alios: mm: memcontrol: make distance between wmark_low and wmark_high configurable

Introduce a new interface, wmark_scale_factor, which defines the
distance between wmark_high and wmark_low.  The unit is in fractions of
10,000. The default value of 50 means the distance between wmark_high
and wmark_low is 0.5% of the max limit of the cgroup.  The maximum value
is 1000, or 10% of the max limit.

The distance between wmark_low and wmark_high have impact on how hard
memcg kswapd would reclaim.
Reviewed-by: NGavin Shan <shan.gavin@linux.alibaba.com>
Reviewed-by: NXunlei Pang <xlpang@linux.alibaba.com>
Signed-off-by: NYang Shi <yang.shi@linux.alibaba.com>

Documentation/admin-guide/cgroup-v2.rst

@@ -1116,6 +1116,15 @@ PAGE_SIZE multiple when read back.
         Memory usage low water mark, which means the available memory is ok.
         For details, please refer to the above wmark_ratio section.
 
+  memory.wmark_scale_factor
+        A read-write single value file which exists on non-root cgroups.
+        The default is 50.
+
+        The gap between wmark_low and wmark_high.  The unit is in fractions
+        of 10,000. The default value of 50 means the distance between wmark_high
+        and wmark_low is 0.5% of the max limit of the cgroup.  The maximum value
+        is 1000, or 10% of max limit.
+
   memory.oom.group
 	A read-write single value file which exists on non-root
 	cgroups.  The default value is "0".

Documentation/cgroup-v1/memory.txt

@@ -93,6 +93,10 @@ Brief summary of control files.
 					      read-only)
  memory.wmark_high		 # high limit (memory usge high water mark,
 					       read-only)
+ memory.wmark_scale_factor       # the gap between wmark_low and wmark_high,
+				   percentage of max limit, default is 50 or
+                                   0.5% of max limit.  The max value is 1000 or
+                                   10% of max limit.
 
 1. History
 

include/linux/memcontrol.h

@@ -292,6 +292,7 @@ struct mem_cgroup {
 
 	unsigned int		wmark_ratio;
 	struct work_struct	wmark_work;
+	unsigned int		wmark_scale_factor;
 
 #ifdef CONFIG_MEMCG_KMEM
         /* Index in the kmem_cache->memcg_params.memcg_caches array */

mm/memcontrol.c

@@ -2948,10 +2948,19 @@ static void setup_memcg_wmark(struct mem_cgroup *memcg)
 	unsigned long max = memcg->high > memcg->memory.max ?
 			    memcg->memory.max : memcg->high;
 	unsigned int wmark_ratio = memcg->wmark_ratio;
+	unsigned int wmark_scale_factor = memcg->wmark_scale_factor;
+	unsigned long gap;
 
 	if (wmark_ratio) {
 		high_wmark = (max * wmark_ratio) / 100;
-		low_wmark = high_wmark - (high_wmark >> 8);
+
+		/*
+		 * Set the memcg watermark distance according to the
+		 * scale factor in proportion to max limit.
+		 */
+		gap = mult_frac(max, wmark_scale_factor, 10000);
+
+		low_wmark = high_wmark - gap;
 
 		page_counter_set_wmark_low(&memcg->memory, low_wmark);
 		page_counter_set_wmark_high(&memcg->memory, high_wmark);
@@ -3791,6 +3800,42 @@ static ssize_t memory_wmark_ratio_write(struct kernfs_open_file *of,
 	return nbytes;
 }
 
+static int memory_wmark_scale_factor_show(struct seq_file *m, void *v)
+{
+	struct mem_cgroup *memcg = mem_cgroup_from_css(seq_css(m));
+	unsigned int wmark_scale_factor;
+
+	wmark_scale_factor = READ_ONCE(memcg->wmark_scale_factor);
+
+	seq_printf(m, "%d\n", wmark_scale_factor);
+
+	return 0;
+}
+
+static ssize_t memory_wmark_scale_factor_write(struct kernfs_open_file *of,
+				char *buf, size_t nbytes, loff_t off)
+{
+	struct mem_cgroup *memcg = mem_cgroup_from_css(of_css(of));
+	int ret, wmark_scale_factor;
+
+	buf = strstrip(buf);
+	if (!buf)
+		return -EINVAL;
+
+	ret = kstrtouint(buf, 0, &wmark_scale_factor);
+	if (ret)
+		return ret;
+
+	if (wmark_scale_factor > 1000 || wmark_scale_factor < 1)
+		return -EINVAL;
+
+	xchg(&memcg->wmark_scale_factor, wmark_scale_factor);
+
+	setup_memcg_wmark(memcg);
+
+	return nbytes;
+}
+
 static void __mem_cgroup_threshold(struct mem_cgroup *memcg, bool swap)
 {
 	struct mem_cgroup_threshold_ary *t;
@@ -4516,6 +4561,12 @@ static struct cftype mem_cgroup_legacy_files[] = {
 		.private = MEMFILE_PRIVATE(_MEM, WMARK_LOW_LIMIT),
 		.read_u64 = mem_cgroup_read_u64,
 	},
+	{
+		.name = "wmark_scale_factor",
+		.flags = CFTYPE_NOT_ON_ROOT,
+		.seq_show = memory_wmark_scale_factor_show,
+		.write = memory_wmark_scale_factor_write,
+	},
 	{
 		.name = "force_empty",
 		.write = mem_cgroup_force_empty_write,
@@ -4817,6 +4868,9 @@ mem_cgroup_css_alloc(struct cgroup_subsys_state *parent_css)
 		memcg->swappiness = mem_cgroup_swappiness(parent);
 		memcg->oom_kill_disable = parent->oom_kill_disable;
 		memcg->wmark_ratio = parent->wmark_ratio;
+		/* Default gap is 0.5% max limit */
+		memcg->wmark_scale_factor = parent->wmark_scale_factor ?
+					    : 50;
 	}
 	if (parent && parent->use_hierarchy) {
 		memcg->use_hierarchy = true;
@@ -6066,6 +6120,12 @@ static struct cftype memory_files[] = {
 		.flags = CFTYPE_NOT_ON_ROOT,
 		.seq_show = memory_wmark_low_show,
 	},
+	{
+		.name = "wmark_scale_factor",
+		.flags = CFTYPE_NOT_ON_ROOT,
+		.seq_show = memory_wmark_scale_factor_show,
+		.write = memory_wmark_scale_factor_write,
+	},
 	{
 		.name = "events",
 		.flags = CFTYPE_NOT_ON_ROOT,