[PATCH 0/2] RCU preempt updates

[PATCH 0/2] RCU preempt updates

[Steven Rostedt]

The following patches have been sitting in the RT tree for some time.
They have been waiting for both RCU preempt and RCU preempt tracing
to get into mainline.

Since those are now in mainline, it is time to send in these patches.

-- Steve

[PATCH 1/2] rcu: just rename call_rcu_bh instead of making it a macro

[Steven Rostedt]

[-- Attachment #1: call_rcu_bh-rename-of-call_rcu.patch --]
[-- Type: text/plain, Size: 1158 bytes --]

Seems that I found a box that has a config that passes call_rcu_bh as a
function pointer (see net/sctp/sm_make_chunk.c), so declaring the
call_rcu_bh has a macro function isn't good enough.

This patch makes it just another name of call_rcu for rcupreempt.

Signed-off-by: Steven Rostedt <srostedt@redhat.org>

---
 include/linux/rcupreempt.h |    8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

Index: linux-compile.git/include/linux/rcupreempt.h
===================================================================
--- linux-compile.git.orig/include/linux/rcupreempt.h	2008-07-30 13:56:28.000000000 -0400
+++ linux-compile.git/include/linux/rcupreempt.h	2008-07-30 13:57:57.000000000 -0400
@@ -57,7 +57,13 @@ static inline void rcu_qsctr_inc(int cpu
 	rdssp->sched_qs++;
 }
 #define rcu_bh_qsctr_inc(cpu)
-#define call_rcu_bh(head, rcu) call_rcu(head, rcu)
+
+/*
+ * Someone might want to pass call_rcu_bh as a function pointer.
+ * So this needs to just be a rename and not a macro function.
+ *  (no parentheses)
+ */
+#define call_rcu_bh	 	call_rcu
 
 /**
  * call_rcu_sched - Queue RCU callback for invocation after sched grace period.

-- 

[PATCH 2/2] rcu: trace fix possible mem-leak

[Steven Rostedt]

[-- Attachment #1: rcu-trace-fix-free.patch --]
[-- Type: text/plain, Size: 1056 bytes --]

In the initialization of the RCU trace module, if
rcupreempt_debugfs_init() fails, we never free the the trace buffer.

This patch frees the trace buffer in case the debugfs fails.

Signed-off-by: Steven Rostedt <srostedt@redhat.com>
---
 kernel/rcupreempt_trace.c |    7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

Index: linux-compile.git/kernel/rcupreempt_trace.c
===================================================================
--- linux-compile.git.orig/kernel/rcupreempt_trace.c	2008-07-15 23:55:47.000000000 -0400
+++ linux-compile.git/kernel/rcupreempt_trace.c	2008-07-30 13:58:41.000000000 -0400
@@ -308,11 +308,16 @@ out:
 
 static int __init rcupreempt_trace_init(void)
 {
+	int ret;
+
 	mutex_init(&rcupreempt_trace_mutex);
 	rcupreempt_trace_buf = kmalloc(RCUPREEMPT_TRACE_BUF_SIZE, GFP_KERNEL);
 	if (!rcupreempt_trace_buf)
 		return 1;
-	return rcupreempt_debugfs_init();
+	ret = rcupreempt_debugfs_init();
+	if (ret)
+		kfree(rcupreempt_trace_buf);
+	return ret;
 }
 
 static void __exit rcupreempt_trace_cleanup(void)

-- 

Re: [PATCH 1/2] rcu: just rename call_rcu_bh instead of making it a macro

[Paul E. McKenney]

On Wed, Jul 30, 2008 at 02:20:54PM -0400, Steven Rostedt wrote:
> Seems that I found a box that has a config that passes call_rcu_bh as a
> function pointer (see net/sctp/sm_make_chunk.c), so declaring the
> call_rcu_bh has a macro function isn't good enough.
> 
> This patch makes it just another name of call_rcu for rcupreempt.

Looks good!

Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>

> Signed-off-by: Steven Rostedt <srostedt@redhat.org>
> 
> ---
>  include/linux/rcupreempt.h |    8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
> 
> Index: linux-compile.git/include/linux/rcupreempt.h
> ===================================================================
> --- linux-compile.git.orig/include/linux/rcupreempt.h	2008-07-30 13:56:28.000000000 -0400
> +++ linux-compile.git/include/linux/rcupreempt.h	2008-07-30 13:57:57.000000000 -0400
> @@ -57,7 +57,13 @@ static inline void rcu_qsctr_inc(int cpu
>  	rdssp->sched_qs++;
>  }
>  #define rcu_bh_qsctr_inc(cpu)
> -#define call_rcu_bh(head, rcu) call_rcu(head, rcu)
> +
> +/*
> + * Someone might want to pass call_rcu_bh as a function pointer.
> + * So this needs to just be a rename and not a macro function.
> + *  (no parentheses)
> + */
> +#define call_rcu_bh	 	call_rcu
> 
>  /**
>   * call_rcu_sched - Queue RCU callback for invocation after sched grace period.
> 
> -- 

Re: [PATCH 2/2] rcu: trace fix possible mem-leak

[Paul E. McKenney]

On Wed, Jul 30, 2008 at 02:20:55PM -0400, Steven Rostedt wrote:
> In the initialization of the RCU trace module, if
> rcupreempt_debugfs_init() fails, we never free the the trace buffer.
> 
> This patch frees the trace buffer in case the debugfs fails.

Looks good as well!

Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>

> Signed-off-by: Steven Rostedt <srostedt@redhat.com>
> ---
>  kernel/rcupreempt_trace.c |    7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> Index: linux-compile.git/kernel/rcupreempt_trace.c
> ===================================================================
> --- linux-compile.git.orig/kernel/rcupreempt_trace.c	2008-07-15 23:55:47.000000000 -0400
> +++ linux-compile.git/kernel/rcupreempt_trace.c	2008-07-30 13:58:41.000000000 -0400
> @@ -308,11 +308,16 @@ out:
> 
>  static int __init rcupreempt_trace_init(void)
>  {
> +	int ret;
> +
>  	mutex_init(&rcupreempt_trace_mutex);
>  	rcupreempt_trace_buf = kmalloc(RCUPREEMPT_TRACE_BUF_SIZE, GFP_KERNEL);
>  	if (!rcupreempt_trace_buf)
>  		return 1;
> -	return rcupreempt_debugfs_init();
> +	ret = rcupreempt_debugfs_init();
> +	if (ret)
> +		kfree(rcupreempt_trace_buf);
> +	return ret;
>  }
> 
>  static void __exit rcupreempt_trace_cleanup(void)
> 
> -- 

Re: [PATCH 1/2] rcu: just rename call_rcu_bh instead of making it a macro

[Ingo Molnar]

* Paul E. McKenney <paulmck@linux.vnet.ibm.com> wrote:

> On Wed, Jul 30, 2008 at 02:20:54PM -0400, Steven Rostedt wrote:
> > Seems that I found a box that has a config that passes call_rcu_bh as a
> > function pointer (see net/sctp/sm_make_chunk.c), so declaring the
> > call_rcu_bh has a macro function isn't good enough.
> > 
> > This patch makes it just another name of call_rcu for rcupreempt.
> 
> Looks good!
> 
> Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>

applied to tip/core/rcu - thanks!

	Ingo

Re: [PATCH 2/2] rcu: trace fix possible mem-leak

[Ingo Molnar]

* Paul E. McKenney <paulmck@linux.vnet.ibm.com> wrote:

> On Wed, Jul 30, 2008 at 02:20:55PM -0400, Steven Rostedt wrote:
> > In the initialization of the RCU trace module, if
> > rcupreempt_debugfs_init() fails, we never free the the trace buffer.
> > 
> > This patch frees the trace buffer in case the debugfs fails.
> 
> Looks good as well!
> 
> Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>

applied to tip/core/rcu - thanks.

	Ingo