Re: [PATCH #upstream-fixes 1/3] pata_hpt366: fix cable detection,

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

Tejun Heo wrote:

> pata_hpt366 is strange in that its two channels occupy two PCI
> functions and both are primary channels and bit1 of PCI configuration
> register 0x5A indicates cable for both channels.

> Signed-off-by: Tejun Heo <tj@kernel.org>
> Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
> Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com>

    Alas, NAK.

> Index: work/drivers/ata/pata_hpt366.c
> ===================================================================
> --- work.orig/drivers/ata/pata_hpt366.c
> +++ work/drivers/ata/pata_hpt366.c
> @@ -211,11 +211,15 @@ static u32 hpt36x_find_mode(struct ata_p
>  
>  static int hpt36x_cable_detect(struct ata_port *ap)
>  {
> -	u8 ata66;
>  	struct pci_dev *pdev = to_pci_dev(ap->host->dev);
> +	u8 ata66;
>  
> +	/*
> +	 * Each channel of pata_hpt366 occupies separate PCI function
> +	 * as the primary channel and bit1 indicates the cable type.
> +	 */

    No, it doesn't for the secondary channel (which is fucntion 1).

>  	pci_read_config_byte(pdev, 0x5A, &ata66);
> -	if (ata66 & (1 << ap->port_no))
> +	if (ata66 & 2)

    You need to use (pdev->devfn & 1) as a right shift count instead. Not very 
sophisticated indeed. :-)

MBR, Sergei