Hello, Linus. The threadgroup locking changes which went in during 4.2 devel cycle added write locking of a percpu_rwsem in cgroup task migration path; unfortunately, that involved expedited rcu syncing which turned out to be too slow and heavy for certain workloads. The patchset which is dependent on this one didn't get committed during that devel cycle, so these two patches can be reverted safely.
Oleg reworked percpu_rwsem for 4.4 so that the writer path is a lot lighter. The reported issue goes away with Oleg's reworked percpu_rwsem and I'll reapply these patches on the for-4.4 branch so that they can land together with Oleg's changes. Thanks. The following changes since commit 6ff33f3902c3b1c5d0db6b1e2c70b6d76fba357f: Linux 4.3-rc1 (2015-09-12 16:35:56 -0700) are available in the git repository at: git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup.git for-4.3-fixes for you to fetch changes up to 0c986253b939cc14c69d4adbe2b4121bdf4aa220: Revert "sched, cgroup: replace signal_struct->group_rwsem with a global percpu_rwsem" (2015-09-16 11:51:12 -0400) ---------------------------------------------------------------- Tejun Heo (2): Revert "cgroup: simplify threadgroup locking" Revert "sched, cgroup: replace signal_struct->group_rwsem with a global percpu_rwsem" include/linux/cgroup-defs.h | 27 +---------- include/linux/init_task.h | 8 ++++ include/linux/sched.h | 12 +++++ kernel/cgroup.c | 110 +++++++++++++++++++++++++++++++++----------- kernel/fork.c | 4 ++ 5 files changed, 110 insertions(+), 51 deletions(-) -- tejun -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/