Date: Thu, 21 Apr 2016 12:17:07 -0300
From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>, Ingo Molnar <mingo@kernel.org>,
        Adrian Hunter <adrian.hunter@intel.com>,
        Brendan Gregg <brendan.d.gregg@gmail.com>,
        Alexander Shishkin <alexander.shishkin@linux.intel.com>,
        Alexei Starovoitov <ast@kernel.org>, David Ahern <dsahern@gmail.com>,
        He Kuang <hekuang@huawei.com>, Jiri Olsa <jolsa@redhat.com>,
        Masami Hiramatsu <mhiramat@kernel.org>,
        Milian Wolff <milian.wolff@kdab.com>,
        Namhyung Kim <namhyung@kernel.org>,
        Stephane Eranian <eranian@google.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Vince Weaver <vincent.weaver@maine.edu>,
        Wang Nan <wangnan0@huawei.com>, Zefan Li <lizefan@huawei.com>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH/RFC] perf core: Allow setting up max frame stack depth
 via sysctl
Message-ID: <20160421151707.GB3677@kernel.org>
References: <20160420224730.GX3677@kernel.org>
 <20160421104858.GH3408@twins.programming.kicks-ass.net>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20160421104858.GH3408@twins.programming.kicks-ass.net>
User-Agent: Mutt/1.5.24 (2015-08-30)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2077
Lines: 62

Em Thu, Apr 21, 2016 at 12:48:58PM +0200, Peter Zijlstra escreveu:
> On Wed, Apr 20, 2016 at 07:47:30PM -0300, Arnaldo Carvalho de Melo wrote:
> > The default remains 127, which is good for most cases, and not even hit
> > most of the time, but then for some cases, as reported by Brendan, 1024+
> > deep frames are appearing on the radar for things like groovy, ruby.
 
> yea gawds ;-)
 
> > +++ b/kernel/events/callchain.c

> > @@ -73,7 +81,7 @@ static int alloc_callchain_buffers(void)
> >  	if (!entries)
> >  		return -ENOMEM;
> >  
> > -	size = sizeof(struct perf_callchain_entry) * PERF_NR_CONTEXTS;
> > +	size = perf_callchain_entry__sizeof() * PERF_NR_CONTEXTS;
> >  
> >  	for_each_possible_cpu(cpu) {
> >  		entries->cpu_entries[cpu] = kmalloc_node(size, GFP_KERNEL,
> 
> And this alloc _will_ fail if you put in a decent sized value..
> 
> Should we put in a dmesg WARN if this alloc fails and
> perf_event_max_stack is 'large' ?

Unsure, it already returns -ENOMEM, see, some lines above, i.e. it
better have error handling up this, ho-hum, call chain, I'm checking...
 
> > @@ -215,3 +223,25 @@ exit_put:
> >  
> >  	return entry;
> >  }
> > +
> > +int perf_event_max_stack_handler(struct ctl_table *table, int write,
> > +				 void __user *buffer, size_t *lenp, loff_t *ppos)
> > +{
> > +	int new_value = sysctl_perf_event_max_stack, ret;
> > +	struct ctl_table new_table = *table;
> > +
> > +	new_table.data = &new_value;
 
> cute :-)

Hey, I found it on sysctl_schedstats() and sysctl_numa_balancing(), as a
way to read that value but only make it take effect if some condition
was true (nr_callchain_events == 0 in this case), granted, could be
better, less clever, but I leave this for later ;-)
 
> > +	ret = proc_dointvec_minmax(&new_table, write, buffer, lenp, ppos);
> > +	if (ret || !write)
> > +		return ret;
> > +
> > +	mutex_lock(&callchain_mutex);
> > +	if (atomic_read(&nr_callchain_events))
> > +		ret = -EBUSY;
> > +	else
> > +		sysctl_perf_event_max_stack = new_value;
> > +	
> > +	mutex_unlock(&callchain_mutex);
> > +
> > +	return ret;
> > +}