[PATCH] net/nfp: add the restrict of setting the mtu

[PATCH] net/nfp: add the restrict of setting the mtu

[Peng Zhang]

1.When the setting mtu is higher than flbufsz, the mtu doesn't work.
But it doesn't have any notice about this restrict.
2.when the setting mtu isn't in the range, it doesn't have any notice.

This patch will add the notice about these restrict.

Signed-off-by: Peng Zhang <peng.zhang@corigine.com>
Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
Signed-off-by: Louis Peens <louis.peens@corigine.com>
---
 drivers/net/nfp/nfp_common.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/drivers/net/nfp/nfp_common.c b/drivers/net/nfp/nfp_common.c
index f8978e803a..2ea9853548 100644
--- a/drivers/net/nfp/nfp_common.c
+++ b/drivers/net/nfp/nfp_common.c
@@ -956,6 +956,20 @@ nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
 		return -EBUSY;
 	}
 
+	/* the setting mtu is in the range */
+	if (mtu < 68 || mtu > hw->max_mtu) {
+		PMD_DRV_LOG(ERR, "the setting mtu cannot be less than 68 or more than %d",
+			    hw->max_mtu);
+		return -ERANGE;
+	}
+
+	/* the setting mtu is lower than flbufsz */
+	if (mtu > hw->flbufsz) {
+		PMD_DRV_LOG(ERR, "the setting mtu must be lower than current mbufsize of %d",
+			    hw->flbufsz);
+		return -ERANGE;
+	}
+
 	/* writing to configuration space */
 	nn_cfg_writel(hw, NFP_NET_CFG_MTU, mtu);
 
-- 
2.27.0

Re: [PATCH] net/nfp: add the restrict of setting the mtu

[Stephen Hemminger]

On Thu, 17 Mar 2022 16:39:17 +0200
Peng Zhang <peng.zhang@corigine.com> wrote:

> +	/* the setting mtu is in the range */
> +	if (mtu < 68 || mtu > hw->max_mtu) {
> +		PMD_DRV_LOG(ERR, "the setting mtu cannot be less than 68 or more than %d",
> +			    hw->max_mtu);
> +		return -ERANGE;
> +	}
> +
> +	/* the setting mtu is lower than flbufsz */
> +	if (mtu > hw->flbufsz) {
> +		PMD_DRV_LOG(ERR, "the setting mtu must be lower than current mbufsize of %d",
> +			    hw->flbufsz);
> +		return -ERANGE;

The ethdev layer already has min/max mtu checks, why is it needed inside common code?

It looks like nfp_net_infos_get() does not set these values.
Please fix it there instead.

PS: 68 is RTE_ETHER_MIN_MTU so please use that.

[PATCH v2] net/nfp: fix mtu settings

[Peng Zhang]

1.When the setting mtu is higher than flbufsz, the mtu doesn't work.
But it doesn't have any notice about this restrict.
2.add the min_mtu and max_mtu in the nfp_net_infos_get() to avoid
the setting mtu isn't in the range

This patch will add these restrict of nfp mtu.

Fixes: d4a27a3 ("nfp: add basic features")
Cc: stable@dpdk.org

Signed-off-by: Peng Zhang <peng.zhang@corigine.com>
Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
Signed-off-by: Louis Peens <louis.peens@corigine.com>
---
v2:
* add the min_mtu and max_mtu in the nfp_net_infos_get()
---

 drivers/net/nfp/nfp_common.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/net/nfp/nfp_common.c b/drivers/net/nfp/nfp_common.c
index f8978e803a..4e48e33a63 100644
--- a/drivers/net/nfp/nfp_common.c
+++ b/drivers/net/nfp/nfp_common.c
@@ -693,6 +693,8 @@ nfp_net_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
 
 	hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
 
+	dev_info->max_mtu = (uint16_t)hw->max_mtu;
+	dev_info->min_mtu = RTE_ETHER_MIN_MTU;
 	dev_info->max_rx_queues = (uint16_t)hw->max_rx_queues;
 	dev_info->max_tx_queues = (uint16_t)hw->max_tx_queues;
 	dev_info->min_rx_bufsize = RTE_ETHER_MIN_MTU;
@@ -956,6 +958,13 @@ nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
 		return -EBUSY;
 	}
 
