2018-11-05 00:53:58

by Vasily Averin

Subject: [PATCH 2/2] mm: avoid unnecessary swap_info_struct allocation

Currently newly allocated swap_info_struct can be quickly freed.
This patch avoid uneccessary high-order page allocation and helps
to decrease the memory pressure.

Signed-off-by: Vasily Averin <[email protected]>
---
mm/swapfile.c | 18 +++++++++++++-----
1 file changed, 13 insertions(+), 5 deletions(-)

diff --git a/mm/swapfile.c b/mm/swapfile.c
index 8688ae65ef58..53ec2f0cdf26 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -2809,14 +2809,17 @@ late_initcall(max_swapfiles_check);

static struct swap_info_struct *alloc_swap_info(void)
{
- struct swap_info_struct *p;
+ struct swap_info_struct *p = NULL;
unsigned int type;
int i;
+ bool force_alloc = false;

- p = kvzalloc(sizeof(*p), GFP_KERNEL);
- if (!p)
- return ERR_PTR(-ENOMEM);
-
+retry:
+ if (force_alloc) {
+ p = kvzalloc(sizeof(*p), GFP_KERNEL);
+ if (!p)
+ return ERR_PTR(-ENOMEM);
+ }
spin_lock(&swap_lock);
for (type = 0; type < nr_swapfiles; type++) {
if (!(swap_info[type]->flags & SWP_USED))
@@ -2828,6 +2831,11 @@ static struct swap_info_struct *alloc_swap_info(void)
return ERR_PTR(-EPERM);
}
if (type >= nr_swapfiles) {
+ if (!force_alloc) {
+ force_alloc = true;
+ spin_unlock(&swap_lock);
+ goto retry;
+ }
p->type = type;
swap_info[type] = p;
/*
--
2.17.1

2018-11-05 00:57:43

by Huang, Ying

Subject: Re: [PATCH 2/2] mm: avoid unnecessary swap_info_struct allocation

Vasily Averin <[email protected]> writes:

> Currently newly allocated swap_info_struct can be quickly freed.
> This patch avoid uneccessary high-order page allocation and helps
> to decrease the memory pressure.

I think swapon/swapoff are rare operations, so it will not increase the
memory pressure much.

Best Regards,
Huang, Ying

> Signed-off-by: Vasily Averin <[email protected]>
> ---
> mm/swapfile.c | 18 +++++++++++++-----
> 1 file changed, 13 insertions(+), 5 deletions(-)
>
> diff --git a/mm/swapfile.c b/mm/swapfile.c
> index 8688ae65ef58..53ec2f0cdf26 100644
> --- a/mm/swapfile.c
> +++ b/mm/swapfile.c
> @@ -2809,14 +2809,17 @@ late_initcall(max_swapfiles_check);
>
> static struct swap_info_struct *alloc_swap_info(void)
> {
> - struct swap_info_struct *p;
> + struct swap_info_struct *p = NULL;
> unsigned int type;
> int i;
> + bool force_alloc = false;
>
> - p = kvzalloc(sizeof(*p), GFP_KERNEL);
> - if (!p)
> - return ERR_PTR(-ENOMEM);
> -
> +retry:
> + if (force_alloc) {
> + p = kvzalloc(sizeof(*p), GFP_KERNEL);
> + if (!p)
> + return ERR_PTR(-ENOMEM);
> + }
> spin_lock(&swap_lock);
> for (type = 0; type < nr_swapfiles; type++) {
> if (!(swap_info[type]->flags & SWP_USED))
> @@ -2828,6 +2831,11 @@ static struct swap_info_struct *alloc_swap_info(void)
> return ERR_PTR(-EPERM);
> }
> if (type >= nr_swapfiles) {
> + if (!force_alloc) {
> + force_alloc = true;
> + spin_unlock(&swap_lock);
> + goto retry;
> + }
> p->type = type;
> swap_info[type] = p;
> /*

2018-11-05 05:20:09

by Vasily Averin

Subject: Re: [PATCH 2/2] mm: avoid unnecessary swap_info_struct allocation

On 11/5/18 3:57 AM, Huang, Ying wrote:
> Vasily Averin <[email protected]> writes:
>
>> Currently newly allocated swap_info_struct can be quickly freed.
>> This patch avoid uneccessary high-order page allocation and helps
>> to decrease the memory pressure.
>
> I think swapon/swapoff are rare operations, so it will not increase the
> memory pressure much.

You are right, typically it should not affect usual nodes.

It's OpenVz-specific usecase.

OpenVz allows hosters to run hundreds of non-trusted containers per node.
Our containers have enabled "virtual swap" functionality,
and container's owners can call sys_swapon without any limits.
Containers can be restarted in any time and we would like to
decrease number of unnecessary high-order memory allocations.

