[PATCH] mm/vmscan.c: drop_slab_node with task's memcg

[PATCH] mm/vmscan.c: drop_slab_node with task's memcg

[Neil Sun]

This patch makes shrink_slab() with task's memcg in drop_slab_node(),
so we can free reclaimable slab objects belongs to memcg /lxc/i-vbe1u8o7
with following command:

cgexec -g memory:/lxc/i-vbe1u8o7 sysctl vm.drop_caches=2

Test with following steps:

root@i-yl0pwrt8:~# free -h
              total        used        free      shared  buff/cache   available
Mem:           62Gi       265Mi        62Gi       1.0Mi       290Mi        61Gi
Swap:          31Gi          0B        31Gi
root@i-yl0pwrt8:~# (cd /tmp && /root/generate_slab_cache)
root@i-yl0pwrt8:~# free -h
              total        used        free      shared  buff/cache   available
Mem:           62Gi       266Mi        60Gi       1.0Mi       2.2Gi        61Gi
Swap:          31Gi          0B        31Gi
root@i-yl0pwrt8:~# cgcreate -g memory:/lxc/i-vbe1u8o7
root@i-yl0pwrt8:~# cgexec -g memory:/lxc/i-vbe1u8o7 /root/generate_slab_cache
root@i-yl0pwrt8:~# free -h
              total        used        free      shared  buff/cache   available
Mem:           62Gi       267Mi        58Gi       1.0Mi       4.1Gi        61Gi
Swap:          31Gi          0B        31Gi
root@i-yl0pwrt8:~# cgexec -g memory:/lxc/i-vbe1u8o7 sysctl vm.drop_caches=2
vm.drop_caches = 2
root@i-yl0pwrt8:~# free -h
              total        used        free      shared  buff/cache   available
Mem:           62Gi       268Mi        60Gi       1.0Mi       2.2Gi        61Gi
Swap:          31Gi          0B        31Gi
root@i-yl0pwrt8:~# sysctl vm.drop_caches=2
vm.drop_caches = 2
root@i-yl0pwrt8:~# free -h
              total        used        free      shared  buff/cache   available
Mem:           62Gi       267Mi        62Gi       1.0Mi       290Mi        61Gi
Swap:          31Gi          0B        31Gi

Signed-off-by: Neil Sun <neilsun@yunify.com>
---
 mm/vmscan.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 562e87cb..81d770a 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -702,7 +702,7 @@ void drop_slab_node(int nid)
 			return;
 
 		freed = 0;
-		memcg = mem_cgroup_iter(NULL, NULL, NULL);
+		memcg = mem_cgroup_from_task(current);
 		do {
 			freed += shrink_slab(GFP_KERNEL, nid, memcg, 0);
 		} while ((memcg = mem_cgroup_iter(NULL, memcg, NULL)) != NULL);
-- 
2.7.4

Re: [PATCH] mm/vmscan.c: drop_slab_node with task's memcg

[kernel test robot]

[-- Attachment #1: Type: text/plain, Size: 2764 bytes --]

Hi Neil,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on v5.12-rc5]
[also build test ERROR on next-20210401]
[cannot apply to hnaz-linux-mm/master]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Neil-Sun/mm-vmscan-c-drop_slab_node-with-task-s-memcg/20210402-184123
base:    a5e13c6df0e41702d2b2c77c8ad41677ebb065b3
config: mips-randconfig-r006-20210402 (attached as .config)
compiler: clang version 13.0.0 (https://github.com/llvm/llvm-project b23a314146956dd29b719ab537608ced736fc036)
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # install mips cross compiling tool for clang build
        # apt-get install binutils-mips-linux-gnu
        # https://github.com/0day-ci/linux/commit/73abc13485ec57de137f60535b5b2495c0bac570
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Neil-Sun/mm-vmscan-c-drop_slab_node-with-task-s-memcg/20210402-184123
        git checkout 73abc13485ec57de137f60535b5b2495c0bac570
        # save the attached .config to linux build tree
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross ARCH=mips 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All error/warnings (new ones prefixed by >>):

