Re: [PATCH 2/3] trace: fix default boot up tracer

[Andrew Morton <akpm@linux-foundation.org>]

On Mon, 02 Feb 2009 21:38:32 -0500 Steven Rostedt <rostedt@goodmis.org> wrote:

> @@ -510,8 +520,25 @@ int register_tracer(struct tracer *type)
>   out:
>  	tracing_selftest_running = false;
>  	mutex_unlock(&trace_types_lock);
> -	lock_kernel();
>  
> +	if (!ret && default_bootup_tracer) {
> +		if (!strncmp(default_bootup_tracer, type->name,
> +			     BOOTUP_TRACER_SIZE)) {
> +			printk(KERN_INFO "Starting tracer '%s'\n",
> +			       type->name);
> +			/* Do we want this tracer to start on bootup? */
> +			tracing_set_tracer(type->name);
> +			default_bootup_tracer = NULL;
> +			/* disable other selftests, since this will break it. */
> +			tracing_selftest_disabled = 1;
> +#ifdef CONFIG_FTRACE_STARTUP_TEST
> +			printk(KERN_INFO "Disabling FTRACE selftests due"
> +			       " to running tracer '%s'\n", type->name);
> +#endif
> +		}
> +	}
> +
> +	lock_kernel();
>  	return ret;
>  }

The fun and games which register_tracer() plays with lock_kernel() tell
us that this function is only called at bootup time and hence could be
__init.  A quick whizz through callers confirms that.

And if register_tracer() is also only callable at bootup, one suspects
that unregister_tracer() isn't useful.  And lo, it has no callers.

This leads one to further surmise that trace_types_lock a) could be
__initdata and b) could be removed (the list is only altered when we're
running single-threaded).  This appears to be the case.