[PATCH net] ethtool: fix the check logic of at least one channel for RX/TX

From: Yinjun Zhang <redacted>

The command "ethtool -L <intf> combined 0" may clean the RX/TX channel
count and skip the error path, since the attrs
tb[ETHTOOL_A_CHANNELS_RX_COUNT] and tb[ETHTOOL_A_CHANNELS_TX_COUNT]
are NULL in this case when recent ethtool is used.

Tested using ethtool v5.10.

Fixes: 7be92514b99c ("ethtool: check if there is at least one channel for TX/RX in the core")
Signed-off-by: Yinjun Zhang <redacted>
Signed-off-by: Simon Horman <redacted>
Signed-off-by: Louis Peens <redacted>
---
 net/ethtool/channels.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/net/ethtool/channels.c b/net/ethtool/channels.c
index 25a9e566ef5c..e35ef627f61f 100644
--- a/net/ethtool/channels.c
+++ b/net/ethtool/channels.c

@@ -175,14 +175,14 @@ int ethnl_set_channels(struct sk_buff *skb, struct genl_info *info)
 
 	/* 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];
+		err_attr = mod_combined ? tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT] :
+					  tb[ETHTOOL_A_CHANNELS_RX_COUNT];
 	else if (!channels.combined_count && !channels.tx_count)
-		err_attr = tb[ETHTOOL_A_CHANNELS_TX_COUNT];
+		err_attr = mod_combined ? tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT] :
+					  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;

-- 
2.20.1

On Tue, 23 Feb 2021 14:24:40 +0100 Simon Horman wrote:

From: Yinjun Zhang <redacted>

The command "ethtool -L <intf> combined 0" may clean the RX/TX channel
count and skip the error path, since the attrs
tb[ETHTOOL_A_CHANNELS_RX_COUNT] and tb[ETHTOOL_A_CHANNELS_TX_COUNT]
are NULL in this case when recent ethtool is used.

Tested using ethtool v5.10.

Fixes: 7be92514b99c ("ethtool: check if there is at least one channel for TX/RX in the core")
Signed-off-by: Yinjun Zhang <redacted>
Signed-off-by: Simon Horman <redacted>
Signed-off-by: Louis Peens <redacted>

Please make sure you CC Michal on ethtool patches.

In case driver decides to adjust max counts - I'd lean towards:

diff --git a/net/ethtool/channels.c b/net/ethtool/channels.c
index 5635604cb9ba..73d267415819 100644
--- a/net/ethtool/channels.c
+++ b/net/ethtool/channels.c

@@ -116,10 +116,10 @@ int ethnl_set_channels(struct sk_buff *skb, struct genl_info *info)
        struct ethtool_channels channels = {};
        struct ethnl_req_info req_info = {};
        struct nlattr **tb = info->attrs;
-       const struct nlattr *err_attr;
        const struct ethtool_ops *ops;
        struct net_device *dev;
        u32 max_rx_in_use = 0;
+       u32 err_attr;
        int ret;
 
        ret = ethnl_parse_header_dev_get(&req_info,

@@ -157,34 +157,34 @@ int ethnl_set_channels(struct sk_buff *skb, struct genl_info *info)
 
        /* ensure new channel counts are within limits */
        if (channels.rx_count > channels.max_rx)
-               err_attr = tb[ETHTOOL_A_CHANNELS_RX_COUNT];
+               err_attr = ETHTOOL_A_CHANNELS_RX_COUNT;
        else if (channels.tx_count > channels.max_tx)
-               err_attr = tb[ETHTOOL_A_CHANNELS_TX_COUNT];
+               err_attr = ETHTOOL_A_CHANNELS_TX_COUNT;
        else if (channels.other_count > channels.max_other)
-               err_attr = tb[ETHTOOL_A_CHANNELS_OTHER_COUNT];
+               err_attr = ETHTOOL_A_CHANNELS_OTHER_COUNT;
        else if (channels.combined_count > channels.max_combined)
-               err_attr = tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT];
+               err_attr = ETHTOOL_A_CHANNELS_COMBINED_COUNT;
        else
-               err_attr = NULL;
+               err_attr = 0;
        if (err_attr) {
                ret = -EINVAL;
-               NL_SET_ERR_MSG_ATTR(info->extack, err_attr,
+               NL_SET_ERR_MSG_ATTR(info->extack, tb[err_attr],
                                    "requested channel count exceeds maximum");
                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];
+               err_attr = ETHTOOL_A_CHANNELS_RX_COUNT;
        else if (!channels.combined_count && !channels.tx_count)
-               err_attr = tb[ETHTOOL_A_CHANNELS_TX_COUNT];
+               err_attr = ETHTOOL_A_CHANNELS_TX_COUNT;
        else
-               err_attr = NULL;
+               err_attr = 0;
        if (err_attr) {
                if (mod_combined)
-                       err_attr = tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT];
+                       err_attr = 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");
+               NL_SET_ERR_MSG_ATTR(info->extack, tb[err_attr], "requested channel counts would result in no RX or TX channel being configured");
                goto out_ops;
        }
 

On Tue, Feb 23, 2021 at 09:01:06AM -0800, Jakub Kicinski wrote:

On Tue, 23 Feb 2021 14:24:40 +0100 Simon Horman wrote:

quoted

From: Yinjun Zhang <redacted>

The command "ethtool -L <intf> combined 0" may clean the RX/TX channel
count and skip the error path, since the attrs
tb[ETHTOOL_A_CHANNELS_RX_COUNT] and tb[ETHTOOL_A_CHANNELS_TX_COUNT]
are NULL in this case when recent ethtool is used.

