mbox series

[mm-unstable,v4,0/5] Improve mem_cgroup_iter()

Message ID 20240905003058.1859929-1-kinseyho@google.com (mailing list archive)
Headers show
Series Improve mem_cgroup_iter() | expand

Message

Kinsey Ho Sept. 5, 2024, 12:30 a.m. UTC 
Incremental cgroup iteration is being used again [1]. This patchset
improves the reliability of mem_cgroup_iter(). It also improves
simplicity and code readability.

[1] https://lore.kernel.org/20240514202641.2821494-1-hannes@cmpxchg.org/
---
v4: Fixed memcg restart bug reported by hughd in v3, patch 4/5
v3: https://lore.kernel.org/20240827230753.2073580-1-kinseyho@google.com/
Removed __rcu tag from patch 2/5 which removes the need for
rcu_dereference(). This helps readability.
v2: https://lore.kernel.org/20240813204716.842811-1-kinseyho@google.com/
Add patch to clarify css sibling linkage is RCU protected. The
kernel build bot RCU sparse error from v1 has been ignored.
v1: https://lore.kernel.org/20240724190214.1108049-1-kinseyho@google.com/

Kinsey Ho (5):
  cgroup: clarify css sibling linkage is protected by cgroup_mutex or
    RCU
  mm: don't hold css->refcnt during traversal
  mm: increment gen # before restarting traversal
  mm: restart if multiple traversals raced
  mm: clean up mem_cgroup_iter()

 include/linux/cgroup-defs.h |  6 ++-
 include/linux/memcontrol.h  |  4 +-
 kernel/cgroup/cgroup.c      | 16 +++----
 mm/memcontrol.c             | 84 ++++++++++++++++---------------------
 4 files changed, 51 insertions(+), 59 deletions(-)