[PATCH net-next 0/3] ethtool: set_channels: add a few more checks

[PATCH net-next 0/3] ethtool: set_channels: add a few more checks

[Jakub Kicinski]

There seems to be a few more things we can check in the core before
we call drivers' ethtool_ops->set_channels. Adding the checks to
the core simplifies the drivers. This set only includes changes
to the NFP driver as an example.

There is a small risk in the first patch that someone actually
purposefully accepts a strange configuration without RX or TX
channels, but I couldn't find such a driver in the tree.

Jakub Kicinski (3):
  ethtool: check if there is at least one channel for TX/RX in the core
  nfp: don't check lack of RX/TX channels
  ethtool: don't call set_channels in drivers if config didn't change

 .../ethernet/netronome/nfp/nfp_net_ethtool.c  |  3 +--
 net/ethtool/channels.c                        | 20 +++++++++++++++++--
 net/ethtool/ioctl.c                           | 11 ++++++++++
 3 files changed, 30 insertions(+), 4 deletions(-)

-- 
2.25.4

[PATCH net-next 1/3] ethtool: check if there is at least one channel for TX/RX in the core

[Jakub Kicinski]

Having a channel config with no ability to RX or TX traffic is
clearly wrong. Check for this in the core so the drivers don't
have to.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 net/ethtool/channels.c | 20 ++++++++++++++++++--
 net/ethtool/ioctl.c    |  5 +++++
 2 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/net/ethtool/channels.c b/net/ethtool/channels.c
index 389924b65d05..3aa4975919d7 100644
--- a/net/ethtool/channels.c
+++ b/net/ethtool/channels.c
@@ -129,13 +129,13 @@ int ethnl_set_channels(struct sk_buff *skb, struct genl_info *info)
 {
 	struct nlattr *tb[ETHTOOL_A_CHANNELS_MAX + 1];
 	unsigned int from_channel, old_total, i;
+	bool mod = false, mod_combined = false;
 	struct ethtool_channels channels = {};
 	struct ethnl_req_info req_info = {};
 	const struct nlattr *err_attr;
 	const struct ethtool_ops *ops;
 	struct net_device *dev;
 	u32 max_rx_in_use = 0;
-	bool mod = false;
 	int ret;
 
 	ret = nlmsg_parse(info->nlhdr, GENL_HDRLEN, tb,
@@ -170,7 +170,8 @@ int ethnl_set_channels(struct sk_buff *skb, struct genl_info *info)
 	ethnl_update_u32(&channels.other_count,
 			 tb[ETHTOOL_A_CHANNELS_OTHER_COUNT], &mod);
 	ethnl_update_u32(&channels.combined_count,
-			 tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT], &mod);
+			 tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT], &mod_combined);
+	mod |= mod_combined;
 	ret = 0;
 	if (!mod)
 		goto out_ops;
@@ -193,6 +194,21 @@ int ethnl_set_channels(struct sk_buff *skb, struct genl_info *info)
 		goto out_ops;
 	}
 
+	/* ensure there is at least one RX and one TX channel */
+	if (!channels.combined_count && !channels.rx_count)
+		err_attr = tb[ETHTOOL_A_CHANNELS_RX_COUNT];
+	else if (!channels.combined_count && !channels.tx_count)
+		err_attr = tb[ETHTOOL_A_CHANNELS_TX_COUNT];
+	else
+		err_attr = NULL;
+	if (err_attr) {
+		if (mod_combined)
+			err_attr = tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT];
+		ret = -EINVAL;
+		NL_SET_ERR_MSG_ATTR(info->extack, err_attr, "requested channel counts would result in no RX or TX channel being configured");
+		goto out_ops;
+	}
+
 	/* ensure the new Rx count fits within the configured Rx flow
 	 * indirection table settings
 	 */
diff --git a/net/ethtool/ioctl.c b/net/ethtool/ioctl.c
index 52102ab1709b..a574d60111fa 100644
--- a/net/ethtool/ioctl.c
+++ b/net/ethtool/ioctl.c
@@ -1676,6 +1676,11 @@ static noinline_for_stack int ethtool_set_channels(struct net_device *dev,
 	    channels.other_count > curr.max_other)
 		return -EINVAL;
 
+	/* ensure there is at least one RX and one TX channel */
+	if (!channels.combined_count &&
+	    (!channels.rx_count || !channels.tx_count))
+		return -EINVAL;
+
 	/* ensure the new Rx count fits within the configured Rx flow
 	 * indirection table settings */
 	if (netif_is_rxfh_configured(dev) &&
-- 
2.25.4

