AW: [fusion] rt_intr_bind

Philippe Gerum rpm at xenomai.org
Fri Aug 5 13:21:01 CEST 2005


Studencki Pawel wrote:
> 
>>	Not this file. "/proc/rtai/register/interrupts" must contain a
> 
> string like "interrupt at N" (N == the second argument 
> 
>>of rt_intr_create()). The exactly same string should be passed to the
> 
> rt_intr_bind().
> 
>>	
>>	If you don't see this file at all (after a moment when
> 
> rt_intr_create() has been called, of course), it means either > the
> interrupt object has not been properly created (see what rt_intr_create()
> returns) or the registry support have not 
> 
>>been configured in fusion (CONFIG_RTAI_OPT_NATIVE_REGISTRY).
> 
> 
> 
> it is very strange...
> 
> output from my kernel module:
> -----------------------------------------
> /fusion # insmod ppc-rtc.ko
> Using ppc-rtc.ko
> rt_intr_create ret = 0
> ppc_rtc_int_handler 1	<------------------- printk in ISR at the beginning
> ppc_rtc_int_handler 2	<------------------- printk in ISR at the end
> rt_intr_enable ret = 0
> ----------------------------------------
> 
> rt_intr_create and rt_intr_enable both return 0, but ISR is called only one
> time.
> Unfortunately I don't have any "/proc/rtai/register/interrupts", but fusion
> is configured with CONFIG_RTAI_OPT_NATIVE_REGISTRY="y".
> 

Ok, what's wrong is that the intr object does not provide export 
information to /proc to the registry yet; I'll fix this. This said, you 
should still be able to bind to interrupt objects from user-space, both 
issues are independent from each other.

-- 

Philippe.




More information about the Rtai mailing list