[PATCH v1] mm: Add configuration to control whether vmpressure notifier is enabled

@@ -388,6 +388,7 @@ a. Enable CONFIG_CGROUPS
 b. Enable CONFIG_MEMCG
 c. Enable CONFIG_MEMCG_SWAP (to use swap extension)
 d. Enable CONFIG_MEMCG_KMEM (to use kmem extension)
+e. Enable CONFIG_MEMCG_VMPRESSURE (to use vmpressure extension)
 
 3.1. Prepare the cgroups (see cgroups.txt, Why are cgroups needed?)
 -------------------------------------------------------------------

@@ -855,7 +856,7 @@ At reading, current status of OOM is shown.
           The number of processes belonging to this cgroup killed by any
           kind of OOM killer.
 
-11. Memory Pressure
+11. Memory Pressure (CONFIG_MEMCG_VMPRESSURE)
 ===================
 
 The pressure level notifications can be used to monitor the memory

@@ -6158,6 +6158,9 @@
 	vmpoff=		[KNL,S390] Perform z/VM CP command after power off.
 			Format: <command>
 
+	vmpressure=	[KNL] Enable or disable vmpressure notifier.
+			Format: <bool>
+
 	vsyscall=	[X86-64]
 			Controls the behavior of vsyscalls (i.e. calls to
 			fixed addresses of 0xffffffffff600x00 from legacy

@@ -257,8 +257,10 @@ struct mem_cgroup {
 
 	unsigned long soft_limit;
 
+#ifdef CONFIG_MEMCG_VMPRESSURE
 	/* vmpressure notifications */
 	struct vmpressure vmpressure;
+#endif
 
 	/*
 	 * Should the OOM killer kill all belonging tasks, had it kill one?

@@ -29,7 +29,8 @@ struct vmpressure {
 
 struct mem_cgroup;
 
-#ifdef CONFIG_MEMCG
+#ifdef CONFIG_MEMCG_VMPRESSURE
+extern bool vmpressure_enable;
 extern void vmpressure(gfp_t gfp, struct mem_cgroup *memcg, bool tree,
 		       unsigned long scanned, unsigned long reclaimed);
 extern void vmpressure_prio(gfp_t gfp, struct mem_cgroup *memcg, int prio);

@@ -48,5 +49,7 @@ static inline void vmpressure(gfp_t gfp, struct mem_cgroup *memcg, bool tree,
 			      unsigned long scanned, unsigned long reclaimed) {}
 static inline void vmpressure_prio(gfp_t gfp, struct mem_cgroup *memcg,
 				   int prio) {}
-#endif /* CONFIG_MEMCG */
+static inline void vmpressure_init(struct vmpressure *vmpr) {}
+static inline void vmpressure_cleanup(struct vmpressure *vmpr) {}
+#endif /* CONFIG_MEMCG_PRESSURE */
 #endif /* __LINUX_VMPRESSURE_H */

@@ -948,6 +948,26 @@ config MEMCG_KMEM
 	depends on MEMCG && !SLOB
 	default y
 
+config MEMCG_VMPRESSURE
+	bool "Memory pressure notifier"
+	depends on MEMCG
+	default y
+	help
+	  Vmpressure extension is used to monitor the memory allocation cost.
+	  The pressure level can be set according to the use scenario and
+	  application will be notified through eventfd when memory pressure is at
+	  the specific level (or higher).
+
+config VMPRESSURE_DEFAULT_DISABLED
+	bool "Require boot parameter to enable memory pressure notifier"
+	depends on MEMCG_VMPRESSURE
+	default n
+	help
+	  If set, memory pressure notifier will be disabled  but can be
+	  enabled through passing vmpressure=1 on the kernel commandline
+	  during boot.
+	  For those who want to use memory pressure notifier flexibly.
+
 config BLK_CGROUP
 	bool "IO controller"
 	depends on BLOCK

@@ -92,7 +92,8 @@ obj-$(CONFIG_MEMTEST)		+= memtest.o
 obj-$(CONFIG_MIGRATION) += migrate.o
 obj-$(CONFIG_TRANSPARENT_HUGEPAGE) += huge_memory.o khugepaged.o
 obj-$(CONFIG_PAGE_COUNTER) += page_counter.o
-obj-$(CONFIG_MEMCG) += memcontrol.o vmpressure.o
+obj-$(CONFIG_MEMCG) += memcontrol.o
+obj-$(CONFIG_MEMCG_VMPRESSURE) += vmpressure.o
 obj-$(CONFIG_MEMCG_SWAP) += swap_cgroup.o
 obj-$(CONFIG_CGROUP_HUGETLB) += hugetlb_cgroup.o
 obj-$(CONFIG_GUP_TEST) += gup_test.o

