[LITMUS^RT] RFC: record interrupt interference with Feather-Trace

[Björn Brandenburg]

On Jan 24, 2012, at 9:46 PM, Andrea Bastoni <bastoni at cs.unc.edu> wrote:

>> +#include <litmus/trace_irq.h>
>> +
>> /*
>>  * We put the hardirq and softirq counter into the preemption
>>  * counter. The bitmask has the following meaning:
>> @@ -186,6 +188,7 @@ extern void rcu_nmi_exit(void);
>>        account_system_vtime(current);        \
>>        add_preempt_count(HARDIRQ_OFFSET);    \
>>        trace_hardirq_enter();            \
>> +        ft_irq_fired();                \
>>    } while (0)
> 
> But isn't this one only tracing NMI's ? :)

Sure looks like it, but I am very certain that I was tracing regular IRQs and not NMIs. I think git got confused with the context since it's a file full of multiline macros and not regular C code. A function with a body certainly wouldn't be declared as 'extern', so the whole snippet is bogus (well, it would apply cleanly, but it is misleading). Also, I added the hooks to the _enter() macros, and not the _exit() macros.

- Björn