Hi, Bean
Thank you for your advice! Moving original pm_runtime_put_sync()
to after "out" label will influence an error path branched from
ups_bsg_verify_query_size(). So I think changing "goto out" to
"break" is a good idea. But in this case we may execute an extra
sg_copy_from_buffer() and an extra kfree() compared with unpatched
version. Does this matter?
Regards,
Dinghao
> > 1 file changed, 3 insertions(+), 1 deletion(-)
> Hi, Dinghao
> >
> > diff --git a/drivers/scsi/ufs/ufs_bsg.c b/drivers/scsi/ufs/ufs_bsg.c index
> > 53dd87628cbe..516a7f573942 100644
> > --- a/drivers/scsi/ufs/ufs_bsg.c
> > +++ b/drivers/scsi/ufs/ufs_bsg.c
> > @@ -106,8 +106,10 @@ static int ufs_bsg_request(struct bsg_job *job)
> > desc_op = bsg_request->upiu_req.qr.opcode;
> > ret = ufs_bsg_alloc_desc_buffer(hba, job, &desc_buff,
> > &desc_len, desc_op);
> > - if (ret)
> > + if (ret) {
> > + pm_runtime_put_sync(hba->dev);
>
> No need to add pm_runtime_put_sync() here, you can change "goto out" to
> "break",
> Or move original pm_runtime_put_sync() to after goto label.
>
> > goto out;
> > + }
> >
> > /* fall through */
> > case UPIU_TRANSACTION_NOP_OUT:
> > --
> > 2.17.1
