Skip to content

kill the racy EXIT_ZOMBIE->EXIT_DEAD->EXIT_ZOMBIE transition

Oleg Nesterov lkml 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest5 patches 2 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/5wait: fix reparent_leader() vs EXIT_DEAD->EXIT_ZOMBIE racekernel/exit.c+11 -40 replies
2/5wait: introduce EXIT_TRACE to avoid the racy EXIT_DEAD->EXIT_ZOMBIE transitioninclude/linux/sched.h kernel/exit.c+22 -290 replies
3/5wait: use EXIT_TRACE only if thread_group_leader(zombie)kernel/exit.c+7 -100 replies
4/5wait: completely ignore the EXIT_DEAD taskskernel/exit.c+6 -50 replies
5/5wait: swap EXIT_ZOMBIE and EXIT_DEAD to hide EXIT_TRACE from user-spacefs/proc/array.c include/linux/sched.h+4 -40 replies

Reviews trailers per patch

None of the 5 patches has a Reviewed-by or Acked-by.

PatchMichal Schmidt
1/5Reported-and-tested-by
2/5Reported-and-tested-by
3/5T-b
4/5T-b
5/5

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [PATCH 0/5] kill the racy EXIT_ZOMBIE->EXIT_DEAD->EXIT_ZOMBIE transition

Hello.

1/7 cc's stable, and imo it is v3.14 material. This hack is reverted by the next patch.

Many thanks to Michal and Jan for investigating.

And it seems that we need more (unrelated) changes in do_wait(), will try to do this tomorrow.

Tejun, unless I missed something WSTOPPED logic is broken if a process has a zombie/ptraced leader, "A zombie ptracee is only visible to its ptracer" is wrong in this case. Plus perhaps some cleanups make sense.

Oleg.

 fs/proc/array.c       |    4 +-
 include/linux/sched.h |    5 ++-
 kernel/exit.c         |   51 +++++++++++++++++++++++-------------------------
 3 files changed, 29 insertions(+), 31 deletions(-)