RE: Interrupt trouble due to IRQ router VIA?

[Kathy Frazier]

Thanks for your response.  I have tried other methods for sleeping to avoid
the race condition you are speaking of, but have not had any luck.  At this
point, I am using the wait_event_interruptible, but this routine never
returns - I must Cntrl C to stop my test program.  So now, the driver looks
something like this:

In routine called from ioctl:

/* a global, declared static int */
RcvIntr = 0;

/* cause H/W to interrupt */
writel(Data, HardwareAddress);
wait_event_interruptible(RcvIntrrcvQ, RcvIntr);

In ISR:

if (USER_INTERRUPTING(pslState)
   {
    ResetReceiverInterrupt(pslState); /* clear the board's interrupt */
    RcvIntr = 1;
    wake_up_interruptible(&RcvIntrrcvQ);
   }

Since I never get into my ISR, I added some [nasty] kernel hacks in
/arch/i386/kernel/irq.c to see if the O/S ever hears of the interrupt.  In a
nutshell:

Global (after the #includes):
static unsigned myIRQ = 0;

In request_irq:

   if (!action)
      return = -ENOMEM;

+  if (strcmp(devname,"pslengrave") == 0)
+     {
+     myIRQ = irq;
+     printk(KERN_INFO "KJF: monitor pslengrave IRQ\n");
+     }

   action->handler = handler;

In do_IRQ:

   kstat.irqs[cpu][irq]++;

+  if ((myIRQ !=0) && (irq == myIRQ)) printk (KERN_INFO "KJF: Get spin lock
for IRQ %d\n", irq);
   spin_lock(&desc->lock);

+  if ((myIRQ != 0) && (irq == myIRQ)) printk (KERN_INFO "KJF: Calling ack
handler for IRQ %d\n", irq);

   desc->handler ->ack(irq);

Told you it was nasty!  Note that my driver requests the IRQ (as specified
in the pci_dev structure) as sharable.  However,in the current
configuration, no other device is using this IRQ. The request_irq routine
completes successfully and displays my message, but no sign of myIRQ in the
do_IRQ routine.

I appreciate any advice on how to solve this.  I am also open to doing any
testing that may be necessary to determine why the OS never hears of the
interrupt.  Once our logic analyzer is freed up, I can verify that the board
is interrupting . . . although we have been that route before when I was
testing in my original developement system.  The device was interrupting
fine . . .

Thanks in advance for your help!

Kathy

-----Original Message-----
From: [e-mail]
[mailto:[e-mail]]On Behalf Of Alan Cox
Sent: Wednesday, July 31, 2002 8:35 PM
To: Kathy Frazier
Cc: [e-mail]
Subject: Re: Interrupt trouble due to IRQ router VIA?

On Wed, 2002-07-31 at 21:15, Kathy Frazier wrote:
> another system.  In this system, the driver times out on the
> interruptible_sleep_on_timeout waiting for the interrupt.  One thing that
I

There is a classic kernel programming error which goes something like
this

my_interrupt()
{
	foo->ready = 1;
	wake_up(&foo_pending);
}

foo_read()
{
	while(!foo->ready && !signal_pending(current))
		interruptible_sleep_on(&foo_pending);

}

What happens then is this

	foo->ready = 0 - true
	signal pending = 0 - ok

		INTERRUPT
			foo->ready=1
			wake up
		END INTERRUPT

	interruptible_sleep_on(&foo_pending);

Waits forever

It could be your timings have changed so such a bug now shows up