This commit removes two #ifdef directives from include/linux/notifier.h
by causing SRCU Tiny to provide a dummy srcu_usage structure and a dummy
__SRCU_USAGE_INIT() macro.

Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: atthias Brugger <matthias.bgg@gmail.com>
Cc: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com>
Cc: "Michał Mirosław" <mirq-linux@rere.qmqm.pl>
Cc: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Cc: Sachin Sant <sachinp@linux.ibm.com>
Cc: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
 include/linux/notifier.h | 11 -----------
 include/linux/srcutiny.h |  7 +++++++
 2 files changed, 7 insertions(+), 11 deletions(-)

diff --git a/include/linux/notifier.h b/include/linux/notifier.h
index 86544707236a..45702bdcbceb 100644
--- a/include/linux/notifier.h
+++ b/include/linux/notifier.h
@@ -73,9 +73,7 @@ struct raw_notifier_head {
 
 struct srcu_notifier_head {
 	struct mutex mutex;
-#ifdef CONFIG_TREE_SRCU
 	struct srcu_usage srcuu;
-#endif
 	struct srcu_struct srcu;
 	struct notifier_block __rcu *head;
 };
@@ -106,7 +104,6 @@ extern void srcu_init_notifier_head(struct srcu_notifier_head *nh);
 #define RAW_NOTIFIER_INIT(name)	{				\
 		.head = NULL }
 
-#ifdef CONFIG_TREE_SRCU
 #define SRCU_NOTIFIER_INIT(name, pcpu)				\
 	{							\
 		.mutex = __MUTEX_INITIALIZER(name.mutex),	\
@@ -114,14 +111,6 @@ extern void srcu_init_notifier_head(struct srcu_notifier_head *nh);
 		.srcuu = __SRCU_USAGE_INIT(name.srcuu),		\
 		.srcu = __SRCU_STRUCT_INIT(name.srcu, name.srcuu, pcpu), \
 	}
-#else
-#define SRCU_NOTIFIER_INIT(name, pcpu)				\
-	{							\
-		.mutex = __MUTEX_INITIALIZER(name.mutex),	\
-		.head = NULL,					\
-		.srcu = __SRCU_STRUCT_INIT(name.srcu, name.srcuu, pcpu), \
-	}
-#endif
 
 #define ATOMIC_NOTIFIER_HEAD(name)				\
 	struct atomic_notifier_head name =			\
diff --git a/include/linux/srcutiny.h b/include/linux/srcutiny.h
index ebd72491af99..06ce65518974 100644
--- a/include/linux/srcutiny.h
+++ b/include/linux/srcutiny.h
@@ -48,6 +48,13 @@ void srcu_drive_gp(struct work_struct *wp);
 #define DEFINE_STATIC_SRCU(name) \
 	static struct srcu_struct name = __SRCU_STRUCT_INIT(name, name, name)
 
+// Dummy structure for srcu_notifier_head.
+struct srcu_usage {
+	char srcuu_dummy;
+};
+
+#define __SRCU_USAGE_INIT(name) { .srcuu_dummy = 0, }
+
 void synchronize_srcu(struct srcu_struct *ssp);
 
 /*
-- 
2.40.1

On Mon, 17 Jul 2023 at 11:03, Paul E. McKenney <paulmck@kernel.org> wrote:
>
> +// Dummy structure for srcu_notifier_head.
> +struct srcu_usage {
> +       char srcuu_dummy;
> +};
> +
> +#define __SRCU_USAGE_INIT(name) { .srcuu_dummy = 0, }

You really should be able to just do

   struct srcu_usage { };
   #define __SRCU_USAGE_INIT(name) { }

which is something we've done for ages for spinlocks in

    include/linux/spinlock_types_up.h

because while we had a gcc bug wrt empty structures, that was ages ago
(ie "gcc-2.x").

See commit a1365647022e ("[PATCH] remove gcc-2 checks") from 2006.

So we've already had these kinds of empty dummy structs for literally
over a decade in active use. Exactly so that you can avoid having to
use #ifdef's etc, and can just always assume you have a spinlock, even
if it doesn't generate any code or any data overhead.

               Linus

On Mon, Jul 17, 2023 at 11:09:56AM -0700, Linus Torvalds wrote:
> On Mon, 17 Jul 2023 at 11:03, Paul E. McKenney <paulmck@kernel.org> wrote:
> >
> > +// Dummy structure for srcu_notifier_head.
> > +struct srcu_usage {
> > +       char srcuu_dummy;
> > +};
> > +
> > +#define __SRCU_USAGE_INIT(name) { .srcuu_dummy = 0, }
> 
> You really should be able to just do
> 
>    struct srcu_usage { };
>    #define __SRCU_USAGE_INIT(name) { }
> 
> which is something we've done for ages for spinlocks in
> 
>     include/linux/spinlock_types_up.h
> 
> because while we had a gcc bug wrt empty structures, that was ages ago
> (ie "gcc-2.x").
> 
> See commit a1365647022e ("[PATCH] remove gcc-2 checks") from 2006.
> 
> So we've already had these kinds of empty dummy structs for literally
> over a decade in active use. Exactly so that you can avoid having to
> use #ifdef's etc, and can just always assume you have a spinlock, even
> if it doesn't generate any code or any data overhead.

Showing my age again, I guess.  ;-)

Thank you for the hint!  I will rework this as you suggest.

							Thanx, Paul