Re: [PATCH] libata: implement ata_wait_after_reset()

[Tejun Heo <htejun@gmail.com>]

Jeff Garzik wrote:
> Tejun Heo wrote:
>> -    msleep(150);
>> +    /* wait a while before checking status */
>> +    ata_wait_after_reset(ap, deadline);
> [...]
>> -    msleep(150);
>> +    /* wait a while before checking status */
>> +    ata_wait_after_reset(ap, deadline);
>>  
>>      /* Before we perform post reset processing we want to see if
>>       * the bus shows 0xFF because the odd clown forgets the D7
>> @@ -3543,8 +3583,8 @@ int sata_std_hardreset(struct ata_port *
>>          return 0;
>>      }
>>  
>> -    /* wait a while before checking status, see SRST for more info */
>> -    msleep(150);
>> +    /* wait a while before checking status */
>> +    ata_wait_after_reset(ap, deadline);
>>  
>>      rc = ata_wait_ready(ap, deadline);
> [...]
>> -    msleep(150);
>> +    /* wait a while before checking status */
>> +    ata_wait_after_reset(ap, deadline);
>>  
>>      /* Before we perform post reset processing we want to see if
>>       * the bus shows 0xFF because the odd clown forgets the D7
>> Index: work/drivers/ata/sata_inic162x.c
>> ===================================================================
>> --- work.orig/drivers/ata/sata_inic162x.c
>> +++ work/drivers/ata/sata_inic162x.c
>> @@ -446,7 +446,7 @@ static int inic_hardreset(struct ata_por
>>          struct ata_taskfile tf;
>>  
>>          /* wait a while before checking status */
>> -        msleep(150);
>> +        ata_wait_after_reset(ap, deadline);
>>  
>>          rc = ata_wait_ready(ap, deadline);
> [...]
> 
> The main thing that bothers me is not the increase in delay, but the
> fact that this create converts a delay/Status-poll sequence into a
> delay/Status-poll/Status-poll sequence.
> 
> ata_wait_after_reset() immediately before ata_wait_ready() seems highly
> redundant.  Why not just poll Status once?

I was trying to minimize code disturbance around reset such that
ata_wait_after_reset() can be drop-in replacement for msleep(150).  This
was for two reasons 1. as this patch was to fix regression I didn't want
to introduce a lot of change into -rcX and 2. I was lazy.  :-)

As dont-consider-0xff-as-port-empty-if-sstatus-available patch fixes the
regression nicely, I think we can delay this to 2.6.23.  I'll merge
ata_wait_after_reset() into ata_wait_ready() (or the other way around).

Thanks.

-- 
tejun