Re: [PATCH] rcu: Have rcu torture test select trace_clock

From: Steven Rostedt
Date: Mon Feb 04 2013 - 13:36:19 EST


On Mon, 2013-02-04 at 08:59 -0800, Paul E. McKenney wrote:
> On Mon, Feb 04, 2013 at 10:52:21AM -0500, Steven Rostedt wrote:
> > The rcu torture test uses trace_clock, and requires that it be built
> > into the kernel. If it is not, then we get the following build error:
> >
> > ERROR: "trace_clock_local" [kernel/rcutorture.ko] undefined!
> >
> > Reported-by: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
> > Signed-off-by: Steven Rostedt <rostedt@xxxxxxxxxxx>
>
> This would work, but would unnecessarily force trace_clock_local()
> to be defined when RCU tracing is disabled. So, would the alternative
> patch I posted work for you?
>

Unfortunately the other patch wasn't Cc'd to LKML (I hate private
emails).

I would suggest the following to be complete, why keep that extra divide
around?

-- Steve

I started with your patch:


rcu: Allow rcutorture to be built at low optimization levels

The uses of trace_clock_local() are dead code when CONFIG_RCU_TRACE=n,
but some compilers might nevertheless generate code calling this function.
This commit therefore ensures that trace_clock_local() is invoked only
when CONFIG_RCU_TRACE=y.

Reported-by: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx>
Signed-off-by: Paul E. McKenney <paulmck@xxxxxxxxxxxxxxxxxx>
Signed-off-by: Steven Rostedt <rostedt@xxxxxxxxxxx>

Index: linux-trace.git/kernel/rcutorture.c
===================================================================
--- linux-trace.git.orig/kernel/rcutorture.c
+++ linux-trace.git/kernel/rcutorture.c
@@ -208,6 +208,20 @@ MODULE_PARM_DESC(rcutorture_runnable, "S
#define rcu_can_boost() 0
#endif /* #else #if defined(CONFIG_RCU_BOOST) && !defined(CONFIG_HOTPLUG_CPU) */

+#ifdef CONFIG_RCU_TRACE
+static u64 notrace rcu_trace_clock_local(void)
+{
+ u64 ts = trace_clock_local();
+ unsigned long __maybe_unused ts_rem = do_div(ts, NSEC_PER_USEC);
+ return ts;
+}
+#else /* #ifdef CONFIG_RCU_TRACE */
+static u64 notrace rcu_trace_clock_local(void)
+{
+ return 0ULL;
+}
+#endif /* #else #ifdef CONFIG_RCU_TRACE */
+
static unsigned long shutdown_time; /* jiffies to system shutdown. */
static unsigned long boost_starttime; /* jiffies of next boost test start. */
DEFINE_MUTEX(boost_mutex); /* protect setting boost_starttime */
@@ -1051,7 +1065,7 @@ static void rcu_torture_timer(unsigned l

idx = cur_ops->readlock();
completed = cur_ops->completed();
- ts = trace_clock_local();
+ ts = rcu_trace_clock_local();
p = rcu_dereference_check(rcu_torture_current,
rcu_read_lock_bh_held() ||
rcu_read_lock_sched_held() ||
@@ -1075,8 +1089,6 @@ static void rcu_torture_timer(unsigned l
}
completed_end = cur_ops->completed();
if (pipe_count > 1) {
- unsigned long __maybe_unused ts_rem = do_div(ts, NSEC_PER_USEC);
-
do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu, ts,
completed, completed_end);
rcutorture_trace_dump();
@@ -1122,7 +1134,7 @@ rcu_torture_reader(void *arg)
}
idx = cur_ops->readlock();
completed = cur_ops->completed();
- ts = trace_clock_local();
+ ts = rcu_trace_clock_local();
p = rcu_dereference_check(rcu_torture_current,
rcu_read_lock_bh_held() ||
rcu_read_lock_sched_held() ||
@@ -1144,9 +1156,6 @@ rcu_torture_reader(void *arg)
}
completed_end = cur_ops->completed();
if (pipe_count > 1) {
- unsigned long __maybe_unused ts_rem =
- do_div(ts, NSEC_PER_USEC);
-
do_trace_rcu_torture_read(cur_ops->name, &p->rtort_rcu,
ts, completed, completed_end);
rcutorture_trace_dump();


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/