From: Ciara Loftus <ciara.loftus@intel.com>
To: dev@dpdk.org
Cc: Ciara Loftus <ciara.loftus@intel.com>
Subject: [dpdk-dev] [PATCH v2] net/af_xdp: don't allow umem sharing for xsks with same ctx
Date: Tue, 13 Oct 2020 13:10:08 +0000 [thread overview]
Message-ID: <20201013131008.4070-1-ciara.loftus@intel.com> (raw)
In-Reply-To: <20201008091729.4321-1-ciara.loftus@intel.com>
AF_XDP PMDs who wish to share a UMEM must have a unique context
(ctx) ie. netdev,qid tuple. For instance, the following will not
work since both PMDs' contexts are identical.
--vdev net_af_xdp0,iface=ens786f1,start_queue=0,shared_umem=1
--vdev net_af_xdp1,iface=ens786f1,start_queue=0,shared_umem=1
Supporting this scenario would require locks, which would impact
the performance of the more typical cases - xsks with different
netdev,qid tuples.
Signed-off-by: Ciara Loftus <ciara.loftus@intel.com>
Fixes: 74b46340e2d4 ("net/af_xdp: support shared UMEM")
---
v2:
* Add doc update
* Fix commit message style issues
* Update commit message with more information
doc/guides/nics/af_xdp.rst | 25 ++++++++++++++++
drivers/net/af_xdp/rte_eth_af_xdp.c | 44 +++++++++++++++++++++++------
2 files changed, 60 insertions(+), 9 deletions(-)
diff --git a/doc/guides/nics/af_xdp.rst b/doc/guides/nics/af_xdp.rst
index be268fe7ff..052e59a3ae 100644
--- a/doc/guides/nics/af_xdp.rst
+++ b/doc/guides/nics/af_xdp.rst
@@ -82,3 +82,28 @@ Limitations
Note: The AF_XDP PMD will fail to initialise if an MTU which violates the driver's
conditions as above is set prior to launching the application.
+
+- **Shared UMEM**
+
+ The sharing of UMEM is only supported for AF_XDP sockets with unique contexts.
+ The context refers to the netdev,qid tuple.
+
+ The following combination will fail:
+
+ .. code-block:: console
+
+ --vdev net_af_xdp0,iface=ens786f1,shared_umem=1 \
+ --vdev net_af_xdp1,iface=ens786f1,shared_umem=1 \
+
+ Either of the following however is permitted since either the netdev or qid differs
+ between the two vdevs:
+
+ .. code-block:: console
+
+ --vdev net_af_xdp0,iface=ens786f1,shared_umem=1 \
+ --vdev net_af_xdp1,iface=ens786f1,start_queue=1,shared_umem=1 \
+
+ .. code-block:: console
+
+ --vdev net_af_xdp0,iface=ens786f1,shared_umem=1 \
+ --vdev net_af_xdp1,iface=ens786f2,shared_umem=1 \
\ No newline at end of file
diff --git a/drivers/net/af_xdp/rte_eth_af_xdp.c b/drivers/net/af_xdp/rte_eth_af_xdp.c
index eaf2c9c873..9e0e5c254a 100644
--- a/drivers/net/af_xdp/rte_eth_af_xdp.c
+++ b/drivers/net/af_xdp/rte_eth_af_xdp.c
@@ -634,16 +634,35 @@ find_internal_resource(struct pmd_internals *port_int)
return list;
}
+/* Check if the netdev,qid context already exists */
+static inline bool
+ctx_exists(struct pkt_rx_queue *rxq, const char *ifname,
+ struct pkt_rx_queue *list_rxq, const char *list_ifname)
+{
+ bool exists = false;
+
+ if (rxq->xsk_queue_idx == list_rxq->xsk_queue_idx &&
+ !strncmp(ifname, list_ifname, IFNAMSIZ)) {
+ AF_XDP_LOG(ERR, "ctx %s,%i already exists, cannot share umem\n",
+ ifname, rxq->xsk_queue_idx);
+ exists = true;
+ }
+
+ return exists;
+}
+
/* Get a pointer to an existing UMEM which overlays the rxq's mb_pool */
-static inline struct xsk_umem_info *
-get_shared_umem(struct pkt_rx_queue *rxq) {
+static inline int
+get_shared_umem(struct pkt_rx_queue *rxq, const char *ifname,
+ struct xsk_umem_info **umem)
+{
struct internal_list *list;
struct pmd_internals *internals;
- int i = 0;
+ int i = 0, ret = 0;
struct rte_mempool *mb_pool = rxq->mb_pool;
if (mb_pool == NULL)
- return NULL;
+ return ret;
pthread_mutex_lock(&internal_list_lock);
@@ -655,20 +674,25 @@ get_shared_umem(struct pkt_rx_queue *rxq) {
if (rxq == list_rxq)
continue;
if (mb_pool == internals->rx_queues[i].mb_pool) {
+ if (ctx_exists(rxq, ifname, list_rxq,
+ internals->if_name)) {
+ ret = -1;
+ goto out;
+ }
if (__atomic_load_n(
&internals->rx_queues[i].umem->refcnt,
__ATOMIC_ACQUIRE)) {
- pthread_mutex_unlock(
- &internal_list_lock);
- return internals->rx_queues[i].umem;
+ *umem = internals->rx_queues[i].umem;
+ goto out;
}
}
}
}
+out:
pthread_mutex_unlock(&internal_list_lock);
- return NULL;
+ return ret;
}
static int
@@ -913,7 +937,9 @@ xsk_umem_info *xdp_umem_configure(struct pmd_internals *internals,
uint64_t umem_size, align = 0;
if (internals->shared_umem) {
- umem = get_shared_umem(rxq);
+ if (get_shared_umem(rxq, internals->if_name, &umem) < 0)
+ return NULL;
+
if (umem != NULL &&
__atomic_load_n(&umem->refcnt, __ATOMIC_ACQUIRE) <
umem->max_xsks) {
--
2.17.1
next prev parent reply other threads:[~2020-10-13 13:36 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-08 9:17 [dpdk-dev] [PATCH] net/af_xdp: Don't allow umem sharing for xsks with same netdev, qid Ciara Loftus
2020-10-08 11:55 ` Ferruh Yigit
2020-10-13 8:52 ` Ferruh Yigit
2020-10-13 13:37 ` Loftus, Ciara
2020-10-13 13:10 ` Ciara Loftus [this message]
2020-10-13 15:57 ` [dpdk-dev] [PATCH v2] net/af_xdp: don't allow umem sharing for xsks with same ctx Ferruh Yigit
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=20201013131008.4070-1-ciara.loftus@intel.com \
--to=ciara.loftus@intel.com \
--cc=dev@dpdk.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).