Re: [PATCH][GIT PULL] tracing: add trace_stack interface

[Frederic Weisbecker <fweisbec@gmail.com>]

On Tue, May 19, 2009 at 08:44:50PM -0400, Steven Rostedt wrote:
> 
> Ingo,
> 
> Please pull the latest tip/tracing/ftrace tree, which can be found at:
> 
>   git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
> tip/tracing/ftrace
> 
> 
> Steven Rostedt (1):
>       tracing: add trace_stack interface
> 
> ----
>  include/linux/kernel.h |    2 ++
>  kernel/trace/trace.c   |   15 +++++++++++++++
>  2 files changed, 17 insertions(+), 0 deletions(-)
> ---------------------------
> commit b85d1029ba534376d203c3ac83a2e5a4d6876cb6
> Author: Steven Rostedt <srostedt@redhat.com>
> Date:   Tue May 19 18:54:16 2009 -0400
> 
>     tracing: add trace_stack interface
>     
>     This patch adds the global function
>     
>       void trace_stack(void)
>     
>     This allows a developer to find where a function is called. For example,
>     if you want to know who calls __netif_reschedule, you can add
>     
>      static inline void __netif_reschedule(struct Qdisc *q)
>      {
>      	struct softnet_data *sd;
>      	unsigned long flags;
>     
>     +	trace_stack();
>      	local_irq_save(flags);
>      	sd = &__get_cpu_var(softnet_data);
>      	q->next_sched = sd->output_queue;
>     
>     And see the following in the trace output:
>     
>      <= __ftrace_trace_stack
>      <= trace_stack



I guess the depth should begin here :-)



>      <= __netif_schedule
>      <= dev_watchdog
>      <= run_timer_softirq
>      <= __do_softirq
>      <= call_softirq
>      <= do_softirq
>      <= irq_exit
>     
>     Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
> 
> diff --git a/include/linux/kernel.h b/include/linux/kernel.h
> index 883cd44..347a0f1 100644
> --- a/include/linux/kernel.h
> +++ b/include/linux/kernel.h
> @@ -511,6 +511,7 @@ extern int
>  __ftrace_vprintk(unsigned long ip, const char *fmt, va_list ap);
>  
>  extern void ftrace_dump(void);
> +extern void trace_stack(void);
>  #else
>  static inline void
>  ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) { }
> @@ -531,6 +532,7 @@ ftrace_vprintk(const char *fmt, va_list ap)
>  	return 0;
>  }
>  static inline void ftrace_dump(void) { }
> +static inline void trace_stack(void) { }
>  #endif /* CONFIG_TRACING */
>  
>  /*
> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> index dd40d23..33ab502 100644
> --- a/kernel/trace/trace.c
> +++ b/kernel/trace/trace.c
> @@ -1046,6 +1046,21 @@ void __trace_stack(struct trace_array *tr,
>  	__ftrace_trace_stack(tr, flags, skip, pc);
>  }
>  
> +/**
> + * trace_stack - dump the stack into the trace buffer
> + *
> + * Use this to record the stack trace into the ring buffer. If you
> + * want to know a caller of a function, place a call to trace_stack
> + * and the ftrace trace file will display who the callers were.
> + */
> +void trace_stack(void)
> +{
> +	unsigned long flags;
> +
> +	local_save_flags(flags);
> +	__trace_stack(&global_trace, flags, 0, preempt_count());
> +}



Stupidly I use a very similar patch while working on reiserfs
but I forgot to post it.

Another trick is to use trace_printk() combined with:

echo stacktrace > trace_option


Frederic.


>  static void ftrace_trace_userstack(struct trace_array *tr,
>  				   unsigned long flags, int pc)
>  {
>