]> git.karo-electronics.de Git - linux-beck.git/commit
workqueue: reimplement WQ_HIGHPRI using a separate worker_pool
authorTejun Heo <tj@kernel.org>
Sat, 14 Jul 2012 05:16:45 +0000 (22:16 -0700)
committerTejun Heo <tj@kernel.org>
Sat, 14 Jul 2012 05:24:45 +0000 (22:24 -0700)
commit3270476a6c0ce322354df8679652f060d66526dc
treedb58846beb7c5e1c1b50b7e2f1c2538320408c26
parent4ce62e9e30cacc26885cab133ad1de358dd79f21
workqueue: reimplement WQ_HIGHPRI using a separate worker_pool

WQ_HIGHPRI was implemented by queueing highpri work items at the head
of the global worklist.  Other than queueing at the head, they weren't
handled differently; unfortunately, this could lead to execution
latency of a few seconds on heavily loaded systems.

Now that workqueue code has been updated to deal with multiple
worker_pools per global_cwq, this patch reimplements WQ_HIGHPRI using
a separate worker_pool.  NR_WORKER_POOLS is bumped to two and
gcwq->pools[0] is used for normal pri work items and ->pools[1] for
highpri.  Highpri workers get -20 nice level and has 'H' suffix in
their names.  Note that this change increases the number of kworkers
per cpu.

POOL_HIGHPRI_PENDING, pool_determine_ins_pos() and highpri chain
wakeup code in process_one_work() are no longer used and removed.

This allows proper prioritization of highpri work items and removes
high execution latency of highpri work items.

v2: nr_running indexing bug in get_pool_nr_running() fixed.

v3: Refreshed for the get_pool_nr_running() update in the previous
    patch.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Josh Hunt <joshhunt00@gmail.com>
LKML-Reference: <CAKA=qzaHqwZ8eqpLNFjxnO2fX-tgAOjmpvxgBFjv6dJeQaOW1w@mail.gmail.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Documentation/workqueue.txt
kernel/workqueue.c