Cgroups v2 have been around for a while and many users have fully adopted them,
so they never use cgroups v1 features and functionality. Yet they have to "pay"
for the cgroup v1 support anyway:
1) the kernel binary contains an unused cgroup v1 code,
2) some code paths have additional checks which are not needed,
3) some common structures like task_struct and mem_cgroup contain unused
   cgroup v1-specific members.

Cgroup v1's memory controller has a number of features that are not supported
by cgroup v2 and their implementation is pretty much self contained.
Most notably, these features are: soft limit reclaim, oom handling in userspace,
complicated event notification system, charge migration. Cgroup v1-specific code
in memcontrol.c is close to 4k lines in size and it's intervened with generic
and cgroup v2-specific code. It's a burden on developers and maintainers.

This patchset aims to solve these problems by:
1) moving cgroup v1-specific memcg code to the new mm/memcontrol-v1.c file,
2) putting definitions shared by memcontrol.c and memcontrol-v1.c into the
   mm/memcontrol-v1.h header,
3) introducing the CONFIG_MEMCG_V1 config option, turned off by default,
4) making memcontrol-v1.c to compile only if CONFIG_MEMCG_V1 is set.

If CONFIG_MEMCG_V1 is not set, cgroup v1 memory controller is still available
for mounting, however no memory-specific control knobs are present.

This patchset is based against mm-unstable tree (b610f75d19a34),
however a version based on mm-stable can be found here:
  https://github.com/rgushchin/linux/tree/memcontrol_v1.1-stable .

rfc:
  https://lwn.net/Articles/973082/

v1:
  - switched to CONFIG_MEMCG_V1 being off by default based on LSFMMBPF
    discussion [1]
  - switched to memcg1_ prefix (Johannes)
  - many minor fixes
  - dropped patches which put struct memcg members under CONFIG_MEMCG_V1
    (will post as a separate patchset)

[1]: https://lwn.net/Articles/974575/

MAINTAINERS                |    2 +
include/linux/memcontrol.h |  156 ++++---
init/Kconfig               |    9 +
mm/Makefile                |    2 +
mm/memcontrol-v1.c         | 2933 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
mm/memcontrol-v1.h         |  132 ++++++
mm/memcontrol.c            | 4169 +++++++++++++++++++++++++++---------------------------------------------------------------------------------------------------------------------------------------------------
mm/vmscan.c                |   10 +-
8 files changed, 3794 insertions(+), 3619 deletions(-)

Suggested-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Roman Gushchin <roman.gushchin@linux.dev>


Roman Gushchin (14):
  mm: memcg: introduce memcontrol-v1.c
  mm: memcg: move soft limit reclaim code to memcontrol-v1.c
  mm: memcg: rename soft limit reclaim-related functions
  mm: memcg: move charge migration code to memcontrol-v1.c
  mm: memcg: rename charge move-related functions
  mm: memcg: move legacy memcg event code into memcontrol-v1.c
  mm: memcg: rename memcg_check_events()
  mm: memcg: move cgroup v1 oom handling code into memcontrol-v1.c
  mm: memcg: rename memcg_oom_recover()
  mm: memcg: move cgroup v1 interface files to memcontrol-v1.c
  mm: memcg: make memcg1_update_tree() static
  mm: memcg: group cgroup v1 memcg related declarations
  mm: memcg: put cgroup v1-related members of task_struct under config
    option
  MAINTAINERS: add mm/memcontrol-v1.c/h to the list of maintained files

 MAINTAINERS                |    2 +
 include/linux/memcontrol.h |  156 +-
 init/Kconfig               |    9 +
 mm/Makefile                |    2 +
 mm/memcontrol-v1.c         | 2933 +++++++++++++++++++++++++
 mm/memcontrol-v1.h         |  132 ++
 mm/memcontrol.c            | 4141 ++++++------------------------------
 mm/vmscan.c                |   10 +-
 8 files changed, 3780 insertions(+), 3605 deletions(-)
 create mode 100644 mm/memcontrol-v1.c
 create mode 100644 mm/memcontrol-v1.h

-- 
2.45.1

Johannes, Michal,

friendly ping. Do you have any concerns regarding this patchset?