+	/* the setting mtu is lower than flbufsz */
+	if (mtu > hw->flbufsz) {
+		PMD_DRV_LOG(ERR, "the setting mtu must be lower than current mbufsize of %d",
+			    hw->flbufsz);
+		return -ERANGE;
+	}
+
 	/* writing to configuration space */
 	nn_cfg_writel(hw, NFP_NET_CFG_MTU, mtu);
 
-- 
2.27.0

[PATCH v3] net/nfp: fix mtu settings

[Peng Zhang]

1.When the setting mtu is higher than flbufsz, the mtu doesn't work.
But it doesn't have any notice about this restrict.
2.add the min_mtu and max_mtu in the nfp_net_infos_get() to avoid
the setting mtu isn't in the range

This patch will add these restrict of nfp mtu.

Fixes: d4a27a3b092a ("nfp: add basic features")
Cc: stable@dpdk.org

Signed-off-by: Peng Zhang <peng.zhang@corigine.com>
Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
Signed-off-by: Louis Peens <louis.peens@corigine.com>
---
v3:
* git summary

v2:
* add the min_mtu and max_mtu in the nfp_net_infos_get()
---

 drivers/net/nfp/nfp_common.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/net/nfp/nfp_common.c b/drivers/net/nfp/nfp_common.c
index f8978e803a..4e48e33a63 100644
--- a/drivers/net/nfp/nfp_common.c
+++ b/drivers/net/nfp/nfp_common.c
@@ -693,6 +693,8 @@ nfp_net_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
 
 	hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
 
+	dev_info->max_mtu = (uint16_t)hw->max_mtu;
+	dev_info->min_mtu = RTE_ETHER_MIN_MTU;
 	dev_info->max_rx_queues = (uint16_t)hw->max_rx_queues;
 	dev_info->max_tx_queues = (uint16_t)hw->max_tx_queues;
 	dev_info->min_rx_bufsize = RTE_ETHER_MIN_MTU;
@@ -956,6 +958,13 @@ nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
 		return -EBUSY;
 	}
 
+	/* the setting mtu is lower than flbufsz */
+	if (mtu > hw->flbufsz) {
+		PMD_DRV_LOG(ERR, "the setting mtu must be lower than current mbufsize of %d",
+			    hw->flbufsz);
+		return -ERANGE;
+	}
+
 	/* writing to configuration space */
 	nn_cfg_writel(hw, NFP_NET_CFG_MTU, mtu);
 
-- 
2.27.0

Re: [PATCH v3] net/nfp: fix mtu settings

[Ferruh Yigit]

On 3/30/2022 4:17 AM, Peng Zhang wrote:
> 1.When the setting mtu is higher than flbufsz, the mtu doesn't work.
> But it doesn't have any notice about this restrict.
> 2.add the min_mtu and max_mtu in the nfp_net_infos_get() to avoid
> the setting mtu isn't in the range
> 
> This patch will add these restrict of nfp mtu.
> 

+Niklas, as he is the maintainer of the PMD.

Can you please prefer uppercase 'MTU' in title/commit log and error log?

I assume scattered_rx is not supported by the device, which enables 
receiving packets bigger than mbuf data size.

> Fixes: d4a27a3b092a ("nfp: add basic features")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Peng Zhang <peng.zhang@corigine.com>
> Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
> Signed-off-by: Louis Peens <louis.peens@corigine.com>
> ---
> v3:
> * git summary
> 
> v2:
> * add the min_mtu and max_mtu in the nfp_net_infos_get()
> ---
> 
>   drivers/net/nfp/nfp_common.c | 9 +++++++++
>   1 file changed, 9 insertions(+)
> 
> diff --git a/drivers/net/nfp/nfp_common.c b/drivers/net/nfp/nfp_common.c
> index f8978e803a..4e48e33a63 100644
> --- a/drivers/net/nfp/nfp_common.c
> +++ b/drivers/net/nfp/nfp_common.c
> @@ -693,6 +693,8 @@ nfp_net_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
>   
>   	hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
>   
> +	dev_info->max_mtu = (uint16_t)hw->max_mtu;
> +	dev_info->min_mtu = RTE_ETHER_MIN_MTU;
>   	dev_info->max_rx_queues = (uint16_t)hw->max_rx_queues;
>   	dev_info->max_tx_queues = (uint16_t)hw->max_tx_queues;
>   	dev_info->min_rx_bufsize = RTE_ETHER_MIN_MTU;
> @@ -956,6 +958,13 @@ nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
>   		return -EBUSY;
>   	}
>   
> +	/* the setting mtu is lower than flbufsz */
> +	if (mtu > hw->flbufsz) {
> +		PMD_DRV_LOG(ERR, "the setting mtu must be lower than current mbufsize of %d",

You may want to start log with uppercase as done in other logs, for 
consistency.
Also may want to print the MTU value in the log.

