From: Maxime Coquelin <maxime.coquelin@redhat.com>
To: Srujana Challa <schalla@marvell.com>,
"dev@dpdk.org" <dev@dpdk.org>,
"chenbox@nvidia.com" <chenbox@nvidia.com>
Cc: Jerin Jacob <jerinj@marvell.com>,
Nithin Kumar Dabilpuram <ndabilpuram@marvell.com>,
Vamsi Krishna Attunuru <vattunuru@marvell.com>
Subject: Re: [EXTERNAL] Re: [PATCH v2 1/3] net/virtio_user: avoid cq descriptor buffer address accessing
Date: Tue, 2 Jul 2024 14:41:33 +0200 [thread overview]
Message-ID: <62446aea-34f3-4fbc-ad42-ddd6b57e5864@redhat.com> (raw)
In-Reply-To: <DS0PR18MB5368A531DA1BD3CCBB9381E6A0DC2@DS0PR18MB5368.namprd18.prod.outlook.com>
On 7/2/24 13:09, Srujana Challa wrote:
>> On 2/29/24 14:29, Srujana Challa wrote:
>>> This patch makes changes to avoid descriptor buffer address accessing
>>> while processing shadow control queue.
>>> So that Virtio-user can work with having IOVA as descriptor buffer
>>> address.
>>>
>>> Signed-off-by: Srujana Challa <schalla@marvell.com>
>>> ---
>>> .../net/virtio/virtio_user/virtio_user_dev.c | 68 +++++++++----------
>>> 1 file changed, 33 insertions(+), 35 deletions(-)
>>>
>>> diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c
>>> b/drivers/net/virtio/virtio_user/virtio_user_dev.c
>>> index d395fc1676..bf3da4340f 100644
>>> --- a/drivers/net/virtio/virtio_user/virtio_user_dev.c
>>> +++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c
>>> @@ -885,11 +885,11 @@ static uint32_t
>>> virtio_user_handle_ctrl_msg_split(struct virtio_user_dev *dev, struct vring
>> *vring,
>>> uint16_t idx_hdr)
>>> {
>>> - struct virtio_net_ctrl_hdr *hdr;
>>> virtio_net_ctrl_ack status = ~0;
>>> - uint16_t i, idx_data, idx_status;
>>> + uint16_t i, idx_data;
>>> uint32_t n_descs = 0;
>>> int dlen[CVQ_MAX_DATA_DESCS], nb_dlen = 0;
>>> + struct virtio_pmd_ctrl *ctrl;
>>>
>>> /* locate desc for header, data, and status */
>>> idx_data = vring->desc[idx_hdr].next; @@ -902,34 +902,33 @@
>>> virtio_user_handle_ctrl_msg_split(struct virtio_user_dev *dev, struct vring
>> *vri
>>> n_descs++;
>>> }
>>>
>>> - /* locate desc for status */
>>> - idx_status = i;
>>> n_descs++;
>>>
>>> - hdr = (void *)(uintptr_t)vring->desc[idx_hdr].addr;
>>> - if (hdr->class == VIRTIO_NET_CTRL_MQ &&
>>> - hdr->cmd == VIRTIO_NET_CTRL_MQ_VQ_PAIRS_SET) {
>>> - uint16_t queues;
>>> + /* Access control command via VA from CVQ */
>>> + ctrl = (struct virtio_pmd_ctrl *)dev->hw.cvq->hdr_mz->addr;
>>> + if (ctrl->hdr.class == VIRTIO_NET_CTRL_MQ &&
>>> + ctrl->hdr.cmd == VIRTIO_NET_CTRL_MQ_VQ_PAIRS_SET) {
>>> + uint16_t *queues;
>>
>> This is not future proof, as it just discards the index and assume the buffer will
>> always be at the same place.
>> We should find a way to perform the desc address translation.
> Can we use rte_mem_iova2virt() here?
It should be safe to use it here.
Can you send a new revision ASAP, which would use this API and not take
the shortcurt, i.e. keep fetching buffer addres from descriptors?