>> mm/vmscan.c:705:11: error: implicit declaration of function 'mem_cgroup_from_task' [-Werror,-Wimplicit-function-declaration]
                   memcg = mem_cgroup_from_task(current);
                           ^
>> mm/vmscan.c:705:9: warning: incompatible integer to pointer conversion assigning to 'struct mem_cgroup *' from 'int' [-Wint-conversion]
                   memcg = mem_cgroup_from_task(current);
                         ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   1 warning and 1 error generated.


vim +/mem_cgroup_from_task +705 mm/vmscan.c

   693	
   694	void drop_slab_node(int nid)
   695	{
   696		unsigned long freed;
   697	
   698		do {
   699			struct mem_cgroup *memcg = NULL;
   700	
   701			if (fatal_signal_pending(current))
   702				return;
   703	
   704			freed = 0;
 > 705			memcg = mem_cgroup_from_task(current);
   706			do {
   707				freed += shrink_slab(GFP_KERNEL, nid, memcg, 0);
   708			} while ((memcg = mem_cgroup_iter(NULL, memcg, NULL)) != NULL);
   709		} while (freed > 10);
   710	}
   711	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 20960 bytes --]

Re: [PATCH] mm/vmscan.c: drop_slab_node with task's memcg

[kernel test robot]

[-- Attachment #1: Type: text/plain, Size: 2663 bytes --]

Hi Neil,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on v5.12-rc5]
[also build test ERROR on next-20210401]
[cannot apply to hnaz-linux-mm/master]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Neil-Sun/mm-vmscan-c-drop_slab_node-with-task-s-memcg/20210402-184123
base:    a5e13c6df0e41702d2b2c77c8ad41677ebb065b3
config: arm-randconfig-r035-20210401 (attached as .config)
compiler: arm-linux-gnueabi-gcc (GCC) 9.3.0
reproduce (this is a W=1 build):
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # https://github.com/0day-ci/linux/commit/73abc13485ec57de137f60535b5b2495c0bac570
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Neil-Sun/mm-vmscan-c-drop_slab_node-with-task-s-memcg/20210402-184123
        git checkout 73abc13485ec57de137f60535b5b2495c0bac570
        # save the attached .config to linux build tree
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross ARCH=arm 

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>

All error/warnings (new ones prefixed by >>):

   mm/vmscan.c: In function 'drop_slab_node':
>> mm/vmscan.c:705:11: error: implicit declaration of function 'mem_cgroup_from_task'; did you mean 'mem_cgroup_from_obj'? [-Werror=implicit-function-declaration]
     705 |   memcg = mem_cgroup_from_task(current);
         |           ^~~~~~~~~~~~~~~~~~~~
         |           mem_cgroup_from_obj
>> mm/vmscan.c:705:9: warning: assignment to 'struct mem_cgroup *' from 'int' makes pointer from integer without a cast [-Wint-conversion]
     705 |   memcg = mem_cgroup_from_task(current);
         |         ^
   cc1: some warnings being treated as errors


vim +705 mm/vmscan.c

   693	
   694	void drop_slab_node(int nid)
   695	{
   696		unsigned long freed;
   697	
   698		do {
   699			struct mem_cgroup *memcg = NULL;
   700	
   701			if (fatal_signal_pending(current))
   702				return;
   703	
   704			freed = 0;
 > 705			memcg = mem_cgroup_from_task(current);
   706			do {
   707				freed += shrink_slab(GFP_KERNEL, nid, memcg, 0);
   708			} while ((memcg = mem_cgroup_iter(NULL, memcg, NULL)) != NULL);
   709		} while (freed > 10);
   710	}
   711	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

[-- Attachment #2: .config.gz --]
[-- Type: application/gzip, Size: 21042 bytes --]

Re: [PATCH] mm/vmscan.c: drop_slab_node with task's memcg

[Michal Hocko]

On Fri 02-04-21 18:38:54, Neil Sun wrote:
> This patch makes shrink_slab() with task's memcg in drop_slab_node(),
> so we can free reclaimable slab objects belongs to memcg /lxc/i-vbe1u8o7
> with following command:

You are changing semantic of the existing user interface. This knob has
never been memcg aware and it is supposed to have a global impact. I do
not think we can simply change that without some users being surprised
or even breaking them.

