Re: [PATCH] i2c: qup: Fix error return code in qup_i2c_bam_schedule_desc()

[Bjorn Andersson]

On Mon 16 Nov 08:10 CST 2020, Zhihao Cheng wrote:

> Fix to return the error code from qup_i2c_change_state()
> instaed of 0 in qup_i2c_bam_schedule_desc().
> 
> Fixes: fbf9921f8b35d9b2 ("i2c: qup: Fix error handling")
> Reported-by: Hulk Robot <hulkci@xxxxxxxxxx>
> Signed-off-by: Zhihao Cheng <chengzhihao1@xxxxxxxxxx>
> ---
>  drivers/i2c/busses/i2c-qup.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/i2c/busses/i2c-qup.c b/drivers/i2c/busses/i2c-qup.c
> index fbc04b60cfd1..5a47915869ae 100644
> --- a/drivers/i2c/busses/i2c-qup.c
> +++ b/drivers/i2c/busses/i2c-qup.c
> @@ -801,7 +801,8 @@ static int qup_i2c_bam_schedule_desc(struct qup_i2c_dev *qup)
>  	if (ret || qup->bus_err || qup->qup_err) {
>  		reinit_completion(&qup->xfer);
>  
> -		if (qup_i2c_change_state(qup, QUP_RUN_STATE)) {
> +		ret = qup_i2c_change_state(qup, QUP_RUN_STATE);

In the case that we entered this block because ret was -ETIMEDOUT then
this will overwrite this and the function will return -EIO.

But in the other paths out of this block ret is being overwritten
anyways, so I think it's fine.

Reviewed-by: Bjorn Andersson <bjorn.andersson@xxxxxxxxxx>

Regards,
Bjorn

> +		if (ret) {
>  			dev_err(qup->dev, "change to run state timed out");
>  			goto desc_err;
>  		}
> -- 
> 2.25.4
>