Re: [PATCH] libata: Add pata_atp867x driver for Artop/Acard ATP867X controllers

["Jung-Ik (John) Lee" <jilee@google.com>]

On Sat, Sep 12, 2009 at 5:38 PM, Jeff Garzik <jgarzik@pobox.com> wrote:
> On 09/12/2009 08:17 PM, Jung-Ik (John) Lee wrote:
>>
>> On Sat, Sep 12, 2009 at 9:49 AM, Robert Hancock<hancockrwd@gmail.com>
>>  wrote:
>>>
>>> On 09/12/2009 04:59 AM, Jung-Ik (John) Lee wrote:
>>>
>>> (snip)
>>>
>>> Looks mostly reasonable to me, other than a few issues:
>>>
>>>> +static void atp867x_set_piomode(struct ata_port *ap, struct ata_device
>>>> *adev)
>>>> +{
>>>> +       struct ata_device *peer = ata_dev_pair(adev);
>>>> +       struct atp867x_priv *dp = ap->private_data;
>>>> +       u8 speed = adev->pio_mode;
>>>> +       struct ata_timing t, p;
>>>> +       int T, UT;
>>>> +       u8 b;
>>>> +
>>>> +       T = 1000000000 / 33333;
>>>> +       UT = T/4;
>>>> +
>>>> +       switch (speed) {
>>>> +       case XFER_PIO_4:
>>>> +       case XFER_PIO_3:
>>>> +       case XFER_PIO_2:
>>>> +       case XFER_PIO_1:
>>>> +       case XFER_PIO_0:
>>>> +       case XFER_PIO_SLOW:
>>>> +               break;
>>>> +       default:
>>>> +               printk(KERN_WARNING "ATP867X: Unsupported speed %#x."
>>>> +                       " Default to XFER_PIO_0.\n", (unsigned)speed);
>>>> +               speed = XFER_PIO_0;
>>>> +       }
>>>> +
>>>> +       ata_timing_compute(adev, speed,&t, T, UT);
>>>> +       if (peer&&    peer->pio_mode) {
>>>> +               ata_timing_compute(peer, peer->pio_mode,&p, T, UT);
>>>> +               ata_timing_merge(&p,&t,&t, ATA_TIMING_8BIT);
>>>> +       }
>>>> +
>>>> +       b = inb(dp->dma_mode);
>>>> +       if (adev->devno&    1)
>>>> +               b = (b&    ~ATP867X_IO_DMAMODE_SLAVE_MASK);
>>>> +       else
>>>> +               b = (b&    ~ATP867X_IO_DMAMODE_MSTR_MASK);
>>>> +       outb(b, dp->dma_mode);
>>>> +
>>>> +#ifdef ATP867X_NO_HACK_PIOMODE
>>>> +       b = atp867x_get_active_clocks_shifted(t.active) |
>>>> +               atp867x_get_recover_clocks_shifted(t.recover);
>>>> +#else
>>>> +       /*
>>>> +        * magic value that works (from doc 6.4, 6.6.9)
>>>> +        */
>>>> +       b = 0x31;
>>>> +#endif
>>>
>>> What's the purpose of this ifdef?
>>
>> The magic value part must go. I'll update the patch.
>>
>>>
>>>> +       if (dp->pci66mhz)
>>>> +               b += 0x10;
>>>> +
>>>> +       if (adev->devno&    1)
>>>> +               outb(b, dp->slave_piospd);
>>>> +       else
>>>> +               outb(b, dp->mstr_piospd);
>>>> +
>>>> +       /*
>>>> +        * use the same value for comand timing as for PIO timimg
>>>> +        */
>>>> +       outb(b, dp->eightb_piospd);
>>>> +}
>>>> +
>>>> +static int atp867x_cable_detect(struct ata_port *ap)
>>>> +{
>>>> +       return ATA_CBL_PATA40_SHORT;
>>>> +}
>>>
>>> Doesn't the controller have a way to do proper 80-wire cable detection?
>>
>> No programmatic way. libata.force should be used for other configurations.
>
> Ideally you would use ATA_CBL_PATA_UNK / ata_cable_unknown, but it seems
> that ATA_CBL_PATA_UNK has grown rather useless in the current kernel.
>
> We should probably (a) update libata-core/libata-sff to do something
> intelligent with ATA_CBL_PATA_UNK, and (b) use ATA_CBL_PATA_UNK in
> pata_atp867x.
>
> But yes, libata.force works as well.  It just seems like we are losing an
> opportunity to accept an existing BIOS-configured 80-wire cable
> configuration by hardcoding ATA_CBL_PATA40_SHORT.

Right. But from my experience relying on BIOS just doesn't seem to
work well in all cases though.

>
> Also, a specific note -- atp867x_cable_detect() should be moved into
> drivers/ata/libata-core.c and made generic.  We have ata_cable_xxx versions
> for everything except 40-short, it seems.

Sounds right to me.

>
>        Jeff
>
>
>
>
>