Re: [PATCH 5/9] libata: implement and use SHT initializers and ops inheritance

[Tejun Heo <htejun@gmail.com>]

Mark Lord wrote:
>>> I think this will need to be re-diff'd against the latest sata_mv,
>>> which now has NCQ support.
>>
>> No problem at all.  Just push it into #upstream.  :-)
> ..
> 
> Oh, it is/was already there.  And I quoted the wrong lines from your patch.
> Let's try again:

Is it?  I don't see it in my copy of #upstream which I synced yesterday.
 Pulling again... Oh.. it's there.  I must have made a mistake while
pulling yesterday.

>>  static struct scsi_host_template mv6_sht = {
>> -       .module                 = THIS_MODULE,
>> -       .name                   = DRV_NAME,
>> -       .ioctl                  = ata_scsi_ioctl,
>> -       .queuecommand           = ata_scsi_queuecmd,
>> -       .can_queue              = ATA_DEF_QUEUE,
>> -       .this_id                = ATA_SHT_THIS_ID,
>> +       ATA_BASE_SHT(DRV_NAME),
>>         .sg_tablesize           = MV_MAX_SG_CT / 2,
>> -       .cmd_per_lun            = ATA_SHT_CMD_PER_LUN,
>> -       .emulated               = ATA_SHT_EMULATED,
>> -       .use_clustering         = 1,
>> -       .proc_name              = DRV_NAME,
>>         .dma_boundary           = MV_DMA_BOUNDARY,
>> -       .slave_configure        = ata_scsi_slave_config,
>> -       .slave_destroy          = ata_scsi_slave_destroy,
>> -       .bios_param             = ata_std_bios_param,
>>  };
> 
> That doesn't look correct.  :)
> Shouldn't it be ATA_NCQ_SHT(DRV_NAME) there?

On top of a984f58dd97f22f5113700322fed311a0ee29947, it is correct
because .can_queue is ATA_DEF_QUEUE and no queue_depth callback is
specified.  Anyways, after review, I'll regenerate the patchset on top
of the current sata_mv.

Thanks.

-- 
tejun