>> Signed-off-by: Vasily Averin <[email protected]>
>> ---
>> mm/swapfile.c | 18 +++++++++++++-----
>> 1 file changed, 13 insertions(+), 5 deletions(-)
>>
>> diff --git a/mm/swapfile.c b/mm/swapfile.c
>> index 8688ae65ef58..53ec2f0cdf26 100644
>> --- a/mm/swapfile.c
>> +++ b/mm/swapfile.c
>> @@ -2809,14 +2809,17 @@ late_initcall(max_swapfiles_check);
>>
>> static struct swap_info_struct *alloc_swap_info(void)
>> {
>> - struct swap_info_struct *p;
>> + struct swap_info_struct *p = NULL;
>> unsigned int type;
>> int i;
>> + bool force_alloc = false;
>>
>> - p = kvzalloc(sizeof(*p), GFP_KERNEL);
>> - if (!p)
>> - return ERR_PTR(-ENOMEM);
>> -
>> +retry:
>> + if (force_alloc) {
>> + p = kvzalloc(sizeof(*p), GFP_KERNEL);
>> + if (!p)
>> + return ERR_PTR(-ENOMEM);
>> + }
>> spin_lock(&swap_lock);
>> for (type = 0; type < nr_swapfiles; type++) {
>> if (!(swap_info[type]->flags & SWP_USED))
>> @@ -2828,6 +2831,11 @@ static struct swap_info_struct *alloc_swap_info(void)
>> return ERR_PTR(-EPERM);
>> }
>> if (type >= nr_swapfiles) {
>> + if (!force_alloc) {
>> + force_alloc = true;
>> + spin_unlock(&swap_lock);
>> + goto retry;
>> + }
>> p->type = type;
>> swap_info[type] = p;
>> /*
>

2018-11-05 10:42:29

by Vasily Averin

Subject: Re: [PATCH 2/2] mm: avoid unnecessary swap_info_struct allocation

I was wrong, openVz blocks sys_swapon/swapoff syscalls inside containers.
Our kernel just emulates /proc/swaps output inside containers,
it is enough for 'swapon' userspace to do not fail and show required info.

So I do not have any special arguments for proposed patch.

On 11/5/18 8:19 AM, Vasily Averin wrote:
> On 11/5/18 3:57 AM, Huang, Ying wrote:
>> Vasily Averin <[email protected]> writes:
>>
>>> Currently newly allocated swap_info_struct can be quickly freed.
>>> This patch avoid uneccessary high-order page allocation and helps
>>> to decrease the memory pressure.
>>
>> I think swapon/swapoff are rare operations, so it will not increase the
>> memory pressure much.
>
> You are right, typically it should not affect usual nodes.
>
> It's OpenVz-specific usecase.
>
> OpenVz allows hosters to run hundreds of non-trusted containers per node.
> Our containers have enabled "virtual swap" functionality,
> and container's owners can call sys_swapon without any limits.
> Containers can be restarted in any time and we would like to
> decrease number of unnecessary high-order memory allocations.
>
>>> Signed-off-by: Vasily Averin <[email protected]>
>>> ---
>>> mm/swapfile.c | 18 +++++++++++++-----
>>> 1 file changed, 13 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/mm/swapfile.c b/mm/swapfile.c
>>> index 8688ae65ef58..53ec2f0cdf26 100644
>>> --- a/mm/swapfile.c
>>> +++ b/mm/swapfile.c
>>> @@ -2809,14 +2809,17 @@ late_initcall(max_swapfiles_check);
>>>
>>> static struct swap_info_struct *alloc_swap_info(void)
>>> {
>>> - struct swap_info_struct *p;
>>> + struct swap_info_struct *p = NULL;
>>> unsigned int type;
>>> int i;
>>> + bool force_alloc = false;
>>>
>>> - p = kvzalloc(sizeof(*p), GFP_KERNEL);
>>> - if (!p)
>>> - return ERR_PTR(-ENOMEM);
>>> -
>>> +retry:
>>> + if (force_alloc) {
>>> + p = kvzalloc(sizeof(*p), GFP_KERNEL);
>>> + if (!p)
>>> + return ERR_PTR(-ENOMEM);
>>> + }
>>> spin_lock(&swap_lock);
>>> for (type = 0; type < nr_swapfiles; type++) {
>>> if (!(swap_info[type]->flags & SWP_USED))
>>> @@ -2828,6 +2831,11 @@ static struct swap_info_struct *alloc_swap_info(void)
>>> return ERR_PTR(-EPERM);
>>> }
>>> if (type >= nr_swapfiles) {
>>> + if (!force_alloc) {
>>> + force_alloc = true;
>>> + spin_unlock(&swap_lock);
>>> + goto retry;
>>> + }
>>> p->type = type;
>>> swap_info[type] = p;
>>> /*
>>