Re: PCI Device Drivers

[Greg KH <greg@xxxxxxxxx>]

On Wed, Jan 04, 2006 at 11:15:50AM -0700, Sonja Tideman wrote:
> Hello,
> 
>  I am trying to debug some problems with a device driver for a cPCI device.
> For some reason, the authors of the driver do not call pci_register_driver
> or pci_enable_device.  Should this cause any problems?

Yes, it will keep the device from working properly (well, the
pci_enable_device() call, the register_driver one is just them not using
the PCI api properly.)

> The driver registers itself via register_chrdev and calls request_irq
> to get the interrupt line.
> 
> The problems I am seeing are when it and the ethernet driver share the same
> interrupt line.  For some reason, the device driver's release function seems
> to occasionally get called (when I don't belive it should be called).

You mean the chrdev release function?  That's up to userspace, when it
closes the file handle.

> The release function frees the irq (via free_irq), but this seems to
> cause the ethernet driver to stop getting interrupts.  Any ideas?

Have a pointer to the source?

thanks,

greg k-h

--
Kernelnewbies: Help each other learn about the Linux kernel.
Archive:       http://mail.nl.linux.org/kernelnewbies/
FAQ:           http://kernelnewbies.org/faq/