From: Jason Wang <jasowang@redhat.com>
To: Eugenio Perez Martin <eperezma@redhat.com>
Cc: Eli Cohen <elic@nvidia.com>, Cindy Lu <lulu@redhat.com>,
qemu-level <qemu-devel@nongnu.org>,
lingshan.zhu@intel.com, Michael Tsirkin <mst@redhat.com>
Subject: Re: [PATCH 18/18] vhost-vdpa: multiqueue support
Date: Thu, 1 Jul 2021 16:15:34 +0800 [thread overview]
Message-ID: <4a9981c4-be51-d221-0b11-0d41376b2b5b@redhat.com> (raw)
In-Reply-To: <CAJaqyWeT+VhXSzu9VA7UrJMFeOCUwNXUoN9-yWZzp9Rg4pBZWQ@mail.gmail.com>
在 2021/7/1 下午2:51, Eugenio Perez Martin 写道:
> On Mon, Jun 21, 2021 at 6:18 AM Jason Wang <jasowang@redhat.com> wrote:
>> This patch implements the multiqueue support for vhost-vdpa. This is
>> done simply by reading the number of queue pairs from the config space
>> and initialize the datapath and control path net client.
>>
>> Signed-off-by: Jason Wang <jasowang@redhat.com>
>> ---
>> hw/net/virtio-net.c | 3 +-
>> net/vhost-vdpa.c | 98 ++++++++++++++++++++++++++++++++++++++++-----
>> 2 files changed, 91 insertions(+), 10 deletions(-)
>>
>> diff --git a/hw/net/virtio-net.c b/hw/net/virtio-net.c
>> index 5074b521cf..2c2ed98c0b 100644
>> --- a/hw/net/virtio-net.c
>> +++ b/hw/net/virtio-net.c
>> @@ -3370,7 +3370,8 @@ static void virtio_net_device_realize(DeviceState *dev, Error **errp)
>>
>> n->max_ncs = MAX(n->nic_conf.peers.queues, 1);
>>
>> - /* Figure out the datapath queue pairs since the bakcend could
>> + /*
>> + * Figure out the datapath queue pairs since the bakcend could
> If we are going to modify the comment we could s/bakcend/backend/.
Will fix.
>
>> * provide control queue via peers as well.
>> */
>> if (n->nic_conf.peers.queues) {
>> diff --git a/net/vhost-vdpa.c b/net/vhost-vdpa.c
>> index cc11b2ec40..048344b4bc 100644
>> --- a/net/vhost-vdpa.c
>> +++ b/net/vhost-vdpa.c
>> @@ -18,6 +18,7 @@
>> #include "qemu/error-report.h"
>> #include "qemu/option.h"
>> #include "qapi/error.h"
>> +#include <linux/vhost.h>
>> #include <sys/ioctl.h>
>> #include <err.h>
>> #include "standard-headers/linux/virtio_net.h"
>> @@ -52,6 +53,8 @@ const int vdpa_feature_bits[] = {
>> VIRTIO_NET_F_HOST_UFO,
>> VIRTIO_NET_F_MRG_RXBUF,
>> VIRTIO_NET_F_MTU,
>> + VIRTIO_NET_F_MQ,
>> + VIRTIO_NET_F_CTRL_VQ,
>
> Hi!
>
> I'm not sure if it's qemu the one that must control it, but I cannot
> use vdpa_sim of linux 5.13 (i.e., with no control vq patches) with
> this series applied:
>
> [ 3.967421] virtio_net virtio0: device advertises feature
> VIRTIO_NET_F_CTRL_RX but not VIRTIO_NET_F_CTRL_VQ
> [ 3.968613] virtio_net: probe of virtio0 failed with error -22
Interesting, looks like a bug somewhere.
We never advertise CTRL_RX in the case of simulator.
>
> Did you mention it somewhere else and I've missed it? or is it
> actually a bug in the device? In this second case, I think we should
> still workaround it in qemu, because old vdpasim_net with no
> VIRTIO_NET_F_CTRL_VQ still works ok without this patch.
Should be a bug, will have a look.
Thanks
>
> Thanks!
