From: Yosuke Iwamatsu <y-iwamatsu@ab.jp.nec.com>
To: barak@neocleus.com
Cc: xen-devel@lists.xensource.com
Subject: Re: Question about MMIO Region Alignment for PCI Pass-through
Date: Wed, 12 Mar 2008 20:48:57 +0900 [thread overview]
Message-ID: <47D7C329.2060706@ab.jp.nec.com> (raw)
In-Reply-To: <9392A06CB0FDC847B3A530B3DC174E7B04D3ED22@mse10be1.mse10.exchange.ms>
Hi Barak,
Thank you for your reply.
Barak Fargoun wrote:
> Hi!
>
> I have created the 'pci-mem-align' patch. The version posted before, had
> some bugs. However, I have fixed those issues, and it works fine for me
> now.
What kinds of bugs did you fix?
It would be nice if I could see your new patch.
> BTW, why are you interested in this?
Because I'm recently working a bit on PV PCI passthrough.
I haven't encountered the mmio region conflict yet, but I think it is a
a potential problem for all the passthrough methods
(VT-d/IOMMU, PV driver domain and nativedom).
Regards,
------------------
Yosuke Iwamatsu
>
> Barak
>
> -----Original Message-----
> From: xen-devel-bounces@lists.xensource.com
> [mailto:xen-devel-bounces@lists.xensource.com] On Behalf Of Yosuke
> Iwamatsu
> Sent: Wednesday, March 05, 2008 11:34 AM
> To: xen-devel@lists.xensource.com
> Subject: [Xen-devel] Question about MMIO Region Alignment for
> PCIPass-through
>
> Hi there,
>
> I have a question about the mmio region access of PCI pass-trough
> devices.
>
> When assigning a PCI device to a guest, pages that contain the device
> mmio regions are mapped to the guest. Then the guest can directly issue
> mmio instructions. However, there is a problem that, if a single page
> has mmio regions of multiple devices, we cannot expose it to the guest.
> So we need to align PCI memory regions to page size (4K) in dom0 kernel
> and make sure that each page has only one device's mmio region.
>
> I found that this problem was discussed a bit in the following thread.
> http://lists.xensource.com/archives/html/xen-devel/2007-10/msg01012.html
> (linux-2.6.18-xen.hg: changest 296)
> Dom0 parameter "pci-mem-align" seems to help aligning mmio regions.
> But, at a quick look, mmio regions appear to be aligned not to PAGE_SIZE
> but to PAGE_SIZE-1. I also have a concern that re-mapping all the PCI
> resources might do harm to some devices unrelated to PCI pass-through.
>
> Does anyone have an idea about this mmio re-mapping issue ?
> (or is anyone already working on this ?)
>
> Regards,
> -------------------
> Yosuke Iwamatsu
> NEC Corporation
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
prev parent reply other threads:[~2008-03-12 11:48 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-03-05 9:33 Question about MMIO Region Alignment for PCI Pass-through Yosuke Iwamatsu
[not found] ` <9392A06CB0FDC847B3A530B3DC174E7B04D3ED22@mse10be1.mse10.exchange.ms>
2008-03-12 11:48 ` Yosuke Iwamatsu [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=47D7C329.2060706@ab.jp.nec.com \
--to=y-iwamatsu@ab.jp.nec.com \
--cc=barak@neocleus.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.