2023-12-13 01:38:59

by Dan Schatzberg

[permalink] [raw]
Subject: [PATCH V4 1/2] mm: add defines for min/max swappiness

We use the constants 0 and 200 in a few places in the mm code when
referring to the min and max swappiness. This patch adds MIN_SWAPPINESS
and MAX_SWAPPINESS #defines to improve clarity. There are no functional
changes.

Signed-off-by: Dan Schatzberg <[email protected]>
---
include/linux/swap.h | 2 ++
mm/memcontrol.c | 2 +-
mm/vmscan.c | 10 +++++-----
3 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/include/linux/swap.h b/include/linux/swap.h
index f6dd6575b905..e2ab76c25b4a 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -407,6 +407,8 @@ extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order,

#define MEMCG_RECLAIM_MAY_SWAP (1 << 1)
#define MEMCG_RECLAIM_PROACTIVE (1 << 2)
+#define MIN_SWAPPINESS 0
+#define MAX_SWAPPINESS 200
extern unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
unsigned long nr_pages,
gfp_t gfp_mask,
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 1c1061df9cd1..9748a6b88bb8 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -4337,7 +4337,7 @@ static int mem_cgroup_swappiness_write(struct cgroup_subsys_state *css,
{
struct mem_cgroup *memcg = mem_cgroup_from_css(css);

- if (val > 200)
+ if (val > MAX_SWAPPINESS)
return -EINVAL;

if (!mem_cgroup_is_root(memcg))
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 506f8220c5fe..2aa671fe938b 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2403,7 +2403,7 @@ static void get_scan_count(struct lruvec *lruvec, struct scan_control *sc,
ap = swappiness * (total_cost + 1);
ap /= anon_cost + 1;

- fp = (200 - swappiness) * (total_cost + 1);
+ fp = (MAX_SWAPPINESS - swappiness) * (total_cost + 1);
fp /= file_cost + 1;

fraction[0] = ap;
@@ -4400,7 +4400,7 @@ static int get_type_to_scan(struct lruvec *lruvec, int swappiness, int *tier_idx
{
int type, tier;
struct ctrl_pos sp, pv;
- int gain[ANON_AND_FILE] = { swappiness, 200 - swappiness };
+ int gain[ANON_AND_FILE] = { swappiness, MAX_SWAPPINESS - swappiness };

/*
* Compare the first tier of anon with that of file to determine which
@@ -4444,7 +4444,7 @@ static int isolate_folios(struct lruvec *lruvec, struct scan_control *sc, int sw
type = LRU_GEN_ANON;
else if (swappiness == 1)
type = LRU_GEN_FILE;
- else if (swappiness == 200)
+ else if (swappiness == MAX_SWAPPINESS)
type = LRU_GEN_ANON;
else
type = get_type_to_scan(lruvec, swappiness, &tier);
@@ -5368,9 +5368,9 @@ static int run_cmd(char cmd, int memcg_id, int nid, unsigned long seq,

lruvec = get_lruvec(memcg, nid);

- if (swappiness < 0)
+ if (swappiness < MIN_SWAPPINESS)
swappiness = get_swappiness(lruvec, sc);
- else if (swappiness > 200)
+ else if (swappiness > MAX_SWAPPINESS)
goto done;

switch (cmd) {
--
2.34.1


2023-12-13 01:58:55

by Huan Yang

[permalink] [raw]
Subject: Re: [PATCH V4 1/2] mm: add defines for min/max swappiness


在 2023/12/13 9:38, Dan Schatzberg 写道:
> [????????? [email protected] ????????? https://aka.ms/LearnAboutSenderIdentification,????????????]
>
> We use the constants 0 and 200 in a few places in the mm code when
> referring to the min and max swappiness. This patch adds MIN_SWAPPINESS
> and MAX_SWAPPINESS #defines to improve clarity. There are no functional
> changes.
>
> Signed-off-by: Dan Schatzberg <[email protected]>
> ---
> include/linux/swap.h | 2 ++
> mm/memcontrol.c | 2 +-
> mm/vmscan.c | 10 +++++-----
> 3 files changed, 8 insertions(+), 6 deletions(-)
>
> diff --git a/include/linux/swap.h b/include/linux/swap.h
> index f6dd6575b905..e2ab76c25b4a 100644
> --- a/include/linux/swap.h
> +++ b/include/linux/swap.h
> @@ -407,6 +407,8 @@ extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
>
> #define MEMCG_RECLAIM_MAY_SWAP (1 << 1)
> #define MEMCG_RECLAIM_PROACTIVE (1 << 2)
> +#define MIN_SWAPPINESS 0
> +#define MAX_SWAPPINESS 200

Do MAX_SWAPPINESS apply for all swapppiness? If so, maybe better change
swappiness sysctl define:
```
sysctl.c:

{
        .procname    = "swappiness",
        .data        = &vm_swappiness,
        .maxlen        = sizeof(vm_swappiness),
        .mode        = 0644,
        .proc_handler    = proc_dointvec_minmax,
        .extra1        = SYSCTL_ZERO,
        .extra2        = SYSCTL_TWO_HUNDRED,
    },

```

Here hard code swappiness in [0, 200], and now add a new define.