Thanks,
Maxime
>
>>
>>>
>>> - queues = *(uint16_t *)(uintptr_t)vring->desc[idx_data].addr;
>>> - status = virtio_user_handle_mq(dev, queues);
>>> - } else if (hdr->class == VIRTIO_NET_CTRL_MQ && hdr->cmd ==
>> VIRTIO_NET_CTRL_MQ_RSS_CONFIG) {
>>> + queues = (uint16_t *)ctrl->data;
>>> + status = virtio_user_handle_mq(dev, *queues);
>>> + } else if (ctrl->hdr.class == VIRTIO_NET_CTRL_MQ &&
>>> + ctrl->hdr.cmd == VIRTIO_NET_CTRL_MQ_RSS_CONFIG) {
>>> struct virtio_net_ctrl_rss *rss;
>>>
>>> - rss = (struct virtio_net_ctrl_rss *)(uintptr_t)vring-
>>> desc[idx_data].addr;
>>> + rss = (struct virtio_net_ctrl_rss *)ctrl->data;
>>> status = virtio_user_handle_mq(dev, rss->max_tx_vq);
>>> - } else if (hdr->class == VIRTIO_NET_CTRL_RX ||
>>> - hdr->class == VIRTIO_NET_CTRL_MAC ||
>>> - hdr->class == VIRTIO_NET_CTRL_VLAN) {
>>> + } else if (ctrl->hdr.class == VIRTIO_NET_CTRL_RX ||
>>> + ctrl->hdr.class == VIRTIO_NET_CTRL_MAC ||
>>> + ctrl->hdr.class == VIRTIO_NET_CTRL_VLAN) {
>>> status = 0;
>>> }
>>>
>>> if (!status && dev->scvq)
>>> - status = virtio_send_command(&dev->scvq->cq,
>>> - (struct virtio_pmd_ctrl *)hdr, dlen, nb_dlen);
>>> + status = virtio_send_command(&dev->scvq->cq, ctrl, dlen,
>> nb_dlen);
>>>
>>> /* Update status */
>>> - *(virtio_net_ctrl_ack *)(uintptr_t)vring->desc[idx_status].addr =
>> status;
>>> + ctrl->status = status;
>>>
>>> return n_descs;
>>> }
>>> @@ -948,7 +947,7 @@ virtio_user_handle_ctrl_msg_packed(struct
>> virtio_user_dev *dev,
>>> struct vring_packed *vring,
>>> uint16_t idx_hdr)
>>> {
>>> - struct virtio_net_ctrl_hdr *hdr;
>>> + struct virtio_pmd_ctrl *ctrl;
>>> virtio_net_ctrl_ack status = ~0;
>>> uint16_t idx_data, idx_status;
>>> /* initialize to one, header is first */ @@ -971,32 +970,31 @@
>>> virtio_user_handle_ctrl_msg_packed(struct virtio_user_dev *dev,
>>> n_descs++;
>>> }
>>>
>>> - hdr = (void *)(uintptr_t)vring->desc[idx_hdr].addr;
>>> - if (hdr->class == VIRTIO_NET_CTRL_MQ &&
>>> - hdr->cmd == VIRTIO_NET_CTRL_MQ_VQ_PAIRS_SET) {
>>> - uint16_t queues;
>>> + /* Access control command via VA from CVQ */
>>> + ctrl = (struct virtio_pmd_ctrl *)dev->hw.cvq->hdr_mz->addr;
>>> + if (ctrl->hdr.class == VIRTIO_NET_CTRL_MQ &&
>>> + ctrl->hdr.cmd == VIRTIO_NET_CTRL_MQ_VQ_PAIRS_SET) {
>>> + uint16_t *queues;
>>>
>>> - queues = *(uint16_t *)(uintptr_t)
>>> - vring->desc[idx_data].addr;
>>> - status = virtio_user_handle_mq(dev, queues);
>>> - } else if (hdr->class == VIRTIO_NET_CTRL_MQ && hdr->cmd ==
>> VIRTIO_NET_CTRL_MQ_RSS_CONFIG) {
>>> + queues = (uint16_t *)ctrl->data;
>>> + status = virtio_user_handle_mq(dev, *queues);
>>> + } else if (ctrl->hdr.class == VIRTIO_NET_CTRL_MQ &&
>>> + ctrl->hdr.cmd == VIRTIO_NET_CTRL_MQ_RSS_CONFIG) {
>>> struct virtio_net_ctrl_rss *rss;
>>>
>>> - rss = (struct virtio_net_ctrl_rss *)(uintptr_t)vring-