Tested using ethtool v5.10.

Fixes: 7be92514b99c ("ethtool: check if there is at least one channel for TX/RX in the core")
Signed-off-by: Yinjun Zhang <redacted>
Signed-off-by: Simon Horman <redacted>
Signed-off-by: Louis Peens <redacted>

Please make sure you CC Michal on ethtool patches.

quoted

diff --git a/net/ethtool/channels.c b/net/ethtool/channels.c
index 25a9e566ef5c..e35ef627f61f 100644
--- a/net/ethtool/channels.c
+++ b/net/ethtool/channels.c

@@ -175,14 +175,14 @@ int ethnl_set_channels(struct sk_buff *skb, struct genl_info *info)
 
 	/* 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];
+		err_attr = mod_combined ? tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT] :
+					  tb[ETHTOOL_A_CHANNELS_RX_COUNT];
 	else if (!channels.combined_count && !channels.tx_count)
-		err_attr = tb[ETHTOOL_A_CHANNELS_TX_COUNT];
+		err_attr = mod_combined ? tb[ETHTOOL_A_CHANNELS_COMBINED_COUNT] :
+					  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;

In case driver decides to adjust max counts - I'd lean towards:

I missed this note while reading the e-mail for the first time so I
thouoght this was just a cleanup. You are right, this would address both
issues.

Michal

On Tue, Feb 23, 2021 at 02:24:40PM +0100, Simon Horman wrote:

From: Yinjun Zhang <redacted>

The command "ethtool -L <intf> combined 0" may clean the RX/TX channel
count and skip the error path, since the attrs
tb[ETHTOOL_A_CHANNELS_RX_COUNT] and tb[ETHTOOL_A_CHANNELS_TX_COUNT]
are NULL in this case when recent ethtool is used.

Tested using ethtool v5.10.

Fixes: 7be92514b99c ("ethtool: check if there is at least one channel for TX/RX in the core")
Signed-off-by: Yinjun Zhang <redacted>
Signed-off-by: Simon Horman <redacted>
Signed-off-by: Louis Peens <redacted>

Reviewed-by: Michal Kubecek <redacted>

On Wed, 24 Feb 2021 01:32:51 +0100 Michal Kubecek wrote:

On Tue, Feb 23, 2021 at 02:24:40PM +0100, Simon Horman wrote:

quoted

From: Yinjun Zhang <redacted>

The command "ethtool -L <intf> combined 0" may clean the RX/TX channel
count and skip the error path, since the attrs
tb[ETHTOOL_A_CHANNELS_RX_COUNT] and tb[ETHTOOL_A_CHANNELS_TX_COUNT]
are NULL in this case when recent ethtool is used.

Tested using ethtool v5.10.

Fixes: 7be92514b99c ("ethtool: check if there is at least one channel for TX/RX in the core")
Signed-off-by: Yinjun Zhang <redacted>
Signed-off-by: Simon Horman <redacted>
Signed-off-by: Louis Peens <redacted>  

Reviewed-by: Michal Kubecek <redacted>

IOW you prefer this to what I proposed?

On Tue, Feb 23, 2021 at 05:02:06PM -0800, Jakub Kicinski wrote:

On Wed, 24 Feb 2021 01:32:51 +0100 Michal Kubecek wrote:

quoted

On Tue, Feb 23, 2021 at 02:24:40PM +0100, Simon Horman wrote:

quoted

From: Yinjun Zhang <redacted>

The command "ethtool -L <intf> combined 0" may clean the RX/TX channel
count and skip the error path, since the attrs
tb[ETHTOOL_A_CHANNELS_RX_COUNT] and tb[ETHTOOL_A_CHANNELS_TX_COUNT]
are NULL in this case when recent ethtool is used.

Tested using ethtool v5.10.

Fixes: 7be92514b99c ("ethtool: check if there is at least one channel for TX/RX in the core")
Signed-off-by: Yinjun Zhang <redacted>
Signed-off-by: Simon Horman <redacted>
Signed-off-by: Louis Peens <redacted>  

Reviewed-by: Michal Kubecek <redacted>

IOW you prefer this to what I proposed?

No, that was my misunderstanding, please see my reply to your e-mail.

Michal

On Tue, 23 Feb 2021 17:02:06 -0800 Jakub Kicinski wrote:

On Wed, 24 Feb 2021 01:32:51 +0100 Michal Kubecek wrote:

quoted

On Tue, Feb 23, 2021 at 02:24:40PM +0100, Simon Horman wrote:  

quoted

From: Yinjun Zhang <redacted>

The command "ethtool -L <intf> combined 0" may clean the RX/TX channel
count and skip the error path, since the attrs
tb[ETHTOOL_A_CHANNELS_RX_COUNT] and tb[ETHTOOL_A_CHANNELS_TX_COUNT]
are NULL in this case when recent ethtool is used.

Tested using ethtool v5.10.

Fixes: 7be92514b99c ("ethtool: check if there is at least one channel for TX/RX in the core")
Signed-off-by: Yinjun Zhang <redacted>
Signed-off-by: Simon Horman <redacted>
Signed-off-by: Louis Peens <redacted>    

Reviewed-by: Michal Kubecek <redacted>  

IOW you prefer this to what I proposed?

Hah, now I missed your email :)