There is a number of things which can/need to be improved/polished on top,
but I'd really like to land this one first, because rebasing of
such a massive patchset is time-consuming and might also introduce
some silly bugs.

Thanks

On Tue, May 28, 2024 at 01:20:52PM -0700, Roman Gushchin wrote:
> Cgroups v2 have been around for a while and many users have fully adopted them,
> so they never use cgroups v1 features and functionality. Yet they have to "pay"
> for the cgroup v1 support anyway:
> 1) the kernel binary contains an unused cgroup v1 code,
> 2) some code paths have additional checks which are not needed,
> 3) some common structures like task_struct and mem_cgroup contain unused
>    cgroup v1-specific members.
> 
> Cgroup v1's memory controller has a number of features that are not supported
> by cgroup v2 and their implementation is pretty much self contained.
> Most notably, these features are: soft limit reclaim, oom handling in userspace,
> complicated event notification system, charge migration. Cgroup v1-specific code
> in memcontrol.c is close to 4k lines in size and it's intervened with generic
> and cgroup v2-specific code. It's a burden on developers and maintainers.
> 
> This patchset aims to solve these problems by:
> 1) moving cgroup v1-specific memcg code to the new mm/memcontrol-v1.c file,
> 2) putting definitions shared by memcontrol.c and memcontrol-v1.c into the
>    mm/memcontrol-v1.h header,
> 3) introducing the CONFIG_MEMCG_V1 config option, turned off by default,
> 4) making memcontrol-v1.c to compile only if CONFIG_MEMCG_V1 is set.
> 
> If CONFIG_MEMCG_V1 is not set, cgroup v1 memory controller is still available
> for mounting, however no memory-specific control knobs are present.
> 
> This patchset is based against mm-unstable tree (b610f75d19a34),
> however a version based on mm-stable can be found here:
>   https://github.com/rgushchin/linux/tree/memcontrol_v1.1-stable .
> 
> rfc:
>   https://lwn.net/Articles/973082/
> 
> v1:
>   - switched to CONFIG_MEMCG_V1 being off by default based on LSFMMBPF
>     discussion [1]
>   - switched to memcg1_ prefix (Johannes)
>   - many minor fixes
>   - dropped patches which put struct memcg members under CONFIG_MEMCG_V1
>     (will post as a separate patchset)
> 
> [1]: https://lwn.net/Articles/974575/

Hi Andrew, can you please add this series to the mm-unstable? Roman and I
plan to add more stuff based on this series, so it would be good to get
some exposure for this series.