>>> desc[idx_data].addr;
>>> + rss = (struct virtio_net_ctrl_rss *)ctrl->data;
>>> status = virtio_user_handle_mq(dev, rss->max_tx_vq);
>>> - } else if (hdr->class == VIRTIO_NET_CTRL_RX ||
>>> - hdr->class == VIRTIO_NET_CTRL_MAC ||
>>> - hdr->class == VIRTIO_NET_CTRL_VLAN) {
>>> + } else if (ctrl->hdr.class == VIRTIO_NET_CTRL_RX ||
>>> + ctrl->hdr.class == VIRTIO_NET_CTRL_MAC ||
>>> + ctrl->hdr.class == VIRTIO_NET_CTRL_VLAN) {
>>> status = 0;
>>> }
>>>
>>> if (!status && dev->scvq)
>>> - status = virtio_send_command(&dev->scvq->cq,
>>> - (struct virtio_pmd_ctrl *)hdr, dlen, nb_dlen);
>>> + status = virtio_send_command(&dev->scvq->cq, ctrl, dlen,
>> nb_dlen);
>>>
>>> /* Update status */
>>> - *(virtio_net_ctrl_ack *)(uintptr_t)
>>> - vring->desc[idx_status].addr = status;
>>> + ctrl->status = status;
>>>
>>> /* Update used descriptor */
>>> vring->desc[idx_hdr].id = vring->desc[idx_status].id;
>
next prev parent reply other threads:[~2024-07-02 12:41 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-02-29 13:29 [PATCH v2 0/3] net/virtio: support IOVA as PA mode for vDPA backend Srujana Challa
2024-02-29 13:29 ` [PATCH v2 1/3] net/virtio_user: avoid cq descriptor buffer address accessing Srujana Challa
2024-06-28 8:07 ` Maxime Coquelin
2024-07-02 11:09 ` [EXTERNAL] " Srujana Challa
2024-07-02 12:41 ` Maxime Coquelin [this message]
2024-07-03 10:03 ` [PATCH v3 0/3] net/virtio: support IOVA as PA mode for vDPA backend Srujana Challa
2024-07-03 10:03 ` [PATCH v3 1/3] net/virtio_user: convert cq descriptor IOVA address to Virtual address Srujana Challa
2024-07-03 10:19 ` Jerin Jacob
2024-07-03 13:28 ` Maxime Coquelin
2024-07-03 13:40 ` Maxime Coquelin
2024-07-03 10:03 ` [PATCH v3 2/3] net/virtio: store desc IOVA address in vring data structure Srujana Challa
2024-07-03 13:41 ` Maxime Coquelin
2024-07-03 10:03 ` [PATCH v3 3/3] net/virtio_user: support sharing vq descriptor IOVA to the backend Srujana Challa
2024-07-03 13:41 ` Maxime Coquelin
2024-07-03 14:34 ` [PATCH v3 0/3] net/virtio: support IOVA as PA mode for vDPA backend Maxime Coquelin
2024-02-29 13:29 ` [PATCH v2 2/3] net/virtio: store desc IOVA address in vring data structure Srujana Challa
2024-02-29 13:29 ` [PATCH v2 3/3] net/virtio_user: support sharing vq descriptor IOVA to the backend Srujana Challa
2024-06-19 9:39 ` [PATCH v2 0/3] net/virtio: support IOVA as PA mode for vDPA backend Srujana Challa
2024-06-28 13:33 ` Maxime Coquelin
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=62446aea-34f3-4fbc-ad42-ddd6b57e5864@redhat.com \
--to=maxime.coquelin@redhat.com \
--cc=chenbox@nvidia.com \
--cc=dev@dpdk.org \
--cc=jerinj@marvell.com \
--cc=ndabilpuram@marvell.com \
--cc=schalla@marvell.com \
--cc=vattunuru@marvell.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).