Re: [PATCH net-next] eth: fbnic: fix memory corruption in fbnic_tlv_attr_get_string()

[Lee Trager]

On 3/7/25 1:28 AM, Dan Carpenter wrote:

> This code is trying to ensure that the last byte of the buffer is a NUL
> terminator.  However, the problem is that attr->value[] is an array of
> __le32, not char, so it zeroes out 4 bytes way beyond the end of the
> buffer.  Cast the buffer to char to address this.
>
> Fixes: e5cf5107c9e4 ("eth: fbnic: Update fbnic_tlv_attr_get_string() to work like nla_strscpy()")
> Signed-off-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
> ---
>   drivers/net/ethernet/meta/fbnic/fbnic_tlv.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/ethernet/meta/fbnic/fbnic_tlv.c b/drivers/net/ethernet/meta/fbnic/fbnic_tlv.c
> index d558d176e0df..517ed8b2f1cb 100644
> --- a/drivers/net/ethernet/meta/fbnic/fbnic_tlv.c
> +++ b/drivers/net/ethernet/meta/fbnic/fbnic_tlv.c
> @@ -261,7 +261,7 @@ ssize_t fbnic_tlv_attr_get_string(struct fbnic_tlv_msg *attr, char *dst,
>   		return -E2BIG;
>   
>   	srclen = le16_to_cpu(attr->hdr.len) - sizeof(*attr);
> -	if (srclen > 0 && attr->value[srclen - 1] == '\0')
> +	if (srclen > 0 && ((char *)attr->value)[srclen - 1] == '\0')
>   		srclen--;
>   
>   	if (srclen >= dstsize) {

Thanks for catching that. While I didn't see any negative effect without 
it I have verified on hardware this patch works.

Reviewed-by: Lee Trager <lee@xxxxxxxxx>