> cgexec -g memory:/lxc/i-vbe1u8o7 sysctl vm.drop_caches=2
> 
> Test with following steps:
> 
> root@i-yl0pwrt8:~# free -h
>               total        used        free      shared  buff/cache   available
> Mem:           62Gi       265Mi        62Gi       1.0Mi       290Mi        61Gi
> Swap:          31Gi          0B        31Gi
> root@i-yl0pwrt8:~# (cd /tmp && /root/generate_slab_cache)
> root@i-yl0pwrt8:~# free -h
>               total        used        free      shared  buff/cache   available
> Mem:           62Gi       266Mi        60Gi       1.0Mi       2.2Gi        61Gi
> Swap:          31Gi          0B        31Gi
> root@i-yl0pwrt8:~# cgcreate -g memory:/lxc/i-vbe1u8o7
> root@i-yl0pwrt8:~# cgexec -g memory:/lxc/i-vbe1u8o7 /root/generate_slab_cache
> root@i-yl0pwrt8:~# free -h
>               total        used        free      shared  buff/cache   available
> Mem:           62Gi       267Mi        58Gi       1.0Mi       4.1Gi        61Gi
> Swap:          31Gi          0B        31Gi
> root@i-yl0pwrt8:~# cgexec -g memory:/lxc/i-vbe1u8o7 sysctl vm.drop_caches=2
> vm.drop_caches = 2
> root@i-yl0pwrt8:~# free -h
>               total        used        free      shared  buff/cache   available
> Mem:           62Gi       268Mi        60Gi       1.0Mi       2.2Gi        61Gi
> Swap:          31Gi          0B        31Gi
> root@i-yl0pwrt8:~# sysctl vm.drop_caches=2
> vm.drop_caches = 2
> root@i-yl0pwrt8:~# free -h
>               total        used        free      shared  buff/cache   available
> Mem:           62Gi       267Mi        62Gi       1.0Mi       290Mi        61Gi
> Swap:          31Gi          0B        31Gi
> 
> Signed-off-by: Neil Sun <neilsun@yunify.com>
> ---
>  mm/vmscan.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index 562e87cb..81d770a 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -702,7 +702,7 @@ void drop_slab_node(int nid)
>  			return;
>  
>  		freed = 0;
> -		memcg = mem_cgroup_iter(NULL, NULL, NULL);
> +		memcg = mem_cgroup_from_task(current);
>  		do {
>  			freed += shrink_slab(GFP_KERNEL, nid, memcg, 0);
>  		} while ((memcg = mem_cgroup_iter(NULL, memcg, NULL)) != NULL);
> -- 
> 2.7.4

-- 
Michal Hocko
SUSE Labs

Re: [PATCH] mm/vmscan.c: drop_slab_node with task's memcg

[Michal Hocko]

On Tue 06-04-21 19:30:22, Neil Sun wrote:
> On 2021/4/6 15:21, Michal Hocko wrote:
> > 
> > You are changing semantic of the existing user interface. This knob has
> > never been memcg aware and it is supposed to have a global impact. I do
> > not think we can simply change that without some users being surprised
> > or even breaking them.
> 
> Yes, do you think add new interface to sysfs is a good way? such as
> /sys/fs/cgroup/memory/lxc/i-vbe1u8o7/memory.kmem.drop_caches

There were other attempts to add a memcg specific alternative to
drop_caches. A lack of a strong usecase has been a reason that no such
attempt has been merged until now. drop_caches is a problematic
interface because it is really coarse and people have learned to (ab)use
it to workaround problem rather than fix them properly.