Re: [PATCH net-next 1/3] ethtool: check if there is at least one channel for TX/RX in the core

[Michal Kubecek]

On Fri, May 15, 2020 at 12:49:00PM -0700, Jakub Kicinski wrote:
> Having a channel config with no ability to RX or TX traffic is
> clearly wrong. Check for this in the core so the drivers don't
> have to.
> 
> Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Reviewed-by: Michal Kubecek <mkubecek@suse.cz>

Only one formal problem:

> @@ -170,7 +170,8 @@ int ethnl_set_channels(struct sk_buff *skb, struct genl_info *info)
>  	ethnl_update_u32(&channels.other_count,
>  			 tb[ETHTOOL_A_CHANNELS_OTHER_COUNT], &mod);
>  	ethnl_update_u32(&channels.combined_count,
> -			 tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT], &mod);
> +			 tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT], &mod_combined);
> +	mod |= mod_combined;
>  	ret = 0;
>  	if (!mod)
>  		goto out_ops;

Bitwise or should do the right thing but using "|=" on bool variables
looks strange.

Michal

Re: [PATCH net-next 1/3] ethtool: check if there is at least one channel for TX/RX in the core

[Jakub Kicinski]

On Fri, 15 May 2020 22:56:56 +0200 Michal Kubecek wrote:
> On Fri, May 15, 2020 at 12:49:00PM -0700, Jakub Kicinski wrote:
> > Having a channel config with no ability to RX or TX traffic is
> > clearly wrong. Check for this in the core so the drivers don't
> > have to.
> > 
> > Signed-off-by: Jakub Kicinski <kuba@kernel.org>  
> 
> Reviewed-by: Michal Kubecek <mkubecek@suse.cz>

Thanks for the reviews!

> > @@ -170,7 +170,8 @@ int ethnl_set_channels(struct sk_buff *skb, struct genl_info *info)
> >  	ethnl_update_u32(&channels.other_count,
> >  			 tb[ETHTOOL_A_CHANNELS_OTHER_COUNT], &mod);
> >  	ethnl_update_u32(&channels.combined_count,
> > -			 tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT], &mod);
> > +			 tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT], &mod_combined);
> > +	mod |= mod_combined;
> >  	ret = 0;
> >  	if (!mod)
> >  		goto out_ops;  
> 
> Bitwise or should do the right thing but using "|=" on bool variables
> looks strange.

Interesting, I never thought about it in this way.

[PATCH net-next 2/3] nfp: don't check lack of RX/TX channels

[Jakub Kicinski]

Core will now perform this check.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
index a5aa3219d112..6eb9fb9a1814 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
@@ -1438,8 +1438,7 @@ static int nfp_net_set_channels(struct net_device *netdev,
 	unsigned int total_rx, total_tx;
 
 	/* Reject unsupported */
-	if (!channel->combined_count ||
-	    channel->other_count != NFP_NET_NON_Q_VECTORS ||
+	if (channel->other_count != NFP_NET_NON_Q_VECTORS ||
 	    (channel->rx_count && channel->tx_count))
 		return -EINVAL;
 
-- 
2.25.4

Re: [PATCH net-next 2/3] nfp: don't check lack of RX/TX channels

[Michal Kubecek]

On Fri, May 15, 2020 at 12:49:01PM -0700, Jakub Kicinski wrote:
> Core will now perform this check.
> 
> Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Reviewed-by: Michal Kubecek <mkubecek@suse.cz>

> ---
>  drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
> index a5aa3219d112..6eb9fb9a1814 100644
> --- a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
> +++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
> @@ -1438,8 +1438,7 @@ static int nfp_net_set_channels(struct net_device *netdev,
>  	unsigned int total_rx, total_tx;
>  
>  	/* Reject unsupported */
> -	if (!channel->combined_count ||
> -	    channel->other_count != NFP_NET_NON_Q_VECTORS ||
> +	if (channel->other_count != NFP_NET_NON_Q_VECTORS ||
>  	    (channel->rx_count && channel->tx_count))
>  		return -EINVAL;
>  
> -- 
> 2.25.4
> 

Re: [PATCH net-next 2/3] nfp: don't check lack of RX/TX channels

[Simon Horman]

On Fri, May 15, 2020 at 12:49:01PM -0700, Jakub Kicinski wrote:
> Core will now perform this check.
> 
> Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Reviewed-by: Simon Horman <simon.horman@netronome.com>

