From: "Jianyong Wu (Arm Technology China)" <Jianyong.Wu@arm.com>
To: Thomas Gleixner <tglx@linutronix.de>,
Paolo Bonzini <pbonzini@redhat.com>
Cc: "Justin He \(Arm Technology China\)" <Justin.He@arm.com>,
"kvm@vger.kernel.org" <kvm@vger.kernel.org>,
"maz@kernel.org" <maz@kernel.org>,
"richardcochran@gmail.com" <richardcochran@gmail.com>,
"yangbo.lu@nxp.com" <yangbo.lu@nxp.com>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"sean.j.christopherson@intel.com"
<sean.j.christopherson@intel.com>,
"john.stultz@linaro.org" <john.stultz@linaro.org>,
"netdev@vger.kernel.org" <netdev@vger.kernel.org>,
nd <nd@arm.com>, "will@kernel.org" <will@kernel.org>,
"kvmarm@lists.cs.columbia.edu" <kvmarm@lists.cs.columbia.edu>,
"linux-arm-kernel@lists.infradead.org"
<linux-arm-kernel@lists.infradead.org>
Subject: RE: [PATCH v5 3/6] timekeeping: Add clocksource to system_time_snapshot
Date: Wed, 16 Oct 2019 09:48:43 +0000 [thread overview]
Message-ID: <HE1PR0801MB1676EC775B7BFA5FC7E4F9D5F4920@HE1PR0801MB1676.eurprd08.prod.outlook.com> (raw)
In-Reply-To: <alpine.DEB.2.21.1910160914230.2518@nanos.tec.linutronix.de>
Hi tglx,
> -----Original Message-----
> From: Thomas Gleixner <tglx@linutronix.de>
> Sent: Wednesday, October 16, 2019 3:29 PM
> To: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Jianyong Wu (Arm Technology China) <Jianyong.Wu@arm.com>;
> netdev@vger.kernel.org; yangbo.lu@nxp.com; john.stultz@linaro.org;
> sean.j.christopherson@intel.com; maz@kernel.org;
> richardcochran@gmail.com; Mark Rutland <Mark.Rutland@arm.com>;
> will@kernel.org; Suzuki Poulose <Suzuki.Poulose@arm.com>; linux-
> kernel@vger.kernel.org; linux-arm-kernel@lists.infradead.org;
> kvmarm@lists.cs.columbia.edu; kvm@vger.kernel.org; Steve Capper
> <Steve.Capper@arm.com>; Kaly Xin (Arm Technology China)
> <Kaly.Xin@arm.com>; Justin He (Arm Technology China)
> <Justin.He@arm.com>; nd <nd@arm.com>
> Subject: Re: [PATCH v5 3/6] timekeeping: Add clocksource to
> system_time_snapshot
>
> On Wed, 16 Oct 2019, Paolo Bonzini wrote:
> > On 15/10/19 22:13, Thomas Gleixner wrote:
> > > On Tue, 15 Oct 2019, Paolo Bonzini wrote:
> > >> On 15/10/19 12:48, Jianyong Wu wrote:
> > >>>
> > >>>
> > >>
> > >> Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
> > >
> > > You're sure about having reviewed that in detail?
> >
> > I did review the patch; the void* ugliness is not in this one, and I
> > do have some other qualms on that one.
> >
> > > This changelog is telling absolutely nothing WHY anything outside of
> > > the timekeeping core code needs access to the current clocksource.
> > > Neither does it tell why it is safe to provide the pointer to random callers.
> >
> > Agreed on the changelog, but the pointer to a clocksource is already
> > part of the timekeeping external API via struct system_counterval_t.
> > get_device_system_crosststamp for example expects a clocksource
> > pointer but provides no way to get such a pointer.
>
> That's a completely different beast, really.
>
> The clocksource pointer is handed in by the caller and the core code validates
> if the clocksource is the same as the current system clocksource and not the
> other way round.
>
> So there is no need for getting that pointer from the core code because the
> caller knows already which clocksource needs to be active to make.the whole
> cross device timestamp correlation work. And in that case it's the callers
> responsibility to ensure that the pointer is valid which is the case for the
> current use cases.
>
I thinks there is something misunderstanding of my patch. See patch 4/6, the reason why I add clocksource is that I want to check if the current clocksouce is
arm_arch_counter in virt/kvm/arm/psci.c and nothing to do with get_device_system_crosststamp.
So I really need a mechanism to do that check.
