Interrupts first transfer to the intr_entry_start code sequence in
kernel.S, but that code is not used in place so placing a probe at that
address will not show anything. Instead, the instructions at intr_entry_start 
through intr_entry_end labels are copied to dynamically allocated memory and
intermixed with the interrupt_id and eoi callouts from the startup code.
See the cpu_interrupt_init() function in ker/x86/init_cpu.c for where
that happens.

On Tue, Jun 16, 2009 at 12:26:30PM -0400, Pietro Descombes wrote:
> Hi everyone,
> 
> I'm trying to find the interrupt entry point. On X68, does the CPU 
> directly jump to the handler pointed to by the 'interrupt_vector[]' or 
> is there a function equivalent to linux 'do_IRQ()' ?
> 
> My goal is to monitor interrupts on my own tracer but I can't find where 
> to put my probes.
> 
> 
> (( 'nano_interrupt.c', 'ker_interrupt.c' even 'x86/kernel.S' didn't help 
> me much. The 'level' and 'itp->id' parameters of interrupt_mask 
> (nano_interrupt.c) are often 1 and 5 respectively, while I expect to 
> receive at least the timer IRQ frequently which should be interrupt '0' 
> on x86. ))
> 
> 
> Cheers,
> Pietro
> 
> 
> Using Momentics 6.4.0, instrumented kernel.
> 
> _______________________________________________
> OSTech
> http://community.qnx.com/sf/go/post31895
> 

-- 
Brian Stecher (bstecher@qnx.com)        QNX Software Systems
phone: +1 (613) 591-0931 (voice)        175 Terence Matthews Cr.
       +1 (613) 591-3579 (fax)          Kanata, Ontario, Canada K2M 1W8