From: Zachary Amsden <zamsden@redhat.com>
To: "David S. Ahern" <daahern@cisco.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>, KVM <kvm@vger.kernel.org>,
Avi Kivity <avi@redhat.com>, Glauber Costa <glommer@redhat.com>,
Linux-kernel <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 09/18] Robust TSC compensation
Date: Tue, 13 Jul 2010 11:45:51 -1000 [thread overview]
Message-ID: <4C3CDE8F.60304@redhat.com> (raw)
In-Reply-To: <4C3CDDD4.6010009@cisco.com>
On 07/13/2010 11:42 AM, David S. Ahern wrote:
>
> On 07/13/10 15:15, Zachary Amsden wrote:
>
>
>>> What prevents a vcpu from seeing its TSC go backwards, in case the first
>>> write in the 5 second window is smaller than the victim vcpu's last
>>> visible TSC value ?
>>>
>>>
>> Nothing, unfortunately. However, the TSC would already have to be out
>> of sync in order for the problem to occur. It can never happen in
>> normal circumstances on a stable hardware TSC except in one case;
>> migration. During the CPU state transfer phase of migration, however,
>>
> What about across processor sockets? Aren't CPUs brought up at different
> points such that their TSCs start at different times?
>
Yes, that's called an unsynchronized TSC. In that case, the
compensation does the best it can based on time since the first TSC
write, but it will never be exact.
next prev parent reply other threads:[~2010-07-13 21:45 UTC|newest]
Thread overview: 61+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-07-13 2:25 KVM timekeeping fixes, V2 Zachary Amsden
2010-07-13 2:25 ` [PATCH 01/18] Make TSC offset writes non-preemptible Zachary Amsden
2010-07-13 21:33 ` Rik van Riel
2010-07-18 14:28 ` Avi Kivity
2010-07-18 14:30 ` Avi Kivity
2010-07-13 2:25 ` [PATCH 02/18] Fix SVM VMCB reset Zachary Amsden
2010-07-13 21:37 ` Rik van Riel
2010-07-13 2:25 ` [PATCH 03/18] TSC reset compensation Zachary Amsden
2010-07-13 22:11 ` Rik van Riel
2010-07-18 14:34 ` Avi Kivity
2010-07-19 20:01 ` Zachary Amsden
2010-07-13 2:25 ` [PATCH 04/18] Make cpu_tsc_khz updates use local CPU Zachary Amsden
2010-07-14 14:41 ` Rik van Riel
2010-07-18 14:45 ` Avi Kivity
2010-07-19 20:06 ` Zachary Amsden
2010-07-20 8:53 ` Avi Kivity
2010-07-20 21:57 ` Zachary Amsden
2010-07-13 2:25 ` [PATCH 05/18] Warn about unstable TSC Zachary Amsden
2010-07-14 15:02 ` Rik van Riel
2010-07-18 14:47 ` Avi Kivity
2010-07-13 2:25 ` [PATCH 06/18] Unify TSC logic Zachary Amsden
2010-07-14 15:53 ` Rik van Riel
2010-07-13 2:25 ` [PATCH 07/18] Fix deep C-state TSC desynchronization Zachary Amsden
2010-07-14 16:14 ` Rik van Riel
2010-07-13 2:25 ` [PATCH 08/18] Add helper functions for time computation Zachary Amsden
2010-07-14 19:02 ` Rik van Riel
2010-07-13 2:25 ` [PATCH 09/18] Robust TSC compensation Zachary Amsden
2010-07-13 20:34 ` Marcelo Tosatti
2010-07-13 21:15 ` Zachary Amsden
2010-07-13 21:42 ` David S. Ahern
2010-07-13 21:45 ` Zachary Amsden [this message]
2010-07-13 23:32 ` Zachary Amsden
2010-07-14 22:33 ` Rik van Riel
2010-07-18 14:52 ` Avi Kivity
2010-07-19 20:39 ` Zachary Amsden
2010-07-13 2:25 ` [PATCH 10/18] Keep SMP VMs more in sync on unstable TSC Zachary Amsden
2010-07-15 2:13 ` Rik van Riel
2010-07-13 2:25 ` [PATCH 11/18] Perform hardware_enable in CPU_STARTING callback Zachary Amsden
2010-07-15 2:15 ` Rik van Riel
2010-07-13 2:25 ` [PATCH 12/18] Add clock sync request to hardware enable Zachary Amsden
2010-07-15 2:32 ` Rik van Riel
2010-07-13 2:25 ` [PATCH 13/18] Move scale_delta into common header Zachary Amsden
2010-07-15 2:35 ` Rik van Riel
2010-07-13 2:25 ` [PATCH 14/18] Fix a possible backwards warp of kvmclock Zachary Amsden
2010-07-15 2:37 ` Rik van Riel
2010-07-13 2:25 ` [PATCH 15/18] Implement getnsboottime kernel API Zachary Amsden
2010-07-15 2:41 ` Rik van Riel
2010-07-18 15:07 ` Avi Kivity
2010-07-13 2:25 ` [PATCH 16/18] Use getnsboottime in KVM Zachary Amsden
2010-07-15 2:43 ` Rik van Riel
2010-07-13 2:25 ` [PATCH 17/18] Indicate reliable TSC in kvmclock Zachary Amsden
2010-07-15 2:44 ` Rik van Riel
2010-07-13 2:25 ` [PATCH 18/18] Add timekeeping documentation Zachary Amsden
2010-07-14 7:16 ` Takuya Yoshikawa
2010-07-14 20:28 ` Zachary Amsden
2010-07-16 13:19 ` KVM timekeeping fixes, V2 Joerg Roedel
2010-07-16 17:20 ` Zachary Amsden
2010-07-16 19:26 ` Joerg Roedel
2010-07-18 14:22 ` Avi Kivity
2010-07-18 15:08 ` Avi Kivity
2010-07-19 8:11 ` Zachary Amsden
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=4C3CDE8F.60304@redhat.com \
--to=zamsden@redhat.com \
--cc=avi@redhat.com \
--cc=daahern@cisco.com \
--cc=glommer@redhat.com \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mtosatti@redhat.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).