Re: [PATCH] trace: Add a free on close control mechanism for buffer_size_kb

[Steven Rostedt <rostedt@goodmis.org>]

Sorry, my mind has been elsewhere and I forgot about your patches.

On Fri, 2011-04-22 at 15:46 -0700, Vaibhav Nagarnaik wrote:
> The proc file entry buffer_size_kb is used to set the size of tracing
> buffer. The memory to expand the buffer size is kernel memory. Consider
> a use case where tracing is handled by a user space utility, which acts
> as a gate keeper for tracing requests. In an OOM condition, tracing is
> considered a low priority task and if the utility gets killed the ring
> buffer memory cannot be released back to the kernel.
> 
> This patch adds an IOCTL on the buffer_size_kb file to set a boolean.
> When this boolean is enabled, closing buffer_size_kb file will cause
> tracing to stop and free up the ring buffer memory.
> 
> The user space process can then open the buffer_size_kb file to set the
> new buffer size for tracing, enable the boolean through IOCTL and keep
> the file open. Under OOM condition, if the process gets killed, the
> kernel closes the file descriptor for buffer_size_kb. The release
> handler stops the tracing and releases the kernel memory automatically.
> 
> Signed-off-by: Vaibhav Nagarnaik <vnagarnaik@google.com>
> ---
>  include/linux/ftrace.h |    3 +
>  kernel/trace/trace.c   |  135 ++++++++++++++++++++++++++++++++++++------------
>  2 files changed, 105 insertions(+), 33 deletions(-)
> 
> diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
> index ca29e03..30c8a23 100644
> --- a/include/linux/ftrace.h
> +++ b/include/linux/ftrace.h
> @@ -114,6 +114,9 @@ struct ftrace_func_command {
>  					char *params, int enable);
>  };
>  
> +/* enable/disable auto free ring buffer on file close */
> +#define TRACE_RINGBUF_FREE_ON_CLOSE _IOW('t', 0x01, int)
> +
>  #ifdef CONFIG_DYNAMIC_FTRACE
>  
>  int ftrace_arch_code_modify_prepare(void);
> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> index d38c16a..c676f17 100644
> --- a/kernel/trace/trace.c
> +++ b/kernel/trace/trace.c
> @@ -2817,6 +2817,42 @@ static int tracing_resize_ring_buffer(unsigned long size)
>  	return ret;
>  }
>  
> +static ssize_t tracing_buffer_resize_atomic(unsigned long size)

Why call it "atomic"? Because you do the mutex? Anyway I hate the name.
It sounds like it can do it without preemption, which it clearly can't
do.

The rest looks fine to me.

-- Steve

> +{
> +	int cpu, ret = 0;
> +
> +	mutex_lock(&trace_types_lock);
> +
> +	tracing_stop();
> +
> +	/* disable all cpu buffers */
> +	for_each_tracing_cpu(cpu) {
> +		if (global_trace.data[cpu])
> +			atomic_inc(&global_trace.data[cpu]->disabled);
> +		if (max_tr.data[cpu])
> +			atomic_inc(&max_tr.data[cpu]->disabled);
> +	}
> +
> +	if (size != global_trace.entries)
> +		ret = tracing_resize_ring_buffer(size);
> +
> +	/* If check pages failed, return ENOMEM */
> +	if (tracing_disabled)
> +		ret = -ENOMEM;
> +
> +	for_each_tracing_cpu(cpu) {
> +		if (global_trace.data[cpu])
> +			atomic_dec(&global_trace.data[cpu]->disabled);
> +		if (max_tr.data[cpu])
> +			atomic_dec(&max_tr.data[cpu]->disabled);
> +	}
> +
> +	tracing_start();
> +	mutex_unlock(&trace_types_lock);
> +
> +	return ret;
> +}
> +
>