From: Long Li <longli@microsoft.com>
To: Peter Zijlstra <peterz@infradead.org>,
"longli@linuxonhyperv.com" <longli@linuxonhyperv.com>
Cc: Ingo Molnar <mingo@redhat.com>,
Keith Busch <keith.busch@intel.com>, Jens Axboe <axboe@fb.com>,
Christoph Hellwig <hch@lst.de>, Sagi Grimberg <sagi@grimberg.me>,
"linux-nvme@lists.infradead.org" <linux-nvme@lists.infradead.org>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: RE: [PATCH 3/3] nvme: complete request in work queue on CPU with flooded interrupts
Date: Wed, 21 Aug 2019 08:37:55 +0000 [thread overview]
Message-ID: <CY4PR21MB0741A817BEB880C8DC526ECFCEAA0@CY4PR21MB0741.namprd21.prod.outlook.com> (raw)
In-Reply-To: <20190820095240.GH2332@hirez.programming.kicks-ass.net>
>>>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, longli@linuxonhyperv.com
>>>wrote:
>>>> From: Long Li <longli@microsoft.com>
>>>>
>>>> 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
>>>> queue interrupts CPU 0 for I/O response 3. processes from CPU 1, 2 and
>>>> 3 keep sending I/Os
>>>>
>>>> CPU 0 may be flooded with interrupts from NVMe device that are I/O
>>>> responses for CPU 1, 2 and 3. Under heavy I/O load, it is possible
>>>> that CPU 0 spends all the time serving NVMe and other system
>>>> interrupts, but doesn't have a chance to run in process context.
>>>
>>>Ideally -- and there is some code to affect this, the load-balancer will move
>>>tasks away from this CPU.
>>>
>>>> To fix this, CPU 0 can schedule a work to complete the I/O request
>>>> when it detects the scheduler is not making progress. This serves multiple
>>>purposes:
>>>
>>>Suppose the task waiting for the IO completion is a RT task, and you've just
>>>queued it to a regular work. This is an instant priority inversion.
This is a choice. We can either not "lock up" the CPU, or finish the I/O on time from IRQ handler. I think throttling only happens in extreme conditions, which is rare. The purpose is to make the whole system responsive and happy.
>>>
>>>> 1. This CPU has to be scheduled to complete the request. The other
>>>> CPUs can't issue more I/Os until some previous I/Os are completed.
>>>> This helps this CPU get out of NVMe interrupts.
>>>>
>>>> 2. This acts a throttling mechanisum for NVMe devices, in that it can
>>>> not starve a CPU while servicing I/Os from other CPUs.
>>>>
>>>> 3. This CPU can make progress on RCU and other work items on its queue.
>>>>
>>>> Signed-off-by: Long Li <longli@microsoft.com>
>>>> ---
>>>> drivers/nvme/host/core.c | 57
>>>> +++++++++++++++++++++++++++++++++++++++-
>>>> drivers/nvme/host/nvme.h | 1 +
>>>> 2 files changed, 57 insertions(+), 1 deletion(-)
>>>
>>>WTH does this live in the NVME driver? Surely something like this should be
>>>in the block layer. I'm thinking there's fiber channel connected storage that
>>>should be able to trigger much the same issues.
Yes this can be done in block layer. I'm not sure the best way to accomplish this so implemented a NVMe patch to help test. The test results are promising in that we are getting 99.5% of performance while avoided CPU lockup. The challenge is to find a way to throttle a fast storage device.
