Messages in this thread | | | Date | Tue, 20 Aug 2019 08:07:27 +0200 | From | Steffen Klassert <> | Subject | Re: [PATCH 0/9] padata: use unbound workqueues for parallel jobs |
| |
On Mon, Aug 12, 2019 at 08:52:15PM -0400, Daniel Jordan wrote: > Hi, > > No objections the first time around[*], so this series is no longer RFC. > The code is the same other than being rebased on recent padata fixes. > Any feedback or testing welcome. > > Thanks, > Daniel > > RFC -> v1: > - Included Tejun's acks. > - Added testing section to cover letter. > > Padata binds the parallel part of a job to a single CPU and round-robins > over all CPUs in the system for each successive job. Though the serial > parts rely on per-CPU queues for correct ordering, they're not necessary > for parallel work, and it improves performance to run the job locally on > NUMA machines and let the scheduler pick the CPU within a node on a busy > system. > > This series makes parallel padata jobs run on unbound workqueues.
Looks good!
I did a test with IPsec pcrypt and it seems to work as expected. Crypto load gets distributed, no network packet reordering.
Acked-by: Steffen Klassert <steffen.klassert@secunet.com>
| |