Re: [PATCH] ftrace: add tracing_thresh support to function_graphtracer (v3)

[Tim Bird]

On 02/25/2010 06:46 PM, Frederic Weisbecker wrote:
> On Thu, Feb 25, 2010 at 03:36:43PM -0800, Tim Bird wrote:
>> Add support for tracing_thresh to the function_graph tracer.  This
>> version of this feature isolates the checks into new entry and
>> return functions, to avoid adding more conditional code into the
>> main function_graph paths.
>>
>> Also, add support for specifying tracing_thresh on the kernel
>> command line.  When used like so: "tracing_thresh=200 ftrace=function_graph"
>> this can be used to analyse system startup.  It is important to disable
>> tracing soon after boot, in order to avoid losing the trace data.
>>
>> Note: the elimination of 'notrace' in the definition of '__init'
>> may be controversial.  This can be removed, or made conditional,
>> if it's a bit too scary, but it worked OK for me.  Tracing during
>> kernel startup still works, just with no visibility of routines
>> declared __init.
> 
> 
> __init functions are notrace otherwise ftrace would hot patch
> callsites of function that have disappeared.
> 
> That said, tracing __init functions is indeed interesting.
> Perhaps we can handle that by removing the __init functions
> from ftrace records each time we release init pages.

Thanks.  Given what Steven said about __init routines,
I'll have to think some more about this.  I'm OK with
just removing that piece of the patch for now.
> 
>> +static int __init set_tracing_thresh(char *str)
>> +{
>> +	unsigned long threshhold;
>> +	int ret;
>> +
>> +	if (!str)
>> +		return 0;
>> +	ret = strict_strtoul(str, 0, &threshhold);
>> +	if (ret < 0)
>> +		return 0;
>> +	tracing_thresh = threshhold * 1000;
>> +	return 1;
>> +}
>> +__setup("tracing_thresh=", set_tracing_thresh);
> 
> Looks like setting this, while the function graph tracer (normal
> mode) is running, will have no effect. That said it's perfectly
> fine as it would be pointless to change this value in the middle
> of the tracing.
> 
The command line is parsed before the tracer is activated, so
tracing_thresh is set when trace_graph_init() is called, which
results in the desired behaviour (that is, if you specify the
tracing_thresh on the command line, you get the duration
filtering on bootup).

...
> Ok, looks good to me.
> 
> Thanks.

Thanks for taking a look at it.
 -- Tim

=============================
Tim Bird
Architecture Group Chair, CE Linux Forum
Senior Staff Engineer, Sony Corporation of America
=============================

--
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/