2020-09-15 04:57:07

by Vijay Balakrishna

[permalink] [raw]
Subject: [v1] mm: khugepaged: avoid overriding min_free_kbytes set by user

set_recommended_min_free_kbytes need to honor min_free_kbytes set by the
user. Post start-of-day THP enable or memory hotplug operations can
lose user specified min_free_kbytes, in particular when it is higher than
calculated recommended value.

Signed-off-by: Vijay Balakrishna <[email protected]>
Cc: [email protected]
---
mm/khugepaged.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 4f7107476a6f..b4b753ba411a 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -2253,7 +2253,7 @@ static void set_recommended_min_free_kbytes(void)
{
struct zone *zone;
int nr_zones = 0;
- unsigned long recommended_min;
+ int recommended_min;

for_each_populated_zone(zone) {
/*
@@ -2280,12 +2280,12 @@ static void set_recommended_min_free_kbytes(void)

/* don't ever allow to reserve more than 5% of the lowmem */
recommended_min = min(recommended_min,
- (unsigned long) nr_free_buffer_pages() / 20);
+ (int) nr_free_buffer_pages() / 20);
recommended_min <<= (PAGE_SHIFT-10);

- if (recommended_min > min_free_kbytes) {
+ if (recommended_min > user_min_free_kbytes) {
if (user_min_free_kbytes >= 0)
- pr_info("raising min_free_kbytes from %d to %lu to help transparent hugepage allocations\n",
+ pr_info("raising min_free_kbytes from %d to %d to help transparent hugepage allocations\n",
min_free_kbytes, recommended_min);

min_free_kbytes = recommended_min;
--
2.28.0


2020-09-15 18:31:41

by Pasha Tatashin

[permalink] [raw]
Subject: Re: [v1] mm: khugepaged: avoid overriding min_free_kbytes set by user

On Tue, Sep 15, 2020 at 12:55 AM Vijay Balakrishna
<[email protected]> wrote:
>
> set_recommended_min_free_kbytes need to honor min_free_kbytes set by the
> user. Post start-of-day THP enable or memory hotplug operations can
> lose user specified min_free_kbytes, in particular when it is higher than
> calculated recommended value.
>
> Signed-off-by: Vijay Balakrishna <[email protected]>
> Cc: [email protected]
> ---
> mm/khugepaged.c | 8 ++++----
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/mm/khugepaged.c b/mm/khugepaged.c
> index 4f7107476a6f..b4b753ba411a 100644
> --- a/mm/khugepaged.c
> +++ b/mm/khugepaged.c
> @@ -2253,7 +2253,7 @@ static void set_recommended_min_free_kbytes(void)
> {
> struct zone *zone;
> int nr_zones = 0;
> - unsigned long recommended_min;
> + int recommended_min;

Hi Vijay,

Please explain in log the type change. user_min_free_kbytes is of
"int" type and is initialized to -1, so comparing to unsigned leads to
undesired effect.


>
> for_each_populated_zone(zone) {
> /*
> @@ -2280,12 +2280,12 @@ static void set_recommended_min_free_kbytes(void)
>
> /* don't ever allow to reserve more than 5% of the lowmem */
> recommended_min = min(recommended_min,
> - (unsigned long) nr_free_buffer_pages() / 20);
> + (int) nr_free_buffer_pages() / 20);
> recommended_min <<= (PAGE_SHIFT-10);
>
> - if (recommended_min > min_free_kbytes) {
> + if (recommended_min > user_min_free_kbytes) {

Took me a while to understand, but this change is correct. It is the
same logic that is done in init_per_zone_wmark_min().

Please join the two patches you have in one patch series.
Reviewed-by: Pavel Tatashin <[email protected]>

Thank you,
Pasha

2020-09-15 19:38:36

by Vijay Balakrishna

[permalink] [raw]
Subject: Re: [v1] mm: khugepaged: avoid overriding min_free_kbytes set by user

On 9/15/2020 11:27 AM, Pavel Tatashin wrote:
> On Tue, Sep 15, 2020 at 12:55 AM Vijay Balakrishna
> <[email protected]> wrote:
>>
>> set_recommended_min_free_kbytes need to honor min_free_kbytes set by the
>> user. Post start-of-day THP enable or memory hotplug operations can
>> lose user specified min_free_kbytes, in particular when it is higher than
>> calculated recommended value.
>>
>> Signed-off-by: Vijay Balakrishna <[email protected]>
>> Cc: [email protected]
>> ---
>> mm/khugepaged.c | 8 ++++----
>> 1 file changed, 4 insertions(+), 4 deletions(-)
>>
>> diff --git a/mm/khugepaged.c b/mm/khugepaged.c
>> index 4f7107476a6f..b4b753ba411a 100644
>> --- a/mm/khugepaged.c
>> +++ b/mm/khugepaged.c
>> @@ -2253,7 +2253,7 @@ static void set_recommended_min_free_kbytes(void)
>> {
>> struct zone *zone;
>> int nr_zones = 0;
>> - unsigned long recommended_min;
>> + int recommended_min;
>
> Hi Vijay,
>
> Please explain in log the type change. user_min_free_kbytes is of
> "int" type and is initialized to -1, so comparing to unsigned leads to
> undesired effect.

Hi Pasha,

I will update log. I uncovered it when trying to understand why
one-line change to address issue raised by Kirill didn't work!

>
>
>>
>> for_each_populated_zone(zone) {
>> /*
>> @@ -2280,12 +2280,12 @@ static void set_recommended_min_free_kbytes(void)
>>
>> /* don't ever allow to reserve more than 5% of the lowmem */
>> recommended_min = min(recommended_min,
>> - (unsigned long) nr_free_buffer_pages() / 20);
>> + (int) nr_free_buffer_pages() / 20);
>> recommended_min <<= (PAGE_SHIFT-10);
>>
>> - if (recommended_min > min_free_kbytes) {
>> + if (recommended_min > user_min_free_kbytes) {
>
> Took me a while to understand, but this change is correct. It is the
> same logic that is done in init_per_zone_wmark_min().
>
> Please join the two patches you have in one patch series.
> Reviewed-by: Pavel Tatashin <[email protected]>

Will do.

Thanks,
Vijay
>
> Thank you,
> Pasha
>