RE: [PATCH 3/3] nvme: complete request in work queue on CPU with flooded interrupts

2019-08-23 Thread Long Li
>>>Subject: Re: [PATCH 3/3] nvme: complete request in work queue on CPU >>>with flooded interrupts >>> >>> >>>> Sagi, >>>> >>>> Here are the test results. >>>> >>>> Benchmark command: >>>>

Re: [PATCH 3/3] nvme: complete request in work queue on CPU with flooded interrupts

2019-08-22 Thread Ming Lei
On Tue, Aug 20, 2019 at 10:33:38AM -0700, Sagi Grimberg wrote: > > > From: Long Li > > > > When a NVMe hardware queue is mapped to several CPU queues, it is possible > > that the CPU this hardware queue is bound to is flooded by returning I/O for > > other CPUs. > > > > For example, consider th

Re: [PATCH 3/3] nvme: complete request in work queue on CPU with flooded interrupts

2019-08-21 Thread Sagi Grimberg
Sagi, Here are the test results. Benchmark command: fio --bs=4k --ioengine=libaio --iodepth=64 --filename=/dev/nvme0n1:/dev/nvme1n1:/dev/nvme2n1:/dev/nvme3n1:/dev/nvme4n1:/dev/nvme5n1:/dev/nvme6n1:/dev/nvme7n1:/dev/nvme8n1:/dev/nvme9n1 --direct=1 --runtime=90 --numjobs=80 --rw=randread --na

RE: [PATCH 3/3] nvme: complete request in work queue on CPU with flooded interrupts

2019-08-21 Thread Long Li
>>>Subject: RE: [PATCH 3/3] nvme: complete request in work queue on CPU >>>with flooded interrupts >>> >>>>>>Subject: Re: [PATCH 3/3] nvme: complete request in work queue on >>>CPU >>>>>>with flooded interrupts >>&g

Re: [PATCH 3/3] nvme: complete request in work queue on CPU with flooded interrupts

2019-08-21 Thread Peter Zijlstra
On Wed, Aug 21, 2019 at 08:37:55AM +, Long Li wrote: > >>>Subject: Re: [PATCH 3/3] nvme: complete request in work queue on CPU > >>>with flooded interrupts > >>> > >>>On Mon, Aug 19, 2019 at 11:14:29PM -0700, lon...@linuxonhyperv.com > >&g

RE: [PATCH 3/3] nvme: complete request in work queue on CPU with flooded interrupts

2019-08-21 Thread Long Li
>>>Subject: Re: [PATCH 3/3] nvme: complete request in work queue on CPU >>>with flooded interrupts >>> >>> >>>> From: Long Li >>>> >>>> When a NVMe hardware queue is mapped to several CPU queues, it is >>>> possib

RE: [PATCH 3/3] nvme: complete request in work queue on CPU with flooded interrupts

2019-08-21 Thread Long Li
>>>Subject: Re: [PATCH 3/3] nvme: complete request in work queue on CPU >>>with flooded interrupts >>> >>>On Mon, Aug 19, 2019 at 11:14:29PM -0700, lon...@linuxonhyperv.com >>>wrote: >>>> From: Long Li >>>> >>>> Whe

Re: [PATCH 3/3] nvme: complete request in work queue on CPU with flooded interrupts

2019-08-20 Thread Sagi Grimberg
From: Long Li When a NVMe hardware queue is mapped to several CPU queues, it is possible that the CPU this hardware queue is bound to is flooded by returning I/O for other CPUs. For example, consider the following scenario: 1. CPU 0, 1, 2 and 3 share the same hardware queue 2. the hardware q

Re: [PATCH 3/3] nvme: complete request in work queue on CPU with flooded interrupts

2019-08-20 Thread Peter Zijlstra
On Mon, Aug 19, 2019 at 11:14:29PM -0700, lon...@linuxonhyperv.com wrote: > From: Long Li > > When a NVMe hardware queue is mapped to several CPU queues, it is possible > that the CPU this hardware queue is bound to is flooded by returning I/O for > other CPUs. > > For example, consider the foll