Re: mv-linux: Problem to implement custom driver interrupt handling

[Andrei Konovalov <akonovalov@ru.mvista.com>]

Eckart Göhler wrote:
> 
> Andrei Konovalov wrote:
>  > Hi,
>  >
>  > In the Linux driver you should not access the interrupt controller
>  > directly.
>  > The relevant XIntc_* calls are done by arch/ppc/syslib/xilinx_pic.c 
> code.
>  > E.g. the particular interrupt is unmasked when one calls request_irq().
>  >
>  > Few more comments below.
> 
> 
> Hi,
> 
> Thanks a lot. Actually I inserted the low-level code below after 
> request_irq() did not work. The note about xilinx_pic.c code (that is 
> located in my implementation in arch/ppc/kernel/xilinx_pic.c) lead me to 

Correct. arch/ppc/syslib/xilinx_pic.c is the 2.6 kernel case.

> the problem origin that should be reported here for the community (which 
> I presume already know the very fact):
> The interrupt numbering generated by the EDK is opposite to the one used 
> by linux, i.e. interrupt number 4 reported in EDK-generated 
> xparameters.h/xparameters_ml300.h becomes 31-4 = 27 when using request_irq.

Yes, this is true for 2.4 kernels.

In 2.6 the "natural" irq numbering is used.
I.e. for irq number of 4 (as per xparameters.h) one should pass
to request_irq()
    31-4 = 27 if using 2.4 kernel
and
   4 if using 2.6 kernel.

Thanks,
Andrei

> Therefore the handler was not called because he was attached to the 
> wrong interrupt, and also was not able to reset the interrupt pending 
> flag, that must be done as you noted below.
> 
> cheers
> 
> eckart