And many other code hard reference 200 into max value of swappiness, like:

```
memcontrol.c:
static int mem_cgroup_swappiness_write(struct cgroup_subsys_state *css,
                       struct cftype *cft, u64 val)
{
    struct mem_cgroup *memcg = mem_cgroup_from_css(css);

    if (val > 200)
        return -EINVAL;

    if (!mem_cgroup_is_root(memcg))
        memcg->swappiness = val;
    else
        vm_swappiness = val;

    return 0;
}
vmscan.c:

/*
 * From 0 .. 200.  Higher means more swappy.
 */
int vm_swappiness = 60;

```


> extern unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
> unsigned long nr_pages,
> gfp_t gfp_mask,
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 1c1061df9cd1..9748a6b88bb8 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -4337,7 +4337,7 @@ static int mem_cgroup_swappiness_write(struct cgroup_subsys_state *css,
> {
> struct mem_cgroup *memcg = mem_cgroup_from_css(css);
>
> - if (val > 200)
> + if (val > MAX_SWAPPINESS)
> return -EINVAL;
>
> if (!mem_cgroup_is_root(memcg))
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index 506f8220c5fe..2aa671fe938b 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -2403,7 +2403,7 @@ static void get_scan_count(struct lruvec *lruvec, struct scan_control *sc,
> ap = swappiness * (total_cost + 1);
> ap /= anon_cost + 1;
>
> - fp = (200 - swappiness) * (total_cost + 1);
> + fp = (MAX_SWAPPINESS - swappiness) * (total_cost + 1);
> fp /= file_cost + 1;
>
> fraction[0] = ap;
> @@ -4400,7 +4400,7 @@ static int get_type_to_scan(struct lruvec *lruvec, int swappiness, int *tier_idx
> {
> int type, tier;
> struct ctrl_pos sp, pv;
> - int gain[ANON_AND_FILE] = { swappiness, 200 - swappiness };
> + int gain[ANON_AND_FILE] = { swappiness, MAX_SWAPPINESS - swappiness };
>
> /*
> * Compare the first tier of anon with that of file to determine which
> @@ -4444,7 +4444,7 @@ static int isolate_folios(struct lruvec *lruvec, struct scan_control *sc, int sw
> type = LRU_GEN_ANON;
> else if (swappiness == 1)
> type = LRU_GEN_FILE;
> - else if (swappiness == 200)
> + else if (swappiness == MAX_SWAPPINESS)
> type = LRU_GEN_ANON;
> else
> type = get_type_to_scan(lruvec, swappiness, &tier);
> @@ -5368,9 +5368,9 @@ static int run_cmd(char cmd, int memcg_id, int nid, unsigned long seq,
>
> lruvec = get_lruvec(memcg, nid);
>
> - if (swappiness < 0)
> + if (swappiness < MIN_SWAPPINESS)
> swappiness = get_swappiness(lruvec, sc);
> - else if (swappiness > 200)
> + else if (swappiness > MAX_SWAPPINESS)
> goto done;
>
> switch (cmd) {
> --
> 2.34.1
>

2023-12-13 16:42:03

by Dan Schatzberg

[permalink] [raw]
Subject: Re: [PATCH V4 1/2] mm: add defines for min/max swappiness

On Wed, Dec 13, 2023 at 09:58:26AM +0800, Huan Yang wrote:
>
> 在 2023/12/13 9:38, Dan Schatzberg 写道:
> > [????????? [email protected] ????????? https://aka.ms/LearnAboutSenderIdentification,????????????]
> >
> > We use the constants 0 and 200 in a few places in the mm code when
> > referring to the min and max swappiness. This patch adds MIN_SWAPPINESS
> > and MAX_SWAPPINESS #defines to improve clarity. There are no functional
> > changes.
> >
> > Signed-off-by: Dan Schatzberg <[email protected]>
> > ---
> > include/linux/swap.h | 2 ++
> > mm/memcontrol.c | 2 +-
> > mm/vmscan.c | 10 +++++-----
> > 3 files changed, 8 insertions(+), 6 deletions(-)
> >
> > diff --git a/include/linux/swap.h b/include/linux/swap.h
> > index f6dd6575b905..e2ab76c25b4a 100644
> > --- a/include/linux/swap.h
> > +++ b/include/linux/swap.h
> > @@ -407,6 +407,8 @@ extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
> >
> > #define MEMCG_RECLAIM_MAY_SWAP (1 << 1)
> > #define MEMCG_RECLAIM_PROACTIVE (1 << 2)
> > +#define MIN_SWAPPINESS 0
> > +#define MAX_SWAPPINESS 200
>
> Do MAX_SWAPPINESS apply for all swapppiness? If so, maybe better change
> swappiness sysctl define:
> ```
> sysctl.c:
>
> {
>         .procname    = "swappiness",
>         .data        = &vm_swappiness,
>         .maxlen        = sizeof(vm_swappiness),
>         .mode        = 0644,
>         .proc_handler    = proc_dointvec_minmax,
>         .extra1        = SYSCTL_ZERO,
>         .extra2        = SYSCTL_TWO_HUNDRED,
>     },
>
> ```
>
> Here hard code swappiness in [0, 200], and now add a new define.

