From: Avi Kivity <avi@redhat.com>
To: kvm@zavalon.com
Cc: kvm@vger.kernel.org
Subject: Re: shared memory between host and VMs (windows xp) using virtio
Date: Sun, 25 Oct 2009 16:03:00 +0200 [thread overview]
Message-ID: <4AE45A94.7060902@redhat.com> (raw)
In-Reply-To: <1256290163.24460.58.camel@roo>
On 10/23/2009 11:29 AM, Nev wrote:
> I need to be able to share memory between guest VMs and the host. I have
> been using VMware workstation 6.0.5. As Cam Macdonell stated in one of
> his posts, VMware dropped the shared memory as part of their VMCI
> interface in later versions. I am the person he refered to in his post
> that is running simulation across multiple VMs.
>
> My requirements are different from Cam's.
> 1. I need to be able to dynamically allocate multiple blocks of shared
> memory of different sizes. Size of each shared block can vary from a few
> bytes to a few MBytes
> 2. I need to be able to have any number of Guest, include zero, and any
> number of host processes, including zero, sharing the memory.
> 3. The virtual address used to access the memory *can* be different for
> each host process and guest process. eg no requirements
> 4. I dont need interrupts.
>
Probably best served with a single shared area with an internal memory
allocator.
> I *think* I will need a host kernel module (device drive?) to service
> the requests.
> The module needs to be accessible from use level host code, and user
> guest code via a Guest device driver, or service and then via virtio to
> the host module.
>
>
No need for host kernel modifications (as Cam's work shows).
> I am assuming there is little or no documentation for virtio, and I need
> to look at the code.
>
> I would like a little help to get started.
> 1. Should I be looking at quemu-kvm or kvm or somewhere else?
> 2. Can I get some hints at what function to look for and where?
> 3. If there is any documentation please indicate where I might find it.
>
virtio is mostly useful for dynamic dma from guest memory; for static
sharing of host memory it isn't a good fit.
> I would prefer to *not* modify the existing kvm code, but would prefer
> to write stand allow modules that interface to the kvm code. I will be
> happy to make the source for these modules publicly available.
>
> Any help will be greatly appreciated.
>
Best starting point is Cam's patches, IMO.
--
error compiling committee.c: too many arguments to function
prev parent reply other threads:[~2009-10-25 14:02 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-10-23 9:29 shared memory between host and VMs (windows xp) using virtio Nev
2009-10-25 14:03 ` 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=4AE45A94.7060902@redhat.com \
--to=avi@redhat.com \
--cc=kvm@vger.kernel.org \
--cc=kvm@zavalon.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