From: Marc Zyngier <marc.zyngier@arm.com>
To: "Tangnianyao (ICT)" <tangnianyao@huawei.com>
Cc: <kvm@vger.kernel.org>, <kvmarm@lists.cs.columbia.edu>,
<linux-kernel@vger.kernel.org>, Linuxarm <linuxarm@huawei.com>,
guoheyi@huawei.com, Christoffer Dall <christoffer.dall@arm.com>
Subject: Re: [RFC] Question about enable doorbell irq and halt_poll process
Date: Wed, 20 Mar 2019 17:02:19 +0000 [thread overview]
Message-ID: <20190320170219.510f2e1e@why.wild-wind.fr.eu.org> (raw)
In-Reply-To: <0fb3c9ba-8428-ea6c-2973-952624f601cc@huawei.com>
On Tue, 19 Mar 2019 21:25:47 +0800
"Tangnianyao (ICT)" <tangnianyao@huawei.com> wrote:
> Hi, all
>
> Using gicv4, when guest is waiting for irq, it sends wfi and traps to kvm.
> When vlpi is forwarded to PE after its_vpe_deschedule, before halt_poll in
> kvm_vcpu_block, halt_poll may increase latency for this vlpi getting to guest.
> In halt_poll process, it checks if there's pending irq for vcpu using pending_last.
> However, doorbell is not enable at this moment and vlpi or doorbell can not set
> pending_last true, to stop halt_poll. It will run until halt_poll time ends, if
> there's no other physical irq coming in the meantime. And then vcpu is scheduled out.
> This pending vlpi has to wait for vcpu getting schedule in next time.
>
> Should we enable doorbell before halt_poll process ?
Enabling doorbells can be quite expensive. Depending on the HW, this is
either:
- a write to memory (+DSB, potential cache maintenance), a write to the
INVLPI register, and a poll of the SYNC register
- a write to memory (+DSB, potential cache maintenance), potentially
a string of DISCARD+SYNC+MAPI+SYNC commands, and an INV+SYNC command
Frankly, you want to be careful with that. I'd rather enable them late
and have a chance of not blocking because of another (virtual)
interrupt, which saves us the doorbell business.
I wonder if you wouldn't be in a better position by drastically
reducing halt_poll_ns for vcpu that can have directly injected
interrupts.
In any case, this is something that we should measure, not guess.
Thanks,
M.
--
Without deviation from the norm, progress is not possible.
next prev parent reply other threads:[~2019-03-20 17:02 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-03-19 13:25 [RFC] Question about enable doorbell irq and halt_poll process Tangnianyao (ICT)
2019-03-20 14:31 ` Heyi Guo
2019-03-20 17:02 ` Marc Zyngier [this message]
[not found] ` <5df934fd-06d5-55f2-68a5-6f4985e4ac1b@huawei.com>
[not found] ` <86zhpc66jl.wl-marc.zyngier@arm.com>
2019-04-04 10:07 ` Tangnianyao (ICT)
2019-04-04 10:59 ` Marc Zyngier
2019-04-23 7:44 ` Tangnianyao (ICT)
2019-04-23 10:00 ` Marc Zyngier
2019-04-29 2:29 ` Tangnianyao (ICT)
2019-05-14 9:02 ` Tangnianyao (ICT)
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=20190320170219.510f2e1e@why.wild-wind.fr.eu.org \
--to=marc.zyngier@arm.com \
--cc=christoffer.dall@arm.com \
--cc=guoheyi@huawei.com \
--cc=kvm@vger.kernel.org \
--cc=kvmarm@lists.cs.columbia.edu \
--cc=linux-kernel@vger.kernel.org \
--cc=linuxarm@huawei.com \
--cc=tangnianyao@huawei.com \
/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).