Re: [PATCH] libata: IORDY handling, quiet handling

[Sergei Shtylyov <sshtylyov@ru.mvista.com>]

Hello, I wrote:

>> @@ -3466,8 +3529,18 @@
>>      tf.command = ATA_CMD_SET_FEATURES;
>>      tf.feature = SETFEATURES_XFER;
>>      tf.flags |= ATA_TFLAG_ISADDR | ATA_TFLAG_DEVICE;
>> +
>> +    /* Older CFA may not support this command */
>> +    if (ata_id_is_cfa(dev->id))
>> +            tf.flags |= ATA_TFLAG_QUIET;
>> +
>>      tf.protocol = ATA_PROT_NODATA;
>> -    tf.nsect = dev->xfer_mode;
>> +
>> +    /* Ancient devices may need us to avoid IORDY */
>> +    if (ata_pio_need_iordy(dev))
>> +        tf.nsect = dev->xfer_mode;
>> +    else
>> +        tf.nsect = 0x01;

>    Note that ATA-1 did *not* yet define this subcode (only 0 for "block 
> transfer"). I think that we should not change the PIO modes at all on 
> non-IORDY drives, otherwise it's becoming pointless and messy -- you're

    After thinking a bit more: we should not do it on non-EIDE drives 
(however, this is probably the same thing).
    If our host by some mischance has no IORDY support and an EIDE drive has 
it and supports disabling (otherwise there's little we can do), we must issue 
subcode 0x01 and tune our host controller to the default PIO mode.

> not setting what you're told here and force the drive's default mode 
> instead... why then change it at all?

    Anyway, all this logic should be one layer higher than this function...

MBR, Sergei