From: Alex Williamson <alex.williamson@redhat.com>
To: Chen Fan <chen.fan.fnst@cn.fujitsu.com>
Cc: izumi.taku@jp.fujitsu.com, qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [RFC v9 08/18] vfio: improve vfio_get_group to support adding group without devices
Date: Tue, 09 Jun 2015 15:23:01 -0600 [thread overview]
Message-ID: <1433884981.4927.143.camel@redhat.com> (raw)
In-Reply-To: <ed5cd2bb1f87eddcde9d54a3ebe32167a476aac3.1433812962.git.chen.fan.fnst@cn.fujitsu.com>
On Tue, 2015-06-09 at 11:37 +0800, Chen Fan wrote:
> Pre-adding all affected groups for aer devices, it could
> ensure the affected groups are owned in VM.
>
> Signed-off-by: Chen Fan <chen.fan.fnst@cn.fujitsu.com>
> ---
> hw/vfio/common.c | 15 ++++++++++++---
> 1 file changed, 12 insertions(+), 3 deletions(-)
>
> diff --git a/hw/vfio/common.c b/hw/vfio/common.c
> index b1045da..4230f83 100644
> --- a/hw/vfio/common.c
> +++ b/hw/vfio/common.c
> @@ -793,8 +793,15 @@ VFIOGroup *vfio_get_group(int groupid, AddressSpace *as)
>
> QLIST_FOREACH(group, &vfio_group_list, next) {
> if (group->groupid == groupid) {
> + if (as && !group->container) {
> + if (vfio_connect_container(group, as)) {
> + error_report("vfio: failed to setup container for group %d",
> + groupid);
> + return NULL;
> + }
> + }
> /* Found it. Now is it already in the right context? */
> - if (group->container->space->as == as) {
> + if (!as || group->container->space->as == as) {
> return group;
> } else {
> error_report("vfio: group %d used in multiple address spaces",
> @@ -828,7 +835,7 @@ VFIOGroup *vfio_get_group(int groupid, AddressSpace *as)
> group->groupid = groupid;
> QLIST_INIT(&group->device_list);
>
> - if (vfio_connect_container(group, as)) {
> + if (as && vfio_connect_container(group, as)) {
> error_report("vfio: failed to setup container for group %d", groupid);
> goto close_fd_exit;
> }
> @@ -859,7 +866,9 @@ void vfio_put_group(VFIOGroup *group)
> }
>
> vfio_kvm_device_del_group(group);
> - vfio_disconnect_container(group);
> + if (group->container) {
> + vfio_disconnect_container(group);
> + }
> QLIST_REMOVE(group, next);
> trace_vfio_put_group(group->fd);
> close(group->fd);
It's an interesting idea, but should we instead separate the container
from vfio_get/put_group() to be done as a separate step? Using the
AddressSpace pointer is a bit obfuscated.
Also, this gets you the group, but the vfio_group_get_external_user()
interface used by vfio-pci to do a bus reset requires that the group is
active, with an iommu. If it didn't, we could potentially be resetting
a non-viable group, with devices still owned by the host or not isolated
by the iommu. So you're a step closer, but if we go back to the example
of a dual port NIC with isolated functions, QEMU may be able to get the
group for the second port, it may even be a viable group, but if the
second port doesn't eventually get connected to the VM, the bus reset
won't be allowed.
next prev parent reply other threads:[~2015-06-09 21:23 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-06-09 3:37 [Qemu-devel] [RFC v9 00/18] vfio-pci: pass the aer error to guest Chen Fan
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 01/18] vfio: extract vfio_get_hot_reset_info as a single function Chen Fan
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 02/18] vfio: squeeze out vfio_pci_do_hot_reset for support bus reset Chen Fan
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 03/18] pcie: modify the capability size assert Chen Fan
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 04/18] vfio: make the 4 bytes aligned for capability size Chen Fan
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 05/18] vfio: add pcie extanded capability support Chen Fan
2015-06-09 21:22 ` Alex Williamson
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 06/18] aer: impove pcie_aer_init to support vfio device Chen Fan
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 07/18] vfio: add aer support for " Chen Fan
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 08/18] vfio: improve vfio_get_group to support adding group without devices Chen Fan
2015-06-09 21:23 ` Alex Williamson [this message]
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 09/18] vfio: add ref for group to support own affected groups Chen Fan
2015-06-09 21:23 ` Alex Williamson
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 10/18] get all affected groups for each device support aer Chen Fan
2015-06-09 21:22 ` Alex Williamson
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 11/18] vfio: add check host bus reset is support or not Chen Fan
2015-06-09 21:23 ` Alex Williamson
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 12/18] pci: add bus reset_notifiers callbacks for host bus reset Chen Fan
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 13/18] vfio: add sec_bus_reset notifier to notify physical bus reset is needed Chen Fan
2015-06-09 21:23 ` Alex Williamson
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 14/18] vfio: improve vfio_pci_hot_reset to support more case Chen Fan
2015-06-09 21:24 ` Alex Williamson
2015-06-16 8:10 ` Chen Fan
2015-06-16 14:08 ` Alex Williamson
2015-06-17 6:28 ` Chen Fan
2015-06-17 15:23 ` Alex Williamson
2015-06-18 10:27 ` Chen Fan
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 15/18] vfio: do hot bus reset when do virtual secondary bus reset Chen Fan
2015-06-09 21:23 ` Alex Williamson
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 16/18] pcie_aer: expose pcie_aer_msg() interface Chen Fan
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 17/18] vfio-pci: pass the aer error to guest Chen Fan
2015-06-09 3:37 ` [Qemu-devel] [RFC v9 18/18] vfio: add 'aer' property to expose aercap Chen Fan
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=1433884981.4927.143.camel@redhat.com \
--to=alex.williamson@redhat.com \
--cc=chen.fan.fnst@cn.fujitsu.com \
--cc=izumi.taku@jp.fujitsu.com \
--cc=qemu-devel@nongnu.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 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.