[RFC][PATCH] tracepoint: Do not waste memory on mods with no tracepoints

[RFC][PATCH] tracepoint: Do not waste memory on mods with no tracepoints

[Steven Rostedt]

No reason to allocate tp_module structures for modules that have no
tracepoints. This just wastes memory.

Fixes: b75ef8b44b1c "Tracepoint: Dissociate from module mutex"
Cc: stable@vger.kernel.org # 3.2+
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
 kernel/tracepoint.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c
index 29f2654..0d4ef26 100644
--- a/kernel/tracepoint.c
+++ b/kernel/tracepoint.c
@@ -636,6 +636,9 @@ static int tracepoint_module_coming(struct module *mod)
 	struct tp_module *tp_mod, *iter;
 	int ret = 0;
 
+	if (!mod->num_tracepoints)
+		return 0;
+
 	/*
 	 * We skip modules that taint the kernel, especially those with different
 	 * module headers (for forced load), to make sure we don't cause a crash.
@@ -679,6 +682,9 @@ static int tracepoint_module_going(struct module *mod)
 {
 	struct tp_module *pos;
 
+	if (!mod->num_tracepoints)
+		return 0;
+
 	mutex_lock(&tracepoints_mutex);
 	tracepoint_update_probe_range(mod->tracepoints_ptrs,
 		mod->tracepoints_ptrs + mod->num_tracepoints);
-- 
1.8.1.4

Re: [RFC][PATCH] tracepoint: Do not waste memory on mods with no tracepoints

[Mathieu Desnoyers]

----- Original Message -----
> From: "Steven Rostedt" <rostedt@goodmis.org>
> To: "LKML" <linux-kernel@vger.kernel.org>
> Cc: "Ingo Molnar" <mingo@kernel.org>, "Rusty Russell" <rusty@rustcorp.com.au>, "Peter Zijlstra"
> <peterz@infradead.org>, "Mathieu Desnoyers" <mathieu.desnoyers@efficios.com>, "Andrew Morton"
> <akpm@linux-foundation.org>
> Sent: Wednesday, February 26, 2014 11:03:13 AM
> Subject: [RFC][PATCH] tracepoint: Do not waste memory on mods with no tracepoints
> 
> 
> No reason to allocate tp_module structures for modules that have no
> tracepoints. This just wastes memory.

Good catch! Thanks!

Acked-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

> 
> Fixes: b75ef8b44b1c "Tracepoint: Dissociate from module mutex"
> Cc: stable@vger.kernel.org # 3.2+
> Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
> ---
>  kernel/tracepoint.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c
> index 29f2654..0d4ef26 100644
> --- a/kernel/tracepoint.c
> +++ b/kernel/tracepoint.c
> @@ -636,6 +636,9 @@ static int tracepoint_module_coming(struct module *mod)
>  	struct tp_module *tp_mod, *iter;
>  	int ret = 0;
>  
> +	if (!mod->num_tracepoints)
> +		return 0;
> +
>  	/*
>  	 * We skip modules that taint the kernel, especially those with different
>  	 * module headers (for forced load), to make sure we don't cause a crash.
> @@ -679,6 +682,9 @@ static int tracepoint_module_going(struct module *mod)
>  {
>  	struct tp_module *pos;
>  
> +	if (!mod->num_tracepoints)
> +		return 0;
> +
>  	mutex_lock(&tracepoints_mutex);
>  	tracepoint_update_probe_range(mod->tracepoints_ptrs,
>  		mod->tracepoints_ptrs + mod->num_tracepoints);
> --
> 1.8.1.4
> 
> 

-- 
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com

Re: [RFC][PATCH] tracepoint: Do not waste memory on mods with no tracepoints

[Rusty Russell]

Steven Rostedt <rostedt@goodmis.org> writes:
> No reason to allocate tp_module structures for modules that have no
> tracepoints. This just wastes memory.
>
> Fixes: b75ef8b44b1c "Tracepoint: Dissociate from module mutex"
> Cc: stable@vger.kernel.org # 3.2+

Really?  CC:stable?  To save an insignificant amount of memory?

The definition of stable seems to be shifting away from "fixes for
problems with significant effects".  I obviously missed the memo.

Rusty.



> Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
> ---
>  kernel/tracepoint.c | 6 ++++++
>  1 file changed, 6 insertions(+)
>
> diff --git a/kernel/tracepoint.c b/kernel/tracepoint.c
> index 29f2654..0d4ef26 100644
> --- a/kernel/tracepoint.c
> +++ b/kernel/tracepoint.c
> @@ -636,6 +636,9 @@ static int tracepoint_module_coming(struct module *mod)
>  	struct tp_module *tp_mod, *iter;
>  	int ret = 0;
>  
> +	if (!mod->num_tracepoints)
> +		return 0;
> +
>  	/*
>  	 * We skip modules that taint the kernel, especially those with different
>  	 * module headers (for forced load), to make sure we don't cause a crash.
> @@ -679,6 +682,9 @@ static int tracepoint_module_going(struct module *mod)
>  {
>  	struct tp_module *pos;
>  
> +	if (!mod->num_tracepoints)
> +		return 0;
> +
>  	mutex_lock(&tracepoints_mutex);
>  	tracepoint_update_probe_range(mod->tracepoints_ptrs,
>  		mod->tracepoints_ptrs + mod->num_tracepoints);
> -- 
> 1.8.1.4

Re: [RFC][PATCH] tracepoint: Do not waste memory on mods with no tracepoints

[Steven Rostedt]

On Thu, 27 Feb 2014 11:00:35 +1030
Rusty Russell <rusty@rustcorp.com.au> wrote:

> Steven Rostedt <rostedt@goodmis.org> writes:
> > No reason to allocate tp_module structures for modules that have no
> > tracepoints. This just wastes memory.
> >
> > Fixes: b75ef8b44b1c "Tracepoint: Dissociate from module mutex"
> > Cc: stable@vger.kernel.org # 3.2+
> 
> Really?  CC:stable?  To save an insignificant amount of memory?

I agree that this rational alone is not sufficient for stable. But this
is required for another patch that warns when tracepoints are not
loaded by a module due to taints. We don't want to warn on all modules.


> 
> The definition of stable seems to be shifting away from "fixes for
> problems with significant effects".  I obviously missed the memo.

Not telling users that tracepoints were not activated, but acting in
every other way like they are is to me a "significant effect". This
just happened to be something that fixing this depended on.

-- Steve