Re: [PATCH] scsi: qla2xxx: Re-use existing error handling path

From: Bart Van Assche
Date: Sun Apr 11 2021 - 11:59:03 EST


On 4/11/21 2:21 AM, Christophe JAILLET wrote:
> The code above this hunk looks spurious to me.
>
> It looks like an error handling path (i.e.
> "if (response_len > bsg_job->reply_payload.payload_len)")
> but returns 0, which is the initial value of 'ret'.
>
> Shouldn't we have ret = -<something> here?

Hmm ... if I read that code path correctly it is on purpose that that
code path returns 0 and error reporting happens by reporting the
EXT_STATUS_BUFFER_TOO_SMALL error code to user space.

> ---
> drivers/scsi/qla2xxx/qla_bsg.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/scsi/qla2xxx/qla_bsg.c b/drivers/scsi/qla2xxx/qla_bsg.c
> index aef2f7cc89d3..d42b2ad84049 100644
> --- a/drivers/scsi/qla2xxx/qla_bsg.c
> +++ b/drivers/scsi/qla2xxx/qla_bsg.c
> @@ -2585,8 +2585,8 @@ qla2x00_get_host_stats(struct bsg_job *bsg_job)
>
> data = kzalloc(response_len, GFP_KERNEL);
> if (!data) {
> - kfree(req_data);
> - return -ENOMEM;
> + ret = -ENOMEM;
> + goto host_stat_out;
> }
>
> ret = qla2xxx_get_ini_stats(fc_bsg_to_shost(bsg_job), req_data->stat_type,

Since the above looks good to me:

Reviewed-by: Bart Van Assche <bvanassche@xxxxxxx>