Skip to contentSkip to conversation

test for "spurious" IRQ ignores possible IRQ_WAKE_THREAD value

DISCUSSION Robert P. J. Day 3 participants 6 messages Sep 14, 2009 – Sep 17, 2009

lore
Share & tools
lore
#0 lore raw

never ashamed to embarrass myself in public, i just noticed the
following. from kernel/irq/spurious.c:

...
static void
__report_bad_irq(unsigned int irq, struct irq_desc *desc,
irqreturn_t action_ret)

{
        struct irqaction *action;

        if (action_ret != IRQ_HANDLED && action_ret != IRQ_NONE) {
                printk(KERN_ERR "irq event %d: bogus return value %x\n",
                                irq, action_ret);

but from include/linux/irqreturn.h, we see *three* possible return
values:

enum irqreturn {
IRQ_NONE,
IRQ_HANDLED,
IRQ_WAKE_THREAD,
};

typedef enum irqreturn irqreturn_t;
#define IRQ_RETVAL(x) ((x) != IRQ_NONE)

is there an inconsistency here?

rday
--

========================================================================
Robert P. J. Day Waterloo, Ontario, CANADA

Linux Consulting, Training and Annoying Kernel Pedantry.

Web page:                                          http://crashcourse.ca
Twitter:                                       http://twitter.com/rpjday

========================================================================

#1 lore raw
Quoted 23 lines · Robert P. J. Day

never ashamed to embarrass myself in public, i just noticed the
following. from kernel/irq/spurious.c:

...
static void
__report_bad_irq(unsigned int irq, struct irq_desc *desc,
irqreturn_t action_ret)

{
        struct irqaction *action;

        if (action_ret != IRQ_HANDLED && action_ret != IRQ_NONE) {
                printk(KERN_ERR "irq event %d: bogus return value %x\n",
                                irq, action_ret);

but from include/linux/irqreturn.h, we see *three* possible return
values:

enum irqreturn {
IRQ_NONE,
IRQ_HANDLED,
IRQ_WAKE_THREAD,
};

Quoted 4 lines

typedef enum irqreturn irqreturn_t;
#define IRQ_RETVAL(x) ((x) != IRQ_NONE)

is there an inconsistency here?

i should have pointed out that there is (apparently) only one place
in the entire tree that uses that third return value:

$ grep -rw IRQ_WAKE_THREAD *
drivers/net/wireless/b43/main.c:	return IRQ_WAKE_THREAD;  <-- there
include/linux/irqreturn.h: * @IRQ_WAKE_THREAD	handler requests to wake the handler thread
include/linux/irqreturn.h:	IRQ_WAKE_THREAD,
Binary file include/linux/.irqreturn.h.swp matches
include/linux/interrupt.h: * IRQTF_WARNED    - warning "IRQ_WAKE_THREAD w/o thread_fn" has been printed
kernel/irq/handle.c:	printk(KERN_WARNING "IRQ %d device %s returned IRQ_WAKE_THREAD "
kernel/irq/handle.c:		case IRQ_WAKE_THREAD:
kernel/irq/manage.c:	return IRQ_WAKE_THREAD;
kernel/irq/manage.c: *	IRQ_WAKE_THREAD which will wake up the handler thread and run
$

for what that's worth.

rday
--

========================================================================
Robert P. J. Day Waterloo, Ontario, CANADA

Linux Consulting, Training and Annoying Kernel Pedantry.

Web page:                                          http://crashcourse.ca
Twitter:                                       http://twitter.com/rpjday

========================================================================

#2 lore raw

[Robert P. J. Day - Mon, Sep 14, 2009 at 04:34:48PM -0400]

| On Mon, 14 Sep 2009, Robert P. J. Day wrote:
| 
| >   never ashamed to embarrass myself in public, i just noticed the
| > following.  from kernel/irq/spurious.c:
| >
| > ...
| > static void
| > __report_bad_irq(unsigned int irq, struct irq_desc *desc,
| >                  irqreturn_t action_ret)
| > {
| >         struct irqaction *action;
| >
| >         if (action_ret != IRQ_HANDLED && action_ret != IRQ_NONE) {
| >                 printk(KERN_ERR "irq event %d: bogus return value %x\n",
| >                                 irq, action_ret);
| >
| >   but from include/linux/irqreturn.h, we see *three* possible return
| > values:
| >
| > enum irqreturn {
| >         IRQ_NONE,
| >         IRQ_HANDLED,
| >         IRQ_WAKE_THREAD,
| > };
| >
| > typedef enum irqreturn irqreturn_t;
| > #define IRQ_RETVAL(x)   ((x) != IRQ_NONE)
| >
| >   is there an inconsistency here?
| 

...

