From: Stefan Hajnoczi <stefanha@gmail.com>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: qemu-devel@nongnu.org, stefanha@redhat.com
Subject: Re: [Qemu-devel] [PATCH 0/4] dataplane: use more of the generic virtio data structures, drop hostmem
Date: Wed, 4 Dec 2013 15:12:53 +0100 [thread overview]
Message-ID: <20131204141253.GB27759@stefanha-thinkpad.redhat.com> (raw)
In-Reply-To: <1381417639-22547-1-git-send-email-pbonzini@redhat.com>
On Thu, Oct 10, 2013 at 05:07:15PM +0200, Paolo Bonzini wrote:
> Now that the memory API is thread-safe, we can use it in
> virtio-blk-dataplane and replace hostmem.[ch]. This series does this,
> and also changes the vring API to use VirtQueueElement (with an eye
> towards migration). With this change, virtio-blk-dataplane is also safe
> against memory hot-unplug.
>
> The next step would be to replace memory_region_find with
> address_space_{map,unmap}, which handle dirtying of memory correctly.
> However these APIs are not thread-safe yet, and neither is the handling
> of dirty memory (Juan's patches may be a start here).
>
> Also, the usage of iov_discard_{front,back} may cause some complication
> when we use address_space_{map,unmap}. We may have to change a bit the
> logic in virtio-blk-dataplane to switch to address_space_{map,unmap}.
>
> If we do not want to do this intermediate step, the first three patches
> can be applied separately from the fourth.
>
> Paolo Bonzini (4):
> vring: create a common function to parse descriptors
> vring: factor common code for error exits
> dataplane: change vring API to use VirtQueueElement
> dataplane: replace hostmem with memory_region_find
>
> hw/block/dataplane/virtio-blk.c | 86 +++++-------
> hw/virtio/dataplane/Makefile.objs | 2 +-
> hw/virtio/dataplane/hostmem.c | 183 -------------------------
> hw/virtio/dataplane/vring.c | 244 +++++++++++++++++++++-------------
> include/hw/virtio/dataplane/hostmem.h | 58 --------
> include/hw/virtio/dataplane/vring.h | 9 +-
> 6 files changed, 193 insertions(+), 389 deletions(-)
> delete mode 100644 hw/virtio/dataplane/hostmem.c
> delete mode 100644 include/hw/virtio/dataplane/hostmem.h
Finally looked into this series. I think we should merge all patches so
we start exercising the thread-safe memory API as soon as possible in
the QEMU 2.0 release cycle.
Before merging I wanted to discuss the vring->broken case which I've
left a comment on.
Stefan
prev parent reply other threads:[~2013-12-04 14:13 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-10-10 15:07 [Qemu-devel] [PATCH 0/4] dataplane: use more of the generic virtio data structures, drop hostmem Paolo Bonzini
2013-10-10 15:07 ` [Qemu-devel] [PATCH 1/4] vring: create a common function to parse descriptors Paolo Bonzini
2013-10-10 15:07 ` [Qemu-devel] [PATCH 2/4] vring: factor common code for error exits Paolo Bonzini
2013-10-10 20:53 ` Richard Henderson
2013-10-11 9:19 ` Paolo Bonzini
2013-10-10 15:07 ` [Qemu-devel] [PATCH 3/4] dataplane: change vring API to use VirtQueueElement Paolo Bonzini
2013-12-04 14:06 ` Stefan Hajnoczi
2013-12-04 17:40 ` Paolo Bonzini
2013-12-05 9:24 ` Stefan Hajnoczi
2013-12-05 10:34 ` Paolo Bonzini
2013-12-06 9:02 ` Stefan Hajnoczi
2013-10-10 15:07 ` [Qemu-devel] [PATCH 4/4] dataplane: replace hostmem with memory_region_find Paolo Bonzini
2013-12-04 14:12 ` Stefan Hajnoczi [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=20131204141253.GB27759@stefanha-thinkpad.redhat.com \
--to=stefanha@gmail.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=stefanha@redhat.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).