Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1179738 > unrolled thread
| Started by | Michal Hocko <mhocko@kernel.org> |
|---|---|
| First post | 2015-07-08 14:40 +0200 |
| Last post | 2015-07-08 14:40 +0200 |
| Articles | 2 — 1 participant |
Back to article view | Back to linux.kernel
[PATCH 0/8 -v3] memcg cleanups + get rid of mm_struct::owner Michal Hocko <mhocko@kernel.org> - 2015-07-08 14:40 +0200
[PATCH 2/8] memcg: get rid of mem_cgroup_root_css for !CONFIG_MEMCG Michal Hocko <mhocko@kernel.org> - 2015-07-08 14:40 +0200
| From | Michal Hocko <mhocko@kernel.org> |
|---|---|
| Date | 2015-07-08 14:40 +0200 |
| Subject | [PATCH 0/8 -v3] memcg cleanups + get rid of mm_struct::owner |
| Message-ID | <pJWEa-2Dy-5@gated-at.bofh.it> |
Hi,
this is the third version of the patch series. The previous version was posted
here: http://marc.info/?l=linux-mm&m=143290066126282 and the first version
here: http://marc.info/?l=linux-mm&m=143264102317318&w=2.
Johannes has suggested to export struct mem_cgroup because some other
things will become easier. So even though this is not directly related
to the patchset I have posted it together because there are some
dependencies.
The first patch exports struct mem_cgroup along with its dependencies +
some simple accessor functions so they can be inlined in the code.
The second patch is a minor cleanup for cgroup writeback code which hasn't
been present in the previous version.
The third patch simply cleans up some extern declarations in memcontrol.h
because we are not consistent in that regard.
The fourth patch moves some more memcg code to vmscan as it doesn't have any
other users.
The fifth patch is from Tejun and it cleans up mem_cgroup_can_attach a
bit and the follow up patches depend on it.
The sixth patch is preparatory and it's touching sock_update_memcg which
doesn't check for cg_proto == NULL. It doesn't have to do it currently
because mem_cgroup_from_task always return non-NULL but the code is awkward
and this will no longer be true with the follow up patch.
The patch number 7 is the core one and it gets rid of mm_struct::owner
in favor of mm_struct::memcg. The rationale is described in the patch
so I will not repeat it here again.
The last patch gets rid of mem_cgroup_from_task which is not needed anymore.
I have tried to compile test this with my usual configs battery +
randconfigs and nothing blown up. I have also runtime tested migration
between two memcgs as fast as possible. Except for pre-existing races no
regressions seemed to be introduced.
I was worried about the user visible change this patch introduces
but Johannes and Tejun seem to be OK with that and nobody complained
since this was posted last time as an RFC. The changelog states
that a potential usecase is not really worth all the troubles the
implementation exposes to the memcg behavior.
changes since v2
- rebased on top of the current mmotm tree
- clenaup mem_cgroup_root_css
changes since v1
- added Suggested-by Oleg has it was him who kicked me into doing
it
- exec_mmap association was missing [Oleg]
- mem_cgroup_from_task cannot return NULL anymore [Oleg]
- mm_match_cgroup doesn't have to play games and it can use
rcu_dereference after mm_struct is exported [Johannes]
- drop mm_move_memcg as it has only one user in memcontrol.c
so it can be opencoded [Johannes]
- functions to associate and drop mm->memcg association have
to be static inline for !CONFIG_MEMCG [Johannes]
- drop synchronize_rcu nonsense during memcg move [Johannes]
- drop "memcg: Use mc.moving_task as the indication for charge
moving" patch because we can get the target task from css
easily [Johannes]
- rename css_set_memcg renamed to css_inherit_memcg because the
name better suits its usage
- dropped css_get(&from->css) during move because it is pinned
already by the mm. We just have to be careful and do not drop
it before we are really done during migration.
Shortlog says:
Michal Hocko (7):
memcg: export struct mem_cgroup
memcg: get rid of mem_cgroup_root_css for !CONFIG_MEMCG
memcg: get rid of extern for functions in memcontrol.h
memcg, mm: move mem_cgroup_select_victim_node into vmscan
memcg, tcp_kmem: check for cg_proto in sock_update_memcg
memcg: get rid of mm_struct::owner
memcg: get rid of mem_cgroup_from_task
Tejun Heo (1):
memcg: restructure mem_cgroup_can_attach()
And diffstat looks promissing as well.
fs/exec.c | 2 +-
include/linux/memcontrol.h | 439 +++++++++++++++++++++++++++++++++----
include/linux/mm_types.h | 12 +-
include/linux/swap.h | 10 +-
include/net/sock.h | 28 ---
kernel/exit.c | 89 --------
kernel/fork.c | 10 +-
mm/debug.c | 4 +-
mm/memcontrol.c | 527 ++++++---------------------------------------
mm/memory-failure.c | 2 +-
mm/slab_common.c | 2 +-
mm/vmscan.c | 98 ++++++++-
12 files changed, 579 insertions(+), 644 deletions(-)
Thanks
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[toc] | [next] | [standalone]
| From | Michal Hocko <mhocko@kernel.org> |
|---|---|
| Date | 2015-07-08 14:40 +0200 |
| Subject | [PATCH 2/8] memcg: get rid of mem_cgroup_root_css for !CONFIG_MEMCG |
| Message-ID | <pJWNR-2GX-23@gated-at.bofh.it> |
| In reply to | #1179738 |
The only user is cgwb_bdi_init and that one depends on CONFIG_CGROUP_WRITEBACK which in turn depends on CONFIG_MEMCG so it doesn't make much sense to definte an empty stub for !CONFIG_MEMCG. Moreover ERR_PTR(-EINVAL) is ugly and would lead to runtime crashes if used in unguarded code paths. Better fail during compilation. Signed-off-by: Michal Hocko <mhocko@kernel.org> --- include/linux/memcontrol.h | 2 -- 1 file changed, 2 deletions(-) diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index f5a8d0bbef8d..680cefec8c2a 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -496,8 +496,6 @@ void mem_cgroup_split_huge_fixup(struct page *head); #else /* CONFIG_MEMCG */ struct mem_cgroup; -#define mem_cgroup_root_css ((struct cgroup_subsys_state *)ERR_PTR(-EINVAL)) - static inline void mem_cgroup_events(struct mem_cgroup *memcg, enum mem_cgroup_events_index idx, unsigned int nr) -- 2.1.4 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
[toc] | [prev] | [standalone]
Back to top | Article view | linux.kernel
csiph-web