>
>> VIRTIO_F_IOMMU_PLATFORM,
>> VIRTIO_F_RING_PACKED,
>> VIRTIO_NET_F_RSS,
>> @@ -82,7 +85,8 @@ static int vhost_vdpa_net_check_device_id(struct vhost_net *net)
>> return ret;
>> }
>>
>> -static int vhost_vdpa_add(NetClientState *ncs, void *be)
>> +static int vhost_vdpa_add(NetClientState *ncs, void *be, int qp_index,
>> + int nvqs)
>> {
>> VhostNetOptions options;
>> struct vhost_net *net = NULL;
>> @@ -95,7 +99,7 @@ static int vhost_vdpa_add(NetClientState *ncs, void *be)
>> options.net_backend = ncs;
>> options.opaque = be;
>> options.busyloop_timeout = 0;
>> - options.nvqs = 2;
>> + options.nvqs = nvqs;
>>
>> net = vhost_net_init(&options);
>> if (!net) {
>> @@ -159,18 +163,28 @@ static NetClientInfo net_vhost_vdpa_info = {
>> static NetClientState *net_vhost_vdpa_init(NetClientState *peer,
>> const char *device,
>> const char *name,
>> - int vdpa_device_fd)
>> + int vdpa_device_fd,
>> + int qp_index,
>> + int nvqs,
>> + bool is_datapath)
>> {
>> NetClientState *nc = NULL;
>> VhostVDPAState *s;
>> int ret = 0;
>> assert(name);
>> - nc = qemu_new_net_client(&net_vhost_vdpa_info, peer, device, name);
>> + if (is_datapath) {
>> + nc = qemu_new_net_client(&net_vhost_vdpa_info, peer, device,
>> + name);
>> + } else {
>> + nc = qemu_new_net_control_client(&net_vhost_vdpa_info, peer,
>> + device, name);
>> + }
>> snprintf(nc->info_str, sizeof(nc->info_str), TYPE_VHOST_VDPA);
>> s = DO_UPCAST(VhostVDPAState, nc, nc);
>>
>> s->vhost_vdpa.device_fd = vdpa_device_fd;
>> - ret = vhost_vdpa_add(nc, (void *)&s->vhost_vdpa);
>> + s->vhost_vdpa.index = qp_index;
>> + ret = vhost_vdpa_add(nc, (void *)&s->vhost_vdpa, qp_index, nvqs);
>> if (ret) {
>> qemu_del_net_client(nc);
>> return NULL;
>> @@ -196,12 +210,52 @@ static int net_vhost_check_net(void *opaque, QemuOpts *opts, Error **errp)
>> return 0;
>> }
>>
>> +static int vhost_vdpa_get_max_qps(int fd, int *has_cvq, Error **errp)
>> +{
>> + unsigned long config_size = offsetof(struct vhost_vdpa_config, buf);
>> + struct vhost_vdpa_config *config;
>> + __virtio16 *max_qps;
>> + uint64_t features;
>> + int ret;
>> +
>> + ret = ioctl(fd, VHOST_GET_FEATURES, &features);
>> + if (ret) {
>> + error_setg(errp, "Fail to query features from vhost-vDPA device");
>> + return ret;
>> + }
>> +
>> + if (features & (1 << VIRTIO_NET_F_CTRL_VQ)) {
>> + *has_cvq = 1;
>> + } else {
>> + *has_cvq = 0;
>> + }
>> +
>> + if (features & (1 << VIRTIO_NET_F_MQ)) {
>> + config = g_malloc0(config_size + sizeof(*max_qps));
>> + config->off = offsetof(struct virtio_net_config, max_virtqueue_pairs);
>> + config->len = sizeof(*max_qps);
>> +
>> + ret = ioctl(fd, VHOST_VDPA_GET_CONFIG, config);
>> + if (ret) {
>> + error_setg(errp, "Fail to get config from vhost-vDPA device");
>> + return -ret;
>> + }
>> +
>> + max_qps = (__virtio16 *)&config->buf;
>> +
>> + return lduw_le_p(max_qps);
>> + }
>> +
>> + return 1;
>> +}
>> +
>> int net_init_vhost_vdpa(const Netdev *netdev, const char *name,
>> NetClientState *peer, Error **errp)
>> {
>> const NetdevVhostVDPAOptions *opts;
>> int vdpa_device_fd;
>> - NetClientState *nc;
>> + NetClientState **ncs, *nc;
>> + int qps, i, has_cvq = 0;
>>
>> assert(netdev->type == NET_CLIENT_DRIVER_VHOST_VDPA);
>> opts = &netdev->u.vhost_vdpa;
>> @@ -216,11 +270,37 @@ int net_init_vhost_vdpa(const Netdev *netdev, const char *name,
>> return -errno;
>> }
>>
>> - nc = net_vhost_vdpa_init(peer, TYPE_VHOST_VDPA, name, vdpa_device_fd);
>> - if (!nc) {
>> + qps = vhost_vdpa_get_max_qps(vdpa_device_fd, &has_cvq, errp);
>> + if (qps < 0) {
>> qemu_close(vdpa_device_fd);
>> - return -1;
>> + return qps;
>> + }
>> +
>> + ncs = g_malloc0(sizeof(*ncs) * qps);
>> +
>> + for (i = 0; i < qps; i++) {