> ---
>  drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
> index a5aa3219d112..6eb9fb9a1814 100644
> --- a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
> +++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
> @@ -1438,8 +1438,7 @@ static int nfp_net_set_channels(struct net_device *netdev,
>  	unsigned int total_rx, total_tx;
>  
>  	/* Reject unsupported */
> -	if (!channel->combined_count ||
> -	    channel->other_count != NFP_NET_NON_Q_VECTORS ||
> +	if (channel->other_count != NFP_NET_NON_Q_VECTORS ||
>  	    (channel->rx_count && channel->tx_count))
>  		return -EINVAL;
>  
> -- 
> 2.25.4
> 

[PATCH net-next 3/3] ethtool: don't call set_channels in drivers if config didn't change

[Jakub Kicinski]

Don't call drivers if nothing changed. Netlink code already
contains this logic.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
 net/ethtool/ioctl.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/net/ethtool/ioctl.c b/net/ethtool/ioctl.c
index a574d60111fa..eeb1137a3f23 100644
--- a/net/ethtool/ioctl.c
+++ b/net/ethtool/ioctl.c
@@ -1669,6 +1669,12 @@ static noinline_for_stack int ethtool_set_channels(struct net_device *dev,
 
 	dev->ethtool_ops->get_channels(dev, &curr);
 
+	if (channels.rx_count == curr.rx_count &&
+	    channels.tx_count == curr.tx_count &&
+	    channels.combined_count == curr.combined_count &&
+	    channels.other_count == curr.other_count)
+		return 0;
+
 	/* ensure new counts are within the maximums */
 	if (channels.rx_count > curr.max_rx ||
 	    channels.tx_count > curr.max_tx ||
-- 
2.25.4

Re: [PATCH net-next 3/3] ethtool: don't call set_channels in drivers if config didn't change

[Michal Kubecek]

On Fri, May 15, 2020 at 12:49:02PM -0700, Jakub Kicinski wrote:
> Don't call drivers if nothing changed. Netlink code already
> contains this logic.
> 
> Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Reviewed-by: Michal Kubecek <mkubecek@suse.cz>

> ---
>  net/ethtool/ioctl.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/net/ethtool/ioctl.c b/net/ethtool/ioctl.c
> index a574d60111fa..eeb1137a3f23 100644
> --- a/net/ethtool/ioctl.c
> +++ b/net/ethtool/ioctl.c
> @@ -1669,6 +1669,12 @@ static noinline_for_stack int ethtool_set_channels(struct net_device *dev,
>  
>  	dev->ethtool_ops->get_channels(dev, &curr);
>  
> +	if (channels.rx_count == curr.rx_count &&
> +	    channels.tx_count == curr.tx_count &&
> +	    channels.combined_count == curr.combined_count &&
> +	    channels.other_count == curr.other_count)
> +		return 0;
> +
>  	/* ensure new counts are within the maximums */
>  	if (channels.rx_count > curr.max_rx ||
>  	    channels.tx_count > curr.max_tx ||
> -- 
> 2.25.4
> 

Re: [PATCH net-next 0/3] ethtool: set_channels: add a few more checks

[David Miller]

From: Jakub Kicinski <kuba@kernel.org>
Date: Fri, 15 May 2020 12:48:59 -0700

> There seems to be a few more things we can check in the core before
> we call drivers' ethtool_ops->set_channels. Adding the checks to
> the core simplifies the drivers. This set only includes changes
> to the NFP driver as an example.
> 
> There is a small risk in the first patch that someone actually
> purposefully accepts a strange configuration without RX or TX
> channels, but I couldn't find such a driver in the tree.

Series applied, thanks Jakub.

And for the record I accept logical 'or' of booleans as valid :-)

Re: [PATCH net-next 0/3] ethtool: set_channels: add a few more checks

[Michal Kubecek]

On Sat, May 16, 2020 at 01:56:58PM -0700, David Miller wrote:
> From: Jakub Kicinski <kuba@kernel.org>
> Date: Fri, 15 May 2020 12:48:59 -0700
> 
> > There seems to be a few more things we can check in the core before
> > we call drivers' ethtool_ops->set_channels. Adding the checks to
> > the core simplifies the drivers. This set only includes changes
> > to the NFP driver as an example.
> > 
> > There is a small risk in the first patch that someone actually
> > purposefully accepts a strange configuration without RX or TX
> > channels, but I couldn't find such a driver in the tree.
> 
> Series applied, thanks Jakub.
> 
> And for the record I accept logical 'or' of booleans as valid :-)

For the record, my point was that "|=" is a bitwise operator, not logical.

Michal