From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner+w=401wt.eu-S1757542AbZDXGCc@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1757542AbZDXGCc (ORCPT <rfc822;w@1wt.eu>);
	Fri, 24 Apr 2009 02:02:32 -0400
Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1756186AbZDXGB4
	(ORCPT <rfc822;linux-kernel-outgoing>);
	Fri, 24 Apr 2009 02:01:56 -0400
Received: from cn.fujitsu.com ([222.73.24.84]:52028 "EHLO song.cn.fujitsu.com"
	rhost-flags-OK-FAIL-OK-OK) by vger.kernel.org with ESMTP
	id S1755881AbZDXGBx (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
	Fri, 24 Apr 2009 02:01:53 -0400
Message-ID: <49F15607.1090707@cn.fujitsu.com>
Date: Fri, 24 Apr 2009 14:02:47 +0800
From: Li Zefan <lizf@cn.fujitsu.com>
User-Agent: Thunderbird 2.0.0.9 (X11/20071115)
MIME-Version: 1.0
To: Steven Rostedt <rostedt@goodmis.org>
CC: linux-kernel@vger.kernel.org, Ingo Molnar <mingo@elte.hu>,
       Andrew Morton <akpm@linux-foundation.org>,
       Peter Zijlstra <peterz@infradead.org>,
       Frederic Weisbecker <fweisbec@gmail.com>,
       Steven Rostedt <srostedt@redhat.com>
Subject: Re: [PATCH 2/6] tracing: increase size of number of possible events
References: <20090424043005.338878133@goodmis.org> <20090424043111.983189535@goodmis.org>
In-Reply-To: <20090424043111.983189535@goodmis.org>
Content-Type: text/plain; charset=GB2312
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

> From: Steven Rostedt <srostedt@redhat.com>
> 
> With the new event tracing registration, we must increase the number
> of events that can be registered. Currently the type field is only
> one byte, which leaves us only 256 possible events.
> 
> Since we do not save the CPU number in the tracer anymore (it is determined
> by the per cpu ring buffer that is used) we have an extra byte to use.
> 
> This patch increases the size of type from 1 byte (256 events) to
> 2 bytes (65,536 events).
> 

You forgot to change this:

@@ -198,7 +198,7 @@ ftrace_define_fields_##call(void)					\
 	struct ftrace_event_call *event_call = &event_##call;		\
 	int ret;							\
 									\
-	__common_field(int, type);					\
+	__common_field(unsigned short, type);				\

And we can apply the check in 3/6 to __common_field(). :)

> It also adds a WARN_ON_ONCE if we exceed that limit.
> 

WARN_ON_ONCE() may not be sufficient IMO:

We can reach 65536 this way (It took me 15 mins):

while (foo_bar.id < 65536) {
	insmod trace-events-sample.ko
	rmmod trace-events-sample.ko
}

Now next id will be 0! Then do this:

console 1:
 # cat /debug/tracing/trace_pipe

console 2:
while (1) {
	insmod trace-events-sample.ko
	echo foo_bar > /debug/tracing/set_event
	rmmod trace-events-sample.ko
}

I got this immediately:

BUG: unable to handle kernel NULL pointer dereference at 0000006f
IP: [<c05210f3>] bstr_printf+0x2ce/0x302
...
Call Trace:
 [<c0476d12>] ? trace_seq_bprintf+0x28/0x41
 [<c0477569>] ? trace_bprint_print+0x58/0x6c
 [<c0472ffc>] ? print_trace_line+0x2c5/0x2df
 [<c0428a41>] ? sub_preempt_count+0x85/0xa0
 [<c04758cf>] ? tracing_read_pipe+0x118/0x191
 [<c04757b7>] ? tracing_read_pipe+0x0/0x191
 [<c04b09f9>] ? vfs_read+0x8f/0x136
 [<c04b0da3>] ? sys_read+0x40/0x65
 [<c0402a68>] ? sysenter_do_call+0x12/0x36

(We can even get other crashes..)

So I think we should fail the initialization of the event, instead of
WARN_ON().

> @@ -537,6 +537,8 @@ int register_ftrace_event(struct trace_event *event)
>   out:
>  	mutex_unlock(&trace_event_mutex);
>  
> +	WARN_ON_ONCE(next_event_type > FTRACE_MAX_EVENT);
> +

This WARN_ON is triggered when we create an event with id == 65535,
but not 65536.