From: MengEn Sun <mengensun@tencent.com>

when we do not config percpu_pagelist_high_fraction the kernel
compute the pcp high_min/max by itself, which may be not easy
to tell the really high_min/max.

we output the pcp high_min/max to the zoneinfo

Reviewed-by: Jinliang Zheng <alexjlzheng@tencent.com>
Signed-off-by: MengEn Sun <mengensun@tencent.com>
---
 mm/vmstat.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/mm/vmstat.c b/mm/vmstat.c
index b5a4cea423e1..1917c034c045 100644
--- a/mm/vmstat.c
+++ b/mm/vmstat.c
@@ -1791,13 +1791,17 @@ static void zoneinfo_show_print(struct seq_file *m, pg_data_t *pgdat,
 		pcp = per_cpu_ptr(zone->per_cpu_pageset, i);
 		seq_printf(m,
 			   "\n    cpu: %i"
-			   "\n              count: %i"
-			   "\n              high:  %i"
-			   "\n              batch: %i",
+			   "\n              count:    %i"
+			   "\n              high:     %i"
+			   "\n              batch:    %i"
+			   "\n              high_min: %i"
+			   "\n              high_max: %i",
 			   i,
 			   pcp->count,
 			   pcp->high,
-			   pcp->batch);
+			   pcp->batch,
+			   pcp->high_min,
+			   pcp->high_max);
 #ifdef CONFIG_SMP
 		pzstats = per_cpu_ptr(zone->per_cpu_zonestats, i);
 		seq_printf(m, "\n  vm stats threshold: %d",
-- 
2.43.5

On Thu, 10 Oct 2024 20:09:36 +0800 alexjlzheng@gmail.com wrote:

> From: MengEn Sun <mengensun@tencent.com>
> 
> when we do not config percpu_pagelist_high_fraction the kernel
> compute the pcp high_min/max by itself, which may be not easy
> to tell the really high_min/max.
> 
> we output the pcp high_min/max to the zoneinfo
> 

Fair enough.  I tuned the changelog a bit:

: When we do not set percpu_pagelist_high_fraction the kernel will compute
: the pcp high_min/max by itself, which makes it hard to determine the
: current high_min/max values.
: 
: So output the pcp high_min/max values to /proc/zoneinfo.