Hi Robert,

It could that IRQ_WAKE_THREAD is just missed here. I suppose it was brough there as thread irq merged. But I think only Thomas know for sure, I definitely miss something :) CC'ed

-- Cyrill

#3 lore raw
Quoted 31 lines · Cyrill Gorcunov

[Robert P. J. Day - Mon, Sep 14, 2009 at 04:34:48PM -0400]

| On Mon, 14 Sep 2009, Robert P. J. Day wrote:
| 
| >   never ashamed to embarrass myself in public, i just noticed the
| > following.  from kernel/irq/spurious.c:
| >
| > ...
| > static void
| > __report_bad_irq(unsigned int irq, struct irq_desc *desc,
| >                  irqreturn_t action_ret)
| > {
| >         struct irqaction *action;
| >
| >         if (action_ret != IRQ_HANDLED && action_ret != IRQ_NONE) {
| >                 printk(KERN_ERR "irq event %d: bogus return value %x\n",
| >                                 irq, action_ret);
| >
| >   but from include/linux/irqreturn.h, we see *three* possible return
| > values:
| >
| > enum irqreturn {
| >         IRQ_NONE,
| >         IRQ_HANDLED,
| >         IRQ_WAKE_THREAD,
| > };
| >
| > typedef enum irqreturn irqreturn_t;
| > #define IRQ_RETVAL(x)   ((x) != IRQ_NONE)
| >
| >   is there an inconsistency here?
| 
Quoted 7 lines

...

Hi Robert,

It could that IRQ_WAKE_THREAD is just missed here. I suppose it was brough there as thread irq merged. But I think only Thomas know for sure, I definitely miss something :) CC'ed

from kernel/irq/handle.c:

		trace_irq_handler_entry(irq, action);
		ret = action->handler(irq, action->dev_id);
		trace_irq_handler_exit(irq, action, ret);

		switch (ret) {
		case IRQ_WAKE_THREAD:
			/*
			 * Set result to handled so the spurious check
			 * does not trigger.
			 */
			ret = IRQ_HANDLED;

So nothing to worry about :)

Thanks,

tglx

#4 lore raw
In reply to Cyrill Gorcunov
Quoted 31 lines · Cyrill Gorcunov

[Robert P. J. Day - Mon, Sep 14, 2009 at 04:34:48PM -0400]

| On Mon, 14 Sep 2009, Robert P. J. Day wrote:
|
| >   never ashamed to embarrass myself in public, i just noticed the
| > following.  from kernel/irq/spurious.c:
| >
| > ...
| > static void
| > __report_bad_irq(unsigned int irq, struct irq_desc *desc,
| >                  irqreturn_t action_ret)
| > {
| >         struct irqaction *action;
| >
| >         if (action_ret != IRQ_HANDLED && action_ret != IRQ_NONE) {
| >                 printk(KERN_ERR "irq event %d: bogus return value %x\n",
| >                                 irq, action_ret);
| >
| >   but from include/linux/irqreturn.h, we see *three* possible return
| > values:
| >
| > enum irqreturn {
| >         IRQ_NONE,
| >         IRQ_HANDLED,
| >         IRQ_WAKE_THREAD,
| > };
| >
| > typedef enum irqreturn irqreturn_t;
| > #define IRQ_RETVAL(x)   ((x) != IRQ_NONE)
| >
| >   is there an inconsistency here?
|
Quoted 7 lines

...

Hi Robert,

It could that IRQ_WAKE_THREAD is just missed here. I suppose it was brough there as thread irq merged. But I think only Thomas know for sure, I definitely miss something :) CC'ed

actually, after a bit more reading, i found this in
kernel/irq/handle.c:

                ...
                switch (ret) {
                case IRQ_WAKE_THREAD:
                        /*
                         * Set result to handled so the spurious check
                         * does not trigger.
                         */
                        ret = IRQ_HANDLED;
                        ...

so it looks like that value of IRQ_WAKE_THREAD is simply "mapped" to IRQ_HANDLED, and perhaps that's done before __report_bad_irq is ever called so that that latter routine never sees a value of IRQ_WAKE_THREAD. but that's just a guess.

rday
--

========================================================================
Robert P. J. Day Waterloo, Ontario, CANADA

Linux Consulting, Training and Annoying Kernel Pedantry.

Web page:                                          http://crashcourse.ca
Twitter:                                       http://twitter.com/rpjday

========================================================================

#5 lore raw

[Robert P. J. Day - Thu, Sep 17, 2009 at 04:08:32PM -0400]
...

| > Hi Robert,
| >
| > It could that IRQ_WAKE_THREAD is just missed here. I suppose it was
| > brough there as thread irq merged. But I think only Thomas know for
| > sure, I definitely miss something :) CC'ed
| 
|   actually, after a bit more reading, i found this in
| kernel/irq/handle.c:
|                 ...
|                 switch (ret) {
|                 case IRQ_WAKE_THREAD:
|                         /*
|                          * Set result to handled so the spurious check
|                          * does not trigger.
|                          */
|                         ret = IRQ_HANDLED;
|                         ...
| 
| so it looks like that value of IRQ_WAKE_THREAD is simply "mapped" to
| IRQ_HANDLED, and perhaps that's done before __report_bad_irq is ever
| called so that that latter routine never sees a value of
| IRQ_WAKE_THREAD.  but that's just a guess.
| 
| rday
| --

yeah, Thomas just pointed it too :) The note_interrupt is called after handle_IRQ_event (except a few drivers which don;t use threaded irq) so it doesnt reach bad irq state with IRQ_WAKE_THREAD, for now at least.

-- Cyrill