All of lore.kernel.org
 help / color / mirror / Atom feed
From: Wei Wang2 <wei.wang2@amd.com>
To: "Roedel, Joerg" <Joerg.Roedel@amd.com>
Cc: "M. Schneider" <ms@it-infrastrukturen.com>,
	"xen-devel@lists.xensource.com" <xen-devel@lists.xensource.com>
Subject: Re: IOMMU support for xen in BIOS of AMD based HP DL385g7 and kernel 3.0.0-rc7
Date: Tue, 26 Jul 2011 13:47:34 +0200	[thread overview]
Message-ID: <201107261347.35313.wei.wang2@amd.com> (raw)
In-Reply-To: <20110725090335.GG21948@amd.com>

On Monday 25 July 2011 11:03:35 Roedel, Joerg wrote:
> On Thu, Jul 21, 2011 at 11:53:50AM -0400, M. Schneider wrote:
> > Am 20.07.2011 14:45, schrieb Roedel, Joerg:
> > > On Sun, Jul 17, 2011 at 06:34:33AM -0400, M. Schneider wrote:
> > >> Please allow me one question about IOMMU support in AMDS's 890FX
> > >> chipset and 6176SE CPUs used in HP DL385g7 servers. These chips
> > >> support IOMMU as far as I know. BIOS of HP DL385g7 doesn't currently
> > >> offer an IOMMU flag.
> > >>
> > >> I have contacted already HP support due to this missing BIOS setting
> > >> options but less or more my request has been ignored (no solution and
> > >> no plans for solving of this important issue  yet) so I am looking for
> > >> a workaround as IOMMU support is required for passing thoruhg of PCI
> > >> devices to domU domains (xen 4.0 / 4.1).
> > >
> > > I am sorry but if there is no IVRS table there is not much we can do.
> > > The IVRS table contains the information which device uses which
> > > request-ids. Without this information the IOMMU can not work reliably.
> >
> > Thank you very much fo your answer Jörg.
> >
> > Some BIOSes (like BIOS of HD DL385g7 server with AMD CPUs) don't have
> > the IOMMU option.
> > I have to correct my prevoius email as HP DL385g7 servers have the AMD
> > SR5690/SP5100 chipset (supporting IOMMU).
> >
> > Is it possible to write a workaround for xen 4.1 or 4.2 to create such
> > IVRS table in code (in case it has static entries for specific server
> > type) and add a boot option like "iommu=force-dl385g7" ?
> >
> > Thank you in advance for any hints.
>
> This is basically possible. The Xen code needs to do what the BIOS fails
> to do. But for Xen this is a lot more difficult to get right because Xen
> can not make the same assumptions about the hardware like a BIOS can.  I
> can easily imagine that this will end in a maintenance nighmare if
> people start using this. I personally wouldn't do that for Linux, but it
> is up to the Xen IOMMU maintainer to decide that for the Xen side. I put
> Wei Wang on Cc, he maintains AMD IOMMU for Xen.

Xen also relies on BIOS to setup IVRS table for iommu device. You could also 
look at Xen 3.2, there were some old bits that enables iommu directly using 
pci configure space. But just as Joerg said, this was not an optimal and 
reliable way to enable iommu, we do not maintain these codes any more.
Thanks,
Wei

> Regards,
>
> 	Joerg

      parent reply	other threads:[~2011-07-26 11:47 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-07-17 10:34 IOMMU support for xen in BIOS of AMD based HP DL385g7 and kernel 3.0.0-rc7 M. Schneider
2011-07-20 12:45 ` Roedel, Joerg
2011-07-21 15:53   ` M. Schneider
2011-07-25  9:03     ` Roedel, Joerg
2011-07-25  8:44       ` M. Schneider
2011-07-26 11:47       ` Wei Wang2 [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=201107261347.35313.wei.wang2@amd.com \
    --to=wei.wang2@amd.com \
    --cc=Joerg.Roedel@amd.com \
    --cc=ms@it-infrastrukturen.com \
    --cc=xen-devel@lists.xensource.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.