khilman@ti.com: changelog edits, clear IRQS_REPLAY in handle_nested_irq()]
> >> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
> >
> > Umm, we also have the converse situation.  We have platforms where the
> > resend has to be done from the child IRQ, and the parent must not be
> > touched.  I hope that doesn't break those.
> 
> I'm assuming the child IRQs you're concerned with are not threaded,
> right?  This patch only addresses nested, threaded IRQs, and these don't
> have a primary handler to run at all, so cannot do any triggering.

And it involves that you activly set the parent irq via the new
interface: irq_set_parent()

You don't have that yet or you don't use that in your future changes,
then you're good. :)

Thanks,

	tglx


--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/include/linux/irq.h b/include/linux/irq.h
index 216b0ba..526f10a 100644
--- a/include/linux/irq.h
+++ b/include/linux/irq.h
@@ -392,6 +392,15 @@  static inline void irq_move_masked_irq(struct irq_data *data) { }
 
 extern int no_irq_affinity;
 
+#ifdef CONFIG_HARDIRQS_SW_RESEND
+int irq_set_parent(int irq, int parent_irq);
+#else
+static inline int irq_set_parent(int irq, int parent_irq)
+{
+	return 0;
+}
+#endif
+
 /*
  * Built-in IRQ handlers for various IRQ types,
  * callable via desc->handle_irq()
diff --git a/include/linux/irqdesc.h b/include/linux/irqdesc.h
index 0ba014c..623325e 100644
--- a/include/linux/irqdesc.h
+++ b/include/linux/irqdesc.h
@@ -11,6 +11,8 @@ 
 struct irq_affinity_notify;
 struct proc_dir_entry;
 struct module;
+struct irq_desc;
+
 /**
  * struct irq_desc - interrupt descriptor
  * @irq_data:		per irq and chip data passed down to chip functions
@@ -65,6 +67,7 @@  struct irq_desc {
 #ifdef CONFIG_PROC_FS
 	struct proc_dir_entry	*dir;
 #endif
+	int			parent_irq;
 	struct module		*owner;
 	const char		*name;
 } ____cacheline_internodealigned_in_smp;
diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c
index 57d86d0..3aca9f2 100644
--- a/kernel/irq/chip.c
+++ b/kernel/irq/chip.c
@@ -272,6 +272,7 @@  void handle_nested_irq(unsigned int irq)
 
 	raw_spin_lock_irq(&desc->lock);
 
+	desc->istate &= ~(IRQS_REPLAY | IRQS_WAITING);
 	kstat_incr_irqs_this_cpu(irq, desc);
 
 	action = desc->action;
diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c
index 4c69326..d06a396 100644
--- a/kernel/irq/manage.c
+++ b/kernel/irq/manage.c
@@ -616,6 +616,22 @@  int __irq_set_trigger(struct irq_desc *desc, unsigned int irq,
 	return ret;
 }
 
+#ifdef CONFIG_HARDIRQS_SW_RESEND
+int irq_set_parent(int irq, int parent_irq)
+{
+	unsigned long flags;
+	struct irq_desc *desc = irq_get_desc_lock(irq, &flags, 0);
+
+	if (!desc)
+		return -EINVAL;
+
+	desc->parent_irq = parent_irq;
+
+	irq_put_desc_unlock(desc, flags);
+	return 0;
+}
+#endif
+
 /*
  * Default primary interrupt handler for threaded interrupts. Is
  * assigned as primary handler when request_threaded_irq is called
diff --git a/kernel/irq/resend.c b/kernel/irq/resend.c
index 6454db7..9065107 100644
--- a/kernel/irq/resend.c
+++ b/kernel/irq/resend.c
@@ -74,6 +74,14 @@  void check_irq_resend(struct irq_desc *desc, unsigned int irq)
 		if (!desc->irq_data.chip->irq_retrigger ||
 		    !desc->irq_data.chip->irq_retrigger(&desc->irq_data)) {
 #ifdef CONFIG_HARDIRQS_SW_RESEND
+			/*
+			 * If the interrupt has a parent irq and runs
+			 * in the thread context of the parent irq,
+			 * retrigger the parent.
+			 */
+			if (desc->parent_irq &&
+			    irq_settings_is_nested_thread(desc))
+				irq = desc->parent_irq;
 			/* Set it pending and activate the softirq: */
 			set_bit(irq, irqs_resend);
 			tasklet_schedule(&resend_tasklet);