Names must be null-terminated strings. If a name which is not 
null-terminated is passed through netlink, strstr() and similar 
functions can cause buffer overrun. This patch fixes this issue 
by returning -EINVAL if a non-null-terminated name is passed.

Signed-off-by: Ma Ke <make_ruc2021@163.com>
---
 net/tipc/node.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/net/tipc/node.c b/net/tipc/node.c
index 3105abe97bb9..a02bcd7e07d3 100644
--- a/net/tipc/node.c
+++ b/net/tipc/node.c
@@ -2519,6 +2519,9 @@ int tipc_nl_node_get_link(struct sk_buff *skb, struct genl_info *info)
 		return -EINVAL;
 
 	name = nla_data(attrs[TIPC_NLA_LINK_NAME]);
+	if (name[strnlen(name,
+			 nla_len(attrs[TIPC_NLA_LINK_NAME]))] != '\0')
+		return -EINVAL;
 
 	msg.skb = nlmsg_new(NLMSG_GOODSIZE, GFP_KERNEL);
 	if (!msg.skb)
-- 
2.37.2

On Fri, 13 Oct 2023 15:04:08 +0800 Ma Ke wrote:
> Names must be null-terminated strings. If a name which is not 
> null-terminated is passed through netlink, strstr() and similar 
> functions can cause buffer overrun. This patch fixes this issue 
> by returning -EINVAL if a non-null-terminated name is passed.
> 
> Signed-off-by: Ma Ke <make_ruc2021@163.com>

You have now sent 14 incorrect fixes to Linux networking. 
And not a single correct one.
I would like to politely ask you to stop.

Ma Ke <make_ruc2021@163.com> wrote:
> Names must be null-terminated strings. If a name which is not 
> null-terminated is passed through netlink, strstr() and similar 
> functions can cause buffer overrun. This patch fixes this issue 
> by returning -EINVAL if a non-null-terminated name is passed.
> 
> Signed-off-by: Ma Ke <make_ruc2021@163.com>
> ---
>  net/tipc/node.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/net/tipc/node.c b/net/tipc/node.c
> index 3105abe97bb9..a02bcd7e07d3 100644
> --- a/net/tipc/node.c
> +++ b/net/tipc/node.c
> @@ -2519,6 +2519,9 @@ int tipc_nl_node_get_link(struct sk_buff *skb, struct genl_info *info)
>  		return -EINVAL;
>  
>  	name = nla_data(attrs[TIPC_NLA_LINK_NAME]);
> +	if (name[strnlen(name,
> +			 nla_len(attrs[TIPC_NLA_LINK_NAME]))] != '\0')
> +		return -EINVAL;

If the existing userspace is passing 0-terminated strings it would be
better to fix the policy (tipc_nl_link_policy) instead (and set NLA_NUL_STRING).

And if not, above change breaks userspace.

On 13/10/2023 08:04, Ma Ke wrote:
> Names must be null-terminated strings. If a name which is not
> null-terminated is passed through netlink, strstr() and similar
> functions can cause buffer overrun. This patch fixes this issue
> by returning -EINVAL if a non-null-terminated name is passed.
> 
> Signed-off-by: Ma Ke <make_ruc2021@163.com>
> ---
>   net/tipc/node.c | 3 +++
>   1 file changed, 3 insertions(+)
> 
> diff --git a/net/tipc/node.c b/net/tipc/node.c
> index 3105abe97bb9..a02bcd7e07d3 100644
> --- a/net/tipc/node.c
> +++ b/net/tipc/node.c
> @@ -2519,6 +2519,9 @@ int tipc_nl_node_get_link(struct sk_buff *skb, struct genl_info *info)
>   		return -EINVAL;
>   
>   	name = nla_data(attrs[TIPC_NLA_LINK_NAME]);
> +	if (name[strnlen(name,
> +			 nla_len(attrs[TIPC_NLA_LINK_NAME]))] != '\0')
> +		return -EINVAL;

The better choice would be to use strncmp() with limit of
TIPC_MAX_LINK_NAME in tipc_node_find_by_name().
This patch fixes tipc_nl_node_get_link(), but the same pattern is used
in tipc_nl_node_set_link() and tipc_nl_node_reset_link_stats(), these
functions also need improvements. Changes to strncmp() and strnstr()
will fix all spots.

>   
>   	msg.skb = nlmsg_new(NLMSG_GOODSIZE, GFP_KERNEL);
>   	if (!msg.skb)