Messages in this thread | | | From | Lai Jiangshan <> | Subject | [PATCH 0/7 V2] workqueue: cleanup for attr management | Date | Tue, 12 May 2015 20:32:28 +0800 |
| |
Hi,
This is the V2 version of the V1 pathset. But it is just the updated version of the patch1&2 of the V1 patchset.
[1/5 V1] is split into [1/7 V2] and [2/7 V2]. [2/5 V1] is split into [3,4,5,6,7/7 V2].
[1/7] extends the wq_pool_mutex lock region in the apply_workqueue_attrs(). It is a basic patch for all the later patches except the [6/7 V2]. [2/7] simplifies wq_update_unbound_numa() since lock requirement is eased.
[3/7] introduces get_pwq_unlocked() for reusing existing pwqs It is also a basic patch for all the later patches except the [6/7 V2]. [4,5,6/7] apply the good code of wq_update_unbound_numa() to apply_workqueue_attrs(): reuse the unchanged per-node/default pwq and reuse the preallocated wq_update_unbound_numa_attrs_buf.
[2,3,4,5,6/7] try to make the per-node allocation are the same in wq_update_unbound_numa() and apply_workqueue_attrs(), but they are not exactly the same. wq_update_unbound_numa() directly returns when the pwq unchanged.
[7/7] adds get_node_unbound_pwq() which uses the per-node allocation behavior of the apply_workqueue_attrs(). So wq_update_unbound_numa() has some overhead introduced. But cpu-hotplug path is cold path, it is Ok.
It removes some comments without adding the corresponding ones back.
Other changed from V1: The unneeded comment about the stableness of the wq->unbound_attrs is removed in [2/7].
alloc_node_unbound_pwq() is renamed to get_node_unbound_pwq(). @use_dfl_when_fail is removed from get_node_unbound_pwq(). get_node_unbound_pwq() is shorter under the help of get_pwq_unlocked().
apply_wqattrs_cleanup() is not protected by wq_pool_mutex
some comment revised as TJ's suguested.
Thanks, Lai
Cc: Tejun Heo <tj@kernel.org>
Lai Jiangshan (7): workqueue: wq_pool_mutex protects the attrs-installation workqueue: simplify wq_update_unbound_numa() workqueue: introduce get_pwq_unlocked() workqueue: reuse the current per-node pwq when its attrs unchanged workqueue: reuse the current default pwq when its attrs unchanged workqueue: reuse wq_update_unbound_numa_attrs_buf as temporary attrs workqueue: add get_node_unbound_pwq()
kernel/workqueue.c | 197 ++++++++++++++++++++++++++--------------------------- 1 file changed, 95 insertions(+), 102 deletions(-)
-- 2.1.0 ge
| |