Re: Questions about request_irq and reading PCI_INTERRUPT_LINE

["Le Wen" <le_wen@hotmail.com>]

On Fri, 18 Mar 2005, Le Wen wrote:

>Hi, there,
>
>I have problem to grab video from my ati all-in-wonder card. The card is in 
>a PII Celeron machine with an on board video card (ATI Technologies Inc 3D 
>Rage IIC AGP). there is no monitor connected with the on board video card. 
>I only hook my AIW card with a monitor.
>
>I use km-0.6 from gatos project. I load this km_drv module, but kernel 
>always complains:
>
>km: IRQ 0 busy
>
>I checked code:
>       km_probe(struct pci_dev *dev, const struct pci_device_id *pci_id)
>
>here dev->irq with a value 0.
>
>When km_probe gets called, it try to request an IRQ0 returns a -EBUSY:
>       kms_irq=dev.irq;
>       result=request_irq(kms->irq, handler, SA_SHIRQ, tag, (void *)kms);
>
>       if(result==-EBUSY){
>               printk(KERN_ERR "km: IRQ %ld busy\n", kms->irq);
>               goto fail;
>       }
>
>
>So I tried to get right IRQ number using:
>       u8 myirq;
>       int rtn=pci_read_config_byte(dev,PCI_INTERRUPT_LINE, &myirq);
>       dev->irq=myirq;
>       kms->irq=dev_irq;
>       result=request_irq(kms->irq, handler, SA_SHIRQ, tag, (void *)kms);
>
>       if(result==-EBUSY){
>               printk(KERN_ERR "km: IRQ %ld busy\n", kms->irq);
>               goto fail;
>       }
>       if(result<0){
>               printk(KERN_ERR "km: could not install irq handler: 
>result=%d\n",result);
>               goto fail;
>       }
>But this time I got:
>
>km: kms->irq=24
>km: could not install irq handler: result=-38
>
>
>My questions are:
>1. I don't know why dev->irq has value of 0?
>

The PCI interface now needs to be enabled first. The IRQ value
returned is BAD until after one calls pci_enable_device(). This
is a BUG, now considered a FEATURE so it's unlikely to be fixed!
There are lots of people who have encountered this problem
with modules that are not in the "distribution".

>2. Is an IRQ number of 24 valid for a Intel PII Celeron?
>

Could be, but it;s probably invalid considering the way you
got it.

>3. What does this result=-38 mean?
>

Probably errno 38, i.e., ENOSYS

>
>Wen, Le
>

Cheers,
Dick Johnson
Penguin : Linux version 2.6.11 on an i686 machine (5537.79 BogoMips).
  Notice : All mail here is now cached for review by Dictator Bush.
                  98.36% of all statistics are fiction.

Thank you Dick!

But pci_enable_device(dev) was called sucussful before read 
pci_read_config_byte():

static int __devinit km_probe(struct pci_dev *dev, const struct 
pci_device_id *pci_id)
{
...
if (pci_enable_device(dev))
                return -EIO;
        printk(KERN_DEBUG "pci_read_config_byte(dev,PCI_INTERRUPT_LINE, 
&testirq)\n");
        u8 myirq;//=0;
        int rtn=pci_read_config_byte(dev,PCI_INTERRUPT_LINE, &myirq);
        kms->dev=dev;
        dev->irq=myirq;
        kms->irq=dev->irq;
and then call
       result=request_irq(kms->irq, handler, SA_SHIRQ, tag, (void *)kms);