[PATCH v3 0/5] proc: improve root readdir latency with many threads

From: Brian Foster
Date: Fri Dec 02 2022 - 12:17:26 EST


Hi all,

Here's v3 of the /proc readdir optimization patches. See v1 for the full
introductary cover letter.

Most of the feedback received to this point has been around switching
the pid code over to use the xarray api instead of the idr. Matt Wilcox
posted most of the code to do that. I cleaned it up a bit and posted a
standalone series for that here [1], but didn't receive any feedback.
Patches 1-3 of this series are essentially a repost of [1].

Patches 4-5 are otherwise mostly the same as v2 outside of switching
over to use the xarray bits instead of the idr/radix-tree.

Thoughts, reviews, flames appreciated.

Brian

[1] https://lore.kernel.org/linux-mm/20220715113349.831370-1-bfoster@xxxxxxxxxx/

v3:
- Drop radix-tree fixups.
- Convert pid idr usage to xarray.
- Replace tgid radix-tree tag set/lookup to use xarray mark.
v2: https://lore.kernel.org/linux-fsdevel/20220711135237.173667-1-bfoster@xxxxxxxxxx/
- Clean up idr helpers to be more generic.
- Use ->idr_base properly.
- Lift tgid iteration helper into pid.c to abstract tag logic from
users.
v1: https://lore.kernel.org/linux-fsdevel/20220614180949.102914-1-bfoster@xxxxxxxxxx/

Brian Foster (5):
pid: replace pidmap_lock with xarray lock
pid: split cyclic id allocation cursor from idr
pid: switch pid_namespace from idr to xarray
pid: mark pids associated with group leader tasks
procfs: use efficient tgid pid search on root readdir

arch/powerpc/platforms/cell/spufs/sched.c | 2 +-
fs/proc/base.c | 17 +--
fs/proc/loadavg.c | 2 +-
include/linux/pid.h | 3 +-
include/linux/pid_namespace.h | 9 +-
include/linux/threads.h | 2 +-
init/main.c | 3 +-
kernel/fork.c | 2 +-
kernel/pid.c | 177 +++++++++++++---------
kernel/pid_namespace.c | 23 ++-
10 files changed, 132 insertions(+), 108 deletions(-)

--
2.37.3