From: Avi Kivity <avi@redhat.com>
To: PICT PUNE <kvm.smartshare@gmail.com>
Cc: Anthony Liguori <anthony@codemonkey.ws>,
virtualization@lists.linux-foundation.org
Subject: Re: Help - KVM TODO_NOV2008_Sharing Memory between guests
Date: Sun, 08 Mar 2009 16:21:14 +0200 [thread overview]
Message-ID: <49B3D45A.8030200@redhat.com> (raw)
In-Reply-To: <b20fe0330903040612m784b273ew7bca3c8ad087990a@mail.gmail.com>
PICT PUNE wrote:
>
>
> On Wed, Feb 25, 2009 at 9:11 PM, Anthony Liguori
> <anthony@codemonkey.ws <mailto:anthony@codemonkey.ws>> wrote:
>
> Avi Kivity wrote:
>
> PICT PUNE wrote:
>
> In reference to the November 2008 todo list on Qumranet
> Website
>
> we are a group of college students working on this todo:
>
> *Add a qemu interface for sharing memory between guests.
> Using a pci device to expose the shared memory is probably
> a good starting point.
> *
> Can anyone suggest us some starting point or a possible
> approach for the same...
>
>
> Write a PCI device that exposes a BAR containing RAM.
> Implement the RAM by mapping a shared memory segment or a
> file. Write a guest driver to use this shared RAM.
>
>
> Or implement a virtio device that can advertise available shared
> memory segments. Have guest allocate its own memory to send to
> host that becomes the shared memory segment.
>
> Expose shared memory mappings via a virtual file system (similar
> to hugetlbfs or ramfs).
>
> Using PCI BARs implies static shared memory mappings. For a long
> running VM, you're likely to want to support dynamic shared memory
> mappings.
>
> Also exposing a simple signaling mechanism with this too would
> allow for shared ring queues to be implemented in userspace
> between guests.
>
> Regards,
>
> Anthony Liguori
>
>
> Hello Mr. Avi Kivity and Mr. Anthony Liguori
>
> Thanks for your early reply. We expecting the same approach.
There were two approaches outlined above. Which one do you mean?
> We are in process of making clusters of free memory and reserve these
> memory as a fake PCI device for the guest OS. this device will be
> then passed using PCI passthrough to the guest from the host.
PCI passthrough does not make much sense in this context, is this what
you meant?
> Can you please help us in faking the free memory cluster as a PCI
> device. Thanking in anticipation.
>
I suggest reading the code of framebuffer devices such as vga.c.
--
error compiling committee.c: too many arguments to function
prev parent reply other threads:[~2009-03-08 14:21 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-02-21 13:35 Help - KVM TODO_NOV2008_Sharing Memory between guests PICT PUNE
2009-02-25 13:38 ` Avi Kivity
2009-02-25 15:41 ` Anthony Liguori
2009-03-04 14:12 ` PICT PUNE
2009-03-08 14:21 ` Avi Kivity [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=49B3D45A.8030200@redhat.com \
--to=avi@redhat.com \
--cc=anthony@codemonkey.ws \
--cc=kvm.smartshare@gmail.com \
--cc=virtualization@lists.linux-foundation.org \
/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).