> +			    hw->flbufsz);
> +		return -ERANGE;
> +	}

Should this be checked in 'nfb_eth_dev_configure()' too, where 
'dev->data->mtu' can be set?

> +
>   	/* writing to configuration space */
>   	nn_cfg_writel(hw, NFP_NET_CFG_MTU, mtu);
>   

RE: [PATCH v3] net/nfp: fix mtu settings

[Nole Zhang]

Thanks for your advice.

> -----Original Message-----
> From: Ferruh Yigit <ferruh.yigit@xilinx.com>
> Sent: 2022年4月21日 2:55
> To: Nole Zhang <peng.zhang@corigine.com>; dev@dpdk.org; Niklas
> Soderlund <niklas.soderlund@corigine.com>
> Cc: Heinrich Kuhn <heinrich.kuhn@corigine.com>; stable@dpdk.org;
> Chaoyong He <chaoyong.he@corigine.com>; Louis Peens
> <louis.peens@corigine.com>
> Subject: Re: [PATCH v3] net/nfp: fix mtu settings
> 
> On 3/30/2022 4:17 AM, Peng Zhang wrote:
> > 1.When the setting mtu is higher than flbufsz, the mtu doesn't work.
> > But it doesn't have any notice about this restrict.
> > 2.add the min_mtu and max_mtu in the nfp_net_infos_get() to avoid the
> > setting mtu isn't in the range
> >
> > This patch will add these restrict of nfp mtu.
> >
> 
> +Niklas, as he is the maintainer of the PMD.
> 
> Can you please prefer uppercase 'MTU' in title/commit log and error log?
> 

Yes,  I will use the MTU replace the mtu.

> I assume scattered_rx is not supported by the device, which enables
> receiving packets bigger than mbuf data size.
> 
> > Fixes: d4a27a3b092a ("nfp: add basic features")
> > Cc: stable@dpdk.org
> >
> > Signed-off-by: Peng Zhang <peng.zhang@corigine.com>
> > Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
> > Signed-off-by: Louis Peens <louis.peens@corigine.com>
> > ---
> > v3:
> > * git summary
> >
> > v2:
> > * add the min_mtu and max_mtu in the nfp_net_infos_get()
> > ---
> >
> >   drivers/net/nfp/nfp_common.c | 9 +++++++++
> >   1 file changed, 9 insertions(+)
> >
> > diff --git a/drivers/net/nfp/nfp_common.c
> > b/drivers/net/nfp/nfp_common.c index f8978e803a..4e48e33a63 100644
> > --- a/drivers/net/nfp/nfp_common.c
> > +++ b/drivers/net/nfp/nfp_common.c
> > @@ -693,6 +693,8 @@ nfp_net_infos_get(struct rte_eth_dev *dev, struct
> > rte_eth_dev_info *dev_info)
> >
> >   	hw = NFP_NET_DEV_PRIVATE_TO_HW(dev->data->dev_private);
> >
> > +	dev_info->max_mtu = (uint16_t)hw->max_mtu;
> > +	dev_info->min_mtu = RTE_ETHER_MIN_MTU;
> >   	dev_info->max_rx_queues = (uint16_t)hw->max_rx_queues;
> >   	dev_info->max_tx_queues = (uint16_t)hw->max_tx_queues;
> >   	dev_info->min_rx_bufsize = RTE_ETHER_MIN_MTU; @@ -956,6
> +958,13 @@
> > nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
> >   		return -EBUSY;
> >   	}
> >
> > +	/* the setting mtu is lower than flbufsz */
> > +	if (mtu > hw->flbufsz) {
> > +		PMD_DRV_LOG(ERR, "the setting mtu must be lower than
> current
> > +mbufsize of %d",
> 
> You may want to start log with uppercase as done in other logs, for
> consistency.
> Also may want to print the MTU value in the log.
> 

Yes,  I will use the MTU replace the mtu.