>> + ncs[i] = net_vhost_vdpa_init(peer, TYPE_VHOST_VDPA, name,
>> + vdpa_device_fd, i, 2, true);
>> + if (!ncs[i])
>> + goto err;
>> }
>>
>> + if (has_cvq) {
>> + nc = net_vhost_vdpa_init(peer, TYPE_VHOST_VDPA, name,
>> + vdpa_device_fd, i, 1, false);
>> + if (!nc)
>> + goto err;
>> + }
>> +
>> + g_free(ncs);
>> return 0;
>> +
>> +err:
>> + if (i) {
>> + qemu_del_net_client(ncs[0]);
>> + }
>> + qemu_close(vdpa_device_fd);
>> + g_free(ncs);
>> +
>> + return -1;
>> }
>> --
>> 2.25.1
>>
next prev parent reply other threads:[~2021-07-01 8:16 UTC|newest]
Thread overview: 51+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-06-21 4:16 [PATCH 00/18] vhost-vDPA multiqueue Jason Wang
2021-06-21 4:16 ` [PATCH 01/18] vhost_net: remove the meaningless assignment in vhost_net_start_one() Jason Wang
2021-06-21 11:45 ` Eli Cohen
2021-06-24 7:42 ` Jason Wang
2021-06-21 4:16 ` [PATCH 02/18] vhost: use unsigned int for nvqs Jason Wang
2021-06-21 11:46 ` Eli Cohen
2021-06-21 4:16 ` [PATCH 03/18] vhost_net: do not assume nvqs is always 2 Jason Wang
2021-06-23 14:49 ` Stefano Garzarella
2021-06-24 6:22 ` Eli Cohen
2021-06-24 7:42 ` Jason Wang
2021-06-21 4:16 ` [PATCH 04/18] vhost-vdpa: remove the unnecessary check in vhost_vdpa_add() Jason Wang
2021-06-23 14:53 ` Stefano Garzarella
2021-06-24 6:38 ` Eli Cohen
2021-06-24 7:46 ` Jason Wang
2021-06-21 4:16 ` [PATCH 05/18] vhost-vdpa: don't cleanup twice " Jason Wang
2021-06-23 14:56 ` Stefano Garzarella
2021-06-21 4:16 ` [PATCH 06/18] vhost-vdpa: fix leaking of vhost_net " Jason Wang
2021-06-23 15:00 ` Stefano Garzarella
2021-06-24 7:06 ` Eli Cohen
2021-06-24 7:10 ` Jason Wang
2021-06-24 7:32 ` Eli Cohen
2021-06-24 7:14 ` Eli Cohen
2021-06-24 7:41 ` Jason Wang
2021-06-21 4:16 ` [PATCH 07/18] vhost-vdpa: tweak the error label " Jason Wang
2021-06-23 15:03 ` Stefano Garzarella
2021-07-06 8:03 ` Jason Wang
2021-07-06 8:10 ` Jason Wang
2021-07-06 8:27 ` Stefano Garzarella
2021-07-06 8:28 ` Jason Wang
2021-06-21 4:16 ` [PATCH 08/18] vhost-vdpa: fix the wrong assertion in vhost_vdpa_init() Jason Wang
2021-06-23 15:04 ` Stefano Garzarella
2021-06-21 4:16 ` [PATCH 09/18] vhost-vdpa: remove the unncessary queue_index assignment Jason Wang
2021-06-23 15:05 ` Stefano Garzarella
2021-06-21 4:16 ` [PATCH 10/18] vhost-vdpa: open device fd in net_init_vhost_vdpa() Jason Wang
2021-06-23 15:07 ` Stefano Garzarella
2021-06-21 4:16 ` [PATCH 11/18] vhost-vdpa: classify one time request Jason Wang
2021-06-21 4:16 ` [PATCH 12/18] vhost-vdpa: prepare for the multiqueue support Jason Wang
2021-06-21 4:16 ` [PATCH 13/18] vhost-vdpa: let net_vhost_vdpa_init() returns NetClientState * Jason Wang
2021-06-21 4:16 ` [PATCH 14/18] net: introduce control client Jason Wang
2021-06-21 4:16 ` [PATCH 15/18] vhost-net: control virtqueue support Jason Wang
2021-06-24 7:42 ` Eli Cohen
2021-06-24 7:44 ` Jason Wang
2021-06-30 17:33 ` Eugenio Perez Martin
2021-07-01 3:03 ` Jason Wang
2021-06-21 4:16 ` [PATCH 16/18] virito-net: use "qps" instead of "queues" when possible Jason Wang
2021-06-21 4:16 ` [PATCH 17/18] virtio-net: vhost control virtqueue support Jason Wang
2021-06-21 4:16 ` [PATCH 18/18] vhost-vdpa: multiqueue support Jason Wang
2021-07-01 6:51 ` Eugenio Perez Martin
2021-07-01 8:15 ` Jason Wang [this message]
2021-07-06 7:46 ` Jason Wang
2021-06-21 4:33 ` [PATCH 00/18] vhost-vDPA multiqueue no-reply
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=4a9981c4-be51-d221-0b11-0d41376b2b5b@redhat.com \
--to=jasowang@redhat.com \
--cc=elic@nvidia.com \
--cc=eperezma@redhat.com \
--cc=lingshan.zhu@intel.com \
--cc=lulu@redhat.com \
--cc=mst@redhat.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 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).