@@ -248,6 +248,7 @@ static inline bool should_force_charge(void)
 		(current->flags & PF_EXITING);
 }
 
+#ifdef CONFIG_MEMCG_VMPRESSURE
 /* Some nice accessors for the vmpressure. */
 struct vmpressure *memcg_to_vmpressure(struct mem_cgroup *memcg)
 {

@@ -260,6 +261,7 @@ struct mem_cgroup *vmpressure_to_memcg(struct vmpressure *vmpr)
 {
 	return container_of(vmpr, struct mem_cgroup, vmpressure);
 }
+#endif
 
 #ifdef CONFIG_MEMCG_KMEM
 extern spinlock_t css_set_lock;

@@ -4794,9 +4796,12 @@ static ssize_t memcg_write_event_control(struct kernfs_open_file *of,
 	} else if (!strcmp(name, "memory.oom_control")) {
 		event->register_event = mem_cgroup_oom_register_event;
 		event->unregister_event = mem_cgroup_oom_unregister_event;
-	} else if (!strcmp(name, "memory.pressure_level")) {
+#ifdef CONFIG_MEMCG_VMPRESSURE
+	} else if (vmpressure_enable &&
+		   !strcmp(name, "memory.pressure_level")) {
 		event->register_event = vmpressure_register_event;
 		event->unregister_event = vmpressure_unregister_event;
+#endif
 	} else if (!strcmp(name, "memory.memsw.usage_in_bytes")) {
 		event->register_event = memsw_cgroup_usage_register_event;
 		event->unregister_event = memsw_cgroup_usage_unregister_event;

@@ -67,6 +67,19 @@ static const unsigned int vmpressure_level_critical = 95;
  */
 static const unsigned int vmpressure_level_critical_prio = ilog2(100 / 10);
 
+DEFINE_STATIC_KEY_FALSE(vmpressure_disabled);
+#ifdef CONFIG_VMPRESSURE_DEFAULT_DISABLED
+bool vmpressure_enable;
+#else
+bool vmpressure_enable = true;
+#endif
+static int __init setup_vmpressure(char *str)
+{
+	return kstrtobool(str, &vmpressure_enable) == 0;
+}
+__setup("vmpressure=", setup_vmpressure);
+
+
 static struct vmpressure *work_to_vmpressure(struct work_struct *work)
 {
 	return container_of(work, struct vmpressure, work);

@@ -246,6 +259,9 @@ void vmpressure(gfp_t gfp, struct mem_cgroup *memcg, bool tree,
 
 	vmpr = memcg_to_vmpressure(memcg);
 
+	if (static_branch_likely(&vmpressure_disabled))
+		return;
+
 	/*
 	 * Here we only want to account pressure that userland is able to
 	 * help us with. For example, suppose that DMA zone is under

@@ -326,6 +342,8 @@ void vmpressure(gfp_t gfp, struct mem_cgroup *memcg, bool tree,
  */
 void vmpressure_prio(gfp_t gfp, struct mem_cgroup *memcg, int prio)
 {
+	if (static_branch_likely(&vmpressure_disabled))
+		return;
 	/*
 	 * We only use prio for accounting critical level. For more info
 	 * see comment for vmpressure_level_critical_prio variable above.

@@ -450,6 +468,11 @@ void vmpressure_unregister_event(struct mem_cgroup *memcg,
  */
 void vmpressure_init(struct vmpressure *vmpr)
 {
+	if (!vmpressure_enable) {
+		static_branch_enable(&vmpressure_disabled);
+		return;
+	}
+
 	spin_lock_init(&vmpr->sr_lock);
 	mutex_init(&vmpr->events_lock);
 	INIT_LIST_HEAD(&vmpr->events);

@@ -465,6 +488,9 @@ void vmpressure_init(struct vmpressure *vmpr)
  */
 void vmpressure_cleanup(struct vmpressure *vmpr)
 {
+
+	if (static_branch_likely(&vmpressure_disabled))
+		return;
 	/*
 	 * Make sure there is no pending work before eventfd infrastructure
 	 * goes away.