>>>
>>>> diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c index
>>>> 6a9dd68c0f4f..576bb6fce293 100644
>>>> --- a/drivers/nvme/host/core.c
>>>> +++ b/drivers/nvme/host/core.c
>>>
>>>> @@ -260,9 +270,54 @@ static void nvme_retry_req(struct request *req)
>>>> blk_mq_delay_kick_requeue_list(req->q, delay); }
>>>>
>>>> +static void nvme_complete_rq_work(struct work_struct *work) {
>>>> + struct nvme_request *nvme_rq =
>>>> + container_of(work, struct nvme_request, work);
>>>> + struct request *req = blk_mq_rq_from_pdu(nvme_rq);
>>>> +
>>>> + nvme_complete_rq(req);
>>>> +}
>>>> +
>>>> +
>>>> void nvme_complete_rq(struct request *req) {
>>>> - blk_status_t status = nvme_error_status(req);
>>>> + blk_status_t status;
>>>> + int cpu;
>>>> + u64 switches;
>>>> + struct nvme_request *nvme_rq;
>>>> +
>>>> + if (!in_interrupt())
>>>> + goto skip_check;
>>>> +
>>>> + nvme_rq = nvme_req(req);
>>>> + cpu = smp_processor_id();
>>>> + if (idle_cpu(cpu))
>>>> + goto skip_check;
>>>> +
>>>> + /* Check if this CPU is flooded with interrupts */
>>>> + switches = get_cpu_rq_switches(cpu);
>>>> + if (this_cpu_read(last_switch) == switches) {
>>>> + /*
>>>> + * If this CPU hasn't made a context switch in
>>>> + * MAX_SCHED_TIMEOUT ns (and it's not idle), schedule a
>>>work to
>>>> + * complete this I/O. This forces this CPU run non-interrupt
>>>> + * code and throttle the other CPU issuing the I/O
>>>> + */
>>>
>>>What if there was only a single task on that CPU? Then we'd never
>>>need/want to context switch in the first place.
>>>
>>>AFAICT all this is just a whole bunch of gruesome hacks piled on top one
>>>another.
>>>
>>>> + if (sched_clock() - this_cpu_read(last_clock)
>>>> + > MAX_SCHED_TIMEOUT) {
>>>> + INIT_WORK(&nvme_rq->work,
>>>nvme_complete_rq_work);
>>>> + schedule_work_on(cpu, &nvme_rq->work);
>>>> + return;
>>>> + }
>>>> +
>>>> + } else {
>>>> + this_cpu_write(last_switch, switches);
>>>> + this_cpu_write(last_clock, sched_clock());
>>>> + }
>>>> +
>>>> +skip_check:
>>>
>>>Aside from everything else; this is just sodding poor coding style. What is
>>>wrong with something like:
>>>
>>> if (nvme_complete_throttle(...))
>>> return;
>>>
>>>> + status = nvme_error_status(req);
>>>>
>>>> trace_nvme_complete_rq(req);
>>>>
next prev parent reply other threads:[~2019-08-21 8:38 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-08-20 6:14 [PATCH 0/3] fix interrupt swamp in NVMe longli
2019-08-20 6:14 ` [PATCH 1/3] sched: define a function to report the number of context switches on a CPU longli
2019-08-20 9:38 ` Peter Zijlstra
2019-08-21 8:20 ` Long Li
2019-08-21 10:34 ` Peter Zijlstra
2019-08-20 9:39 ` Peter Zijlstra
2019-08-20 6:14 ` [PATCH 2/3] sched: export idle_cpu() longli
2019-08-20 6:14 ` [PATCH 3/3] nvme: complete request in work queue on CPU with flooded interrupts longli
2019-08-20 9:52 ` Peter Zijlstra
2019-08-21 8:37 ` Long Li [this message]
2019-08-21 10:35 ` Peter Zijlstra
2019-08-20 17:33 ` Sagi Grimberg
2019-08-21 8:39 ` Long Li
2019-08-21 17:36 ` Long Li
2019-08-21 21:54 ` Sagi Grimberg
2019-08-24 0:13 ` Long Li
2019-08-23 3:21 ` Ming Lei
2019-08-24 0:27 ` Long Li
2019-08-24 12:55 ` Ming Lei
2019-08-20 8:25 ` [PATCH 0/3] fix interrupt swamp in NVMe Ming Lei
2019-08-20 8:59 ` John Garry
2019-08-20 15:05 ` Keith Busch
2019-08-21 7:47 ` Long Li
2019-08-21 9:44 ` Ming Lei
2019-08-21 10:03 ` John Garry
2019-08-21 16:27 ` Long Li
2019-08-22 1:33 ` Ming Lei
2019-08-22 2:00 ` Keith Busch
2019-08-22 2:23 ` Ming Lei
2019-08-22 9:48 ` Thomas Gleixner
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CY4PR21MB0741A817BEB880C8DC526ECFCEAA0@CY4PR21MB0741.namprd21.prod.outlook.com \
--to=longli@microsoft.com \
--cc=axboe@fb.com \
--cc=hch@lst.de \
--cc=keith.busch@intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-nvme@lists.infradead.org \
--cc=longli@linuxonhyperv.com \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=sagi@grimberg.me \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).