Re: [PATCH 1/2] [libata] sata_mv: Remove PCI dependency

[Jeff Garzik <jeff@garzik.org>]

saeed bishara wrote:
> -	if (unlikely(irq_stat & PCI_ERR)) {
> +	if (unlikely(irq_stat & PCI_ERR) && HAS_PCI(host)) {
>  		mv_pci_error(host, mmio);
>  		handled = 1;
>  		goto out_unlock;	/* skip all other HC irq handling */

the unlikely() should cover the entire expression.



>  static int __init mv_init(void)
>  {
> -	return pci_register_driver(&mv_pci_driver);
> +	int rc;
> +#ifdef CONFIG_PCI
> +	rc = pci_register_driver(&mv_pci_driver);
> +	if (rc)
> +		return rc;
> +#endif
> +	return 0;
>  }

I would do

{
	int rc = -ENODEV;

	#ifdef CONFIG_PCI
	rc = pci_register_driver(...);
	#endif

	return rc;
}

to ensure sane non-SoC, non-PCI behavior (which this patch now enables).

Finally, in Kconfig, even when removing the CONFIG_PCI dependency, the 
driver suddenly has other unsatified dependencies:  CONFIG_HAS_DMA and 
CONFIG_HAS_IOMEM.  Those two need to be added to the Kconfig dep list.

I would have made these minor corrections myself, but git-am (main 
kernel patch-apply tool, for git users) doesn't seem to like the patch:

Applying sata_mv: Remove PCI dependency

fatal: corrupt patch at line 59