On Sat, Mar 9, 2019 at 6:54 PM Wolfram Sang
<wsa+renesas@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> Make sure we report no buffer for 0-length messages. This can only
> happen if threshold is set to 0 which is kind of bogus but we should
> handle this situation nonetheless.
>
> Reported-by: Hsin-Yi Wang <hsinyi@xxxxxxxxxxxx>
> Fixes: e94bc5d18be0 ("i2c: add helpers to ease DMA handling")
> Signed-off-by: Wolfram Sang <wsa+renesas@xxxxxxxxxxxxxxxxxxxx>
> ---
> drivers/i2c/i2c-core-base.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/i2c/i2c-core-base.c b/drivers/i2c/i2c-core-base.c
> index cb6c5cb0df0b..22c8f38bf68c 100644
> --- a/drivers/i2c/i2c-core-base.c
> +++ b/drivers/i2c/i2c-core-base.c
> @@ -2268,7 +2268,8 @@ EXPORT_SYMBOL(i2c_put_adapter);
> */
> u8 *i2c_get_dma_safe_msg_buf(struct i2c_msg *msg, unsigned int threshold)
> {
How about also printing some warning if threshold is zero? Since
caller might expect
that if msg->len >= threshold, they will get valid pointer. But they
will get valid pointer
if (msg->len, threshold) is (>0, 0) and NULL if (=0, 0). So warn them
that the threshold
is bogus.
> - if (msg->len < threshold)
> + /* also skip 0-length msgs for bogus thresholds of 0 */
> + if (msg->len < threshold || msg->len == 0)
> return NULL;
>
> if (msg->flags & I2C_M_DMA_SAFE)
> --
> 2.19.1
>