> > +			    hw->flbufsz);
> > +		return -ERANGE;
> > +	}
> 
> Should this be checked in 'nfb_eth_dev_configure()' too, where 'dev->data-
> >mtu' can be set?
> 

You mean in the nfp_net_configure(), we need through the rxmode->mtu to check the mtu?
I will consider it.

Thanks for your advice again.
> > +
> >   	/* writing to configuration space */
> >   	nn_cfg_writel(hw, NFP_NET_CFG_MTU, mtu);
> >

[PATCH v4] net/nfp: make sure MTU is never larger than mbufsize

[Peng Zhang]

Setting a MTU larger than mbufsize is not supported by the device but
not prohibited by the driver. This change adds a restriction to the
driver to prevent setting an MTU that is too large.

While at it define the minimum MTU in the device information to describe
the complete supported MTU range.

Fixes: d4a27a3 ("nfp: add basic features")
Cc: stable@dpdk.org

Signed-off-by: Peng Zhang <peng.zhang@corigine.com>
Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Niklas Söderlund <niklas.soderlund@corigine.com>
Reviewed-by: Walter Heymans <walter.heymans@corigine.com>
------
Depends-on: patch-109914 ("net/nfp: update how MAX MTU is read")
---
 drivers/net/nfp/nfp_common.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/drivers/net/nfp/nfp_common.c b/drivers/net/nfp/nfp_common.c
index 52fbda1a79..be68c25fb8 100644
--- a/drivers/net/nfp/nfp_common.c
+++ b/drivers/net/nfp/nfp_common.c
@@ -176,6 +176,13 @@ nfp_net_configure(struct rte_eth_dev *dev)
 		return -EINVAL;
 	}
 
+	/* Checking MTU set */
+	if (rxmode->mtu > hw->flbufsz) {
+		PMD_INIT_LOG(INFO, "MTU (%u) larger then current mbufsize (%u) not supported",
+				    rxmode->mtu, hw->flbufsz);
+		return -ERANGE;
+	}
+
 	return 0;
 }
 
@@ -702,6 +709,7 @@ nfp_net_infos_get(struct rte_eth_dev *dev, struct rte_eth_dev_info *dev_info)
 	 */
 	dev_info->max_rx_pktlen = NFP_FRAME_SIZE_MAX;
 	dev_info->max_mtu = hw->max_mtu;
+	dev_info->min_mtu = RTE_ETHER_MIN_MTU;
 	/* Next should change when PF support is implemented */
 	dev_info->max_mac_addrs = 1;
 
@@ -961,6 +969,13 @@ nfp_net_dev_mtu_set(struct rte_eth_dev *dev, uint16_t mtu)
 		return -EBUSY;
 	}
 
+	/* MTU larger then current mbufsize not supported */
+	if (mtu > hw->flbufsz) {
+		PMD_DRV_LOG(ERR, "MTU (%u) larger then current mbufsize (%u) not supported",
+			    mtu, hw->flbufsz);
+		return -ERANGE;
+	}
+
 	/* writing to configuration space */
 	nn_cfg_writel(hw, NFP_NET_CFG_MTU, mtu);
 
-- 
2.27.0

Re: [PATCH v4] net/nfp: make sure MTU is never larger than mbufsize

[Andrew Rybchenko]

On 5/11/22 04:15, Peng Zhang wrote:
> Setting a MTU larger than mbufsize is not supported by the device but
> not prohibited by the driver. This change adds a restriction to the
> driver to prevent setting an MTU that is too large.
> 
> While at it define the minimum MTU in the device information to describe
> the complete supported MTU range.
> 
> Fixes: d4a27a3 ("nfp: add basic features")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Peng Zhang <peng.zhang@corigine.com>
> Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
> Signed-off-by: Louis Peens <louis.peens@corigine.com>
> Signed-off-by: Niklas Söderlund <niklas.soderlund@corigine.com>
> Reviewed-by: Walter Heymans <walter.heymans@corigine.com>

Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>

Applied, thanks.