Thanks
Jianyong
> From your other reply:
>
> > Why add a global id? ARM can add it to archdata similar to how x86
> > has vclock_mode. But I still think the right thing to do is to
> > include the full system_counterval_t in the result of
> > ktime_get_snapshot. (More in a second, feel free to reply to the other
> email only).
>
> No, the clocksource pointer is not going to be exposed as there is no
> guarantee that it will be still around after the call returns.
>
> It's not even guaranteed to be correct when the store happens in Wu's patch
> simply because the store is done outside of the seqcount protected region.
Yeah, all of the elements in system_time_snapshot should be captured in consistency. So
I think the consistency will be guaranteed if the store ops added in the seqcount region.
>
> Vs. arch data: arch data is an opaque struct, so you'd need to store a pointer
> which has the same issue as the clocksource pointer itself.
>
> If we want to convey information then it has to be in the generic part of
> struct clocksource.
>
> In fact we could even simplify the existing get_device_system_crosststamp()
> use case by using the ID field.
>
> Thanks,
>
> tglx
_______________________________________________
kvmarm mailing list
kvmarm@lists.cs.columbia.edu
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
next prev parent reply other threads:[~2019-10-17 10:29 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-10-15 10:48 [RFC PATCH v5 0/6] Enable ptp_kvm for arm64 Jianyong Wu
2019-10-15 10:48 ` [PATCH v5 1/6] psci: Export psci_ops.conduit symbol as modules will use it Jianyong Wu
2019-10-15 17:24 ` Mark Rutland
2019-10-16 3:54 ` Jianyong Wu (Arm Technology China)
2019-10-15 10:48 ` [PATCH v5 2/6] ptp: Reorganize ptp_kvm modules to make it arch-independent Jianyong Wu
2019-10-16 7:10 ` Paolo Bonzini
2019-10-16 10:04 ` Jianyong Wu (Arm Technology China)
2019-10-15 10:48 ` [PATCH v5 3/6] timekeeping: Add clocksource to system_time_snapshot Jianyong Wu
2019-10-15 16:37 ` Paolo Bonzini
2019-10-15 20:13 ` Thomas Gleixner
2019-10-15 22:36 ` Paolo Bonzini
2019-10-16 7:28 ` Thomas Gleixner
2019-10-16 9:48 ` Jianyong Wu (Arm Technology China) [this message]
2019-10-16 10:23 ` Thomas Gleixner
2019-10-15 20:12 ` Thomas Gleixner
2019-10-15 22:35 ` Paolo Bonzini
2019-10-16 10:01 ` Jianyong Wu (Arm Technology China)
2019-10-16 10:26 ` Thomas Gleixner
2019-10-15 10:48 ` [PATCH v5 4/6] psci: Add hvc call service for ptp_kvm Jianyong Wu
2019-10-16 7:24 ` Paolo Bonzini
2019-10-16 7:31 ` Thomas Gleixner
2019-10-16 7:42 ` Paolo Bonzini
2019-10-16 10:20 ` Jianyong Wu (Arm Technology China)
2019-10-15 10:48 ` [PATCH v5 5/6] ptp: arm64: Enable ptp_kvm for arm64 Jianyong Wu
2019-10-15 16:39 ` Paolo Bonzini
2019-10-16 3:52 ` Jianyong Wu (Arm Technology China)
2019-10-16 7:10 ` Paolo Bonzini
2019-10-16 7:13 ` Paolo Bonzini
2019-10-16 10:11 ` Jianyong Wu (Arm Technology China)
2019-10-15 10:48 ` [PATCH v5 6/6] kvm: arm64: Add capability check extension for ptp_kvm Jianyong Wu
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=HE1PR0801MB1676EC775B7BFA5FC7E4F9D5F4920@HE1PR0801MB1676.eurprd08.prod.outlook.com \
--to=jianyong.wu@arm.com \
--cc=Justin.He@arm.com \
--cc=john.stultz@linaro.org \
--cc=kvm@vger.kernel.org \
--cc=kvmarm@lists.cs.columbia.edu \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=maz@kernel.org \
--cc=nd@arm.com \
--cc=netdev@vger.kernel.org \
--cc=pbonzini@redhat.com \
--cc=richardcochran@gmail.com \
--cc=sean.j.christopherson@intel.com \
--cc=tglx@linutronix.de \
--cc=will@kernel.org \
--cc=yangbo.lu@nxp.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).