From: Alex Williamson <alex.williamson@redhat.com>
To: Ross Deleon <rdeleon@marvell.com>
Cc: "kvm@vger.kernel.org" <kvm@vger.kernel.org>
Subject: Re: Could multiple PCIe devices in one IOMMU group be added to different KVM VM separately?
Date: Thu, 23 Jul 2020 15:55:36 -0600 [thread overview]
Message-ID: <20200723155536.52655375@w520.home> (raw)
In-Reply-To: <BY5PR18MB3282589C431572AC89EC531CA3760@BY5PR18MB3282.namprd18.prod.outlook.com>
On Thu, 23 Jul 2020 21:38:31 +0000
Ross Deleon <rdeleon@marvell.com> wrote:
> Hi:
> I want to know could multiple PCIe devices in one IOMMU group be
> added to different KVM VM separately? Mother board: ASUS 390-A(intel
> supports VT-D) host OS: Ubuntu 18.04 with kernel 4.18.0-15
> KVM-QEMU: version 2.11.1
> I have got failed message like *card1 is used by VM1, and card2 is in
> the card1's group and added to VM2, then VM2 can't boot* I have tried
> vfio driver, but it didn't work, so what should I do? try SR-IOV? or
> update KVM-QEMU or update kernel?
No. The IOMMU group is the smallest set of devices that are DMA
isolated from other groups. An IOMMU group is the unit of assignment
for vfio and also the granularity with which we manage IOMMU context.
The processor and chipset support for PCIe ACS (Access Control Services)
plays a significant role in getting the smallest granularity in IOMMU
grouping. Consumer systems often do not support ACS to the same degree
as a server, therefore if you have independent devices that are grouped
together, the supported answer might be to get new hardware, or
possibly move the device to a different slot. Intel consumer CPUs do
not support ACS, but the chipsets do make use of several quirks that
can sometime provide ACS equivalent isolation. Thanks,
Alex
prev parent reply other threads:[~2020-07-23 21:56 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-07-23 21:38 Could multiple PCIe devices in one IOMMU group be added to different KVM VM separately? Ross Deleon
2020-07-23 21:55 ` Alex Williamson [this message]
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=20200723155536.52655375@w520.home \
--to=alex.williamson@redhat.com \
--cc=kvm@vger.kernel.org \
--cc=rdeleon@marvell.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).