* [PATCH v2 1/1] net/tap: add a check that Rx/Tx have the same num of queues
@ 2022-01-12 7:23 Nobuhiro MIKI
2022-01-18 12:31 ` Ferruh Yigit
0 siblings, 1 reply; 3+ messages in thread
From: Nobuhiro MIKI @ 2022-01-12 7:23 UTC (permalink / raw)
To: keith.wiles; +Cc: dev, i.maximets, dmarchan, Nobuhiro MIKI
Users can create the desired number of RxQ and TxQ in DPDK. For
example, if the number of RxQ = 2 and the number of TxQ = 5,
a total of 8 file descriptors will be created for a tap device,
including RxQ, TxQ, and one for keepalive. The RxQ and TxQ
with the same ID are paired by dup(2).
In this scenario, Kernel will have 3 RxQ where packets are
incoming but not read. The reason for this is that there are only
2 RxQ that are polled by DPDK, while there are 5 queues in Kernel.
This patch add a checking if DPDK has appropriate numbers of
queues to avoid unexpected packet drop.
Signed-off-by: Nobuhiro MIKI <nmiki@yahoo-corp.jp>
---
v2: fix commit message
I had first discussed this issue in OVS [1], but changed my mind
that a fix in DPDK would be more appropriate.
[1]: https://mail.openvswitch.org/pipermail/ovs-dev/2021-November/389690.html
---
drivers/net/tap/rte_eth_tap.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/net/tap/rte_eth_tap.c b/drivers/net/tap/rte_eth_tap.c
index 5bb472f1a6..02eb311e09 100644
--- a/drivers/net/tap/rte_eth_tap.c
+++ b/drivers/net/tap/rte_eth_tap.c
@@ -940,6 +940,14 @@ tap_dev_configure(struct rte_eth_dev *dev)
RTE_PMD_TAP_MAX_QUEUES);
return -1;
}
+ if (dev->data->nb_rx_queues != dev->data->nb_tx_queues) {
+ TAP_LOG(ERR,
+ "%s: number of rx queues %d must be equal to number of tx queues %d",
+ dev->device->name,
+ dev->data->nb_rx_queues,
+ dev->data->nb_tx_queues);
+ return -1;
+ }
TAP_LOG(INFO, "%s: %s: TX configured queues number: %u",
dev->device->name, pmd->name, dev->data->nb_tx_queues);
--
2.24.4
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v2 1/1] net/tap: add a check that Rx/Tx have the same num of queues
2022-01-12 7:23 [PATCH v2 1/1] net/tap: add a check that Rx/Tx have the same num of queues Nobuhiro MIKI
@ 2022-01-18 12:31 ` Ferruh Yigit
2022-01-19 7:49 ` Nobuhiro MIKI
0 siblings, 1 reply; 3+ messages in thread
From: Ferruh Yigit @ 2022-01-18 12:31 UTC (permalink / raw)
To: Nobuhiro MIKI, keith.wiles; +Cc: dev, i.maximets, dmarchan
On 1/12/2022 7:23 AM, Nobuhiro MIKI wrote:
> Users can create the desired number of RxQ and TxQ in DPDK. For
> example, if the number of RxQ = 2 and the number of TxQ = 5,
> a total of 8 file descriptors will be created for a tap device,
> including RxQ, TxQ, and one for keepalive. The RxQ and TxQ
> with the same ID are paired by dup(2).
>
> In this scenario, Kernel will have 3 RxQ where packets are
> incoming but not read. The reason for this is that there are only
> 2 RxQ that are polled by DPDK, while there are 5 queues in Kernel.
> This patch add a checking if DPDK has appropriate numbers of
> queues to avoid unexpected packet drop.
>
> Signed-off-by: Nobuhiro MIKI <nmiki@yahoo-corp.jp>
It makes sense to add this check, since the driver logic seem already has
the assumption that Rx & Tx queue numbers will be same.
But can you please update the tap documentation for this limitation/restriction?
'doc/guides/nics/tap.rst'.
> ---
> v2: fix commit message
>
> I had first discussed this issue in OVS [1], but changed my mind
> that a fix in DPDK would be more appropriate.
> [1]: https://mail.openvswitch.org/pipermail/ovs-dev/2021-November/389690.html
> ---
> drivers/net/tap/rte_eth_tap.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/drivers/net/tap/rte_eth_tap.c b/drivers/net/tap/rte_eth_tap.c
> index 5bb472f1a6..02eb311e09 100644
> --- a/drivers/net/tap/rte_eth_tap.c
> +++ b/drivers/net/tap/rte_eth_tap.c
> @@ -940,6 +940,14 @@ tap_dev_configure(struct rte_eth_dev *dev)
> RTE_PMD_TAP_MAX_QUEUES);
> return -1;
> }
> + if (dev->data->nb_rx_queues != dev->data->nb_tx_queues) {
> + TAP_LOG(ERR,
> + "%s: number of rx queues %d must be equal to number of tx queues %d",
> + dev->device->name,
> + dev->data->nb_rx_queues,
> + dev->data->nb_tx_queues);
> + return -1;
> + }
>
> TAP_LOG(INFO, "%s: %s: TX configured queues number: %u",
> dev->device->name, pmd->name, dev->data->nb_tx_queues);
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v2 1/1] net/tap: add a check that Rx/Tx have the same num of queues
2022-01-18 12:31 ` Ferruh Yigit
@ 2022-01-19 7:49 ` Nobuhiro MIKI
0 siblings, 0 replies; 3+ messages in thread
From: Nobuhiro MIKI @ 2022-01-19 7:49 UTC (permalink / raw)
To: Ferruh Yigit; +Cc: keith.wiles, dev, i.maximets, dmarchan
On Tue, Jan 18, 2022 at 12:31:51PM +0000, Ferruh Yigit wrote:
> On 1/12/2022 7:23 AM, Nobuhiro MIKI wrote:
> > Users can create the desired number of RxQ and TxQ in DPDK. For
> > example, if the number of RxQ = 2 and the number of TxQ = 5,
> > a total of 8 file descriptors will be created for a tap device,
> > including RxQ, TxQ, and one for keepalive. The RxQ and TxQ
> > with the same ID are paired by dup(2).
> >
> > In this scenario, Kernel will have 3 RxQ where packets are
> > incoming but not read. The reason for this is that there are only
> > 2 RxQ that are polled by DPDK, while there are 5 queues in Kernel.
> > This patch add a checking if DPDK has appropriate numbers of
> > queues to avoid unexpected packet drop.
> >
> > Signed-off-by: Nobuhiro MIKI <nmiki@yahoo-corp.jp>
>
>
> It makes sense to add this check, since the driver logic seem already has
> the assumption that Rx & Tx queue numbers will be same.
>
> But can you please update the tap documentation for this limitation/restriction?
> 'doc/guides/nics/tap.rst'.
Thanks for the review.
I've updated doc/guides/nics/tap.rst and sent v3 patch.
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2022-01-19 7:49 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-01-12 7:23 [PATCH v2 1/1] net/tap: add a check that Rx/Tx have the same num of queues Nobuhiro MIKI
2022-01-18 12:31 ` Ferruh Yigit
2022-01-19 7:49 ` Nobuhiro MIKI
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).