Messages in this thread Patch in this message | | | Date | Tue, 26 May 2009 23:50:46 +0200 (CEST) | From | Thomas Gleixner <> | Subject | Re: [PATCH 3/3] ftrace: add tracepoint for itimer |
| |
On Fri, 22 May 2009, Xiao Guangrong wrote: > +TRACE_EVENT(itimer_start, > + > + TP_PROTO(int which, struct itimerval *value), > + > + TP_ARGS(which, value), > + > + TP_STRUCT__entry( > + __field( int, which ) > + __field( unsigned long long, it_value ) > + __field( unsigned long long, it_interval ) > + __string( comm, current->comm ) > + ), > + > + TP_fast_assign( > + __entry->which = which; > + __entry->it_value = (which == ITIMER_REAL ? > + ktime_to_ns(timeval_to_ktime(value->it_value)) : > + timeval_to_cputime(&value->it_value)); > + __entry->it_interval = (which == ITIMER_REAL ? > + ktime_to_ns(timeval_to_ktime(value->it_interval)) : > + timeval_to_cputime(&value->it_interval)); > + __assign_str(comm, current->comm); > + ),
Again, conversion of raw values needs to be done when we analyse the trace and not in the fast path.
....
> + } else{ } else {
please > tsk->signal->it_real_incr.tv64 = 0; > + trace_itimer_cancel(ITIMER_REAL); > + } > > spin_unlock_irq(&tsk->sighand->siglock); > break; > @@ -186,7 +191,10 @@ again: > jiffies_to_cputime(1)); > set_process_cpu_timer(tsk, CPUCLOCK_VIRT, > &nval, &cval); > - } > + trace_itimer_start(ITIMER_VIRTUAL, value); > + }else
Please run your patches through checkpatch.pl
Thanks,
tglx
| |