Thanks,
Shakeel

On Tue, May 28, 2024 at 01:20:52PM GMT, Roman Gushchin wrote:
> Cgroups v2 have been around for a while and many users have fully adopted them,
> so they never use cgroups v1 features and functionality. Yet they have to "pay"
> for the cgroup v1 support anyway:
> 1) the kernel binary contains an unused cgroup v1 code,
> 2) some code paths have additional checks which are not needed,
> 3) some common structures like task_struct and mem_cgroup contain unused
>    cgroup v1-specific members.
> 
> Cgroup v1's memory controller has a number of features that are not supported
> by cgroup v2 and their implementation is pretty much self contained.
> Most notably, these features are: soft limit reclaim, oom handling in userspace,
> complicated event notification system, charge migration. Cgroup v1-specific code
> in memcontrol.c is close to 4k lines in size and it's intervened with generic
> and cgroup v2-specific code. It's a burden on developers and maintainers.
> 
> This patchset aims to solve these problems by:
> 1) moving cgroup v1-specific memcg code to the new mm/memcontrol-v1.c file,
> 2) putting definitions shared by memcontrol.c and memcontrol-v1.c into the
>    mm/memcontrol-v1.h header,
> 3) introducing the CONFIG_MEMCG_V1 config option, turned off by default,
> 4) making memcontrol-v1.c to compile only if CONFIG_MEMCG_V1 is set.
> 
> If CONFIG_MEMCG_V1 is not set, cgroup v1 memory controller is still available
> for mounting, however no memory-specific control knobs are present.
> 
> This patchset is based against mm-unstable tree (b610f75d19a34),
> however a version based on mm-stable can be found here:
>   https://github.com/rgushchin/linux/tree/memcontrol_v1.1-stable .
> 
> rfc:
>   https://lwn.net/Articles/973082/
> 
> v1:
>   - switched to CONFIG_MEMCG_V1 being off by default based on LSFMMBPF
>     discussion [1]
>   - switched to memcg1_ prefix (Johannes)
>   - many minor fixes
>   - dropped patches which put struct memcg members under CONFIG_MEMCG_V1
>     (will post as a separate patchset)
> 
> [1]: https://lwn.net/Articles/974575/
> 
> MAINTAINERS                |    2 +
> include/linux/memcontrol.h |  156 ++++---
> init/Kconfig               |    9 +
> mm/Makefile                |    2 +
> mm/memcontrol-v1.c         | 2933 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> mm/memcontrol-v1.h         |  132 ++++++
> mm/memcontrol.c            | 4169 +++++++++++++++++++++++++++---------------------------------------------------------------------------------------------------------------------------------------------------
> mm/vmscan.c                |   10 +-
> 8 files changed, 3794 insertions(+), 3619 deletions(-)
> 
> Suggested-by: Matthew Wilcox (Oracle) <willy@infradead.org>
> Signed-off-by: Roman Gushchin <roman.gushchin@linux.dev>
> 
> 
> Roman Gushchin (14):
>   mm: memcg: introduce memcontrol-v1.c
>   mm: memcg: move soft limit reclaim code to memcontrol-v1.c
>   mm: memcg: rename soft limit reclaim-related functions
>   mm: memcg: move charge migration code to memcontrol-v1.c
>   mm: memcg: rename charge move-related functions
>   mm: memcg: move legacy memcg event code into memcontrol-v1.c
>   mm: memcg: rename memcg_check_events()
>   mm: memcg: move cgroup v1 oom handling code into memcontrol-v1.c
>   mm: memcg: rename memcg_oom_recover()
>   mm: memcg: move cgroup v1 interface files to memcontrol-v1.c
>   mm: memcg: make memcg1_update_tree() static
>   mm: memcg: group cgroup v1 memcg related declarations
>   mm: memcg: put cgroup v1-related members of task_struct under config
>     option
>   MAINTAINERS: add mm/memcontrol-v1.c/h to the list of maintained files
> 
>  MAINTAINERS                |    2 +
>  include/linux/memcontrol.h |  156 +-
>  init/Kconfig               |    9 +
>  mm/Makefile                |    2 +
>  mm/memcontrol-v1.c         | 2933 +++++++++++++++++++++++++
>  mm/memcontrol-v1.h         |  132 ++
>  mm/memcontrol.c            | 4141 ++++++------------------------------
>  mm/vmscan.c                |   10 +-
>  8 files changed, 3780 insertions(+), 3605 deletions(-)
>  create mode 100644 mm/memcontrol-v1.c
>  create mode 100644 mm/memcontrol-v1.h
> 
> -- 
> 2.45.1
>

On Fri, Jun 07, 2024 at 08:29:20AM -0700, Shakeel Butt wrote:
> Hi Andrew, can you please add this series to the mm-unstable? Roman and I
> plan to add more stuff based on this series, so it would be good to get
> some exposure for this series.

I've a v2 with 2 minor fixes (compilation issue with !CONFIG_MMU and a small
comments fix). I can rebase it on top of the current mm-unstable and post, please,
let me know when it's a good moment to do it.

Thanks!

On Fri, Jun 07, 2024 at 09:13:46AM GMT, Roman Gushchin wrote:
> On Fri, Jun 07, 2024 at 08:29:20AM -0700, Shakeel Butt wrote:
> > Hi Andrew, can you please add this series to the mm-unstable? Roman and I
> > plan to add more stuff based on this series, so it would be good to get
> > some exposure for this series.
> 
> I've a v2 with 2 minor fixes (compilation issue with !CONFIG_MMU and a small
> comments fix). I can rebase it on top of the current mm-unstable and post, please,
> let me know when it's a good moment to do it.
> 

Please post the v2. I think we are already late for the next window but
let's request Andrew to add this to mm-unstable for the follow up planned
work. 

Shakeel