What is your usecase?
-- 
Michal Hocko
SUSE Labs

Re: [PATCH] mm/vmscan.c: drop_slab_node with task's memcg

[Neil Sun]

On 2021/4/6 19:39, Michal Hocko wrote:
> On Tue 06-04-21 19:30:22, Neil Sun wrote:
>> On 2021/4/6 15:21, Michal Hocko wrote:
>>>
>>> You are changing semantic of the existing user interface. This knob has
>>> never been memcg aware and it is supposed to have a global impact. I do
>>> not think we can simply change that without some users being surprised
>>> or even breaking them.
>>
>> Yes, do you think add new interface to sysfs is a good way? such as
>> /sys/fs/cgroup/memory/lxc/i-vbe1u8o7/memory.kmem.drop_caches
> 
> There were other attempts to add a memcg specific alternative to
> drop_caches. A lack of a strong usecase has been a reason that no such
> attempt has been merged until now. drop_caches is a problematic
> interface because it is really coarse and people have learned to (ab)use
> it to workaround problem rather than fix them properly.
> 
> What is your usecase?
> 

We have some lxc containers running on the server, when mysqld running 
backup jobs in the container, page cache will grow up and eat up all 
unused memory in the container, then some new jobs come, we can see that 
tasks are busy on allocing memory with reclaiming, so we want to drop 
page cache after mysql backup job for individual container, it will 
speed up allocing memory when new jobs come.

This patch only drop slab cache but not page cache, this can be the
first step if people really need this interface.

---
Neil

Re: [PATCH] mm/vmscan.c: drop_slab_node with task's memcg

[Michal Hocko]

On Tue 06-04-21 22:34:02, Neil Sun wrote:
> 
> 
> On 2021/4/6 19:39, Michal Hocko wrote:
> > On Tue 06-04-21 19:30:22, Neil Sun wrote:
> > > On 2021/4/6 15:21, Michal Hocko wrote:
> > > > 
> > > > You are changing semantic of the existing user interface. This knob has
> > > > never been memcg aware and it is supposed to have a global impact. I do
> > > > not think we can simply change that without some users being surprised
> > > > or even breaking them.
> > > 
> > > Yes, do you think add new interface to sysfs is a good way? such as
> > > /sys/fs/cgroup/memory/lxc/i-vbe1u8o7/memory.kmem.drop_caches
> > 
> > There were other attempts to add a memcg specific alternative to
> > drop_caches. A lack of a strong usecase has been a reason that no such
> > attempt has been merged until now. drop_caches is a problematic
> > interface because it is really coarse and people have learned to (ab)use
> > it to workaround problem rather than fix them properly.
> > 
> > What is your usecase?
> > 
> 
> We have some lxc containers running on the server, when mysqld running
> backup jobs in the container, page cache will grow up and eat up all unused
> memory in the container, then some new jobs come, we can see that tasks are
> busy on allocing memory with reclaiming, so we want to drop page cache after
> mysql backup job for individual container, it will speed up allocing memory
> when new jobs come.
> 
> This patch only drop slab cache but not page cache, this can be the
> first step if people really need this interface.

Have you considered using high limit for the pro-active memory reclaim?
It really seems odd to drop a certain category of memory without aging
information we already do have. I do understand the start time overhead
concern but it seems to be a much better approach to drop old objects
rather than hammer a very specific type of memory.
-- 
Michal Hocko
SUSE Labs

Re: [PATCH] mm/vmscan.c: drop_slab_node with task's memcg

[Neil Sun]

On 2021/4/6 22:39, Michal Hocko wrote:
> 
> Have you considered using high limit for the pro-active memory reclaim?

Thanks, Michal, do you mean the procfs interfaces?
We have set vm.vfs_cache_pressure=1000 and so on.
would you please take an example?

--
Neil

Re: [PATCH] mm/vmscan.c: drop_slab_node with task's memcg

[Michal Hocko]

On Tue 06-04-21 23:12:34, Neil Sun wrote:
> 
> 
> On 2021/4/6 22:39, Michal Hocko wrote:
> > 
> > Have you considered using high limit for the pro-active memory reclaim?
> 
> Thanks, Michal, do you mean the procfs interfaces?
> We have set vm.vfs_cache_pressure=1000 and so on.
> would you please take an example?

No, I've meant high memory limit available in the memcg v2 interface:
Documentation/admin-guide/cgroup-v2.rst.
-- 
Michal Hocko
SUSE Labs