>
> > static int ufs_bsg_verify_query_size(struct ufs_hba *hba,
> > unsigned int request_len,
> > - unsigned int reply_len,
> > - int desc_len, enum query_opcode desc_op)
> > + unsigned int reply_len)
> > {
> > int min_req_len = sizeof(struct ufs_bsg_request);
> > int min_rsp_len = sizeof(struct ufs_bsg_reply);
> >
> > - if (desc_op == UPIU_QUERY_OPCODE_WRITE_DESC)
> > - min_req_len += desc_len;
> > -
>
> I think this calling convention cleanup should go into a ѕeparate
> prep patch with its own changelog.
Done
>
> > + descp = kzalloc(*desc_len, GFP_KERNEL);
> > + if (!descp)
> > + return -ENOMEM;
> > +
> > + if (desc_op == UPIU_QUERY_OPCODE_WRITE_DESC)
> > + sg_copy_to_buffer(job->request_payload.sg_list,
> > + job->request_payload.sg_cnt, descp,
> > + *desc_len);
>
> So we always need to bounce buffer here? Is there any good reason
> we can't pass through the sglist to the low-level functions? and
> share the low-level code used in ->queuecommand?
This is a device management command - there isn't really a data phase here,
And we are not using any component of queuecommand,
Specifically not the scsi command sg list.
>
> Also isn't this an ABI change for the write side? Are we sure there
> are not existing users relying on it? Is there an API document
> that needs to be updated?
Right. I did updated Documentation/scsi/ufs.txt.
It's in the beginning of the patch, you might have missed it.
I am counting that ufs-utils will be the first utility to use ufs-bsg,
and it's waiting for this series to get accepted first.
>
> Last but not least the commit log needs to be a lot more detailed.
Done.
