Re: [PATCH 08/10] block: cleanup rq->data_len usages

[Tejun Heo <tj@kernel.org>]

Hello, Boaz.

Boaz Harrosh wrote:
>>  	/* The req and req->next_rq have not been completed */
>> -	BUG_ON(blk_end_bidi_request(req, 0, dlen, next_dlen));
>> +	BUG_ON(blk_end_bidi_request(req, 0, blk_rq_bytes(req),
>> +				    blk_rq_bytes(req->next_rq)));
>>  
> 
> Just blk_end_request_all() actually. If you let blk_end_request_all
> also do bidi

I wrote in the other reply but I don't want to make any other change
than converting direct accesses to accessors in this patch, so that we
can _know_ this patch is safe.  Patches in this series are already all
over the place.  But, yeah, cleaning this up would be great.  Care to
send a patch?

>>  	scsi_release_buffers(cmd);
>>  
> 
> and then it can be in-lined into caller, where blk_pc_request(req)
> just do blk_end_request_all regardless (and only the residual
> setting is conditional)
> 
> I'll send a patch to scsi_lib later once this settles a bit.

Ah.. cool.  :-)

>> @@ -966,7 +965,7 @@ static int scsi_init_sgtable(struct request *req, struct scsi_data_buffer *sdb,
>>  	BUG_ON(count > sdb->table.nents);
>>  	sdb->table.nents = count;
>>  	if (blk_pc_request(req))
>> -		sdb->length = req->data_len;
>> +		sdb->length = blk_rq_bytes(req);
>>  	else
>>  		sdb->length = blk_rq_sectors(req) << 9;
> 
> Is this true. I thought they must be the same now. I was actually
> anticipating this if() removed.

Replied in the other reply.

>>  	/* FIXME: should be include in osd_sense_info */
>>  	if (in_resid)
>> -		*in_resid = or->in.req ? or->in.req->data_len : 0;
>> +		*in_resid = or->in.req ? blk_rq_bytes(or->in.req) : 0;
> 
> +		*in_resid = or->in.req ? or->in.req->resid_len : 0;
> 
>>  
>>  	if (out_resid)
>> -		*out_resid = or->out.req ? or->out.req->data_len : 0;
>> +		*out_resid = or->out.req ? blk_rq_bytes(or->out.req) : 0;
> 
> +		*out_resid = or->out.req ? or->out.req->resid_len : 0;
> 
>>  
>>  	return ret;
>>  }
> 
> OK This segment is wrong. It should be moved to the residual count patch
> (PATCH 3/10) the assignment into *in_resid/*out_resid should shourly triggered
> a warning ;-)

Ah... thanks a lot for spotting it.  Will fix.

Thanks.

-- 
tejun