[PATCH v4 00/15] sched: introduce for_each_process_rculock and for_each_thread_rculock

[Ye Liu]

Hi, Peter

Friendly ping on this series, posted three weeks ago. You previously
requested splitting the combined kernel/ conversion into per-subsystem
patches, which has been done in v4.
No rush — just want to make sure this is on your radar.

Thanks,
Ye Liu

在 2026/9/21 17:20, Ye Liu 写道:
> From: Ye Liu <liuye at kylinos.cn>
>
> Introduce for_each_process_rculock(), for_each_thread_rculock() and
> for_each_process_thread_rculock() macros that combine the existing
> iteration macros with scoped_guard(rcu), so that the RCU read lock
> is automatically acquired before iteration and released when the
> loop exits — including via break, goto, or return.
>
> The rest of the series converts manual rcu_read_lock()/
> rcu_read_unlock() and guard(rcu)() pairs across mm/, kernel/, fs/,
> lib/ and security/ to use the new macros.
>
> Suggested by Michal Hocko for the oom_kill path [2].
>
> Signed-off-by: Ye Liu <liuye at kylinos.cn>
>
> ---
> Changes since v3 [4]:
>   - Fix patch 5 (cpu/hotplug): title and commit message incorrectly
>     said "thread iterator" and "for_each_thread_rculock" while the
>     code converts for_each_process() to for_each_process_rculock().
>
> Note: Patches 4, 11, and 13 received automated reviews from sashiko-bot
> in the v3 thread [4] that were verified as false positives:
>
>   - Patch 4: claimed the folio lock is dropped before collect_procs(),
>     leading to a NULL deref on folio->mapping; the lock is held in all
>     call paths.
>   - Patch 11: claimed mixing goto/scope-based cleanup in
>     alloc_retstack_tasklist(); the goto free label handles memory
>     cleanup, not lock management.
>   - Patch 13: claimed mixing goto/scope-based cleanup in
>     __set_oom_adj(); goto err_unlock manages oom_adj_mutex while
>     scoped_guard(rcu) manages the RCU read lock — different resources.
>
> Changes since v2 [3]:
>   - Split the kernel/ patch into per-subsystem patches, as the
>     combined patch could not be applied (Peter Zijlstra, Steven
>     Rostedt).
>   - Move changelog below the --- separator so it is excluded from
>     the commit message (Lorenzo Stoakes).
>   - Indent loop body in macro definitions to show scoped_guard
>     scope (Lorenzo Stoakes).
>
> Changes since v1 [1]:
>   - Rename macros from *_rcu to *_rculock, as suggested by Steven
>     Rostedt and acked by Thomas Gleixner, to avoid confusion with
>     existing *_rcu() list iterators that expect the caller to hold
>     the RCU read lock.
>   - Improve the comment on for_each_process_thread_rculock() to
>     document that 'break' only exits the inner loop and 'goto' is
>     needed to exit both loops (Thomas Gleixner).
>   - Rename the stale 'unlock:' label to 'out:' in hung_task.c, as
>     noted by Günther Noack.
>   - Clarify in patch 4 that page_pgoff() is safe outside the RCU
>     read-side critical section (SJ Park).
>   - CC all relevant maintainers on every patch (Lorenzo Stoakes).
>   - Drop the mm: prefix from patch 1, as the macros are in
>     include/linux/sched/signal.h (Michal Hocko).
>
> [1] https://lore.kernel.org/all/20260904083001.553587-1-ye.liu@linux.dev/
> [2] https://lore.kernel.org/all/20260813092933.562028-1-ye.liu@linux.dev/
> [3] https://lore.kernel.org/all/20260907081334.1152889-1-ye.liu@linux.dev/
> [4] https://lore.kernel.org/all/20260911075800.491472-1-ye.liu@linux.dev/
>
> Ye Liu (15):
>   sched: introduce for_each_process_rculock and for_each_thread_rculock
>   mm/oom_kill: convert process/thread iterators to for_each_*_rculock
>   mm/ksm: convert process iterator to for_each_process_rculock
>   mm/memory-failure: convert process iterator to
>     for_each_process_rculock
>   cpu/hotplug: convert process iterator to for_each_process_rculock
>   freezer: convert thread iterator to for_each_thread_rculock
>   hung_task: convert process/thread iterators to for_each_*_rculock
>   locking/lockdep: convert process/thread iterators to
>     for_each_*_rculock
>   rcu: convert process/thread iterator to
>     for_each_process_thread_rculock
>   sched: convert process/thread iterators to for_each_*_rculock
>   tracing/fgraph: convert process/thread iterator to
>     for_each_process_thread_rculock
>   unwind: convert process/thread iterator to
>     for_each_process_thread_rculock
>   fs: convert process/thread iterators to for_each_*_rculock
>   lib: convert process iterator to for_each_process_rculock
>   security/landlock: convert thread iterator to for_each_thread_rculock
>
>  fs/proc/base.c               |  4 +---
>  fs/resctrl/rdtgroup.c        |  8 ++------
>  include/linux/sched/signal.h | 26 ++++++++++++++++++++++++++
>  kernel/cpu.c                 |  4 +---
>  kernel/freezer.c             |  4 +---
>  kernel/hung_task.c           | 11 ++++-------
>  kernel/locking/lockdep.c     |  4 +---
>  kernel/rcu/update.c          |  4 +---
>  kernel/sched/core.c          |  3 +--
>  kernel/sched/debug.c         |  4 +---
>  kernel/trace/fgraph.c        |  8 ++------
>  kernel/unwind/deferred.c     |  3 +--
>  lib/is_single_threaded.c     |  5 +----
>  mm/ksm.c                     |  4 +---
>  mm/memory-failure.c          | 16 ++++------------
>  mm/oom_kill.c                | 20 +++++---------------
>  security/landlock/tsync.c    |  8 ++------
>  17 files changed, 55 insertions(+), 81 deletions(-)
>
> --
> 2.25.1

-- 
Thanks,
Ye Liu