[Slirp] [PATCH] slirp: ncsi: compute checksum for valid data length

[Philippe Mathieu-Daudé]

Hi Prasad,

On 12/30/19 7:39 AM, P J P wrote:
> From: Prasad J Pandit <pjp at fedoraproject.org>
> 
> 'ncsi_calculate_checksum' computes an optional checksum value for
> the ncsi response packet by reading the data as series of 2 byte
> words. But it receives the data length in no of bytes. Fix the for

I suppose by 'no' you mean 'number'.

> loop to run for half the iterations to compute checksum for valid
> data bytes and avoid OOB access.
> 
> Reported-by: Xingwei Lin <linyi.lxw at antfin.com>
> Signed-off-by: Prasad J Pandit <pjp at fedoraproject.org>
> ---
>   src/ncsi.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/src/ncsi.c b/src/ncsi.c
> index 6864b73..ddd980d 100644
> --- a/src/ncsi.c
> +++ b/src/ncsi.c
> @@ -47,7 +47,7 @@ static uint32_t ncsi_calculate_checksum(uint16_t *data, int len)
>        * 32-bit unsigned sum of the NC-SI packet header and NC-SI packet
>        * payload interpreted as a series of 16-bit unsigned integer values.
>        */
> -    for (i = 0; i < len; i++) {
> +    for (i = 0; i < len / 2; i++) {

We could replace '2' by sizeof(data[0]) but since the type will never 
change, 2 is easier.

>           checksum += htons(data[i]);
>       }
>   

The libslirp README says [*] you can send patches to the list, but last 
time I sent something the maintainers told me they prefer a GitLab merge 
request, I didn't insisted and my patch got lost. But then if you open a 
GitLab merge request the list isn't Cc'ed and patch aren't reviewed by 
subscribers, so I dunno what's best.

Anyway:
Reviewed-by: Philippe Mathieu-Daudé <philmd at redhat.com>

[*] https://gitlab.freedesktop.org/slirp/libslirp