Yes, MAX_SWAPPINESS is a hard limit. I'm not sure what you're
proposing here - the SYSCTL_ZERO and SYSCTL_TWO_HUNDRED values are a
little different than the defines I added. I think most of the value
is just consistently using the defines in the core mm code.

>
> And many other code hard reference 200 into max value of swappiness, like:
>
> ```
> memcontrol.c:
> static int mem_cgroup_swappiness_write(struct cgroup_subsys_state *css,
>                        struct cftype *cft, u64 val)
> {
>     struct mem_cgroup *memcg = mem_cgroup_from_css(css);
>
>     if (val > 200)
>         return -EINVAL;
>
>     if (!mem_cgroup_is_root(memcg))
>         memcg->swappiness = val;
>     else
>         vm_swappiness = val;
>
>     return 0;
> }

This one is already fixed in my patch.

2023-12-14 02:00:57

by Huan Yang

[permalink] [raw]
Subject: Re: [PATCH V4 1/2] mm: add defines for min/max swappiness


在 2023/12/14 0:41, Dan Schatzberg 写道:
> [Some people who received this message don't often get email from [email protected]. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]
>
> On Wed, Dec 13, 2023 at 09:58:26AM +0800, Huan Yang wrote:
>> 在 2023/12/13 9:38, Dan Schatzberg 写道:
>>> [????????? [email protected] ????????? https://aka.ms/LearnAboutSenderIdentification,????????????]
>>>
>>> We use the constants 0 and 200 in a few places in the mm code when
>>> referring to the min and max swappiness. This patch adds MIN_SWAPPINESS
>>> and MAX_SWAPPINESS #defines to improve clarity. There are no functional
>>> changes.
>>>
>>> Signed-off-by: Dan Schatzberg <[email protected]>
>>> ---
>>> include/linux/swap.h | 2 ++
>>> mm/memcontrol.c | 2 +-
>>> mm/vmscan.c | 10 +++++-----
>>> 3 files changed, 8 insertions(+), 6 deletions(-)
>>>
>>> diff --git a/include/linux/swap.h b/include/linux/swap.h
>>> index f6dd6575b905..e2ab76c25b4a 100644
>>> --- a/include/linux/swap.h
>>> +++ b/include/linux/swap.h
>>> @@ -407,6 +407,8 @@ extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
>>>
>>> #define MEMCG_RECLAIM_MAY_SWAP (1 << 1)
>>> #define MEMCG_RECLAIM_PROACTIVE (1 << 2)
>>> +#define MIN_SWAPPINESS 0
>>> +#define MAX_SWAPPINESS 200
>> Do MAX_SWAPPINESS apply for all swapppiness? If so, maybe better change
>> swappiness sysctl define:
>> ```
>> sysctl.c:
>>
>> {
>> .procname = "swappiness",
>> .data = &vm_swappiness,
>> .maxlen = sizeof(vm_swappiness),
>> .mode = 0644,
>> .proc_handler = proc_dointvec_minmax,
>> .extra1 = SYSCTL_ZERO,
>> .extra2 = SYSCTL_TWO_HUNDRED,
>> },
>>
>> ```
>>
>> Here hard code swappiness in [0, 200], and now add a new define.
> Yes, MAX_SWAPPINESS is a hard limit. I'm not sure what you're
> proposing here - the SYSCTL_ZERO and SYSCTL_TWO_HUNDRED values are a
I mean, MAX_SWAPPINESS and SYSCTL_TWO_HUNDRED  all limit swappiness,
but  are different definitions.

If swappiness change 200 into 400, you need to change here extra2 and your
MAX_SWAPPINESS. It's wierd.
> little different than the defines I added. I think most of the value
> is just consistently using the defines in the core mm code.
>
>> And many other code hard reference 200 into max value of swappiness, like:
>>
>> ```
>> memcontrol.c:
>> static int mem_cgroup_swappiness_write(struct cgroup_subsys_state *css,
>> struct cftype *cft, u64 val)
>> {
>> struct mem_cgroup *memcg = mem_cgroup_from_css(css);
>>
>> if (val > 200)
>> return -EINVAL;
>>
>> if (!mem_cgroup_is_root(memcg))
>> memcg->swappiness = val;
>> else
>> vm_swappiness = val;
>>
>> return 0;
>> }
> This one is already fixed in my patch.